Some preliminary studies of the thermal properties of Devon cob walls

Full text

1
Some preliminary studies of the thermal properties of Devon cob walls.
Steven Goodhew, Richard Griffiths*, David Short and Linda Watson.
*Author to whom correspondence should be addressed
Joint Schools of the Built Environment, School of Architecture, University of Plymouth,
Hoe Centre, Notte Street, Plymouth, Devon, PL1 2AR, UK.
Tel: +44(0)1752 233605, Fax: +44(0)1752 233634 e-mail: rgriffiths@plymouth.ac.uk
This paper was published in the Proceedings of the Terra 2000 8th International Conference on the study and conservation of earthen
architecture, Torquay. James and James, London 2000, pp 139 - 143.
ABSTRACT
Little has been published about the thermal properties of existing mass earth walls. In order to model the
thermal behaviour of an enclosure made from traditional cob wall both the thermal conductivity and thermal
diffusivity are required. The earth building research team at Plymouth has been investigating a simple
thermal probe technique, to be used in the field, to determine these two thermal properties accurately,
quickly and with the minimum of disturbance to the sample wall.
This paper presents some preliminary results from three studies of real cob walls found at Bovey Tracy,
Rezare and Sandford. Measured values of the thermal conductivity and diffusivity are given, while the
Building Research Establishment Admittance method software is used to calculate the thermal
transmittance (U-value), the thermal admittance (Y-value), the decrement factor and the lag time in hours
for the walls. Comparisons are drawn with the thermal data of other forms of wall construction, like light-
weight concrete block and brick.
Key words: thermal cob measurement earth walls
Introduction
The University of Plymouth’s School of Architecture Centre for Earthen Architecture (CEA) has carried out
research into the various aspects of vernacular architecture typical to Devon. The local monolithic form of
unbaked walling is known as cob and has featured in several projects investigated by the CEA. The
relationship between the moisture content of these walls and their structural properties has been
investigated by Greer, (1996). The pathology of structural failure of cob walls has been researched and
patterns of failure have been analysed by Keefe (1998) and Keefe et al (1999). Links have been made
between the age, building form and the geographical location of cob buildings using Geographic
Information Systems (GIS), by Ford et al (1999). A study of the effects of straw content upon the
mechanical properties of cob walls is on-going and will be complete soon, Coventry, unpublished. This
paper concerns the thermal properties of cob, as there is little published data, and a technique has been
developed to measure the properties in situ. The background to this study may be found in Goodhew
(1999).
The two thermal properties of a building material that affect the thermal performance of a building are (i) the
thermal conductivity, (

), and (ii) the thermal capacity, (c). When studying the thermal performance of a
building material, both the rate of heat transfer, which is a function of conductivity, and the quantity of heat
required to raise the temperature of the material, the thermal capacity, are required.

2
This paper will report the application of a time dependent probe technique to measure these two thermal
properties. The probe is placed in the material and receives a constant supply of power. The rise in probe
temperature will depend upon the rate at which heat can flow away from the probe, and on the thermal
capacity of the material. By recording the probe temperature as a function of time for several minutes, the
theory suggests that both the thermal conductivity and the thermal diffusivity can be obtained from these
measurements. The thermal diffusivity () is the thermal conductivity divided by the volumetric heat
capacity, or density , multiplied by the thermal capacity of the material. When looking for a suitable
technique for studying the thermal behaviour of cob walls, ideally five criteria are to be satisfied. The
measurement procedure (1) must be non-destructive, (2) it must not influence the material being
measured, (3) must give representative values for both (

) and (), (4) should be rapid and affordable and
(5) allow in situ testing.
A time dependent thermal probe technique theoretically meets these criteria. For example, if a small
diameter probe is used, then it may be assumed that one 3 mm diameter hole 70 mm long will not damage
a wall that may have a thickness as large as 650 mm. The measurements can be carried out with a
sufficient delay as not to affect or be affected by previous or future readings. The technique will give values
for thermal conductivity and thermal diffusivity. Previous researchers have used time periods of less than
an hour with relatively inexpensive equipment. The short experimental periods required for data collection
allied with the low power inputs needed, allow a small test apparatus with a self contained power supply to
be used. This gives the technique the unique advantage of portability and flexibility in carrying out in situ
studies. Moreover, because of the short time scale used in the probe measurements, with associated
small rises in temperature in the specimen, the sample moisture content is less disturbed.
Previous applications of this time dependent thermal probe technique have provided a number of solutions
to many of the theoretical problems that are inherent in its successful use. However, one vital element is
still an obstacle to obtaining realistic thermal data from in situ tests, that of the interface between the probe
and the material being examined. The effect of this constraint, the thermal probe conductance (H) and the
effects of different power levels supplied to the probe are explored in other publications. This paper
concerns some of the preliminary studies of real walls, while the development of the technique under
laboratory conditions will be given elsewhere.
The measurement technique.
The time dependent thermal probe technique consists of a line heat source supplied with constant power Q
Watts per unit length of the probe. This probe is placed in a specimen, which is assumed to be of infinite
extent and at a uniform initial temperature. The rise in temperature  of the probe is recorded and
analysed. Following many previous authors, for example Batty et al (1984) and Blackwell (1954), a graph
of rise in probe temperature as a function of the natural logarithm of the elapsed heating time becomes a
straight line. The slope of the line is a function of the specimen thermal conductivity, while the intercept on
the temperature axis is a function of the thermal diffusivity and the contact thermal conductance between
probe and specimen. Laboratory studies have shown that using careful measurements and modern
computing routines, that are readily available on desk top computers to collect and analyse the data, it is
possible to determine both thermal parameters,  and , without the need to make the assumption that the
probe - specimen conductance is very large. This is usually the assumption made by workers employing
this technique.

3
The probe consisted of a stainless steel tube 70 mm long and 3 mm in diameter, with a mild steel hair-pin
heater mounted along the tube axis. The probe was sealed at one end with an inert fast-setting resin
based adhesive. At the other a plastic grommet protects the cabling for the power leads and thermocouple.
This thermocouple, a type E, was mounted at the mid point of the heater. Current and potential leads
connected the heater to the stabilised direct current supply and the datalogger monitored both the heater
current and potential difference, as well as the probe temperature. This information was recorded every
second for experimental times extending from 100 seconds before the heater current is established, to
3600 seconds of heating. As has been previously stated the simplicity, size and operation of the probe
lend itself to the in situ testing of cob walls. To ensure the accuracy of the measured thermal data four
issues are addressed:
 The thermal conductance between the probe and the specimen should be a large as possible, although
it does not have to be assumed infinite.
 The initial temperature of the sample must be known and steady, and hence the data collection for 100
s prior to switching on the heater current.
 The lack of variation of background and sample temperature needs to be confirmed.
 The power supplied to the probe heater must remain constant.
The measurement and recording equipment was based around a Geologger Datataker, model 615. This
instrument can record temperatures from a range of different thermocouple types at different time steps.
Small electrical current and voltage values can also be measured and stored to ensure an accurate
determination of the power dissipated in the probe heater. This experimental arrangement was
supplemented with other devices, such as mercury-in-glass thermometers to check the environmental
conditions around the specimen.
The measurement technique was tested in the laboratory, by studying materials with known thermal
properties under controlled conditions, before the field work was undertaken.
Field Studies
Field Study One: Cob barn at REZARE, near Launceston, Cornwall, grid reference SX359793
The first field study was carried out upon the walls of 150 year old cob barn situated north of Rezare and
south of Launceston. The barn is structurally sound, has a replacement profiled metal sheeted roof, but is
in need of some minor repairs to the internal part-flooring. The upper storey of the barn was selected as
the site for the study with measurements taken from the internal face of the north wall. The barn measures
10 m long, 5 m wide, and has cob walls of height varying between 2.0 m and 2.4 m built on a stone plinth.
The height of this stone plinth varied depending upon the slope of the ground. Because of the barn’s close
proximity to a fast flowing stream, the lower parts of the structure are sometimes flooded in the winter
months and all readings were taken at sufficient height above this level to reduce any affects upon the
conductivity measurements. The data was collected in February 1998.
A hole of the same dimensions as the probe was carefully drilled by hand into the wall and the probe
inserted taking care not to jeopardise the thermal contact between the probe wall and the sides of the
drilled hole. The exterior end of the probe and surrounding wall area were covered with some expanded
polystyrene foam insulation to reduce external thermal changes influencing the probe results. The cob
walls were approximately 450 to 500 mm thick and so any thermal affects on the probe from the other
exterior surface were reduced by the large amount of cob material between the probe end and the outside.

4
The time dependent probe measurements were undertaken in the evening, allowing 24 hours between
each data collection, so that the wall and probe could be allowed to regain thermal equilibrium.
Field Study Two: Cob block summerhouse, Bovey Tracey, Devon, grid reference SX816784.
This study was undertaken on separate cob blocks making up the structural walls of a partially completed
summerhouse. The single storey summerhouse is 2.5 m long, 1.5 m wide and approximately 2.0 m high.
The unbaked earth section of the wall varies from 1.2 m to 1.4 m high, with wall thickness 220 mm founded
upon a stone and concrete plinth approximately 0.8 m in height. At the time of data collection the roof
structure had not been completed and the walls were protected with temporary boarding to prevent rain
damage. The walls were built from cob blocks using soil rich in clay from the Tedburn St Mary area close
to Exeter. The data was collected in February 1997 and March 1998.
A similar testing routine to that undertaken for the Rezare study was adopted, the probe was shielded with
insulation and a 24 hour period was allowed between measurements.
Field Study Three: Cob barn/stable Frogmire, Sandford, near Crediton, Devon, grid reference SS825015.
The third wall studied was built from a different soil, and forms part of a complex of buildings with a well
documented history just north of Crediton in Mid-Devon. The buildings date from the late sixteenth century
and have been dated from various architectural features. This building has cob walls made from the local
soil based upon red sandstone and has a lower clay content than both the cob summer house blocks, and
the cob barn at Rezare. The probe site for this study was an internal wall facing north with very little
temperature variation due to solar gain. The wall is about 2.5 m high and 550 mm thick. Two thermal
probes were used in this study, placed 1 m and 2 m up the wall face from the floor and insulated from the
exterior, as previously. The data collection carried at Frogmire was initially in drilled holes without any
compound used between the probe and sample. The later experiments used a compound in the probe
holes to improve the thermal contact between the probe and wall material.
The data was collected in the summer 1998.
Data analysis and results
The raw data was down-loaded from the field datalogger using a lap-top computer. This data was transferred
to an Excel spread-sheet, and standard routines contained in Excel were used to analyse the data. This
analysis had a number of elements, and these are now described.
The probe temperature was converted to a rise in temperature , and this rise was plotted as a function of the
natural logarithm of the elapsed time t seconds. This graph enabled a visual check of the data, to confirm that
there were no anomalies, and that the trace was of the expected shape. The theoretical analysis of this time
dependent thermal problem shows that the rise in probe temperature as a function of the heating time has an
approximate form:
 = A { ln t + B + (1/t) [ C ln t + D] + (1/t)2 F} 1
which is true after a short period of time has elapsed, in this case after about one minute. Moreover, at longer
times, say 150 to 3000 seconds, the terms in equation 1 that contain 1/t and 1/t2 may be ignored. Over this

5
time interval we would expect the graph to display a straight line of slope A and intercept on the temperature
axis of AB, where:
A = Q / ( 4  ) and B = [ ln (4  / r2 ) -  + 2  / ( r H) ] 2
where r is the radius of the probe,  is Euler’s constant, 0.57722, and H is the probe to specimen contact
thermal conductance. A regression analysis was performed on the data and values of A and B determined.
The thermal conductivity is obtained from the value of A. However, the expression for B contains two
unknowns, the thermal diffusivity , and the probe conductance H. Following Blackwell (1954) a second graph
was produced where Blackwell’s parameter YB is plotted against the square root of the elapsed heating time at
short times. The resulting straight line has an intercept on the YB axis at a value of H. With H, we can return to
the expression for B, in 2 above, and determine the thermal diffusivity . The parameter YB was found to be
sensitive to small errors in the time origin, and the linear relationship between YB and t0.5 was only observed
from about 10 to 60 seconds. We therefore used a second approach which determined all three unknowns, ,
, and H, using the iterative method in the Solver routine on Excel. This routine looks for the optimum values
of the three parameters by a guided trial and error method. We have used this Solver routine with all four
constants in equation 1, that is using A, B, C and D. The curve fitting has been successful from relatively short
times, 50 seconds, up to and beyond 3000 seconds. Densities of the samples, , were determined in the
laboratory on small samples of the walls.
For the three cob walls, these preliminary studies have given values of the thermal conductivity and thermal
diffusivity, as in table 1. The specific heat capacity, c, of the wall materials was calculated from c =      ).
Many values used to represent the thermal conductivity of unbaked earth walling are often approximations
derived from other materials of a similar density, although some work has been done by Minke (1994), who
refers to many properties of soils and their behaviour when constituted into unbaked earth walling. Minke
discusses the relationship between the density of the soil used and the thermal conductivity of the resulting
wall. He states that as the density increases from 1200 to 2000 kgm-3 the thermal conductivity also increases
from 0.47 to 0.93 Wm-1K-1. While the thermal conductivity values achieved in this work are smaller than
suggested by Minke, nevertheless the trend of increasing conductivity with density is observed.
Thermal behaviour
Much anecdotal evidence exists suggesting that buildings constructed from unbaked earth maintain a ‘steady’
internal thermal environment. However, very little work has been carried out to confirm this. The use of earth
walling materials for ‘damping’ variations in temperature within buildings hints at the importance of establishing
the value of the thermal capacity of unbaked walling, Facey (1997) and Padfield (1998). The investigation of
the behaviour of these buildings using dynamic thermal analysis is dependent upon the establishment of a firm
value for this thermal capacity. Clearly, with measured values of conductivity and capacity it is possible to
model the time dependent behaviour of cob buildings. We would expect cob walls to have a large damping
effect on thermal waves passing through them; therefore we would expect small decrement factors and long
lag times. Again, in order to model the time dependent thermal behaviour of a cob building, the non steady-
state thermal transmittance is required, along with the other parameters that describe the transient heat flow in
materials. The steady-state air to air thermal transmittance is the well known U-value, while the time

6
dependent equivalent is the Admittance, or Y-value Wm-2K-1. Two other parameters will be reported here, the
decrement factor, f, and the lag time, . The decrement factor is a measure of the decay suffered by a cyclic
wave of heat as it moves through a material, while the lag time gives the delay in hours for the heat maximum
to appear on the other side. Since the density of the cob wall is not known precisely, it is fortunate that to
calculate the admittance, decrement factor and lag time, we required the volumetric heat capacity, or simply
the ratio of  / . Bloomfield (1983) has written software to perform these calculations. His program,
“thermalfactors”, which is part of the Building Research Establishment (BRE) Admittance software, enables U,
Y, f and  to be calculated, and the variations of these quantities with wall thickness can be explored. Table 2
shows these values, calculated using the experimental results for the cob walls at Rezare, Bovey Tracy and
Sandford present in table 1. The results confirm our expectations; the walls all show small decrement factors
and long lag times. Some information concerning solid brick and solid concrete block walls is also included in
this table for comparison. Here, it is assumed that a wall of similar dimensions to the cob walls is constructed
entirely from the one material, say outer leaf brickwork, or inner leaf lightweight concrete block work. Table 2
shows the time dependent thermal properties for 250 mm and 500 mm thick brickwork and concrete block
work of two densities. The thermal data for these calculations on brick and concrete were taken from table
A3.15, CIBSE (1986).
Although not shown in table 2, it was observed that the admittances all had lead times with magnitudes
between 1 and 2 hours, and the internal admittances had similar magnitudes and lead times.
The data in table 2 shows that the U values of the cob walls decrease with increasing thickness, whilst the
admittances remain constant with an average value of 3.8 Wm-2K-1. The decrement factor falls and the lag
time increases as the thickness increases, confirming the intuitive view of cob walled buildings. The range of
decrement factors and lag times are comparable to those calculated for massive brickwork and concrete block
work walls of similar thickness. Finally, again for comparison, the bottom section of table 2 shows the values of
two cavity wall constructions, one with an air gap, the other with 75 mm of glass fibre insulation. The values
were taken from the table A3.25 CIBSE (1986) on the thermal properties of building structures. It is interesting
to note that a Rezare cob wall 1500 mm thick would be required to achieve a U value of 0.3 Wm-2K-1, the value
quoted for a 293 mm thick brickwork-glass fibre insulation -concrete block work-plaster wall. To meet the
current United Kingdom Building Regulations, Part L 1995, a wall U value of 0.45 Wm-2K-1 would be achieved
by Rezare cob wall 900 mm thick, and this may be compared with 100 mm lightweight concrete block, 50 mm
urea formaldehyde foam, 100 mm medium concrete block and 13 mm light weight plaster, which has the same
U value, an admittance of 3.5 Wm-2K-1, (lag time 2 hours), decrement factor 0.28 with lag time 9 hr.
An important issue not addressed in this work is the dependence of these thermal properties on moisture
content. The relatively large fluctuations of moisture content of the walling is confirmed by Trotman (1993),
suggesting that the in situ transient thermal technique would be best suited to measure a representative figure
for the thermal conductivity and thermal diffusivity. A further study might investigate the variation of thermal
properties with season as a way of assessing the influence of moisture and external environmental
temperature on these cob walls.
REFERENCES:

7
Batty W J, Probert S D, Ball M and O'Callaghan P W., 1984 The use of the thermal probe technique for the
measurement of the apparent thermal conductivities of moist materials, in Applied Energy, 18, 301-317.
Blackwell J H., 1954 A transient-flow method for determination of thermal constants of insulating materials in
bulk, in J Appl Phys, 25:2, 137-144.
Bloomfield D P., 1983 Building Research Establishment micro-computing package for prediction of
building temperatures and heat/cooling loads using the admittance method : Thermalfactors program,
Watford, Building Research Establishment.
CIBSE, 1986 Section 3, Thermal properties of building structures, in Design Guide A, Chartered Institute of
Building Service Engineers, London.
Facey W., 1997 Back to earth, adobe building in Saudi Arabia, published by Al-Turath, in association with The
London Centre of Arab Studies, PO Box 68200, Riyadh.
Ford M, Elkadi H and Watson L.,1999 The relevance of GIS in the evaluation of vernacular architecture, in
Journal of Architectural Conservation. Donhead Publishing, Dorset UK, 5:3, 64-75. ISSN 1355-6207.
Goodhew S., 1999 The Thermal properties of Cob Buildings in Devon, unpublished PhD thesis, University of
Plymouth.
Greer M J A., 1996 The effect of moisture content and composition on the compressive strengths and rigidity of
cob made from the soil of the Breccia measures near Teignmouth, Devon, PhD thesis, University of Plymouth.
Keefe L., 1998 An investigation into the causes of structural failure in traditional cob buildings, MPhil thesis,
University of Plymouth.
Keefe L, Watson L and Griffiths R., 1999 Possible causes of structural failure in traditional cob buildings, in
Terra 2000 8th International Conference on the Study and Conservation of Earthen Architecture, James and
James Ltd London. (In press)
Minke G., 1994 Lehmbau - Handbuch, Der Baustoff Lehm und seine Anwendung, Okobuch Verlag, Staufen
bei Freiburg, p55.
Padfield T., 1998 Casting mud in the debate on museum environmental standards,
http://www.natmus.min.dk/cons/tp/mudbuf/mudbuf1.htm
Trotman, P., 1993 Dampness in cob walls, paper presented to Devon Earth Building Association Seminar -
The building and repair of cob buildings, held in Exeter on 9 December.

8
TABLE 1. Preliminary measured values of the thermal properties of cob walls.
Thermal
Thermal
Density
Specific heat
conductivity
diffusivity
capacity
 / Wm-1K-1
 / m2s-1
 / kgm-3
c / Jkg-1K-1
Bovey Tracy cob
0.37
3.2 10-7
1230
950
Rezare cob
0.44
3.7 10-7
1460
830
Sandford cob
0.55
4.9 10-7
1800
630
TABLE 2. Calculated thermal properties of walls as a function of thickness.
[Calculations assume internal and external surface resistance of 0.12
and 0.06 m2KW-1 respectively, and 1 cycle per day.]
Wall construction
Thickness
U value
Admittance
Decrement
Lag time
mm
Wm-2K-1
Wm-2K-1
Factor
hr
Rezare cob
480
0.8
3.8
0.06
-17.4
250
1.3
3.9
0.37
-8.6
500
0.8
3.8
0.05
-18.1
Bovey Tracy cob
220
1.3
3.6
0.42
-7.9
250
1.2
3.6
0.34
-9.1
500
0.7
3.6
0.04
-19.3
Sandford cob
550
0.8
4.0
0.06
-17.3
250
1.6
4.0
0.45
-7.4
500
0.9
4.0
0.09
-15.6
For comparison :
Brickwork outer leaf
250
2.1
4.7
0.46
-6.9
500
1.3
4.7
0.11
-14.3
Brickwork inner leaf
250
1.7
4.4
0.40
-7.9
500
1.0
4.4
0.07
-16.4
Concrete block heavyweight
250
3.0
6.0
0.36
-7.4
500
2.1
5.9
0.09
-14.2
Concrete block lightweight
250
0.7
2.3
0.41
-8.4
500
0.4
2.3
0.05
-18.6
For comparison :
105mm brick, 25mm air gap, 100mm
heavy concrete block,13mm light plaster
243
1.6
4.3
0.31
-8.0
105mm brick,75mm glass fibre, 100mm
light concrete block, 13mm light plaster
293
0.33
2.4
0.39
-9.0