Which Weather Data Should You Use for Energy Simulations of Commercial Buildings?

Abstract

Users of energy simulation programs have a wide variety of weather data from which to choose–from locally recorded weather data to preselected ‘typical’ years, often a bewildering range of options. In the last five years, several organizations have developed new typical weather data sets including WYEC2, TMY2, CWEC, and CTZ2. Unfortunately, neither how these new data influence energy simulation results nor how they compare to recorded weather data is well documented.
This paper presents results from the DOE-2.1E hourly energy simulation program for a prototype office building as influenced by local measured weather data for multiple years and several weather data sets for eight U.S. locations. We compare the influence of the various weather data sets on simulated annual energy use and costs and annual peak electrical demand, heating load, and cooling load. Statistics for temperature, heating and cooling degree-days, and solar radiation for the different locations and data sets are also presented. Where possible, the author explains the variation relative to the different designs used in developing each data set. The variation inherent in actual weather data and how it influences simulation results is also shown. Finally, based on these results, the question is answered: which weather data should you use?

Full text

TO-98-2-2 1
TO-98-2-2 ASHRAE 1998 TRANSACTIONS 104 Part 2
Which Weather Data Should You Use
for Energy Simulations
of Commercial Buildings?
Drury B. Crawley
Member ASHRAE
ABSTRACT
Users of energy simulation programs have a wide
variety of weather data from which to choose–from
locally recorded weather data to preselected ‘typical’
years, often a bewildering range of options. In the last
five years, several organizations have developed new
typical weather data sets including WYEC2, TMY2,
CWEC, and CTZ2. Unfortunately, neither how these
new data influence energy simulation results nor how
they compare to recorded weather data is well
documented.
This paper presents results from the DOE-2.1E
hourly energy simulation program for a prototype office
building as influenced by local measured weather data
for multiple years and several weather data sets for eight
U.S. locations. We compare the influence of the various
weather data sets on simulated annual energy use and
costs and annual peak electrical demand, heating load,
and cooling load. Statistics for temperature, heating and
cooling degree-days, and solar radiation for the different
locations and data sets are also presented. Where
possible, the author explains the variation relative to the
different designs used in developing each data set. The
variation inherent in actual weather data and how it
influences simulation results is also shown. Finally,
based on these results, the question is answered: which
weather data should you use?
INTRODUCTION
In the last five years, American Society of Heating,
Refrigerating, and Air-Conditioning Engineers
(ASHRAE), National Renewable Energy Laboratory
(NREL), WATSUN Simulation Laboratory, and
California Energy Commission (CEC) have released new
or updated typical weather data sets for use in simulating
building energy performance: WYEC2, TMY2, CWEC,
and CTZ2, respectively. Each of these data sets contain a
year of hourly data (8,760 hours) synthesized to represent
long-term statistical trends and patterns in weather data
for a longer period of record. Each developer designed
its data sets to meet a particular need. ASHRAE
designed the WYEC2 data set to represent typical
weather patterns. NREL updated the TMY data sets to
represent the most recent period of record available for
use in work that require solar radiation data. WATSUN
Simulation Laboratory created the CWEC weather data
sets for use by the National Research Council (NRC)
Canada in developing and complying with their new
National Energy Code for Buildings. The CEC updated
their CTZ weather data to make them better represent
design conditions within each climate region and for use
in demonstrating compliance with the California Title 24
energy standards. All groups intended their weather data
sets to be usable with energy performance simulation
programs. A recent study by Haberl (1995) compared
measured weather data in calibrated DOE-2 simulations
versus TMY data.
For each of the four weather data sets—WYEC2,
TMY2, CWEC, and CTZ2—the developers used
standard methodologies to determine which data would
be used from the actual weather data period of record.
The methods were virtually the same in the four cases—
the true differences are related to the different weights
applied to weather variables in the selection process. But
these methods do not attempt to evaluate either the
impact on energy simulation results of the new data sets
or how these data sets compare to actual weather data or
other existing typical data sets. In this paper, we focus
on the TMY2 and WYEC2 data sets, comparing results
with actual weather data through energy simulation
results.
WEATHER DATA SETS
Over the past 20 years, several groups have
developed hourly weather data sets specifically designed
Drury B. Crawley is a program manager for the U.S. Department of Energy in Washington, D.C.

TO-98-2-2 2
for use in building energy simulations. One of the earliest
is Test Reference Year (TRY) (NCDC 1976). TRY con-
tains dry-bulb temperature, wet-bulb temperature, dew
point, wind direction and speed, barometric pressure,
relative humidity, cloud cover and type, and a placeholder
for solar radiation; however, no measured or calculated
solar data are included. When used for building energy
simulations, the simulation program must calculate the
solar radiation based on the cloud cover and cloud type
information available in the TRY data. Simulation pro-
grams may not deal with the complex interactions, use
simplified methods, or methods that are out of date.
Another weakness of the TRY data set was the method
used to select data. The TRY data are an actual historic
year of weather, selected using a process where years in the
period of record (~1948-1975) which had months with
extremely high or low mean temperatures were pro-
gressively eliminated until only one year remained. This
tended to result in a particularly mild year that, either by
intention or default, excludes extreme conditions. TRY
data are available for 60 locations in the United States.
To deal with the limitations of TRY, particularly the
lack of solar data, the National Climatic Data Center
(NCDC) worked together with Sandia National Laboratory
(SNL) to create a new data set, Typical Meteorological
Year (TMY). TMY include, in addition to the data con-
tained in TRY, total horizontal and direct normal solar
radiation data for 234 U.S. locations (NCDC 1981).
Twenty-six locations have measured solar data; solar data
for the other 208 locations were calculated from cloud
cover and type. This eliminated the need for the simu-
lation programs to separately calculate solar data. Data in
this set consist of 12 months selected from an approxi-
mately 23-year period of record (~1952-1975, available
data varies by location) to represent typical months. The
method used is similar to that used for the TRY, but
individual months are selected rather than entire years.
The TMY months were selected based on a monthly
composite weighting of solar radiation, dry-bulb tempera-
ture, dew point, and wind velocity as compared to the long-
term distribution of those values. Months that were closest
to the weighted long-term distribution were selected. The
resulting TMY data files each contain months from a
number of different years.
In the late 1970s, the CEC developed a data set
specifically for use in complying with the new Title 24
building energy regulations. They mapped the climatic
regions of California, dividing it into 16 regions. CEC
then created a weather data set, California Thermal Zones
(CTZ), with a weather file for each region. The CTZ are
based on the TMY format with several CTZ files derived
from a specific TMY location. In 1992, the CEC updated
their CTZ data set, creating CTZ2 (CEC 1992), with data
in the new ASHRAE WYEC2 format. In creating the
CTZ2, the temperatures from the original CTZ data set
were adjusted to make the temperature profile match the
ASHRAE design conditions for the particular location
(ASHRAE 1993). More recently, the CEC developed a
method to adjust the CTZ2 data to another location (CEC
1994). Essentially this procedure modifies the tempera-
ture in the existing CTZ2 weather file to match the design
conditions for another city.
From 1970 through 1983, ASHRAE commissioned
research projects RP-100, RP-239, and RP-364 (Crow
1970, 1980, 1983) to create a weather data set to represent
more typical weather patterns than either a single repre-
sentative year or an assemblage of months. This weather
data set, known as Weather Year for Energy Calculations
(WYEC) (ASHRAE 1985), uses the TRY format but it
includes solar data (measured where available, otherwise
calculated based on cloud cover and type). The basic
method used to select data for WYEC was to determine,
for each month of the year, the single, real month of hourly
data whose mean dry-bulb air temperature was closest to
the average dry-bulb temperature for that month in the 30-
year period of record. If the mean dry-bulb air temperature
for the individual month was within 0.2F of the mean for
the 30 months in the period of record, that individual
month was used unless it included unusual or extreme
weather patterns or events. If an unusual weather event
was found in that month, the next closest month was
examined for unusual patterns—until a month was found.
If none of the 30 months in the period of record was within
0.2F of the mean dry -bulb temperature, then the month
with a mean closest to the mean of the 30 months was
selected. Then, individual days from other months were
substituted when those days helped bring the mean for the
month closer to the 30-year mean. This process continued
until one or more substitute days brought the now modified
month mean dry bulb temperature to within 0.2 F of the 30-
year mean. In general, no WYEC file needed more than 3
substitute days for any month to match the 30-year mean.
WYEC data for 51 locations (46 locations in the United
States and five in Canada) were completed in late 1983.
In the early 1990s, ASHRAE began to update the
WYEC data set. Beginning with the format for the TMY
data set, the WYEC data set format was first extended by
including calculated hourly illuminance data and data
quality and source flags. Other major changes included
updating the calculated solar radiation data and adjusting
the data from solar time to local time. A recently updated
models for calculating solar radiation from cloud cover
data (Perez et al. 1990, 1991) was used to recalculate the
solar radiation components and illuminance data. The
updated WYEC data set became known as WYEC2, for
WYEC version 2 (ASHRAE 1997a). In addition, the 26
TMY locations with measured solar data were updated to
include illumination data and to correct the time shift.
NREL, working with ASHRAE, processed the existing
WYEC and TMY data to create the 77 final WYEC2
format files (Stoffel and Rymes 1997, 1998).
In 1993, NREL created a new long-term solar
radiation data set based on the 1961-1990 period of record
known as the National Solar Radiation Data Base

TO-98-2-2 3
(NSRDB) (Maxwell 1990). In conjunction with the
National Climatic Data Center (NCDC), they published a
combined set of hourly weather and solar data for the
1961-1990 period of record. These data are known as
Solar and Meteorological Surface Observational Network
(SAMSON) (NCDC 1993) and include 30 years of hourly
data for 239 locations, most of those in the original TMY
data set. As with the TMY data set, only 26 locations have
measured solar data for at least a portion of the 30-year
period of record. For the remaining 213 locations, solar
radiation values were calculated from cloud cover based on
the Perez model (1990, 1991). Separately, NREL updated
the TMY data set based on the new period of record (1961-
1990) available in SAMSON—creating the TMY2 data set
(NREL 1995).
In 1992, NRC Canada commissioned the WATSUN
Simulation Laboratory at the University of Waterloo to
create an hourly weather data set for Canadian locations.
They used the long-term data set developed by the
Atmospheric Environment Service, Environment Canada in
a TMY methodology, formatting the resultant data set in
the WYEC2 format. WATSUN created data for 49
locations (WATSUN 1992).
In Europe, a data set for European locations (European
Test Reference Year) (Commission of the European
Community 1985) was created using a methodology
similar to that used by NCDC and SNL to derive the TMY.
Petrakis (1995) recently recommended revised procedures
for generating Test Meteorological Years from observed
data sets in Europe.
SIMULATION METHODOLOGY
For the work described in this paper, the author
simulated an office building using the DOE-2.1E energy
simulation program (Winkelmann et al. 1994). The build-
ing model was kept identical for all weather data sets with
HVAC equipment sizing based on design conditions in the
ASHRAE Handbook—Fundamentals (ASHRAE 1993).
The office building modeled is a 48,000 ft2, three-story
building typical of recent envelope-dominated, low-rise
buildings built in the U.S. For lighting, efficient 0.8 W/ft2,
T-8 fluorescent, 2-lamp, 2 x 4 fixtures with electronic
ballasts and occupancy sensors were assumed. Office
equipment was assumed at a level of 1.0 W/ft2 for
computers, laser printers, photocopiers, and facsimile
machines. The building envelope assumed a 40% fene-
stration-to-wall ratio with glazing varying by location—
primarily single-pane, tinted/reflective in southern
locations and double-pane, tinted in northern locations.
The minimum occupied outside air ventilation rate was 20
cfm/person. The air system simulated was a VAV reheat
system with an enthalpy-controlled outside air economizer.
The central plant included 0.55 kW per ton centrifugal
chillers and a 90% efficiency gas-fired boiler. Energy
costs were calculated by DOE-2.1E from local utility rates.
Actual hourly weather data (SAMSON, 1961-1990,
30-year period of record) and typical weather data sets
with hourly data (TRY, TMY, TMY2, WYEC, and
WYEC2) were used in the simulations. Eight U.S.
locations were selected to cover a range of typical climatic
patterns: Denver, Los Angeles, Miami, Minneapolis, New
York, Phoenix, Seattle, and Washington, D. C.
For each of the eight locations, Table 1 first shows the
99% (winter) and 2-1/2% (summer) design temperature
values from Chapter 24 of the 1993 ASHRAE Handbook—
Fundamentals (ASHRAE 1993). The second line for each
location shows the new annual 99.6% (heating) and 1%
(cooling) design temperatures for each location from
Chapter 26 of the 1997 ASHRAE Handbook—
Fundamentals (ASHRAE 1997b). Note that for
Washington, D.C., the design temperatures from the 1993
Fundamentals are for Washington National Airport—there
were no design conditions for Washington Dulles Airport
in the 1993 Fundamentals. The design conditions listed in
Table 1 for Washington Dulles Airport are from nearby
Sterling, Virginia. The next portion of Table 1 for each
location includes maximum, average, median, and mini-
mum values for the 99% (winter) and 2-1/2% (summer)
design temperatures, heating and cooling degree-days, and
solar radiation calculated from the 1961-1990 SAMSON
data. For the SAMSON data:
♦ Maximum refers to the highest value in the 30-year
set, i.e., the maximum winter design temperature is the
winter design temperature for the warmest year in the
30-year set.
♦ Average is the average of the annual values for the 30
years.
♦ Median is the median of the annual values for the 30
years.
♦ Minimum is the lowest value among the 30 years, i.e.,
the minimum winter design temperature is the design
winter temperature for the coldest year of the 30.
Similar statistics derived from the typical weather data
sets are also shown in Table 1. In Table 1 and Figures 9
through 24, ‘WYEC2 (TMY)’ means WYEC2 data derived
from TMY files and ‘WYEC2 (WYEC)’ means WYEC2
data derived from WYEC files. Note that not all weather
data types were available for all locations; no TRY or
WYEC2 (TMY) data were available for Denver and no
WYEC2 (TMY) data were available for Los Angeles and
Minneapolis. WYEC2 (TMY) data were only available for
locations with measured solar data (the original 26
SOLMET locations). These data are left blank for
consistent presentation among locations in Table 1 and
Figures 9-24.
RESULTS
Figures 1 through 8 show the office building
simulation results using the 30 years of SAMSON weather
data in terms of annual end-use energy performance and

TO-98-2-2 4
TABLE 1
Statistics for Weather File Types and SAMSON Weather Data
Location Statistic or
File Type
Winter 99%
Dry bulb
Temperature
(F)
Summer 2.5%
Dry bulb
Temperature
(F)
Annual Heating
Degree-Days,
65 F
Annual Cooling
Degree-Days,
65F
Daily Average
Direct Normal
Solar (Btu/h)/ft2
Daily Average
Horizontal Solar
(Btu/h)/ft2
Denver 1993 HOF -5 91
Colorado 1997 HOF (99.6/1%)
-3 90
Maximum 8 93 6780.5 934.5 1897.2 1525.0
Average -3 91 6015.5 650.3 1714.8 1450.8
Median -2 91 6042.3 668.8 1743.5 1453.3
1961-
Minimum -15 84 4936.5 279.5 1376.1 1348.6
TRY -- -- -- -- -- --
TMY -4 90 6114 566 2044.3 1591.2
TMY2 -2 91 6007 623 1743.2 1467.0
WYEC -4 91 5941 631 1875.7 1573.2
WYEC2 (TMY)
WYEC2 (WYEC) -3 91 5936 631 1612.2 1514.9
Los Angeles 1993 HOF 41 80
California 1997 HOF (99.6/1%)
43 81
Maximum 47 84 1915.5 933.5 1694.8 1632.7
Average 42.6 78.8 1401.6 591.7 1532.1 1568.1
Median 42.0 78.5 1376.3 535.5 1546.4 1564.8
1961-
Minimum 39 74 976.5 284.5 1365.2 1499.7
TRY 42 78 1518.0 391.5 1331.5 1392.2
TMY 42 78 1506.5 466.5 1693.7 1611.6
TMY2 43 77 1291.0 469.5 1563.6 1579.4
WYEC 41 77 1704.0 459.0 1662.6 1608.8
WYEC2 (TMY) -- -- -- -- -- --
WYEC2 (WYEC) 41 77 1704.0 459.0 1373.2 1553.6
Miami 1993 HOF 44 90
Florida 1997 HOF (99.6/1%)
46 90
Maximum 54 92 345.0 4741.0 1453.7 1630.9
Average 44.4 89.4 190.5 4138.7 1254.0 1532.0
Median 44.5 89.0 194.8 4119.5 1274.2 1531.5
1961-
Minimum 37 87 17.5 3438.0 990.8 1344.4
TRY 44 89 147.0 4262.5 863.7 1367.5
TMY 43 89 188.5 4031.0 1231.7 1482.0
TMY2 48 89 141.0 4126.5 1307.2 1557.2
WYEC 42 89 227.0 4005.0 1047.6 1478.0
WYEC2 (TMY) 43 89 188.5 4032.5 1071.0 1477.5
WYEC2 (WYEC) 42 89 227.0 4005.0 1049.9 1470.2
Minneapolis 1993 HOF -16 89
Minnesota 1997 HOF (99.6/1%)
-16 88
Maximum -5 95 9105.0 1124.5 1574.6 1343.9
Average -15.7 87.9 8002.9 695.9 1265.6 1234.0
Median -16.5 88.0 8077.3 688.3 1250.4 1228.7
1961-
Minimum -24 84 6435.0 401.0 1041.1 1167.2
TRY -25 90 8345.5 911.5 1069.0 1160.2
TMY -17 90 8095.0 759.5 1182.3 1169.6
TMY2 -15 86 7985.5 634.0 1299.1 1257.0
WYEC -20 88 8070.5 750.5 1123.3 1170.8
WYEC2 (TMY) -- -- -- -- -- --
WYEC2 (WYEC) -19 88 8070.0 750.5 1135.4 1161.4

TO-98-2-2 5
TABLE 1 (Continued)
Statistics for Weather File Types and SAMSON Weather Data
Location Statistic or
File Type
Winter 99%
Dry bulb
Temperature
(F)
Summer 2.5%
Dry bulb
Temperature
(F)
Annual Heating
Degree-Days,
65 F
Annual Cooling
Degree-Days,
65F
Daily Average
Direct Normal
Solar (Btu/h)/ft2
Daily Average
Horizontal Solar
(Btu/h)/ft2
New York 1993 HOF 11 89
New York 1997 HOF (99.6/1%)
13 89
Maximum 21 92 5465.0 1324.5 1215.0 1321.9
Average 10.1 87.9 4977.6 1067.1 1095.7 1265.3
Median 10.5 88.0 5009.3 1082.0 1095.0 1268.6
1961-
Minimum 3 84 3976.5 751.5 933.8 1182.7
TRY 14 86 4520 1059 963.7 1180.8
TMY 13 84 5058 824 953.2 1092.1
TMY2 9 87 5090 1002 1069.8 1268.5
WYEC 14 87 4941 1034 786.6 1094.6
WYEC2 (TMY) 13 84 5052 825 854.5 1090.5
WYEC2 (WYEC) 14 87 4941 1034 789.7 1066.7
Phoenix 1993 HOF 31 107
Arizona 1997 HOF (99.6/1%)
34 108
Maximum 41 111 2043.5 5125.0 2331.0 1866.8
Average 33.9 107.8 1210.2 4052.2 2154.8 1817.8
Median 34.0 108.0 1222.3 4112.5 2171.1 1826.4
1961-
Minimum 27 104 649.0 3087.0 1844.1 1669.0
TRY 30 106 1476 3390 1892.0 1679.2
TMY 31 106 1391 3641 2187.2 1873.0
TMY2 34 108 1154 3815 2187.7 1839.0
WYEC 32 107 1356 3661 2159.6 1870.0
WYEC2 (TMY) 31 106 1389 3644 2232.6 1868.5
WYEC2 (WYEC) 32 107 1356 3661 2199.1 1863.6
Seattle 1993 HOF 21 80
Washington 1997 HOF (99.6/1%)
23 81
Maximum 31 86 5674.5 338.0 1106.6 1140.5
Average 23.7 81.5 4927.7 162.9 932.5 1055.2
Median 25.5 82.0 4844.8 167.8 947.4 1056.4
1961-
Minimum 13 76 4338.0 52.0 664.3 1000.1
TRY 27 84 5373.5 142.0 675.7 933.8
TMY 24 81 5299.5 106.0 878.2 1031.8
TMY2 29 80 4867.0 127.0 966.4 1061.5
WYEC 24 81 5295.5 106.0 878.8 1030.8
WYEC2 (TMY) 20 81 5222.5 97.0 916.5 1054.0
WYEC2 (WYEC) 20 81 5222.5 97.0 908.1 1047.2
Washington, D. C. 1993 HOF (National)
14 91
(Dulles Airport) 1997 HOF (99.6/1%)
9 90
(Sterling, Virginia) Maximum 18 95 5538.0 1470.0 1367.4 1402.8
Average 7.0 89.9 5017.3 1042.4 1173.7 1303.2
Median 6.5 90.0 5034.8 1019.8 1172.3 1311.1
1961-
Minimum 0 87 3993.0 766.5 1020.8 1177.4
TRY (National) 13 91 4112.5 1525.5 1025.0 1231.9
TMY 7 90 4865.5 1054.0 1131.2 1215.3
TMY2 8 89 5233.0 1044.0 1171.4 1300.5
WYEC 7 90 4865.5 1062.5 1023.2 1213.5
WYEC2 (TMY) 12 90 4236.0 1425.0 1000.0 1212.3
WYEC2 (WYEC) 12 90 4236.0 1425.0 982.6 1201.7

TO-98-2-2 6
0
10
20
30
40
50
60
70
80
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
Average
Total Annual Energy Consumption, kBtu/ft2-y
Supply Fans
Cooling Electric
Heating Natural Gas
Interior Lights
Other
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
Average
Total Annual Energy Cost, $/ft2-y
Natural Gas Charges
Electric Energy Charges
Peak Demand Charges
Figure 1 Effect of Actual Weather Variation on Annual Energy Consumption and Costs in Denver, Colorado.
0
10
20
30
40
50
60
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
Average
Total Annual Energy Consumption, kBtu/ft2-y
Supply Fans
Cooling Electric
Heating Natural Gas
Interior Lights
Other
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1.60
1.80
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
Average
Total Annual Energy Cost, $/ft2-y
Natural Gas Charges
Electric Energy Charges
Peak Demand Charges
Figure 2 Effect of Actual Weather Variation on Annual Energy Consumption and Costs in Los Angeles, California.
0
10
20
30
40
50
60
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
Average
Total Annual Energy Consumption, kBtu/ft2-y
Supply Fans
Cooling Electric
Heating Natural Gas
Interior Lights
Other
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
Average
Total Annual Energy Cost, $/ft2-y
Natural Gas Charges
Electric Energy Charges
Electric Demand Charges
Figure 3 Effect of Actual Weather Variation on Annual Energy Consumption and Costs in Miami, Florida.

TO-98-2-2 7
0
10
20
30
40
50
60
70
80
90
100
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
Average
Total Annual Energy Consumption, kBtu/ft2-y
Supply Fans
Cooling Electric
Heating Natural Gas
Interior Lights
Other
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
Average
Total Annual Energy Cost, $/ft2-y
Natural Gas Charges
Electric Energy Charges
Electric Demand Charges
Figure 4 Effect of Actual Weather Variation on Annual Energy Consumption and Costs in Minneapolis, Minnesota.
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
Average
Total Annual Energy Consumption, kBtu/ft2-y
Supply Fans
Cooling Electric
Heating Natural Gas
Interior Lights
Other
0.00
0.25
0.50
0.75
1.00
1.25
1.50
1.75
2.00
2.25
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
Average
Total Annual Energy Cost, $/ft2-y
Natural Gas Charges
Electric Energy Charges
Peak Demand Charges
Figure 5 Effect of Actual Weather Variation on Annual Energy Consumption and Costs in New York, New York.
0
10
20
30
40
50
60
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
Average
Total Annual Energy Consumption, kBtu/ft2-y
Supply Fans
Cooling Electric
Heating Natural Gas
Interior Lights
Other
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
Average
Total Annual Energy Cost, $/ft2-y
Natural Gas Charges
Electric Energy Charges
Peak Demand Charges
Figure 6 Effect of Actual Weather Variation on Annual Energy Consumption and Costs in Phoenix, Arizona.

TO-98-2-2 8
0
10
20
30
40
50
60
70
80
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
Average
Total Annual Energy Consumption, kBtu/ft2-y
Supply Fans
Cooling Electric
Heating Natural Gas
Interior Lights
Other
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
Average
Total Annual Energy Cost, $/ft2-y
Natural Gas Charges
Electric Energy Charges
Electric Demand Charges
Figure 7 Effect of Actual Weather Variation on Annual Energy Consumption and Costs in Seattle, Washington.
0
10
20
30
40
50
60
70
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
Average
Total Annual Energy Consumption, kBtu/ft2-y
Supply Fans
Cooling Electric
Heating Natural Gas
Interior Lights
Other
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
Average
Total Annual Energy Cost, $/ft2-y
Natural Gas Charges
Electric Energy Charges
Electric Demand Charges
Figure 8 Effect of Actual Weather Variation on Annual Energy Consumption and Costs in Washington, DC
.
energy costs by fuel type for the eight locations. The left
graph in each figure show the component end-use (heating,
cooling, lighting, fans, and other) energy performance (in
thousands of Btu/ft2-yr) for each of the 30 years. The
right-hand bar shows the average energy performance for
the 30 years. The right graph shows the components
(natural gas and electricity energy and demand charges) of
total annual energy costs (in $/ft2-yr) as simulated for each
of the 30 years. The right-hand bar shows the average
energy costs for the 30 years. Table 2 presents the
average, minimum, and maximum annual energy
consumption and costs (summarized from Figures 1
through 8) along with average, minimum, and maximum
annual peak electric demand, and annual peak cooling and
heating loads.
Figures 1 through 8 demonstrate that buildings in
locations that are either heating-dominated (Minneapolis)
or have a significant amount of both heating and cooling
(Denver, New York, Seattle, and Washington, D.C.)
exhibit a higher relative variation in annual energy con-
sumption year-to-year. Milder or cooling-dominated
climates (Los Angeles, Miami, and Phoenix) demonstrate
relatively less overall variation in year-to-year energy
consumption. However, as shown in the summary data in
Table 2, the range of annual energy performance across the
eight locations only varies from –11% to 7% for the
SAMSON 30-year period of record. Annual energy cost is
a function of the three components shown on the right-hand
portion of the figures: natural gas costs (usually based on
natural gas consumption), and electric energy and demand
charges. Local utility rates weight electricity consumption
and peak demand differently depending on what is more
expensive to the utility, consumption or peak demand.
Throughout the eight locations, the annual energy costs
vary widely, from a low of $0.42 for Denver to a high of
$2.12 for New York. Overall, the year-to-year variation in
annual energy cost for the eight locations is less than half
the variation in energy consumption noted

TO-98-2-2 9
TABLE 2
Variation in Simulated Annual Energy Consumption, Energy Costs, Peak Electric Demand, and
Peak Loads for SAMSON Weather Data
Average
(Min/Max percent change from average)
Total Annual Energy Annual Peak
Location Consumption,
kBtu/ft
2
-yr Costs,
$/ft
2
-yr Electric
Demand, W/ft2 Cooling Load,
(Btu/h)/ft2 Heating Load,
(Btu/h)/ft2
Denver,
Colorado 66.1
(-7.7%/6.7%) 0.42
(-4.6%/3.3%) 4.1
(-2.3%/1.4%) 17.6
(-8.8%/9.0%) 32.5
(-16.1%/8.9%)
Los Angeles,
California 49.9
(-3.0%/4.0%) 1.59
(-1.7%/1.7%) 4.1
(-4.7%/4.9%) 19.5
(-21.2%/34.1%) 20.1
(-21.7%/21.4%)
Miami,
Florida
50.3
(-1.8%/1.8%) 1.11
(-2.1%/1.9%) 4.7
(-1.1%/1.0%) 28.0
(-8.2%/8.9%) 15.6
(-74.8%/75.3%)
Minneapolis,
Minnesota 81.4
(-11.0%/7.0%) 0.92
(-4.4%/2.6%) 4.4
(-4.5%/2.2%) 24.0
(-18.7%/19.5%) 36.9
(-6.4%/11.9%)
New York,
New York 67.0
(-8.7%/4.0%) 2.12
(-1.5%/1.6%) 4.4
(-2.3%/2.0%) 24.0
(-11.9%/15.2%) 32.0
(-13.5%/14.9%)
Phoenix,
Arizona 52.4
(-2.3%/2.9%) 1.31
(-2.1%/2.1%) 4.7
(-1.7%/2.6%) 28.0
(-7.6%/10.7%) 19.4
(-54.9%/34.5%)
Seattle,
Washington
63.9
(-3.9%/6.5%) 0.58
(-2.3%/3.6%) 4.0
(-3.5%/2.1%) 18.1
(-17.8%/18.5%) 25.7
(-11.3%/19.5%)
Washington,
D. C. 63.8
(-8.1%/4.3%) 1.23
(-3.0%/2.0%) 4.5
(-3.7%/1.7%) 24.5
(-14.4%/16.7%) 30.6
(-13.6%/13.4%)
above, only –4.6% to 3.6%. Interestingly, annual peak
electrical demand variation is similar to that for energy
costs, -4.7% to 4.9%. Similar to annual energy con-
sumption, the least variation is apparent in cooling-
dominated climates (Miami and Phoenix). But climates
with a mix of heating and cooling (Denver, New York,
Seattle, and Washington) showed less variation in peak
demand. Unlike energy consumption, peak demand varies
considerably more in Los Angeles, a location with
relatively mild but variable weather conditions. Similar to
Los Angeles, Seattle has higher variation in electric
demand. Because the simulated building is gas-heated,
electrical demand variation is less than that of energy
consumption in heating-dominated climates such as
Minneapolis.
Figures 9 through 16 compare similar results for the
weather data type sets in terms of energy performance and
energy cost for the eight locations. The weather data type
sets are contrasted with average, minimum, and maximum
values shown in Table 2 (from the SAMSON 30-year
simulations in Figures 1 through 8). The left graph in each
figure shows total energy performance (thousands of
Btu/ft2-yr) for each of the weather data file types. The
three lines on the graph are the maximum, average, and
minimum energy performance from the SAMSON
simulations (Table 2 and Figures 1 through 8). The right
graph shows the total annual energy costs ($/ft2-yr) as
simulated for the weather data file types. The three lines
on the right graph show the maximum, average, and
minimum energy costs from the simulations of 30 years of
SAMSON data (from Table 2 and Figures 1 through 8).
Table 3 presents summary information from Figures 9
through 16 for annual energy consumption and costs, and
annual peak electric demand and cooling and heating
loads. For annual energy consumption and costs and peak
electric demand, the table shows the average value for the
30-year SAMSON simulations from Table 2 along with the
percentage change from the average value for each weather
data type. The two right-hand columns show the variation
exhibited in annual peak heating and cooling loads
calculated from the simulations. The values in the design
size rows are the peak cooling or heating requirement for
HVAC equipment sizing based on design conditions in the
1993 ASHRAE Handbook—Fundamentals (ASHRAE
1993). The values in the SAMSON and the weather data
type rows are the percent change from the design size
values.
The variation of energy consumption for the weather
data types shown in Figures 9 through 16 and summarized
in Table 3 is less than that shown for the 30-year period of
record. The range of variation across the eight locations is
-2.3% to 5.4%; excluding the TRY results, the range of
variation among the weather data types is –1.9% to 3.2%.
Because the TRY period of record (~1945-1973) and the

TO-98-2-2 10
60
61
62
63
64
65
66
67
68
69
70
71
72
TRY TMY TMY2 WYEC WYEC2 (TMY) WYEC2
(WYEC)
Weather File Type
Annual Energy Consumption, kBtu/ft2-y
Maximum
Average
Minimum
0.39
0.40
0.41
0.42
0.43
0.44
TRY TMY TMY2 WYEC WYEC2 (TMY) WYEC2
(WYEC)
Weather File Type
Annual Energy Costs, $/ft2-y
Maximum
Average
Minimum
Figure 9 Comparison of Annual Energy Consumption and Costs for Weather File Types and SAMSON Weather Data in
Denver, Colorado.
48
49
50
51
52
TRY TMY TMY2 WYEC WYEC2 (TMY) WYEC2
(WYEC)
Weather File Type
Annual Energy Consumption, kBtu/ft2-y
Maximum
Average
Minimum
1.55
1.56
1.57
1.58
1.59
1.60
1.61
1.62
TRY TMY TMY2 WYEC WYEC2 (TMY) WYEC2
(WYEC)
Weather File Type
Annual Energy Costs, $/ft2-y
Maximum
Average
Minimum
Figure 10 Comparison of Annual Energy Consumption and Costs for Weather File Types and SAMSON Weather Data in
Los Angeles, California.
49.0
49.5
50.0
50.5
51.0
51.5
TRY TMY TMY2 WYEC WYEC2 (TMY) WYEC2
(WYEC)
Weather File Type
Annual Energy Consumption, kBtu/ft2-y
Maximum
Average
Minimum
1.09
1.10
1.11
1.12
1.13
1.14
TRY TMY TMY2 WYEC WYEC2 (TMY) WYEC2
(WYEC)
Weather File Type
Annual Energy Costs, $/ft2-y
Maximum
Average
Minimum
Figure 11 Comparison of Annual Energy Consumption and Costs for Weather File Types and SAMSON Weather Data in
Miami, Florida.

TO-98-2-2 11
72
74
76
78
80
82
84
86
88
TRY TMY TMY2 WYEC WYEC2 (TMY) WYEC2
(WYEC)
Weather File Type
Annual Energy Consumption, kBtu/ft2-y
Maximum
Average
Minimum
0.88
0.89
0.90
0.91
0.92
0.93
0.94
0.95
0.96
TRY TMY TMY2 WYEC WYEC2 (TMY) WYEC2
(WYEC)
Weather File Type
Annual Energy Costs, $/ft2-y
Maximum
Average
Minimum
Figure 12 Comparison of Annual Energy Consumption and Costs for Weather File Types and SAMSON Weather Data in
Minneapolis, Minnesota.
61
62
63
64
65
66
67
68
69
70
TRY TMY TMY2 WYEC WYEC2 (TMY) WYEC2
(WYEC)
Weather File Type
Annual Energy Consumption, kBtu/ft2-y
Maximum
Average
Minimum
2.07
2.08
2.09
2.10
2.11
2.12
2.13
2.14
2.15
2.16
TRY TMY TMY2 WYEC WYEC2 (TMY) WYEC2
(WYEC)
Weather File Type
Annual Energy Costs, $/ft2-y
Maximum
Average
Minimum
Figure 13 Comparison of Annual Energy Consumption and Costs for Weather File Types and SAMSON Weather Data in
New York, New York.
51
52
53
54
TRY TMY TMY2 WYEC WYEC2 (TMY) WYEC2
(WYEC)
Weather File Type
Annual Energy Consumption, kBtu/ft2-y
Maximum
Average
Minimum
1.28
1.29
1.30
1.31
1.32
1.33
1.34
TRY TMY TMY2 WYEC WYEC2 (TMY) WYEC2
(WYEC)
Weather File Type
Annual Energy Costs, $/ft2-y
Maximum
Average
Minimum
Figure 14 Comparison of Annual Energy Consumption and Costs for Weather File Types and SAMSON Weather Data in
Phoenix, Arizona.

TO-98-2-2 12
61
62
63
64
65
66
67
68
69
TRY TMY TMY2 WYEC WYEC2 (TMY) WYEC2
(WYEC)
Weather File Type
Annual Energy Consumption, kBtu/ft2-y
Maximum
Average
Minimum
0.56
0.57
0.58
0.59
0.60
TRY TMY TMY2 WYEC WYEC2 (TMY) WYEC2
(WYEC)
Weather File Type
Annual Energy Costs, $/ft2-y
Maximum
Average
Minimum
Figure 15 Comparison of Annual Energy Consumption and Costs for Weather File Types and SAMSON Weather Data in
Seattle, Washington.
58
59
60
61
62
63
64
65
66
67
TRY (National
Airport) TMY TMY2 WYEC WYEC2 (TMY) WYEC2
(WYEC)
Weather File Type
Annual Energy Consumption, kBtu/ft2-y
Maximum
Average
Minimum
1.18
1.19
1.20
1.21
1.22
1.23
1.24
1.25
1.26
TRY (National
Airport) TMY TMY2 WYEC WYEC2 (TMY) WYEC2
(WYEC)
Weather File Type
Annual Energy Costs, $/ft2-y
Maximum
Average
Minimum
Figure 16 Comparison of Annual Energy Consumption and Costs for Weather File Types and SAMSON Weather Data in
Washington, DC.
0.80
0.85
0.90
0.95
1.00
1.05
1.10
1.15
SAMSON
TRY
TMY
TMY2
WYEC
WYEC2 (TMY)
WYEC2 (WYEC)
SAMSON
TRY
TMY
TMY2
WYEC
WYEC2 (TMY)
WYEC2 (WYEC)
Peak Cooling Peak Heating
Peak Load as Fraction of Average Peak Load
Design Size
Design Size
Figure 17 Comparison of Annual Peak Loads in Denver,
Colorado.
0.75
0.85
0.95
1.05
1.15
1.25
1.35
SAMSON
TRY
TMY
TMY2
WYEC
WYEC2 (TMY)
WYEC2 (WYEC)
SAMSON
TRY
TMY
TMY2
WYEC
WYEC2 (TMY)
WYEC2 (WYEC)
Peak Cooling Peak Heating
Peak Load as Fraction of Average Peak Load
Design Size
Design Size
Figure 18 Comparison of Annual Peak Loads in Los
Angeles, California.

TO-98-2-2 13
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1.60
1.80
SAMSON
TRY
TMY
TMY2
WYEC
WYEC2 (TMY)
WYEC2 (WYEC)
SAMSON
TRY
TMY
TMY2
WYEC
WYEC2 (TMY)
WYEC2 (WYEC)
Peak Cooling Peak Heating
Peak Load as Fraction of Average Peak Load
Design Size
Design Size
Figure 19 Comparison of Annual Peak Loads in Miami,
Florida.
0.85
0.90
0.95
1.00
1.05
1.10
1.15
1.20
SAMSON
TRY
TMY
TMY2
WYEC
WYEC2 (TMY)
WYEC2 (WYEC)
SAMSON
TRY
TMY
TMY2
WYEC
WYEC2 (TMY)
WYEC2 (WYEC)
Peak Cooling Peak Heating
Peak Load as Fraction of Average Peak Load
Design Size
Design Size
Figure 21 Comparison of Annual Peak Loads in New
York, New York.
0.80
0.85
0.90
0.95
1.00
1.05
1.10
1.15
1.20
SAMSON
TRY
TMY
TMY2
WYEC
WYEC2 (TMY)
WYEC2 (WYEC)
SAMSON
TRY
TMY
TMY2
WYEC
WYEC2 (TMY)
WYEC2 (WYEC)
Peak Cooling Peak Heating
Peak Load as Fraction of Average Peak Load
Design Size
Design Size
Figure 23 Comparison of Annual Peak Loads in Seattle,
Washington.
0.80
0.85
0.90
0.95
1.00
1.05
1.10
1.15
1.20
SAMSON
TRY
TMY
TMY2
WYEC
WYEC2 (TMY)
WYEC2 (WYEC)
SAMSON
TRY
TMY
TMY2
WYEC
WYEC2 (TMY)
WYEC2 (WYEC)
Peak Cooling Peak Heating
Peak Load as Fraction of Average Peak Load
Design Size
Design Size
Figure 20 Comparison of Annual Peak Loads in
Minneapolis, Minnesota.
0.45
0.65
0.85
1.05
1.25
1.45
SAMSON
TRY
TMY
TMY2
WYEC
WYEC2 (TMY)
WYEC2 (WYEC)
SAMSON
TRY
TMY
TMY2
WYEC
WYEC2 (TMY)
WYEC2 (WYEC)
Peak Cooling Peak Heating
Peak Load as Fraction of Average Peak Load
Design Size
Design Size
Figure 22 Comparison of Annual Peak Loads in Phoenix,
Arizona
0.85
0.90
0.95
1.00
1.05
1.10
1.15
1.20
SAMSON
TRY
TMY
TMY2
WYEC
WYEC2 (TMY)
WYEC2 (WYEC)
SAMSON
TRY
TMY
TMY2
WYEC
WYEC2 (TMY)
WYEC2 (WYEC)
Peak Cooling Peak Heating
Peak Load as Fraction of Average Peak Load
Design Size
Design Size
Figure 24 Comparison of Annual Peak Loads in
Washington, DC

TO-98-2-2 14
TABLE 3
Comparison of Simulated Annual Energy Consumption, Energy Costs, Peak Electric Demand,
and Peak Loads for Weather Data Files Types and SAMSON Weather Data
Total Annual Energy Annual Peak
Location Weather File Type
Consumption,
kBtu/ft2-yr
(percent of
SAMSON
Average)
Costs, $/ft
2
-yr
(percent of
SAMSON
Average)
Electric Demand,
W/ft2 (percent of
SAMSON
Average)
Cooling Load,
(Btu/h)/ft2
(percent of
design size)
Heating Load,
(Btu/h)/ft2
(percent of
design size)
SAMSON Average
66.1 0.42 4.1 6.1% -9.2%
Design Size -- -- -- 16.6 35.8
TRY -- -- -- -- --
TMY -0.7% -2.2% -0.6% -0.3% -0.3%
TMY2 -0.9% -1.7% 0.4% 7.6% -9.2%
WYEC -1.9% -2.2% -0.1% 1.4% -7.0%
WYEC2 (TMY) -- -- -- -- --
Denver,
Colorado
WYEC2 (WYEC) -1.2% -1.7% 0.4% 2.9% -7.5%
SAMSON Average
49.9 1.59 4.1 14.2% -25.6%
Design Size -- -- -- 17.1 27.1
TRY 0.4% -0.2% -0.1% 5.9% -32.9%
TMY -0.2% -0.4% 0.3% 17.0% -35.6%
TMY2 -0.8% -0.5% -0.3% 7.1% -32.0%
WYEC 1.2% 0.1% 2.0% 30.5% -37.0%
WYEC2 (TMY) -- -- -- -- --
Los Angeles,
California
WYEC2 (WYEC) 1.2% -0.1% 1.3% 22.8% -35.5%
SAMSON Average
50.3 1.11 4.7 11.1% -43.5%
Design Size -- -- -- 25.2 27.5
TRY -0.2% 0.2% -0.1% 13.6% -43.7%
TMY -0.8% -0.9% -0.3% 7.9% -22.2%
TMY2 -0.6% -0.3% 0.6% 16.3% -23.7%
WYEC -0.8% -0.7% -0.5% 8.8% -37.5%
WYEC2 (TMY) -0.6% -0.6% 0.1% 10.5% -21.8%
Miami,
Florida
WYEC2 (WYEC) -0.8% -0.7% -0.4% 6.9% -38.5%
SAMSON Average
81.4 0.92 4.4 13.2% -4.3%
Design Size -- -- -- 21.2 38.6
TRY 5.4% 3.3% 0.6% 22.6% -2.8%
TMY 2.3% 1.2% 1.4% 20.9% -0.7%
TMY2 -0.4% -0.6% -2.2% -1.5% -7.3%
WYEC 1.6% 1.4% -1.0% 2.4% -5.1%
WYEC2 (TMY) -- -- -- -- --
Minneapolis,
Minnesota
WYEC2 (WYEC) 1.4% 1.2% -1.8% -0.8% -5.8%
SAMSON Average
67.0 2.12 4.4 12.4% -3.7%
Design Size -- -- -- 21.3 33.2
TRY -1.4% -0.9% -0.8% 8.4% -11.2%
TMY 1.1% -0.9% -3.1% -3.3% -7.0%
TMY2 0.2% -0.6% 0.2% 11.7% 3.2%
WYEC 3.2% 1.2% -0.7% 9.0% -1.6%
WYEC2 (TMY) 1.6% -1.9% -6.1% -11.6% 0.1%
New York,
New York
WYEC2 (WYEC) 3.2% 1.1% -0.7% 8.5% -1.6%

TO-98-2-2 15
TABLE 3 (Continued)
Comparison of Simulated Annual Energy Consumption, Energy Costs, Peak Electric Demand,
and Peak Loads for Weather Data Files Types and SAMSON Weather Data
Total Annual Energy Annual Peak
Location Weather File Type Consumption,
kBtu/ft2-yr
(percent of
SAMSON
Average)
Costs, $/ft
2
-yr
(percent of
SAMSON
Average)
Electric Demand,
W/ft2
(percent of
SAMSON
Average)
Cooling Load,
(Btu/h)/ft2
(percent of design
size)
Heating Load,
(Btu/h)/ft2
(percent of
design size)
SAMSON Average 52.4 1.31 4.7 9.3% -33.3%
Design Size -- -- -- 25.7 29.1
TRY 1.0% -1.3% 0.7% 15.3% -15.6%
TMY 0.2% -1.0% -0.4% 4.5% -26.5%
TMY2 -0.1% -0.4% -0.6% 8.4% -27.7%
WYEC 0.2% -0.6% -0.2% 8.5% -46.3%
WYEC2 (TMY) 0.0% -1.2% -0.8% 2.5% -28.1%
Phoenix,
Arizona
WYEC2 (WYEC) 0.0% -0.9% -0.2% 7.3% -48.5%
SAMSON Average 63.9 0.58 4.0 12.3% -15.3%
Design Size -- -- -- 16.1 30.3
TRY 3.9% 1.9% 1.0% 19.6% -14.0%
TMY 2.5% 1.4% -0.6% 7.6% -18.5%
TMY2 -0.2% -0.1% -0.3% 5.3% -20.6%
WYEC 2.8% 1.5% -0.9% 5.3% -14.0%
WYEC2 (TMY) 2.5% 1.4% -0.7% 7.3% -17.8%
Seattle,
Washington
WYEC2 (WYEC) 2.7% 1.5% -1.4% 2.0% -16.5%
SAMSON Average 63.8 1.23 4.5 9.8% -7.0%
Design Size -- -- -- 22.3 32.9
TRY -2.3% -1.3% -1.4% 0.6% -9.4%
TMY 0.2% -0.3% -0.7% 2.2% -5.1%
TMY2 1.4% 0.7% 1.5% 19.6% -7.3%
WYEC -0.9% 0.1% 0.9% 23.6% -12.0%
WYEC2 (TMY) 0.3% -0.2% -0.5% 4.4% -5.0%
Washington,
D. C.
WYEC2 (WYEC) -0.9% -0.1% 0.7% 23.8% -12.2%
SAMSON period of record (1961-1990) differ, TRY data
could include years that are either hotter or colder than
those in the SAMSON data. For example, the TRY data
for Minneapolis (Figure 12) resulted in significantly
higher energy consumption and costs; in fact, the energy
costs were outside the range of values from the SAMSON
data. The TRY data had a winter design condition below
that of all the SAMSON data and solar data on the low
end of the range as well. Unlike Figures 1 through 8,
Figures 9 through 16 exhibit a relatively higher range of
variability in energy costs but the range of variation is still
small, ranging from –2.2% to 3.3% including the TRY
data. With the exception of Washington, D.C., the TMY2
consistent provide a closer match to the average energy
consumption of the SAMSON data. With a few
exceptions (New York, Seattle, and occasionally WYEC
and WYEC2), simulations using the typical weather data
sets under-predict the energy consumption and energy
costs.
The last set of Figures (17 through 24) presents
another aspect of the impact of weather selection on
energy performance simulation: annual peak cooling and
heating loads. These figures compare the variation in
peak annual cooling and heating loads from the
simulations using the 30 years of SAMSON data and the
weather data file types. The left graph of each figure
shows the annual peak-cooling load as a fraction of the
average annual peak-cooling load for the 30 years. The
right side shows similar information for the annual peak-
heating load. The horizontal line shown on each graph is
the calculated peak design size based on the design
conditions (2-1/2% for cooling and 99% for heating) for
the location from the 1993 Fundamentals (ASHRAE
1993). For the SAMSON simulations (1961-1990), the

TO-98-2-2 16
mean of the annual peak loads is shown as a diamond near
the center of the left-hand vertical line on each graph. The
vertical line represents the range of annual peak loads for
the SAMSON simulations—maximum to minimum. The
loads from the weather data sets are shown as a fraction of
the mean of the annual peak loads from the SAMSON
simulations. The values for the weather data files types are
shown as a scatter of diamonds to the right of the
SAMSON vertical line.
As would be expected, annual peak cooling and
heating load vary more than do either the annual energy
consumption (summed hourly energy) or the annual
energy costs (monthly peak demand and summed hourly
energy consumption). The peaks depend on the how
much the building is affected by the hourly temperature
fluctuation and incident solar radiation. The range of
percentage variation of the peak loads as a function of the
design sizing (see above) is shown in the two right-hand
columns of Table 3.
In reviewing Figures 17 through 24, several
observations about the peak cooling and heating loads
become apparent. First, the heating design size values are
generally higher than the peak heating loads of both the
30-year data set and the typical weather data sets. On the
other hand, cooling design size is generally close to or less
than the peak cooling loads. This seems to be related to
the use of the more conservative 99% design conditions
for heating and the more generous sizing allowed for
heating by the commercial building energy standards
(ASHRAE 1989). The energy standards allow heating
equipment to be sized up to 40% larger than the annual
peak-heating load calculated based on the design
conditions. On the other hand, the energy standards only
allow cooling equipment to be sized up to 20% larger than
the calculated annual peak-cooling load. For cooling, the
combination of less conservative 2-1/2% design
conditions and the lower over sizing allowance means that
for a few hours every year, the cooling equipment may not
be able to meet the load.
Overall the variation in annual peak cooling load
ranged from 11.5% below the design size to 30.5% above.
Note that in all the locations, the range of cooling loads
from the SAMSON simulations was greater than that of
the weather data sets. For annual peak heating loads, the
variation among the weather data sets ranges from 48.5%
below the design size to 3.2% above. The locations with
the greatest heating over sizing were those with relatively
low heating loads: Los Angeles, Miami, and Phoenix.
SUMMARY
As described above, the range of annual energy
consumption and costs and peak cooling and heating loads
due to actual weather variation over a 30-year period can
be significant; in this case, the SAMSON data set. For the
eight locations in this study,
• annual energy consumption varied as much as –
11.0%% to 7.0%,
• annual energy cost varied from –4.6% to 3.6%,
• annual peak electrical demand varied from –4.7% to
4.9%,
• annual peak cooling loads ranged from 11.5% below
the design size to 30.5% above, and
• annual peak heating loads ranged from 48.5% below
the design size to 3.2% above.
The variation in energy consumption is similar to that
reported by Haberl (1995) for measured and TMY
weather data. Haberl showed DOE-2 predicted energy
consumption values that were consistently 5% to 10 %
higher than the measured energy consumption.
Before beginning to discuss the weather data types
results, it is important to note again that the design
conditions in the 1993 Fundamentals (ASHRAE 1993)
and the TRY, TMY, WYEC, and WYEC2 data sets have
roughly the same period of record—~1945-1973. The
design conditions in the 1997 Fundamentals (ASHRAE
1997b), SAMSON 30-year and TMY2 have the same
period of record, 1961-1990, and data source. Thus, these
three should exhibit similar results.
Of the six weather data types studied in this work
[TRY, TMY, TMY2, WYEC, and WYEC2 (TMY and
WYEC)], TRY showed the most variation, higher and
lower (except in mild Los Angeles and hot Miami). This is
demonstrated in the annual energy costs for Minneapolis
(Figure 12) where the TRY values exceed the maximum
from the SAMSON simulations. The TRY is a year
outside the SAMSON period of record with more severe
winter design conditions (-25F) than the lowest of the
30-year period for SAMSON (-24F). The TRY also had
higher than average (SAMSON) heating and cooling
degree-days, and the annual average solar radiation was
toward the low end of the range for SAMSON. Another
example is Washington, D.C. where both the annual
energy consumption and energy costs are lower than all
the other weather types even though it is still within the
30-year range shown for SAMSON. It has lower than
average (SAMSON) heating degree-days and higher than
even the SAMSON maximum cooling degree-days—
another example of the year selected for the TRY being
outside the SAMSON 30-year period.
For the six weather data types, the range of variation
from the SAMSON average and the design size, as shown
in Table 3, is as follows.
• TRY, annual energy consumption from –2.3% to
5.4%; annual energy costs from –1.3% to 3.3%;
annual peak electric demand from –1.4% to 1%;
annual peak cooling load from 0.6% to 22.6%; and
annual peak heating load from –43.7% to –2.8%.

TO-98-2-2 17
• TMY, annual energy consumption from –0.8% to
2.5%; annual energy costs from –1.7% to 1.4%;
annual peak electric demand from –3.1% to 0.6%;
annual peak cooling load from –3.3% to 20.9%; and
annual peak heating load from –35.6% to –0.3%.
• TMY2, annual energy consumption from –0.9% to
1.4%; annual energy costs from –1.7% to 0.7%;
annual peak electric demand from –2.2% to 1.5%;
annual peak cooling load from –1.5% to 19.6%; and
annual peak heating load from –32.0% to 3.2%.
• WYEC, annual energy consumption from –1.9% to
3.2%; annual energy costs from –2.2% to 1.5%;
annual peak electric demand from –1.0% to 2.0%;
annual peak cooling load from 1.4% to 30.5; and
annual peak heating load from –46.3% to –1.6%.
• WYEC2 (TMY), annual energy consumption from –
0.6% to 2.5%; annual energy costs from –1.9% to
1.4%; annual peak electric demand from –6.1% to
0.1%; annual peak cooling load from –11.6% to
10.5%; and annual peak heating load from –28.1% to
0.1%.
• WYEC2 (WYEC), annual energy consumption from
–1.2% to 3.2%; annual energy costs from –1.7% to
1.5%; annual peak electric demand from –1.8% to
1.3%; annual peak cooling load from -0.8% to 23.8%;
and annual peak heating load from –48.5% to –1.6%.
By limiting the selection method for the WYEC to
dry-bulb temperature, the resulting data set is not as
representative of the period of record. Note that the solar
radiation data for WYEC in Table 1 are often near the
high or low end of the range for the SAMSON
(recognizing that WYEC and SAMSON have different
periods of records).
Simulations using the TMY2 data set more
consistently match the simulation results for the SAMSON
30-year period than any other data set. Some of this can
be attributed to the TMY2 and SAMSON having the same
period of record and data source. This suggests that a
data selection method, such as TMY2, that evaluates a
composite weighting of each month for multiple variables
(solar radiation, dry bulb temperature, dew point
temperature, and wind velocity) provides better simulation
results, i.e., closer to the mean for the period of record. In
several of the locations that are more temperature-
dominated than solar-dominated (Los Angeles and
Washington, D.C.), the TMY2 appears not to match the
long-term temperature averages as well. This suggests
that the weights assigned to the weather variables should
be adjusted. The TMY2 design temperatures further
support this—they occasionally do not match those from
the 1997 Fundamentals (see Table 1)—even though both
are derived from the same period of record and data set.
RECOMMENDATIONS
Users of energy simulation programs should avoid
using single year, TRY-type weather data. No single year
can represent the typical long-term weather patterns.
More comprehensive methods that attempt to produce a
synthetic year to represent the temperature, solar
radiation, and other variables within the period of record
are more appropriate and will result in predicted energy
consumption and energy costs that are closer to the long-
term average. Both TMY2 and WYEC2 use this type of
method, are based on improved solar models, and more
closely match the long-term average climatic conditions.
We have several recommendations for developers of
future weather data sets. The TMY2 method appears to
work well in most cases but the resultant files may need to
be adjusted to match the long-term average statistics more
closely, the mean of the 30-year period of record in this
case. A second approach would be to create a typical
weather file that has three years: typical (average),
cold/cloudy, and hot/sunny. This would capture more
than the average or typical conditions and provide
simulation results that identify some of the uncertainty and
variability inherent in weather.
The method used in this paper needs to be attempted
on a broader geographic scale with more typical weather
data sets and actual weather data. In a similar study of
residential buildings, Huang (1998) evaluated the impact
of weather data on heating and cooling loads. The author
also believes that a similar approach should be taken to
determine if weather data selection methods affect energy
and loads in smaller, enveloped-dominated and larger,
internal-load dominated commercial buildings (<10,000
ft2 and >100,000 ft2).
Which weather data should you use for simulating
commercial buildings? From this study, we believe that
either the TMY2 or WYEC2 data sets will provide users
with energy simulation results that most closely represent
typical weather patterns.
REFERENCES
ASHRAE. 1985. Weather Year for Energy Calculations.
Atlanta: American Society of Heating, Refrigerating
and Air-Conditioning Engineers, Inc.
ASHRAE. 1989. ANSI/ASHRAE/IESNA Standard 90.1-
1989, “Energy Efficient Design of New Buildings
Except Low-Rise Residential Buildings.” Atlanta:
ASHRAE.
ASHRAE. 1993. 1993 ASHRAE Handbook—
Fundamentals. Atlanta: ASHRAE.
ASHRAE. 1997a. WYEC2 Weather Year for Energy
Calculations 2, Toolkit and Data, Atlanta: ASHRAE.
ASHRAE. 1997b. 1997 ASHRAE Handbook—
Fundamentals. Atlanta: ASHRAE.

TO-98-2-2 18
California Energy Commission. 1992. Climate Zone
Weather Data Analysis and Revision Project, P400-
92-004. Sacramento: California Energy Commission.
California Energy Commission. 1994. The Adjustment of
Title 24 Climate Zone Weather Data to Reflect Cities’
ASHRAE Design Day Conditions, Staff Report, 20
July 1994. Sacramento: California Energy
Commission.
Commission of the European Community. 1985. Test
Reference Years, Weather data sets for computer
simulations of solar energy systems and energy
consumption in buildings, CEC, DG XII. Brussels,
Belgium: Commission of the European Community.
Crow, Loren W. 1970. “Summary Description of Typical
Year Weather Data, Chicago Midway Airport,”
ASHRAE Research Project RP-100. Atlanta:
ASHRAE.
Crow, L.W. 1980. “Development of Hourly Data for
Weather Year for Energy Calculations (WYEC),
Including Solar Data, at 21 Weather Stations
throughout the United States,” ASHRAE Transactions
87 (1). Atlanta: ASHRAE.
Crow, L.W. 1983. “Development of Hourly Data for
Weather Year for Energy Calculations (WYEC),
Including Solar Data, at 29 Stations throughout the
United States and 5 Stations in Canada,” ASHRAE
Research Project RP-364 Final Report, November
1983. Atlanta: ASHRAE.
Haberl, Jeff S., Doug J. Bronson, and Dennis L. O’Neal.
1995. “Impact of Using Measured Weather Data vs.
TMY Weather Data in a DOE-2 Simulation,” in
ASHRAE Transactions 101 (2). Atlanta: ASHRAE.
Huang, Y. Joe. 1998. “The Impact of Different Weather
Data on Simulated Residential Heating and Cooling
Loads,” in ASHRAE Transactions 104 (2). Atlanta:
ASHRAE.
Maxwell, E. 1990. “Producing a 1961-1990 Solar
Radiation Data Base for the United States,” in Solar
‘90 Technical Papers: Proceedings of the 1990
Annual Conference, American Solar Energy Society.
Boulder: American Solar Energy Society.
NCDC. 1976. Test Reference Year (TRY), Tape
Reference Manual, TD-9706, September 1976.
Asheville, North Carolina: National Climatic Data
Center, U.S. Department of Commerce.
NCDC. 1981. Typical Meteorological Year User’s
Manual, TD-9734, Hourly Solar Radiation—Surface
Meteorological Observations, May 1981. Asheville,
North Carolina: National Climatic Data Center, U.S.
Department of Commerce.
NCDC. 1993. Solar and Meteorological Surface Obser-
vation Network, 1961-1990, Version 1.0, September
1993. Asheville, North Carolina: National Climatic
Data Center, U.S. Department of Commerce.
National Renewable Energy Laboratory (NREL). 1995.
User’s Manual for TMY2s (Typical Meteorological
Years), NREL/SP-463-7668, and TMY2s, Typical
Meteorological Years Derived from the 1961-1990
National Solar Radiation Data Base, June 1995, CD-
ROM. Golden, Colorado: National Renewable
Energy Laboratory.
Perez, R., P. Ineichen, R. Seals, J. Michalsky and R.
Stewart. 1990. “Modeling Daylight Availability and
Irradiance Components from Direct and Global
Irradiance,” in Solar Energy, Vol. 44, pp. 271-289.
Perez, R. R., P. Ineichen, E. Maxwell, F. Seals and A.
Zelenda. 1991. “Dynamic Models for Hourly
Global-to-direct Irradiance Conversion,” in 1991
Solar World Congress: Proceedings of the Biennial
Congress of the International Solar Energy Society,
Vol. 1 (II): pp. 951-956.
Petrakis, M. 1995. “Test Meteorological Years:
Procedures for Generation of TMY from Observed
Data Sets,” in Natural Cooling Techniques, Research
Final Report, PASCOOL, Alvarez et al., ed.
Brussels, Belgium: Commission of the European
Commission.
Stoffel, T.L., and M. Rymes. 1997. Production of the
Weather Year for Energy Calculations Version 2
(WYEC2), NREL TP-463-20819. Golden, Colorado:
National Renewable Energy Laboratory.
Stoffel, Thomas L., and Martin D. Rymes. 1998.
“Production of the Weather Year for Energy
Calculations version 2 (WYEC2) Data Files,” in
ASHRAE Transactions 104 (2). Atlanta: ASHRAE.
WATSUN Simulation Laboratory. 1992. Engineering
Data Sets of Hourly Weather Observations in
WYEC2 Format (WYEC2 Files) and Canadian
Weather for Energy Calculations (CWEC Files),
User’s Manual, 15 September 1992, prepared for
Environment Canada Atmospheric Environment
Service and National Research Council Canada.
Waterloo, Ontario: WATSUN Simulation Labora-
tory, University of Waterloo.
Winkelmann, F.C., W.F. Buhl, B. Birdsall, A. E. Erdem,
and K. Ellington. 1994. DOE-2.1E Supplement, DE-
940-11218. Lawrence Berkeley Laboratory,
Berkeley, California. Springfield, Virginia: National
Technical Information Service.