ESL-HH-88-09-47

FIELD VALIDATION OF THE ASHRAE TRANSFER FUNCTION METHOD

FOR CALCULATING COOLING LOAD

H.J. BRAUD, P.E. TED QUILLE JASON C. SHIH, P.E.

Professor Research Associate Professor
Louisiana Agricultural Experiment Station School of Architecture
Louieiana State University Agricultural Center Louisiana State Univereitv
Agricultural ~ngineeringDepartment Baton Rouge. LA
Baton Rouge, Louieiana

ABSTRACT - government agencies and universitiee. Gadgil et al.

(1) demonstrated that the majority of theee large
The American Society of Heating, Refrigerating mainframe programs were equivalent. In fact, many
and Air Conditioning Engineers (ASHRAE) Tranefer of the program algorithms were based on ASHRAE
Function Method (TFM) was validated using two proceduree.
identical wood frame residential-type etructuree,
each containing an east and weet-facing room. Each An hour-by-hour profile of the cooling load ie of
r o w had a calibrated, thermostatically controlled interest ae an energy consumption predictor since
window air conditioning unit and two south-facing instantaneous heat gains are closely matched with
windows. The study included a parametric analysia the instantaneoue cooling load and epace heat
of the thermoetat setpoint and fenestration load extraction rate. In addition to giving the maximum
effect on apace heat extraction rate and cooling cooling load which the air-conditioner must meet,
energy coneumption. Some discrepancies such a cooling load profile identifies the time of
vithstanding, the transfer function method predicted day at which it occurs, and the method allows
the hourly heat extraction ratee quite well. The estimation of daily energy coneumption to provide
principal diecrepanciee appeared to be the the cooling.
difference in daily curve amplitude and a phaee-like
shift of one to two hours. The heat etorage Before describing the Tranefer Function Method,
capacity of the unoccupied test buildings wae less some definitions are relevant at thie point.
than predicted by the TFM model. Accuracy of the
tranefer function coefficients to model the 1. Space heat gain is the instantaneoue rate at
roof-ceiling combination was questionable due to the which heat enters or is generated in an encloeed
emall attic air epace which was not accurately space.
described in the ASHRAE table of coefficients.
2. The apace cooling load is the rate at which heat
INTRODUCTION must be removed to maintain a constant epace
temperature.
Cooling loads account for the major portion of
the residential utility bill during summer in the 3. The heat extraction rate is the rate of heat
South. Sizing of air-conditioning equipment to removal by the air-conditioning eyetem from the
handle the loads is accomplished with one of many space, and it equals the epace cooling load only if
atandard procedures such as the American Society of the epace temperature remains constant. In a normal
Beating, Refrigerating and Air-Conditioning daily cycle temperature changee slowly, so that at
Engineers (ASRRAE) Cooling Load Temperature any time the heat extraction rate can be taken as
Difference (CLTD) method or the Air Conditioning equal to the cooling rate.
Contractore of America (ACCA) Manual J method.
These analytical proceduree estimate the maximum The ASHRAE Tranefer Function Method (TFU) was
cooling load by using steady state type equatione developed by Stephenson and Mitalae (2). The model
based on fundamental laws of conduction heat utilizes transfer functione, aleo called weighing
transfer. These methods are obviously approximate, factors or thermal reevonse factors. comvuted from
ns such important factors ae heat etorage and rigorous calculations involving the governing
outside surface orientation are only marginally equations of transient heat transfer applied to
included in the calculatione. A more precise method typical residential structuree. In developing the
ie the ASHRAE Transfer Function Method (TFM). This Transfer Function Method. Stephenson and Mitalae
model utilizes transfer functione, computed from considered the governing equatione for a surface
rigoroue calculatione involving the governing heat balance of the six eurfaces of a room and those
equations of heat tranefer applied to typical governing the transient heat conduction within the
residential structures, to calculate epace cooling building components adjoining those surfaces.
loads and heat extraction ratee on an hour-by-hour Initial determination of the coefficients involved a
basis. matrix formulation. However, a more recent
formulation of the coefficients utilized
There are many other rigoroue hour-by-hour 2-transforms (Stephenson and Mitalae, (3); Mitalae,
cooling load programs developed by the varioue (4)).

Proceedings of the Fifth Symposium on Improving Building Systems in Hot and Humid Climates, Houston, TX, September 12-14, 1988
 ESL-HH-88-09-47

A useful validation of the ASBRAE model for transfer coefficients used represented less attic
residential interests would involve a relatively ineulation than was the actual case.
small, tight etructure that is still representative
of residential structures and one in which the more M s MEASUREMENT
uncertain heat gain components, such as air
infiltration and internal sensible and latent loads, SPACE HEAT GAIN
may be neglected entirely in the modelling. Such a (Conduction heat gain through
structure was available st Louisiana State wall, ceiling and floor;
Univergity, School of Architecture. In fact. two fenestration load)
300 ft identical, fully instrumented wood-frame II
buildings, with two test rooms each, were used.

OBJECTIVES
Z - transfer function
I
COOLING LOAD
I
I
I
The following were the project objectives: Z - tranefer function
I
1. Validate the A S H M E Transfer Function Method of

- traneferI function
predicting the hourly heat extraction by the
air-conditioning system. Z compare
I
2. Test certain aspects of the model by

a. changing the thermostat setpoint temperature. Fig. 1. Relation of transfer function to field
b. changing the fenestration load. measurements of room cooling load and
heat extraction rate.
COMPUTER MODEL OF THE TRANSFER FUNCTION METHOD
TEST BUILDINGS
Figure 1 showe a schematic relation of the
transfer function method for predicting cooling The validation procedure consisted of measuring
load, space extraction rate and ultimately indoor hourly temperatures and cooling energy consumption
space temperature. It is essentially the logic in two identical, skid-mounted, test houses, each
diagram for the computer model written to utilize containing an east and west-oriented room with an
the transfer function method. Reat gain instrumentation room between. The buildinge ware of
calculations were based on conduction through light wood-frame construction, and the walls were
exterior walle, floor. and the roof-ceiling insulated with R-11 of fiberglass insulation. the
combination. Except in the case of the floor, these attic and floor with R-19 of the same ineulation.
calculations were based on sol-air temperatures. Figure 2 gives details. Each room initially
Solar and conduction heat gain through windows was contained two windows. However, since the
the only other component of heat gain considered. fenestration area to floor area ratio was considered
Infiltration was assumed to be negligible because to be much larger than usual residential
all windows and air-conditioners were tightly construction, one window in each room was covered,
sealed. Also, all doors contained magnetic strips. sealed, and insulated to wall levels. Thie left a
In addition, the rooms were always unoccupied and fenestration area to floor area ratio of 11.5.
contained no internal heat or moisture. Complete Unlike most residential structures, however, there
model information and validation results were was no roof overhang or exterior or interior
reported by Quille (5). shading. This absence of shading was actually
advantageous when it came to modelling solar gains.
The test buildings had 114 slope pitched roofs. Each room had a window-type air-conditioning unit
Due to the absence of tabulated transfer (Heat Controller, Inc.. Jackson. MI) of 4,600 Btu/hr
coefficients for pitched roofs in the ASHRAE capacity. Thermostats for each air-conditionerowere
handbooks, (6,7). the roofs were modelled as being Accuetate, Model TKS-AH-23 with precision ( 0.5 F)
horizontal. The ASERAE equations for conduction sensors, mounted on the interior walle of a room.
heat gain were used to calculate the heat gain
through exterior walle and the roof-ceiling INSTRUMENTATION
combination. The AS€lRAE coefficients associated
with the designation of "light" construction was The pertinent measurements to the analysis were
used , as opposed to "medium" or "heavy" air-conditioning power usage and inside temperature.
construction. The latter was accomplished with a shielded
thermocouple (Type J) located close to the interior
The transfer function coefficients were choeen instrumentation room wall. This point was chosen as
from ASHRAE tables to match as closely as possible it was away from direct air currents set up by the
the actual structure. Thie was relatively easy for air-conditioner and free from direct solar gains.
the walle because there were 103 different wall All sensors were polled every seven seconds and the
sections in the ASHRAE literature from which to data were recorded and then averaged over hourly
choose transfer coefficients. But, as pointed out intervals on the hour with a data logger (Doric
above, the roof was of low elope while ASERAE Scientific, Inc., Digitrend 235. San Diego, CA).
coefficients are given for horizontal. and the Other parameters measured were wall and ceiling

Proceedings of the Fifth Symposium on Improving Building Systems in Hot and Humid Climates, Houston, TX, September 12-14, 1988
 ESL-HH-88-09-47

-
9'-10' #-/om
3\on f-9- 3io- /'##
I I r I I I I I t
I i I I I I

h
h

-LA

II

A# & a 3 ' ~ 1 6 ! # r l %

Fig. 2. Floor plan and wall section of test building.

surface temperaturea, temperature at the thermostat The first run values were measured during a five
location. solar radiation and attic temperaturea. minute on-time and the second run during a ten
minute on-time. Normal on-time was closer to five,
Solar radiation data were taken with a minutes during air-conditioner operation.
pyranometer (Epply Laboratory, Inc., Model PSP,
Newport, R.I.) mounted on the vertical surface of EXPERIMENTAL PROCEDURE
the south building.
Three tests were run to compare measured hourly
Recent research (8) has ehown that attic cooling energy to that predicted by the transfer
insulation temperatures may be higher than attic air method model. The teat period was August and
temperatures during midday houre. Thus, roof September, 1987.
eurface, attic air, and attic insulation
temperatures at a depth of 1 inch were measured. 1. The first experiment was a base case in that all
four rooms had identical thermostat setpointe (78'~)
Electric energy consumed by an air-conditioner and identical windows (double pane). Thus, in
was converted to millivolts by a watt transducer. addition to comparison with the ASHRAE model
Power consumption was calibrated with a watt-hour predictions, inherent room differences were
meter (Ohio Semitronica Inc., Model WH3, Columbus, determined. The test wae repeated later with all
Ohio). single pane windows.

COP DETERMINATION FOR THE AIR-CONDITIONING UNITS 2. The second experiment involved a atudy of the
influence of thermostat setpoint on heat extraction
The coefficient of performance (COP) is defined and cooling load. The north buildin was operated
as a ratio of the total heat removed by a cooling
8
at 78'~ and the south operated at 74 F indoor
unit divided by the electrical energy input. This temperature.
ratio had to be determined for each individual room
air-conditioning unit, since the cooling unit power 3. This was a window study. Rooms E-S and W-S of:
consumption would be measured during experiments. the south building had double pane windows which was
For COP determination, a metal duct was fabricated compared to single pane windows in the north
and placed on the air diecharge of each room cooling building.
unit. Simultaneously with the temperature
meaeurements, air speed was determined with a hot DETERMINATION OF INHERENT DIFFERENCES IN TEST ROOMS
vire anemometer (TSI, Model 1610-12, St. Paul, MN).
This inetrument had previouely been calibrated in a The first analysis examined the inherent
wind tunnel. The COP value8 determined from these differences that could be discerned between the four
measurements are reported in Table 1. test rooms with the same thermostat settings and
same window treatment. As the computer model '
Room E-N W-N E-S W-S a s ~ u m e dcloudless skies, only those days which were
the sunniest and cloudless were chosen, Auguet 20.
Run 1 1 .26 1.21 1.24 1.37 22. 24. and 28. for all double pane windows. and
Run 2 1.55 1.45 1.53 1.40 Sept. 10, 12, and 13, for all single pane windows.

Table 1. Coefficient of Performance (COP) for Data collected and model predictione are
the Four Room Air-conditioning Units. presented in Figure 3 and Table 2. Very erratic ,

Proceedings of the Fifth Symposium on Improving Building Systems in Hot and Humid Climates, Houston, TX, September 12-14, 1988
 ESL-HH-88-09-47

operation of the air-conditioning unit in west room

W-S was noticed during the tests with double pane
windows. Eventual failure of the thermostat 4m -
required that it be replaced. Therefore, data for
this test room were recorded but not used in 100 -
subsequent analyses.

Referring to Figure 3 for the east rooms, the

principal discrepancies between the heat extraction - -2
IQ)
rates predicted by the TFM model and measured appear 9
to be the difference in amplitude and the phase-like
shift of one to two hours. This shift, which begins
to appear soon after daybreak, suggests that the
heat storage capacity of the buildings was less than
the model. This would also help explain the
x
difference in peak loads. Room E-N (double pane)
peak load was 7% larger than E-S and 41% higher than
the TFM model peak, Table 2. Room E-S had a 32%
higher than model peak for double pane and 18% with 2 4 6 8 10 I2 14 16 la 10 1R 84
single pane. The difference between east rooms was SOLAR TIME I tlounl
only 7% for double pane and 18% for single pane.
The west room W-S had a 11% (single pane) higher
peak load than the model predicted, and room W-N had
a 10% lower peak. Room difference was 19% single Pig. 3. Experimental and predicted heat
pane. These results were used to normalize the data extraction rates for rooms E-N and E-S.
in the following experiments. Double pane windown.

Table 2. Comparison of Measured Peak Heat Extraction and Cooling Energy Coneumption to TFM Madel.
- -

Room Room
Set Window East-N Diff. East-S Diff Model-E W-N Diff. W-S Diff. Model-W

BASE STUDY

Peak watts 780 Double 699 32% 530

78' Double - 41% 530

Peak watts 78' Single 798 18% 679

78' Single - -3% 679

Energy watt-hr 78' Double 6069 7% 5655

78' Double - 4% 5655

Energy watt-hr 78' Single 5393 -22% 6902

78' Single - -30% 6902

THERMOSTAT SET POINT

Peak watts 78' Double -- -11% 583

Difference 8%
Peak watts 74' Double 769 22% 630

Energy watt-hr 78' Double - 37% 6063

Difference - - 18%
Energy watt-hr 74' Double 5936 17% 7164

WINDOW STUDY

Peak watts 78' Single - 21% 699

Difference - - 9%
Peak watts 78' Double 732 15% 639

Energy watt-hr 78' Single - 12% 7066

Difference - - 8%
Energy watt-hr 78' Double 5430 17% 6529

*Data void due to faulty thermostat.

Proceedings of the Fifth Symposium on Improving Building Systems in Hot and Humid Climates, Houston, TX, September 12-14, 1988
 ESL-HH-88-09-47

Besides h e a t s t o r a g e , o t h e r f a c t o r s might e x p l a i n thermoetat e e t p o i n t t o 7h°F, Table 2. The meaeured

diecrepancies between t h e meaeured and p r e d i c t e d hourly i n c r e a e e wae 47% a f t e r adjuetment based upon
data. A f l a t roof was assumed i n t h e modelling of t h e room d i f f e r e n c e 8 found i n t h e base study.
t h e r o o f - c e i l i n g combination. Also, f o r t h e Energy coneumption i n c r e a e e f o r lowering t h e
ceiling-attic-roof combination, a c e i l i n g U-value thermoetat s e t p o i n t t o 7h°F was 18% p r e d i c t e d by t h e
l a r g e r than t h e a c t u a l one was ueed i n t h e model. model, while t h e meaeured d a i l y i n c r e a s e was 49% f o r
However, a e McQuiston (9) noted, p a r t l y due t o t h e t h e west room.
a t t i c a i r apace r e s i s t a n c e ueed i n t h e r o o f - c e i l i n g
heat gain transfer function coefficients, t h e i r INFLUENCE OF FENESTRATION LOAD
a p p l i c a b i l i t y t o t h i s b u i l d i n g component i e l i m i t e d .
The t h i r d experiment examined room h e a t
Over t h e period from 1100 houre t o 1800 houre, e x t r a c t i o n r a t e e f o r two d i f f e r e n t window types.
f o r t h e two e a s t roome, t h e average meaeured h e a t s i n g l e pane windowe i n t h e n o r t h b u i l d i n g and double
e x t r a c t i o n r a t e was 12.3% higher than p r e d i c t e d , and pane windowe i n t h e south. The experimental d a t a
f o r one weet room W-N 11.4% higher. covered t h e daye September 14-15, and 17, Figuree
5-6.
Close agreement i n t o t a l d a i l y cooling energy
were i n evidence. Measured energy f o r t h e e a e t
roome were 7% and 4% h i g h e r than t h e p r e d i c t e d by
t h e model. Correspondingly, f o r t h e weet room, t h e
meaeured energy wae 1%h i g h e r than p r e d i c t e d w i t h
double pane b u t 31% below w i t h s i n g l e pane.

INFLUEUCE OF THERMOSTAT SETPOINT

The e f f e c t e of lowering t h e room thermoetat

s e t p o i n t were i n v e s t i g a t e d . To achieve t h i e , t h e
n o r t h b u i l d i n g roome had t h e i r thermoetate e e t t o
7 8 O ~ , and roome of t h e mouth b u i l d i n g had t h e i r
e e t p o i n t e e e t t o 74'~. Figure 4 provides graphic
reeulte.

A e h i f t i n time of t h e peak load i e eeen again.

The d i f f e r e n c e i n peak load i s smaller t h a n i n t h e
previoum t e n t , becaume t h e e l i g h t l y l e e e sunny
weather conditione during t h i s experiment. Peak Fig. 5. I n f l u e n c e of f e n e e t r a t i o n load on
h e a t e x t r a c t i o n r a t e e f o r Rooms E-N (78'~) and E-S meaeured and p r e d i c t e d h e a t e x t r a c t i o n
(74'~) were 11%and 22% s m a l l e r than p r e d i c t e d , r a t e s f o r rooms E-N and E-S.
r e e p e c t i v e l y . The corresponding f i g u r e s f o r Room
W-N was 1% l a r g e r than predicted. T o t a l d a i l y h e a t
e x t r a c t i o n energy v n r i e d coneiderably, probably
900 -
because of i d e a l sunny c o n d i t i o n s . BOO-
-

-
;TOO -

E 000 -
2
p ,oo -

--
4
0

4m-
w
x
I-

W
200-

SOLAR TIME I Hoursl

Fig.6. I n f l u e n c e of f e n e e t r a t i o n load on meaeured

and p r e d i c t e d h e a t e x t r a c t i o n r a t e e f o r
Fig. 4. Influence of thermoetat e e t p o i n t on room W-N.
meaeured and p r e d i c t e d h e a t e x t r a c t i o n
r a t e e f o r r o o w E-N and E-S. Double pane Overall, t h i e t e a t displayed the cloeeet
windows. agreement between meaeured and p r e d i c t e d d a t e .
Meeeured peak loade were s t i l l m a r g i n a l l y h i g h e r
The model p r e d i c t e d e n 8% i n c r e a e e i n peek pt then modelled, 21% f o r s i n g l e pane and 15% f o r
e x t r a c t i o n f o r an e a e t room upon going from 78 F double pane, e a s t roome. For t h e weet r o o m , t h e

Proceedings of the Fifth Symposium on Improving Building Systems in Hot and Humid Climates, Houston, TX, September 12-14, 1988
 ESL-HH-88-09-47

data yielded a 15% larger measured peak for single rooms and 10% for the west rooms. Finally.
pane and 18% larger peak for double pane. predicted hourly temperatures tended to exceed the
meaeured onee by one degree. This was relatively
The main interest was the energy savinga to be close agreement considering the difficulty in
made by reducing the window load. The TPH model precisely modelling thermostat operation.
predicted an 8% savings for double pane over single
pane. After data adjustment in light of the room REFERENCES
differences found in the base study, the measured
savings for double pane over single pane was 12% in Gadgil. A., D. Goldstein, R. Kammarud, and J.
east rooma and 10% in west rooms. Moss. 1979. Residential building simulation
model comparison using several building energy
analyeis programs. Proceedings of the 4th
National Paeeive Solar Conference, pp.
The ASRRAE transfer function method of calculating 187-190.
heat extraction rates was validated using two
identical teat buildings. Three test runs were Stephenson. D.G. and G.P. Mitalas. 1967.
performed: Cooling load calculationa by thermal responae
factor method; and Room thermal reeponse
1. The four test rooms were operated with the same factors. ASHRAE Transactions. Pt. 1, pp.
indoor thermostat setting of 78'~ and with the 111.1.1-111.1.7. and 111.2.1-111.2.10.
same windows to discern inherent room
differences. Stephenson. D.G. and G.P. Mitalas. 1971.
Calculation of heat conduction tranafer
2. The second test was a atudy of thermostat functions for multi-layer slaba. ASBRAE
setpoint with two r o w s set at 78°F and the Transactione, Pt. 2. pp. 117-126.
other two at 74'~.
Mitalas, G.P. 1978. Comments of the 2-tranefer
3. The third study compared single pane to double function method for calculating heat tranefer
pane windows. in buildings. ASHRAG Transactione. pt. 1, pp.
667-673.
A pronounced shift in phase and amplitude between
the measured and predicted heat extraction rates Quille, T.J. 1987. Validation of the ASHIUE
curves in the early daya (August) of the experiment Transfer function method of calculating space
suggested that heat storage and peak solar heat extraction rates. Master's report. Ag.
conditions were not modelled accurately then. Engineering Dept., Louisiana State Univereity,
However, in the later days of the experiment Baton Rouge, LA. Unpublished.
(September) these differences diminished. Since the
prevailing external temperatures remained more or ASHRAE. 1985. Handbook of fundamentals.
leee the same over the period August 20 to September American Society of Beating, Refrigerating and
17. this suggests that the model is very sensitive Air-conditioning Engineers Inc.. Atlanta. GA.
to solar altitude.
ASRRAE. 1981. Handbook of Fundamentale.
Some discrepancies withstanding. the tranefer American Society of Heating. Refrigerating and
function method predicted the hourly heat extraction Air-conditioning Engineere, Inc.. Atlanta, GA.
rates quite well. The principal diacrepancies
appeared to be the difference in daily curve Rish. J.W. 1985. A numerical and experimental
amplitude and a phase-like shift of one to two investigation of coupled radiative and
hours. The heat storage capacity of the unoccupied conductive transient heat tranafer in fibroue
test buildings was less than predicted by the TFM insulations. Ph.D. dieaertation. Univ. of
model. Accuracy of the transfer function Missiesippi.
coefficients to model the roof-ceiling combination
was questionable due to the small attic air space McQuieton. F.C. 1986. A study and review of
which was not accurately described in the ASRRAE existing data to develop a standard methology
table of coefficients. for residential heating and cooling load
calculations. ASHRAE Transactione.
A parametric study waa done on the influence of
thermostat setpoint on room heat extraction. The
setpoint value seema to have a greater effect than
predicted by the model. The model predicted an 8%
increase in peak heat extraction upon going from a
thermoetat eetpoint of 78'~ to 74'~. The measured
increase was 47%. The predicted increase in total
heat energy for the thermostat setpoint variation
was 18%. The measured increase was 49%.

The ASHRAE model predicted an 8% energy savings

for the double pane windows over single pane
windows. The measured savings were 12% for east

Proceedings of the Fifth Symposium on Improving Building Systems in Hot and Humid Climates, Houston, TX, September 12-14, 1988