Energy Efficient Window Systems. Effects on Energy Use and Daylight in Buildings.

Abstract

Cutting edge, world leading, research is carried out in a wide range of subjects including medicine,nano technology, synchrotron radiation, geo sciences and environmental sciences.

Full text

xxxxx
1
Energy-Efficient Window
Systems
Effects on Energy Use and Daylight
in Buildings
Helena Bülow-Hübe
Doctoral Dissertation

Energy-Efficient Window Systems
2
Key words
window, glazing, low-emittance coating, building, energy de-
mand, heating, cooling, solar protection, shading device, solar
energy transmittance, thermal transmittance, simulation, day-
light, user aspects, operative temperature, comfort, perception
© copyright Helena Bülow-Hübe and Division of Energy and Building Design.
Lund University, Lund Institute of Technology, Lund 2001.
Layout: Hans Follin, LTH, Lund
Cover Photo: Jean-Yves Dion
Printed by KFS AB, Lund 2001
Report No TABK--01/1022
Energy-Efficient Window Systems. Effects on Energy Use and Daylight in Buildings.
Department of Construction and Architecture, Lund University, Division of Energy and
Building Design, Lund
ISSN 1103-4467
ISRN LUTADL/TABK--1022-SE
Lund University, Lund Institute of Technology
Department of Construction and Architecture
Division of Energy and Building Design Telephone: +46 46 - 222 73 52
P.O. Box 118 Telefax: +46 46 - 222 47 19
SE-221 00 LUND E-mail: ebd@ebd.lth.se
Sweden Home page: www.byggark.lth.se

Abstract
3
Abstract
This thesis deals with energy-efficient windows in Swedish buildings.
Parametric studies were performed in the dynamic energy simulation tool
Derob-LTH in order to study the effects of window choices on energy
use and indoor climate for both residential and office buildings. A steady-
state program was used to evaluate two years of measurements of energy
use and indoor temperatures of an energy-efficient row-house. Two be-
havioural studies regarding (1) daylight transmittance, view and room
perception using super-insulated windows and (2) the satisfaction with
the daylight environment and the use of shading devices in response to
daylight/sunlight were conducted in full-scale laboratory environments
exposed to the natural climate.
Results show that as the energy-efficiency of buildings increase, win-
dow U-values must decrease in order not to increase the annual heating
demand, since the heating season is shortened, and useful solar gains
become smaller. For single-family houses with a window-to-floor area
ratio of 15 % and insulated according the current Swedish building code,
the U-values should thus on average be lower than 1.0 W/m2K. For houses
insulated according to 1960s standard, the U-value may on average be
1.6 W/m2K. For colder climates (northern Sweden), the U-values should
be somewhat lower, while slightly higher U-values can be tolerated in
milder climates of south Sweden. Thermal comfort during winter is im-
proved for energy-efficient windows. However, overheating problems exist
for both super-insulated houses and highly glazed office buildings show-
ing a need for very low U-values in combination with low g-values. Day-
light experiments indicate that the use of two low-emittance coatings
tints the transmitted daylight enough to be appreciated, and colours may
be perceived as more drab and rooms more enclosed. A compromise be-
tween energy-efficiency and daylighting may be needed, and it is sug-
gested that only one coating be used except when very high energy-effi-
ciency is required.

Energy-Efficient Window Systems
4

Contents
5
Contents
Key words
2
Abstract
3
Contents
5
List of symbols
9
List of articles
11
Foreword
13
1 Introduction
15
1.1 How to read this thesis 15
1.2 Goals 15
1.3 Methods 16
1.4 Limitations 18
1.5 The context 18
1.5.1 Energy related environmental problems 18
1.5.2 Energy use in buildings 20
1.6 Main topic area 1: The role of windows in the energy system 22
1.7 Main topic area 2: Daylighting and view 23
2 Technology status of windows
27
2.1 Performance requirements 28
2.1.1 Sunlight and daylight penetration 28
2.1.2 View out and view in 28
2.1.3 Thermal insulation 29
2.1.4 Air flow, ventilation control and condensation 29
2.1.5 Rain and snow protection 30
2.1.6 Sound insulation 30
2.1.7 Mechanical strength and rigidity 31
2.1.8 Durability 33
2.1.9 Fire protection and fire escape 36
2.1.10 Burglary protection 36
2.1.11 Insect protection 36
2.1.12 Operation, window cleaning and child safety 37
2.1.13 Aesthetically appealing 37
2.1.14 Economical 38
2.1.15 Sustainability 39

Energy-Efficient Window Systems
6
2.2 Background to current window design 40
2.2.1 Changes to building code requirements 42
2.2.2 Technology procurement of energy-efficient windows 45
2.2.3 Aesthetical development 48
2.3 Windows of today 49
2.3.1 Guarantees 52
2.3.2 Quality labelling, P-labelling of windows 53
2.3.3 Energy labelling 53
2.4 Windows of tomorrow 55
3 Basic window physics
57
3.1 UV-transmittance 59
3.2 Light transmittance 59
3.3 Solar energy transmittance 59
3.4 Multiple panes and angle-dependent properties of glass 60
3.5 Glazing for energy-efficiency 62
3.6 Glazing for solar control 66
3.7 Thermal insulation of windows 66
3.7.1 Glazing 67
3.7.2 Sash and frame 74
3.7.3 Total window U-values 76
4 Windows and daylight
81
4.1 General lighting terms 81
4.1.1 Illuminance, E 81
4.1.2 Luminance, L 82
4.1.3 Daylight factor, DF 82
4.1.4 Glare 82
4.2 The sun as the source of daylight 83
4.2.1 Luminance and radiance models of the sky 83
4.2.1 Luminous efficacy 87
4.3 Daylight calculation methods 88
4.3.1 Hand calculation methods 89
4.3.2 Radiosity methods 89
4.3.3 Ray-tracing methods 90
4.4 Daylighting software 90
4.4.1 Pure daylighting programs 90
4.4.2 Thermal programs with daylighting routines 91
4.4.3 Derob-LTH daylight module 93
4.5 Daylight utilisation 94
4.6 Lighting quality and visual comfort 96
4.6.1 Common recommendations for illuminance and luminance. 98
4.7 Psychological aspects of windows 99
4.7.1 View 100
4.7.2 Window size, shape and position 101
4.7.3 Window transmittance and tint 102

Contents
7
4.7.4 Avoidance of glare 105
4.7.5 Sunlight penetration 106
4.8 Non-visual effects of light 107
4.8.1 Physiological effects of solar radiation on the (human) skin 107
4.8.2 Physiological effects of daylight and artificial illumination entering the eye108
4.8.3 Psychological effects of light and colour 109
5 Windows and energy
111
5.1 Single family house 112
5.1.1 Window types 114
5.1.2 Insulation levels and ventilation 116
5.1.3 Effects of window choice on energy demands and indoor temperatures 120
5.1.4 Cost efficiency of window replacement 127
5.1.5 Effects of orientation 130
5.1.6 Effects of site/climate 131
5.1.7 Effects of reduced emittance 135
5.2 Single-person office room 138
5.2.1 Electric lighting savings through daylight utilisation 143
5.3 Office space fully glazed on three sides 145
6 Conclusions and recommendations for further research
155
6.1 Technology status of windows 156
6.2 Energy-efficient windows: the compromise between
energy demand and daylight quality? 157
6.3 Shading devices 158
6.4 Thermal comfort 159
6.5 Tools for daylight calculation 159
6.6 Further research 160
Summary
161
Acknowledgements
169
References
171

Energy-Efficient Window Systems
8

List of symbols
9
List of symbols
α
3deduction of window U-value with respect to insolation (W/m2K)
ε
(hemispherical) emittance (-)
ε
eff effective emittance (-)
λ
wavelength (m)
µ
viscosity (kg/m,s)
θ
incidence angle (°)
ρ
density (kg/m3)
σ
Stefan Boltzmann's constant (5,67·10-8 W/m2K4)
Ψ
linear thermal transmittance (W/m,K)
Aarea (m2) or absorptance (%)
Acog projected area of centre-of-glass (m2)
Aenv aggregate area of surfaces towards heated indoor air (m2)
Aeog projected area of edge-of-glass (m2)
Afprojected area of frame (including sash) (m2)
Aheat heated usable floor area (m2)
Awaggregate area of windows, doors etc. (m2)
Awin projected area of window (m2)
Ain inward flowing fraction of absorbed energy (-)
Aout outward flowing fraction of absorbed energy (-)
dgap width (m)
DF daylight factor (%)
Eilluminance (lux)
Gdegree-hours (°h)
gtotal solar energy transmittance (-)
annual mean value of g
hheat transfer coefficient (W/m2K)
hcconvective heat transfer coefficient (W/m2K)
heexternal heat transfer coefficient (W/m2K)
hiinternal heat transfer coefficient (W/m2K)
hrradiative heat transfer coefficient (W/m2K)
hrb black-body radiative heat transfer coefficient (W/m2K)
Isolar irradiance (W/m2)
IdH diffuse horizontal irradiance (W/m2)
g

Energy-Efficient Window Systems
10
INdirect normal (beam) irradiance (W/m2)
IN,max theoretical direct normal irradiance for clear sky (W/m2)
Kluminous efficacy (lm/W)
kthermal conductivity (W/m,K)
Lluminance (cd/m2)
lgperimeter of visible glass area (m)
Nu Nusselt number (-)
Pilighting load for hour i (W)
Pmax installed lighting power (W)
Preal,i realistic lighting load for hour i (W)
Qnet energy transport during the heating season (Wh/m2)
Qloss thermal heat loss during the heating season (Wh/m2)
Qsolar solar heat gain during the heating season (Wh/m2)
qheat transfer (W/m2)
Rreflectance (%) or thermal resistance (m2K/W)
Rgap thermal resistance of gap between panes (m2K/W)
Rglass thermal resistance of glass pane (m2K/W)
Rse external surface resistance (m2K/W)
Rsi internal surface resistance (m2K/W)
Rsol solar reflectance (%)
Rtot total thermal resistance (m2K/W)
Rvis visible reflectance (%)
Saccumulated solar irradiation (Wh/m2)
SSP sunshine probability (-)
t1, t2surface temperatures (K or °C)
tbbalance temperature of a building (°C)
tmmean temperature (K or °C)
Ttransmittance (%)
Tsol,dir solar direct transmittance (%)
Tsol,tot total solar energy transmittance (%)
Tuv transmittance of UV-radiation (%)
Tvis transmittance of visible radiation (%)
Uthermal transmittance (W/m2K)
Uave average, area weighted, thermal transmittance (W/m2K)
Uave,req required average thermal transmittance (W/m2K)
Ucog thermal transmittance of centre-of-glass (W/m2K)
Ueog thermal transmittance of edge-of-glass (W/m2K)
Ufthermal transmittance of frame (including sash) (W/m2K)
Uwin thermal transmittance of window (W/m2K)
Uwin,p practical thermal transmittance of window (W/m2K)

List of articles
11
List of articles
I Bülow-Hübe, H. (1993). The Solar Village in Dalby – Evaluation
of an Energy-Efficient Row House with Attached Sunspace. Pro-
ceedings of the 3rd Symposium Building Physics in the Nordic Coun-
tries. Copenhagen, Denmark, Sept 13-15 1993. Vol.
2
, 427-431.
II Bülow-Hübe, H. (1998). The Effect of Glazing Type and Size on
Annual Heating and Cooling Demand for Swedish Offices. Proc.
of Renewable Energy Technologies in Cold Climates ’98. Montréal,
Québec, Canada, May 4-6 1998. 188-193.
III Bülow-Hübe, H. (1995). Subjective reactions to daylight in rooms:
Effect of Using Low-emittance Coatings on Windows. Lighting
Research and Technology.
27
(1), 37-44.
IV Bülow-Hübe, H. (2000). Office Worker Preferences of Exterior
Shading Devices: A Pilot Study. Proc. of EuroSun 2000, Copenha-
gen, Denmark, June 19-22 2000. 7 pp.
V Bülow-Hübe, H. (2001) Validation of daylight module in DEROB-
LTH. (submitted to Energy and Buildings).

Energy-Efficient Window Systems
12

Foreword
13
Foreword
When I started my career as researcher in 1990, the intent was never to
come this far as a doctoral dissertation. I was driven by interest in the
issues I encountered. My first taste of the academic world was through a
delegation appointed by the Swedish government with the aim to look at
the “environmental status” and development of a region in Sweden, West-
ern Scania. I participated in the project on energy, where we made predic-
tions of how the energy supply and demand could change over a longer
time period. At that time, the abolishment of Swedish nuclear power had
not yet begun, and one of the aims of the energy project was to see if it
was at all possible, through energy-efficiency improvements, and changes
to the supply system, to meet this goal over a time frame of 20 years.
Given my background as a civil engineer, the energy project awoke my
current interest in the (smaller) energy system of the building. My second
project was therefore to analyse measurements of energy use and tem-
peratures of an energy-efficient row house with some ecological features.
The house was built with a higher insulation standard than required by
the building code. This included for example windows with low-emittance
coatings and a sunspace. The energy demand for this row-house (100
kWh/m2,yr) was lower than the average house, but still not as low as
some of our most optimistic predictions in the energy project above.
Why was this so? One reason was the leakage rate, another was probably
the windows used.
With the high insulation standard of this row house and, for that case,
for most new built houses around 1990, one of the weaker chains was
still the window. The U-value of walls were perhaps on the order of 0.2
W/m2K, while normal windows at that time where approx. 2 W/m2K.
Thus, windows looses 10 times more energy per unit area, at least during
the night. NUTEK, the National board for Technical an Industrial De-
velopment realised this, and decided to improve Swedish window stand-
ards by challenging the industry with a “competition” or technical pro-
curement program. In 1992, they elected two winners.

Energy-Efficient Window Systems
14
The two winning windows both had quadruple panes of glass, of which
two or three panes had low-emittance coatings. This resulted in total
window U-values (including the sash and frame) of less than 0.8 W/
m2K.
However, how far can one go in the hunt for energy-efficiency im-
provements? These windows had a daylight transmittance of less than
60 %, and the low-e coatings made them look green. Was this too much,
or was this acceptable? I studied this in two full-scale rooms using 95
subjects as my “measuring instrument”.
Later I performed a study involving 50 subjects to look at the lighting
preferences of office workers regarding two external shading devices.
The latest project that I have been involved in, deals with the develop-
ment of a computer tool to estimate daylight levels in a room. I believe
that it is important and necessary to intergrate daylight and energy calcu-
lations, in order to estimate the potential for control and regulation of
systems, but it is not sufficient. The human aspects are still not accurately
known. Glare is one aspect, but the constantly changing levels of day-
light is something that is profoundly built into our understanding of the
real world, and is probably most stimulating, and affecting our well-be-
ing in a positive way. Attempts to interweave my knowledge as an engi-
neer with acquired skills on the understanding of human nature, has thus
always been my leading star.
Helena Bülow-Hübe, October 2001

Introduction
15
1 Introduction
1.1 How to read this thesis
The major body of the work behind this thesis is presented in the 5 arti-
cles attached at the end of this book, and also partly in chapter 2 and 5. A
background for the work, a summary of its contents and the limitations
of the work is given is this chapter. Chapter two is a review regarding
mainly performance requirements and also a form of state-of-the-art re-
garding energy aspects of windows. It is based on an earlier report pub-
lished in Swedish (Bülow-Hübe, 1996). Chapter 3 can be read independ-
ently, and is included for those who are not familiar with energy and
physics. It is purposely written for a general audience. Chapter 4 deals
with windows from the daylighting point of view. It begins with some
general knowledge regarding daylight and the lighting of buildings for
those not familiar with the subject. Thereafter follows some notes on the
calculation of daylight and a short presentation of daylighting software.
Benefits and drawbacks of daylight utilisation are presented both from a
technical point of view, as well as from a psychological one. Chapter 5
deals with windows from an energy point of view, and contains paramet-
ric studies on the effect of window choice on annual energy demands and
peak loads. Aspects on economy, daylight utilisation and thermal com-
fort are also included. Conclusions from this work and recommenda-
tions for further research are finally given in chapter 6.
1.2 Goals
There are several motives for using energy-efficient windows, ranging
from the global/national level (e.g. reducing environmentally harmful
emissions) to the individual level (e.g. lower heating bills, better thermal
comfort). However, windows are here to provide for daylight and view.
Therefore, this thesis focuses on the following two, sometimes conflict-
ing, topic areas:

Energy-Efficient Window Systems
16
1) to provide for a good thermal protection against the outdoor envi-
ronment with a minimum of used energy
2) to provide for a good visual daylight environment which satisfies
human needs.
The main goals have been to estimate the potential energy savings with
energy-efficient window systems, and to study the resulting daylight en-
vironment, in order to be able to find the optimum solution in various
Swedish settings and climates. Therefore, the thesis deals mainly with
windows and glazing that have a rather high visual transmittance, and
good thermal insulation (i.e. low U-value). Five articles were written on
different aspects of this subject, and are attached at the end of this book.
One of the articles deals with shading devices, which are usually needed
to prevent from glare and overheating.
Articles I-II falls within the first topic area and represents traditional
engineering work. Articles III and IV represent the second topic area.
Article V is concerned with the calculation of daylight in an energy simu-
lation program. It represents one important but not sufficient attempt to
link together the two topic areas. Further studies on the interaction be-
tween the two topic areas are necessary in order to learn how we in vary-
ing situations can provide for a good thermal protection against the out-
door environment that at the same time yields a low energy use and a
good visual environment, satisfying human needs.
1.3 Methods
The role of the window in the energy balance of buildings has been stud-
ied, mainly by simulations. Two years of measurments of energy use and
indoor temperatures of an inhabited row-house were also available. Com-
puter tools have thus been an important aid to systematic studies on the
effect of e.g. window insulation, solar energy transmittance, window size,
orientation and climate. In some cases other factors such as internal loads
and ventilation rates have been studied to put the potential energy sav-
ings from the window into perspective.
In this thesis two different types of programs have been used, (1) a
steady-state program for the estimation of heating demands on a monthly
basis, the BKL-method (article I), and (2) a dynamic program (on an
hourly basis) for the simulation of heating and cooling demands, indoor
temperatures etc., Derob-LTH, (chapter 5, articles II, V).

Introduction
17
The BKL-method, developed by Källblad (1994), has been the basis
for a commercial program ENORM which is commonly used among
Swedish consultants. The BKL-method is quick and simple to use. How-
ever, it only estimates heating demands. Cooling demands and indoor
temperatures cannot be calculated with this program.
Derob-LTH is a tool used frequently at the department of Construc-
tion and Architecture. It origins from Austin, Texas, (Arumi-Noé &
Wysocki, 1979; Arumi-Noé, 1979), but has been further developed at
the department for Construction and Architecture at Lund University,
Lund Institute of Technology (LTH) over the last 15 years. Its advantages
and capabilities to predict heating and cooling demands accurately has
been demonstrated in several validation studies and comparisons with
full-scale measurements (Wall, 1996 and Wall & Bülow-Hübe, 2001).
Derob-LTH currently has a pleasant user-interface in the MS Windows
environment (Kvist, 2000).
The choice of programs fell on these two since they have both been
developed and validated at our department. There is thus a large know-
ledge about the methods used and the limitations of the two programs.
When large glazed surfaces are studied, for example the current trend of
fully glazed façades in office buildings (see chapter 5), or in atria, it is
important to have tools that uses a geometrical description of the build-
ing, and that treat the distribution of solar radiation within a space in a
detailed way, as demonstrated by Wall (1997).
In articles III and IV, I have used methods from a research field called
environmental psychology. This field is concerned with the environment
as a determinant or influence on behaviour and mood. It is also con-
cerned with the consequences of behaviour on the environment (Bell et
al., 1996). People, or subjects, have been used in a laboratory setting in
order to measure certain aspects of the perceived environment. These
aspects were e.g. the daylight and perception of a room using two differ-
ent types of glazing. So called repeated measures (within-person) designs
were used, which means that each person judged the two situations in
order to reduce the variance, and the number of people required for the
experiment.

Energy-Efficient Window Systems
18
1.4 Limitations
The limitations of this thesis are the following: Windows are studied
from a Swedish perspective, which means that mainly Swedish window
types have been selected in various parametric studies and in Chapters 2
and 5. Also, only Swedish climates have been used in the simulation of
energy demands. However, this does not mean that the results can not be
transfered to other climates. Especially southern Sweden has a climate
comparable to several other north-European countries.
Further, the system border has been put around the heated rooms,
which means that only the energy requried for heating or cooling the
room to a certain temperature has been calculated. The energy demands
have not been converted to bought energy except for in article I or to
primary energy, since this requires several assumptions regarding the sup-
ply system (both in the building and in society in general), and on the
mix of primary energy sources. However, a recent summary over
efficiencies and specific emission levels can be found in a report from the
Swedish Environmental Protection Agency (Naturvårdsverket, 1999).
In the two studies where subjects where used to measure certain as-
pects of the environment (article III and IV), these were both performed
as full-scale laboratory studies. This gives more realistic results than per-
forming scale model experiments, which have been quite common in
early work within the field of environmental or architectural psychology.
In some ways it is better than performing field studies (e.g. in real of-
fices), since confounding factors can be limited and controlled for.
One drawback is however the short exposure time, each subject has
only stayed in each room between 10-15 minutes. Long-term effects of
lighting, such as headaches and eye-strain, cannot be captured in such
short experiments. Rather, it is the immediate impression of the daylight
situation of the room that is captured.
1.5 The context
1.5.1 Energy related environmental problems
Energy use in buildings is strongly connected to serious environmental
problems such as the greenhouse effect, acid rain, eutrophication of land
and waters etc., through the burning of fossil fuels. Nuclear prolifera-
tion, north south problems etc. are also strongly linked to energy use
(Goldemberg et al., 1988).

Introduction
19
The Swedish problem is how to arrive at the goals of nuclear phase-
out and not exploiting the last 4 wild rivers for hydropower without in-
creasing carbon dioxide emissions. There are several other goals as well,
concerning protection of valuable biotopes and emissions reductions of
e.g. volatile organic compounds, particles, sulphur dioxide and nitrogen
oxides, ammonia and heavy metals (Naturvårdsverket, 1999). Improving
the energy-efficiency has been identified as the main key to be able to
maintain our current standard of living, and still allow for a reduction of
harmful emissions of e.g. CO2, NOx and SOx, or at least to maintain
them at current levels (Goldemberg et al., 1988; Mills, 1991). The obser-
vation made is that “electricity per se is not of interest, but rather the
demand for electricity is a reflection of the demand for services it can
provide: hot showers, cold herring, clean clothes, illumination, motive
power, maintenance of comfortable indoor climate, data storage/retrieval
and so on” (Bodlund et al., 1989). The term electricity use above can of
course be replaced by energy use in general, since we are interested in all
the services provided or the tasks accomplished in society.
It may seem difficult to meet the sometimes conflicting environmen-
tal goals. However, several studies have shown that it would be possible
to do so, but that it would require quite dramatic changes to the energy
system (Naturvårdsverket, 1999; Bodlund et al., 1989).
In the studies cited above, the energy use is calculated by an end-use
accounting model. Energy use in society is first divided into sectors of
use, (e.g. industries, agriculture, housing etc), then into areas of services
provided or activities (lighting, heating, clothes washing etc). Tomorrow’s
energy demand is calculated by multiplying each activity with its inten-
sity, which is a reflection of the service desired (Mills, 1991). By applying
different intensities for tomorrow, and by multiplying with expected
growth rates and summing over all sectors, scenarios for total future en-
ergy demand are created. The calculated energy demands are then matched
with different supply scenarios, describing some available options regarding
the main types of energy carriers (primary energy sources) chosen, and
the technologies used to produce electricity and district-heat. Finally,
emissions and costs can be calculated.
These types of studies can also be made for individual countries or
even regions or communities. This was done in 1991 for the region of
Western Scania (Gustavsson et al., 1992) and also for the two largest
communities in that region: Malmö (Johansson, 1990) and Helsingborg
(Bülow-Hübe, 1990). On the demand side we started by collecting sta-
tistics of the current energy-use, and made predictions of the energy-use

Energy-Efficient Window Systems
20
for 2010 1 for three different levels of efficiency improvements. The cal-
culated energy demands where then matched with four supply scenarios,
with increasing efficiency and use of renewable sources. The supply sce-
narios ranged from condensing power with natural gas to cogeneration
with biomass and extensive use of wind power.
In our scenarios for Western Scania we could see that by a consequent
use of the most energy-efficient technologies on the market, cogeneration,
and by introducing biomass on a large scale (short-rotation forests) and
wind-power, it would be possible to phase out nuclear power, not in-
creasing hydropower and maintaining strong economic growth. At the
same time 75 % reductions in carbon dioxide emissions, and 50 % re-
ductions in acidifying gases were achieved compared to the levels of 1988
(Gustavsson et al., 1992).
It must be stressed that it is not predictions of future energy use or
emissions that are made, rather it is a method to identify which measures
are needed to achieve a certain goal, e.g. a reduction of greenhouse gas
emissions or nuclear phase out.
Today, 10 years later, only one of twelve nuclear reactors has been shut
down, and the energy supply system has not changed dramatically. Some
efficient power plants for reserve power have even been shut down, and
Sweden imports electricity generated with coal in inefficient condensing
plants during the winter. It seems like the aim of abolishing nuclear power
by the year 2010 will be very hard to reach, given the slow start during
the last 10 years.
Today’s problem seems to be how to get rid of the barriers preventing
the scenarios from becoming our future reality. For example, more en-
ergy-efficient technologies do enter the market all the time, but if the rate
is high enough to be able to reach the goal we set up for the year 2010 has
not been further studied here. It is likely that incentive programmes are
needed to speed up the process. It also seems like energy-efficiency im-
provements are not the only solution to achieve a sustainable society,
lifestyle changes are probably also necessary.
1.5.2 Energy use in buildings
The total energy use within the building and services sector accounted in
1999 for 150 TWh, or approximately 40 % of Sweden’s total use of en-
ergy. The industry sector accounts for another 40 %, and transportation
1. The end year 2010 was partly chosen because this is the year when nuclear power
should be totally phased out, according to a public referendum in 1980.

Introduction
21
for the remaining 20 %. To this comes distribution and conversion losses,
and foreign maritime trade. The building sector thus has a potentially
large effect on the environment due to its large share of total energy use.
Efficiency improvements have however already lead to a decreased use
of specific gross energy expressed as kWh/yr,m2 of heated floor area. This
is clearly demonstrated by the fact that the energy use within the building
and services sector has remained practically constant since 1970, even if
the heated area has increased by over 50%! (Energimyndigheten 2000;
Byggforskningsrådet, 1995). To give some more figures, the specific gross
energy use (including space heating, domestic hot water and household/
operation electricity) in 1970 was about 340 and 330 kWh/yr,m2 for
one- and two-dwelling buildings and multi-dwelling buildings respec-
tively, and about 380 kWh/yr,m2 for service buildings. In 1994, the spe-
cific energy use had decreased to 210 and 220 kWh/yr,m2 for one- and
two-dwelling buildings and multi-dwelling buildings respectively, and
about 300 kWh/yr,m2 for service buildings. These values apply to the
whole building stock. For new buildings (produced after 1986), the spe-
cific final energy use is about 150 and 175 kWh/yr,m2 for one- and two-
dwelling buildings and multi-dwelling buildings respectively, and about
220 kWh/yr,m2 for service buildings (Byggforskningsrådet, 1995). The
heating demands of buildings thus show a quite remarkable improve-
ment during the last 30 years. New experimental buildings (see below)
also demonstrate that significant further improvements can be achieved,
at least for future buildings.
The electricity use in the sector for buildings and services has however
increased dramatically from 1970 to today: by more than a factor of 3
(Energimyndigheten, 2000). This is of course partly due to an increase of
activities. The switch of primary energy sources that have taken place
between 1970 and today is probably even more important: from mainly
oil in 1970, to especially nuclear power, and increased hydro power and
biomass today. The increase in electricity use also reflects some other
changes, for example an increased use of mechanical ventilation in build-
ings and an increased use of air-conditioning systems and an increased
use of electrical equipment (computers, copy-machines, TVs, videos etc.)
in both offices and homes.
In article I, measurement from an energy-efficient two-storey row house
in Solbyn, Dalby, was evaluated, see also Bülow-Hübe & Blomsterberg
(1992). The dwelling, located at a gable, had a usable floor area of 116
m2. The specific use of bought energy was 100 kWh/m2 of electricity,
including space heating (direct electric heating), domestic hot water and
ventilation, and household electricity. The main features of the dwelling
was an insulation standard above the requirements in the building code,

Energy-Efficient Window Systems
22
controlled ventilation with air-to-air heat recovery, attached sunspace to-
wards south, and small windows towards north. The energy efficiency of
the heat recovery unit was high, about 77 %. The envelope U-values were:
attic ceiling U=0.11, external walls U=0.17, slab on ground U=0.20 and
windows U=1.5 W/m2K. It was shown that the main features leading to
a rather low specific energy use was the heat-recovery and the increased
insulation of the building envelope. The attached sunspace contributed
in an insignificant way to the low energy use, mainly because it was at-
tached on the outside of the well insulated external wall, and of single
glazing. The specific gross energy use was thus about half compared to
the average multi-dwelling unit, and about 30-40 % lower compared to
buildings erected after 1986.
Probably the most energy-efficient row-house built so far in Sweden is
designed to have a specific use of bought energy (all electricity) of only
45-50 kWh/yr,m2. This experimental project was erected in 2001 in Lindås
outside of Gothenburg, and measurement results of energy use are ex-
pected within a year or two. These houses are even better insulated than
Solbyn, extremely air-tight, and uses a mechanical ventilation system with
a very efficient air-to air heat recovery unit. The windows are super-insu-
lated (U=0.85 W/m2K) and solar energy is expected to provide about
50 % of the energy needed for domestic hot water (Maria Wall, Lund
University, personal communication, Aug 2001). If the low predicted
energy use can be achieved, these houses will only use half of the energy
used in Solbyn!
1.6 Main topic area 1: The role of
windows in the energy system
In the type of energy scenarios described in section 1.5.1, the individual
house and its energy demand is treated much like a black box. How these
improvements can be made are left to the building engineers and archi-
tects. The role of the window is not specified by itself, rather it is assump-
tions on the combined effect of energy-efficiency improvements to the
building envelope that is considered, e.g. higher insulation levels, better
windows and air-tightness, heat-recovery of exhaust air etc.
In article II the importance of the window on the energy demand of
an office room has been especially studied. It is shown that the annual
cooling demand of a single-person office room in a Swedish climate is
largely influenced by (in descending order) glazing size, window orienta-
tion, ventilation rate, internal load and daylight utilisation. The heating

Introduction
23
demand is mainly affected by the ventilation rate, climate, orientation
and glazing type. It is demonstrated that daylight utilisation has a poten-
tial of reducing cooling demands without also increasing the heating de-
mands. Further, it is demonstrated that south facing super-insulated win-
dows will gain energy over the year, i.e. they are better than having an
opaque wall. The cooling demand is not higher for the super-insulated
window than for other windows of moderate to high U-values.
In chapter 5 it is again shown that the cooling demand is largely af-
fected by the glazing size, both for the annual cooling demand and for
peak cooling loads. The solar energy transmittance of the glazing also
plays a major role. However, the annual heating demand is mainly influ-
enced by the U-value, and not by the solar energy transmittance. Ther-
mal comfort is also largely dependent on the glazing. During the cooling
season, glazing size and solar energy transmittance are important param-
eters, while the glazing size and U-value are important during the heating
season.
1.7 Main topic area 2: Daylighting and
view
The admission of daylight through windows and the provision of a view
out are the primary functions of windows. If daylighting can be used in a
larger extent to replace artificial lighting, it might be seen as “renewable”
lighting. In earlier studies this was referred to as daylight utilisation, but
later the more specific term “daylight responsive (linked) lighting sys-
tems” was invented. This term refers to advanced control systems that
regulate the light output of the artificial lighting system in response to
the incoming daylight. In this report I have chosen to use the shorter
term daylight utilisation as a synonym for daylight responsive lighting
systems.
Computer tools are valuable to be able to estimate the effect of day-
light responsive lighting systems. Preferably, it should be possible to cal-
culate effects on heating and cooling loads simultaneously with lighting
energy savings. Some solutions exist, for example DOE-2, which is now
being modernised and merged with BLAST into the new simulation en-
gine Energy-Plus (Crawley et al., 2001). Adeline (a lighting simulation
platform, see chapter 4 and Erhorn & Stoffel, 1996) can also be used, but
calculations on lighting energy savings have to be done previous to the

Energy-Efficient Window Systems
24
energy simulation, which is a drawback. Radiance has been used in inter-
active loops together with both TRNSYS and ESP-r (Kovach-Hebling et
al., 1997; Clarke et al., 1997).
A first step to provide for daylight simulations in Derob-LTH, and
later to allow for daylight responsive control of shading devices within
Derob-LTH, is presented in article V. The presented model has some
limitations compared to full-fledged lighting simulation programs like
Radiance, but it still seems to satisfy many needs and may become a
useful tool.
No matter how important the energy-efficiency improvements are seen
in the global context, we must not introduce new technologies that might
possibly be harmful to our health or reduce the comfort and satisfaction
of occupants. Then we have not provided the same service as before.
For energy-efficient windows the question arises mainly around the
effects of introducing one or several low-emittance coatings. The low-e
coatings are made to reduce the thermal losses while affecting the visual
transmittance as little as possible. However, they cannot totally achieve
the same transmittance as ordinary clear glass. Some coatings have a slightly
green or blue tone (silver-based coatings) and some have a slightly brownish
tone (tin-oxide coatings). The effects of using several low-e coatings on
the perception and daylight of a space were therefore studied (article III).
The study shows that quadruple-pane glazing with two low-e coatings
have a significant effect on the daylight transmitted to a room compared
to triple-pane clear glazing, both regarding the amount of daylight, and
also regarding its spectral composition. The results show that people could
distinguish between the two situations, and found the room with the
super-insulated window to be more enclosed and darker, and the day-
light was perceived as more tinted. A situation with a similar triple-pane
glazing with two low-e coatings was never studied with subjects, but
measurements of the spectrum of the transmitted daylight revealed that
the spectrum was closer to that of the super-insulated quadruple glazing,
than to that of the triple-clear window (Bülow-Hübe, 1994). This sug-
gests that it is the use of the two low-e coatings that have the most pro-
found effect on the spectral composition, and thus on perception, while
the fourth, clear glazing only lowers the daylight level slightly. Therefore,
also triple-pane super-insulated window with two low-e coatings may
have negative effects on daylight and perception.
Solar shading devices is another energy-efficient technology, which is
used together with windows. With a good solar shading system, cooling
demands can be reduced dramatically (up to 80 %), see Dubois (1998).
Cooling systems may even be omitted, which will dramatically lower the
cost for the HVAC system (both first cost and running costs). Visual

Introduction
25
comfort, satisfaction and view out are aspects that need to be considered
when solar shading systems are chosen. In an attempt to start to learn
more about user behaviour and their preferences regarding shading de-
vices, a pilot study was performed on two shading devices, an awning
and an exterior Venetian blind (article IV). This study showed no signifi-
cant difference between the two systems, although the awning was appre-
ciated as slightly easier to operate. The effects on view out where moder-
ate and equal for the two devices. Interestingly enough, it was not possi-
ble to find any correlations between the amount of daylight entering the
room (illuminance on desk) or between the luminance of the sky seen
through the window, and to how much occupants decided to pull the
shading devices. Only if there was a sunlight patch somewhere in the
room, was there a weak correlation. This suggests that finding control
algorithms that take human response into account, may be very hard to
find.

Energy-Efficient Window Systems
26

Technology status of windows
27
2 Technology status of
windows
This chapter is an updated and shortened version of an earlier report in
Swedish, see Bülow-Hübe (1996).
The main purpose of a window is to admit light into a building, provide
for a view out and to protect us from the sometimes-harsh outdoor cli-
mate. However, there are many more aspects to window design than this.
Once I was told that a window has to satisfy about 20 different func-
tions, and they can all be fulfilled by building a wall instead of a window.
In the Swedish performance requirements for windows there are namely
no requirements of windows regarding daylight admission!
However, without daylight penetration we can no longer call it a win-
dow. Windows in buildings play a major role in providing quality, com-
fort and satisfaction. The different performance requirements for win-
dows can be summarised in a list:
• Sunlight and daylight penetration
• View out and view in
• Thermal insulation
• Control of air flow and ventilation
• Control of water vapour flow
• Protection against rain and snow
• Sound insulation
• Mechanical strength and rigidity
• Durability
• Fire protection
• Fire escape
• Burglary protection
• Insect protection
• Easy to open
• Window cleaning
• Child safety
• Aesthetically appealing
• Economical
• Sustainability

Energy-Efficient Window Systems
28
2.1 Performance requirements
2.1.1 Sunlight and daylight penetration
The primary purpose of a window is to admit daylight, and to create a
visual contact between inside and outside. This should be done without
distorting the colour of the transmitted light. In residential buildings it is
often desirable to capture the heat from the sun during the heating sea-
son for passive climatization. At the same time the room heat should be
kept indoors. In office buildings solar radiation may be more of a prob-
lem, since it will increase an often-existing cooling demand. Here, a high
daylight transmittance along with a low solar transmittance is desirable.
The UV-part of the radiation is usually not wanted, since it bleaches
textiles, wallpapers etc.
The right to a direct access to daylight is often stipulated in the build-
ing codes, as in the Swedish code (BBR 1999): “Rooms were people stay
more than temporary, shall have a good access to direct daylight. This is
valid for space containing work places, if it is not unreasonable in consid-
eration to the type of activity. Dwellings shall have access to direct sun-
light.” The following advice is given in the code regarding the size of
windows: The window glass area should be at least 10 % of the floor area.
If building parts or other buildings block the daylight more than 20° of
the view angle, the glass area should be increased.
2.1.2 View out and view in
The provision of view out is closly linked with the previous requirement.
To be able to see the changes of the light and weather, to watch over
children etc. are essential aspects of a window. There may also be more
unconscious benefits to windows than previously believed. These con-
cern the influence on life satisfaction as well as environmental satisfac-
tion (Kaplan, 1983, 1985). A nice view with greenery has also been asso-
ciated with faster recovery in post-surgical hospital wards (Ulrich, 1984).
In general, there are four general benefits of windows: (1) access to
environmental information; (2) access to sensory change; (3) a feeling of
connection to the world outside; and (4) restoration and recovery
(Heerwagen, 1990).
Privacy is another issue, which must be dealt with in the architectural
design. Heerwagen uses the two concepts visual access and visual expo-
sure. Visual access is directly linked to the ability of occupants to see out.
Visual exposure is the – sometimes unwanted – possibility to be seen.

Technology status of windows
29
There must be a balance between access and exposure that is appropriate
to the context and for the personal preferences of the occupant. Heerwagen
draws the following matrix between visual access and exposure, Fig. 2.1.
VISUAL ACCESS
High Low
High
The goldfish bowl The interrogation room
(can see and be seen)(cannot see, but
can be seen)
VISUAL
EXPOSURE Low
Ideal The cave
(can see, without (cannot see,
being seen) cannot be seen)
Figure 2.1 Visual Access and Visual Exposure Matrix (after Heerwagen, 1990).
2.1.3 Thermal insulation
The basic principle is that room heat is lost through the window when it
is warmer indoors than outdoors. The thremal performance is described
by the U-value. This is a measure of the heat flux through the window per
unit surface area and degree temperature difference between inside and
outside. It is given in (W/m2K). It is sometimes called the dark U-value,
since it only accounts for heat being lost through the window (e.g. night-
time) and not for incoming solar radiation. Today, when a window U-
value is given, it usually applies to the whole window, including the sash
and frame (Uwin). Glass manufacturers however, usually only gives the
centre-of-glass U-value (Ucog) for the glazing combination itself.
Since the U-value of the glass in modern windows is usually better
than that of the frame, the total U-value should always be stated, since a
bad sash/frame construction can spoil the U-value of an otherwise ac-
ceptable glazing combination.
2.1.4 Air flow, ventilation control and condensation
Windows should be airtight to avoid air leakage, which can affect heat
losses, sound insulation, comfort and risk of condensation. Placing draught
excluders between sash and frame does this. To avoid moisture transport
from inside to outside, the strip is placed on the inner side of the win-
dow. Otherwise, humid air can enter between the panes in a coupled

Energy-Efficient Window Systems
30
window, and condensate on the inside of the cold outer pane. To reduce
dirt accumulation, a dust-absorbing strip is often placed between the
sashes, which allows for some ventilation.
A new phenomenon, which has appeared with highly insulated win-
dows, is condensation forming on the outside of the outer pane (NUTEK,
1995). The condensation can happen during clear nights, and in loca-
tions where the window “sees” a large part of the sky. In the radiation
exchange between the sky and the window, energy is lost to the sky. Since
the heat transport out through the window is small, the outer glass thus
becomes colder than the surrounding air, and condensation is formed.
The frequency of this phenomenon has been studied by Jonsson (1995).
It was found that it usually appears in the spring and autumn during
periods when the air is very humid and the temperature swings between
day and night are high. The condensation starts to form at the bottom of
the glass (the coldest part), and in some cases it is spread over the whole
glass height. The condensation becomes visible in the morning, but usu-
ally dries up a few hours after sunrise. The phenomena starts to appear at
a Ucog of approximately 1-1.3 W/m2K but becomes more frequent with
lower U-values.
2.1.5 Rain and snow protection
Protection against rain and snow penetration is done primarily through
the constructive design of the window, for example with grooves to re-
duce pressure differences and to drain incoming rain water etc. (Fig. 2.2),
and through a proper mounting of the window in the wall. For example,
it is important to make sure that water penetrating the outer panel is
drained outwards, and does not remain at the top of the window frame.
2.1.6 Sound insulation
The two main properties affecting the sound insulation of a window is
the distance between the panes, and the glass thickness. A large air gap is
desirable, since a coupled window with 30-40 mm glass distance has ap-
proximately the same sound insulation as a triple insulating glass unit
(IGU) with two air gaps of 12 mm. It is also preferable to have glass panes
of different thickness, and – in triple-pane windows – to have air gaps of
different thickness (Göransson, 1995).

Technology status of windows
31
Figure 2.2 Groove to reduce pressure differences and to drain rain water (left).
Details at top and bottom of window showing rain screen at a stud
frame with brick cladding (from Mur 90, 1991) (right).
Replacing the air in IGU’s for a gas, for example sulphur hexaflouride
(SF6), can also improve the acoustical properties. However, for traffic
noise, the sound insulation may even decrease when using SF6, and is
therefore not recommended (Jonasson, 1994).
Air-tightness is very important, and several strips improve the sound
insulation. Air-inlets in the frame can drastically reduce the sound insu-
lation. When a very high sound insulation is required, the mounting of
the glass to the sash, as well as the design of sash and frame becomes
important. Windows with good sound insulation can become heavy and
harder to operate.
Total sound insulation is usually not wanted. Informative noise such
as hearing when the mail arrives, or when a child is calling is important.
2.1.7 Mechanical strength and rigidity
It is necessary to consider the window as a whole, and make sure that the
sash and frame have adequate dimensions for the load of the glass and of
normal use. The window must also be able to sustain different external
loads such as wind load. Sometimes there are higher requirements on the
strength than that of normal float glass. Apart from increasing the glass
thickness there are a number of alternative glazing products available, see
below. After Carlson (1992) and Button & Pye (1993).

Energy-Efficient Window Systems
32
Toughened (tempered) glass
By heating the glass to 650 °C, and then cooling it rapidly, compression
stress is built into the surfaces, and tensile stress into the core of the glass.
The bending and tensile strength is thus increased by 4-5 times. It is
neither harder (scratches just as easily) or stiffer (bends down just as much)
as ordinary glass, but it can bend more before it breaks. When toughened
glass fractures, many small pieces (dice) without sharp edges are formed.
This makes it suitable as safety glass. Toughened glass must be cut or
otherwise processed before tempering, since it shatters at all attempts of
processing after tempering. It is also sensitive to mechanical damage at
the edges, or for a sharp object penetrating the compression zone. It can
be used as safety glass for example in offices which are glazed all the way
from the floor and in shop-windows.
Heat strengthened glass
Heat strengthened glass is manufactured in a similar way to toughened
glass, but the cooling process is slower. The strength thus becomes twice
that of ordinary glass. At breakage, larger pieces of glass are formed, which
resemble the fracture of ordinary glass. Therefore, it cannot be used as
safety glass. As with toughened glass, heat strengthened glass cannot be
processed after heat treatment. A main use is for façades.
Laminated glass
Laminated glass is manufactured by bonding two or more sheets of glass
together with a plastic material or resin. Laminated glass built by ordi-
nary float glass is not stronger than ordinary glass, but at fracture the
pieces are kept in place by the plastic foil. The glass pane’s ability to
remain within the construction is also improved. Laminated glass is of-
ten used in glass roofs to prevent the glass from falling down at a poten-
tial fracture, or anywhere where there is a risk that people may fall through
due to a difference in floor levels, for example on balconies. Another use
is for shop-windows, where the plastic film can also be supplied with an
additional UV-filter to reduce bleaching.
Wired glass
Another way of ensuring that the pieces of glass are kept in place at frac-
ture is by embedding a steel wire mesh within the sheet of glass at manu-
facture (in a rolling process). The mechanical strength is lower than for
ordinary glass, and it is also more sensitive to temperature induced stress.
For glass that can become sunlit, the fixing point is critical. Wired glass
should be avoided in places where projected shadows occur. It is used for
fire-protection, burglar protection etc.

Technology status of windows
33
2.1.8 Durability
Window durability
Large facility managers often have requirements on long maintenance
intervals. Thus, the Swedish market for pure wooden windows is reduced
to mainly single family houses. Today, larger facility managers mainly
choose wooden windows with an outer aluminium cladding, aluminium
windows, or to a small extent, plastic windows (Hans Öqvist, SNIRI,2
personal communication, June, 1995).
During the late 1970’s a large number of damages to rather new win-
dows were brought to public attention. The damaged windows were mainly
found in multi-family housing from the 1960’s, in the so-called million-
programme 3 (miljonprogrammet). A number of large investigations re-
garding the size and cause of the damage were carried out. The demands
on durability were then raised by building owners and from the side of
the authorities.
The damage was often caused by a combination of factors, where the
increasingly faster and more industrialised construction of course played
a part. New types of housing, mainly taller buildings, new building tech-
nology, window placement in unprotected locations (pelting rain) were
other factors. The main cause was however lost demands of the treatment
of the timber throughout the production chain, and new types of paint.
At this time the production volume was high, which lead to an increased
rationalisation within the window industry. There were no demands on
the quality or treatment of the raw timber material used in windows,
hence young quickly grown pieces were used in windows. Timber which
had been stored in water was accepted which later led to an increased
spreading of the damage.
Today there has been an improvement, and today’s wooden windows
have better durability than those of the million-programme do. Many
wooden windows are today delivered with aluminium cladding, which
prevents water penetration into the wood, which of cause is the main
cause of rot. A large part of the existing windows have also been covered
with an external metal cladding. In order for this to work, the cladding
must allow for a proper ventilation, an air space of at least 6 mm is rec-
2. SNIRI, The National Association of the Swedish Joinery Factories is the Swedish
trade organisation for producers of joinery, doors, windows, kitchen interiors, staircases
and special interior designs.
3. Caused by a large shortage of housing during the 1950’s, the Swedish government
issued a housing policy with the goal to build one million apartments during a period of
10 years. This was also done, and the result is referred to as the million-programme.

Energy-Efficient Window Systems
34
ommended, with air inlets and outlets. Today it is also recommended
that the wood is primed with oil and oil-based paint (alkyd or linseed oil)
to create a water-repellent ground layer.
According to Gunilla Billgren, Wasakronan (personal communication,
May 1995), a large part of the knowledge that was gained during the late
1970’s and beginning of the 1980’s is falling into oblivion. But there is a
lot of knowledge to be found in somewhat older literature. Today the
discussions are mainly about the timber raw material and about the paint
systems used, and less about the constructive aspects of windows and
window and wall assembly.
According to Karin Wennerståhl, SP (personal communication, May
1995) it is important to separate between the durability of paint systems
and of the wood itself. Earlier, the chemists only focussed on the film of
paint, which was one of the causes of the widespread use of latex paints
for outdoor use, since they were so weather durable. The fact that they
did not work so well together with the wood was an expensive experience
gained a few years later. Today the pendulum has swung in the other
direction and the main approach is from the viewpoint of building phys-
ics or wood technology. The attention is now directed towards the dura-
bility of the wood, and the paint is mostly considered as just a protective
layer.
Insulating glass unit durability
Another issue is the durability of insulating glass units. If the sealant of
the insulating glass fails, this will result in air and moisture penetration.
Milk-white glass is a characteristic sign of such a failure of the IGU.
The edge of the insulated glass unit is the weakest part. Traditionally,
a metal spacer is used to keep the glass panes at the desired distance. The
spacer can be of galvanized steel, extruded aluminium or other low-con-
ductivity materials. The metal spacer is attached to the glass with a
polyisobutylene (butyl) sealant, which also acts as the diffusion barrier
(Wolf & Waters, 1993). An additional sealant of e.g. polysulphide is ap-
plied for extra mechanical stability, i.e. a dual-seal unit, Fig. 2.3. Since
spacers are usually hollow, they are filled with a desiccant to avoid con-
densation forming within the cavity of the unit by the moisture entrapped
at the time of production. Today, the metal spacers usually have bent
corners, and are welded together at one of the long edges. Together with
the dual-seal system, this greatly reduces the risk of potential puncture of
the IGU. (Earlier, single-seal systems were common and the spacer frame
was usually made by four bars connected with corner-keys).

Technology status of windows
35
Butyl
Polysulphide
Dessicant
Spacer
Figure 2.3 Section through the edge of an insulated glass unit.
In addition, modern IGUs often have low-emittance coatings to reduce
radiation losses through the window. The cavity between the panes is
then usually filled with a heavy gas (e.g. argon) to reduce convection and
conduction losses. This puts extra demands on the long-term stability of
the sealed edge of the IGU in order to retain the thermal performance
during the service life of the IGU. One question is whether the gas con-
centration within the cavity is the same as that claimed by manufactur-
ers, another is how quickly or slowly this concentration will decrease over
time (gas retention capacity).
There are two common techniques for filling the units with gas: the
lance filling (gas displacement) method and the vacuum chamber method.
In the first method, two holes are drilled at one side of the IGU. Gas is
filled into the lower hole, and air is exhausted through the upper hole.
When the gas concentration in the upper whole is high enough, the fill-
ing process is terminated, and the holes are sealed with e.g. rivets. In the
second method, the IGU is sealed within a vacuum chamber filled with
gas. The level of gas filling depends on the attainable vacuum level and
the time allowed for filling the chamber with gas (Elmahdy & Yusuf,
1995). Both methods may thus lead to an underfilling of gas within the
cavity. Elmahdy & Yusuf claim that filling levels of 95-98 % are attain-
able, but when testing a large number of samples, levels of 50 % or lower
can be found. In their study of 42 IGU’s produced in North-America
covering 7 types of spacers, the initial argon levels where found to vary
considerably (Elmahdy & Yusuf, 1995). While 52 % of the units had
initial concentrations of over 90 %, 10 % of the units where below 70 %.
All units had lower concentrations than the 95 % or higher claimed by
the manufacturers. After a series of accelerated ageing, high humidity
and volatile (fogging) tests, most of the units retained the argon gas. A
loss between 1 to 5 % was observed which would correspond to a maxi-

Energy-Efficient Window Systems
36
mum loss of 1 % per year, since the accelerated ageing test was assumed
to correspond to 5 years of normal use. While two units were found to be
tight, they had very low initial fill levels (30 and 50 %) which means that
the filling process was inadequate. A few others lost all of the argon dur-
ing the tests due to pin-holes in the sealant or to defective corner-keys.
Wolf (1988) reckons that dual-sealed insulating glass units can have a
service life of over 25 years. Today, when corner keys are abandoned, and
if mounted properly (ventilated and dry) representatives from the indus-
try believes in a service life of up to 60 years.
2.1.9 Fire protection and fire escape
For some spaces and premises there are certain requirements on fire-re-
sistance of building components. Windows can be divided into different
categories regarding fire-resistance. For windows with stipulated require-
ments on fire-protection special glass is needed, for example toughened
glass, laminated glass or wired glass.
There also exists special fire-protective glazing which are built on ei-
ther of two main principles: (1) by a phase-change induced by heat or (2)
by reduced transmittance. The first principle is used in glazing with a
water-based gel. When the glass pane closest to the fire becomes hot, it
separates from the gel and granulates. The gel thereafter stands for the
fire resistance until it dries out. The other principle is used in glazing
with one or more layers of water glass integrated between the panes. This
glazing function in the way that the water glass rises and forms an opaque
heat shield when the glass has reached a temperature of approx. 120°C.
The fire-protective requirements can also imply that a window must
function as a means for evacuation, i.e. fire escape.
2.1.10 Burglary protection
To increase the protection against burglary, insurance companies some-
times require the use of key locks on windows. Usually these require-
ments apply to personal property insurance above a certain sum and for
windows under a certain height from the ground (e.g. 4 m).
2.1.11 Insect protection
In Sweden there are no requirements for insect protection in windows,
but in other countries, for example in North America, it is very common
to provide windows with a net against mosquitoes and other insects.

Technology status of windows
37
2.1.12 Operation, window cleaning and child safety
These demands are more or less linked together. Windows that can be
opened must be operated in an easy way that does not require too much
force, so that all groups of people can handle them. Further, it should be
possible to clean all sides in an easy way. Windows that can be opened in
premises where children can stay must have some sort of locking device
to prevent them from opening the windows and falling out. These re-
quirements do not apply to windows on the ground floor.
2.1.13 Aesthetically appealing
Windows are often called the eyes of a building. The window plays a
major role in the appearance of a building, which the architect shapes.
He/she has a difficult task to coordinate demands on aesthetics, mainte-
nance, durability, economy etc., to pursue his idea with the building.
Many new materials are involved, and the window manufacture is highly
industrialized, see also sec. 2.2.3.
In the renovation of older buildings it is important to pay attention to
the original idea with the building.Windows are often replaced. The rea-
sons can be draughty windows, high maintenance costs, high energy costs
etc. The new windows are sometimes simplified with respect to number
of lights, colour, thickness and design of sash and frame, etc. compared
to the original ones. This may influence the way people perceive the build-
ing. Olsson-Jonsson (1988) has shown that a simplification of the win-
dow influences many aspects of the perception: for example the mean-
ingfulness and pleasantness of the façade is reduced. More lights than
originally will on the other hand increase the perception of articulation
and detailing, while the meaningfulness of the façade is decreased.
There are usually many aesthetical and other qualities (e.g. high wood
quality) in windows from approximately 1950 and earlier, which make
renovation both desirable and worthwhile, instead of putting in new win-
dows. Several methods to facilitate renovation have been developed, and
a new trade has been introduced – window craftsmen – combining the
skills of the carpenter, painter, glazier, plasterer and blacksmith (Pearson,
1994).
By using these new skills, and combining them with modern tech-
nology in a sensitive way, additional qualities can be introduced. Fred-
lund (1999) showed that a renovated double-pane window from 1880
could reach a total U-value of 1.60 W/m2K by replacing the clear inner

Energy-Efficient Window Systems
38
pane with an energy saving glass, i.e. a glass with a low-emittance coating
(Sec 3.5). The achieved U-value was even better than that of a triple-
glazed window from 1982, which had a U-value of 1.83 W/m2K!
2.1.14 Economical
First cost is often a considerable factor in the choice of windows. In Table
2.1 an example of the investment cost of a bedroom window is given.
The example shows that the investment cost is about 5 % of the total
investment cost per square m assuming a total cost of 10.000 SEK/m2 or
10 % of the direct building cost excl VAT and clients cost. Table 2.1 also
includes an example for the annual cost of the window. Despite some
uncertainties (e.g. that capital costs are strongly related to actual interest
rates, and that thermal losses through windows vary with U-values, ori-
entation etc.) the fact still remains that the annual cost for a window can
be around 50 SEK/m2,yr or approximately 5 % of the housing cost. (Bengt
Hansson, Lund University, personal communication, Aug 2001).
Table 2.1 An example of investment and annual cost for a window, 1.2
by 1.2 m, (inward opening 1+2 construction) in a 10 m2 bed-
room. Examples provided by Bengt Hansson, Lund University
(personal communication, Aug 2001).
Production cost:
Material (purchase, transport, insurance) 4 200 SEK
Labour (assembly, yarning, jointing and trimming) 400 SEK
Site cost 552 SEK
Contractors fee 412 SEK
Total production cost 5 564 SEK
Total specific production cost, (5564 SEK/m2 /10 m2=) 556 SEK/m2
Annual cost:
(Expected service life 50 years, average interest rate 6 %, maintenance
cost 450 SEK every 20 years).
Capital cost 353 SEK
Running cost 100-200 SEK
Maintenance cost 12 SEK
Total annual cost 465-565 SEK
Total specific annual cost 46-56 SEK/m2

Technology status of windows
39
For large facility managers, demands on durability and maintenance are
often deciding as to what type of window is chosen. Windows with very
low U-values are seldom chosen solely on the fact that heating bills will
be lower during the life-time of a building, since they who are responsible
for the purchase seldom are those who will pay for the running costs of
the building. The incentives are here diverging. If the synergistic effects
of energy-efficient windows are considered, (e.g. simplified heating sys-
tems, better thermal comfort), then this may be a strong enough reason
for purchase.
2.1.15 Sustainability
There are several definitions of the term sustainability or sustainable de-
velopment, see for example the following web site:
(www.sustainable.doe.gov/overview/definitions.shtml, 2001-08-22).
However, all definitions encompass ecological, economical and social
aspects. Perhaps the most well-accepted definition comes from the United
Nations Conference on Environment and Development (UNCED) in
Rio de Janeiro in 1992:
“Sustainable development meets the needs of the present
without compromising the ability of future generations to meet
their own needs.” – United Nations World Commission on
Environment and Development.
The above mentioned performance requirements considers mainly the
first use, or the first lifetime, of the window. Several of these require-
ments are also appropriate for the sustainability requirement. There are
however some other aspects to the term sustainability, which have not
been previously mentioned. These regard for example the use of energy
and resources during both production and maintenance. Environmental
impacts after first use should also be considered. Can the window be
reused and is it suited for material recycling? Both reuse and material
recycling may cause the need of disassembly. If the window is reused, is it
well suited for the new building (good thermal insulation, etc.)? Regard-
ing material recycling, one main question which arises is whether the
different materials can be separated without contaminating each other? A
discussion around the terms recycling and disassembly of building com-
ponents can be found in Thormark (2001).

Energy-Efficient Window Systems
40
2.2 Background to current window design
In Sweden, due to the harsh winter climate, single-pane windows were
abandoned early on. The first solution was that the traditional, outward
opening window was equipped with an inner sash, which was mounted
directly on the inside of the frame during winter, Fig. 2.4. Such windows
have been found in buildings already from the late 17th century. Towards
the end of the 19th century, such inner sashes became equipped with
hinges. These loose inner sashes existed until the 1920’s. In 1889, Flodquist
and Hallberg patented the coupled double-pane sash, which became very
common from 1910 and on. Windows were now often side-hung, in-
ward opening since this was more practical in the taller buildings of the
growing cities, see Antell & Lisinski (1988). The coupled double pane
window was thus the most common type in housing from 1910 and
forward. The heat loss was approximately halved with the introduction
of the double pane, since the U-value drops from about 6 (single pane) to
about 3 W/m2K. 4
Figure 2.4 Double-pane window from 1883 with a separate, loose, inner sash.
(From Antell & Lisinski, 1988).
In 1956 came the Suez-crisis, where the for oil transport so important
Suez-canal was closed during several months. Even if the feared shortage
of oil was exaggerated, it was an alarm clock. Between approximately
1956-58, the Swedish government issued advantageous loans for those
who installed triple-pane windows, and used extra insulation. During
this short period the majority of new single-family housing were built
with triple-pane windows. However, it was with three coupled sashes,
4. If the frame and sash is accounted for, the total U-value is somewhat lower. A
coupled window from 1930 was found to have a U-value of 2.6 before and 2.3 W/m2K
after renovation, while a window similar to that in Fig 2.4 had a U-value of 2.4 before
and 2.1 W/m2K after renovation (Fredlund, 1999).

Technology status of windows
41
which meant that 6 sides needed cleaning! When these subsidy loans
were removed, the coupled double-pane window again became “stand-
ard” (Bo Adamson, personal communication, April, 2001).
During the 1970’s, Sweden was heavily dependent on imported oil.
The oil crisis in 1973 was thus very noticeable to the Swedish society.
One solution to reduce the dependency of oil was to use the energy more
efficiently, another was to shift over to other energy sources. Sweden thus
implemented a quite rigorous energy-code in 1975, as a direct answer to
the oil crisis. This code (SBN 75) laid down rather strict requirements on
U-values for individual building components (walls, windows, roofs, floors
etc.), see Table 2.2, and established limitations of window size (max. 15 %
of the floor area). For windows, the 1975 code required a window U-
value of 2 W/m2K, thus three panes were needed to accomplish this. The
technology was known and tried in the late 1950’s, but was not competi-
tive enough to become prevailing on its own. Actually, insulating glass
units where already patented in 1865, in Ohio, USA (Wolf, 1988). In
Sweden, the furniture designer and architect Bruno Mathsson had ex-
perimented with double-glazed sealed and fixed units and floor heating
in houses that he designed already in the 1940s. Ventilation was provided
through openable vents above the windows. In 1950, he built an exposi-
tion hall for his furniture, where he further developed his ideas and also
after American inspiration built buildings with whole façades of glass,
with triple-glazed sealed units and electric floor heating. The window
pane he invented was called the “Bruno pane”, (Böhn-Jullander, 1992).
In 1975, sealed insulating glass units was a common technology also in
Sweden. This meant that window cleaning was much easier. However,
the traditional coupled window was kept, but modified in that way that
the inner sash was equipped with a double-glazed IGU. This was prob-
ably due to that the in Sweden much popular Venetian blind could be
kept between the glass panes. Another reason for keeping the coupled
construction is the lower replacement cost at potential failure. Especially
large facility managers use this as an argument. (If the inner IGU breaks,
the tenant has to pay, but if the outer pane brakes, the facility manager
pays).
Today, a U-value of 2 can easily be reached in a double pane window
with a low-emittance coating, but in 1975, the technology was not fully
developed. Triple-pane windows have thus been “standard” in Sweden
since the end of the 1970’s, and the manufacturing lines are adapted to
producing these windows. During the years there have been (1) triple-
pane windows in three coupled sashes (6 sides to clean!), (2) 1+2 construc-
tions with a double IGU in the inner sash and a single clear glazing in the
outer sash, and (3) a triple IGU in one single sash. While the first con-

Energy-Efficient Window Systems
42
struction has disappeared from the market, the two others exist side by
side, but are partly marketed for different target groups (e.g. larger facil-
ity managers versus private home-owners). (See also Sec. 2.3).
Today, there is also a trend to go back to double glazing, especially
within the commercial sector, since rather low U-values can be achieved
with modern solar control/low-emittance glazing. Glazing systems be-
come both lighter and cheaper if the third pane can be omitted, which is
especially desirable for the builders in the current trend of very large glazed
areas.
2.2.1 Changes to building code requirements
A building energy code usually lays down the bottom line for energy-
efficiency. Sometimes this stimulates the market to improve the energy-
efficiency of products, sometimes these improvements are market-driven.
In the case of windows, the building code of 1975 (SBN 75), which came
into force 1977, certainly acted as the driving force for permanently in-
troducing the triple-pane window. The required U-values where speci-
fied for each building component, with slightly higher requirements for
northern Sweden, Table 2.2.
As mentioned above, the 1973 oil crisis was the start of introducing
energy requirements into the building code as a means to improve the
energy-efficiency of buildings. Large efforts were also made to improve
existing buildings. Extra façade insulation, improved air-tightness of win-
dows and window replacement were typical examples. Simultaneously,
the mentioned shift from oil to other energy sources occurred. For resi-
dential housing in 1970, the main heating system was individual oil fur-
naces connected to a water-heating system. Such systems could be con-
verted quite easily by replacing the oil furnace with e.g. a combined elec-
tricity and oil or wood furnace. A few solar systems appeared, and heat
pumps using heat from outdoor air, drilled bore holes or the ground were
later introduced. District-heating systems were also growing rapidly in
size, connecting both new areas, and existing housing. However, perhaps
the most radical change to other energy sources was made possible through
nuclear power. At the beginning of the 1980’s, 12 Swedish reactors were
coming on line, one by one. Suddenly Sweden was faced with a large
electricity surplus. (The projections of growth rates in electricity demand,
which the nuclear build out was based on, were found to be strongly
exaggerated). Therefore, a stricter supplement to the Building code was

Technology status of windows
43
introduced in 1984, the so called ELAK 5code (Byggforskningsrådet,
1987). In residential housing, which complied with this code, direct elec-
tric heating was allowed. The code required wall U-values of 0.17 W/
m2K and heat recovery from exhaust air. Today, when the phase-out of
nuclear power has begun, this has introduced a new problem, since it is
very costly to convert these direct-electric heating systems to other en-
ergy-carriers. Energy-efficient windows (Uwin<1 W/m2K) can be one im-
portant answer, since radiators placed under windows can be omitted,
allowing for simpler heating solutions.
The building codes of 1980 and 1985 (SBN 80, SBN 85) brought no
changes to the section on energy compared to SBN 75. In 1988 the build-
ing code was reformed again with the introduction of NR 1 (BFS
1988:18). This had a new, systems approach. Where the old code had
laid down requirements on individual building components, the new
code focussed on the requirements or performance of the whole build-
ing, and on fulfilling certain functions, without specifying exactly how
this should be done. The new requirement was shaped as a maximum
permissible average U-value for the building, e.g. the sum of individual
U-values times their surface area divided by the total enveloping area,
Table 2.3. For windows this meant that the old requirement of Uwin<2
was omitted. Solar gains through windows were accounted for by reduc-
ing the dark U-value of windows with an orientation dependent value,
α
3, see Table 2.4. In theory, this new approach means that windows are
allowed to have a higher U-value than 2, if only the requirement on the
average U-value of the house is fulfilled.
NR 1 has been revised a couple of times, the current code is the BBR 99
(1998). The changes compared to NR 1 regarding energy is however very
marginal. A chapter has been added regarding the electricity efficiency of
buildings, but this goal has not been quantified.
5. ELAK: ELAnvändningsKommittén förslag för direktelvärmda småhus. (The en-
ergy use committee’s proposal for direct electric heating in one or two-family houses).

Energy-Efficient Window Systems
44
Table 2.2 Example of U-value requirements (W/m2K) for residential build-
ings according to the building code of 1975 (SBN 75). Tem-
perature zone I-II was south of a line through the cities of
Strömstad - Örebro – Gävle, Zone III-IV was north of this line.
Building part Temperature zone
I-II III-IV
Wall towards the ambient or through
earth towards the ambient 0.25 0.30
Roof or attic ceiling with roof above 0.17 0.20
Floor towards the ambient 0.17 0.20
Floor towards closed outdoor-ventilated
crawl space 0.30 0.30
Floor directly on ground 0.30 0.30
Window and door towards the ambient
Non-glazed part of door 1.00 1.00
Window and window in door
(including sash and frame) 2.00 2.00
Wall and floor towards space heated to
between +10°C and 0°C 0.50 0.50
Wall and floor towards space heated to
between +18°C and +10°C 1.00 1.0
Table 2.3 Example of U-value requirements for buildings according to the
building code of 1988 (NR 1), which is still valid (BBR 1999).
Uave,req for dwellings = 0.18 + 0.95 Aw /Aenv
Uave,req for non-residential premises = 0.24 + 0.95 Aw /Aenv
The maximum proportion of the area Aw which may be taken into
consideration is 0.18 Aheat .
Uave,req maximum permissible average total thermal transmittance (W/m2K).
Awaggregate area (m2) of windows, doors and similar, calculated over the
external frame dimensions.
Aenv aggregate area (m2) of the surfaces, in contact with the heated indoor
air, of enclosing elements of structure. The term enclosing element of
structure refers to elements which separate the heated parts of
dwellings or non-residential premises from the external air, the ground
or partly heated or unheated spaces.
Aheat heated usable floor area (m2) as defined in SS 02 10 53, (i.e. measured
from the inner side of exterior walls, and incl. of interior walls).

Technology status of windows
45
Table 2.4 Values of
α
3, the allowed subtraction on dark U-values of win-
dows according to window orientation in NR 1 and BBR 99.
Window orientation
α
3
SO - SV 1.2
SO - NV, SV - NV 0.7
NO - NV 0.4
If window orientation is unknown 0.7
2.2.2 Technology procurement of energy-efficient
windows
In 1992, the National Board for Industrial and Technical Development
(NUTEK) performed a technology procurement program to promote
super-insulated windows (NUTEK, 1992). The background was that
window development had more or less ceased since the development of
the triple-pane window during the late 1970’s. Occasionally, low-e coated
glass was used. To compare, modern walls usually have a U-value of 0.2
W/m2K, or 10 times better than windows per unit surface area, since
window U-values typically were around 2.0. The requirements in the
competition was a total window U-value of 0.9 W/m2K, with a bonus for
a U-value lower than 0.8. The requirements on both daylight transmit-
tance and total solar energy transmittance was 60 % (at normal inci-
dence).
Two windows were elected winners, both were quadruple-pane wood,
or mainly wood, windows with 2 or 3 low-e coatings and argon gas fill-
ings, see Fig 2.5. They had U-values of 0,88 and 0,73 respectively which
was below the required value. It was found harder to achieve the required
short-wave transmittance, both the daylight transmittance (55-54 %) and
the solar energy transmittance (51-44 %) was a bit to low. A few thou-
sand of these windows were installed in refurbishing and new projects,
which was also part of the technology procurement program; i.e. a guar-
anteed market for these windows.

Energy-Efficient Window Systems
46
Figure 2.5 One of the award winning windows with total U-value of 0,88
W/m2K. Vertical section showing the thermal break in frame and
sash with polyurethane imbedded in a PVC-profile. The quadruple
glazing consisted of: 4 mm – 94 mm Air – 4 mm LE – 16 mm Ar
– 4 mm – 16 mm Ar – LE 4 mm, where LE indicates the position
of the low-e coating and Ar stands for Argon gas.
However, the “market” did not welcome these windows with open arms.
Architects accused the windows of being clumsy and having to poor a
daylight transmittance. (These windows look very green placed before a
white wall). The green tint came partly from the fact that four 4 mm glass
panes were used, but mostly from the use of two to three low-e coatings.
In a study I performed, it was shown that people perceived a room with
such a window as darker and more enclosed than a room with an ordi-
nary triple-pane window (without coatings) and the daylight was per-
ceived as more tinted (article III), which thus supported the accusations
of poor daylight quality of these windows. These windows were also
heavier, leading to the need for special mounting tools on the building
site.
The benefits of the window was, apart from the lower heating bill, the
much improved thermal comfort during winter, see Wallentén (1993),
allowing for simplified heating systems (the radiator under the window
becomes unnecessary) or larger glazing surfaces, (NUTEK, 1993, 1995).
Also, the sound insulation was improved. However, when these synergistic
effects do not benefit the constructor, it is hard for him/her to motivate a
larger first cost.

Technology status of windows
47
NUTEK therefore continued to support the development of these
windows. The U-value requirement was raised from 0.9 to 1.0 W/m2K
to allow for triple-pane windows, see Fig 2.6. The name was simultane-
ously changed from super-insulated to energy-efficient windows. Two
low-e coatings were however still necessary to achieve the desired U-value,
since a rather traditional wooden frame and sash construction was used.
The daylight transmittance was thereby improved slightly. Argon or kryp-
ton gas fillings, and traditional spacer materials were mostly used. In one
case a stainless steel spacer with a slightly lower conductance was used.
Figure 2.6 Example of window from the second round of the technology pro-
curement program with total U-value of 1 W/m2K. The triple glaz-
ing consisted of: 4 mm LE – 70 mm Air – 3 mm – 15 mm Ar – LE
3 mm.
Condensation appearing on the outside of the outer pane was a new phe-
nomenon, which was observed on some of the first super-insulated win-
dows installed (see section 2.1.4). This was studied further, see e.g. Jonsson
(1995). This phenomenon is likely to appear on some occasions already
on windows that have a centre-of-glass U-value of approximately 1.0 W/
m2K, but the frequency will increase with decreasing Ucog. If taken as a
sign of a well-insulated window, this might be acceptable, but for a few
people it is perceived as very disturbing and unacceptable. Within the
Swedish glazing and window industry, it is still a hot topic for debate.

Energy-Efficient Window Systems
48
2.2.3 Aesthetical development
There is certainly an enormous difference between today’s windows and
those from one hundred years ago. A review of window architecture, typical
shapes, styles, wood dimensions etc. can be found in Antell & Lisinski
(1988) and Stockholms byggnadsnämnd (1988). Of course, fashion has
changed dramatically over the years (Fig. 2.7), but that is not the whole
reason why windows look the way they do today.
At the beginning of the 1900’s, each carpenter still put his own char-
acter to the windows he made, by the choice of profiles and fine detailing
of the sash and frame. Architects were also heavily involved into the de-
sign of windows, and its fine details, which can be seen on old architec-
ture drawings. The window and its trimmings was built and painted on
site, and was treated as a whole. Later, around 1940, when the window
manufacturing had became more industrialised, the design of windows
had to conform to a standard in the Swedish standardisation system. The
first window standard was issued in 1945, and was revised from time to
time. Naturally, this lead to a very high uniformity in window design,
with the same details on sash and frame.
Around 1970, window manufacturers no longer had to conform to a
standard, and suddenly (without really knowing it) it was up to them to
design the window. Architects or industrial designers were hardly involved
in the design process. Instead, demands on wind and snow penetration,
durability, and all the others steered the design process. The result is the –
with few exceptions – criticised clumsy, heavy windows we see today
(Hjorth, 1992).
After NUTEK in 1992 had conducted their technology procurement
for energy-efficient windows, I thought this would also lead to a develop-
ment on the aesthetical side. NUTEK granted some money so that archi-
tects could be involved in the manufacturers’ design process, and some
good examples of aesthetically appealing energy-efficient windows were
shown. Another example from approximately the same time was the ar-
chitect driven project “Good building components”, in which coupled
double-pane windows with one low-e coating were developed (Byggforsk-
ningsrådet, 1993). However, the impact seem to have been only tempo-
rary, and today almost 10 years later, the design of new windows is – with
few exceptions – more or less the same as during the 1980’s.

Technology status of windows
49
Figure 2.7 Changes of window styles from 1880 to 1980. (Modified from Björk
et al., 1984).
2.3 Windows of today
In Sweden, the triple-pane wood window is still dominating, either as a
1+2 construction or as a triple-pane IGU. Aluminium cladding on wood
windows is common among larger facility managers, while ordinary wood
windows are still the most commonly used type in single-family housing.

Energy-Efficient Window Systems
50
Aluminium windows and vinyl windows still have small market shares.
The total volume of window sales in Sweden year 2000 was approxi-
mately 1.2 million window lights, of which 66 % were wood windows,
30 % were wood-aluminium, 3 % aluminium and 1 % vinyl windows
(Leif G Gustafsson, SNIRI, personal communication, April, 2001).
Unfortunately, there are no easily available statistics for the share of
low-e coated windows. After having communicated (April-May, 2000)
with the 7 largest window manufactures, covering about 80 % of the
market, I estimate that around 70 % of the windows sold in Sweden
2001 were delivered with a low-e coating. The share of coated windows
shows a quite remarkable increase since 1995, see discussion of Elitfönster
below. The market share for such windows is probably highest within the
small-house industry (e.g. prefabricated houses). However, about 50–70 %
of current sales goes directly to consumers (via building material retail-
ers) for the renovation of older houses and here the picture is more di-
verse. Since these consumers are mostly interested in a low price, they are
less susceptible to low U-value arguments (Leif G Gustafsson, SNIRI,
personal communication, April, 2001). This was emphasised at SP win-
dows who can see a decline in low-e glass during the summer period,
while it is higher during autumn and spring. Among the companies I
spoke to, there is also a widespread “fear” of selling windows with very
low U-values to the consumer market due to the condensation phenom-
ena discussed in Sec. 2.1.4. While some companies deliver windows with
both one low-e coating and argon gas as standard, others have omitted
the argon gas to reduce the risk of condensation. The average U-value of
windows sold today is thus estimated to roughly between 1.3 and 1.6 W/
m²K.
Of the windows sold today, triple-glazed windows accounted for over
80 % of the volume, and double-glazed windows for the remaining 20 %.
For the triple-glazed windows, 75 % were sold with a triple IGU, and
25 % with a coupled 1+2 construction. A similar situation was found for
double-glazed windows: 76 % were with a IGU and 24 % were coupled
which means that coupled windows continue to decrease. Of the 7 inter-
viewed companies, only the smallest one have a large production of dou-
ble-pane IGU windows (70 % of their sales). These were then equipped
with one low-e coating, but usually not with argon.
This shows that it is still hard to convince Swedish consumers that a
double-pane window can have an equivalent or slightly better U-value
than a clear triple-glazed one. However, since the low-e coated share of
triple-glazed windows has increased over the last few years, the consumer
will for the most part also get a better product with the triple-glazed
window. In a national perspective it is also important to keep the triple-
glazed construction in order to continue the improvements on energy-

Technology status of windows
51
efficiency. The risk is that the market share for double-glazed windows
increases, something that some of the interviewed companies believe will
happen.
It is sometimes claimed that the market share for low-e coated glass in
Sweden is much lower than in other European countries, and should
thus be increased. This might have been true, but today the market share
seems to be quite high. The background with the early introduction of a
triple-pane window must also be remembered. The market for low-e coated
glass has also fluctuated according to the general economic state of soci-
ety: During the building recession around 1992-1998 the market share
for low-e coated windows probably reached a temporary low. One exam-
ple is given for one large Swedish window manufacturer, Elitfönster
(Anders Browall, Elitfönster, personal communication, 1994, 2001).
Elitfönster’s share of low-e coated glass was 23 % in 1990, which
dropped to 13 % in 1994. At the same time the largest part of the already
low share went on export to Germany, leaving the Swedish market almost
without coated glass. In 1990, the export to Germany was also much
smaller. Before the recession, the low-e coated glass mainly went to the
small-house industry. In 1994, this construction had almost stopped.
However, in 1998, Elitfönster changed their production when they in-
stalled a new production line for their insulated glass units. They have
since March 1998 sold triple-pane windows with one low-e coating and
one argon gas filling as their standard product. The insulated glass unit is
sealed in a vacuum chamber filled with argon, and the traditional filling
process through two holes in the spacer has thus been omitted. A clear
triple-pane window now costs extra! Their windows have a total window
U-value of 1.2-1.5 depending on window type and size.
This picture is supported by statistics given from the glass and glazing
manufacturing industry. Their production data from the last eleven years
show that the low-e coated share of IGU’s took a marked step upward in
1999 and was about 45 % in year 2000 (Lars Genberg, Pilkington, per-
sonal communication, June 2001), see Fig. 2.8. Elitfönster’s move to in-
troduce low-e coated glass as a standard product in 1998 is the main
reason to this large increase, since it has also pushed other companies to
follow.
While Swedish window manufacturers have long been conservative
regarding the use of traditional metal spacers, at least one glazing manu-
facturer has recently started offering one “warm-edge” alternative, what
they call a thermo-plastic spacer. However, none of the window compa-
nies that I spoke to have taken it in to their standard production, even if
several have it as an alternative. The conservatism is partly due to a fear of

Energy-Efficient Window Systems
52
introducing a new technology that would possibly jeopardise the life-
time of the IGU. Another reason is the cost. One manufacturer esti-
mated that the extra cost for the consumer is around 150 SEK/m2.
00
500
1 000
1 500
2 000
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
Year
Low-e coated
Total IGU
Production
(
thousand m
2
)
Figure 2.8 Statistics over insulated glass unit production in Sweden from 1990
until today.
2.3.1 Guarantees
Today some companies give a 10-year guarantee on windows. The P-
label is often the basis for the guarantee, see below. The window must
also be stored, mounted and maintained according to the specifications
of the manufacturer. The guarantee concerns wood rot, condensation
between panes in insulating glass units (IGU) and the function of win-
dows and their fittings. In other cases the normal construction guaran-
tee, which comprises two years, apply.
A guarantee of five years is normally given on rectangular IGUs. How-
ever, in the case of an extended guarantee (10 years) of windows, the
guarantee is also usually prolonged to 10 years for the IGU itself. The
guarantee concerns condensation between the panes.

Technology status of windows
53
2.3.2 Quality labelling, P-labelling of windows
The P-label is the name of a system for certification and quality control
of industrially manufactured products issued by the Swedish Testing and
Research Institute (SP). A P-labelled window must fulfil a number of
different quality criteria drawn up by SP (Brolin, 1987, 1990). The sys-
tem implies that the manufacturer performs internal quality control and
that it submits to continuous spot tests from SP to ensure that the pro-
duction always fulfils the quality criteria. P-labelling thus acts as a sort of
warrantor for good quality, and is sometimes used as a basis of issued
product guarantees, see above.
The general function criteria for a P-labelled window concern the fol-
lowing: air tightness, rain tightness, safety against wind load, thermal
insulation, condensation, mechanical strength and stability, manoeu-
vrability, fire safety and temperature stability (metal windows). There are
also certain requirements on the timber raw material for wood windows,
profiles of sash and frame regarding water drainage, surface cladding,
surface treatment, glueing, fittings, glass panes and glazing, putty, sealant
and strips, and finally handling instructions. As an extra, windows can be
classified regarding both fire safety and sound insulation, and they can be
tested regarding burglary protection. According to SNIRI (the Swedish
association of joinery shops), over 70 % of all wooden windows pro-
duced today are P-labelled (Leif G Gustafsson, SNIRI, personal commu-
nication, April, 2001).
2.3.3 Energy labelling
There are a number of different glass types, glass distances and gases,
frame and sash constructions etc. which make it difficult to estimate the
thermal performance and solar and daylight properties of windows. In an
attempt to establish common standards for the calculation and labelling
of windows, The National Fenestration Rating Council (NFRC) was
formed in the US in 1989. The standards regard calculation and labelling
of window U-values, total solar energy transmittance (or solar heat gain
coefficient, SHGC), shading coefficient, daylight transmittance, and con-
densation. The computer tool WINDOW 4.1 (LBL, 1992) was devel-
oped at LBNL to facilitate calculations of window performance. (The
newest version will be WINDOW 5, it exists now as a beta-version).
Today there are five different NFRC standards. One advantage of the
NFRC-labelling system is that it is cheaper than laboratory testing. The
label is shown in Fig. 2.9. Since the US is a federation of individual

Energy-Efficient Window Systems
54
states, it is up the each state to decide on following the NFRC standards
or not. As of April 2000, eight states require NFRC certification and
labelling of some window products. An additional 21 states or jurisdic-
tions have adopted the 1995 Model Energy Code or the 1998 Interna-
tional Energy Conservation Code, both which reference NFRC as the
preferred source for fenestration performance ratings (NFRC, 2000).
Figure 2.9 Energy labelling according to NFRC (left) and the Swedish symbol
for energy-efficient (mainly electrical) equipment (right).
In Europe, there is not really an equivalent of the NFRC rating system at
the moment, but energy labelling is currently a question of debate among
researchers, see e.g. Duer et al. (2000). There already exist a number of
different standards for the calculation of U-values, light and solar trans-
mittance etc., see for example (EN 673, ISO 10077-1:2000, ISO/DIS
10077-2 / prEN 10077-2, ISO/DIS 15099, ISO 9050:1990).
In practise, SP in Sweden in their contact with window manufacturers
works quite a lot like the American system, allowing manufacturers to
perform calculations of U-values instead of intensive laboratory testing.
Programs that can be used are mainly VISION/FRAME from Enermodal
Engineering, WINDOW from LBNL, USA, Canada and WIS from Eu-
rope, available via University College Dublin, Ireland.
NUTEK has developed a symbol (Eloff strömsnål) for energy-effi-
cient products, mainly electrical equipment, which can also be applied to
energy-efficient windows, see section 2.2.2. Windows to which this sym-
bol may be applied, must have a well thought-out aesthetical design and
fulfil the following technical requirements: U-value (whole window) <
1.0 W/m2K; glass ratio > 65 %; daylight transmittance > 63 % (normal
incidence); no change of the colour perception (from inside to outside);

Technology status of windows
55
no optical distortion (from inside to outside); the colour of the outside
glass surface must not be remarkably different to that of ordinary glass;
frame thickness < 140 mm; good sound insulation; guarantee of at least
10 years; service life of at least 30 years. Further, the window must fulfil
the requirements on rain and air-tightness, security against wind load,
mechanical strength and stability, manoeuvrability, and risk of conden-
sation according to Swedish standard. Child safety and fittings must be
according to the Swedish building code. It must be possible to clean all
glass surfaces within reach from the inside according to prevalent meth-
ods. Measurements should be done on windows of the size 1.0 × 1.2 m.
The aesthetical appeal is judged by a group of architects appointed by
NUTEK (Agneta Persson, personal communication, Oct 1995).
2.4 Windows of tomorrow
What will be the window of tomorrow? I can already say that I do not
want to elaborate on this highly hypothetical question. However, it is
difficult not to mention by a single line technologies that are under study
by scientists of today, for example vacuum glazing, aerogels, chromogenics
etc.
If the Kyoto-protocol about emission reductions shall be fulfilled, and
we plan to keep our current standard of living, the solution seems to lie
within new technologies, or at least with a proper use of the technologies
available today and through good building design. To reduce winter heat-
ing demands, further U-value reductions of windows might be solved by
new technologies like aerogels or vacuum glazing, see Duer (2001), since
the thin film technologies probably do not have much more to offer than
what already exist on today’s markets. However, there is a large potential
(and challenge) already within available technologies, to make the aver-
age technology equal to that of the best on the market.
Overheating and daylight (glare) problems in the summer require so-
lutions that can control throughput of solar energy and daylight, here
chromogenics has to compete against traditional shading devices and their
control systems. The solar energy control problem is the most straight-
forward to solve, since it deals only with thermodynamics. The daylight
issue is more complex, since here we need to deal with the dynamics of
people and to gain their acceptance or satisfaction of the system.

Energy-Efficient Window Systems
56

Basic window physics
57
3 Basic window physics
The heat flow through a window is a complex process. Room heat is
transported outwards through the window construction when it is warmer
indoors than outdoors, see Fig 3.1 showing the principles of long-wave
radiation, conduction and convection.
Figure 3.1 Principles of heat transport (during darkness) through a triple-pane
window showing radiation, conduction, and convection.
Short-wave solar radiation is transported inwards during daytime. The
major part is of course when there is direct solar radiation, but also the
diffuse parts (from sky and ground) give significant contributions. Part
of this radiation is ”visible” and provides lighting indoors. Therefore, it
is of interest to divide the electromagnetic spectrum into wavelength in-
tervals (Roos, 1994), see also Fig 3.2:
1) λ < 380 nm (UV-radiation). Non-visible ultraviolet radiation that
has little meaning for the energy balance of buildings. It can how-
ever be harmful for people, plants and textiles.
conduction
radiation
convection

Energy-Efficient Window Systems
58
2) 380 nm < λ < 780 nm (visible radiation). The most important
wavelength interval that contains approximately 50 % of the solar
radiation. Ordinary window glass has a high transmittance in this
interval.
3) 780 nm < λ < 2500 nm (near-infrared radiation). The part of solar
radiation reaching the surface of the earth which is not visible.
Approximately 40 % of the energy content from the sun are found
within this interval.
4) λ > 2500 nm (IR-radiation). All surfaces at room temperature emit
energy in this interval. Ordinary window glass is opaque for these
wavelengths. The radiation is however absorbed and then re-radi-
ated both inwards and outwards. A major part of the heat loss
through an ordinary window happens in this way.
Figure 3.2 Spectrum for a) a black body at four temperatures, b) extraterres-
trial solar radiation, c) typical absorption of the whole atmosphere,
d) relative sensitivity of the human eye (after Grankvist, 1981).
(1000 nm = 1
µ
m).

Basic window physics
59
3.1 UV-transmittance
The transmittance of UV-radiation, Tuv, through ordinary glass is rather
low. For example, it is usually said that it is not possible to get a suntan
behind a window, see also section 4.8.1. However, even if the transmittance
of ordinary glazing is very low, the energy content of these short wave-
lengths is still high, and may cause bleaching of textiles and paintings. If
this is of special concern, a special UV-filtering glass can be laminated to
the window, to further reduce bleaching. Some bleaching may still occur,
since sunlight up to the red part of the spectrum is known to cause bleach-
ing (IESNA, 1993).
3.2 Light transmittance
The transmittance of solar energy within the visible region, weighted
against the photopic sensitivity of the eye, is called the light transmit-
tance, Tvis, but the term LT also appears in the literature. For ordinary
clear float glass, approximately 90 % of the light that hits the surface at
normal incidence is transmitted. Approximately 8 % of the energy is re-
flected (R = 4 % at each surface), and the remaining 2-3 % is absorbed as
heat in the glass. The more window panes that are used, the lower is of
course the transmittance.
3.3 Solar energy transmittance
The transmission of solar radiation within wavelength intervals 1-3 is
called the solar energy transmittance. A distinction is made between the
directly transmitted energy, solar direct (or primary) transmittance Tsol,dir
(which is comparable to the light transmittance) and energy gain from
secondary heat transfer processes (absorption). The secondary part con-
sists of the fraction of absorbed energy in the glazing that is transported
inwards to the room, Ain, see Fig. 3.3. When the secondary part Ain is
added to the directly transmitted part, we call this the total solar energy
transmittance, Tsol,tot. Expressed as a ratio it is also denoted the g-value (g
for gain) or SHGC (solar heat gain coefficient). It is usually only slightly
lower than the corresponding light transmittance, but can for special so-
lar control glass be significantly lower, see sec 3.6. For single clear float
glass Tsol,dir is approximately 83 % and Tsol,tot is 86 %.

Energy-Efficient Window Systems
60
T
sol,dir
T
sol,to
t
A
in
A
out
I
A
1
A
2
A
3
R
+
Figure 3.3 Scheme over reflected, directly transmitted and absorbed solar ra-
diation in a triple-pane window. Multiple reflections not shown.
3.4 Multiple panes and angle-dependent
properties of glass
To be able to calculate the transmittance of a multiple-pane window, it is
necessary to take into account multiple reflection between the panes.
Polarisation of light must also be considered. The calculation procedure
for the transmittance for different wavelength regions is described in ISO
9050:1990.
A rough estimation of the transmittance is to multiply the transmittance
of the individual panes. An ordinary (clear glass) double-pane window
will thus have a light transmittance of about 80 % and a triple-pane win-
dow 72-73 %.
The transmittance is also dependent on the incidence angle: It is larg-
est for normal incidence, remains fairly constant to about 50-60°, and
then drops quickly to 0 % at 90° incidence angle, see Fig. 3.4. If the
physical properties of the glass are known (thickness, refraction and ex-
tinction coefficients), the angle-dependent properties can be calculated
using Fresnel’s equations and Snell’s Law.

Basic window physics
61
0
10
20
30
40
50
60
70
80
90
100
0 102030405060708090
Angle of incidence (°)
T
R
A
T, R, A (%)
Figure 3.4 The angle dependence of transmittance, reflectance and absorptance
of clear float glass for incidence angle
θ
(from the surface normal).
For special glazing, such as absorbing or coated glass, the angle-depend-
ent properties are usually slightly different than for clear glass. They can
be calculated if all material properties of the coatings (thickness, refrac-
tive indices) are known, but it is a more tedious work. Material properties
must also be known on a spectral basis, i.e. for each wavelength. Since
such data are not readily available, it becomes the work of a material
scientist. Some attempts have been made to fit polynomials to the angle-
dependent curves of the g-value, in order to facilitate this calculation
(Roos, 1997; Karlsson & Roos, 2000; Karlsson, Rubin & Roos, 2001).
These methods still require some knowledge about the materials used in
the coatings, but this could probably be built into an expert-system de-
pending on the relationships between g, Tvis, emittance, etc.
When the transmittance for a single glass is calculated or measured, it
is necessary to weigh the results for each wavelength against a “standard-
ized” solar spectrum. For the light transmittance, D65 is a widely used
spectrum. For the solar energy transmittance, references to two solar spectra
are given in the ISO 9050 (Perry Moon air mass 2 and CIE 85), and both
of these are widely used by manufacturers in Europe. In the US, a differ-
ent spectrum is used, ASTM E87-891, which corresponds to ISO 9845-
1:1992, see Fig. 3.5. Standardisation work is in progress with the aim to

Energy-Efficient Window Systems
62
change from the spectra refereed to within ISO 9050 to the spectrum
given in ISO 9845. The industry is however resisting this change, since
the values for g can be as much as 3-4 % higher when calculated with the
ISO 9845 spectrum. (Arne Roos, Uppsala University, personal communi-
cation, 2000). Care must therefore be taken when performance data on
products from different manufacturers are compared. Hopefully, this prob-
lem will eventually disappear when everyone conforms to the same calcu-
lation procedure.
0
200
400
600
800
1000
1200
1400
0 500 1000 1500 2000 2500 3000 3500 4000
Wavelength (nm)
Moon solar source
ISO9845 Direct normal
Solar spectral irradiance (W/m²,nm)
Figure 3.5 Solar spectral irradiance for two spectra: Perry Moon and ISO 9845
Direct normal irradiance, Tab 1, col. 2. The source for Perry Moon
was found in Optics5 from LBNL.
3.5 Glazing for energy-efficiency
Low-emittance (low-e) coatings are special types of spectrally selective
coatings designed to increase the energy-efficiency of windows. The main
characteristic of these coatings is that they have a high reflectivity in the
long-wave part of the spectrum, thus giving them a low emittance,
ε
(<20 %) in the same wavelength region. While normal glass absorbs most
of the heat radiation from the room surfaces (
ε
= 84 %) and re-radiates
outwards and inwards, low-e coated glass suppresses the radiation out-
wards, resulting in the long-wave radiation being reflected back to the

Basic window physics
63
room. The increase in heat insulation is equal to, or better, than adding
an extra pane of glass. Normal low-e glass is designed to have a transmit-
tance within the visible region as close as possible to that of ordinary
glass, Fig. 3.6. However, depending on the type of coating, they can give
a slight tint to the glass towards e.g. brown or green in transmission, and
a brown or pink tint in reflection.
Two main types of coatings appear on today’s market: soft and hard.
The soft coatings are sensitive to wear, window cleaning etc., and must
be protected in an insulated glazing unit (IGU). They are usually made
of a thin silver layer (Ag), on the order of 100Å, giving them an emittance
of about 10 %. Recent developments is the race between manufacturers
for extremely low emittances; 4 % or less is reached by a thick or double
silver coating.
The hard coatings are durable, which enables the use of these as single
panes or in coupled double-pane windows. The hard coatings are made
of a doped tin-oxide (SnO2), on the order of 4000Å thick. The emittance
is approx. 15-16 %. At the same time, the solar transmittance, Tsol,tot, is
higher than for the silver coatings, see Table 3.1.
Table 3.1 Approximate physical data of some commonly used glass types:
Visual transmittance (Tvis); Visual reflectance, exterior and in-
terior side (Rvis,1; Rvis,2); Direct solar transmittance (Tsol,dir);
Solar reflectance, exterior and interior side (Rsol,1; Rsol,2);
emittance, exterior and interior side (
ε
1;
ε
2). The coatings are
placed on the side they normally appear in a glazing combina-
tion.
Glass type Tvis Rvis,1 Rvis,2 Tsol,dir Rsol,1 Rsol,2
ε
1
ε
2
Clear float 0.90 0.08 0.08 0.83 0.07 0.07 0.84 0.84
Low-e (Ag) 0.85 0.05 0.04 0.62 0.16 0.20 0.84 0.10
Low-e (Ag) 0.85 0.08 0.06 0.58 0.22 0.28 0.84 0.04
Low-e (SnO2) a) 0.83 0.10 0.11 0.71 0.10 0.12 0.84 0.16
Absorbing green 0.75 0.07 0.07 0.46 0.05 0.05 0.84 0.84
Absorbing grey 0.44 0.05 0.05 0.45 0.05 0.05 0.84 0.84
Solar control 0.72 0.10 0.17 0.45 0.35 0.29 0.06 0.84
Adv. solar control 0.75 0.08 0.09 0.35 0.49 0.33 0.02 0.84
Adv. solar control 0.56 0.09 0.15 0.26 0.45 0.34 0.02 0.84
a) hard coating, can be used as single pane

Energy-Efficient Window Systems
64
0
10
20
30
40
50
60
70
80
90
100
0 500 1000 1500 2000 2500
Wavelength (nm)
Clear float Low-e coated Ag
Low-e coated SnO2 Solar control
UV Visible Near-IR
T (%)
0
10
20
30
40
50
60
70
80
90
100
0 500 1000 1500 2000 2500
Wavelength (nm)
UV Visible Near-IR
R (%)
Figure 3.6 Transmittance (top) and reflectance (bottom) for ordinary float glass
and some types of low-e coatings.

Basic window physics
65
From a physical point of view, it would be more accurate to classify low-e
coatings as thin or thick. Tin-oxide coatings would then belong to the
thick category, while all silver coatings are (more or less) thin. The spec-
tral selectivity of a tin-oxide coating stems from its material properties
(e.g. low electron density). Silver-based coatings, which have a high elec-
tron density, need to be thin in order to be transparent. Thus, by chang-
ing the thickness of the silver coatings, quite different transmittance spectra
can be achieved, see Fig. 3.7.
A low-e coating is usually applied on the outside of the inner pane, to
achieve a high g-value (pos. n-1). In super-insulated triple-pane windows
with two low-e coatings they are usually applied to the outside of the
inner pane (pos. 5), and to the inside of the outer pane (pos. 2). The mid-
pane is better left uncoated, since the rather high absorption of solar
radiation would otherwise cause excessive temperatures, which could lead
to a failure of the IGU.
0
10
20
30
40
50
60
70
80
90
0 500 1000 1500 2000 2500 3000
Wavelength (nm)
Ag 7
Ag 12
Ag 17
AgAg
T (%)
Figure 3.7 Transmittance for silver-based coatings of various thickness. Film
thickness in (nm). (Ag) for single silver layers, and (AgAg) for dou-
ble silver layers. Data supplied by Joakim Karlsson, Uppsala Uni-
versity.

Energy-Efficient Window Systems
66
3.6 Glazing for solar control
Traditionally, solar control glass was achieved by adding a metal oxide
(for example iron, cobalt or selenium oxide) to the glass melt to create a
body tinted (un-coated) absorbing glass, for example green or grey glass.
This glass was placed as the outer pane in a window combination, and
the absorbed heat would be mostly re-radiated and convected to the out-
side. Later came the coated glass with active parts of stainless steel (SS) or
titanium nitride (TiN), for example reflective coatings (high visual re-
flectance). These coatings are hard, and can be used in single layers.
Another type of solar control glass tries to combine solar control prop-
erties with energy efficiency. Such glazing is usually based on a soft silver
layer, giving it a low emittance, but which also requires that it is put in an
IGU. Such glazing is placed as the outermost pane, coating facing in-
ward.
The newest development in solar control glazing is coated glass which
have a very high ratio between Tvis and Tsol,tot, which is approaching the
physical limit of 2. This means that such glazing let in a large part of the
daylight, but cuts out most of the solar radiation in the near-infrared
region. These advanced coatings have multiple layers, where the active
part is a double silver layer (AgAg). They are soft (must be placed in an
IGU), and have a very low emittance (down to 2 %), making them a
combination of solar control and energy-efficient coatings. They are placed
as the outer pane (coating inwards) in an IGU in order to achieve a low g-
value.
3.7 Thermal insulation of windows
The thermal insulation of a window is usually measured by its U-value,
or thermal transmittance, which is the heat flux (in W) through the win-
dow per unit surface area (m2) at a temperature difference between inside
and outside of 1 degree (K or °C). Thus, the lower the U-value, the better
the insulation.
The overall window U-value, Uwin, is the energy loss from indoor air
to outdoor air divided by the total window area and the temperature
difference. It can be calculated as the area-weighted sum of the U-values
for the centre-of glazing, Ucog , and for the sash/frame, Uf . Thermal-bridge
effects around the edges of the glass (two-dimensional heat flow) are treated
in either of two separate ways:

Basic window physics
67
(1) The edge of glass is given a higher U-value, Ueog, than the centre (Eq.
3.1a). The edge effect is according to ASHRAE assumed to stretch 63.5
mm into the glass, but in some work 100 mm is used.
feogcog
ffeogeogcogcog
win AAA
UAUAUA
U++
++
=(W/m2K) (3.1a)
(2) A linear “U-value”, or thermal transmittance
Ψ
(W/m,K), account-
ing for the edge effects is multiplied by the perimeter of visible glass lg,
and added to the overall U-value (Eq 3.1b).
fg
gffcogg
win AA
LUAUA
U+
++
=
Ψ
(W/m2K) (3.1b)
where Acog, Aeog and Af are the projected areas of the different regions in
the window and Ag is the total glass area ( = Acog + Aeog ). While the first
method is mostly used in north America, the second method is mostly
used in Western Europe.
3.7.1 Glazing
The heat through the glazing is due to (1) long-wave radiation exchange
between the individual panes and the panes and their surroundings and
to (2) convection in the gaps of the glazing system and at the exterior and
interior surfaces. In a double-pane window radiation is dominating
(approx. 70 %). The thermal resistance of the glazing can be expressed as
the sum of the resistances of the different gaps, Rgap , and of the indi-
vidual glass panes, Rglass , plus the internal and external surface resistances,
Rsi and Rse :
∑∑ +++= sesiglassgaptot RRRRR (m2K/W) (3.2)
The U-value is the inverse of the thermal resistance Rtot :
Ucog= 1/Rtot (W/m2K) (3.3)
The surface resistances can be determined either by calculations, see for
example Arasteh et al. (1989) or by using standardised values taken from
a building code. In calculations, the convective/conductive part is often
separated from the radiative part, but in the codes, the two effects are
usually combined. In Sweden the Building code values are:

Energy-Efficient Window Systems
68
External surface resistance Rse = 0.04 m2K/W
Internal surface resistance Rsi = 0.13 m2K/W
Note that the inverse of the surface resistance is called the heat transfer
coefficient h. The building code values above gives he =25 and hi =8.
Heat transfer in gaps
The long-wave radiation exchange between two panes with respective
temperatures t1 and t2 (in Kelvin) is described by:
[]
4
2
4
12,1 ttq eff −=
σε
(W/m2) (3.4)
where
ε
eff is the effective emittance between the surfaces and
σ
is Stefan-
Boltzmann’s constant (5.67×10-8 W/m2K4).
The effective emittance
ε
eff is determined by the hemispherical
emittance
ε
of the two surfaces as follows:
1
11
1
21
−+
=
εε
ε
eff (-) (3.5)
Equation 3.4 above can also be expressed as:
q1,2 = hrb
ε
eff (t1 – t2) = hr (t1 – t2) (W/m2) (3.6)
where hrb is the radiative heat transfer coefficient between two black bod-
ies (
ε
=1), t1, t2 are the surface temperatures and hr is the radiative heat
transfer coefficient. hrb is given in Fig. 3.8 for a temperature difference
between the surfaces of 10°C.
The heat transfer in the gap due to convection is determined by:
q1,2 = hc (t1 – t2) (W/m2) (3.7)
where hc is the convective heat transfer coefficient. It accounts both for
conduction and convection in the gap (i.e. conduction when the air is
standing still, convection when the air is moving). It is defined as:
hc=k Nu / d (W/m2K) (3.8)

Basic window physics
69
where k is the thermal conductivity for the gas in the gap (W/m,K) (see
Table 3.2), Nu is the Nusselt number and d is the width of the gap (m).
The Nusselt number is a function of the height and width of the gap
(aspect ratio), of the Rayleigh number, and of the inclination of the win-
dow. It has been determined experimentally, and the following works are
usually used: (Hollands et al., 1976; ElSherbiny et al., 1982; Fergusen &
Wright, 1984).
3.0
4.0
5.0
6.0
7.0
8.0
-20-10 0 10203040
Mean temperature t
m
(°C)
Black body radiation transfer coefficient, h
rb
(W/m²K)
Figure 3.8 Black body radiation heat transfer coefficient hrb for varying mean
temperatures of two surfaces, and a temperature difference of 10°C.
Table 3.2 Thermal conductivity k, density
ρ
, and viscosity
µ
for some com-
monly used gases in windows.
Gas k
ρµ
W/m,K (×10-2) kg/m3kg/m,s (×10-5)
Air 2.41 1.29 1.73
Argon 1.62 1.70 2.11
Krypton 0.86 3.74 2.28
Xenon 0.52 5.89 2.26
CO21.46 1.98 1.39
SF61.30 6.70 1.45

Energy-Efficient Window Systems
70
Values of hc are shown in Fig 3.9. for vertical panes and for different gap
widths and different gases.
The thermal resistance of the gap can now be calculated as
ceffrbcr
gap hhhh
R+
=
+
=
ε
11 (m2K/W) (3.9)
The resistance of ordinary glass Rglass is approximately 0.001 m2K/W per
mm.
0.0
1.0
2.0
3.0
4.0
5.0
6.0
0 1020304050607080
Gap width, d (mm)
Air
Argon
Krypton
Convective heat transfer coefficient h
c
(W/m
2
K)
Figure 3.9 The convective heat transfer coefficient as a function of gas and gap
width. Calculated for a vertical window of 1.2 m height, a mean
temperature of 15°C, and a temperature difference of 10°C ac-
cording to ElSherbiny et al. (1982).
The total resistance of a glazing combination of for example a triple-pane
IGU (T4-12) can then be calculated as follows:
Rtot=Rse+Rpane1+Rgap1+Rpane2+Rgap2+Rpane3+Rsi (m2K/W) (3.10)
The resistance of each gap in this triple-pane window is approximately
equal to (suppose that tm1 = 5°C, and tm2 = 12°C, gap width = 12 mm,
ordinary glass gives
ε
eff = 0.72):
Rgap1 = 1/(4.88⋅0.72+2) = 0.181 m2K/W
Rgap2 = 1/(5.26⋅0.72+2) = 0.173 m2K/W

Basic window physics
71
The resistance of the whole glazing combination then becomes:
Rtot = 0.04+0.001⋅4⋅3+0.181+0.173+0.13 = 0.536 m2K/W.
This is equivalent to a Ucog of 1.87 W/m2K. If the inner pane is replaced
by a low-e coating with
ε
= 10 %, Ucog becomes 1.32, and if the air in the
same gap is replaced with argon, Ucog drops to 1.13. These values can be
compared to results obtained by a detailed calculation (including tem-
perature distribution calculation) as given in Table 3.3. The methodol-
ogy given above can thus be used to roughly estimate the U-value for any
glazing combination, the limitation is that the temperature distribution
of the individual panes is not known, which affects mainly the radiative
losses.
Thus, for the glazing, the heat loss may be decreased by adding more
panes, by applying low-emittance coatings to the glass to reduce radia-
tion losses, and by using heavier gases (such as argon, krypton or even
xenon) to reduce convection losses. The distance between the panes also
affects the U-value somewhat, see Fig. 3.10. This shows that the com-
monly used gap width 12 mm is not the optimum, at least not from an
energy point of view.
0.5
1.0
1.5
2.0
2.5
3.0
0 5 10 15 20 25 30
Distance between panes (mm)
Air
Argon
Krypton
U-value, U
cog
(W/m²K)
Figure 3.10 Ucog as a function of gap width for a double-pane unit with one
low-e coating. Typical distance in IGU’s of 12 mm is marked with
a cross and minimum values with a triangle. (Calculations per-
formed in WINDOW 4.1,
ε
=4 %, T=0/20°C, wind speed 5 m/s).

Energy-Efficient Window Systems
72
The U-value is often thought of as a constant, but it is in fact tempera-
ture dependent, which was already demonstrated in Fig. 3.8. Fig 3.11
show some examples of this temperature dependency for some glazing
combinations. It is evident that glazing with 15 mm gap distance, espe-
cially double glazing, is very sensitive to the temperature difference
∆
t
across the glazing, while some triple-glazing is almost unaffected. This
have some implications for the U-values reported by manufacturers, which
are often based on
∆
t=15°C according to the European standard EN
673. In Sweden a
∆
t of 20°C (0/20°C) has often been used, and when U-
values are measured in a guarded hot-box, -5/25°C have previously been
used as boundary conditions. In the NFRC documentation, the window
U-value shall be calculated at -18/21°C. Using a high temperature differ-
ence can be motivated for cold climates and for keeping measurement
errors small. Other boundary conditions which may vary between differ-
ent sources of information are the internal and external surface resistances,
which are sometimes fixed, sometimes calculated.
0.8
0.9
1.0
1.1
1.2
1.3
1.4
1.5
1.6
1.7
15202530
Temperature difference, in/out (°C)
D4-15 e4%
D4-12 e4%+Ar
D4-15 e4%+Ar
T4-12 e4%+Ar
T4-15 e4%+Ar
U-value, U
cog
(W/m²K)
Figure 3.11 Effects on Ucog due to the temperature difference between inside
and outside.
The requirements on good thermal insulation may come into conflict
with both the solar and the daylight admittance, since a better thermal
insulation is achieved through the use of more panes and/or low-e coat-
ings, Table 3.3. Thus, the better the thermal insulation, the worse the

Basic window physics
73
transmittance. Depending on how the improved thermal insulation is
achieved (for example the type of low-e coating used), the beneficial solar
gains may be reduced more or less, since a lower U-value is usually ac-
companied by a lower g-value, see Fig. 3.12. However, a lower U-value
usually more than outweighs the lower solar and daylight gains, see arti-
cle II.
Table 3.3 Ucog,, g and Tvis for some glazing combinations with clear glass
and low-e coatings (le) of various emittances. (Calculations per-
formed in WINDOW 4.1 for three sets of inner and outer tem-
peratures, wind speed 5 m/s). The low-emittance coating is in
all cases placed on the outside of the innermost pane (pos. n-1),
except for 4le-30-(D4-12). In that case the le-coating is placed
on the inside of the single outer pane (pos. 2). When two coat-
ings are used, the second one is placed on the inside of the out-
ermost pane (pos. 2). D stands for Double and T for Triple
insulating glass unit. (For example, D4-12 is a double pane unit
with 4 mm glass and 12 mm gap width). If any gas fillings are
used, they are marked with Ar for argon and Kr for krypton.
The gas is always in the same gap as the low-e coating.
Emittance Glazing Ucog Ucog Ucog gT
vis
of pane combination (-5/25°C) (0/20°C) (2.5/17.5°C)
ε
=84 % D4-12 2.84 2.79 2.76 0.76 0.82
D4-30 2.83 2.71 2.63 0.76 0.82
T4-12 1.87 1.85 1.84 0.68 0.74
4-30-(D4-12) 1.84 1.79 1.75 0.68 0.74
ε
=16 % D4-12 le 1.91 1.87 1.85 0.71 0.75
D4-30 le 1.97 1.81 1.70 0.72 0.75
4le-30-(D4-12) 1.41 1.32 1.26 0.61 0.69
ε
=10 % D4-12 le 1.79 1.75 1.73 0.65 0.77
D4-12 le+Ar 1.48 1.41 1.40 0.65 0.77
T4-12 le 1.33 1.32 1.31 0.58 0.70
T4-12 le+Ar 1.14 1.12 1.11 0.57 0.70
4-30-(D4-12 le+Ar) 1.11 1.08 1.07 0.58 0.70
T4-12 2le+2Ar 0.81 0.80 0.80 0.47 0.66
T4-12 2le+2Kr 0.68 0.62 0.62 0.47 0.66
ε
=4 % D4-12 le 1.64 1.60 1.58 0.59 0.77
D4-12 le+Ar 1.31 1.24 1.22 0.59 0.77
D4-15 le+Ar 1.32 1.18 1.11 0.60 0.77
D4-12 le+Kr 1.19 1.06 0.98 0.60 0.77
T4-12 le 1.24 1.23 1.22 0.53 0.70
T4-12 le+Ar 1.03 1.00 1.00 0.53 0.70
T4-12 2le+2Ar 0.70 0.69 0.69 0.42 0.65
T4-12 2le+2Kr 0.56 0.50 0.46 0.42 0.65

Energy-Efficient Window Systems
74
For daylighting the issue is more complex. Lower light levels may, in the
worst case, lead to an increased use of artificial lighting. This problem
can be resolved by using slightly larger windows. The filtering effect or
the colouring of the daylight that happens when one and in particular
several low-e coatings are used, can however not be compensated by larger
windows, and this conflict thus poses larger difficulties. The problem is
further described in article III.
y = 0.2081Ln(x) + 0.5463
R
2
= 0.9177
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.01.02.03.04.05.06.0
U
cog
-value (W/m
2
K)
g
-value (-)
Figure 3.12 The g-value (total solar energy transmittance) as a function of Ucog.
Values calculated for various glazing combinations in WINDOW
4.1 (0/20°C, 5 m/s).
3.7.2 Sash and frame
The heat losses for the sash and frame is mainly due to conduction. Thus,
the thermal properties of the frame material is important, as well as the
geometry. Therefore, wood frames are rather good, since wood has a low
conductivity, especially in comparison to aluminium frames. However,
as the glazing U-values are starting to drop from around 2 W/m2K to 1
or even lower, the glazing becomes better than the traditional wooden
sash/frame. It then becomes important to reduce heat losses also in the
sash and frame in order to achieve a low window U-value, Uwin. This can
be done by using more highly insulating materials in the sash and frame,
see Table 3.4. The design can also be changed, for example the IGU can
be embedded deeper into the sash to increase the path length that the

Basic window physics
75
heat has to travel, or the frame can be made deeper. Today there are sev-
eral programs to study two or three-dimensional heat flow, for example
HEAT2&3 (Blomberg, 1996), FRAMEPlus, THERM and others,
whereby it is possible to study the heat losses with different details and
materials in the design.
Table 3.4 Thermal conductivity of some materials for spacers, thermal
brakes, frame insulation materials and frames. From Thyholt et
al. (1994).
Thermal Main
conductivity usage as
Material (W/m,K)
Aluminium 220 spacer/frame
Steel, galvanized 48 spacer
Steel, stainless 14.3 spacer
Polyamide, reinforced 0.40-0.65 thermal break
Cast polyurethane, reinforced 0.20-0.30 thermal break
PVC extruded profiles 0.16 thermal break
Polycarbonate 0.20-0.23 frame insulation
Polystyrene 0.14-0.18 frame insulation
Wood 0.12-0.14 frame
Polyurethane foam 0.02-0.03 frame insulation
The traditional metal spacer (either aluminium or galvanized steel) in the
IGU has a high conductance and is a significant thermal bridge in the
unit itself. Nowadays, there are several spacers with a lower conductance
on the world market, Fig. 3.13. They are usually marketed as “warm-
edge” technologies. The improvement on Uwin is often rather marginal
(∆Uwin = 0.1–0.2 W/m2K), but the risk of condensation on the bottom
of the inside pane is greatly reduced, (Jonsson, 1985 and Frank, 1994),
see also Table 3.5. The usage in Sweden has thus far been very limited,
but an increased interest can be traced at the moment, especially for a
thermo plastic spacer marketed three years ago (Bally & Lenhardt, 1999).
The thermal effect of different frames and edge seal technologies have
been studied by several authors (e.g. Carpenter & McGowan, 1993;
Thyholt et al., 1994; Reilly, 1994). Depending on materials chosen in
the frame and spacer the frame U-value, Uf , can vary significantly. Thyholt
reports Uf -values between 5.4 and 1.8 W/m2K, depending on frame and
spacer material for operable two-pane windows (aluminium windows
without thermal brakes excluded).

Energy-Efficient Window Systems
76
Table 3.5 Linear thermal transmittance for different spacers and two in-
sulating glass units, from (Frank, 1984). Calculations performed
with a wood frame with Uf =1.6 W/m2K
Linear thermal transmittance
Ψ
(W/m,K)
Double IGU Double IGU low-e
Spacer Ucog=2.7 W/m2KUcog=1.2 W/m2K
Aluminium 0.046 0.057
Silicone foam 0.020 0.023
Silicone foam with integral 0.023 0.028
stainless steel spacer
Swiggle strip 0.029 0.035
Double aluminium, 0.035 0.047
thermally broken
U-shaped stainless steel 0.029 0.033
Fibreglass, hollow 0.030 0.039
Figure 3.13 Alternative spacer products or “warm-edge” technologies.
3.7.3 Total window U-values
Despite the technological development, not that much has happened re-
garding the design of sash and frame during the past decades in Sweden.
In the NUTEK procurement program (see Sec. 2.2.2), there was one so
called energy-efficient window (Uwin < 1 W/m2K) that used a stainless
steel spacer (a triple IGU window) but the coupled windows used ordi-
nary galvanized steel spacers. There were also some examples where the
thermal bridge of the wood frame was broken by another material, e.g.
polyurethane, but otherwise these solutions have been used restrictively,
since it has not been certain how they will behave in the long run regard-
ing for example moisture transport and service life. Wood frames are still
the most common construction, along with double or triple IGU units
with traditional metal spacers. The resulting U-values for different areas
of the window for three commonly sold window types are shown in Table
3.6. The table shows that it is harder to achieve a low Uwin in a triple IGU

Basic window physics
77
window than in a coupled window (1+2), because the thermal bridge
along the edge-of-glass is more pronounced. In fixed windows the frame
height is lower, which usually results in higher U-values for the frame.
These U-values can also be used to estimate Uwin for other window sizes
than that shown. This is demonstrated in Fig 3.14 which shows the esti-
mated Uwin for square windows of different sizes. The effect on the total
U-value is rather large: The total U-value increases by 0.23-0.30 W/m2K
when the window size is reduced from 1.5 m side length to 0.5 m side
length.
Table 3.6 U-values for some modern Swedish wooden windows of size
1.18 × 1.18 m as calculated in the FRAME program. Wood
frames and dual seal IGU with galvanized steel spacers. (Source:
Elitfönster AB).
Coupled Triple pivoted Triple fixed
window 1+2 window window
Projected area Area U-value Area U-value Area U-value
(m2) (W/m2K) (m2) (W/m2K) (m2) (W/m2K)
Frame, 0.281 1.38 0.321 1.44 0.178 1.74
top/side
Frame, 0.129 1.74 0.124 2.16 0.063 1.63
bottom
Edge-of-glass, 0.172 1.24 0.169 1.57 0.188 1.72
top/side
Edge-of-glass, 0.064 1.28 0.062 1.44 0.068 1.48
bottom
Centre-of-glass 0.747 1.00 0.716 1.00 0.895 1.00
Total 1.392
1.19
1.392
1.29
1.392
1.24

Energy-Efficient Window Systems
78
1.0
1.1
1.2
1.3
1.4
1.5
5 7 9 11 13 15
Side length (dm)
Fixed triple
Coupled (1+2)
U-value, U
win
, of square window (W/m
2
K)
Figure 3.14 Total window U-value as a function of window size. A square win-
dow is assumed.
In Germany there has been a development of super-insulated windows
for use in so called passive housing (Feist, 1995). There are probably a
dozen window types with a Uwin of around 0.7 W/m2K on the German
market today. These windows often have sash and frame almost purely
made of polyurethane or other highly insulating materials, Fig. 3.15. The
glazing is usually a triple-pane construction with two low-e coatings and
krypton gas fillings.
When the U-value of the sash/frame is to be improved, it is important
that this is done without introducing more clumsy designs, which are
not desirable from the architectural point of view. Expected service life is
another important issue. It is questionable if these issues are resolved in,
at least some of, the German super-windows available on today’s market.

Basic window physics
79
Figure 3.15 Two examples of German windows with total window U-values
around 0.7 W/m2K.

Energy-Efficient Window Systems
80

Windows and daylight
81
4 Windows and daylight
Windows are the eyes of a building. They provide lighting and view out,
but also the sometimes unwanted possibility of view in. This chapter will
discuss windows from several aspects, all regarding daylight. After intro-
ducing some general lighting terms and the source of daylight this chap-
ter continues by discussing calculation methods for the prediction of
daylight levels and the concept of daylight utilisation. Human aspects are
also discussed, both in terms of visual quality, perception and psychol-
ogy. Finally, the so called non-visual effects on humans are briefly re-
viewed.
The major factors affecting the daylight in a room are the size, shape
and position of windows and the room depth. Further factors are the
transmittance of the glazing and any external obstructions such as shad-
ing devices, opposing buildings or vegetation. Generally, tall windows
compared to wide windows of the same size admit the light further into
the room. Dividing the window area into several surfaces, preferably on
opposing walls, is often considered favourable since it gives a more even
and pleasant impression.
4.1 General lighting terms
In order to understand this chapter on daylight, it is important to start by
defining some commonly used lighting terms, here adapted from Illumi-
nating Engineering Society of North America IESNA (1993). For a
broader discussion on lighting terms, vision, etc., the reader is referred to
the general literature on this subject.
4.1.1 Illuminance, E
Illuminance describes the amount of luminous flux arriving at a surface,
i.e. the incident flux per unit area. It is measured in lux. It is a commonly
used unit, since it is easy to measure.

Energy-Efficient Window Systems
82
4.1.2 Luminance, L
Luminance describes the light reflected off a surface and is directly re-
lated to the perceived “brightness” of a surface in a given direction. It
depends not only on the illuminance on an object and its reflective prop-
erties, but also on its projected area on a plane perpendicular to the plane
of view. Thus luminance is what we see, not illuminance. However, the
perceived brightness of objects depend, apart from their luminance, also
on the state of adaptation of the eye. Luminance is measured in lumens
per square meter per steradian 6 or in candelas per square meter (cd/m2).
4.1.3 Daylight factor, DF
The daylight factor DF is defined as the ratio of indoor daylight illumi-
nance to the simultaneous exterior illuminance on a horizontal plane
from the whole of an unobstructed sky of assumed or known luminance
distribution. Since the daylight penetration is lowest during overcast
weather, the daylight factor is calculated for this condition. This also
implies that the daylight factor is largely unaffected by window orienta-
tion. Since the exterior illuminance varies constantly, the daylight factor
has been considered as a rather good measure of the available daylight in
a space. Although it is not possible to judge the quality of a space based
only on the daylight factor, some general guidelines can be given. A day-
light factor below 1 % is considered too low, a minimum of 2 % has
sometimes been used in building codes, between 2-5 % is considered
good, at 5 % daylight autonomy is assumed to be reached, and above
10 % glare problems are likely to occur. Under sunny conditions a simi-
lar approach can be used, which is then called the sunlight factor
(Christoffersen et al., 1999a).
4.1.4 Glare
In everyday language, glare is a word used to describe an unpleasant visual
experience. In the more stringent scientific context, glare is the unwanted
visual effects caused by large differences in luminance levels within the
field of view, or by strong light sources close to the direction of view.
Both skylight and sunlight can become glare sources when seen directly
6. The unit of measure of solid angles.

Windows and daylight
83
through a window, or reflected off surfaces. A distinction is usually made
between disability and discomfort glare (Hopkinson et al., 1966; IESNA,
1993).
Disability glare
A reduction in contrast of an image due to light being scattered in the
eye, causing impaired vision. A typical example is the reduction in vis-
ibility from oncoming car headlights. A long corridor lit only at the end
by a window is another example where vision can be obscured.
Discomfort glare
Discomfort glare is a sensation of annoyance or pain caused by high or
nonuniform distributions of brightness in the field of view. Discomfort
glare may be caused by viewing a light source directly (direct glare) or by
viewing a reflection of the light source from a specular / semispecular
surface (indirect glare). The expression ‘veiling reflections’ is used for the
indirect glare (IESNA, 1993).
While disability glare is very obvious to the individual, discomfort
glare often goes by unnoticed. Even when it causes headaches and eye-
strain, the source of these symptoms (i.e. a bad lighting situation) might
not be identified.
4.2 The sun as the source of daylight
4.2.1 Luminance and radiance models of the sky
The dominant weather type in Northern Europe is the overcast or par-
tially overcast days. Clear sunny days are rare, see measured solar radia-
tion data for Lund 1988. Figs. 4.1 and 4.2. Therefore, the diffuse sky is
the main source for lighting a room, not direct sunlight. Sunlight can
however provide pleasant aesthetical effects when present. In order to
perform accurate daylight calculations, the luminance distribution of the
sky must be known. Since measurements of sky distributions still are not
commonly available for more than a few sites worldwide, we are usually
obliged to use standardised sky distributions.

Energy-Efficient Window Systems
84
0
100
200
300
400
500
600
700
800
900
1000
1 3 5 7 9 11131517192123252729
Day
IN
IdH
Irradiance (Wh/m²,h)
Figure 4.1 Measured direct normal, IN, and diffuse horizontal, IdH, irradi-
ance (Wh/m2,h) in June for the reference year Lund 1988.
0
100
200
300
400
500
600
700
800
900
1000
135791113151719212325272931
Day
IN
IdH
Irradiance (Wh/m²,h)
Figure 4.2 Measured direct normal, IN, and diffuse horizontal, IdH, irradi-
ance (Wh/m2,h) in December for the reference year Lund 1988.

Windows and daylight
85
For overcast conditions, the two most used models are the CIE 7 overcast
sky adopted in 1955 and the much older uniform sky model. The uni-
form (isotropic) sky, defined already by Lambert in 1760, is equally bright
in every point, while the graded CIE overcast sky (also called the Moon
and Spencer formula) is three times brighter at the zenith than at the
horizon (CIE, 1994), see Fig. 4.3. There are also several other sky models
suggested by other researchers, for example by Muneer & Angus, Perez et
al., Coombes & Harrison, Perraudeau and Hooper & Brunger. With the
uniform sky model the resulting ratio between the vertical and horizontal
outdoor illuminance is 0.5, while it is 0.396 with the CIE standard over-
cast sky.
For clear blue skies a model proposed by Kittler has been adopted as
the standard, the CIE clear sky with or without sun. The luminance dis-
tribution, shown in Fig. 4.4, is characterised by a bright aura of light
around the sun, areas of horizon brightening, and a deep blue patch ap-
proximately 90° from the sun, moving around the sky with the move-
ments of the sun. For clear skies with a slight haze, the luminance distri-
bution of the sky is more even than for the perfectly blue sky. This is
taken care of in the formula by a turbidity factor, indicating how turbid
the atmosphere is.
Between the two extremes clear and fully overcast, there are in reality
an infinite variety of sky luminance distributions. In a report by Kittler et
al. (1988), a whole set of sky types have been classified and analysed for
both frequency of occurrence and absolute levels. The presented standard
luminance distributions where based on luminance scans of the skies in
Berkeley, Tokyo and Sydney. According to Kittler, the current CIE stand-
ards for the overcast as well as the clear skies are justified, but are to be
seen as extremes rather than averages. The uniform sky can occur on
unique occasions, especially in dense fog. Further, the uniform sky can
represent and ideal mean sky linking the decreasing gradation of overcast
skies with the increasing gradation of clear skies respectively. That the
uniform sky may be a better representation than the CIE sky for overcast
weather was recently concluded by Muneer (1998) in a study of Japanese
data. According to Muneer, the available daylight will be significantly
underestimated within building interiors with the CIE overcast sky model.
On the other hand, in a comparison of several formulae for overcast con-
ditions performed by Enarun & Littlefair (1995) it was shown that for
7. CIE is the abbreviation of Commision Internationale de l’Eclairage, an interna-
tional organisation for lighting issues.

Energy-Efficient Window Systems
86
conditions where the sun disc is totally invisible, the CIE overcast sky
still performs the best, although it may depend on how the overcast days
are selected.
Figure 4.3 The CIE overcast sky is three times brighter at the zenith than at
the horizon (from Kittler et al., 1988).
Figure 4.4 The CIE clear sky has a clear solar corona and a dark blue patch
about 90° from the sun (from Kittler et al., 1988).
As well as there are sky models available for daylight purposes, there are
also models of the radiance distribution of skies for solar engineering
purposes. The main difference is of course that the daylight models use
the photometric term luminance, which means that it is defined for visual
light weighted by the sensitivity of the human eye, while the other mod-
els uses radiance which is defined for the whole solar spectrum. Gener-
ally, solar sky models seem to be less detailed than daylight sky models.
Duffie & Beckman (1991) present only three clear sky models, one iso-
tropic model and two anisotropic ones: the Hay & Davies model and the
HDKR model (the Hay, Davies, Klucher, Reindl model). However, they

Windows and daylight
87
do mention the existence of other models by for example Skartveit &
Olseth and by Perez. The isotropic model derived by Liu & Jordan in-
cludes three components: beam, isotropic diffuse, and solar radiation
diffusely reflected from the ground.
The Hay & Davies model estimates the fraction of the diffuse radia-
tion that is circumsolar and simply adds this fraction to the beam radia-
tion. Thus they consider that all of the diffuse radiation can be repre-
sented by the two parts isotropic and circumsolar. The ratio between the
two parts is determined by the use of an anisotropy index which is the
ratio of the measured beam radiation to the extraterrestrial solar radia-
tion. The anisotropy index is thus a measure of the transmittance of the
atmosphere. When the sky is very clear the anisotropy index will be high,
and most of the diffuse radiation will be assumed to be forward scattered.
When there is no beam, the Hay & Davies model becomes equal to the
uniform sky. The HDKR model is an extension of the Hay & Davies
model that adds a term accounting for horizon brightening.
4.2.1 Luminous efficacy
Luminous efficacy, K, is the ratio of light output (in lumens) to energy
use (W). It is used to characterise the efficacy of light sources such as
incandescent or fluorescent lighting, and also for skylight and sunlight.
The luminous efficacy of daylight is thus expressed as the ratio of illumi-
nance (lux) to irradiance (W/m2). When only solar radiation data is avail-
able for a site – which is quite common – the luminous efficacy can be
used to translate the solar data to illuminance levels.
Values for luminous efficacies have been estimated by several authors
by correlating simultaneous measurements of solar radiation and illu-
minance, see for example Littlefair (1985) for a good review. Since the
luminous efficacy can depend on solar altitude, cloud cover, and amount
of aerosol and water vapour content in the atmosphere, different values
apply for different sky conditions. Some values for clear and overcast
skies are given below. Values for intermediate (partly cloudy) and average
skies have also been suggested, see for example Littlefair (1985, 1988),
but they have been omitted here. Generally, the values fall between the
two extremes clear and overcast.

Energy-Efficient Window Systems
88
Clear skies
Direct (beam) luminous efficacy
In several studies, measured values of direct luminous efficacies are corre-
lated with solar altitude. Some authors have also found correlations with
aerosol and water vapour content. Littlefair (1985) cite values between
70 to 105 lm/W for increasing solar altitudes. In later measurements (in
Garston, Hertfordshire, UK) he found values from 70 lm/W at 10° to 95
lm/W at 60° solar altitude (Littlefair, 1988). However, average values
often perform on par with more advanced models, as demonstrated by
Muneer & Angus (1995), who give an average value of 104 lm/W for
Edinburgh, UK. A similar value was observed in data from Vaerlose,
Denmark, 103 lm/W (Petersen, 1982, cited in Christoffersen, 1999a).
Diffuse sky (cloudless) luminous efficacy
Clear sky radiation (i.e. from the blue sky vault) has a higher luminous
efficacy than direct solar radiation. A typical value between 120-140 was
cited by Littlefair (1985), while both Muneer & Angus and Littlefair
(1988) measured an average of 144 lm/W, similar to Petersen’s measured
average of 146 lm/W.
Global luminous efficacy
For the global luminous efficacy of clear sky and sun, Littlefair (1985)
gave typical values in the range 95-115 lm/W. Later, Littlefair (1988)
measured on average 107 lm/W, while Muneer & Angus measured 110
lm/W and Petersen obtained a value of 113 lm/W.
Overcast skies
For the overcast sky condition, Littlefair gave a typical range of 105-120
lm/W, fairly independent of solar altitude. Muneer & Angus’ measured
average was 115 lm/W, equal to that of Littlefair (1988), while Petersen
found a value of 121 lm/W.
4.3 Daylight calculation methods
To accurately estimate the daylight distribution within a space is a com-
plex task. The source of daylight, the sky and the sun, is constantly vary-
ing, from minute to minute, and from season to season. In our climate, it
is mostly the cloudy sky that is used for design purposes, but in other

Windows and daylight
89
climates e.g. in sunny climates, the sun can also be used as an illumina-
tion source. As previously mentioned, the luminance distribution of the
cloudy sky varies with cloud cover. However, due to lack of better data, a
standard overcast sky e.g. CIE overcast sky, is usually used in the design
of buildings.
Daylight distribution can be estimated in scale models, but there are
also several lighting programs along with hand calculation methods that
can be used. An overview of these methods is given in Christoffersen et
al. (1999a). With advanced lighting and thermal programs it is also pos-
sible to estimate the performance of different daylighting systems and
control strategies and to evaluate the impact on the overall building en-
ergy use (e.g. illumination, heating and cooling).
4.3.1 Hand calculation methods
At the Building Research Station in England hand calculation methods
for the determination of the daylight factor in both side-lit and top-lit
rooms were developed over 50 years ago (Hopkinson et al., 1966). The
method is referred to as the BRS Daylight Protractors, since they devel-
oped special transparent templates (protractors), which facilitated the
determination of the DF directly on drawings. The method is also de-
scribed by Löfberg (1987).
The method is based on the observation that the daylight in a point
can be determined as the sum of three components: light from the visible
part of the sky (the sky component), light reflected directly onto the
point from the surroundings outside the window (outdoor reflected com-
ponent) plus light reflected from the room surfaces (indoor reflected com-
ponent). This method is also called the split-flux method.
4.3.2 Radiosity methods
In radiosity methods, the light distribution is determined by calculating
the light flow between small surfaces areas, patches, that see each other.
The distribution over a (fictious) surface is then smoothed out between
the different patches. View or form factor calculations are therefore syn-
onymous to radiosity methods. Usually, all surfaces are assumed to have
a perfectly diffuse reflectance (Lambertian surfaces). It is computationally
much faster than ray-tracing. However, if the patches are shrunk to in-
finitesimal points, ray-tracing is actually performed (Ashdown, 1996).
Superlite is an example of a program using the radiosity method.

Energy-Efficient Window Systems
90
4.3.3 Ray-tracing methods
Ray-tracing is a method to determine the light distribution in which rays
of light are followed in space, through multiple reflections until they are
distinguished. Rays can be followed forwards from the light source to the
surfaces or backwards from the viewer to the light source. In forward ray-
tracing a large part of the sent out rays are absorbed along the way, and
therefore never reaches the eye or camera. This is eliminated by following
the rays backwards, from the viewer, into the environment until they
reach a light source. Backward ray-tracing is thus more efficient than
forward ray-tracing. The drawback is that only values for the specified
view are calculated. If a new view is chosen, the calculation must be re-
done. Ray-tracing gives the optimal opportunities for calculating light
distribution accurately, both regarding illuminance and luminance, col-
our effects, specular surfaces, glare, photorealistic images (renderings) etc.
Examples of ray-tracing programs are Radiance, Genelux, Lightscape and
others.
4.4 Daylighting software
4.4.1 Pure daylighting programs
There are several lighting calculation programs on the market today. From
very simple, intuitive programs like LESO-Dial (Paule et al., 2000) –
aimed at architects for very early design decisions – to very advanced
rendering programs like Radiance (Ward Larson & Shakespeare, 1998),
which is considered as the most accurate architectural rendering program
available. Radiance was created at the Lawrence Berkeley National Labo-
ratory, and can be downloaded free over the Internet. Apart from creating
very realistic ‘photographic’ renderings, also illuminance levels, luminance
distributions, various glare indices etc. can be calculated. Radiance itself
is just a powerful calculation engine that requires users with very ad-
vanced skills. In order to facilitate the rendering process several “menu
shells” have been developed for Radiance. One such example is Adeline,
a platform for the two programs Superlite and Radiance (Erhorn & Stoffel,
1996). This shell also incorporates pre-processors to import the geom-
etry from a CAD program, and post-processors to calculate hourly light-
ing electricity use for various control systems, Superlink and Radlink.
These data can then be used as input in a thermal simulation program
like TRNSYS, DOE2, SUNCODE, TSBI3 or even Derob-LTH.

Windows and daylight
91
Other menu shells for Radiance are for example Desktop Radiance,
which integrates Radiance with the CAD program AutoCAD, and
Rayfront, which can be run as an AutoCAD or Intellicad extension, or as
a standalone program. Lightscape and Genelux are other examples of
advanced commercial programs, both using the radiosity approach with
major ray-tracing extensions.
In a validation study the following programs were compared with each
other and with scale model measurements: Superlite, Genelux, Radiance
and Leso-Dial (Fontoynont et al., 1999). The comparison showed that
direct illuminance inside buildings can be calculated with an accuracy of
about 5 %, and that total illuminance (including multiple reflections)
can be calculated with and error of less than 10 %. The programs were
thus reasonably consistent with each other, despite the different calcula-
tion methods employed (both radiosity, ray-tracing and simple techniques
were used). However, it was shown that calculation results are very sensi-
tive to the input data, e.g. the light source description, material photom-
etry, building geometry and other simulation parameters. Especially the
exact luminance distribution of the outdoor environment was pointed
out to be very important, in particular of the sky close to the horizon and
the luminance of the ground. It was striking to note that the radiosity
programs performed on par with the more advanced ray-tracing programs,
but then the geometries and the diffuse surface reflectances were also well
suited for radiosity programs.
4.4.2 Thermal programs with daylighting routines
Since about half of the solar energy is within the visible range (daylight),
the use of daylight and solar energy are strongly interconnected. Daylight
can be used to replace artificial lighting, see section 4.5, and visual com-
fort criteria may often decide when shading devices need to be used. The
trend has thus been to include daylight routines into energy simulation
programs. These incorporated daylight modules vary from being very
simple to quite advanced.
One program that early on implemented daylighting calculation rou-
tines into the thermal simulation program was the American program
DOE-2. It used a separate lighting module for calculating daylight fac-
tors for standard sky conditions (using the split-flux method), discom-
fort glare indices, and use of electric lighting for a few control strategies.
Links were provided to the thermal simulation program for adjusting
shading devices and internal loads on an hourly basis (Winkelmann &
Selkowitz, 1985).

Energy-Efficient Window Systems
92
Today, the two separate programs DOE-2 and BLAST are being inte-
grated into one modern calculation engine: EnergyPlus (Crawley et. al.,
2001). The existing daylighting module in DOE-2 is being ported to this
new version. The module can take into consideration interior illumi-
nance from windows and skylights, dimming of electric lighting, effects
of dimming on heating and cooling, and glare simulation and control.
Another program which incorporates a daylight module to estimate
the thermal effects of daylighting is ENERGY-10 (SBIC, 1996). In the
newly released Danish energy simulation program BSIM2000, there is
also a daylight routine for estimating the daylight factor (Grau & Wittchen,
1999). This is however limited to estimation of daylight factors for the
uniform sky model.
A different approach has been to start with a free standing light simu-
lation program, and then produce an output file that can be passed to
practically any energy simulation program. The lighting simulation plat-
form Adeline (Erhorn & Stoffel, 1996) have such links to thermal pro-
grams. For example, the post-processor Superlink can calculate the an-
nual hourly lighting electricity use for different control strategies, and
this file can be used as an internal load file in the thermal calculation.
However, this method has the drawback of lacking interactivity with the
thermal calculation since the light calculations are performed first. This
means that the daylight program cannot be used to control shading de-
vices. Another problem is that the power demand for continuous dim-
ming systems are underestimated when dimmed, since the program does
not take into consideration the high base load of a fully dimmed system,
see Christoffersen (1995).
Others have tried to overcome the problem of lacking interactivity by
building direct links between thermal programs and the advanced light-
ing program Radiance, e.g. Clarke et al. (1997), Janak (1997) used Radi-
ance together with ESP-r. A similar approach has been used between Ra-
diance and TRNSYS (Kovach-Hebling et al., 1997). The two largest draw-
backs are: (1) the still fairly long calculation times required in Radiance
when high accuracy is wanted and when all daylight hours of the year
have to be calculated; (2) that Radiance is such an advanced program
that it takes an expert to perform these calculations. Solutions to the first
problem have been suggested, for example the use of a set of pre-calcu-
lated daylight coefficients, covering all hours of a year (Tregenza & Wa-
ters, 1983; Reinhart & Herkel, 2000).

Windows and daylight
93
4.4.3 Derob-LTH daylight module
A daylight calculation module for the thermal program Derob-LTH was
developed, see article V. It currently works as a stand-alone post-proces-
sor to Derob-LTH but the intention for the future is to integrate it more
closely with the thermal program. The aim of the module is to provide
input for daylight-responsive control of shading devices. The work on
the daylight module coincides with a modernisation of the whole struc-
ture of the Derob-LTH calculation engine, a work that is still in progress.
The main motivation for this modernisation is that the current version
of the Derob-LTH program does not allow for a changing geometry dur-
ing the simulation, and thus not for control of shading devices. The com-
puter code has over the years also become badly arranged and hard to
work with. The new calculation engine will in many ways be improved.
The current limitation of 27 surfaces enclosing a volume will for example
be removed, and time-steps can be varied.
The proposed daylight estimation method is based on the fact that
Derob-LTH already calculates the distribution of solar radiation (direct
and diffuse components) in a space using a radiosity method. The solar
radiation distribution is then translated to visual radiation via the lumi-
nous efficacy. Last, the radiation levels are amplified using the ratio of
visual-to-solar transmittance to yield the final illuminance level.
Validation was performed for overcast and clear days against mainly
Radiance for a side-lit room and for a simple atrium. For the overcast sky,
the accuracy is acceptable for both vertical and horizontal windows for
the midpoint of the room. For the sunny sky, Derob-LTH accurately
predicts the size and illuminance level of the sunpatch, at least for the
tested vertical window. For the purpose of using it for daylight-linked
control of shading devices, the accuracy of the developed model seems
sufficient. However, if it is to be used as a proper daylight tool further
work is needed mainly on eliminating the diffuse glazing transmittance
of the diffuse radiation component, and to increase the number of nodes
used for the diffuse radiation distribution. If this is done, the luminance
of the surfaces could be calculated easily, which would also make calcula-
tions of for example glare indices possible.

Energy-Efficient Window Systems
94
4.5 Daylight utilisation
That daylight can be used to replace artificial lighting have since long
been identified as an interesting electricity saving technique, and many
articles have been produced on the subject, see for example (Verderber &
Rubinstein; 1984, Szerman, 1994; Christoffersen, 1995). What makes
daylight utilisation (or a daylight responsive/-linked lighting system) so
interesting compared to other electricity saving alternatives is that day-
light is most abundant during spring, summer and autumn, and of course
during normal office hours, i.e. the same hours when overheating in of-
fices is likely to occur. It therefore has the potential of also reducing cool-
ing demands without simultaneously increasing the heating demands,
see for example article II.
The potential of the electric lighting savings is of course strongly re-
lated to many parameters such as window size, glazing type and orienta-
tion, room geometry, sensor position, control system, installed lighting
power etc. Therefore, estimated lighting savings found in the literature
vary greatly. In comparisons between simulated and measured savings it
has also been found that real savings may be lower than predicted ones,
see for example Andresen et al. (1995). They showed that measured light-
ing energy savings were significantly lower than estimations performed
through lighting calculations in Superlite/ Superlink (measured savings
were 16-32 % compared to 34-47 % calculated for continuous dimming
systems). Apart from factors mentioned in the report (e.g. use of shading
devices not accounted for in the simulation, climatic data differed), the
discrepancy might also be due to an error in the Superlink routine, see
Christoffersen (1995), which leads to lighting loads which are much too
low. Simulations performed for the preparation of article II will also dem-
onstrate this, see section 5.2.1.
There it is shown that lighting energy saving can be expected to be
around 55 % for offices with rather large windows (when 50 % of the
façade wall is glazed) and around 40 % when 30 % of the façade is glazed.
It was however concluded that this might be an optimistic estimation,
since it was compared with a case were the lights were on all day. In a real
building, people are not always in their offices, and some people do not
always turn on their own lights.
If lighting energy savings and the synergistic effects on cooling loads
are the major economical benefits of daylight utilisation, what other ad-
vantages or potential drawbacks are there? First, daylight utilisation im-
plies the presence of windows in the immediate surrounding of the
workplace. This has many both psychological and physiological advan-

Windows and daylight
95
tages as will be described below. The quality of natural illumination may
also be highly desirable. Daylight entering through windows located on
the sidewall of a building provide a directional component to the general
illumination which can contribute to the modelling of objects (Collins,
1976).
The drawbacks are to my opinion few, but they may include how
people perceive the electric lights at a dimmed state, and the effects on
room perception in general. There are for example studies that show that
if the sky is bright, people will want to increase the indoor lighting in
order to reduce contrast (Inui & Miyata, 1973). Therefore, any control
algorithm that is developed, must not only aim at reaching a desired
illuminance level, but must take peoples reaction and satisfaction of the
system into consideration. People also have an inbuilt understanding of
the main direction of light as coming from above. This was illustrated
very clearly by Hesselgren (1987) in a famous photograph of metal bumps
on a riveted cylindrical tank. When viewed the right way, the highlights
and shadows gave the impression that the bumps were dents but when
the picture was viewed upside down, the dents were turned into bosses,
and the rivets looked like dents. According to Hesselgren the most pref-
erable light direction is about 30° to the horizontal plane. Without a
directional component the shadows on objects in the room will be unfa-
vourable, and with a completely uniform lighting there will be no lustre,
which make objects appear dead. Therefore, if the daylight is evened out
by a sophisticated lighting installation that meet every engineers dream
of a perfectly even illuminance, this might go against our intuitive un-
derstanding of light, and thus be perceived as unpleasant, unsatisfactory
or awkward.
Finally, it is evident that daylight utilisation requires rooms that are
not too deep. An old rule of thumb has been that the room depth should
not exceed 2-2.5 the distance from the floor to the top of the window,
which yields rooms of about 4 to 6 m deep. Many modern buildings have
floor plans much deeper than that, 18-20 m is not uncommon. No mat-
ter how much glazing is put into the façade, this will leave a core of the
building that will have to be artificially lit. Even if there are new ways of
bringing daylight into deep buildings, e.g. light pipes, the contact with
the outside will still be lost.

Energy-Efficient Window Systems
96
4.6 Lighting quality and visual comfort
Naturally, visual comfort and a good quality of the lit environment are
desirable in any setting. However, visual comfort has become a word of
fashion, and is often used without too much reflection upon how it should
be achieved. Often it is used as an analogy to thermal comfort. When the
research on visual comfort is critically analysed, it is clear that we still
know very little about what criteria should be fulfilled in order to achieve
a good visual comfort (i.e preferred luminance, luminance distribution,
uniformity, flicker rate, spectral power distribution).
Christoffersen et al. (1999a) criticise the wording visual comfort, since
it can lead to the belief that there is a form of neutral comfort situation,
just like in the thermal situation, where neither more or less lighting is
desired. They claim that the basic difference between the two concepts
lies in the fact that lighting, and especially daylighting, is much more
complex. What they also could have said is that vision is the most sophis-
ticated of our senses. Our perception is based on the reception and inter-
pretation of a number of dynamic visual and sensory impressions from
everything that lays in the field of view. Contrary to the thermal comfort
situation – where each change in temperature will lead to reduced com-
fort – the lighting situation can always be improved according to
Christoffersen.
Hopkinson et al. (1966) made the following definition of visual com-
fort:
“The term visual comfort describes the lack of the psychological
sense of pain, irritation or distraction, but visual comfort does
not aim at covering sensations of aesthetical appeal or discomfort
of the surroundings.”
Lighting quality is another term that is often used instead of visual com-
fort, but it is not quite synonymous. Sometimes, lighting quality covers
only the following factors: the adaptation of the eye, the colour of the
light source (colour temperature) and its colour rendering, the main di-
rection of the light, ability to reveal shapes, shadows and lustre, absence
of glare and flicker, etc.
Ljuskultur AB8 in Stockholm has developed a simple light chart to
help occupants judge their own light quality at work, Fig. 4.5. Although
it was developed for artificial lighting, it can also be used for the com-
bined daylit/artificially lit workplace. The light chart is a simple round
8. Ljuskultur AB is an informational body for the Swedish lighting industries.

Windows and daylight
97
disc to be put on the table, and it has eight questions and small reading
tests printed on it. The purpose is to roughly test if the illuminance level
is sufficient, if the surroundings are sufficiently bright, if the colour ren-
dering of the artificial light is good, if the work place is free of disturbing
shadows and to check for direct and indirect glare. It is intended as a
starting point for a more thorough evaluation of the workplace.
Figure 4.5 Light chart from Ljuskultur AB, Stockholm.
Liljefors & Ejhed (1990) hold the opinion that the following factors can
be used systematically for a visual evaluation of the lighting quality: light
level (how bright or dark it is in the room), light distribution (where is it
darker/brighter), shadows (where do they fall and their character), re-
flexes (where they are and their character), glare (where it is and how
obvious), light colour (if the colour of the light is perceived as warm/cold
etc.) and colour (if they look natural or distorted).
Veitch & Newsham (1996a, b) broadens the term lighting quality with
the following definition: “The degree to which the luminous environ-
ment supports the following requirements of the people who will use the
space: visual performance, post-visual performance (task performance and
behavioural effects other than vision), social interaction and communi-
cation, mood state (happiness, alertness, satisfaction, preference), health
and safety, and aesthetic judgments”.
By this wide definition lighting quality is not directly measurable, but
it directly relates the lit space to the tasks accomplished by the people
working there.

Energy-Efficient Window Systems
98
4.6.1 Common recommendations for illuminance
and luminance.
In order to ensure that a task can be performed with normal speed and
accuracy a minimum required illuminance level needs to be established.
Several documents provide tables or flow charts for finding the proper
illuminance for various tasks, e.g. The Illuminance Selection Procedure
(IESNA, 1993), the Lighting Schedule (CIBSE 1994), and Belysning
inomhus (Ljuskultur, 1990). For paper-based (reading and writing) of-
fice work, the recommended illuminance is usually around 500 lux. Com-
puter-based work may need lower levels, e.g. 300 lux. There are however
several studies that indicate that preferred light levels might be much
higher, up to 2000 lux or even 5000 lux, and may increase with age, see
review by Velds (1999).
In order to avoid annoying glare, the luminance of the visual field
should not be too high. Basically all recommendations for luminance
distribution ratios are based on an old paper by Luckiesh (1944), Velds,
1999). The widely used recommendations are that the luminance ratios
should not exceed the following:
between paper task and adjacent VDT
(Video Display Terminal) 3:1 or 1:3
between task and adjacent surroundings 3:1 or 1:3
between task and remote surroundings 10:1 or 1:10
between luminaires, windows or skylights
and adjacent surfaces 40:1
The same values are given in guidelines for good lighting quality in of-
fices by NUTEK (1994). The NUTEK requirement for maximum al-
lowed luminance level is 1000 cd/m2 within the field of view and 2000
cd/m2 outside the normal field of view. This is based on preserving vis-
ibility of video display terminals (VDT) that have a luminance on the
order of 100 cd/m2.
When windows and daylight are introduced in the workplace, it be-
comes difficult to fulfil such strict limitations on luminance ratios and
maximum luminances, since the sky can easily have a luminance of 10'000
cd/m2 and above. Thin white clouds reflecting the sunlight may even
have a luminance above 100'000 cd/m2. This will require the use of some
kind of shading device. Studies on discomfort glare by Velds (1999) have
however shown that it is difficult to establish a maximum tolerable sky
luminance that is acceptable to a majority of the individuals.

Windows and daylight
99
4.7 Psychological aspects of windows
The last two sections started to touch upon human aspects of daylighting.
In this section a literature review of the psychological benefits and draw-
backs of having windows will be given. Aspects that relate to the work in
articles III to IV are given separate subsections.
Generally, people prefer windowed space over windowless environ-
ments (Collins, 1976). People also tend to prefer daylight to artificial
light. In a recent survey of 1800 Danish office workers in 20 buildings,
the three most positive aspects of windows were: ability to see out, ability
to follow changes of weather, and ability to air a room (Christoffersen et
al., 1999b). Further benefits can also be attributed to windows. As Rusak
et al. (1996) puts it, “the obvious benefits of windows are the desirability
of having a view to the external world, an expansive sense of contact with
the outdoors, positive mood effects of seeing sunshine, and desire for the
warmth and atmosphere provided by sunlight in interiors at certain times
of day”. Perceived spaciousness has also been found to increase with the
presence of windows (Inui & Miyata, 1973).
Heerwagen (1990) suggests that there are four general benefits of win-
dows: (1) access to environmental information; (2) access to sensory
change; (3) a feeling of connection to the world outside and (4) restora-
tion and recovery. Heerwagen speculates that the first benefit may have
to do with our evolutionary past, that is was crucial for survival and
health to keep track of time of day, weather and other changing environ-
mental data. However, change is a basic characteristic of the natural world.
In contrast, interior space is deliberately kept at constant temperatures,
ventilation rates, illumination levels etc. Doubts have started to grow
that constant levels are not the optimum. There is evidence that sensory
change is fundamental to perception and may well be essential for effi-
cient functioning of the brain. Sensory change can also give a pleasant
experience independent of the data provided and, for many people, win-
dows are the only source of changing levels of sensation. The third point
stated by Heerwagen is that windowless space makes people feel enclosed
and shut off from the outside world. “Windows provide access to events
and situations in the world beyond our walled boundaries”. The fourth
point, restoration and recovery, is based on some evidence that nature,
and especially trees, have restorative effects, e.g. Ulrich (1984).
Heerwagen also points out that windows are not always beneficial.
View out also brings the possibility of view in. For some people privacy is
extremely important and windows create the possibility for unwanted

Energy-Efficient Window Systems
100
visual exposure. A discussion around the balance between visual access
and visual exposure, and her visual exposure matrix was already given in
Sec. 2.1.2.
4.7.1 View
Today, when mechanical ventilation and artificial lighting can provide us
with necessary air and light, view remains as perhaps the most important
aspect of windows. Apart from connecting us with the outside, view con-
tent has been associated with a number of different health outcomes.
Wilson (1972) found that post-operative delirium was more than halved
among surgical patients with access to windows, and concluded that win-
dows were highly desirable for the prevention of sensory deprivation.
This was later supported in a similar study by Ulrich (1984) where pa-
tients with a natural view had fewer complications, recovered faster, and
needed less pain-killers than those overlooking a brick wall.
Markus (1967) divides the view into three elements; (1) upwards, the
sky as the main source of light; (2) the horizontal part providing view of
landscape or city; (3) downwards, the view of ground and the activities
going on upon it. Almost all views have a horizontal stratification, i.e.
they contain a layer of ground, a layer of city or landscape and a layer of
sky. According to Markus, at least some small portion of all layers should
be present in a view, rather than one layer alone. He suggested that verti-
cal windows would allow for this. Also, tall buildings may require a lower
window placement for top floors than for bottom floors. Moreover, di-
viding lines between layers are very important and windows which are
otherwise generous but which obstruct these dividing lines lose much of
their orientating effect.
The satisfaction with the view is strongly related to the contents of the
view. In the study by Christoffersen (1999b), natural landscapes, trees,
vegetation and sky gave a stronger response of satisfaction than did park-
ing lots, tall buildings and industry. Further, satisfaction was related with
floor level of the occupant, with satisfaction increasing for those sitting
higher up in buildings.
Similar results were found by Markus (1967), in an assessment of 400
office workers in a 12-storey open-plan office in Bristol. About 88 % of
the subjects preferred views of the distant city and landscape, while only
12 % preferred a view of ground level buildings or of the sky. Markus
also found that people who sat far from the window expressed a greater
desire to sit near the windows, but that this may have been a complex
response conditioned partly by status, partly a feeling of real view depri-
vation, and partly by thermal or lighting conditions.

Windows and daylight
101
Cooper et al. (1973) found that subjects rated views with a greater
variety of distant objects as more satisfying. View content, height above
ground and age of the subject also affected judgements of the adequacy
and pleasantness of the view.
4.7.2 Window size, shape and position
In the next chapter it is demonstrated that large windows may lead to
thermal comfort problems and high energy costs both winter and sum-
mer. Therefore, window size may be restricted in order to meet demands
of energy efficiency.
Ne’eman and Hopkinson (in Collins, 1976) studied the minimum
acceptable window size using scale model experiments. They found that
for 50 % of the subjects, the smallest acceptable window size was 25 % of
the window wall, while a window size of 35 % was accepted by 85 % of
the subjects. Further, the acceptable window size was affected by several
parameters such as view content, distance from window, window height
and visual angle.
Keighley (in Collins, 1976) found that people preferred wide rather
than tall windows, contrary to predictions by Markus (1967). He also
found that window areas of 10 % or less of the window wall area was
highly unsatisfactory. Satisfaction increased for windows occupying 20 %
of the window wall, and was highest for 30 % or above.
In the recent Danish study, Christoffersen et al. (1999b) found that
more than 80 % of the employees were satisfied with their window (i.e.
glazing) size, which varied between 18 to 49 % of the façade area. Despite
this, there was a significant relationship between window size and assess-
ment of adequate size. For large window areas, the number of complaints
of ‘too large’ windows increased, as well as there were more complaints
with ‘too small’ windows when window size decreased. Thus, the opti-
mum window size was found for a glazing-to-wall area ratio (GWAR) of
about 30 %. Window size was also the one aspect that people wanted to
change the most. This was especially marked in buildings with very nar-
row windows (0.5-0.6 m). Window placement was also surveyed. In one
building with a large sill height (1.35 m), 72 % of the surveyed desired a
lower window placement. Further, the satisfaction of the indoor tem-
perature decreased with window size, and for west facing offices the dis-
satisfaction was higher than the satisfaction with GWARs above 25 %.

Energy-Efficient Window Systems
102
4.7.3 Window transmittance and tint
Glazing for solar control may severely reduce light transmittance as well
as changing the spectral composition of the transmitted daylight (tint).
There are several hearsay stories of complaints regarding such glazing,
both concerning mood effects and problems with colour rendering. For
example, one architectural firm was said to be very disappointed with
their hand-coloured drawings when they saw them outdoors for the first
time, as they were delivering them to the customer. However, only a few
studies have been performed on the effects of low transmittance glazing.
Cooper, et al. (1973) studied the effects of absorbing and reflecting
glass. In a pilot experiment little adverse effect was found from solar ab-
sorbing glass. However, two types of solar reflecting glass with light trans-
missions of 12 % and 15 % elicited complaints about the dark depressing
view out, the need to rely on artificial light and distracting reflections
from the glass on dark days. In their main experiment, 902 office workers
in 11 buildings were surveyed. They found no effects of the glazing on
the reported pleasantness or brightness of the view or on interior colours,
except for a small adverse effect on pleasantness and view for reflective
glass. However, only one of the eleven buildings had glazing with the
same low light transmittance as had evoked adverse effects in the pilot
study. Cooper et al. concluded that the glazing material is not noticed if
the view to the outside is sufficiently interesting, except for cases where
more than one glazing material is used.
Boyce et al. (1995) performed a simple experiment on minimum ac-
ceptable glazing transmittance, defined by the criterion that 85 % of peo-
ple consider the transmittance acceptable. A scale model of an office was
placed overlooking a parking lot, a building and parts of the sky. A set of
window panes + filters with varying transmittance could be positioned in
front of the window hole. The resulting transmittances ranged from 9 to
82 %. For each glazing combination the 25 subjects where asked too look
around the scale model office and say whether the glazing was acceptable
for a modern office or not. Boyce et al. concluded that the minimum
acceptable glazing transmittance lies in the range 25 % to 38 %. The vari-
ation was associated with glass type and sky condition, where a bright-
ness-enhancing glass was acceptable at a lower transmittance than a
spectrally neutral or brightness-reducing glass.
In article III, a super-insulated quadruple-pane window with two low-
emittance coatings was compared to a clear, triple-pane window. The
light transmittance of the glazing was 55 % and 73 % respectively. With-
out being told of the real purpose of the study, 95 subjects were able to
distinguish between the room with the clear triple-pane window and the

Windows and daylight
103
room with the super-insulated window. The room with the super-insu-
lated window was perceived as darker and more enclosed, and the day-
light as more tinted. The difference was thus large enough for people to
detect, but how people will be affected on the long term is still an unan-
swered question.
Similar well-insulated windows appear on today’s market. Usually, these
windows have three glass panes and two silver-based low-e coatings. The
daylight spectrum transmitted through such a glazing was compared to
the clear triple-pane window (Bülow-Hübe, 1994). The difference in the
tinting effect was not found to vary significantly from the quadruple-
pane window, Figs. 4.6 and 4.7. It was concluded that the two low-e
coatings were mainly responsible for the tinting of the daylight, and not
the fourth glass pane.
0.000
0.005
0.010
0.015
0.020
0.025
300 400 500 600 700 800
Wavelength (nm)
clear triple
super triple
super quadruple
Indoor irradiance (W/m²,nm)
Figure 4.6 Indoor spectral irradiance on desktop for superinsulated quadruple-
pane and triple-pane windows (both with two low-e coatings) and
a normal clear triple-pane window. Measurement in Lund 940506
at noon, clear sky with sunshine, average of 6 scans.

Energy-Efficient Window Systems
104
0.40
0.45
0.50
0.55
0.60
0.65
0.70
0.75
0.80
0.85
0.90
0.95
1.00
380 430 480 530 580 630 680 730 780
Wavelength (nm)
super triple / clear triple
super quadruple / clear triple
Irradiance ratio (-)
Figure 4.7 Irradiance ratio between the clear triple-pane window and two
superinsulated windows: quadruple-pane and triple-pane with two
low-e coatings. Measurement in Lund 940506 at noon, clear sky
with sunshine, average of 6 scans.
Daylight of different wavelengths gives rise to different colour sensations.
The following division is widely accepted: violet, ranging from 380 to
436 nm, blue from 436 to 495, green from 495 to 566, yellow from 566
to 589, orange from 589 to 627, and red from 627 to 780 nm (Küller,
1981). The relative sensitivity of the eye in the photopic state (daytime
vision) is highest for 555 nm, which corresponds to a green sensation. In
Fig. 3.7 it was demonstrated that silver-based low-e coatings, depending
on thickness, have a rather sharp transmittance peek between 530 to 560
nm, which thus corresponds to a green or yellow-green sensation. Clear
glass itself also has a transmittance peak in the green, which stems from
iron oxide contained in the glass. The peek is however much less sharp,
Fig. 3.6. The filtering effect of the glass is emphasised when several panes
and especially when several coatings are used, since the effect is multipli-
cative. The Swedish discussion around daylight quality and super-insu-
lated windows has mainly revolved around the absolute transmittance (in
percent) (Sec.2.2.2). To my opinion, there are however two main effects
of using more coatings and glass panes in a window: (1) the general illu-
VBGYOR

Windows and daylight
105
mination level or daylight factor will be lower; and (2) the filtering effect
or distortion of the spectrum will become more pronounced. Using slightly
larger windows can easily solve the first problem of low illuminance.
However, the sky luminance will remain the same. Resolving the second
one is much more difficult, and further studies are needed to study the
long-term satisfaction of such windows in real environments.
4.7.4 Avoidance of glare
In the Christoffersen study (1999b) the optimum window size was found
for a GWAR of 30 %. It was not directly concluded if this had to do with
glare or overheating at large glazing-to-wall area ratios. However, a sig-
nificant relationship was found between window size and disturbing glare,
which increased with increasing window size. It is however not obvious
that large windows will increase the risk of glare, since they at the same
time will increase the general illumination level in the room, and thus
reduce contrast. The most effective measure is to reduce sky luminance
without reducing window area (Ludlow, 1976). Flexible shading systems
such as window blinds, retractable awnings, and curtains can reduce glare
when sky luminance is high, and be removed on dull days to increase
daylighting. Daylighting systems (e.g. anidolic systems, laser-cut panels,
holographic optical elements) are other modern alternatives.
In the study by Christoffersen et al. (1999b) the office workers were
sometimes or often bothered by glare. Nevertheless, they wanted to sit
near to windows, and over 70 % of the surveyed had the computer placed
there.
In article IV it was hypothesised that the sky luminance seen through
the window in the close surrounding of the VDT would indicate when
shading devices were needed. However, neither this nor the interior desk
illuminance had a significant effect on the use of shading devices. How-
ever, the presence of a sunlight patch anywhere in the room showed a
significant relationship to the use of shading devices, although it only
explained a small portion of the variance. It was also obvious that there
was a large individual variation in the amount of glare that was tolerated.
This has also been found in previous studies, e.g. Velds (1999).
Office work has changed dramatically over the last decades with the
massive introduction of the personal computer and VDT in virtually all
of today’s workplaces. Indeed, in the Christoffersen study 95 % of the
employees used a computer at work, and they did so for 55 % of their
working time. The main effect of this is that the visual task is no longer
mainly on paper lying horizontally on the office desk but vertical on the
VDT. Thus, the main sight line is not downwards but almost horizontal

Energy-Efficient Window Systems
106
(preferably slightly downwards), which often brings the window into the
immediate line of sight. In open-plan fully-glazed offices it may be very
difficult to find a good placement for the VDT, since the sky seen through
the windows, also at a large distance, may be a cause of annoying glare.
The old rule of thumb for luminance ratios of 10:3:1 for artificial
illumination given in Sec. 5.6.1 will be very hard to reach with large
windows without the use of some type of shading device. Osterhaus (2001)
agrees with the findings of Christoffersen that windows are an extremely
important feature in offices and that daylight in general seems to pose
less of a problem than one might otherwise expect. Indeed, more research
is needed to establish preferred luminance ratios and to help understand
factors contributing to discomfort glare from daylight.
4.7.5 Sunlight penetration
Given the discussions of glare and overheating in the previous sections,
one may conclude that a total avoidance of sunlight would be the best.
However, some sunlighting is usually wanted since it has positive effects
on mood. According to a review by Collins (1976), almost all people
want to have sunshine in their dwellings, at least in Northern Europe
where we are not spoilt by an abundance of sunny days. Studies have
indicated that sunshine may be even more important than view, illumi-
nation, balconies etc. It is easy to understand that sunlighting is desired
in the home where movement is unrestricted: one can decide to sit in the
sunpatch to benefit from the heat, or avoid it if the lighting is too bright.
In an office a sunlight patch may be more disturbing since people are tied
to a rather small working surface, especially today with a computer screen
on the desk. Even so, the recent survey by Christoffersen (1999b) indi-
cated a strong preference for sunlight in the office, 60 % desire some
sunlight during one or several seasons. This supports the findings of a
much earlier study by Markus (1967), where an overwhelming majority
(86 %) of 400 surveyed office workers in a highly-glazed building pre-
ferred sunshine all year round.
Boubekri & Boyer (1991) studied the effects of window size and sun-
light penetration on office workers’ emotional state and degree of satis-
faction. They failed to demonstrate that window size affected mood and
satisfaction, but sunlight penetration significantly affected the feeling of
relaxation. The relaxation was highest for moderate sunlight penetration,
and decreased with both small and large sunlight patches.

Windows and daylight
107
4.8 Non-visual effects of light
For many years, environmental and lighting design was based on the
assumption that light only affected visual performance. The research on
the central vision of the eye began already in the beginning of the 20th
century, and is very extensive. One has also known that both lighting and
colours contributes to the impression of architectural space. At the same
time research has demonstrated that light entering the eye has a number
of other effects on humans. It has been shown that there is an activation
of various organs in the brain, especially the pineal gland, pituitary gland,
and the reticular formation of the brain stem. Light affects the diurnal
rhythm, metabolism, pulse frequency, blood pressure and the production
of hormones. Light may even increase the immune defence against cer-
tain types of infections.
The awareness and the importance of the non-visual effects of light-
ing, and especially of daylight, is growing rapidly. Many popular articles
around light therapy as a means to cure winter depression is appearing in
the newspapers and magazines. The lighting industry has caught on, and
is nowadays selling bright light sources for home use. In some countries,
the natural sleep hormone melatonin is being offered freely as a dietary
supplement – marketed for insomnia, winter depression, slowing the age-
ing process and as a cure for jet lag.
The non-visual effects of light (and colour) have been summarised in
an extensive bibliography by Küller (1981). He has divided the effects of
light into three main headings: physiological effects of solar radiation on
the (human) skin, physiological effects of daylight and artificial illumi-
nation entering the eye, psychological effects of light and colour. The
second of these areas has been extensively treated in an updated bibliog-
raphy (Küller & Küller, 2001). Some of the psychological effects associ-
ated with daylight have been mentioned in the previous section, and are
thus not repeated here.
4.8.1 Physiological effects of solar radiation on the
(human) skin
There are several well-known effects of solar radiation on the human
skin. For example, UV-C radiation (100-280 nm) has a strong germi-
cidal effect and can cause superficial erythema and conjunctivitis. UV-B
radiation (280-315 nm) causes vitamin D formation in the body (vital
for calcium intake), and has erythemal (reddening of the skin) and
pigmenting (tanning) effects. Ordinary window glass absorbs essentially

Energy-Efficient Window Systems
108
all radiation within this range. However, UV-A radiation (315-400 nm)
passes through most types of glass, but has no effects on vitamin D for-
mation. Although UV-A light is less effective as a tanning agent, it is
most often used in tanning equipment. UV light is also known to cause
skin cancer (Küller, 1981, Rusak, et al., 1996).
4.8.2 Physiological effects of daylight and artificial
illumination entering the eye
Light entering the eye has a synchronizing effect on the diurnal and sea-
sonal rhythms prevalent in human beings, for example the sleep-wake
cycle and the production of hormones. Sunlight has been shown to in-
hibit the secretion of melatonin – a sleep hormone produced by the pin-
eal gland in the brain – and adaptation to the day–night cycle and to
seasonal changes in day length are thus mediated by this mechanism. If
our internal clock is not reset daily, we go into our own sleep-wake cycle
(which is usually longer than 24 hours), we become circadic. This was
first shown amongst totally blind people and also amongst mine workers
(Hollwich, 1979). While bright artificial light (approx. 2500 lux) has
been shown to suppress melatonin levels, ordinary indoor light levels
may not do so. (Wetterberg, 1978, Boyce & Kennaway, 1987). Some
psychiatric disorders such as SAD (seasonal affective disorder) have also
been associated with the lack of daylight (Küller & Küller, 2001; Tonello,
2001).
Küller & Lindsten (1992) reported that the seasonal hormone cycle of
morning cortisol (a stress hormone) in school children situated in class-
rooms without daylight differed from that of children in ordinary class-
rooms. There also seemed to be some effects on sociability and body
growth, and they recommend that classrooms without windows should
be avoided for permanent use. In another study on the subterranean en-
vironment, Küller & Wetterberg (1996) found some interesting differ-
ences in spaces below and above ground. On average, the illumination
was twice as high in offices above than below ground. The level of morn-
ing cortisol displayed a substantial annual variation in personnel above
ground, but the variation was much less pronounced in personnel below
ground. The variation between daytime and nighttime melatonin levels
was much larger for underground personnel, and on average they slept
half an hour longer than personnel above ground.
While there seem to be ample evidence that light entering the eye acts
as a synchronizer for biological clocks, it is not yet quite certain whether
it is the amount of light, its spectral composition or variations in the day
length that are decisive. The current belief is that light intensity is the

Windows and daylight
109
most important factor. This hypothesis is intuitively correct when one
compares typical indoor and outdoor levels, especially for spaces far away
from windows: indoor levels can range from 50 lux (typically found in a
corridor etc.) to 200-500 lux for office work and about 1500 lux for very
detailed work. Outdoors we may experience anything between 5'000-
10'000 lux on an overcast day up to 100'000 lux on a sunny day. The
amount of illumination we are exposed to in artificially lit environments
is thus only a tiny fraction of the exposure outdoors, and it is evident that
daylight acts as a very strong signal. Many authors seem to consider that
there are possible health risks of spending too much time indoors. It is
however difficult to increase interior light levels for several reasons: we do
not want to waste electrical energy, and it is more difficult to create a
pleasant, glare free atmosphere with bright artificial lighting. Küller et al.,
(1999) has demonstrated that people who work further than 2 m from a
window are more likely to experience SAD or SAD-like symptoms. Küller
recommends that people should sit near to a window, and/or to take a
daily walk outside, preferably in the morning.
There are also effects of light on the activation of the central and auto-
nomic nervous systems, i.e on the physiological arousal level. One exam-
ple is flicker from conventional fluorescent light sources. Although the
flicker is not percepted by the visual system, it may influence the basic
brainwave pattern (EEG) of the central nervous system. This has been
suspected to cause undue stress resulting in headaches and eye-strain
(Wilkins et al, 1989, Küller & Laike, 1998).
4.8.3 Psychological effects of light and colour
There is a long history of speculation that light and colour can influence
task performance, comfort and well-being. A large part of the work in
this field concern the artificially lit environment and preferences for vari-
ous conditions. There are several bibliographies and reviews for the inter-
ested reader (Küller, 1981; Rusak et al., 1996; Veitch, 2001). The evalu-
ation of discomfort glare indices also takes a central position. A discus-
sion of glare indices is presented in Velds (1999) and Dubois (2001).

Energy-Efficient Window Systems
110

Windows and energy
111
5 Windows and energy
In order to illustrate the effect of window choice on the heating demand,
cooling demand and indoor temperatures, some calculation examples for
both residential and office buildings are given in this chapter.
There exist some simple methods to estimate the net energy gain
through windows, taking into account both thermal losses, Qloss, and
solar gains, Qsolar, e.g. (Karlsson & Brunström, 1987; Roos & Karlsson,
1994; Nielsen & Svendsen, 2000; Karlsson, Karlsson & Roos, 2001).
These methods usually incorporate some form of degree-day method in
order to look at the energy gain over the heating season only. The heating
season can be either fixed, or defined by the balance temperature of the
house. In the (Björn) “Karlsson” window formulae, the net annual en-
ergy transport through the window, Q, is defined as:
() ()
EEORVVVRODU
W8*W6J444 −=−= (5.1)
where
J
is the annual mean value of the total solar energy transmittance
(-), S is the total solar radiation impinging on the window summed up to
the balance temperature (Wh/m2,yr), U is the U-value of the window
(W/m2K), and G is the degree-hours summed up to the balance tempera-
ture of the building (°h/yr). The balance temperature is defined as the
average outdoor temperature above which the building does not need to
be actively heated, and thus differs slightly from the usual definition. For
the climate in Älvkarleby (lat 60°N) Roos & Karlsson (1994) found that
J
could be represented by the g-value for 55° incidence angle. With
climatic data for Älvkarleby 1985, an indoor temperature of 20°C and a
balance temperature of 13°C, the equation becomes (in kWh/yr,m2 win-
dow area), (Merkell, 1989):
South windows: 127465 ⋅−⋅= 8J4 (5.2a)
East windows: 127324 ⋅−⋅= 8J4 (5.2b)
West windows: 127234 ⋅−⋅= 8J4 (5.2c)
North windows: 127130 ⋅−⋅= 8J4 (5.2d)

Energy-Efficient Window Systems
112
Later, Joakim Karlsson improved this method by calculating the energy
balance on an hourly basis, thus employing the correct angular depend-
ent g-value, g(
θ
), for each hour. The method was released as a simple
computer tool, WinSel (Karlsson, 2000).
In all of these methods it is required that either the length of the heat-
ing season or the balance temperature is defined from previous experi-
ence, since it will depend on the general insulation level of the house, the
ventilation rate, and the internal gains. Further, these methods usually
only look at the window alone, and not on the interaction with the whole
building. Even if the formulae can be adapted to look at cooling de-
mands, it will still be difficult to estimate potential overheating problems
since ventilation and internal load schemes, storage effects in the thermal
mass of the building etc. are difficult to account for. I have therefore
chosen to perform the calculations in the dynamic energy simulation
program Derob-LTH, which is described elsewhere in this report, see for
example Sec. 1.3 and articles II and V.
5.1 Single family house
A detached 1.5 storey single-family house without basement was mod-
elled. The ground floor is 8.1 by 11.9 m, and the living area is 150 m2.
The geometry of the house is identical to the reference house selected
within IEA/SHC Task 28/ Annex 38: “Sustainable Solar Housing” (Smeds
& Wall, 2001a).
The total window area is 22 m2, which gives a window-to-floor area
ratio (WFAR) of 15 %. The house is thus moderately glazed but rather
typical for Swedish houses built from 1975 an on. The relationship be-
tween glazing and frame area was assumed to be 70/30 %. General data
of the house are given in Table 5.1 and the geometry of the Derob-model
is shown in Fig. 5.1.
In order to illustrated the effect of different window U-values on the
heating demand for houses from various time ages, three insulation levels
were chosen corresponding to typical levels from 1960, 1980, and today
(2000). A hypothetical case (2020) for an extremely well insulated build-
ing was also constructed – perhaps the standard of year 2020? The insu-
lation levels in this house correspond to those of some newly erected row
houses in Lindås, outside of Gothenburg, designed by EFEM Arkitekt-
kontor, see also Chapter 1. For simplicity, no thermal bridges accounting
for extra heat losses at connections between wall-roof, floor-wall etc. have
been accounted for. To obtain the predicted low heating demands of es-

Windows and energy
113
pecially the houses 2000 and 2020 will in reality require extreme care in
the design of these connections, in order to avoid thermal bridges that
can otherwise significantly increase the thermal losses.
In the first set of simulations, the houses were all placed in the climate
of Stockholm (climate year 1988), and the main façade facing south. The
houses were later also studied for the climates of Lund (1988) and Luleå
(1988). Thus, by the case ‘Stockholm 2000’ is meant a house with the
insulation levels of year 2000, placed in the Stockholm climate for refer-
ence year 1988.
The internal loads were taken from IEA/SHC Task 28, and in all cases
but 2020 the case REF 2+2 was chosen (Smeds & Wall, 2001b). This is a
case that models a family of two adults and two children, and with older
household equipment. For the house 2020 the high performance case,
HP 2+2, was used, and this is a case that reflects the internal loads for the
same family, but with modern energy-efficient household appliances.
The case Stockholm 2000 was studied for the main façade oriented in
all four major directions (S, E, N and W). The choice between windows
7-9 (see Table 5.2) was also evaluated for this case.
The calculations focus on the heating demand of the houses as a func-
tion of window U-value, and do not include electricity use for the me-
chanical ventilation needed. However, modern systems use about 20-35
W per fan, and in air-to-air heat recovery systems two fans are needed,
thus the annual electricity demand for the ventilation system is rather
small, between 175 and 310 kWh. Older systems may use more electric-
ity, due to higher power of the fans. In for example Solbyn, the energy use
for the ventilation system was measured to 900 kWh/yr, including inter-
mittent defrosting of the heat recovery unit (see article I).
Table 5.1 Information of the 1.5-storey detached single family house.
General Window areas
City, climate year Stockholm 1988 South 9 m2
Latitude 59°N East 3 m2
Ventilated volume 85 % of 400 m3North 1 m2
Living area 150 m2 (96+54) West 9 m2
Occupants 2 adults + 2 children
Thermostat setting 20°C, constant

Energy-Efficient Window Systems
114
Figure 5.1 Derob-model of the 1.5-storey house. The light-coloured surfaces
are shading elements that model the window reveal and the eaves.
Window frames and doors are modelled as rectangular wall sur-
faces.
5.1.1 Window types
Five glazing combinations were selected along the U(g)-curve presented
in Fig. 3.12. The graph is again presented here, showing the selected
glazing combinations as uncoloured circles, Fig. 5.2. Five realistic frames
were matched with these glazings to create the five main window types 1-
5: (1) a coupled double-pane window; (2) a clear triple-pane IGU win-
dow; (3) a triple IGU window with one low-e coating and argon; (4) a
1+3 window with one low-e coating and argon; and (5) a triple IGU
window with 2 low-e coatings and krypton in both cavities. The frame
U-values Uf were calculated from the practical window U-values, Up,win
and the center-of-glass U-values Ucog. This means that all additional heat
losses due to thermal bridges etc. are accounted for in Uf . Window types
1-3 are “standard” window types on today’s market, window 4 is usually
not produced, and window 5 can be specially ordered.
Some alternative windows were also studied in a few other cases. Win-
dow type 6 was selected as a renovation alternative of window 1 where
the clear inner pane is replaced with a hard tin-oxide low-e coating. Win-
dow types 7-9 are all double-pane insulated glass units with low-e coat-
ings and argon, the only difference is that the emissivity of the coating is
successively reduced from 16 % to 10 % and finally 4 %. These coatings
can all be found on the market today, although the 16 % alternative is
usually not used in IGUs, and the 10 % emittance coating is more and
more being replaced by the 4 % coating. Window type 10 is another

Windows and energy
115
commonly sold window type today, it corresponds to a 1+2 construction
(coupled) with one low-e coating and argon gas. The thermal and solar
properties of the selected windows are given in Table 5.2.
Consumer prices for the windows were gathered from the window
manufacturer SP Fönster, Table 5.2 (Anders Jonsäll, SP fönster, personal
communication, 2001). However, since the quadruple-pane window (#4)
is not in production, an assumption was made that it would be 500 SEK
more expensive than window type 10. These prices were then used to
study the economy of replacing windows in different situations.
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.01.02.03.04.05.06.0
U
cog
-value (W/m
2
K)
g
-value (-)
#2
#1
#3
#5
#6
#9
#4
#10
#8
#7
Figure 5.2 Selected glazing types. The main glazing types (1-5) are chosen along
the dotted g(U) curve. Glazing 6 is similar to 1, but with a hard
low-e coating on the inner pane. Glazings 7-9 are all double-pane
low-e coated IGUs with argon, but the emittance is succesively de-
creased from 16 %, to 10 % and 4 %. Glazing 10 is similar to 3,
but the emittance of the low-e coating is 4 instead of 10 %.

Energy-Efficient Window Systems
116
Table 5.2 Thermal and solar properties and consumer price (excl VAT) of
the 10 selected window types. The U-values of the window are
based on a window 1.0 × 1.2 m. T stands for triple, and D for
double-pane IGU, e.g. T4-12 is a triple IGU with 4 mm glass
and 12 mm gap widths. Since a practical U-value of windows is
usually given as Uwin + 0.05, this additional heat loss is accounted
for in the frame U-value. The window prices are based on the
same size as the U-value, and are the prices of white painted
windows.
Window Description Ucog UfUp,win gPrice
(W/m2K) (-) (SEK)
1 4-30-4 2.71 2.09 2.50 0.76 1495
2 T4-12 1.85 2.14 1.95 0.68 2860
3 T4-12 e10%+Ar 1.12 2.30 1.52 0.57 3092
4 4-46-(T4-15 e4%+Ar) 0.76 1.49 1.01 0.48 3724
5 T4-12 2 e4%+2Kr 0.50 2.14 1.06 0.42 3546
6 4-30-4 e16% 1.81 2.09 1.90 0.72 1688
7 D4-15 e16%+Ar 1.50 2.57 1.86 0.71 2195
8 D4-15 e10%+Ar 1.35 2.57 1.76 0.65 2162
9 D4-15 e4%+Ar 1.18 2.57 1.60 0.60 2162
10 4-30-(D4-12 e4%+Ar) 0.97 1.80 1.25 0.53 3224
5.1.2 Insulation levels and ventilation
The chosen ventilation and infiltration rates and the U-values of the build-
ing components for the different houses are given in Table 5.3. A brief
characterisation of the houses is given below.
House from 1960
The house from 1960 was assumed to have 9.5 cm of wall insulation, 12
cm in the roof and 7 cm on the slab on ground. With window type 1 as
standard, the house neither fulfils the requirements of SBN 75/80 nor
those of NR 1/BBR 99, see Sec 2.2.1 for a discussion of Swedish build-
ing codes.
The ventilation was assumed to be 0.5 ach of the ventilated volume.
Since many houses from this time do not have mechanical ventilation,
no heat recovery was assumed.
House from 1980
The house from 1980 was assumed to have 19 cm of wall insulation,
24.5 cm in the roof and 5-7 cm under the slab on ground. With window
type 2, the house fulfils the requirements of SBN 75/80 but not of NR 1.

Windows and energy
117
The ventilation was assumed to be 0.5 ach of the ventilated volume,
with 50 % heat recovery.
House from 2000
In order to find typical U-values of modern houses, a Swedish manu-
facturer of prefabricated houses was contacted (Myresjöhus). The U-val-
ues for the house 2000 were selected as identical to the current production
of Myresjöhus. Therefore, the walls have 24 cm of insulation, the hori-
zontal roofs have 45 cm of loose fill insulation, and there is about 10-15
cm of insulation under the slab. With window type 3, the house fulfils
the requirements of both SBN 75/80 and NR 1/BBR 99.
The ventilation was assumed to be 0.5 ach of the ventilated volume,
with 50 % heat recovery.
House from 2020
The insulation levels in this house are rather typical for so called “passive
houses”, houses that do not need more than 15 kWh/yr,m2 floor area for
heating, and where the total energy demand (heating, domestic hot water
and household electricity) does not exceed 42 kWh/yr,m2. The U-values
for this house roughly corresponds to walls with 45-50 cm of insulation,
the horizontal roofs have 60 cm loose fill insulation, and there is about
30 cm of insulation under the slab.
With window type 5 the house more than fulfils the requirements of
NR 1/BBR 99. The ventilation was assumed to be 0.5 ach of the venti-
lated volume, with 85 % heat recovery. The infiltration rate was halved to
0.05 ach. Further, the internal load corresponds to very energy-efficient
equipment plus people as before, resulting in lower free heat gains.
Heating demand and useful solar gains
The annual heating demand for the Stockholm climate was simulated in
Derob-LTH for the four houses described above. Every house was simu-
lated with a window type typical for each house, see Table 5.3. The useful
solar gains were found as the difference of two consecutive runs, without
and with solar radiation from the climate file. The results are presented in
Fig. 5.3. This clearly demonstrates that the annual heating demand is
dramatically reduced when the insulation of the building envelope is im-
proved. Compared to the house of 1960, the heating demands (aux. heat-
ing) are reduced by 47% for the house 1980, by 59% for the house 2000
and by 83% for the house 2020! Since the thermal losses are much smaller
for the newer houses, transmitted solar radiation covers an increasingly
larger fraction of the total demand for space heating.

Energy-Efficient Window Systems
118
Table 5.3 Ventilation rates and insulation levels for the four houses. The
average U-values Uave and Uave,req are calculated according to
NR1/BBR99, which includes a subtraction of 0.7 for all win-
dow U-values. This subtraction of 0.7 is however not used in
the dynamic simulations.
Construction year of house
1960 1980 2000 2020
Ventilation (ach) 0.5 0.5 0.5 0.5
Heat recovery efficiency 0 %50 %50 %85 %
Infiltration (ach) 0.1 0.1 0.1 0.05
Building comp. Area (m2)U-value (W/m2K)
Floor on ground 96.4 0.30 0.25 0.19 0.10
Ext. walls 106 0.47 0.25 0.17 0.08
Walls to attic 50.0 0.36 0.16 0.16 0.11
Roof, 45°20.2 0.36 0.16 0.16 0.11
Roof, horizontal 82.1 0.34 0.17 0.12 0.08
Doors 3.6 2.19 1.00 0.80 0.80
Windows 22.0 2.50 1.95 1.52 1.06
Uave (NR1/BBR 99) 0.454 0.268 0.195 0.107
Uave,req (NR1/BBR 99) 0.244 0.244 0.244 0.244
Fulfils SBN75/80 no yes yes yes
Fulfils NR1/BBR 99 no no yes yes
Annual energy (kWh/yr)
0
5000
10000
15000
20000
25000
30000
1960, #1 1980, #2 2000, #3 2020, #5
House and window type
Useful solar gain
Aux. heating
Figure 5.3 Simulated annual heating demand with sun (aux. heating) and use-
ful solar gains in (kWh/yr) for four houses in Stockholm built bet-
ween 1960-2020 with window types typical for each time period.

Windows and energy
119
The heating demand with and without sun, and the resulting useful solar
gains are also presented on a monthly basis for the four houses, Fig. 5.4.
The incident solar radiation S (on the outside of the glass areas) is also
shown, and it is of course the same for the four houses. These graphs
illustrate several interesting facts: (1) in the winter, especially between
December and February there is very little solar radiation available. The
auxiliary heating demand (simulation with sun) is thereby not much lower
than a simulation without sun during this period; (2) during March,
April, October and November, all houses benefit from solar gains in a
substantial way, and the peak for useful solar gains occurs in April and
October; (3) for the super-insulated house, the incident radiation is much
higher than the heating demand already in April, and the house risks to
get overheating problems that may last until the end of September, see
also sec. 5.1.3. However, the g-value of the window is also lower, which is
beneficial since the transmitted solar radiation is reduced. The ratio be-
tween useful solar gains and the incident radiation S is shown in Fig. 5.5,
and this graph illustrates the combined effect of reduced U-values and
reduced g-values of highly insulated windows (and buildings).
Total energy (kWh/month)
0
500
1000
1500
2000
2500
3000
3500
4000
4500
JFMA MJJA SOND
Month
Heating, no sun
Heating w. sun
Useful solar gain
Inc. solar radiation
Stockholm 1960,
window 1
Total energy (kWh/month)
0
500
1000
1500
2000
2500
JFMA MJJA SOND
Month
Stockholm 1980, window 2
Total energy (kWh/month)
0
500
1000
1500
2000
2500
JFMA MJJA SOND
Month
Stockholm 2000, window 3
Total energy (kWh/month)
0
500
1000
1500
2000
2500
JFMA MJJA SOND
Month
Stockholm 2020, window 5
Figure 5.4 Monthly heating demand with and without sun, resulting useful
solar gains and incident solar radiation (on the outside of the glass
area) in (kWh/month) as a function of window U-value and con-
struction year of house. Stockholm climate.

Energy-Efficient Window Systems
120
Useful solar gain to incident radiation S ratio (-)
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
JFMA MJ JA SOND
Month
1960, #1
1980, #2
2000, #3
2020, #5
Figure 5.5 Ratio between useful solar gain and total incident radiation S (on
the outside of the window glass) (-) as a function of window type
and construction year of house. Stockholm climate.
5.1.3 Effects of window choice on energy demands
and indoor temperatures
The effects of the choice of window type (U-value and g-value) will be
further studied below. The previous section demonstrated that the total
heating demand is very different for the four houses, Fig. 5.3. Here, and
in the following sections, new sets of simulations are presented were the
heating demand for each house has been determined for the window
types 1-5, ranging from double-clear windows to energy-efficient low-e
coated triple and quadruple-pane windows.
General results for houses in Stockholm
Figure 5.6 shows that the heating demand increases almost linearly with
the total window U-value. The heating demand for each house without
windows was also calculated, and is shown in the same graph (no win-
dows). In these cases the window area (both glass and frame) was re-
placed by opaque walls with the same U-value as the rest of the exterior
walls. It is clear that solar radiation gains through windows give signifi-
cant contributions to the heating of the house, since the heating demand
is about the same for a house without windows compared to a house with
windows (which have a significantly higher U-value).

Windows and energy
121
Heating demand (kWh/yr)
0
5000
10000
15000
20000
25000
00.511.522.53
Window U-value, Up,win (W/m²K)
1960
1980
2000
2020
No windows
Stockholm
Figure 5.6 Total heating demand (kWh/yr) as a function of window U-value
and construction year of house. Calculations performed for the Stock-
holm climate.
In order to clearly illustrate the breakpoint for which window U-value is
required in order for the window to be better than the wall, the incre-
mental heating demand was calculated. This was done by – for each house
– taking the energy demand associated with each window type and sub-
tracting the corresponding energy demand for the house without win-
dows (when the whole window area is replaced by wall area) and divide
by the total window area. The resulting incremental heating demand for
the four houses in the Stockholm climate is shown in Fig. 5.7. For the
poorly insulated house (1960), it is always better to replace the wall area
with window area as long as the U-value of the windows is about 1.6 W/
m2K or lower. For the house from 1980, the window must have a U-
value of 1.2 or lower to be better than the wall. For the house of today
(2000) the U-value should be lower than 1.0 and for the extremely well-
insulated house (2020), the U-value should be better than 0.65.
The peak heating load shows a linear relationship with the total win-
dow U-value, see Fig. 5.8. The peak load is always higher for a windowed
house, and this is natural since the peak load occurs when there is no
solar radiation, and the windows are replaced by walls of lower U-value.

Energy-Efficient Window Systems
122
-100
-50
0
50
100
150
200
250
00.511.522.53
Window U-value, U
p,win
(W/m²K)
1960
1980
2000
2020
Incremental heating demand (kWh/yr,m² window)
Stockholm
Figure 5.7 Incremental heating demand (kWh/yr,m2 window area) as a func-
tion of window U-value and construction year of house. Simulations
for Stockholm and window types 1-5 according to Table 5.2.
Peak heating load (kW)
0
1
2
3
4
5
6
7
8
9
00.511.522.53
Window U-value, U
p,win
(W/m²K)
1960
1980
2000
2020
No windows
Stockholm
Figure 5.8 Peak heating load (kW) as a function of window U-value and con-
struction year of house. Simulations for Stockholm and window
types 1-5 according to Table 5.2

Windows and energy
123
What happens with the indoor climate when both the U-values of the
house and the windows are improved? Naturally, the indoor tempera-
tures will increase. This is demonstrated in Fig 5.9, which shows the
duration curves of the operative temperature for the ground floor for
window type 3 for the four houses. The simulations were done for a
constant ventilation rate all year round and with the heat recovery unit in
place all through the year, also during the non-heating season (even if it is
normally better not to use it during the summer).
Stockholm, window type 3
15
20
25
30
35
40
0 1000 2000 3000 4000 5000 6000 7000 8000
Hours above temperature
1960
1980
2000
2020
Operative temperature (°C)
Figure 5.9 Global operative temperature (°C) of the ground floor (midpoint of
room) as a function of construction year of house. Window type 3 in
the Stockholm climate.
First of all, one can see that the minimum operative temperature is about
17-18°C, even if the air temperature never goes below 20°C (thermostat
set-point). According to the building code, the operative temperature
should be above 18°C (BBR 1999) and this is the case for the house 2000
(the requirements on the operative temperature are further described at
the end of sec. 5.3). During the heating season, the improved insulation
levels are demonstrated as fewer hours with operative temperatures below
20°C. During the non-heating season it is evident that overheating prob-
lems may occur as the insulation levels increase. Especially the super-
insulated house suffers from extreme overheating problems. Some shad-
ing will naturally occur, from curtains or blinds, outdoor vegetation and

Energy-Efficient Window Systems
124
surrounding buildings. If the ventilation is increased during the summer
(e.g. by-passing the heat recovery unit), the situation will be improved.
However, for extremely well-insulated buildings the over-heating prob-
lem must not be neglected, and external shading devices may become a
necessity.
As the g-value is reduced, the amount of primarily transmitted solar
radiation decreases. This is independent of the insulation level of the
house, and only depends on the climate and window type, see Fig 5.10.
Although the difference is large, the net result on the operative tempera-
tures is quite small, since the window area is just a small share of the total
surface area of the house. This is demonstrated in Fig. 5.11, which shows
the operative temperatures for the house 2000 for windows 1, 3 and 5.
During the non-heating season, there is not a significant difference be-
tween the window types, although the lower g-value of window 5 is re-
flected as a slightly lower operative temperature. During the heating-sea-
son, this window naturally results in a higher operative temperature, since
it has the lowest U-value. However, close to the window, the differences
between windows of various U-values will be much larger than shown for
this large volume which has a relatively small window area. This can be
seen in the Figs. 5.12-5.14 produced by the COMFORT program, a post-
processor to Derob-LTH.
Primary transmission of solar radiation (kWh/month)
0
100
200
300
400
500
600
700
800
900
1000
jan feb mar apr may june july aug sep okt nov dec
Month
#1
#2
#3
#4
#5
Figure 5.10 Primary transmission of solar radiation (direct and diffuse compo-
nents) (kWh/month) through all windows of the house as a func-
tion of window type. Stockholm climate.

Windows and energy
125
15
20
25
30
35
40
0 1000 2000 3000 4000 5000 6000 7000 8000
Hours above temperature
window 1
window 3
window 5
Operative temperature (°C)
Stockholm 2000
Figure 5.11 Global operative temperature of the ground floor (midpoint of room)
as a function of window type for the house Stockholm 2000.
Figure 5.12 Global operative temperature distribution (°C) at 1.8 m above floor
for the ground floor for house 2000, window type 1 at an outdoor
temperature –10°C, no solar radiation. Generated by the post-proc-
essor COMFORT.

Energy-Efficient Window Systems
126
Figure 5.13 Global operative temperature distribution (°C) at 1.8 m above floor
for the ground floor for house 2000, window type 3 at an outdoor
temperature –10°C, no solar radiation. Generated by the post-proc-
essor COMFORT.
Figure 5.14 Global operative temperature distribution (°C) at 1.8 m above floor
for the ground floor for house 2000, window type 5 at an outdoor
temperature –10°C, no solar radiation. Generated by the post-proc-
essor COMFORT.

Windows and energy
127
5.1.4 Cost efficiency of window replacement
The following examples are made to illustrate the possible energy savings
of either replacing old windows or selecting slightly better ones than first
thought, and the associated costs. The investment costs are annualised
using an interest rate of 6 % and an assumed service life of 30 years,
which gives an annuity factor of 0.0726. The annualised investment costs
can then be directly compared to the annual savings (the energy saving
times the energy price) to see if an alternative is cost-effective or not.
Alternatively, the annual savings can be divided by the annuity factor to
obtain the maximum “allowed” investment cost for a measure to be cost-
effective, and this method is employed here. The house is assumed to
have 16 windows, and for simplicity, we assume that the cost for each
window is equal to that of a 1 × 1.2 m window.
House from 1960
In this example we assume that the original windows, coupled double
pane, are in such a bad condition that either renovation or replacement is
necessary. In the simplest alternative we assume that the window can be
renovated. With rather small measures, the window can be upgraded by
replacing e.g. the inner pane of the window by a hard low-e coating (win-
dow 6). In the other (replacement) alternatives, we assume that the house-
owner has decided to buy a triple IGU window (#2), and we only look at
the additional cost for selecting better alternatives (i.e. windows with a
lower U-value). Therefore, the calculated costs do not mirror the whole
cost of renovation or replacement, since costs associated with e.g. new
sealants, trimmings, paint work in the window recess etc. is considered to
be done in any case, and are thus not accounted for. Figure 5.15 shows
the annual savings in heating demand calculated for the Stockholm cli-
mate.
In the renovation alternative (1->6, double clear to hard coated dou-
ble-pane window) Fig. 5.15 shows that the annual saving is 90 kWh/
window. With three assumed levels of the variable cost for heating energy
the low-e pane may cost the consumer (incl 25 % VAT), 301 SEK for an
energy price of 1 SEK/kWh, 211 SEK if the energy price is 0.7 SEK/
kWh and 120 SEK if the energy price is 0.4 SEK/kWh assuming a serv-
ice life of 30 years and a 6 % interest rate. This might therefore be a cost-
effective measure (except for the lowest energy price) since a low-e coated
glass applied at the factory cost about 240 SEK extra per window. How-

Energy-Efficient Window Systems
128
ever, small glaziers may charge the consumer anything between 300 to
800 SEK per window, since the ordered quantities are small, and with
such prices, it is not a cost-effective alternative.
0
20
40
60
80
100
120
1->6
2->3
2->4
2->5
2->7
2->8
2->9
2->10
Heat demand reduction (kWh/ r, window)
Stockholm 1960
Figure 5.15 Annual savings in heating energy per window as a function of win-
dow type replacement. Stockholm 1960. Window types according
to Table 5.2.
In the replacement alternatives, the energy savings have been calculated
as the additional saving compared to window type 2 (triple-pane clear),
Fig 5.15. The allowed additional investment costs have been calculated
for various energy prices, Table 5.4. The real additional investment cost
for these alternatives have been estimated based on the purchase prices
given in Table 5.2, to serve as a guide for cost comparison. Alternatives 3
and 7-10 all seem justified from this simple cost analysis. Especially win-
dows 7-9 seem very cost effective, since the first cost of these windows are
even lower than for the triple-pane IGU window chosen as the base case
(window 2). This explains why the real estimated costs are negative for
these alternatives. However, it must be remembered that the energy sav-
ings are smaller compared to other alternatives, e.g. window 10.

Windows and energy
129
Table 5.4 Allowed replacement cost for various energy-efficiency improve-
ments compared to estimated real costs for a house from 1960
in Stockholm. Cost-efficient alternatives are shown in bold-face.
Window replacement alternative
Allowed replacement cost
Energy price 2->3 2->4 2->5 2->7 2->8 2->9 2->10
(SEK/kWh)
1.0
764 1423 1267 457 533 638 1126
0.7
535
996
887 320 373 447 788
0.4
306
569 507
183 213 255
450
Estimated cost 290 1080 858 -831 -873 -873 455
House from 2000
In this example we consider a house that is to be built today, therefore the
house 2000 is used as a basis for the calculations. Since today’s houses
often are built with better windows than window 2, we consider that
window 3 (triple, low-e + Ar) is our first choice. However, for a small
additional cost, we can also choose other alternatives that will provide
additional savings (i.e. windows 4, 5, and 10). Window type 9 (double,
low-e + Ar) has also been included as one possible alternative. Even if the
U-value is slightly worse, the price seems very attractive. It was also made
sure that the house 2000 will fulfil the requirements of the current build-
ing code with any of these windows. Otherwise, the same principles are
applied as in the example from 1960. The annual savings in heating en-
ergy are shown in Fig. 5.16.
The allowed and real additional investment costs are found in Table
5.5. Alternatives 4 and 5 do not seem justified since the allowed cost is
always lower than the real estimated cost. Alternative 10 seem to be cost-
effective or – for the lowest energy price – close to being so. The negative
values for type 9 perhaps need some explanation. The allowed cost is
negative, since the quality of this window is slightly lower than for win-
dow 3 (i.e. higher U-value which leads to a higher heating demand).
However, this window (a double-pane alternative) is much less expensive
in purchase (negative estimated cost), therefore it might be chosen even if
the thermal comfort and heating demand will be slightly worse than for
the other alternatives.

Energy-Efficient Window Systems
130
-20
-10
0
10
20
30
40
50
60
3->4 3->5 3->9 3->10
Heat demand reduction (kWh/
y
r, window)
Stockholm 2000
Figure 5.16 Annual savings in heating energy per window as a function of win-
dow type replacement. Stockholm 2000. Window types according
to Table 5.2.
Table 5.5 Allowed additional cost for various energy-efficiency im-
provements compared to estimated real costs. Cost-efficient al-
ternatives are shown in bold-face.
Alternative window choice
Allowed additional cost
Energy price 3->4 3->5 3->9 3->10
(SEK/kWh)
1.0 677 534
-161 364
0.7 474 373
-113 255
0.4 271 213
-64
146
Estimated cost 790 568 -1163 165
5.1.5 Effects of orientation
With the original layout, the selected house has its main (long) façade
towards south. The window area towards south and west is 41 % re-
spectively, 14 % is against east and 5 % towards north. Therefore, the
heating demand will be approximately the same for the main façade to-

Windows and energy
131
wards both south and east orientation. Consequently, results for the main
façade towards west and north orientation will also be similar, Fig 5.17.
The higher heating demand for the northerly orientation, compared to
the southerly, corresponds to an increase of 6-7 % for all window types.
It is clear that the window U-value has a much larger impact on the
heating demand than the orientation, at least for this moderatley glazed
house.
Heating demand (kWh/yr)
0
2000
4000
6000
8000
10000
12000
14000
SENW
Main facade orientation
window 1
window 2
window 3
window 4
window 5
Stockholm 2000
Figure 5.17 Annual heating demand (kWh/yr) as a function of main façade
orientation. House Stockholm 2000.
5.1.6 Effects of site/climate
As the window choice may be affected by the local climate, some of the
previous simulations were redone for the climates of both Lund (lat 56°N)
and Luleå (lat 65°N), using the climate years of 1988.
First, the annual heating demand was calculated. The results for cer-
tain combinations of house insulations and window types are presented
in Table 5.6. It is clear that the total demand for heating increases quite
dramatically for northern cities in Sweden. Compared to Lund, the heat-
ing demand for Stockholm is about 25 % higher, and for Luleå it is about
70 % higher.

Energy-Efficient Window Systems
132
Table 5.6 Annual heating demand (kWh/yr,m2 floor area), as a function
of insulation level of house and window type.
Construction year of house, window type
Site 1960, #1 1980, #2 2000, #3 2020, #5
Lund 112.8 58.4 44.0 15.9
Stockholm 140.1 74.7 57.4 23.2
Luleå193.2 106.9 83.4 36.7
The incremental heating demand as a function of window U-value was
simulated using the same methodology as in Sec. 5.1.3 in order to deter-
mine which U-values are required in order not to increase the heating
demand compared to a house without windows. The results for Lund are
similar to Stockholm, but the milder climate is reflected as slightly higher
allowed U-values of windows, see Fig 5.18 and Table 5.7. In Luleå, the
winter is colder and longer than in both Stockholm and Lund. There-
fore, the maximum allowed U-values are even lower for Luleå, see Fig
5.19 and Table 5.7.
-100
-50
0
50
100
150
200
250
0 0.5 1 1.5 2 2.5 3
Window U-value, U
p,win
(W/m²K)
1960
1980
2000
2020
Incremental heating demand (kWh/yr,m² window)
Lund
Figure 5.18 Incremental heating demand (kWh/yr,m2 window area) as a func-
tion of window U-value and construction year of house. Simulations
performed for the Lund climate.

Windows and energy
133
-100
-50
0
50
100
150
200
250
00.511.522.53
Window U-value, U
p,win
(W/m²K)
1960
1980
2000
2020
Incremental heating demand (kWh/yr,m² window)
Luleå
Figure 5.19 Incremental heating demand (kWh/yr,m2 window area) as a func-
tion of window U-value and construction year of house. Simulations
performed for the Luleå climate.
Table 5.7 Maximum allowed window U-value (W/m2K) as a function of
construction year of house on the criterion that the windows
should not increase the total heating demand compared to an
opaque house.
Construction year of house
Site 1960 1980 2000 2020
Lund 1.76 1.33 1.14 0.76
Stockholm 1.56 1.18 1.00 0.65
Luleå1.36 1.05 0.91 0.57
A summary of potential overheating problems is given in Table 5.8.
Surprisingly, the overheating problems seem to increase as the house is
moved further north. This might be explained by the fact that the sum-
mer is short but very intense, and generally the solar radiation during the
summer months is higher in Luleå than in both Lund and Stockholm
(Wall, 1994). Further, the solar altitude is lower which gives a smaller
incidence angle to the window, and hence a larger part of the solar irra-
diation is transmitted. Since the house Luleå 2020 does not behave as
expected, the duration curve of the operative temperature was plotted for

Energy-Efficient Window Systems
134
this house and for the climates Lund and Stockholm as well, Fig 5.20.
Here it is clearly seen that the maximum operative temperature is higher
in Luleå, but the peak is much more sharp. These extremely high opera-
tive temperatures will in reality not be tolerated. People will provide for
ventilation by opening doors and windows and/or install some type of
shading device, e.g. an awning.
Table 5.8 Number of hours with an operative temperature above 27°C as
a function of climate and insulation level of house. Window
type 3.
Construction year of house
Site, climate year 1960 1980 2000 2020
Lund, 1988 122 281 598 2919
Stockholm, 1988 253 540 853 2974
Luleå, 1988 375 792 959 2021
Operative temperature (°C)
15
20
25
30
35
40
45
50
0 1000 2000 3000 4000 5000 6000 7000 8000
Hours above temperature
Lund 2020
Stockholm 2020
Luleå 2020
Figure 5.20 Global operative temperature (°C) of the ground floor (midpoint)
as a function of climate. Window type 3 in house 2020.
However, the effect seen for Luleå might be due to that the climate years
are not fully comparable concerning a few summer months, since they
were mainly selected to be typical over the whole year (Wall, 1994). For

Windows and energy
135
example, in both Lund and Stockholm, the solar radiation in May was
slightly above the 10-year period average from which the climate years
were chosen (1983-1992). The June values were average, and the July
values were lower than average. For Luleå however, the solar radiation
was average to high for the same months. Also in April the radiation was
higher than average in Luleå. However, the differences in solar radiation
and outdoor temperatures for the selected climate years have not been
studied in detail.
5.1.7 Effects of reduced emittance
The effect of varying emittance of the low-e coating was studied for the
case Stockholm 2000. This is interesting since the improved U-value is
accompanied by a lower g-value as well. Three similar double-pane win-
dows with argon (7, 8, 9) were compared, with the respective emittances
of 16 %, 10 % and 4 %. The resulting annual heating demands for the
three window types are given in Table 5.9. The net energy savings for
windows 8 and 9 (compared to window 7) were 101 and 240 kWh/yr
respectively.
Table 5.9 Annual heating demand (kWh/yr) and net energy savings (kWh/
yr), for window types 7-9 in the house Stockholm 2000.
Window type
Annual demand and saving (kWh/yr) (7) (8) (9)
Heating demand 9031 8930 8791
Net energy saving 0 101 240
As the net energy savings of windows 8 and 9 are hard to evaluate on
their own, the effects of the improved U-value and the reduced g-value
were studied separately. The U-value effect was studied by a new series of
simulations under conditions with solar radiation in the following way:
Two fictitious glazing types were constructed (8’ and 9’), with the same
emittances (i.e. U-values) as windows 8 and 9 respectively, but with the
g-value of glazing 7 (g = 0.71). The g-value effect was studied in the same
manner by creating two other fictitious glazings (8” and 9”) with the
same g-values as windows 8 and 9 respectively, but with the U-value of
glazing 7 (
ε
= 16 %). The performance of the real and the fictitious glazings
are given in Table 5.10, and the corresponding annual energy demands
and the net savings are found in Table 5.11. The U-value effect gives

Energy-Efficient Window Systems
136
savings of 239 and 499 kWh/yr for windows 8 and 9 respectively. The
energy savings due to the g-value effect are negative, -137 and -265 kWh/
yr respectively, which is due to the decreasing solar gains.
As the net energy saving is the sum of the U-value and g-value effect,
it should be equal to the first simulation given in Table 5.9, and this is
also roughly correct, Table 5.11.
Table 5.10 Thermal and solar transmittance of the original and fictitious
glazings.
Glazing type
(7) (8) (9) (8') (9') (8") (9")
Ucog (W/m2K) 1.5 1.35 1.18 1.35 1.18 1.5 1.5
g (-) 0.71 0.65 0.60 0.71 0.71 0.65 0.60
Table 5.11 Annual heating demand (kWh/yr) and expected energy savings
for (I) the U-value improvement alone (kWh/yr), and (II) for
the g-value reduction alone (kWh/yr). Net energy saving (kWh/
yr) for the combined effect of (I) and (II). Window types 7-9 in
the house Stockholm 2000.
Glazing type
I) U-value effect, Equal g-value (7) (8’) (9’)
Heating demand 9031 8792 8532
Expected saving 0 239 499
II) g-value effect, Equal U-value (7) (8”)(9”)
Heating demand 9031 9168 9296
Expected saving 0 -137 -265
Net energy saving (7) (8) (9)
(I+II) 0 102 234
Finally, a comparison with the ‘Karlsson’ window formulae, outlined in
the beginning of this chapter, was performed. First, the degree-hours G
for this house and climate were determined, using the approach given in
Eq 5.1. In order to do this, the balance temperature of the building must
be determined. This was done by plotting the simulated hourly heating
demand against the outdoor temperature for the house 2000 (with win-
dow 7), see Fig 5.21. The balance temperature 13.6°C was found from
were the regression line crosses the x-axis, and the corresponding degree-
hours G are 113.4 k°Ch.

Windows and energy
137
Stockholm 2000, window 7
y = -135.29x + 1845.2
R
2
= 0.608
0
500
1000
1500
2000
2500
3000
3500
4000
4500
-20-100 102030
Outdoor temperature (°C)
Heating power demand (Wh/h)
Figure 5.21 Heating power demand (Wh/h) for the house Stockholm 2000 as a
function of the outdoor temperature. Window type 7. All hours with
zero heating demand have been eliminated from the graph.
The hourly solar radiation was equally summed up to the balance tem-
perature tb, which gives S(13.6)=274 kWh/m2,yr. This was easy to do,
since the total solar radiation on the exterior side of transparent surfaces
is given in the results-file of the Derob simulation. This value of S in-
cludes some shading of both the roof and the reveal, since a small reveal
was modelled in the Derob simulations (protruding 10 cm out from the
façade at 10 cm distance from the glazing edge), see Fig 5.1. It is also an
average for all window orientations, in this particular case south = 41 %,
west = 41 %, east = 14 %, and north = 5 %. The energy savings that can be
expected from the improved U-value,
∆
Qloss, and from the solar gains,
∆
Qsolar , were then determined as:
*$84
JODVVFRJORVV
⋅⋅∆=∆ (5.3a)
6$J4
JODVVVRODU
⋅⋅
∆
=
∆
(5.3b)
The value of the solar transmittance for the windows was taken as the g-
value for 55° incidence angle. The respective g(55°) values for glazings 7-
9 were thus approximately determined to 0.63, 0.58 and 0.53. However,
since this is only an approximate method, the values for normal inci-

Energy-Efficient Window Systems
138
dence could also have been used, since the g-value difference between the
glazings are similar for the two angles. The glazing area was 15.5 m2. The
resulting savings are shown in Table 5.12.
Table 5.12 Expected net energy savings for the ‘Karlsson’ window formu-
lae with the effects of U-value improvement and g-value reduc-
tion separated. Window types 7-9 in the house Stockholm 2000.
Window type
Expected saving, Karlsson’s method (7) (8) (9)
U-value effect,
∆
Qloss (G=113.4 k°Ch) 0 263 561
g-value effect,
∆
Qsolar (S=274.3 kWh/m2,yr) 0 -212 -424
Net energy saving 0 51 137
The discrepancy between the dynamic computer simulations and the
Karlsson method is rather small regarding the incremental heat losses
(around 10 %, compare Tables 5.11 and 5.12). In the Derob-LTH simu-
lations, the U-values are temperature dependent (see chapter 3), and the
dynamic effects of varying temperatures and internal gains are all ac-
counted for, the building never reaches steady-state conditions. The solar
radiation is however the part that varies the most. The reduction of the
solar gains using the degree-hour method is 55-60 % larger compared to
the estimations using Derob-LTH. Here, the choice of the balance tem-
perature seems to be very critical, and a lower balance temperature should
be applied. However, the balance temperature that approximately matches
the Derob-results regarding the insolation is as low as 7°C. All in all, the
net energy savings using the Karlsson method are thus only about half of
those estimated by Derob-LTH. Since the Karlsson formulae is an ap-
proximate method, these kinds of discrepancies are to be expected.
5.2 Single-person office room
In article II is presented a calculation example of a single-person office
room with the glazing-to-wall area ratio (GWAR) varying from 0 to 50 %.
It was demonstrated that the cooling demand increases more rapidly than
the heating demand as the GWAR increases. This will be made clearer in
the following section, where some further calculations based on the same
example are made. The geometry of the office space is again repeated
here, Fig. 5.22.

Windows and energy
139
2.9 m
2.7 m
4.2 m
GWAR 0 %
30%
10% 20% 50%
Figure 5.22 The geometry of the office model with Glazing-to-Wall Area Ratios
of 0, 10, 20, 30 and 50 %.
In the examples below, the incremental heating and cooling demands are
calculated for both a south and north facing office room for five glazing
types. Four of them are identical to those in article II. A fifth glazing type
is added which is a triple glazing with one low-e coating and argon, iden-
tical to that of window type 3 in the residential example above, see Table
5.13. The incremental energy demand is here defined from the non-glazed
wall (GWAR 0 %), and presented as the annual demand per square meter
of floor area.
Table 5.13 Window alternatives, thermal and optical performance. The total
window U-value applies to a window 1.1 × 1.3 m, used in the
case GWAR 30 %. Glass distance 12 mm. Ucog calculated at 0/
20°C, 5 m/s in Window 4.1. Coating position is counted from
the outermost glass surface and inwards.
Glazing type No. of Coating Gas Ucog UfUwin gT
vis
panes position filling (W/m2K) (-) (%)
double, clear 2 - Air 2.79 2.16 2.61 0.75 80
triple, clear 3 - Air 1.85 2.16 1.94 0.66 72
triple, e10% 3 5 Air 1.32 2.16 1.56 0.57 69
triple, e10%+Ar 3 5 Ar 1.12 2.16 1.42 0.57 69
triple, 2e10%+2Kr 3 2, 5 Kr 0.62 1.70 0.94 0.50 66

Energy-Efficient Window Systems
140
For the south facing office, the incremental heating demand varies sig-
nificantly with both GWAR and window U-value, Fig 5.23. As expected,
the clear double-pane window displays a significant increase in heating
demand as the quite well-insulated wall is replaced by a larger and larger
window of a significantly higher U-value. The clear triple-pane window
displays a moderate increase, while the two low-e triple-pane windows
are hardly affected by changes to window area. It is again clearly shown
that the superwindow will gain energy over the year, as demonstrated by
the decreased heating demand for large GWARs.
While the incremental heating demand varies significantly with win-
dow U-value, the incremental cooling demand is almost equal for all
window types, Fig 5.24. This is explained by the relatively small differ-
ence in g-value between the cases. The clear double-pane window has the
highest g-value, and consequently displays the highest increase in cooling
demand. However, the superwindow does not show the lowest increase
in cooling demand, even if its g-value is lowest in the group. Since its U-
value is also very low, less energy is also lost to the outside. This example
demonstrates that there can be a fine balance between solar gains and
thermal losses.
Incr. Heating Demand (kWh/yr,m² floor area)
-20
-10
0
10
20
30
40
50
0 1020304050
Glazing-to-Wall Area Ratio (%)
Double, clear
Triple, clear
Triple, low-e
Triple, low-e,
Ar
Triple, 2 low-
e, 2 Kr
South
Figure 5.23 Incremental heating demand (kWh/yr,m2 floor area) as a function
of GWAR and window type for a south facing office in Lund.

Windows and energy
141
Incr. Cooling Demand (kWh/yr,m² floor area)
0
10
20
30
40
50
0 1020304050
Glazing-to-Wall Area Ratio (%)
Double, clear
Triple, clear
Triple, low-e
Triple, low-e,
Ar
Triple, 2 low-
e, 2 Kr
South
Figure 5.24 Incremental cooling demand (kWh/yr,m2 floor area) as a function
of GWAR and window type for a south facing office in Lund
For a north facing office room, the incremental heating demand is of
course very large for the clear, double-glazing. Only the superwindow
shows a moderate increase in heating demand as the window area is en-
larged, Fig 5.25. One can also see that the superwindow towards north
behaves similar to the clear triple-window towards south, compare Figs.
5.23 and 5.25.
Although small, there does exist a cooling demand for north facing
offices. The incremental cooling demand is similar for all glazing types,
Fig 5.26.

Energy-Efficient Window Systems
142
Incr. Heating Demand (kWh/yr,m² floor area)
0
10
20
30
40
50
60
70
0 1020304050
Glazing-to-Wall Area Ratio (%)
Double, clear
Triple, clear
Triple, low-e
Triple, low-e,
Ar
Triple, 2 low-e,
2 Kr
North
Figure 5.25 Incremental heating demand (kWh/yr,m2 floor area) as a function
of GWAR and window type for a north facing office in Lund.
Incr. Cooling Demand (kWh/yr,m² floor area)
0
10
20
30
40
50
0 1020304050
Glazing-to-Wall Area Ratio (%)
Double, clear
Triple, clear
Triple, low-e
Triple, low-e,
Ar
Triple, 2 low-
e, Kr
North
Figure 5.26 Incremental cooling demand (kWh/yr,m2 floor area) as a function
of GWAR and window type for a north facing office in Lund.

Windows and energy
143
5.2.1 Electric lighting savings through daylight
utilisation
The effect of daylight utilisation (see Sec 4.5) on heating and cooling
demands was studied in article II. This required that simulations of the
electric lighting use were performed. The simulations were not described
in article II, but are presented here. The lighting simulations were per-
formed using Superlite and Superlink in the ADELINE 2.0 program pack-
age (Erhorn & Stoffel, 1997). Thereby, output files with hourly power
needs for the lighting were created, using hourly values of sunshine prob-
ability based on the Lund climate.
The sunshine probability, SSP, is an attempt to link the actual weather
pattern to the set of standard skies used in Superlite (i.e. CIE overcast sky,
and CIE clear sky with and without sun), see Szerman (1994). Since the
Superlink program had a problem with the format of the Swedish weather
file (for direct normal radiation), the SSP was derived from the weather
file in the following manner:
SSP(i) = IN(i)/IN,max 0 ≤ SSP ≤ 1 (5.4)
where SSP(i) is the percentage of sunshine during hour i, IN(i) is the
direct solar radiation for the ith hour of the year, obtained from the weather
data, and IN,max is the maximum possible direct radiation during the same
hour (assuming a perfectly clear sky). A routine for generating the SSP-
values from the weather file was built which involved calculating solar
angles, and the direct radiation through Linke turbidities, relative opti-
cal air mass and optical thickness of the atmosphere according to a model
described by Grenier et al. (1994).
Electric lighting savings were calculated for two lighting installations
of different efficiency: (1) high efficiency with a lighting power density of
10 W/m2 (used in article II), and (2) low efficiency with a lighting power
density of 18 W/m2. In both cases it was assumed that the installation
would provide 500 lux in the midpoint of the room at desk level.
When a continuous dimming strategy is used, the lighting is dimmed
exactly to the amount of available daylight. In Superlink, the power de-
mand of the lighting is reduced in direct proportion to the light output
(one to one), i. e when the light output goes from 100 to 0 %, the power
demand goes from 100 to 0 %. This is not correct and has been described
by Christoffersen (1995). Reality, is not so ideal: the lights are designed
to be most efficient (highest lumens per Watt) at full power. For a typical
fluorescent light source with electronic ballast, the power demand goes
from 100 to about 30 % when the lighting is dimmed from 100 to 10 %.

Energy-Efficient Window Systems
144
The power demand for the lighting was therefore recalculated according
to the following formula, adopted from specifications by Philips Light-
ing (1997), before creating internal load files for Derob-LTH:
for Pi / Pmax ≥ 0.10: Preal,i = (0.75⋅Pi / Pmax + 0.25)⋅Pmax (5.5a)
for Pi / Pmax < 0.10: Preal,i = 0.325⋅ Pmax (5.5b)
where Pi is the lighting load for the ith hour according to Superlink (Wh/
h), Pmax is equal to the installed power (W), and Preal,i is the recalculated
(realistic) lighting load for the same hour (Wh/h). This means that the
lowest lighting load (base load) during working hours was 32.5 % of the
installed power. The results calculated according to Eq. 5.5 are called
“realistic” dimming.
In the original Superlink results the base load was close to zero. In
order to see how much this would affect the calculated lighting savings
the original loads in Superlink were also evaluated, and the results are
referred to as ”ideal” dimming. Additional control strategies, e.g. switch-
ing off the lighting when it has been fully dimmed for some time inter-
val, were not applied. All calculations were performed for the four major
orientations (S, N, E, W) and the resulting savings shown in Fig 5.27 are
the average values for the four orientations.
The savings in Fig. 5.27 should be compared to the annual electricity
use for lighting without dimming which was 23 kWh/m2,yr and 43 kWh/
m2,yr for the two lighting power densities 10 and 18 W/m2 respectively.
Naturally, the largest savings occur for the largest windows. For GWAR
50 % they amount to 13 kWh/m2,yr for the high-efficiency lighting and
to 24 kWh/m2,yr for the low-efficiency alternative, which in both cases
corresponds to a saving of approx. 55 %. However, the potential error of
not recalculating the lighting load (the ideal dimming cases) would have
lead to estimated savings of almost 80 %, which is significantly higher
than for the realistic cases.
The difference in savings with respect to window type is however al-
most negligible: The double, clear window yield at most 10 % higher
savings, and never more than 2.5 kWh/m2,yr compared to the super-
window.
It should be emphasised that all savings are optimistic estimates since
they are compared to having the lighting on all day. In a real case, the
lighting will sometimes be switched off.

Windows and energy
145
Annual lighting energy savings (kWh/yr,m² floor area)
0
5
10
15
20
25
30
35
40
0 1020304050
Glazing-to-Wall Area Ratio (%)
Double, clear, ideal
Double, clear, realistic
Superwindow, realistic
18 W/m²
10 W/m²
Figure 5.27 Annual electric lighting savings (kWh/yr,m2 floor area) calculated
in Superlite/Superlink as a function of Glazing-to-Wall Area Ratio
for two lighting power densities (10 and 18 W/m2) and two glaz-
ing types (double, clear and superwindows). Two cases of continu-
ous dimming are shown: realistic and ideal. Average of S, N, E,
and W orientations for Lund 1988.
5.3 Office space fully glazed on three sides
In the previous sections only clear windows and windows with dedicated
low-e coatings were compared. However, for offices the current trend is
to use very large glazed surfaces, up to 100 % of the wall area. This poses
a potentially huge stress on the cooling system if no solar control meas-
ures are taken. Therefore, for such buildings the term energy-efficient
takes on a new meaning: energy-efficiency is now achieved by a low g-
value to reduce or totally avoid the need for air-conditioning during the
summer. However, a low U-value is also necessary to prevent bad thermal
comfort during the dark and cold winter period, since a wall U-value of
around 0.2 W/m2K is replaced by a much higher U-value of the glass,
normally on the order of 1.0-1.2 W/m2K.
A low g-value can be achieved through the use of solar control glass,
shading devices or a combination of both. Exterior devices have a much
larger potential to reduce solar gains since the absorbed heat is dissipated

Energy-Efficient Window Systems
146
to the ambient air. However, these solutions are sometimes not wanted
due to various local factors, e.g. maintenance, wind sensitivity, aesthetics
etc. Nevertheless, some type of shading device is always needed, if not for
solar control, at least to control glare, see Chapter 4.
Below is presented an example of an office space in Lund glazed on
three sides; towards east, west and north. The space is located on the top
floor and in the north-west corner of a larger office building. The east
wall faces a courtyard, and is therefore partly shaded during the morning
hours, see Fig 5.28. The dimensions of the space are 4.2 by 7.8 m and the
ceiling height is 3.2 meter. The intended use of this space is as a confer-
ence room for 12 people, but it could also be used as an office space for
two people. The assumed internal loads and ventilation/infiltration rates
are given in Table 5.14. The inlet temperature of the air was constant at
17°C.
Figure 5.28 Solar views of the modelled space, fully glazed towards east, west
and north, and a neighbouring office space. Seen from the direction
of the sun in the morning (6:30) and afternoon (17:30), true solar
time.
Table 5.14 Assumed internal loads (W) and ventilation/ infiltration rates
during summer and winter periods.
12 people 2 people
summer winter summer winter remark
hour 1/4-30/9 1/10-31/3 1/4-30/9 1/10-31/3
00-07 0 0 0 0
07-08 500 700 500 700 lighting
08-18 1700 1900 700 900 lights + people
18-19 500 700 500 700 lighting
19-24 0 0 0 0
ventilation 144 144 40 40 (l/s)
infiltration 0.25 0.25 0.25 0.25 (ach)

Windows and energy
147
A clear triple glazing (#1) was selected as a base case for the calculations.
This was then compared to three other glazing alternatives: (#2) a dou-
ble-pane glazing with an advanced solar control coating with high visual
transmittance; (#3) a double-pane, advanced solar control glazing with
slightly lower visual and solar transmittance than #2, but with a similar
U-value; and (#4) where a third, low-e, pane was added to #2 in order to
obtain a lower U-value. The glazing data are given in Table 5.15. In an
additional case (#2+screen), an exterior solar screen with a transmittance
of 27 % and absorptance of 19 % was applied to glazing alternative 2 to
study the efficiency of exterior solar shading devices.
Table 5.15 Glazing alternatives, thermal and optical performance.
Glazing Description No. of Coating Gas Gap width Ucog gT
vis
panes position filling (mm) (W/m2K) (-) (%)
#1 clear glass 3 2 Air 12 1.65 0.68 72
#2 solar A 2 2 Ar 15 1.14 0.36 64
#3 solar B 2 2 Ar 15 1.12 0.27 51
#4 solar A+le4% 3 2, 5 Ar 12 0.67 0.21 43
Peak cooling load (W) (12 people)
0
1000
2000
3000
4000
5000
6000
7000
1 3 5 7 9 11 13 15 17 19 21 23
Hour
#1
#2
#3
#4
#2 +
screen
Figure 5.29 Peak cooling load as a function of glazing type and exterior shading
for internal loads and ventilation corresponding to 12 people. Fic-
titious climate with maximum outdoor temperature of 28°C.

Energy-Efficient Window Systems
148
The peak cooling load was calculated for a fictitious climate corresponding
to a maximum outdoor temperature of 28°C, sinking to 17°C at night,
and a rather intense solar radiation for the latitude in Lund. (IN = 906,
IdH = 116 W/m2, solar height of June 11). The space was cooled to 25°C.
Fig. 5.29 shows that there is a small morning peak, due to the partly
shaded east façade, and a larger afternoon peak due to the unobstructed
view of the west façade. The triple, clear glazing (#1) gives rise to a very
high peak load of 6.6 kW, which will be almost impossible to get rid of
with traditional cooling baffles, since the space is rather small (33 m2).
However, glazings #2 and #4 reduces the peak load by around 40 % (to
4.1 and 4.0 kW), and with glazing #3, the cooling load is halved to 3.2
kW. In the additional case #2+screen, an exterior fabric screen was ap-
plied on the west façade only. This reduces the afternoon peak by almost
75 % (to 1.8 kW) compared to clear triple glazing.
The annual energy demand was calculated for the Lund 1988 climate
with internal loads and ventilation corresponding to two persons. Since
this space may not be fully occupied as a conference room every day of
the year, this seemed like a more conservative assumption. The thermo-
stat set-points were 20°C for heating and again 25°C for cooling. As Fig.
5.30 shows, the annual heating demand depends mainly on the window
U-value. It is highest for glazing 1, glazings 2 and 3 (with similar U-
values) give similar results, while the better window 4 gives a significantly
lower heating demand. This is also demonstrated in Fig. 5.31, which
shows that the annual heating demand depends linearly of the U-value.
Glazing 2 has a slightly higher heating demand with the screen than with-
out, since the screen was applied throughout the year (i.e. lower benefi-
cial solar gain).
Fig. 5.29 demonstrated that the cooling demand depends mostly on
the g-value, however, the relationship is not perfectly linear, Fig 5.32.
Although glazing 4 has a low g-value, it also has a lower U-value than the
other glazings. This prolongs the cooling season, and the peak load be-
comes similar to glazing 2, which has a much higher g-value.
To the calculated demands for space heating and cooling, the energy
for keeping the inlet air constant at 17°C should be added. This requires
235 kWh/yr,m2 floor area of heating, and 7 kWh/yr,m2 of cooling en-
ergy. However, these demands will normally be cut down via heat recov-
ery systems in the HVAC system.

Windows and energy
149
Energy demand (kWh/month) (2 people)
0
200
400
600
800
1000
1200
1400
janfebmaraprmayjunejulyaugsepoctnovdec
Month
#1
#2
#3
#4
heating
coolin
g
Figure 5.30 Monthly demand for heating and cooling (kWh/month) as a func-
tion of glazing type. Lund 1988.
Heating demand (kWh/yr,m² floor area)
0
50
100
150
200
250
00.511.52
Glazing U-value (W/m²K)
#1
#2
#3
#2+screen
#4
Figure 5.31 Annual heating demand (kWh/yr,m2 floor area) as a function of
glazing U-value.

Energy-Efficient Window Systems
150
Cooling demand (kWh/yr,m² floor area)
0
20
40
60
80
100
120
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Glazing g-value (-)
#2+screen
#3
#4
#2
#1
Figure 5.32 Annual cooling demand (kWh/yr,m2 floor area) as a function of
glazing g-value.
The resulting thermal comfort in the space is demonstrated in the fol-
lowing graphs, which were produced by the Derob-LTH post-processor
COMFORT, Figs 5.33-35. They show the global operative temperature at
1.2 m for (1) a summer case (+19°C, and intense solar radiation during
an afternoon hour) and (2) a winter case (-5°C and no solar radiation).
The winter case does not include any radiative heating, as all the energy
for heating is supplied directly to the air.
In the summer case (Figs. 5.33-34) the operative temperature of the
space is uncomfortably high during a clear sunny afternoon, even if the
air temperature is cooled to 25°C. This is because the internal glass sur-
faces become very warm. A low g-value of the window system is essential
during sunny conditions. In the first glazing alternative (#1) (Fig. 5.34,
top), the inner side of the sunlit pane is 39°C. It is reduced slightly to
34°C for glazing 2, to 32°C for alternative 3, but increases to 36°C for
glazing 4. Only the case 2 + screen gives a surface temperature similar to
the non-sunlit glass surfaces, 28°C, Fig. 5.33. This demonstrates the great
difficulty of achieving a good thermal comfort during summer periods
with intense solar radiation.
In the winter case (-5°C), the triple clear glazing will yield an unsatis-
factory thermal environment, since the surface temperature of the glaz-
ing will be low (12°C), and the average operative temperature is only

Windows and energy
151
17°C, Fig 5.35 (top). With glazing 2, a slightly better thermal environ-
ment is achieved, but still the operative temperature is low, on average
18°C, Fig 5.35 (middle). The corresponding surface temperature of the
glazing is about 14°C. In order to achieve an operative temperature of
above 18°C, a lower U-value is needed. This is done by glazing 4, where
the operative temperature is about 19°C, and the glazing surface is about
16°C, Fig 5.35 (bottom).
The choice of the outside temperature (-5°C) is very modest, but on
the other hand it is a temperature that occurs quite frequently in Lund.
According to the building code (BBR 99), the operative temperature
should be calculated for a dimensioning outdoor temperature DUT20
that statistically only occurs once every 20 years. The requirement is that
the operative temperature shall not go below 18°C for certain points in
the room as specified in the code (e.g. 1 m from the window). DUT20 is
determined from the normal average temperature of the month of Janu-
ary, and from the time constant of the building according to Swedish
Standard SS 02 43 10. In January, the average outdoor temperature is
0.7°C in Lund, -3.5°C in Stockholm and –10°C in Luleå. The corre-
sponding values for DUT20 for a building with a short time constant (25
h) becomes –11°C for Lund, –19°C for Stockholm and –28°C for Luleå.
For such low outside temperatures, it will be even more difficult to com-
ply with the requirements in the building code for fully glazed buildings.
All in all, this example demonstrates that fully glazed buildings are very
dependent on the outside environment, and it can become very costly to
achieve a good thermal environment throughout the year.
Figure 5.33 Global operative temperature at an outside temperature of 19°C, a
direct normal radiation of 905 W/m2, and a diffuse radiation of
68 W/m2, June 11th at 16.00 hrs., for window type 2 with an
exterior screen towards west.

Energy-Efficient Window Systems
152
Figure 5.34 Global operative temperature at an outside temperature of 19°C, a
direct normal radiation of 905 W/m2, and a diffuse radiation of
68 W/m2, June 11th at 16.00 hrs., for three glazing g-values: 0.68
(top), 0.36 (middle) and 0.27 (bottom).

Windows and energy
153
Figure 5.35 Global operative temperature at an outside temperature of –5°C,
no solar radiation, for three glazing U-values: 1.65 (top), 1.14
(middle) and 0.67 W/m2K (bottom).

Energy-Efficient Window Systems
154
To summarize this chapter, we have seen that the window has a signifi-
cant impact on the energy balance of a building. The residential example
demonstrates that when the total window U-value is reduced from 2.5 to
1.0 W/m2K, the annual heating demand is reduced by 100 kWh/yr,m2
window area in Lund, by 125 kWh/yr,m2 in Stockholm and by 170 kWh/
yr,m2 in Luleå, and these savings are independent of the building type
(e.g. insulation, ventilation rates and internal gains). However, they de-
pend on the actual distribution of window areas in different orientations.
The simulations performed for the house in Stockholm with the main
façade towards north instead of south showed that the annual heating
demand will decrease by 140 kWh/yr,m2 window area for the same U-value
reduction as above (from 2.5 to 1.0). The larger saving towards north is
explained by the lower solar gains, which points to the well-known fact
that windows towards north should have a lower U-value. The potential
savings by using more energy-efficient windows are thus not only de-
pendent on the U-value. The lower solar gain that follows from the lower
g-value may approximately half the potential savings due to the improved
U-value alone. Further, for extremely well-insulated buildings, the re-
duced g-value is beneficial, since the solar radiation during the summer
may quickly create a severe overheating problem.
In the second example of a single-person office room, the effect of
increasing window areas as well as the orientation effect was clearly dem-
onstrated. However, the super-insulated windows gained energy over the
year towards south, without simultaneously increasing the cooling de-
mand in the summer. Finally, the fully glazed building showed that it is
very difficult to establish a constant indoor temperature in such build-
ings, it swings between being either too cold or too hot.

Conclusions and recommendations for further research
155
6 Conclusions and
recommendations for
further research
What is the optimum window? The truth is that the best window will
always depend on the situation. Each site has a different climate, and
there is for example a large difference between northern and southern
Sweden. Each building has a certain insulation level, and internal load.
The orientation of windows is also important, especially for poorly to
medium insulated houses. Due to solar gains, the window may have a
higher U-value than the wall without increasing the annual heating load.
The maximum permissible U-value will thus depend on the general insu-
lation of the house, the climate and the window orientation etc. In south-
erly unobstructed locations, triple-pane clear glazing may – on an annual
heating basis – perform on par with super-insulated windows towards
north. Further, since the U-value reductions are usually accompanied by
lower solar gains (e.g. g-values) it is not certain that the window with the
lowest U-value will always the best. However, for extremely well-insu-
lated buildings, the U-value should always be low, since the heating sea-
son is very short, and useful solar gain is small.
Thermal comfort problem will arise when windows are large, both
during summer and winter. In such situations a low U-value is required
to solve the winter problem, and a low g-value together with shading
devices will solve the summer problem. The daylight issue should also be
considered. Perhaps one low-e coating is sufficient in many cases, since
two coatings will tint the light more than one. Which compromise is
made in each situation is eventually up for the client to make. Tools are
therefore needed to provide realistic data on the predicted behaviour of a
planned building. To integrate daylight tools into thermal programs is
thus one attempt to try to interlink the two sides of the same coin: the
energetic side and the daylight side.

Energy-Efficient Window Systems
156
6.1 Technology status of windows
Triple-pane clear windows have for many years been the de facto stand-
ard in Sweden. They were introduced on a large scale at the end of the
1970s, when a strict energy code was enforced in the wake of the 1973 oil
crisis. The code required a window U-value of 2 W/m2K, and since low-
emittance coated glass was not available at that time, the triple-pane win-
dow was the only solution.
During the last few years there has been a dramatic increase in the use
of low-e coatings. The NUTEK procurement program for energy-effi-
cient windows in 1992 was partly responsible for this. But especially in
1998 when a large window manufacturer changed the production into
using low-e coated glass and argon (in triple-pane glazing) as the stand-
ard product, the market share took a marked jump upwards. Today, about
45 % of all insulating glass units (IGU) are equipped with coated glass,
and an estimation is that the average U-value for windows sold today in
Sweden is between 1.3 and 1.6 W/m2K. However, the best windows on
today’s market have a total U-value of 0.7-1.0 W/m2K.
The emittance of commercial low-e coated glass has decreased in re-
cent years, and today soft silver-based coatings have an emittance of around
4 %. When used in a double-pane unit together with argon gas, and an
optimised gap width of 15 mm, the U-value of the glazing can be as low
as 1.1. It is then possible to achieve a window with a centre-of-glass U-
value better than triple clear glazing and similar to that of triple low-e
coated glazing. Although the market share for double-pane windows is
still small, we might see more and more of them in the future, since the
cost is lower. However, the centre-of-glass U-value of optimised double-
pane glazing is very sensitive to varying inside and outside temperatures
(boundary conditions), and may differ by as much as 0.2 W/m2K for a
temperature difference across the glazing of 15 and 30°C respectively. It
is also more difficult to achieve a low total window U-value than in for
example a 1+2 construction, since the frame losses are more pronounced.
If, from the national perspective, further improvements to window U-
values are requested this will not be possible if double-pane low-e coated
windows are introduced on a large scale.
New spacers, so called warm-edge technologies, can reduce the ther-
mal losses around the perimeter of the IGU, but their market share is still
too small to be seen in the statistics. The extra cost (about 150 SEK/
window, consumer price) is still holding back the development. These
spacers will reduce the total window U-value somewhat, but especially

Conclusions and recommendations for further research
157
the risk of condensation at the bottom edge of the glazing. In double-
pane windows the extra cost is compensated by a lower glass cost, and the
market share may increase quickly in the near future.
There are many performance requirements that a window has to fulfil,
for example on thermal insulation, sound insulation, water and moisture
penetration, ventilation, durability, mechanical strength and rigidity, etc.
The main function is however to provide for daylight and view. As the
energy-efficiency of windows is improved, the daylight transmittance will
be reduced, and the transmitted daylight spectrum might become more
and more distorted (like putting a selective filter on the glass). This is
since the improved U-value is achieved by using three or four panes, and
by using one or several low-e coated panes. These two goals may however
come into conflict with each other.
6.2 Energy-efficient windows:
the compromise between energy
demand and daylight quality?
This work has demonstrated that low-e coatings have a large potential for
energy-efficiency improvements in windows, especially in combination
with argon or krypton gas fillings. However, it is clear that at some point
there has to be made a compromise between daylight quality and heating
energy savings.
From the energy point of view, decreasing emittance is accompanied
by reduced g-value and solar gain. The potential saving is thereby ap-
proximately halved compared to the U-value reduction alone. In normal
residential housing, solar gains provide valuable contributions to the heat-
ing demand, and for southerly orientations especially, the main goal is
thus not to obtain the lowest possible U-value.
For extremely well-insulated residential buildings the heating season
is dramatically shortened. During the dark Swedish winter months, there
is practically no solar radiation of significance available, therefore a mini-
mum U-value should be strived for. Due to the short heating season, the
solar contribution quickly leads to an overheating problem during spring
and autumn. Thus, a low g-value may even be desirable.
From the daylight point of view, my calculations show that the differ-
ence in daylight transmittance, and the potentially larger lighting elec-
tricity use with darker windows, is more than outweighed by the savings
made on the heating energy. Super-insulated low-e windows thus seem to
have only desirable effects, since the lower daylight transmittance can be

Energy-Efficient Window Systems
158
compensated by using larger windows without large sacrifices on the heat-
ing bill. However, the reported change to the daylight spectrum is more
serious, with its potential effect on colour perception, colour rendering,
satisfaction and perhaps even well-being. Larger windows can not com-
pensate this effect, since it depends on the characteristics of the glass
combination itself, which acts as a filter. Since these windows typically
use two silver-based low-e coatings, which already have a transmittance
peak in the green area, this effect becomes more pronounced. The use of
more than one coating in a window can therefore be questioned and
should be further studied. Applying coatings to glass with a low iron
content has been discussed as one possible solution, but has not been
evaluated in this work.
6.3 Shading devices
Energy efficiency is not only about achieving low U-values. In modern
office buildings, especially today with the quickly growing trend of fully-
glazed facades, the cooling demand can be significant also in Swedish
climates. Solar gains then become a nuisance for the employee and a
stress on the cooling system. Even if all sunlight should not be removed,
glare and overheating should be avoided. It is important that the g-value
of the whole system is low, either through solar control glazing or by
using shading devices or a combination of both. Some type of shading
device will always be necessary to control glare, but placed on the outside
it will have a much larger potential to reduce cooling loads since the
absorbed heat will be dissipated to the outside air.
Whatever system is chosen, it must be designed to also meet the needs
of the employees, e.g. reduce the solar radiation, provide view, reduce
glare, be easily manoeuvrable, allow for individual control etc. The im-
pression so far is that visual criteria (e.g. disability or discomfort glare)
are the main motivation for individual to use shading devices, at least in
offices, while thermal criteria are of a second concern. However, the visual
criteria that decide when shading devices in offices should be used have
not yet been found.

Conclusions and recommendations for further research
159
6.4 Thermal comfort
With a dynamic energy simulation tool like Derob-LTH, it is possible to
study the full interaction between the window, the space, the ventilation
and internal loads generated by the occupants of the space. Especially the
effects on indoor temperatures, thermal comfort, and peak loads can be
studied during any climatic condition. This is the major advantage of a
dynamic tool compared to steady-state programs like the BKL-method
and for example degree-hour based methods for windows both of which
take solar gains into account.
It is quite clear that energy-efficient windows not only brings a lower
heating bill, but also provides a better thermal comfort during the winter,
since the surface temperature of the glass increases by several degrees.
With energy-efficient windows it has been demonstrated that radiative
heating is no longer necessary under windows to avoid cold draughts.
The glazing areas can also be made larger, and again we might see in-
creased window areas in residential buildings, and more daylight in build-
ings.
However, when the glazing areas increase drastically, like in the exam-
ple with the office fully-glazed on three sides, thermal comfort problems
during the winter (as well as summer) may easily arise. Even if modern
solar control glazing combines several advantageous factors (low emittance,
low g-value and rather high visual transmittance), the U-value of the win-
dow walls will be much to high to achieve a pleasant operative tempera-
ture. This is easy to understand when we consider that a well-insulated
wall (U-value around 0.2) is replaced with glazing of much higher U-
value (perhaps around 1.1-1.3). The window wall must either have an
even lower U-value (can be achieved with coated triple-glazing), or some
other measures must be taken to heat the internal glass surfaces (e.g. elec-
trically heated glass, radiative/convective heating, forced convective heat-
ing). In order not to increase the thermal losses, the best solution, (except
for decreasing the glazed area), is of course to use a very low U-value of
the glazing.
6.5 Tools for daylight calculation
We already have good tools to estimate the thermal behaviour of existing
and planned buildings. Examples demonstrated in this thesis have shown
what kind of information can be gained through a dynamic (hourly)
simulation tool like Derob-LTH. The attempt to build a daylight mod-
ule to this program is a natural step, since half of the solar radiation lies

Energy-Efficient Window Systems
160
within the visible range. The program Derob-LTH is well suited for inte-
grating a daylight calculation routine, since it already calculates the dis-
tribution of solar radiation using view factors (i.e. a radiosity approach).
The proposed method uses the view factor calculation in the existing
post-processor COMFORT (Källblad, 1996). The radiation intensity
(direct and diffuse components) is then translated to visual radiation via
the luminous efficacy. Last, the radiation levels are amplified using the
ratio of visual-to-solar transmittance to yield the final illuminance level.
Validation was performed for overcast and clear days against mainly Ra-
diance for a side-lit room and for a simple atrium. For the overcast sky,
the accuracy is acceptable for both vertical and horizontal windows for
the midpoint of the room. For the sunny sky, Derob-LTH accurately
predicts the size and illuminance level of the sunpatch, at least for the
tested vertical window. For the purpose of using it for daylight-linked
control of shading devices, the accuracy of the developed model seems
sufficient.
6.6 Further research
As the technological development never stops, there will always be a need
for further research to analyse the consequences of the new technologies.
Several good computer tools are already available for studies on both
the thermal and the daylight environment. Here, new demands from the
users are driving the development further. One example is the daylight
module in Derob-LTH, which was developed within the Solar Shading
Project and validated by the author. This module still needs further de-
velopment work to become a really useful tool, for example regarding
calculation of luminance. This requires that the code in the underlying
Derob-LTH calculation engine is further developed regarding the trans-
mittance, and the number of nodes per surface, of diffuse radiation. More
work is also needed to be able to allow for moveable shading devices, and
to be able to study the effects of various control strategies.
Regarding the daylight issues, there is certainly room for more re-
search. Studies have already indicated the numerous effects of daylight,
e.g. on perception, mood, and health, the so-called non-visual effects of
light. The perception of a space with tinted glass, especially with modern
solar control glass, and other low transmission glass would be interesting
to study in the field. Also, the criteria underlying visual comfort in daylit
spaces are not yet fully understood. If these were found, it would facili-
tate the development of control strategies for both daylight-linked light-
ing and shading systems.

Summary
161
Summary
This thesis deals with energy-efficiency in buildings and especially with
the role of windows. The main focus is on the following two, sometimes
conflicting, goals:
1) to provide for a good thermal protection against the outdoor envi-
ronment with a minimum of used energy
2) to provide for a good visual daylight environment which satisfies
human needs.
The thesis deals mainly with windows that have a rather high visual trans-
mittance and good thermal insulation (i.e. low U-value). The aim of the
work has been to identify optimal window choices that satisfy both of the
above mentioned goals.
The thesis is divided into one text part and one article part, which
contains 5 conference and journal papers. The work conducted by the
author in the text part is as follows: Chapter 2 is a literature review that
describes the many performance requirements of windows and defines
the current status of Swedish windows. Chapter 3 is a short course on
window physics for those not familiar with the subject. Chapter 4 is a
review of literature on daylight in buildings. Chapter 5 provides several
examples demonstrating the effect of various window choices on heating,
cooling, indoor temperatures and thermal comfort for both residential
and office buildings.
Parametric studies have been performed in the dynamic energy simu-
lation tool Derob-LTH in order to study the effects of window choices
on energy use and indoor climate. A steady-state program (the BKL-
method) was used to evaluate two years of measurements of energy use
and indoor temperatures of an energy-efficient row-house. Two behav-
ioural studies regarding (1) daylight transmittance, view and room per-
ception using super-insulated windows and (2) the satisfaction with the
daylight environment and the use of shading devices in response to day-

Energy-Efficient Window Systems
162
light/sunlight have also been conducted. Environmental or architectural
psychology methods were employed in full-scale laboratory environments
exposed to the natural climate.
The main limitation is that windows have been studied from a Swed-
ish perspective only. This means that typical Swedish window types have
been used in various parametric studies, and only Swedish climates have
been used. Further, the calculated energy demands only concern the en-
ergy for either heating or cooling the room air to a certain temperature.
Conversion losses and primary energy demands have not been calculated.
Although the studies have been performed from a Swedish perspective,
several results can be transferred to other climates.
One motive for studying energy efficient windows is that energy use
in buildings is strongly connected to serious environmental problems
such as the green house effect, acid rain and eutrophication of land and
waters. Sweden also faces a major challenge with the goal of abolishing
nuclear power, which now produces almost half of the electricity demand.
The total energy use within the building and services sector accounts for
about 40 % of the total energy use in Sweden. The window can thus play
a significant role in reducing the energy use in buildings, or at least make
sure that it does not increase.
Two-years of measurements of a row house in Dalby have shown that
it is possible to build well functioning dwellings with a low specific use of
space heating (33 kWh/yr,m2 useable floor area). This is lower than the
current standard of new housing, even if these houses were built over 10
years ago. The low energy use was achieved by good thermal insulation
(better than the common standard), acceptable air-tightness and heat re-
covery of the exhaust air. Solar gains were substantial during April, May,
June and September, and only short periods with overheating occurred.
The U-values of the envelope were: roof=0.11; walls=0.17; floor=0.20
and windows=1.5 W/m2K. The windows had three panes and one low-e
coating and were thus better than required in the building code.
A window has to fulfil many performance requirements, for example
on aesthetics, thermal insulation, sound insulation, water and moisture
penetration, ventilation, durability, mechanical strength and rigidity, etc.
The main function is nevertheless to provide for daylight and view out.
Triple-pane clear windows have for many years been the de facto stand-
ard in Sweden. They were introduced on a large scale at the end of the
1970s, when a strict energy code was enforced in the wake of the 1973 oil
crisis. The code required a window U-value of 2 W/m2K, and since low-
emittance coated glass was not available at that time, the triple-pane win-
dow was the only solution.

Summary
163
During the last few years there has been a dramatic increase in the use
of low-e coatings. The NUTEK procurement program for energy-effi-
cient windows in 1992 was partly responsible for this. But especially in
1998 when a large window manufacturer changed the production into
using low-e coated glass and argon (in triple-pane glazing) as the stand-
ard product, the market share took a marked jump upwards. Today, about
45 % of all insulating glass units (IGU) are equipped with coated glass,
and an estimation is that the average U-value for windows sold today in
Sweden is between 1.3 and 1.6 W/m2K. However, the best windows on
today’s market have a total U-value (Uwin) of 0.7-1.0 W/m2K.
The emittance of commercial low-e coated glass has decreased in re-
cent years, and today the centre-of-glass U-value (Ucog) can be as low as
1.1 for low-e coated double-pane glazing with argon. The U-value is then
better than for triple, clear glazing (Ucog=1.85) and similar to that of
triple low-e coated glazing (Ucog=1.0-1.3). Although the market share for
double-pane windows is still small, we might see a growing trend that
double-pane windows will replace triple-pane ones, since the cost is lower.
However, it is more difficult to achieve a low total window U-value than
in for example a coupled 1+2 construction, since the frame losses are
more pronounced.
New spacers, so called warm-edge technologies, can reduce the ther-
mal losses around the perimeter of the IGU, but the market share is still
very low, probably due to the extra cost. These spacers will reduce Uwin
somewhat, but especially the risk of condensation at the bottom edge of
the glazing.
If the national interest is to further improve the energy-efficiency of
windows, e.g. by introducing stricter requirements than the current build-
ing code, this will be difficult to achieve if double-pane low-e coated
windows are introduced on a large scale.
The energy balance of a window can be calculated via simple formulae
using degree-hours and cumulated solar radiation summed up to the bal-
ance temperature of the building. The balance temperature is the outside
temperature above which the building does not need to be actively heated.
The simple methods are good, since they make the calculation very trans-
parent and easy to understand. However, it may be difficult to estimate
the balance temperature correctly, and thus the net energy transport
through windows. This is avoided with detailed dynamic energy simula-
tion programs, in which the full interaction with storage effects, ventila-
tion rates, internal loads schemes etc. are accounted for. Further, much
more knowledge is gained on potential overheating problems, cooling
loads, thermal comfort etc., since the temperature of all interior as well as
exterior surfaces are determined on an hourly basis.

Energy-Efficient Window Systems
164
In a calculation example of a detached 1.5-storey single-family house
it was demonstrated that the heating demand may be reduced by a factor
of 5 for super-insulated buildings compared to buildings from 1960. The
lower heating demand is equivalent to a lower balance temperature, a
much shorter heating season, and thereby lower useful solar gains. Dur-
ing the remaining heating-season there will be very little available solar
radiation, since the Swedish winters have no available solar radiation of
significance. At least between November and February the days are short,
the solar altitude is extremely low, and there are many overcast days. There-
fore, as the general insulation levels of the house increases, the window
must have a lower U-value in order not to increase the annual heating
demand compared to a house without windows. For a house insulated
according to 1960 standard, Uwin should be lower than 1.8 in Lund, 1.6
in Stockholm and 1.4 W/m2K in Luleå. For a typical house of today,
Uwin should be lower than 1.1 in Lund, 1.0 in Stockholm and 0.9 W/
m2K in Luleå. For extremely well-insulated buildings, Uwin should be
lower than 0.8 in Lund, 0.7 in Stockholm and 0.6 W/m2K in Luleå.
As the heating-season becomes shorter and shorter, severe overheating
problems may occur, especially in so-called passive housing. Increased
summer ventilation and external shading devices can avoid this problem.
A lower U-value is accompanied with a lower solar energy transmit-
tance (g-value), since low-e coated glass reduces the solar transmittance
outside of the visual spectrum. It was demonstrated that the net energy
saving was approximately halved for double-pane windows with low-e
coatings of decreasing emittances due to the accompanied g-value reduc-
tion. For poorly insulated to ordinary houses a high g-value is desirable
and there this effect must be taken into account. For extremely well-
insulated buildings the lower g-value is welcomed, since it helps to re-
duce unwanted solar gain. In such buildings it is essential to focus on a
low window U-value.
In a comparison between clear and low-e coated glass in office build-
ings with cooling systems, it was shown that triple-pane glazing with one
coating will not increase the heating demand for south facing offices
when the glazed area increases. Energy-efficient windows (2 low-e coat-
ings) will even gain energy over the year. At the same time the cooling
demand is approximately the same for all glazing types. An energy-effi-
cient window towards north displays the same annual increase in heating
demand as a triple-clear glazing towards south. This implies that better
windows should be installed in northerly orientations, but not necessar-
ily towards the south. However, during the cold winter period, only the
energy-efficient window will provide a good thermal comfort.

Summary
165
Thermal comfort problems will also arise in the growing trend of fully
glazed offices. During the summer, intense overheating will occur if no
measures are taken to reduce the transmittance of solar radiation, e.g. by
shading devices and/or solar control glazing. Shading devices will always
be needed to reduce glare, but placed on the outside they have a much
greater potential to reduce cooling needs and indoor operative tempera-
tures.
During the winter the operative temperature will be too low when the
well-insulated wall is totally replaced by double-pane glazing. Measures
can be taken to heat the glass surface, i.e. electrically heated glass, radiative/
convective heating and forced convective heating. From an energy point
of view, the best way is to use coated triple-pane glazing to achieve a very
low U-value. However, condensation on the external side of the glazing
can then be expected at some periods during the year, especially during
early autumn.
Daylight in buildings has many positive aspects apart from providing
light for vision. A literature review on daylight found that the obvious
benefits of windows are the contact they provide with the outside: to be
able to see the changing weather and the activities going on outdoors,
positive mood and provision of warmth and atmosphere of seeing sun-
shine. The four general benefits of windows have been classified by
Heerwagen (1990) as: (1) access to environmental information; (2) ac-
cess to sensory change; (3) a feeling of connection to the world outside
and (4) restoration and recovery.
Research also indicates that there are other effects of daylight apart
from vision and mood. It has been shown that light leads to an activation
of various organs in the brain, and that light affects the diurnal rhythm,
metabolism, pulse frequency, blood pressure and secretion of hormones.
For example Küller et al. (1999) demonstrated that people who work
further than 2 m from a window are more likely to experience SAD (sea-
sonal affective disorder) or SAD-like symptoms.
Glare is a visual comfort problem associated with windows. It derives
both from sunlight entering a space, and from a direct view of the bright,
unobstructed sky. Generally, it is the luminance distribution that will
determine how we perceive a space, and not the illuminance distribution.
Glare is thus linked to excessive luminance levels, and large contrasts.
Visual comfort is a popular term today. However, it is hard to find what
criteria shall be fulfilled in order for a person to have visual comfort. It is
not certain that the commonly used rule of thumb of luminance ratios of
10:3:1 applies to daylight, since various studies have shown that people
tend to tolerate more glare when the source is daylight.

Energy-Efficient Window Systems
166
Daylight utilisation is a term to describe when daylight is used to
replace artificial lighting. A control system for electric lighting that reacts
to the daylight levels in a space is called a daylight responsive lighting
system. This system is of particular interest, since cooling loads can be
reduced without simultaneously increasing heating loads.
As the energy-efficiency of windows is improved, the daylight trans-
mittance is reduced. The effects on potential electric lighting savings were
estimated via lighting calculations in Superlite/Superlink on a single-per-
son office room. This demonstrated that with a continuous dimming
system the savings might be around 40 % for the moderate Glazing-to-
Wall Area Ratio (GWAR) 20 %, and 55 % for the larger GWAR 50 %.
The difference between various clear and low-e coated window choices
was however small.
Another effect of decreasing window U-values is that the transmitted
daylight spectrum might become more and more distorted (like putting
a selective filter on the glass). This is since the improved U-value is achieved
by using three or four panes, and by using one or several low-e coated
panes. The effect on the perception of daylight, view and the room in
general was therefore studied for two identical rooms. The only differ-
ence between the rooms was the window: one room had a quadruple-
pane super-insulated window, and the other had a triple-pane, clear win-
dow. The results showed that people were clearly able to distinguish be-
tween the two windows. They perceived the room with the super-insu-
lated window as darker and more enclosed, and the daylight as more
tinted. The super-insulated window also affected the colours of the room
and of the view, making them look more subdued and drab. The effect
was attributed to the two low-emittance coatings and not to the fourth
glass pane.
As mentioned above, solar shading devices are often necessary in of-
fice environments to control glare and perhaps also to reduce cooling
loads. Simple decorative interior curtains and traditional mid-pane
Venetian blinds can therefore be classified as shading devices. Examples
of external shading devices are awnings, fabric screens and external
Venetian blinds. Some products are better than other to reduce solar gain,
and this is for example studied in the Solar Shading Project at Lund
University (Wall & Bülow-Hübe, 2001). The effects on daylight and
view out, and how the occupant wants to control the shading devices in
relation to the daylight environment has been less studied. A simple pilot
study was therefore conducted in order to compare user preferences for
two external shading devices: one Venetian blind and one awning. No
significant difference between the two systems was found, although the
awning was found easier to adjust and operate. It was not possible to find

Summary
167
any relationship as to how much the shading device was pulled down and
traditional lighting parameters like the desktop illuminance and the sky
luminance seen through the window. However, the existence of a sun-
light patch showed some, although weak, correlation to the percentage of
the window that was covered. A large individual variation as to how much
glare was tolerated could also be observed.
A daylight module to Derob-LTH was developed in cooperation with
Kurt Källblad, who performed the programming, and he is also the au-
thor of the COMFORT program, on which the daylight module was
built. The method is based on a radiosity method where the distribution
of solar radiation (direct and diffuse components) in a space is calculated
and translated to visual radiation via the luminous efficacy. Last, the ra-
diation levels are amplified using the ratio of visual-to-solar transmit-
tance to yield the final illuminance level. Validation was performed for
overcast and clear days against mainly Radiance for a side-lit room and
for a simple atrium. For the overcast sky, the accuracy is acceptable for
both vertical and horizontal windows for the midpoint of the room. For
the sunny sky, Derob-LTH accurately predicts the size and illuminance
level of the sunpatch, at least for the tested vertical window. For the pur-
pose of using it for daylight responsive control of shading devices, the
accuracy of the developed model seems sufficient.
What is the optimum window? The truth is that the best window will
always depend on the situation. Each site has a different climate, and
there is for example a large difference between northern and southern
Sweden. Each building has a certain insulation level, and internal load.
The orientation of windows is also important, especially for poorly to
medium insulated houses. Due to solar gains, the window may have a
higher U-value than the wall without increasing the annual heating load.
The maximum permissible U-value will thus depend on the general insu-
lation of the house, the climate and the window orientation etc. In south-
erly unobstructed locations, triple-pane clear glazing may – on an annual
heating basis – perform on par with super-insulated windows towards
north. Further, since the U-value reductions are usually accompanied by
lower solar gains (e.g. g-values) it is not certain that the window with the
lowest U-value will always the best. However, for extremely well-insu-
lated buildings, the U-value should always be low, since the heating sea-
son is very short, and useful solar gain is small.
Thermal comfort problem may arise during winter when window U-
values are high. These problems are accentuated as the window areas in-
crease. Highly glazed buildings and extremely well-insulated buildings
will also suffer from summer problems with high operative temperatures

Energy-Efficient Window Systems
168
and high cooling needs. In such situations a low U-value is required to
solve the winter problem, and a low g-value together with shading de-
vices will solve the summer problem.
Küller's findings that seasonal affective disorder (SAD) or SAD-like
symptoms are lower for people working closer than 2 m from windows.
is then another argument for using energy-efficient windows, since the
thermal comfort experienced close to such windows is improved. The
daylight issue should also be considered. Perhaps one low-e coating is
sufficient in many cases, since two coatings will tint the light more than
one. Solar control coatings can be expected to affect the tint more than
low-e coatings, and more research is needed on the effects on perception
of daylight and rooms.
Which compromise is made in each situation is eventually up for the
client to make. Tools are therefore needed to provide realistic data on the
predicted behaviour of a planned building. To integrate daylight tools
into thermal programs is thus one attempt to try to interlink the two
sides of the same coin: the energetic side and the daylight side.

Acknowledgements
169
Acknowledgements
I wish to express my gratitude to my advisors during my last year of
study, Maria Wall and Björn Karlsson for their enthusiasm, good ideas
and wittiness. Especially Björn widened my knowledge on bandy and the
bible. I am also very thankful towards Rikard Küller, an excellent tutor
during my behavioural studies, who has always given me self-confidence
and support throughout my years of study. Special thanks go to all my
colleagues at Energy and Building Design, and at the former department
of Building Science, especially Petter Wallentén, Marie-Claude Dubois
and Kurt Källblad. Arne Roos and Joakim Karlsson from Uppsala Uni-
versity have been very helpful in providing data and knowledge regarding
coated glazing. Along the way several other people as well as organisa-
tions have provided valuable information. I thank all of you, none men-
tioned, none forgotten.
Financial support has also been given both from the Swedish National
Energy Administration and the Swedish Council for Building Research
(now Formas, the Swedish Research Council for Environment, Agricul-
tural Sciences and Spatial Planning).
Finally, I would like to express my greatest gratitude towards my fam-
ily: Jean-Yves, Erika and Chloé, and my parents Lisbeth and Staffan, who
all have liberated me from endless household chores and given me great
support and freedom, especially during the intense period of finalizing
this report.

Energy-Efficient Window Systems
170

References
171
References
Andresen, I., Aschehoug, Ö. & Thyholt, M. (1995). Computer simulations
of energy consumption for lighting, heating and cooling for an office us-
ing different light control strategies. (Report STF62 A95015).
Trondheim, Norway: SINTEF Arkitektur og byggteknik.
Antell, O., & Lisinski, J. (1988). Fönster. Historik och råd vid renovering.
(Windows. History and advice for renovation). Rapport RAÄ 1988:1.
Stockholm: Riksantikvarieämbetet. (in Swedish).
Arasteh, D., Reilly, S. & Rubin, M. (1989). A Versatile Procedure for
Calculating Heat Transfer Through Windows. ASHRAE Trans.
95
Pt
2. 755-765.
Arumi-Noé, F. & M. Wysocki, (1979). The DEROB system volume I,
User’s manual for the DEROB system. Austin, Texas, USA: University
of Texas, School of Architecture, Numerical Simulation Laboratory.
Arumi-Noé, F. (1979). The DEROB system volume II, Explanatory notes
and theory. Austin, Texas, USA: University of Texas, School of Archi-
tecture, Numerical Simulation Laboratory.
Ashdown, I. (1996). Lighting for Architects. Computer Graphics World
19
(8) 38-44.
Bally, M. A. & Lenhardt, K. (1999). TPS – Thermo Plastic Spacer Sys-
tem – A New, Revolutionary and Future-Oriented Production Tech-
nology for the High-Tech Insulating Glass of Tomorrow. Proceed-
ings of Glass Processing Days, 13-16 June’99, Tampere, Finland.
BBR 1999. (1998). Boverkets Byggregler. (Building regulations of the Na-
tional Board of Housing, Building and Planning) BFS 1993:57 med
ändringar t.o.m. BFS 1998:38. Boverket.
Bell, P. A., Greene, T. C., Fisher, J. D. & Baum, A. (1996). Environmen-
tal Psychology. 4th ed. Fort Worth, TX, USA: Hartcourt Brace Col-
lege Publishers.

Energy-Efficient Window Systems
172
Björk, C., Kallstenius, P. & Reppen, L. (1984). 2nd ed. Så byggdes husen
1880-1980. (So were the houses built 1880-1980). (T1:1984). Stock-
holm, Sweden, Stockholms stadsbyggnadskontor och Statens råd för
byggnadsforskning. (in Swedish).
Blomberg, T. (1996). Heat Conduction in Two and Three Dimensions.
Computer Modelling of Building Physics Applications. (Report TVBH-
1008). Lund, Sweden: Lund University, Department of Building
Physics.
Bodlund, B., Mills, E., Karlsson, K. & Johansson, T. B. (1989). The
Challenge of Choices: Technology Options for the Swedish Electric-
ity Sector. In Electricity: Efficient End-Use and New Generation Tech-
nologies, and Their Planning Implications. Johansson, T. B, Bodlund,
B. & Williams, R. H. (eds.). Lund: Lund University Press, pp. 883-
947.
Brolin, H. (1987, 1990). Regler för P-märkning av fönster. (Rules for P-
labelling of windows). Report 1987:09Rev90. Borås: Statens Prov-
ningsanstalt. (in Swedish).
Boubekri, M., & Boyer L. L. (1992). Effect of window size and sunlight
presence on glare. Lighting Research and Technology
24
(2) 69-74.
Boyce, P. & Kennaway, D. J. (1987). Effects of Light on Melatonin Pro-
duction. Biol. Psychiatry
22
473-478.
Boyce, P., Eklund, N., Mangum, S., Saaalfield, C. & Tang, L. (1995).
Minimum acceptable transmittance of glazing. Lighting Research &
Technology
27
(3) 145-152.
Button, D. & Pye, B. (eds.). (1993). Glass in building: Guide to modern
architectural glass performance. Oxford: Butterworth Architecture.
Byggforskningsrådet (1987). Energi i byggd miljö. (Energy in the built
environment). (Report G16:1987). Stockholm, Sweden: The Swed-
ish Council for Building Research. (in Swedish).
Byggforskningsrådet (1993). Arkitektens utvidgade ansvar. (The extended
responsibility of the architect. (Report T16:1993). Stockholm, Swe-
den: The Swedish Council for Building Research. (in Swedish).
Byggforskningsrådet (1995). Energiboken. (The energy book). (Report
T21:1995). Stockholm, Sweden: The Swedish Council for Building
Research. (in Swedish).

References
173
Bülow-Hübe, H. (1990). Helsingborgs kommuns energisystem nu och år
2010. (The energy system of Helsingborg now and year 2010). Lund:
Miljödelegationen Västra Skåne, Energiprojektet. (in Swedish).
Bülow-Hübe, H. & Blomsterberg, Å. (1992). Solbyn i Dalby. Utvärdering
av en energisnål radhuslägenhet med glasrum. (The Solar Village in
Dalby. Evaluation of an energy-efficient row house with sunspace).
(Rapport TABK--92/3003). Lund: Lunds tekniska högskola, Inst. för
byggnadskonstruktionslära. (in Swedish).
Bülow-Hübe, H. (1994). Dagsljusaspekter på superisolerade fönster. (Day-
light aspects of super-insulated windows). (Rapport TABK--94/3024).
Lund: Lunds tekniska högskola, Inst. för byggnadskonstruktionslära.
(in Swedish).
Bülow-Hübe, H. (1996). Fönster. En kunskapssammanställning. (Windows.
A summary of knowledge). (Rapport TABK--96/3036). Lund: Lunds
universitet, Lunds tekniska högskola, Inst. för byggnadskonstruktions-
lära. (in Swedish).
Böhn-Jullander, I. (1992). Bruno Mathsson Möbelkonstnären Glashus-
arkitekten Människan. (Bruno Mathsson The furniture artist The
glass house architect The man). Bokförlaget Signum, Lund. (in
Swedish).
Carlson, P.-O. (1992). Glas - Möjligheternas byggmaterial. (Glass – The
building material of opportunities). (Report T16:1992). Stockholm:
Byggforskningsrådet.
Carpenter S. & McGowan, A. (1993). Effect of Framing Systems on the
Thermal Performance of Windows. ASHRAE Transactions
99
(1) 907-
914.
Christoffersen, J. (1995). Daylight Utilisation in Office Buildings. (PhD
thesis, SBI Report 258). Hørsholm, Denmark: Danish Building Re-
search Institute.
Christoffersen, J., Petersen, E. & Johnsen, K. (1999a). Beregningsværktøjer
til analyse af dagslysforhold i bygninger. (Calculation tools for analysis
of daylight conditions in buildings). (SBI Rapport 277). Hørsholm,
Denmark: Statens Byggeforskningsinstitut. (in Danish).

Energy-Efficient Window Systems
174
Christoffersen, J., Petersen, E., Johnsen, K., Valbjørn, O. & Hygge, S.
(1999b). Vinduer og daglys – en feltundersögelse i kontorbygninger. (Win-
dows and daylight – a field survey of office buildings). (SBI Rapport
318). Hørsholm, Denmark: Statens Byggeforskningsinstitut. (in
Danish).
CIBSE (1994). CIBSE Code for interior lighting. London, Great Britain:
The Chartered Institution of Building Services Engineers.
CIE (1994). Spatial Distribution of Daylight – Luminance Distributions of
Various Skies. Commission Internationale de l’Eclairage CIE, No. 110.
Clarke, J. A., Hand, J. W. & Janak, M. (1997). Results from the Inte-
grated Performance Appraisal of Daylight-Europe Case Study Build-
ings. Proc. of Right Light 4, 4th European Conference on Energy-Effi-
cient Lighting. Copenhagen, Denmark, 19-21 Nov 1997.
1
117-124.
Collins, B. L. (1976). Review of the psychological reactions to windows.
Lighting Research & Technology
8
(2) 80-88.
Cooper, J. R., Wiltshire, T. & Hardy, A. C. (1973). Attitudes toward the
use of heat rejecting/low light transmission glasses in office build-
ings. Proc. CIE TC-42 Daylighting Symposium, Architectural Design
Processes, Istanbul. (Bruxelles: Comite National Belge de l’Eclairage).
Crawley, D.B., Lawrie, L.K., Winkelmann, F.C., Buhl, W.F., Huang, Y.J.,
Pedersen, C.O., Strand, R.K., Liesen, R.J., Fisher, D.E., Witte, M.J.
& Glazer, J. (2001). EnergyPlus: creating a new-generation building
energy simulation program. Energy and Buildings
33
(4) 319-331.
Dubois, M.-C. (1998). Solar-protective Glazing for Cold Climates: A Para-
metric Study of Energy Use in Offices. (Report TABK--98/3053). Lund,
Sweden: Lund University, Lund Institute of Technology, Dept. of
Building Science.
Dubois, M.-C. (2001). Impact of Shading Devices on Daylight Quality in
Offices: Simulations with Radiance. (Report TABK--01/3062). Lund
University, Dept. of Construction and Architecture, Division of En-
ergy and Building Design. Lund (Sweden). (forthcoming).
Duer, K. (2001). Characterisation of advanced windows. Determination of
thermal properties by measurements. (Report BYG•DTU R-007).
Lyngby: Technical University of Denmark, BYG•DTU.
Duer, K., Svendsen, S. & Mogensen, M. M. (2000). Energy Labelling of
Glazings and Windows in Denmark. Proc. of EuroSun 2000, Copen-
hagen, Denmark, June 19-22 2000.

References
175
Duffie, J. A. & Beckman, W. A. (1991). 2nd ed. Solar Engineering of
Thermal Processes. Wiley Interscience.
Elmahdy, A. H. & Yusuf, S. (1995). Determination of Argon Concentra-
tion and Assessment of the Durability of High-Performance Insulat-
ing Glass Units Filled with Argon Gas. ASHRAE Transactions
101
(pt1)
1026-1037.
ElSherbiny, S. M., Raithby, G. D. & Hollands, K. G. T. (1982). Heat
Transfer by Natural Convection Across Vertical and Inclined Air Lay-
ers. Journal of Heat Transfer, Feb
1982,
104
96-102
EN 673 (1997). Glass in building - Determination of thermal transmit-
tance (U value) - Calculation method.
Enarun, D. & Littlefair, P. (1995). Luminance models for overcast skies:
Assessment using measured data. Lighting Research and Technology.
27
(1) 53-58.
Energimyndigheten. (2000). Energiläget 2000 (Energy in Sweden 2000).
Eskilstuna, Sweden: The Swedish National Energy Administration.
Erhorn, H. & J. Stoffel, (eds.) (1996). ADELINE 2.0, Advanced daylighting
and electric lighting integrated new environment, Documentation of the
software package ADELINE, version 2.0 (9 volumes), IEA Solar Heat-
ing and Cooling Program, Task 12: Solar Building Energy Analysis
Tools.
Feist, W. (1995). The Passive Houses at Darmstadt/Germany. Darmstadt,
Institute Housing and Environment (IWU).
Fergusen, J. E. & Wright, J. L. (1984). Vision: A computer program to
evaluate the thermal performance of super windows, Solar Energy pro-
gram, National Research Council Canada, Report No Passice-10, June
1984.
Fontoynont, M., Laforgue, P., Mitanchey, R., Aizlewood, M., Butt, J.,
Carroll, W., Hitchcock, R., Erhorn, H., De Boer, J., Dirksmöller,
M., Michel, L., Paule, B., Scartezzini, J. L., Bodart, M. & Roy, G.
(1999). Validation of daylighting computer programs. IEA SHC Task
21/ ECBCS Annex 29 Daylight in buildings. Vaulx-en-Velin Cedex,
FR: Ecole Nationale des Travaux Public de l’Etat. Département Génie
Civil et Bâtiment.
Frank, T. (1994). Thermische Verbesserung des Glasrandverbundes. Status-
seminar “Energiforschung im Hochbau”, 15/16 Sep 1994. (in Ger-
man).

Energy-Efficient Window Systems
176
Fredlund, B. (1999). Lågemissionsglas och renovering förbättrar äldre fönsters
värmeisolering. (Low-emittance glass and renovation improves older
windows heat insulation). (Report TABK--99/3055). Lund, Sweden:
Lund University, Lund Institute of Technology, Dept. of Building
Science. (in Swedish).
Goldemberg, J., Johansson, T. B., Reddy A. K. N. & Williams, R. H.
(1988). Energy for a Sustainable World. New Delhi: Wiley-Eastern.
Granqvist, C. G. (1981). Radiative heating and cooling with spectrally
selective surfaces. Applied Optics
20
(15) 2606-2615.
Grau, K. & Wittchen, K. B. (1999). Building Design System and CAD
integration. In Proceedings of IBPSA Building Simulations ‘99, 13-15
Sept 1999, Kyoto, Japan.
Grenier, J. C., De la Casinière, A., & Cabot, T. (1994). A spectral model
of Linke’s turbidity factor and its experimental implications, Solar
Energy
52
(4) 301-313.
Gustavsson, L. Johansson, B. & Bülow-Hübe, H. (1992). An Environ-
mentally Benign Energy Future for Western Scania, Sweden. Energy
17
(9), 809-822.
Göransson, C. (1995). Fönstret en viktig del av fasadens ljudisolering.
(The window is an essential part of the sound insulation of façades).
Provning & Forskning (1)7. (in Swedish).
Heerwagen, J. (1990). Windows, windowlessness and simulated view.
EDRA 21 269-280 (Environmental Design Research Association).
Hesselgren, S. (1987). On Architechture. An architectural theory based on
psychological research. Lund, Sweden: Studentlitteratur.
Hollands, K. G. T., Unny, T. E., Raithby, G. D. & Konicek, L. (1976).
Free Convective Heat Transfer Across Inclined Air Layers. Journal of
Heat Transfer, May 1976,
98
189-193.
Hollwich, F. (1979). The Influence of Ocular Light Perception on Metabo-
lism in Man and in Animal. New York, Springer Verlag.
Hopkinson, R. G., Petherbridge, P. & Longmore, J. (1966) Daylighting.
University College, London and Building Research Station, Garston,
Watford, England. London: Heinemann Ltd.
Hjorth, B. (1992). Utan fönster ingen arkitektur. (Without windows no
architecture). Arkitektur (4) 52-57. (in Swedish).

References
177
IESNA (1993). Lighting Handbook. 8th edition. New-York: Illuminating
Engineering Society of North America (IESNA).
Inui, M. & Miyata, T. (1973). Spaciousness in interiors. Lighting Re-
search & Technology
5
(2) 103-111.
ISO 9050:1990. Glass in building – Determination of light transmittance,
solar direct transmittance, total solar energy transmittance and ultravio-
let transmittance, and related glazing factors.
ISO 9845-1:1992. Solar energy – Reference solar spectral irradiance at the
ground at different receiving conditions – Part 1: Direct normal and
hemispherical solar irradiance for air mass 1,5.
ISO 10077-1:2000. Thermal performance of windows, doors and shutters –
Calculation of thermal transmittance – Part 1: Simplified method.
ISO/DIS 10077-2. Thermal performance of windows, doors and shutters –
Calculation of thermal transmittance – Part 2: Numerical method for
frames.
ISO/DIS 15099. Thermal performance of windows, doors and shading de-
vices – Detailed calculations.
Janak, M (1997). Coupling Building Energy and Lighting Simulation.
Proc. of IBPSA, Building Simulation ’97 Prague, Czech Republic, Sep
8-10 1997. Vol
2
313-319.
Johansson, B. (1990). Malmö kommuns energisystem nu och år 2010. (The
energy system of Malmö now and year 2010). Lund: Miljödelega-
tionen Västra Skåne, Energiprojektet. (in Swedish).
Jonasson, H. (1994). Luftljudsisolering. (Airborne sound insulation). Borås:
Sveriges Provnings- och Forskningsinstitut. (in Swedish).
Jonsson, B. (1985). Heat transfer through windows. During the hours of
darkness with the effect of infiltration ignored. (Document D13:1985).
Stockholm: Swedish Council for Building Research.
Jonsson, B. (1995). Utvändig kondens på fönster. Mätningar i Borås 1994.
(External condensation on windows. Measurements in Borås 1994).
(SP Report 1995:01). Borås: Sveriges Provnings- och Forsknings-
institut. (in Swedish).
Kaplan, R. (1983). The role of nature in the urban context. In Altman, I.
& Wohlwill, J. F. (eds.). Behavior and the Natural Environment. (pp.
127-159). New York: Plenum Press.

Energy-Efficient Window Systems
178
Kaplan, R. (1985). Nature at the doorstep: residential satisfaction and
the nearby environment. Journal of Architecture and Planning Research.
2
115-127.
Karlsson, B. & Brunström, C. (1987). Characterization of transparent
insulation materials. Proc. of Building Physics in the Nordic Countries.
Lund, Sweden, Aug 24-26 1987.
Karlsson, J. (2001). WinSel- a general window selection- and energy rat-
ing tool, Proc. of WREN-2000. Brighton, UK, July 1-7, 2000.
Karlsson, J., Karlsson, B. & Roos, A. (2001). A simple model for assess-
ing the energy performance of windows. Energy and Buildings
33
(7)
641-651.
Karlsson, J & Roos, A. (2000). Modelling the angular behaviour of the
total solar energy transmittance of windows. Solar Energy
69
(4) 321-
329.
Karlsson, J., Rubin, M. & Roos, A. (2001). Evaluation of predictive
models for the angle-dependent total solar energy transmittance of
glazing materials. Solar Energy
71
(1) 23-31.
Kittler, R., Darula, S. & Perez, R. (1998). A set of standard skies. Charac-
terizing daylight conditions for computer and energy conscious design.
Final report. Institute of Construction and Architecture, Slovak Acad-
emy of Sciences and Atmospheric Sciences Research Center, State
University of New York. Bratislava: Polygrafia SAV.
Kovach-Hebling, A., Goller, M., Herkel, S. & Wienold, J. (1997). As-
sessing the Energy Saving Potential of Daylighting Technologies for
Non-residential Buildings in Germany. Proc. of Right Light 4, 4th
European Conference on Energy-Efficient Lighting. Copenhagen, Den-
mark, 19-21 Nov 1997. Vol 2, 115-118.
Kvist, H. (2000). DEROB-LTH for MS Windows, User Manual version
99.02+3. Lund: Lund University, Lund Institute of Technology, Dept
of Construction and Architecture.
Küller, R. (1981). Non-visual effects of light and colour. Annotated bibliog-
raphy. (Document D15:81). Stockholm Sweden: Swedish Council
for Building Research.
Küller, R. & Küller, M. (2001). The influence of daylight and artificial
light on diurnal and seasonal variations in humans. A bibliography. Tech-
nical Report No. 139. Vienna: International Commission on Illumi-
nation (CIE).

References
179
Küller, R. & Laike, T. (1998). The impact of flicker from fluorescent
lighting on well-being, performance and physiological arousal. Ergo-
nomics
41
(4) 433-447.
Küller, R., Ballal, S. G., Laike, T. & Mikkelides, B. (1999). Shortness of
daylight as a reason for fatigue and sadness. A Cross-Cultural Com-
parison. Proceedings CIE 24th Session, Warsaw, 1999. Vol. 1 Pt. 2 pp.
291-294.
Küller R. & Lindsten C. (1992). Health and behavior of children in
classrooms with and without windows. Journal of Environmental Psy-
chology
12
(4) 305-317.
Küller, R. & Wetterberg, L. (1996). The subterranean work environment:
Impact on well-being and health. Environment International
22
(1)
33-52.
Källblad, K. (1994). The BKL-METHOD Computer Programme. User
Manual. (Report TABK--94/3019). Lund: Lund University, Lund
Institute of Technology, Dept of Building Science.
Källblad, K. (1996). Comfort. A Computer Program for Thermal Comfort
Simulation. (Report TABK--96/3035). Lund, Sweden: Lund Univer-
sity, Lund Institute of Technology, Dept of Building Science.
LBL (1992). WINDOW 4.0: Program Description. A PC Program for
analyzing the thermal performance of fenestration products. (Report LBL-
32091, TA-285). Berkeley, CA: Lawrence Berkeley Laboratory.
Liljefors, A. & Ejhed, J. (1990). Bättre belysning. (Better lighting). (Rap-
port T17:1990). Stockholm: Byggforskningsrådet. (in Swedish).
Littlefair, P. (1985). Luminous efficacy of daylight: a review. Lighting
Research and Technology.
17
(4) 162-181.
Littlefair, P. (1988). Measurements of the luminous efficacy of daylight.
Lighting Research and Technology.
20
(4) 177-188.
Ljuskultur. (1990). Belysning inomhus, riktlinjer och rekommendationer.
(Indoor lighting, guidelines and recommendations). Stockholm. (in
Swedish).
Luckiesh, M. (1944). Brightness engineering. Illuminating Engineering,
February 1944, 75-92.
Ludlow, A. M. (1976). The functions of windows in buildings. Lighting
Research and Technology
8
(2) 57-65.

Energy-Efficient Window Systems
180
Löfberg, H.-A. (1987). Räkna med dagsljus. (Count with daylight). Gävle:
Statens institut för byggnadsforskning. (in Swedish).
Markus, T. (1967). The Function of Windows – A Reappraisal. Building
Science
2
97-121.
Merkell, A.-L. (1989). Årligt energiflöde genom fönster samt k-värden hos
fönster med plastfolier och lågemitterande beläggningar. (Annual energy
flow through windows and U-values of windows with plastic foils
and low-emittance coatings). Älvkarleby, Sweden: Vattenfall, Älv-
karlebylaboratoriet. (in Swedish).
Mills, E. (1991). Evolving Energy Systems. Technology Options and Policy
Mechanisms. (Doctoral Dissertation). Lund, Sweden: Lund Univer-
sity, Department of Environmental and Energy Systems Studies.
Muneer, T. & Angus, R. C. (1995). Luminous efficacy: Evaluation of
models for the United Kingdom. Lighting Research and Technology
27
(2) 71-77.
Muneer, T. (1998). Evaluation of the CIE overcast sky model against
Japanese data. Energy and Buildings
27
(2) 175-177.
MUR 90 (1991). Murverkshandboken. Häfte 5 C. Byggnadsteknisk
utformning. (Masonry Handbook. Booklet 5 C. Technical construc-
tion details). Sveriges Tegelindustriförening/TCK AB.
Naturvårdsverket. (1999). Hållbar energiframtid? Långsiktiga miljömål med
systemlösningar för el och värme. (Sustainable energy future? Long term
environmental goals and systems solutions for electricity and heat).
Final report of the SAME project. ISBN 91-620-4965-8. (in Swed-
ish).
Nielsen T. R. & Svendsen, S. (2000) Determination of net energy gain
from glazings and windows. Proc. of EuroSun 2000, Copenhagen,
Denmark, June 19-22 2000.
NFRC (2000). National Fenestration Rating Council Update, Vol.
9
(2) 2.
March/April 2000.
NR 1 (1988). Nybyggnadsregler. Föreskrifter och allmänna råd. (New Build-
ing Regulations. Directions and general advice). BFS 1988:18
Boverket, Stockholm: Allmänna Förlaget
NUTEK (1992). Teknikupphandling av energieffektiva fönster. (Technol-
ogy procurement of energy-efficient windows). Stockholm: NUTEK,
Effektivare energianvändning. (in Swedish).

References
181
NUTEK (1993, 1995). 2nd ed. Möjligheter med installationssystem och
energieffektiva fönster. (Opportunities for building services systems
and energy-efficient windows) (Rapport T 1993:4). Stockholm:
NUTEK, Effektivare energianvändning. (in Swedish).
NUTEK (1994). Lighting Design Requirements - Office Lighting. Stock-
holm: Dept of Energy Efficiency, Swedish National Board for Indus-
trial and Technical Development.
NUTEK (1995). Kondens på fönster. (Condensation on windows). Stock-
holm: NUTEK, Effektivare energianvändning. (in Swedish).
Olsson-Jonsson, A. (1988). Förbättring av fönsters värmeisolering samt
upplevelse av fönsterbyten. (Improvement of windows’ heat insulation
and perception of window replacement). (Rapport R40:1988). Stock-
holm: Byggforskningsrådet. (in Swedish).
Osterhaus, W. K. E. (2001). Discomfort glare from daylight in computer
offices: What do we really know? Proc. of Lux Europa 2001, Reykja-
vik, Iceland.
Paule, B., Scartezzini, J.-L., Reynolds, M. & Baker, N. (2000). Handling
daylight in the early design stage. Proc. of EuroSun 2000, Copenha-
gen, Denmark, June 19-22 2000.
Pearson, D. (1994). Earth to Spirit. In Search of Natural Architecture. Lon-
don, UK: Gaia Books Ltd.
Philips Lighting (1997). Product specifications on HF dimmable reac-
tors.
prEN 10077-2. Thermal performance of windows, doors and shutters - Cal-
culation of thermal transmittance - Part 2: Numerical method for frames
(equal to ISO/DIS 10077-2:1998)
Reilly, S. (1994). Spacer Effects on Edge-of-Glass and Frame Heat Trans-
fer. ASHRAE Transactions
100
(1) 1718-1723.
Reinhart, C. F. & Herkel, S. (2000). The simulation of annual daylight
illuminance distributions – a state-of-the-art comparison of six RA-
DIANCE-based methods. Energy and Buildings
32
(2) 167-187.
Roos, A. (1994). Bestämning av optiska egenskaper hos fönster. (Determi-
nation of optical performance of windows). Uppsala: Uppsala
Universitet, Teknikum. (in Swedish).

Energy-Efficient Window Systems
182
Roos, A. & Karlsson, B. (1994). Optical and thermal characterization of
multiple glazed windows with low U-values. Solar Energy
52
(4) 315-
325.
Roos, A. (1997). Optical characterization of coated glazings at oblique
angles of incidence: measurements versus model calculations. Jour-
nal of Non-Crystalline Solids
218
247-255.
Rusak, B., Eskes, G. A. & Shaw, S. R. (1996). Lighting and Human Health.
A Review of the Literature. Ottawa, Ontario, Canada: Mortgage and
Housing Corporation.
SBIC (1996). Designing Low-Energy Buildings: Passive Solar Strategies &
ENERGY-10 Software. Integrating Daylighting, Energy-Efficient Equip-
ment, and Passive Solar Design in Commercial, Institutional, and Resi-
dential Buildings. Washington DC, USA: Passive Solar Industries
Council.
SBN 75 (Swedish Building Code 1975). Supplement 1. Energy conser-
vation etc. Stockholm, Sweden: National Board of Physical Planning
and Building. (in Swedish).
SBN 80 (Swedish Building Code 1980) Stockholm, Sweden: National
Board of Physical Planning and Building. (in Swedish).
SBN 85 (Swedish Building Code 1985) Stockholm, Sweden: National
Board of Physical Planning and Building. (in Swedish).
Smeds, J. & Wall, M. (2001a). IEA Task 28 / Annex 38: Sustainable
Solar Housing, Subtask B: Reference Detached House, Internal docu-
ment, sep 2001.
Smeds, J. & Wall, M. (2001b). IEA Task 28 / Annex 38: Sustainable
Solar Housing, Subtask B: Internal Gains, Assumptions for simula-
tions, Working document, sep 2001.
Stockholms Byggnadsnämnd 1987 (1988). Fönster & balkonger vid
ombyggnad. (Windows and Balconies at Renovation) Stockholm. (in
Swedish).
Szerman, M. (1994). Auswirkung der Tageslichtnutzung auf das energetische
Verhalten von Bürogebäuden (The effect of daylight utilisation on the
energy performance of office buildings) (PhD Thesis), Universität
Stuttgart, Konstruktive Bauphysik, Stuttgart, Germany. (in German).

References
183
Thormark, C. (2001). Recycling Potential and Design for Disassembly in
Buildings. (PhD thesis Report TABK--01/1021). Lund: Lund Uni-
versity, Lund Institute of Technology, Dept of Construction and Ar-
chitecture, Div of Building Science.
Thyholt, M., Andresen, I., Hugdal, B. & Aschehoug, Ø. (1994). Frame
and Edge Seal Technology. A State of the Art Survey. IEA Solar Heating
and Cooling Programme, Task 18, Case Study B9, Working Docu-
ment T18/B9/WD1/94 STF62 A94004 Trondheim, NO: SINTEF
Arkitektur og byggteknikk.
Tonello, G. (2001). Lighting, Mood, and Seasonal Fatigue in Northern
Argentina. Comparison to Countries Close to and Further from the Equa-
tor. (Doctoral Dissertation). Lund, Sweden: Lund University, Lund
Institute of Technology, Environmental Psychology Unit.
Tregenza, P. R. & Waters, I. M. (1983). Daylight coefficients. Lighting
Research and Technology
15
(2) 65-71.
Ulrich, R. S. (1984). View from the window may influence recovery from
surgery. Science
224
420-421.
Veitch, J. A. & Newsham, G. R. (1996a). Determinants of lighting quality
I: State of the science. (NRCC-39866). Paper presented at the 1996
Annual Conference of the Illuminating Engineering Society of North
America, Cleveland, OH, August 5-7 1996.
Veitch, J. A. & Newsham, G. R. (1996b). Determinants of lighting quality
II: research and recommendations. (NRCC-40343). American Psycho-
logical Association 104th Annual Convention, Toronto, Ontario, 8/
9/96, 1-55.
Veitch, J. A. (2001). Psychological processes influencing lighting quality.
Journal of the Illuminating Engineering Society
30
(1) 124-140.
Velds, M. (1999). Assessment of lighting quality in office rooms with
daylighting systems. (PhD-thesis). Delft, Holland: Technische
Universiteit Delft.
Verderber, R. R. & Rubinstein, F. M. (1984). Mutual impacts of lighting
controls and daylighting applications Energy and Buildings
6
(2) 133-
140.
Wall, M. (1994). Projekteringshjälpmedel för inglasade rum. Klimat och
energi. (Design tools for glazed spaces. Climate and energy) (Rapport
TABK--94/3023). Lund, Sweden: Lunds universitet, Lunds tekniska
högskola, Inst f byggnadskonstruktionslära. (in Swedish).

Energy-Efficient Window Systems
184
Wall, M. (1996). Climate and Energy Use in Glazed Spaces. (Doctoral
Dissertation Report TABK--96/1009). Lund, Sweden: Lund Univer-
sity, Lund Institute of Technology, Dept. of Building Science.
Wall, M. (1997). Distribution of solar radiation in glazed spaces and
adjacent buildings. A comparison of simulation programs. Energy
and Buildings
26
129-135.
Wall, M. & Bülow-Hübe H. (eds.). (2001). Solar protection in buildings.
(Report TABK--01/3060). Lund, Sweden: Lund University, Lund In-
stitute of Technology, Dept. of Construction & Architecture.
Wallentén, P. (1993). Fönstrets betydelse för den riktade operativa tem-
peraturen. (The importance of the window for the directed operative
temperature). Lund, Sweden: Lund University, Lund Institute of Tech-
nology, Dept. of Building Science. (in Swedish).
Ward Larson, G. & Shakespeare R. (1998). Rendering with Radiance: The
Art and Science of Lighting Visualization. San Francisco, CA: Morgan
Kaufmann Publishers.
Wetterberg, L. (1978). Melatonin in humans. Psychological and clinical
studies. J Neural Transm
13
289-310.
Wilkins, A. J., Nimmo-Smith, I., Slater, A. I. & Bedocs, L. (1989). Fluo-
rescent lighting, headaches and eye-strain. Lighting Research and Tech-
nology
21
11-18.
Wilson, L. M. (1972). Intensive care delirium. The effect of outside dep-
rivation in a windowless unit. Archives of Internal Medicine
130
225-
226.
Winkelmann, F. C. & Selkowitz, S. (1985). Daylighting Simulation in
the DOE-2 Building Energy Analysis Program. Energy and Buildings
8
(4) 271-286.
Wolf, A. (1988). New Development in the Field of Insulating Glass Units.
Construction and Building Materials
2
(3) 134-144
Wolf, A. T. & Waters, L. J. (1993). Factors governing the life expectancy
of dual-sealed insulating glass units. Construction and Building Mate-
rials
7
(2) 101-107.

Article I
185
I

Energy-Efficient Window Systems
186

Article I
187

Energy-Efficient Window Systems
188

Article I
189

Energy-Efficient Window Systems
190

Article I
191

Energy-Efficient Window Systems
192

Article II
193
II

Energy-Efficient Window Systems
194

Article II
195

Energy-Efficient Window Systems
196

Article II
197

Energy-Efficient Window Systems
198

Article II
199

Energy-Efficient Window Systems
200

Article III
201
III

Energy-Efficient Window Systems
202

Article III
203

Energy-Efficient Window Systems
204

Article III
205

Energy-Efficient Window Systems
206

Article III
207

Energy-Efficient Window Systems
208

Article III
209

Energy-Efficient Window Systems
210

Article III
211

Energy-Efficient Window Systems
212

Article IV
213
IV

Energy-Efficient Window Systems
214

Article IV
215
OFFICE WORKER PREFERENCES
OF EXTERIOR SHADING DEVICES:
A PILOT STUDY
Helena Bülow-Hübe
Dept. of Construction and Architecture, Div. of Energy and Building Design, Lund University, P.O. Box 118,
221 00 Lund, Sweden, +46-46 222 73 56, +46-46 222 47 19, helena.bulow-hube@bkl.lth.se
Abstract – Solar shading devices are commonly used in offices to reduce cooling loads and glare from
windows, but they also affect daylighting and the view to the outside. In this study, the function and op-
eration of an awning and an exterior venetian blind as well as their influence on the view out was assessed
by fifty office workers. The preferred position of the shading devices, the interior illumination and the
weather conditions were recorded. An interior dimmable lighting fixture was introduced in order to see if
there was a difference between the shading devices regarding the need of complementary electric lighting.
The awning was found easier to adjust than the venetian blind. Both shading devices somewhat affected
the view to the outside. The shading devices were used frequently to avoid glare from the window. How-
ever, preliminary results show no or weak correlation between common lighting concepts such as interior
illuminance or sky luminance and how much the shading device was pulled down. On the other hand,
there was a weak relationship between the existence of sunlight patches in the room and the position of
the shading device. Further, the awning showed to be sensitive to wind, creating a disturbing “flickering”
sunlight patch on sunny, windy days. There was no difference between the shading systems in the use of
complementary electric lighting.
1. INTRODUCTION
Shading devices are often used in buildings, perhaps
mainly to reduce cooling energy use, but also to control
glare and daylighting. The control of daylighting is actu-
ally very central because it is linked to occupants’ satis-
faction and performance.
At Lund University’s, Dept. of Construction and Ar-
chitecture, there is a large ongoing project on solar shad-
ing devices in buildings (Wall & Fredlund, 1999). This
project primarily deals with the thermal aspects of shad-
ing devices. However, the daylight aspects are equally
important, not only the effects on illuminance levels, but
also on the view out and the perception of a room. Fur-
ther, increased knowledge on the preferences of occu-
pants would be useful in the selection process of shading
devices and also to improve automatic control systems.
According to Littlefair (1999) shading of windows is
needed for three main reasons: to reduce overheating, to
reduce glare from windows and to provide privacy. Even
so, some sunlighting may still be wanted. The positive
impacts of sunlight is to enhance the visual, emotional,
and psychological well-being of occupants, or using it as
a heat source (Boubekri et al., 1991). However, their
study largely failed to demonstrate the effect of window
size or sunlight patches on office worker’s mood and
satisfaction. However, all subjects were exposed to sunny
conditions and never to a condition without sunlight
penetration. Compared to having no access to windows or
view out, people generally prefer windowed space,.
(Collins, 1976).
1.1 Glare
Among the mentioned negative impacts of windows is
glare. Since the luminance of the sky may well be several
times higher than that of the interior walls – even on an
overcast day – glare discomfort can arise from a direct
view of the sky (Chauvel et al. 1982). They also found
that glare from windows is perceived differently than
glare from large artificial sources, due to the psychologi-
cal differences in the contents of the field of view. Chau-
vel et al. also found large individual differences in the
tolerance of glare. In another study on sunlight penetra-
tion, glare was only moderately affected by window size
(Boubekri & Boyer, 1992).
1.2 Occupant behaviour
In a study on office worker’s behaviour, Rubin et al.
(1978) changed the position of venetian blinds during
weekends to either fully up, or down, closed, and then
studied the occupant’s response by taking external photo-
graphs of the facades of the building. They found that
most blind positions were changed only once per week.
Moreover, they were generally put back in the same po-
sition as before the treatment. The most significant influ-
ence was that of the orientation: on the north side, blinds
were generally kept more open than on the south side.
There were also some effects, although more subtle, of
climatic season and view out.
In another study, Vine et. al. (1998) compared occu-
pant response and satisfaction of an automated blind with
an auto user control mode (manual override of auto
mode) and with full manual control. Although no statisti-
cal analysis was made of the subjects’ responses, over
75% of the subjects preferred more daylight in the auto

Energy-Efficient Window Systems
216
mode. They were generally satisfied with the lighting in
the auto user control mode, but experienced some glare in
the manual mode.
Boyce (1997) claims that if people sitting near to a
window have expectations of thermal or visual discom-
fort to occur, and if they consider that their electric light-
ing is adequate, they will leave the blinds down, unless
they have strong values about the environment. He fur-
ther believes that few people have such values. He calls
this seemingly lack of response to changing environ-
mental conditions for human inertia. For any new auto-
mation system to be successful, this inertia must be used
to its advantage. He suggests that a simple timer might be
enough: one that for example switches off the lighting at
a time in the morning when the daylight is usually suffi-
cient, or pulls up a blind at dawn.
2. EXPERIMENTAL DESIGN AND METHODS
2.1 Aim of the study
The aim of this study was to investigate the function,
operation and effect on daylight of a couple of solar
shading devices. Further, when people are allowed to
control the shading devices we wanted to see how they
decide to use them in relation to the outdoor climate.
Another issue was whether different shading devices
need more or less complementary electric lighting. The
experiment was considered as a pilot study to identify
typical positions (or settings) of the shading devices for
use in later studies.
Since this was considered as a pilot study to develop a
test method only two different shading devices were
included: one awning and one exterior venetian blind.
2.2 Test rooms and solar shading devices
At a laboratory at the Dept. of Construction and Archi-
tecture, there were already two identical south-facing
office rooms, 3.0×3.6×2.45 m (W×D×H), used in an
earlier study by Bülow-Hübe (1994). New office desks in
blond wood were purchased, the walls were repainted in a
warm white colour (NCS 0003-Y20R), trimmings and
ceilings were white, and the linoleum floor had colours in
beige-blue-brown. (R
wall
= 0.8, R
ceil
= 0.9, R
floor
= 0.4,
R
desk
= 0.5).
Each room was furnished and equipped with a com-
puter to resemble a real office room. (Fig. 1). The lighting
consisted of a pendant direct/indirect luminaire with
dimmable HF-ballast, one T8 36 W facing upwards, and
two downwards. The control mechanism was a potenti-
ometer placed on the desk. Measured workplane illumi-
nance was 900 lux at full light output (potentiometer
setting = 35), fully dimmed it was 25 lux (setting = 0).
The 1.2×1.3 m, triple-glazed window was in one room
equipped with an exterior retractable venetian blind with
80 mm aluminium slats. (Fig. 2). In the other room, it
was equipped with an exterior retractable awning with a
beige and brown striped fabric. (Fig. 3). In both rooms
the shading device could be operated from the inside by
two buttons placed on the desk. One button was for re-
tracting the shading device (up position), the other for
closing it (down position). For the venetian blind, the
adjustment of the slat angle was done with the same two
buttons, which meant that to change the slat angle, the
position of the bottom slat had to be changed somewhat.
(Figs. 1-3).
Figure 1 Plan of the offices with 2 points for
illuminance measurements as indicated.
Figure 2 Interior of the office with a venetian blind.
2.3 Method for room assessment
The subjects performed two tasks. The first task consisted
in adjusting the shading device until a pleasant daylight
situation was created. A simple questionnaire was filled
in containing three questions: (1) How well does the
operation system of the shading device work? (bad–good)
(2) How satisfied are you with the lighting conditions?

Article IV
217
(unsatisfied–satisfied) (3) How does the shading device
affect your possibility to see out? (not at all–to a large
extent).
The second task was to see whether the lighting situa-
tion was improved with electric lighting, and if so adjust
the light level until a pleasant situation was reached. This
was followed by a last question: (4) How satisfied are
you with the lighting conditions? (unsatisfied–satisfied).
Figure 3 Detail of the office with an awning.
On all 4 questions, the answers were given on seven-
grade scales with the word-pair given in brackets above
on either side of the scale.
2.4 Procedure and subjects
The study was conducted in September and October
1998. The rooms were assessed by 50 subjects in a bal-
anced design with repeated measures. This meant that
each subject assessed both rooms at one occasion.
Upon arrival to the laboratory, the subject was given a
short introduction to the experiment. Thereafter, he/she
was shown into the first room where further instructions
were given on the computer screen. At the start of each
experiment, the shading device was always fully open
(up) and the lighting was turned off.
Before the experiment started, the experiment leader
noted the weather and lighting conditions, time and tem-
perature and measured the interior illumination levels
(see below). The subject was then left alone to perform
the first task.
The experiment leader then returned to note the posi-
tion of the shade and repeat the lighting measurements
before the subject could proceed with task 2. When this
was finished, the lighting measurements were again re-
peated and the potentiometer setting was recorded. After
that, the subject was guided to the other room to repeat
the same procedure.
The subjects were recruited from the School of Archi-
tecture, and consisted of office workers: clerks, research-
ers and doctoral students. They were all used to working
close to a window. Totally, 24 women and 26 men be-
tween 23 and 64 years participated (mean = 43.3;
SD = 10.8).
2.4 Measurements
As a reference of the outdoor lighting conditions, both the
interior horizontal illuminance level and sky luminance
seen through the window of a neighbouring office room
were measured continuously by two Hagner Universal
photometers (S2) connected to a data logger.
Before and after each room assessment, the weather
situation, lighting conditions in the room, time and indoor
temperature were recorded. The illuminance level was
measured at the desk (work surface) (pt. 1) and vertically
on the wall in front of the person, 1.2 m above the floor
(pt. 2, Fig. 1), with a hand-held Hagner digital luxmeter
(EC1). If there was a sun patch in the visual field, this
was noted as follows: 1 = no, 2 = sometimes (varying
conditions), 3 = yes. The weather situation was rated on a
four-grade scale: 1 = sun, some lighter clouds may exist,
2 = sunny, there are clouds that sometimes cover the sun,
3 = cloudy, there are some blue spots were the sun can be
seen, and 4 = overcast, the sky is covered by thick clouds.
Further, the cloudiness was rated on an 8-grade scale, that
defines how many eighths of the sky are covered by
clouds.
If the electric lighting was turned on, the potentiometer
setting was recorded.
The position of the shades were assessed in the fol-
lowing way: For the awning we constructed and mounted
a protractor on the boom which allowed for reading the
boom angle with the accuracy of ± 1°. The boom angle
ba was later transformed to the awning’s slope by the
simple relationship: slope = 45 + ba/2. (Fig. 4).
For the venetian blind, a scale was drawn next to the
window, so that the bottom slat position could be read
from the interior to the accuracy of ± 5 cm, while the slat
angle was estimated manually.
2.6 Data analysis
The data from the rating scales was treated by analysis of
variance using the SPSS MANOVA procedure (Norusis,
1993). The design included both within-subject and be-
tween-subject variance.
Further, regression analysis was used to study the rela-
tionship between the position of the shading device and
the lighting and weather conditions.
Since the position of the shading devices had been
assessed, a transformation of these data was made to
determine which portion of the window surface was cov-
ered by the shade, called here the coverage of the shade.
For the awning it was made in a simple way: The
measured boom angle was transformed to a percentage:
when the awning was fully retracted, this was interpreted
as a bare window (coverage = 0 %). The fully down po-
sition was interpreted as fully closed awning (coverage
= 100 %). (Figure 4).

Energy-Efficient Window Systems
218
Figure 4 Section through window with awning, and
definition of boom angle and slope.
For the venetian blind two factors were weighed together
to estimate the coverage: (1) the distance of the bottom
slat to the top of the window and (2) the slat angle. (Fig-
ure 5). From the position of the viewer, the percentage of
how much of the window that was covered by the blind
had previously been estimated for different slat angles.
This value was then multiplied by the distance (1) (in per
cent) to give the coverage.
Figure 5 Section through window with exterior venetian
blind, and examples of slat angles.
3. RESULTS
3.1 Weather, light and temperature
The illuminance levels, sky luminance, indoor tempera-
ture and weather recordings before and after the assess-
ments in the two rooms are summarised in Table 1. The
differences between the two rooms are very small and are
not significant. Therefore, the environmental conditions
must be considered equal between the two rooms (awning
and venetian blind).
Table 1 Summary of measurements of environmental
conditions before and after assessments. (Mean values).
Awning Venetian blind
Illuminance level
(lux)
Before After Before After
Point 1 2590 675 2610 580
Point 2 950 460 875 400
Ref. room, pt 1 2500 2400 2960 2870
Sky luminance
(cd/m2)
14400 16400 14700 15200
Indoor temp. (°C) 20,7 20,7 20,4 20,4
Weather type (1–4) 2,60 2,66 2,60 2,60
Cloudiness (1–8) 4,36 4,38 4,36 4,38
Sun in visual field
(1–3)
2,06 1,58 2,08 1,66
3.2 Perception of shading devices
The operation system of both the awning and the venetian
blind was perceived to function well (mean = 6.4/4.9
awning/ ven. blind respectively on the seven-grade scale),
but the awning was most easy to operate (p = 0.000). The
subjects were rather satisfied with their lighting situation
after task 1 (m = 5.5/5.7). After having tried the electric
lighting in task 2, they were somewhat more satisfied
with the lighting condition than before (m = 5.7/6.0). The
possibility to see out was somewhat affected by the solar
shading devices (m = 3.4/3.7). However, there were no
significant differences between the shading devices in
questions 2–4.
There was a grouping effect of age in question 2 (the
satisfaction of the lighting condition): the older the sub-
ject, the more satisfied (p = 0.008). The subjects had then
been divided into three age groups: (1) 23–38 years,
N = 16 ; (2) 39-50 years, N = 18; (3) 51-64 years, N = 16.
The answers were also checked for interaction effects
regarding sex and age, but no such effects were found.
3.3 The position of the shading devices
The shading devices were used frequently to control
glare. They were not only used on clear sunny days, but
also on overcast days. Typical positions of the shading
devices are perhaps best described by frequency distribu-
tions as for the awning’s slope in Figure 6. This shows
that the awning was used by all but 7 subjects, and that

Article IV
219
the most frequent position was when the boom angle was
close to slope 45° (0 degrees boom angle). Only 7 sub-
jects chose to pull it down significantly more than that.
0
2
4
6
8
10
12
14
86.5°-82.5°
82°-75°
74.5°-67.5°
67°-60°
59.5°-52.5°
52°-45°
44.5°-37.5°
37°-30°
29.5°-22.5°
22°-17.5°
Awning's slope (°)
Frequency (N)
Figure 6 Frequency distribution of the slope of the
awning.
For the venetian blind, most subjects did not pull it down
fully. Over 50 % of the subjects pulled it down less than
70 cm compared to the glazing height of 120 cm. (Fig. 7).
Normally, an automatic motorised blind will be pulled
down fully, and the manual override is limited to adjust-
ing the slat angles.
Concerning the slat angles, 75 % of the subjects chose
a slat angle of 30° or larger. Only on 4 occasions did the
subjects choose a negative (sky view) slat angle (Fig. 8).
Beyond slat angles of approximately 45°, the view
through the blind becomes very limited.
0
1
2
3
4
5
6
7
8
9
10
0-9
10-19
20-29
30-39
40-49
50-59
60-69
70-79
80-89
90-99
100-109
110-115
Bottom slat position (cm)
Frequency (N)
Figure 7 Frequency distribution of bottom slat position
of venetian blind.
At a linear regression analysis between the coverage of
the shading device and the measured parameters, no rela-
tionships were found between illuminance levels or sky
luminance. However, a relationship was found between
the coverage and the existence of sunlight patches in the
field of view. This was found both for the awning and for
the venetian blind. The cloudiness also appeared in the
regression equation for the blind. The regression equa-
tions could however only explain a small part of the
variation (adj. R
2
= 0.22–0.34).
Since the existence of sunlight patches appeared in the
regression equations, two new variables were introduced:
the azimuth of the sun’s position (i.e. the angle between
the horizontal projection of the sun and the south axis)
and the perpendicular distance from the wall to the end of
the sunlight patch. However, they did not appear in the
regression analysis.
0
2
4
6
8
10
12
14
16
(-80°)-(-75°)
(-74°)-(-60°)
(-59°)-(-45°)
(-44°)-(-30°)
(-29°)-(-15°)
(-14°)-0°
1°-15°
16°-30°
31°-45°
46°-60°
61°-75°
76°-80°
Slat angle (°)
Frequency (N)
Figure 8 Frequency distribution of slat angles of vene-
tian blind.
Another test was made with a logarithmic transformation
of the measured lighting data. Both the logarithm (to the
base of ten) of the desk illuminance and the sky lumi-
nance appeared in the regression equation, but only for
the venetian blind. The adj. R
2
–value was also low (0.34).
Since these variables only appeared for the blind, the
interpretation of this regression equation was unclear.
3.4 Artificial lighting
The artificial lighting was used in about 30 % of the
cases, just as often in connection with the awning as with
the venetian blind. There was no significant difference in
the use of this complementary lighting between the awn-
ing and the venetian blind. The potentiometer was used
frequently to control the light level, and the average set-
ting was 19 which corresponds to an additional 350 to
500 lux. (The uncertainty is due to how long the lighting
has been turned on).
3.5 Comments
The subjects were encouraged to give their own com-
ments on the questionnaire, and some of the more com-
mon ones have been put together here:
Regarding the artificial lighting: the user’s ability to
dim the electric lighting generated several positive com-
ments, but it became obvious that the chosen lighting
installation was not optimal. Many subjects commented
on the fact that there was no individual light source, just
the ceiling mounted luminaire. Most people wanted more
light on the desk to be able to read, than on the computer
screen, and this was not possible with the chosen solu-
tion. When the subjects chose a setting for the lighting it
was obvious that most persons did this according to the

Energy-Efficient Window Systems
220
computer task, but more light was really needed for paper
tasks.
Regarding the operation of the shades: The awning was
more easy to adjust than the venetian blind as previously
mentioned. Most subjects agreed that the venetian blind
would have been more easy to operate if the function for
adjusting the slat angle had been separated from the
function of bringing the blind up or down, as was the
case. The motor pulling the venetian blind up and down
was also perceived as being too slow.
A few people said that they made a compromise be-
tween glare and the possibility to see out: they would
have been more comfortable with the lighting situation if
they had pulled down the shade even more, but they
chose a more open position in order not to loose too
much of the view out.
On windy days it became apparent that the awning was
much more wind sensitive than the venetian blind. This
lead to a disturbing noise created by the fabric, but even
more disturbing was the light flicker of the sunlight
patch. On sunny afternoons, the sunpatch could often not
be totally removed on the desk due to the oblique angle of
the sun. As the sunpatch was in the field of view, the
flickering effect that was created when the awning was
blowing up and down in the wind gusts was rather dis-
turbing.
The two shading devices also created quite different
impressions of the two rooms. While the grey slats of the
blind did not affect the colours in the room, the fabric of
the awning gave a yellowish tint to the whole room. One
person remarked that it reminded her of an old striped
men’s pyjamas, while for another person it recalled happy
memories of childhood camping trips. A few others
commented on the blinds: for them, the wide slats created
associations with prison bars.
4. DISCUSSION
This study demonstrates the difficulty in predicting when
and how much solar shading devices need to be pulled
down, in order to create a good interior lighting environ-
ment. Glare or contrasts are probably responsible for
when solar shading devices need to be used, but there
seems to be a large individual spread as to how much
glare people tolerate. This is in line with the findings of
Chauvel et al. (1982). Given more measuring points on
luminances in the field of view, it would perhaps have
been possible to find relationships between these and the
use of the shading devices, but in this study no relation-
ships between the sky luminance or the interior illumina-
tion level and the use of the shading devices were found.
One parameter showed a weak relationship to the use of
the shading device: the existence of a sunpatch in the
field of view. But this could only explain a small portion
of the variance. Since the variance among people is large,
even more subjects and more weather situations would
also have been needed.
Generally, solar shading is needed as soon as the sun
enters the room, since the sunpatch will often directly, or
indirectly cause disturbing glare and reflexes in the com-
puter screen. One example is when the sunpatch is on the
wall behind the subject, it will be so strongly lit that it
will cause disturbing reflexes on the screen. This agrees
with the opinions of Littlefair (1999).
The placement of the computer and of the furniture in
relation to the window will of course strongly influence
the glare situation in each individual case. This will, in
turn, affect when and to what extent shading is needed. It
is however clear that computer tasks require some sort of
glare control during a major part of the day, be it interior
or exterior shading devices or curtains, single or in com-
bination.
The fact that the shading devices and the electric
lighting could be controlled was perceived as very posi-
tive. This is a general conclusion in experiments of simi-
lar nature: individual control over physical parameters in
a person’s environment are preferred to having no control
(Bell et al, 1996).
This study also shows that there is no simple relation-
ship between the use of electric lighting and the lighting
parameter that is most often used to estimate the potential
for energy savings of electric lighting through dimming:
the interior illuminance on the work surface. Only when
it became very dark outdoors (and indoors) was there a
trend that the subjects used the electric lighting more
frequently. Also here was there a large individual varia-
tion. However, it did not matter whether it was an awning
or a venetian blind: the same amount of additional elec-
tric lighting was preferred.
Another conclusion is that individual task lighting
should be present so that the lighting on the paper task
can be different from that on the computer screen since
more lighting is generally preferred on the paper than on
the screen.
Clearly, shading devices can have effects on mood and
the general perception of a room. Which effects, and if
these are enough to affect satisfaction and performance
remain to be answered.
This study indicates that several aspects of shading
devices must be considered. Even if the solar shading
properties of shading devices are central, it is also neces-
sary to pay attention to the daylight properties, effects on
view, presence of sunlight patches, adjustability, etc. For
example we found that awnings caused disturbing flick-
ering sunlight patches on sunny, windy days, an effect
which was not present for the venetian blind. However, in
measurements, Wallentén (1999) found that light col-
oured awnings had better shading properties than exterior
venetian blinds.
5. CONCLUSIONS
The main conclusions from this study are:

Article IV
221
• It is difficult to predict the use or need for shading
devices by common measurable factors such as interior
illuminance and sky luminance.
• There is some correlation between the use of shading
devices and the existence of sunlight patches in the room.
• Shading devices are necessary to control glare in the
working environment.
6. ACKNOWLEDGEMENTS
The author gratefully acknowledges Professor Rikard
Küller and Dr. Torbjörn Laike (Lund University, Envi-
ronmental Psychology Unit) for their good advice during
this study.
REFERENCES
Bell, P. Greene, T., Fisher, J. & Baum, A. (1996). Envi-
ronmental Psychology, 4
th
edn. Harcourt Brace College
Publishers, Fort Worth, USA.
Boubekri, M. Hulliv, R. B. Boyer, L. L. (1991). Impact of
window size and sunlight penetration on office workers’
mood and satisfaction. Environment and Behavior 23(4)
474-493.
Boubekri, M. and Boyer, L. L. (1992). Effect of window
size and sunlight presence on glare. Lighting Res. Tech-
nol. 24(2) 69-74.
Bülow-Hübe, H. (1995). Subjective reactions to daylight
in rooms: Effect of using low-emittance coatings on win-
dows. Lighting Res. Technol. 27(1) 37-44.
Chauvel, P. Collins, J. B. Dogniaux, R. Longmore, J.
(1982). Glare from windows: current views of the prob-
lem. Lighting Res. Technol. 14(1) 31-46.
Collins, J. B. (1976). Review of the psychological reac-
tion to windows. Lighting Res. Technol. 8(2) 80-88.
Littlefair, P. (1999). Solar shading of buildings. Building
Research Establishment, Garston, UK.
Norusis, M. J. (1993). SPSS for Windows. Software
Documentation, SPSS Inc.
Rubin, A. I. Collins, B. L. Tibbott, R. L. (1978). Window
Blinds as a Potential Energy Saver–A Case Study. NBS
Building Science Series 112. National Bureau of Stan-
dards, Washington, U.S.A.
Wall M. and Fredlund B. (1999). Solskydd i byggnader.
Verksamhet 1997-1999. (Solar Shading of Buildings:
Research 1997-1999). Report TABK—99/3057. Lund
Institute of Technology, Dep. of Construction and Archi-
tecture, Lund, Sweden.
Vine, E. Lee, E. Clear, R. DiBartolomeo, D. Selkowitz,
S. (1998). Office worker response to an automated vene-
tian blind and electric lighting system: a pilot study. En-
ergy and Buildings (28) 215-218.
Wallentén, P. (1999). Egenskaper hos solskydd – gener-
aliserade mätresultat. (Performance of shading devices –
generalised results of measurements). In Wall M. and
Fredlund B. (1999). Solskydd i byggnader. Verksamhet
1997-1999. (Solar Shading of Buildings: Research 1997-
1999). Report TABK—99/3057. Lund Institute of Tech-
nology, Dep. of Construction and Architecture, Lund,
Sweden.
Errata
The following reference should be added to the list above.
Boyce, P. (1997). Promoting Energy-Efficient Lighting: The Need for
Parallel Processing. Right Light 4, Proc. of the 4th European Confer-
ence on Energy-Efficient Lighting, 19-21 Nov, 1991, Copenhagen,
Denmark. pp 309-313. ISBN 87-87071-73-8

Energy-Efficient Window Systems
222

Article V
223
V

Energy-Efficient Window Systems
224

Article V
225
Validation of daylight module in Derob-LTH
(submitted to Energy & Buildings)
Helena Bülow-Hübe
Department of Construction & Architecture, Lund University
P.O. Box 118, SE-221 00 Lund, Sweden
Abstract
A daylight module has been developed for the energy simulation program Derob-
LTH. Validation has been performed for overcast and clear days against mainly
Radiance for a side-lit room and for a simple atrium. The proposed method is
based on a radiosity method where the distribution of solar radiation (direct and
diffuse components) in a space is calculated and translated to visual radiation
via the luminous efficacy. Last, the radiation levels are amplified using the ratio
of visual-to-solar transmittance to yield the final illuminance level. For the over-
cast sky, the accuracy is acceptable for both vertical and horizontal windows for
the midpoint of the room. For the sunny sky, Derob-LTH accurately predicts
the size and illuminance level of the sunpatch, at least for the tested vertical
window. For the purpose of using it for daylight-linked control of shading de-
vices, the accuracy of the developed model seems sufficient.
1. Introduction
1.1 Background
Within the research and consulting field there has traditionally been a border
between energy and daylight simulations. Requirements on efficient use of en-
ergy in buildings, especially demands to minimise the cooling of commercial
buildings, have however urged a need to estimate the potential of daylight utili-
sation. This has lead to a trend in energy simulation programs to address the
daylight issue and to include routines for the amount of daylight entering a
space in order to estimate potential electricity/energy savings for daylight re-
sponsive lighting and/or shading control systems (Winkelmann & Selkowitz,
1985, Crawley et al., 2001, SBIC, 1996). This is a natural step since about half

Energy-Efficient Window Systems
226
of the solar radiation reaching us at the earth’s surface is within the visible range.
Tools are thus needed to be able to predict the simultaneous benefits of daylight
on overall energy use and peak loads.
Another way of working around the problem has been to start with a light-
ing simulation program, for example ADELINE (Erhorn & Stoffel, 1996), in
order to calculate the hourly lighting energy use, which then is used as input for
an energy simulation. However, this method has the drawback of lacking
interactivity with the thermal calculation since the daylight calculation is per-
formed first. Further, the geometry of the space has to be built up twice.
At Lund University’s department of Construction & Architecture, there is
an ongoing project on solar shading devices (Wall & Bülow-Hübe, 2001). This
project aims at measuring the total solar energy transmittance (g-value) of vari-
ous shading systems, to develop calculation methods, and also to produce a
validated yet simple calculation tool for the prediction of g-values, energy de-
mands (heating and cooling) and indoor temperatures of arbitrary shading sys-
tems. A first version for exterior shading devices has already been released, ParaSol
v1.0, (Wallentén et al., 2000). However, ParaSol is merely a user-friendly inter-
face to an underlying calculation engine, the thermal program Derob-LTH, in
which the developed calculation methods are implemented. In order to make
daylight-linked control of shading devices possible in future releases of ParaSol,
there was a need to develop a daylight calculation module to Derob-LTH, and
this sub-project is reported here.
1.2 Brief description of Derob-LTH
The dynamic energy simulation program Derob-LTH is based on a three-di-
mensional description of the building. It has been proven to accurately estimate
energy demands and temperatures in several measurement projects (Wall, 1996
and Wall & Bülow-Hübe, 2001). The advantage of incorporating a daylight
module into Derob-LTH is that daylight calculations can be performed simul-
taneously with the energy calculations, and thus be used as a control parameter
for moveable shading devices and in the future possibly even electric lighting.
In Derob-LTH, the building model is built of convex surfaces, from trian-
gles to five-sided polygons. A radiosity method (view factors) is used to calculate
the distribution of diffuse radiation, and a ray-tracer follows the direct (beam)
radiation until it hits the first surface inside the space. Then, the reflected radia-
tion is treated as perfectly diffused. Multiple reflections are accounted for, which
means that the reflections are followed until all radiation is either absorbed in
the room surfaces or re-transmitted out again through windows.
The Derob program was first developed in Austin, Texas (Arumi-Noé &
Wysocki, 1979 and Arumi-Noé, 1979). The version Derob-LTH is being devel-
oped and maintained at Lund University, Dept of Construction and Architec-
ture (Kvist, 1999). Improvements during later years include the window mod-

Article V
227
ule, and the shading of diffuse radiation. The window module assigns a thermal
node to each glass surface, and treats convective and radiative heat transfer be-
tween panes in a detailed way (Källblad, 1998). The computer code is also being
modernised at the moment, in order to allow for control strategies of shading
devices, to allow for varying time-steps, and to facilitate future development
and maintenance.
A post-processor, COMFORT, is coupled to Derob-LTH (Källblad, 1996).
This program illustrates graphically the variations in Predicted Mean Vote (PMV),
Predicted Percentage of Dissatisified (PPD) based on ISO 7730, and Global
Operative Temperatures. COMFORT has further limitations than Derob-LTH
on the space that can be analysed, since only one volume at a time, and only
parallelepiped rooms, can be analysed.
In the COMFORT program, the geometry of the room, air and surface tem-
peratures, diffuse solar radiation reflected from the surfaces, and the beam solar
radiation transmitted through windows are taken into account. The observation
point is an infinitesimal cube whose surfaces have equal surface temperatures,
absorptivities and emissivities.
The hypothesis was that by just looking at the solar radiation, it would be
possible to convert the COMFORT program into a simple daylight program.
Due to the construction of the observation point, it would be possible to look at
illuminance values for both horizontal and vertical points at fictitious surfaces.
2. Modelling assumptions
2.1 Treatment of solar radiation
Derob-LTH models the sky as an isotropic (uniform) sky, where the intensity is
changed in relation to the diffuse horizontal radiation. Further, the solar radia-
tion in the weather file is given as direct normal (beam) radiation IN, and diffuse
horizontal radiation, IdH. Solar angles are calculated 4 times per month. The
diffuse and ground reflected radiation reaching the window is calculated accord-
ing to the view factors as seen by the window. Derob-LTH allows shading screens
of arbitrary shape to be put in front of the building, as long as they are convex
polygons. The transmittance is variable from 0 to 100%. This makes them suit-
able for modelling fabrics as in for example awnings. Derob-LTH calculates
shading of both direct and diffuse radiation of such shading screens, see (Wall &
Bülow-Hübe, 2001).
Once diffuse radiation has passed through the window, it is applied as a
diffuse source over the whole window surface. The diffuse radiation of interior
surfaces is also treated so that each surface emits an equal amount of radiation
over the whole surface (only one node per surface element). This is a limitation
of Derob-LTH compared to lighting programs, which usually divide each sur-
face into several nodes.

Energy-Efficient Window Systems
228
The direct radiation is followed through the window until it hits the first
surface in the room. For this direct radiation, 81 (9×9) nodes are applied to each
surface. After the first indoor reflection the light is treated as perfectly diffused.
This is another limitation, which is however shared with other lighting pro-
grams that uses the radiosity method of calculation, e.g. Superlite.
Another limitation when trying to use Derob-LTH as a lighting calculation
program is the choice of the uniform sky model. In energy simulation programs
the uniform sky or the Hay & Davies model are commonly used (Duffie &
Beckman, 1991). In lighting programs however, two (or three) models are used
for an annual calculation depending on whether an hour is classified as sunny or
overcast. The very first difference is that lighting programs uses models that are
based on the luminance distribution of the sky, since the light distribution in a
room strongly depends on this, thus photometric units are used. For overcast
days, the CIE standard sky is the defacto standard, even if the uniform sky often
is an option as well, as it is in Superlite and Radiance. The CIE sky model has a
luminance that is three times higher at the zenith than at the horizon. For nor-
mal side-lit rooms the CIE sky will thus give lower light levels on the work
surface than the uniform sky, an the contrary will happen for atria. While some
authors believe that the CIE model is a good choice for overcast weather (Enarun
& Littlefair, 1995), others seem to be of the opinion that it is an extreme case
(Kittler et al., 1998). Muneer (1988) supported this later opinion in a study of
Japanese data, which showed that the CIE sky always underestimates the aver-
age vertical illuminance. The author was of the opinion that the uniform sky is
a better representation of the overcast situation.
In lighting programs the CIE clear sky with or without sun is usually used
on clear days. This model accounts for circumsolar brightening, (also accounted
for in the Hay & Davies model), the dark spot seen 90° across the sky from the
sun, and horizontal brightening. There are also models for intermediate skies,
and for annual calculations the light distribution is usually based on some sort
of linear combination of various sky models (Littlefair, 1988a). Generally, light-
ing programs are more detailed regarding sky models than energy programs.
Also, the overcast day is very much in focus in lighting programs, while the
sunny day is more interesting for solar engineering purposes.
2.2 Conversion of solar radiation to ”visible” radiation
Since Derob-LTH is a tool for energy simulations, climatic data in a weather file
is always included in a calculation, see Sec 2.1. To convert the irradiance values
in the file to illuminance, the luminous efficacy, K, can be used. This is the ratio
of illuminance to irradiance, and has been measured around the world. Since
the luminous efficacy is dependent of solar altitude, cloud cover, and amount of
aerosol and water vapour content in the atmosphere, several models have been
presented, see for example (Littlefair, 1985 and 1988b) and (Muneer & Kinghorn,

Article V
229
1997). However, Muneer and Angus (1995) have shown that simple models
using average values perform on par with more advanced models. Therefore, in
Derob-LTH average values from measurements in Vaerlose, Denmark (Petersen,
1982) are applied as default values:
Clear sky (IN > 200 W/m2):
For direct radiation IN,K = 103 lm/W
For diffuse radiation IdH,K = 146 lm/W.
Cloudy and intermediate sky:
For direct radiation IN,K = 103 lm/W
For diffuse radiation IdH,K = 121 lm/W.
Other constant values of the luminous efficacy can easily be applied, since they
can be manipulated every hour via the graphical interface. The clear sky level
can also be modified. In the future, it would be a simple task to apply more
advanced models of the luminous efficacy.
2.3 Daylight module interface
The Derob-LTH daylight module DAYLIGHT is currently a stand alone post-
processor to the energy simulation program Derob-LTH. In order to perform
daylight calculations, the volume has to be restricted to a parallel-epiped room.
The calculations can be done for an arbitrary hour of the year, specified by the
user via the graphical interface. Currently, it is possible to look at the light distri-
bution over a fictitious surface in any of the six major orientations at a height
above floor between 0.1 and H-0.1 m in steps of 0.1 m, where H is the room
height.
A critical input is the ratio between the visual Tvis and the direct (or primary)
solar transmittance Tsol,dir of the glazing, here called the VTS (visual-to-solar)
ratio. Our hypothesis was that this ratio could be applied after the solar radia-
tion distribution has been determined, which would allow for “simultaneous”
energy and daylight simulations, thus saving some calculation time, and still
achieving a decent accuracy in the daylight calculations.
The output of the daylight module is a colour graph that shows the distribu-
tion of the illuminance over the specified plane in the chosen direction. This can
either be saved as a bmp-file, or the illuminance values of the grid points can be
saved to a file. The grid width can be changed in steps of 0,1 m, and by setting
the calculation step to anything between 1 and 32 (display) pixels the accuracy
of the view factor/interpolation/ calculation can be varied. A low number will
increase the accuracy, but also the time needed for displaying the graph. A set-
ting of 5 pixels assures a high accuracy, at only a few seconds of calculation time.

Energy-Efficient Window Systems
230
2.4 Limitations in Derob-LTH data input
Today, only the direct solar transmittance and reflectance of windows is given as
input to Derob-LTH. The absorption in the panes and total solar energy trans-
mittance (Tsol,tot or g-value) is calculated by the program. In order to convert the
solar radiation to visible radiation, the higher visual transmittance has to be
taken into account. In the future, this should be given as input to the program
already in the glass library. For the time being, the user has to input the visual-
to-solar ratio, VTS. The main problem with this is that the VTS varies with
incidence angle and glazing type. Especially for modern coated glass (e.g. solar
control glass), the ratio can be quite high. For overcast hours the hemispherical
VTS should be applied, and for sunny hours, the value for the actual incidence
angle towards the window should be chosen.
Another limitation is that the absorptivity of internal surfaces is defined over
the whole short-wave spectrum. The assumption used at present, is that the
absorptivity (or reflectivity) of the surfaces is the same in the visible part of the
spectrum. Normally, the reflectance is slightly higher in the visible region than
over the whole solar spectrum.
3. Validation method
Validation of the daylight module has mainly been performed against other
lighting simulation software, especially Radiance (via the Adeline menu shell,
version 2.0NT). All Radiance runs were performed with parameters set to achieve
a very high accuracy, see (Dubois, 2001). In a few cases Superlite and LESO-
DIAL (Paule et al., 1999) were used. Measurements from the Daylight Labora-
tory at Danish Building and Urban Research (DBUR) in Hoersholm were also
useful.
3.1 Rectangular room: DBUR daylight laboratory
A rectangular room with one façade aperture was modelled. The geometry was
identical to the DBUR Daylight Laboratory in Hoersholm, north of Copenha-
gen (Fig. 1). The facility consists of a full-scale twin room with illumination
sensors both indoors and outdoors (Christoffersen, 2001). A model of the labo-
ratory was built in Radiance, and measurements were used to check this original
Radiance model (see also Dubois, 2001). The reflectance values for interior sur-
faces were Rceil=76 %, Rwall=80 %, Rfloor=10 % (Basic case). The initial Radi-
ance model was then simplified with respect to the outdoor surroundings in
order to facilitate comparisons with Derob-LTH and other programs. This meant

Article V
231
that in all results shown below, the computer simulations were done for a room
with a free horizon and a ground reflectance of 20 %, and without a window
niche.
The sensitivity to different surface reflectances R was studied for two ex-
treme cases: (1) a very white room, R=80 % for all surfaces (White case); and (2)
a completely black room, R=1 % for all surfaces (Black case). The illumination
was evaluated for a work plane height of 0,8 m above the floor.
The window in the laboratory was a double-pane insulating glass unit with
one low-e coating. In both programs the glazing was modelled as two individual
layers, with a resulting visual transmittance of 72 % and reflectance of 14 %
(Radiance), and a direct solar transmittance of 49 % and forward reflectance of
27 % (Derob-LTH) for normal incidence. Although some work has been done
to find simple formulas to model the angle dependent g-value for coated glaz-
ing, this work does not yet include the visual transmittance and reflectance
(Karlsson, et al 2001). The low-e coated glass was thus treated as an ordinary
Fresnel pane in Radiance, and in Derob-LTH the angular dependent properties
were given in the form of a table. The resulting angular dependent properties for
both the visual and solar range, calculated in Derob-LTH, is given in Table 1.
This shows that VTS varies significantly with the incidence angle for the double
low-e glazing. For diffuse light (overcast hours) the hemispherical value of 1.83
applies. Table 1 also shows the values for single clear glass used in Sec. 3.2.
Table 1 Visual and solar transmittance values and the visual-to-solar ratio
for two glazing types as modelled in Derob-LTH.
Double-pane low-e coated window (DBUR laboratory)
Incidence angle 0° 10° 20° 30° 40° 50° 60° 70° 80° 90° Hemis.
Tvis 71.6 71.5 71.2 70.7 69.6 67.2 61.7 48.2 22.8 0.0 62.8
Tsol,dir 49.1 49.0 48.6 47.6 46.2 43.5 38.1 27.5 12.4 0.0 34.3
VTS 1.46 1.46 1.47 1.49 1.51 1.54 1.62 1.75 1.84 - 1.83
Single pane clear float (Simple atrium)
Incidence angle 0° 10° 20° 30° 40° 50° 60° 70° 80° 90° Hemis.
Tvis 90.0 90.0 89.9 89.6 88.9 87.0 82.3 70.4 43.8 0.0 82.6
Tsol,dir 83.0 82.9 82.7 82.2 81.2 79.1 74.3 63.0 38.1 0.0 75.1
VTS 1.08 1.09 1.09 1.09 1.09 1.10 1.11 1.12 1.15 - 1.10

Energy-Efficient Window Systems
232
780 1420
850
1790
6000
3000
3500
Figure 1 Geometry of the DBUR laboratory.
In Derob-LTH, the radiation distribution within the space is calculated using
view factors, and from the intensity of each surface. Since each surface only has
one node regarding the diffuse part of the radiation, this will tend to even out
the light distribution in the room. Especially at the back of the room, the light-
ing will be slightly overestimated. Therefore, three Derob geometries were con-
structed in order to study the effect of increasing surface sub-division. In the
first case (Case×1) no surface subdivision was performed. In the next case
(Case×2) the side walls, floor and ceiling areas were each divided in two surfaces
in order to increase the accuracy of the radiation distribution. Finally, the side
walls, floor and ceiling were each divided into 6 surfaces (Case×6), which means
that the current limitation of maximum 27 surfaces enclosing a volume was
reached (Fig. 2).
Case×1 Case×2 Case×6
Figure 2 Geometry of the three Derob-LTH cases with increasing surface
subdivision.

Article V
233
3.2 Simple atrium
A simple deep atrium was also modelled (Fig 3). The geometry was similar to
the BRE scale model described by Fontoynont et al. (1999). One difference was
that the roof aperture was equipped with a single glazing, since Derob-LTH
does not allow for an open (non-glazed) aperture, see Table 1 for optical proper-
ties. Another difference was that the floor level was always at the bottom of the
deep atrium, but virtual photocells were applied at different “floor” levels, corre-
sponding to the BRE scale model measurements. Therefore, it was not possible
to perform validations directly against the Fontoynont report, instead this was
done against Radiance simulations, using the same geometry as the Derob-LTH
model. The single glazing had Tvis=90 % and Tsol,dir=83 %. The VTS ratio was
thus 1,10 for diffuse light. The reflectance of interior surfaces (walls and floor)
was successively changed from black (4 %) to dark grey (30 %), light grey (48 %)
and white (85 %).
The Derob-LTH model was built similar to case Basic×6 of the DBUR labora-
tory, i.e. with 6 surfaces of equal size on each of the four walls of the atrium.
1
2
3
4
5
6
0
Glazed roof
Virtual
photo
cells
0,25 m
3,5 m
3,5 m
3,5 m
3,5 m
3,5 m
3,5 m
3,5 m
12,0 m
12,0 m
24,75 m
Figure 3 Geometry of the simple atrium.
4. Results
4.1 DBUR model: Overcast day
4.1.1 Comparisons for basic case
The daylight factor of the DBUR lab was calculated in Derob-LTH, LesoDIAL,
Superlite and Radiance. Measurements for a CIE overcast day were also avail-
able. For the Derob-LTH calculations, an hour was found in the climate file
were there was no direct irradiance, and where the diffuse irradiance IdH was
100 W/m2. With the luminous efficacy set to 121 lm/W, this corresponds to an
overcast sky of 12100 lux. Fig. 3 shows that Derob-LTH underpredicts the day-

Energy-Efficient Window Systems
234
light factor close to the window, and overpredicts it at the back of the room. The
effect of increasing surface sub-division is shown in Table 2. For case Basic×2,
the daylight factor increases with about 0.5 percentage-points at 1 m from the
window, and decreases with the same amount at 5 m from the window com-
pared to case Basic×1. However, the results between the two cases Basic×2 and 6
were just marginally different.
0
5
10
15
20
25
0123456
Distance from window (m)
Daylight factor (%)
Slite - CIE overcast
Leso-DIAL
Measured
Radiance - CIE overcast
Derob-LTH Basic*1
Figure 4 Comparison of Derob-LTH with other programs and measurements
in Hørsholm, DK. For Superlite (Slite), Radiance and Leso-DIAL
the calculations were done for the CIE overcast sky model.
Since Derob-LTH uses an isotropic (uniform) sky model, the Radiance calcula-
tions were redone for this sky. With the uniform sky model the daylight factor
will always be higher, about 1.5 percentage points at 1 m from the window, and
0,4 percentage points at the back of the room (Fig. 5). The uniform sky assump-
tion thus seems to explain why the illuminance is higher at the back of the room
in Derob-LTH, but it does not explain the low illuminance seen close to the
window.
A hypothesis was that this is due to the way diffuse radiation is treated. In
Derob-LTH, the window is modelled as a diffusing surface regarding the diffuse
part of solar radiation. Thus, for a vertical window, the mostly downward flow-
ing light flux is redistributed to being forward flowing, which is illustrated in
Fig. 6. For energy calculations this is an acceptable simplification, but for day-
light calculations this has to be investigated, since the overcast day is the ”design
day” and thus used for analysis of daylight.

Article V
235
0
5
10
15
20
25
0123456
Distance from window (m)
Daylight factor (%)
Radiance - uniform sky
Radiance - CIE overcast
Derob-LTH Basic*6
Figure 5 Comparison of daylight factor between Derob-LTH case Basic
×
6
and Radiance with the uniform and CIE sky model respectively.
Figure 6 Schematic drawing showing the principle of transmittance of verti-
cal glazing for diffusing glass (left) and clear glass (right). On the
left side of the glazing is shown the intensity of the radiation arriv-
ing at a small surface element from a uniform sky and ground.
First, a comparison was made with a Radiance calculation where a diffusing
curtain with 100% diffuse transmittance was placed close to the inside of the
window. The uniform sky was chosen, since Derob-LTH uses this model. From
Table 2, and Fig. 7 it is seen that the diffusing curtain reduces the daylight factor
in Radiance from 11.1 to 8.1 % at 1 m from the window, which is close to the
Derob-LTH result of 7.6 %. With the VTS ratio set to 1.83 (hemispherical), the
average relative error was -8 %, which is quite satisfying.

Energy-Efficient Window Systems
236
Table 2 Daylight factor for the basic case (Rwall=80 %, Rceil=76 %,
Rfloor=10 %) for different sky models and window types (Radi-
ance) and surface sub-divisions (Derob-LTH). The ratio VTS
was set to 1.83.
Radiance Derob-LTH
Distance CIE overcast, Uniform sky, Uniform sky, Basic×1 Basic×2 Basic×6
from window clear window clear window w. diffusing
(m) curtain
1 9.66 11.13 8.06 7.06 7.47 7.56
2 3.39 4.43 4.07 3.41 3.70 3.71
3 1.75 2.40 2.65 2.48 2.47 2.43
4 1.16 1.61 1.96 2.17 1.84 1.80
5 0.92 1.28 1.64 2.03 1.57 1.51
From Fig. 7 it is also seen that the intersection point between Derob-LTH and
a Radiance run for a uniform sky and clear window is very close to the midpoint
of the room. For the CIE sky model, the intersection point is at 1.7 m distance
from the window.
4.1.2 Sensitivity to surface reflectance
The comparison between Derob-LTH and Radiance shown below were done
for equal boundary conditions, i.e. uniform sky and a diffusing curtain inside
the window. The Derob-runs were done with the geometry for Case×6, i.e. 6
surfaces on side walls. From Fig. 8 it can be seen that Derob-LTH responds well
to a change in surface reflectance, and the results are very similar to Radiance
results. The absolute differences are small, see Fig. 9, and correspond to relative
differences between –4 % and –8%. The somewhat erratic shape of the curves
in Fig. 9 is intrinsic to the stochastic ray-tracing procedure used in Radiance.

Article V
237
0
5
10
15
20
25
0123456
Distance from window (m)
Daylight factor (%)
Radiance - uniform sky
Radiance - w. diffusing window
Derob-LTH Basic*6
Figure 7 Comparison of daylight factor between Derob-LTH and Radiance
with and without a diffusing window. Uniform sky model.
0
2
4
6
8
10
12
14
16
18
20
0123456
Distance from window (m)
Daylight factor (%)
Radiance white 80%
Derob-LTH, white 80%
Radiance black 1%
Derob-LTH, black 1%
Figure 8 Comparison of daylight factor for black and white room. Window
with diffusing curtain and uniform sky model. The geometry of
Case
×
6 was used in the Derob calculation.

Energy-Efficient Window Systems
238
-0.8
-0.7
-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0.0
0.1
0123456
Distance from window (m)
Absolute difference (Der-Rad) (%-points)
White 80% Basic 76/80/10% Black 1%
Figure 9 Relative difference in daylight factor between Derob-LTH and
Radiance for similar boundary conditions. The somewhat erratic
shape of the curves are intrinsic to the ray-tracing procedure used in
Radiance.
4.2 DBUR model: Sunny day
The illuminance was calculated for a clear sunny day in Derob-LTH and Radi-
ance. March 12 at 12 for Lund 1988 was chosen, since it was an hour with high
direct and low diffuse radiation. The measured irradiance was IN =922.3 W/m2,
and IdH=76.0 W/m2. The solar altitude calculated in Derob-LTH was 28.4°
and the azimuth –8.5°. The measured direct normal irradiance IN thus corre-
sponds to 439.0 W/m2 on the horizontal plane, and a horizontal global illumi-
nance of 56 315 lux with the assumed luminous efficacies mentioned above.
For the Derob-LTH calculations, the geometry of Case×6 was again used, and
the VTS ratio was set to 1.49 (incidence angle 30°). Since the direct part is
dominating for this hour, the Radiance calculations were performed with clear,
not diffusing glass.
The indoor illuminance expressed as a percentage of the outdoor global illu-
minance is shown in Fig. 10 for the basic case. The general trend is that Derob-
LTH shows a good agreement of the illuminance in the sunpatch, especially for
the basic and black cases. Behind the sunpatch, it seems like Derob-LTH
overpredicts the lighting when the inter-reflected component is high (white case),
and underpredicts it slightly for the other cases, see Fig. 11. The differences seen
are probably due to modelling differences regarding the sky model (direct com-
ponent), and the treatment of the diffuse radiation in Derob-LTH. The angular
dependent transmittance might also be different. Another difference is that the

Article V
239
VTS ratio was set to 1.49 for both direct and diffuse radiation, while the factor
1.83 should be applied to the diffuse part. This means that the diffuse illumi-
nance component is underestimated in Derob-LTH.
1
10
100
0123456
Distance from window (m)
Percentage of outdoor illuminance (%)
Derob-LTH, basic
Radiance, basic
Figure 10 Percentage of indoor illuminance to outdoor global illuminance (so-
lar factor) for sunny conditions, March 12 at 12. Comparison be-
tween Derob-LTH and Radiance for the basic case (internal re-
flectance 76/80/10 %).
-5
-4
-3
-2
-1
0
1
2
3
4
5
0123456
Distance from window (m)
A
bsolute difference (Der-Rad) (%-points)
White
Basic
Black
Figure 11 Absolute difference in solar factor (percent of outdoor global illumi-
nance) between Derob-LTH and Radiance for sunny conditions,
March 12 at 12 at three internal reflectances: white (80 %), basic
(76/80/10 %), and black (1 %).

Energy-Efficient Window Systems
240
4.3 Simple atrium: Overcast day
Since the simple atrium only has a horizontal glazed roof the incoming diffuse
radiation is already isotropic when reaching the window in the case where the
uniform sky model is used. Therefore, it does not suffer from the limitation that
the glazing diffuses the diffuse part of the solar radiation, see Fig. 12. Hence, the
Radiance calculations were performed for clear glass. Figs. 13 and 14 show a
comparison of results between Derob-LTH and two sky models in Radiance for
the black and white room respectively. The agreement is rather good for the
uniform sky. This is demonstrated again in Fig. 15, which shows the relative
error between Derob-LTH and Radiance for the uniform sky model for 4 differ-
ent surface reflectances. The error is smallest for the white atrium, but never
exceeds 10 % in any case. When the CIE overcast sky is used, the daylight factor
is always higher. At the bottom of the atrium, level 0, the difference is about 2-
3 percentage points. Close to the top, level 6, the difference is larger, about 5-6
percentage points.
Figure 12 Schematic drawing showing the principle of radiation from a uni-
form sky arriving at a surface element of a horizontal diffusing
glazing, and leaving it diffused.

Article V
241
Black atrium (4 %)
0
10
20
30
40
50
60
70
80
90
100
0123456
Cell level
Daylight factor (%)
Radiance CIE
Radiance uniform
Derob-LTH
Figure 13 Daylight factors for black atrium (R=4 %) calculated in Derob-
LTH compared to Radiance (for the two sky models CIE overcast
sky and uniform sky respectively).
White atrium (85 %)
0
10
20
30
40
50
60
70
80
90
100
0123456
Cell level
Daylight factor (%)
Radiance CIE
Radiance uniform
Derob-LTH
Figure 14 Daylight factors for white atrium (R=85 %) calculated in Derob-
LTH compared to Radiance (for the two sky models CIE overcast
sky and uniform sky respectively).

Energy-Efficient Window Systems
242
-5.0
-4.0
-3.0
-2.0
-1.0
0.0
1.0
2.0
0123456
Cell level
A
bsolute difference (%-points)
White 85% Light grey 48%
Dark grey 30% Black 4%
Figure 15 Absolute difference in daylight factor between Derob-LTH and Ra-
diance simulations for various interior reflectances. Uniform sky
model.
5. Discussion
The developed Derob-LTH daylight module shows a good agreement with simu-
lation results in Radiance for similar boundary conditions. The proposed visual-
to-solar ratio VTS works rather well, but the process should be automated, since
it is important to choose the VTS according to the current incidence angle. It
was also seen that the calculated illuminance was very sensitive to the VTS ratio,
since this ratio can be high for real coated glazing. The exact modelling of the
angle dependent transmittance of the glazing must thus be known both for the
visual and the solar range. Further work is needed to develop such models for
real glazing, especially for modern coated glass.
On overcast days, Derob-LTH suffers from some limitations, which regard
the way diffuse radiation is handled. Firstly, Derob-LTH uses the uniform sky
model, which was abandoned quite long ago within the lighting community.
The use of the uniform sky compared to the CIE overcast sky will overestimate
the lighting in rooms with vertical windows, and underestimate it in rooms with
horizontal windows, such as atria. This is due to the nature of the CIE sky,
which for the same outdoor illuminance will have a brighter sky at the zenith,
and a darker sky at the horizon. However, since purely overcast days are only a
part of all available daylight hours in a year, the problem is perhaps not that
severe. There is also a great variety of the luminance distribution of real skies,
and both the CIE and the uniform sky are just two theoretical varieties.

Article V
243
Secondly, Derob-LTH diffuses the diffuse radiation when it passes through
the glazing. This will not affect the amount of radiation entering a room, but
may lead to a change in the main direction of the light flow. This poses no
problem for horizontal windows facing an unobstructed uniform sky. However,
for vertical windows which “see” part of the ground and just half of the sky, the
mainly downward flowing light will be redistributed to being forward flowing.
Therefore, the lighting on a horizontal work plane will be underestimated close
to the window, and more light will instead reach the ceiling and the back of the
room. This seems like the largest drawback of the current limitations in Derob-
LTH. However, for a point close to the midpoint of the room, these two main
differences cancel each other out.
During sunny conditions, Derob-LTH performs rather well, especially within
the sunpatch. The Derob-values can therefore be used as a trigger for control-
ling shading devices. Behind the sunpatch, the errors are larger, especially for
extremely white rooms with a large inter-reflected component.
Historically, the lighting community has been very focused on daylight fac-
tors and thus on overcast days, since these are the most critical for lighting a
room. From an energy point of view, the clear sunny day is most interesting,
either in order to capture the solar heat, or to prevent from overheating. During
a whole year however, a large number of hours fall between these extremes. For
typical climate years in Sweden (Lund, Stockholm, Gothenburg and Luleå 1988),
slightly less than half of all hours with measurable diffuse radiation can be clas-
sified as overcast (criterion IN<10 W/m2), and about 30-35% of the hours are
clear (criterion IN>200 W/m2). Therefore, it will become necessary to automa-
tize the selection of the proper sky model, or combination of these. This will
require some further work regarding the proper criteria for the selection process
(such as sunshine probability or cloud ratio), and the implementation of these
into Derob-LTH.
6. Conclusions
A daylight module has been developed and validated for Derob-LTH. The pro-
posed method is based on a radiosity method where the, in the energy program
already calculated distribution of solar radiation in a space, is translated to visual
radiation via a luminous efficacy. Last, the radiation levels are amplified using
the ratio of visual-to-solar transmittance to yield the final illuminance level. For
the overcast sky, the accuracy is acceptable for both vertical and horizontal win-
dows for the midpoint of the room. For the sunny sky, Derob-LTH accurately
predicts the size and illuminance level of the sunpatch, at least for the tested
vertical window. For the purpose of controlling shading devices, the accuracy of
the developed model seems sufficient.

Energy-Efficient Window Systems
244
In order to develop the model into a full-fledged lighting calculation pro-
gram, further work is needed mainly regarding the transmittance of diffuse light,
automatic surface subdivision and perhaps also regarding sky modelling. How-
ever, some previously mentioned works seem to indicate that the choice of the
uniform sky is justified (Kittler et al, 1998; Muneer, 1998).
Further work is also needed regarding input of visual data for the glazing,
and for selecting the proper combination of sky models for intermediate days to
facilitate annual energy calculations.
Acknowledgements
The author would like to thank Kurt Källblad who developed the Derob-LTH
daylight module, Marie-Claude Dubois for her assistance in Radiance model-
ling and Bengt Hellström for providing useful discussions. I also acknowledge
the gracious support of the Swedish Council for Building Research and The
Swedish National Energy Administration.
References
F. Arumi-Noé, M. Wysocki, The Derob system volume I, User’s manual for the
Derob system, University of Texas, School of Architecture, Numerical Simu-
lation Laboratory, Austin, TX, USA, 1979.
F. Arumi-Noé, The Derob system volume II, Explanatory notes and theory,
University of Texas, School of Architecture, Numerical Simulation Labora-
tory, Austin, TX, USA, 1979.
J. Christoffersen, Danish Building and Urban Research – Daylight Laboratory,
(internal report), Danish Building and Urban Research, Hørsholm, Den-
mark 2001.
D.B. Crawley, L.K. Lawrie, F.C. Winkelmann, W.F. Buhl, Y. Joe Huang, C.O.
Pedersen, R.K. Strand, R.J. Liesen, D.E. Fisher, M.J. Witte, J. Glazer,
EnergyPlus: Creating a new-generation building simulation program, En-
ergy & Buildings 33 (4) (2001) 319-332, Special Issue: Proceedings of the
6th International IBPSA Conference, Kyoto, Japan, 13-15 September, 1999.
M.-C. Dubois, Impact of shading devices on daylight quality in offices,
Simulations with Radiance, Manuscript to Report TABK--01/3062, Div.
of Energy and Building Design, Lund Institute of Technology, Lund Uni-
versity, Lund, Sweden, 2001. (forthcoming).
J.A. Duffie, W.A. Beckman, Solar Engineering of Thermal Processes, 2nd ed,
Wiley Interscience, 1991.

Article V
245
D. Enarun, P. Littlefair, P. Luminance models for overcast skies: Assessment
using measured data, Lighting Research and Technology, 27
(1) (1995) 53-
58.
H. Erhorn, J. Stoffel, (Eds.), ADELINE 2.0, Advanced daylighting and electric
lighting integrated new environment, Documentation of the software pack-
age ADELINE, version 2.0 (9 volumes), IEA Solar Heating and Cooling
Program, Task 12: Solar Building Energy Analysis Tools, 1996.
M. Fontoynont, P. Laforgue, R. Mitanchey, M. Aizlewood, J. Butt, W. Carroll,
R. Hitchcock, H. Erhorn, J. De Boer, M. Dirksmöller, L. Michel, B. Paule,
J.L. Scartezzini, M. Bodart, G. Roy, Validation of daylighting computer
programs, IEA SHC Task 21/ ECBCS Annex 29 Daylight in buildings,
Ecole Nationale des Travaux Public de l’Etat, Département Génie Civil et
Bâtiment, Vaulx-en-Velin Cedex, FR, 1999.
J. Karlsson, M. Rubin, A. Roos, Evaluation of predictive models for the angle-
dependent total solar energy transmittance of glazing materials, Solar En-
ergy, 71 (1) (2001) 23-31.
R. Kittler, S. Darula, R. Perez, A set of standard skies. Characterizing daylight
conditions for computer and energy conscious design, Final report, Insti-
tute of Construction and Architecture, Slovak Academy of Sciences and
Atmospheric Sciences Research Center, State University of New York.
Bratislava, Polygrafia SAV, 1998.
H. Kvist, Derob-LTH for MS Windows, User manual Ver 99.02, Dept. of Con-
struction & Architecture, Lund Institute of Technology, Lund University,
Lund, Sweden, 1999.
K. Källblad, Comfort – A Computer Program for Thermal Comfort Simula-
tion, Report TABK--96/3035, Dept. of Building Science, Lund Institute of
Technology, Lund University, Lund, Sweden, 1996.
K. Källblad, Thermal Models of Buildings. Determination of Temperatures,
Heating and Cooling Loads. Theories, Models and Computer Programs,
Doctoral Thesis, Report TABK--98/1015, Dept. of Building Science, Lund
Institute of Technology, Lund University, Lund, Sweden, 1998.
P.J. Littlefair, The luminous efficacy of daylight: a review, Lighting Research and
Technology 17 (4) (1985) 169-181.
P.J. Littlefair, Predicting hourly internal daylight illuminances for dynamic en-
ergy modelling, BRE 114/6/4, BRS 262/88, Building Research Establish-
ment, Garston, UK, 1988.
P.J. Littlefair, Measurements of the luminous efficacy of daylight, Lighting Re-
search and Technology 20
(4) (1988) 177-188.

Energy-Efficient Window Systems
246
T. Muneer, Evaluation of the CIE overcast sky model against Japanese data,
Energy and Buildings 27 (2) (1998) 175-177.
T. Muneer, R.C. Angus, Luminous efficacy: Evaluation of models for the United
Kingdom, Lighting Research and Technology 27 (2) (1995) 71-77.
T. Muneer, D. Kinghorn, Luminous efficacy of solar irradiance: Improved mod-
els, Lighting Research and Technology 29
(4) (1997) 185-191.
B. Paule, J.-L. Scartezzini, M. Reynolds, N. Baker, Handling daylight in the
early design stage. In: Proc. of EuroSun 2000, Copenhagen, Denmark, June
19-22, 2000.
E. Petersen, Solstråling og dagslys – målt og beregnet, (Solar radiation and day-
light – measured and calculated), Report nr 34, Lysteknisk Labora-
torium,.Lyngby, Denmark, 1982. (in Danish).
SBIC, Designing Low-Energy Buildings: Passive Solar Strategies & ENERGY-
10 Software. Integrating Daylighting, Energy-Efficient Equipment, and
Passive Solar Design in Commercial, Institutional, and Residential Build-
ings, Passive Solar Industries Council, Washington DC, USA, 1996.
M. Wall, Climate and Energy Use in Glazed Spaces, Doctoral Dissertation, Re-
port TABK--96/1009, Dept. of Building Science, Lund Institute of Tech-
nology, Lund University, Lund, Sweden, 1996.
M. Wall, H. Bülow-Hübe, (Eds.), Solar protection in buildings, Report TABK-
-01/3060, Lund, Sweden: Div. Energy and Building Design, Div Building
Science, Lund Institute of Technology, Lund University, 2001.
P. Wallentén, H. Kvist, M.-C Dubois, ParaSol-LTH: A User-friendly Computer
Tool to Predict the Energy Performance of Shading Devices. In: Proc. of the
International Building Physics Conference, Eindhoven, Holland, Septem-
ber 2000.
F.C. Winkelmann, S. Selkowitz, Daylighting Simulation in the DOE-2 Build-
ing Energy Analysis Program, Energy and Buildings 8 (4)(1985) 271-286.

Article V
247

Energy-Efficient Window Systems
248