SERI/TP-254-1953UC Category: 59cDE83009402

Measured Versus PredictedPerformance of the SERlTest House: A VaUdation Study

R.JudkoffD. WortmanJ. Burch

May 1983

To be presented at theNational Heat Transfer ConferenceSeattle, WashingtonJuly 1983

Prepared under Task No. 1053.00WPA No. 304

Solar Energy Research InstituteA Div isi on of Midwest Resea rch Institute

1617 Cole BoulevardGolden, Colorado 80401

Prepared for the

U.S. Department of EnergyContract No. EG-77-C-01-4042

Printed in the United States 'of AmericaAvailable from:

National Technical Information ServiceU.S. Department of Commerce

5285 Port Royal RoadSpringfield, VA 22161

Price:Microfiche $4.50

Printed Copy $7.00

NOTICE

This report was prepared as an account of work sponsored by the United StatesGovernment. Neither the United States nor the United States Department of Energy,nor any of their employees, nor any of their contractors, subcontractors, or theiremployees, makes any warranty, express or implied, or assumes any legal liabilityor responsibility for the accuracy, completeness or usefulness of any information,apparatus, product or process disclosed, or represents that its use would notinfringe privately owned rights.

MEASURED VERSUS PREDICTED PERFORMANCE OF

THE SERI TEST HOUSE: A VALIDATION STUDY

R. JUdkoff

D. Wortman

J. Burch

ABSTRACT

For the past several years the United States Department of Energy (DOE)

Passive and Hybrid Solar Division has sponsored work to improve the relia-

bility of computerized building energy analysis simulations. Under the

auspices of what has come to be called the Class A Monitoring and Validation

program, the Solar Energy Research Institute (SERI) has engaged in several

areas of research that includes: (1) developing a validation methodology;

(2) developing a performance monitoring methodology designed to meet the

specific data needs for validating analysis/design tools; (3) constructing and

monitoring a lOOO-ft 2, multizone, skin-load-dominated test building;

(4) constructing and monitoring a two-zone test ceLl ; and (5) making sample

validation studies using the DOE-2.1, BLAST-3.0, and SERIRES-l.O computer

programs. This paper reports the results obtained in comparing the measured

thermal performance of the building to the performance calculated by the

building energy analysis simulations. It also describes the validation

methodology and the Class A data acquisition capabilities at SERI.

The Class A, B, and C perf.ormance moni-toring programs were initiated· in 1979because of the demand from researchers andindustry for passive and hybrid building per-formance d~ta at various levels ofdetail (J.). Class A monitoring providesdetailed--data (approximately 200 channels perbuilding) under controlled conditions at afew sites for algorithm development and vali-dation of building energy analysis simulationprograms. Class B provides limited detail(about 20 channels per building) in approx-imately 100-200 occupied buildings for fieldtesting passive and hybrid designs andstatistically evaluating simplIfied designtools. Class C provides utility bill dataand a survey of occupant reactions.

SERl I S involvement in validating build-ing energy analysis simulations (BEAS)tesulted fro~ two comparative studies con-ducted in 1980 and 1981 (2,3). These studiesshowed significant disagr;e;ent between fourstate-of-the-art simulations: DOE-2.1,BLAST-3.0, DEROB-4.0, and SUNCAT-2.4 whengi~en equivalent input for a simple, direct-gain building with a high and low mass par~metric option (Figure 1). The studies alsoindicated the need for high quality,cont ro Ll.ed validation data and a validationmethodology. SERl assumed responsibility fordefining the data acquisition criteria for

Sol'ir Energy Reseat"ch Institute;1617 Cole Blvd.; Golden, CO 80401

1

SERI/TP-1953

validation, developing a validation meth-odology, and constructing a Class A data col-lection facility. Class A facilities werealso constructed at the National Bureau ofStandards (NBS) and several universitie~.

VALlDAnOlf METHODOLOGY

The overall validation methodology usesthree different kinds of tests (4): (1) ana-lytical verification (S)t (2) empirical val-idation, and (3) code=to-code comparisons.The advantages and disadvantages of thesethree techniques are shoen in Table 1.

Each comparison between measured andcalculated performance represents a singledata point in an immense N-dimensional para-meter space. We are constrained to estab-lishing very few data points within thisspace, yet, we must somehow be assured thatthe results at these points are not coin-cidental and do represent the validity of thesimulation elsewhere in the parameterspace. The analytical and comparative tech-niques minimize the uncertainty of the extra-polations we must make around the limitednumber of Class A empirical data points it ispossible to sample. These extrapolations areclassified in Table 2.

Figure 2 shows the process by which weuse the analytical empirical and comparativetechniques together. The first step is ::0run the code against the analytical test

SERI/TP-1953

cases. This checks the numerical solution ofmajor heat transfer models in the code. If adiscrepancy occurs, the source of the dif-ference must be corrected before any furthervalidation is done.

1. Differences between the actual weathersurrounding the building and the statis-tical weather input used with BEAS.

2. Differences between the actual effect ofoccupant behavior and those effectsassumed by the user.

3. User error in deriving building inputfiles.

DATA COLLECTION METHODOLOGY

External Error Types

The third step involves checking thecode against several prevalidated buildingen~rgy analysis simulations (BEAS) in anumber of comparative studies. If the codepasses all three steps, it can be consideredvalidated for the range of climates andbuilding types represented by thesestudies. The prevalidated BEAS will havesuccessfully passed steps one and two andwill have shown substantial agreement for allthe comparative study cases. These compar-ative study cases will use Class B data wherepossible. SERI is currently prevalidatingthe DOE, BLAST, and SERIRES programs as partof its Class A empirical validation project.

There are many levels of validationdepending on the degree of control exercisedover the possible sources of error in a sim-ulation. These error sources consist ofseven types divided into two groups:

LomassCool

LomassHeat

HlmassCool

HlmassHeat

Figure 1. Phase II Comparative Study:Albuquerque

c=J SUNCAT·2.4I

tzzzi DOE·2.1tt'SC:Z£j BLAST-3.0

tz::LZ2Zl DEROB·4 17

r-t

II ~ r1

40

30

20

10

o

70

60

50

The next step is to run the code againstClass A empirical validation data and to cor-rect discrepancies. A quantified definitionof these discrepancies has been proposed byLos Alamos National Laboratory (LANL) (6).SERI and several other Class A sites ~recurrently collecting these data.

Load (10 x 6 Btu. y(1)90

80

Table 1. Validation Techniques

Technique Advantages Disadvantages

ComparativeRelative te~t of modeland solution process

No input uncertaintyAny level of complexityInexpensiveQuick, many comparisons

possible

No truth standard

AnalyticalTest of numericalsolution

No input uncertaintyExact truth standard given

the simplicity of the modelInexpensive

Nv test of modelLimited to cases for

which analytical solu-tions can be derived

EillpiricalTest of model andsolution process

Approximate truth standardwithin accuracy of dataacquisition system

Any level of complexity

Measurement involves somedegree of inputuncertainty

Detailed measurements ofhigh quality are expen-sive and time consuming

A limited number of datasites are economicallypractical

2

SERI/TP-1953

4. Differences between the actual thermaland physical properties of the buildingand those input by the user (generallyfrom ASHRAE handbook values).

Internal Error Types

derived independently by several experiencedusers and then cross-checked until collectiveagreement is reached to control error 3.Thermophysical properties are directlymeasured through destructive and nondestruc-tive testing to control error 4. Once all

Analytical Verification

Figure 2. Validation Method

RepairCode

RepairCode

IdentifySource

Empirical Validation inat Least Two VeryDifferent Climates

Code Validated Within Rangeof Cases Defined byComparative Studies

Comparative Study Cases AgainstSeveral "Validated Codes" and

Class B Where Possible

5. Differences between the actual thermaltransfer mechanisms taking place in thereal building and the simplified modelof those mechanisms in the simulation.

6. Errors or inaccuracies in the numericalsolution of the models.

7. Coding errors.

At the most basic level, the actual long-term energy usage of a building is comparedto that calculated by the computer programwith no attempt to eliminate sources of dis-crepancy. This level is similar to how theBEAS would actually be used in practice and,therefore, is favored by many representativesof the building industry. However, it isdifficult to Ln t e r pr et; the results of thiskind of validation exercise because all pos-sible error sources are simultaneouslyoperative. Even if good agreement isobtained between measured and calculated per-formance, the poss Lb t Lf.t y of offsettingerrors prevents dra~ing conclusions about theaccuracy of the method of calculation. Moreinformative levels of validation are achievedby controlling or eliminating various com-binations of error types. At the mostdetailed level, all known sources of errorare controlled to identify and quantifyunknown error sources. This is the approachtaken in Class A data acquisition forvalidation.

Detailed meteorological and microclimate'lleasurements are taken at the s I te to elim-inate error 1. The buildings are kept unoc-cupied to eliminate error 2. Input files are

Table 2. Types of Ex~rapolation

Obtainable Data Points Extrapolation

A few climates ~~ny climates

Short-term (e.g., monthly) total energy usage Long-term (e.g., yearly) total energy usage

Short-term (hourly) temperatures and/or flux Long-term (yearly) total energy usage

A few buildings representing a few sets ofvariable mixes

Many bUildings representipg many sets ofvariable mixes

Small-scale, simple test cells and buildings Large-scale complex buildings

3

external error types have been controlled, itis possible to isolate internal errors.

To validate the key thermodynamicmodels, which comprise errors 5 and 6, twodifferent kinds of data are needed. First,data must be taken to define the overallbuilding energy performance. This overallsystem level includes zone air and globe tem-perature data and (if temperature controlled)auxiliary energy measurements. These datasummarize building energy performance.Second, data must be taken at the energytransport mechanism level. Energy transportmechanisms are summarized in Table 3. Wherethis is not possible because of state-of-the-art measurement limitations or where noacceptable models exist for a mechanism, themechanism may be physically suppressed as wasdone in our test cell for ground coupling.This two-level approach allows us to identifythose mechanism inaccuracies that lead tosystem level errors.

To ensure that all major transport mech-anisms are monitored, we provide for internalconsistency checks. Failure to achieveclosure on the measured heat balance Qi n= Qout + Qstored can be attributed only tof au Lty data or to important mechanisms notrepresented in the measurements.

Table 3. Energy Transport MechaniSIUI

CONDUCTION: Measure Temperatures andConduction Fluxes

Structural elementsSkin and interzonal opaque wallsGlazings

Ground coupling

CONVECTION: Tracer Gas, Special ExperimentsFilm coefficients

Inside surfaces: free convectionOutside surfaces: forced convection

Air XotionInfiltrationZone to zone

Natural convection through doorwaysNatural convection through cracks

Stratification

~\DIATION: Measure Radiant FluxesInfrared surface coupling

Internal surfacesExternal surfaces (sky temperature)

SolarExternal absorptionGlazing transmission and absorptionInternal absorption

4

SERI/TP-1953

SERI CLASS A DATA FACILITY

The SERI Class A validation facilit2consists of two structures: a 1000-ftresidence and a 120-ft2 two-zone test cell.These two structures are instrumented withapproximately 250 sensors each to achieve thedegree of experimental control previouslydiscussed. The sensors include type Jthermocouples, heat flux transducers, Halleffect watt-hour meters, Kip & Zonen andEppley pyranometers, and an Eppleypyrheliometer. ',Jind speed, direction, andhumidity are also measured.

Details of the house and the test cellare provided in two handbooks (7,8).Figure 3 shows the plan and south elevationof the house. The cell and the house weredesigned to complement each other and otherClass A facilities. The approach in the cellwas to suppress all difficult mechanisms.These included ground coupling, interzonaland cavity convection, stratification, andinfiltration. The house was operated in amore realistic fashion, and attempts weremade to measure such difficult transportpaths as ground coupling via a crawlspace andmultizone infiltration. The crawlspace con-figuration was chosen to complement the floorslab configuration at NBS. For lllUltizoneinfiltration, we initiated a project todevelop an apparatus capable of continuousmultizone infiltration monitoring (9). Aprototype of this apparatus has been ~ollecting data since April 1982. Table 4 shows themeasurement approach taken for variousmechanisms in the house and the cell.

We moni tored the house and cell througha number of configurational changes in thewinter and spring of 1982. In the case ofthe house, this consisted of several con-servation and solar retrofits including:(1) insulation blown into walls and attic,(2) batt insulation on foundation walls incraWlspace. (3) storm windows, (4) caulkingand weatherstripping, (5) orientation oflargest glazed areas to south, and(6) addition of rhe rmaI mass to south-fadngrooms. These retrofits reduced the effectivecrack area as measured by 1 blower door fromapproximately 200 to 50 in. (see Figure 4).

We will continue to collect and analyzedata from the house and cell. Completeresults from the fiscal year 1982 work are inWortman et a.l , (lQ).

SERI/TP-1953

oKitchen Utility Bdrm #3 Bdrm #2

Dining Am

Figure 3&.

Living Am

Validation Test Residence:

44'4"

Bdrm #1

Floor Plan

~I"..,

Partition io. -- --0_

FinishedI r II I

Ceiling25 ft2I II (:,

SERI/TP-1953

Table 4. Measurellent Approaches

Measurement ApproachMeehan!..

Wall Conduction

Code ApproachTest Cell House

Basic assumption

Wall conductivities

Mass Storage

Ground coupling

Boundary Conditions

One-dimensional flow

Inputs t constanta

Not directly available; can becomputed froll temperature

One-dlm.enalonal flow to groundtemperature t neglecting edgeeffects

Insulate edges where possible D;ne-dillenlllional flow assumedto enaure one-dimenalonalflow

Measurements to directly Same as celldetermine Uwa l l' Ulayer

Compute from temperature data, Compute from temperature data I-10 rakes in mass -2 locations per zone

Eliminate entirely. Study in detail , flux and tem-perature at -10 locatioDS

Varied approaches, froll con- Meaautestant (SUNCAT) or wind-driven hconvonly or wind and sky infrared

Interior surfaces

Exterior surfaces

Zone-Related Effects

Zone mixing

Interzonal adveeetceand conduc tion

Occupancy effects

Furnishings

Internal hWllidity

System Effects

Heating systems

Varied approaches froll heat• ceesc , (e.g_, SUNCAT) toexplicl t IR + convec tioncorrelations (e.g_, DEROS)

Alway. isothemal

Uncertain, approximate algo-rithms for advection; wallconduc tion included

Schedules input, _joruncertainty

Neglec ted or approximate

r.tent heat usually included

Set point.s; ramp

Measure he a nv separately;define effective interiortemperature I snc! computeinfrared flux (QIR)

·skyQla • Taround; deduceon average

Destratify to force zone tobe 1&othe1'1l&1

Measure conduction directly;advection minimized bycareful caulking

None

None

Not measured

Measure ~eater with elec-trical inputs of k.nownefficiency, T} - 1. 0; smalldeadband

Measure films only on glazing,same techniques ae for cell

Same a8 cell

"Oestratify continually(FY 1982); Itudy destrat-if1cation in FY 1983

Measure conduction; closeddoors between cells(FY 1982). Study naturaladvec tion in FY 1983

None

Unfurnished

Not lIIeasured

Electric heacer a , to becomputer-controlled fornight setback at night

Night ventilation

Solsr Radiation

Sciledule or constraint .for MoDeVnight; volume flow V 1& illput

Measure Vance by tracerdecay

Descriptive inputs

Tilted surface irradiance

Need I beam• Ga

Various models, mostly iso-tropic or anisotropic

Measure Ibeaml Cia directly Same as cell

Exterior; measure south irra- External: same as celldiance broken into south skyand ground diffuse components

6

Internal: floor, north wall,east wall

Internal: measure verticaltransmi tted, each orienta-tion; and floor and mid-wallirradiance in living room

SERI/TP-1953

Table 4. Measurement Approaches (Concluded)

Mechanism Code ApproachTe!lt Cell

Measurement Approach

House

EDT!raIMent-Related Processes(Continued)

Gla>:ing transmisatons Besm transmission calculatedfrom input index of refrac-tion and extinction coeffi-cient, diffuse transmission~ Some input or defaultconstant

Measure beam and diffuset ransmiss10n direct ly;extract best fit indexof refraction and extinc-tion coefficient from data.Done only occasionally.

Same as cell. for the sout"g l ass only, before and alterstorm giazinl\'s

Ground reflections Input "GR Meat'H~ (lGR continuously;once

(lGR is sOITIe 88 for cell, usecell data

Solar glazing back. 108ses Calculatable frOll variousmodels, or input constant(SUNCAT)

Mea8ure cell albedo directlyfot' clear. clOUdy cond f tions

No albedo measur"",ents

Input ""locity, direction;8SSU1lll! sa_ value for filmcalculation and infiltrationl1IOdel, very uncertain

Measure at two heights at-100 yards £rolll cell;uncertain microscaleproblelll8

- AY::a~:l~~y~~:tion tn- Reduce effects by tigltt

construction

Same as cell

Other: humidity, pressure Inputs used for at r heatcapac1 ty, latent losds

Adequate di rect mea8ure SAme as cell

Precipitation No impact on thermal models Field site obaervation, plus(lGR data effects

Sa.... aa cell

7. Window Conductance: This case was notrun because measured window conductanceswere the same as those given by theASHRAE Handbook of Fundamentals.

4. Ground Albedo: Same as base case exceptmeasured ground albedo was used in thecalculation of radiation incident uponglazed surfaces.

5. Set Point: Same as base case except acorrection was made to the thermostatset point based on the average tem-perature of air in the zone when theheater actually turned on.

energymeasured temperature andperformance of the building.

Results

Figure 6 shows the root mean square(RMS) difference between measured and pre-dicted temperatures in zone 2 of the housefor all 9 cases. Zone 2 is the southernliving room and has a massive floorsurface. In general the results from zone 2are typical of the results from the wholebuilding. Case 1 has RMS errors of between

Figure 5 shows the whole-house heatingload (in kWh) during the week of April 20-26.1982. The loads predicted by the DOE-2.1A,BLAST-3.0, and SERIRES computer programs areshown along with the measured load forcases 1 through 9. In case 1, where handbookinput values were used, the code predictionswere high by 59%-66% compared with measuredloads. In case 5, where the correction wasmade for the actual thermostat set point, thecode predictions were high by 47%-52%. Incase 9, where all known measured input valueswere used, the code predictions were low by10%-17%. In general, the predictions weremost accurate for case 9.

not runspect r-rmwas notassumed

Same as baseand ceiling

6. Wall and Roof Conductance:case except measured wallconductances were used.

8. Absorptivity: This case wasbecause the measured solarabsorptivity on opaque surfacessignificantly different thanvalues.

9. Measured: All of the measured values incases 2 through 6 were used. This caserepresents the highest degree of controlover external error sources and shouldpresumably yield results closest to the

7

SERI/TP-1953

200 g

IA-B Flue and Return Register Sealed g

IB-C Attic and Walls Insulated

s- C-D External Window Frames Caulkedg 150 D-E Internal Storm Windows Installed'" E-F Bricks Installed and Baseboard Caulkede

trr~:~Bath and Hall Baseboards Caulked-c

Q) Crawl Space Hatch ClosedCl

'" 100 Bedroom Floor Temporarily Sealed""'"Q)~Q)

.?:(3~ 50Ui

0A B C 0 E F G H

Figure 4. Blower-Door results for SKKI Retrofit House

I2 345 I 7 a

CASE NUMBERa = BLAST3.0 0 = SER IRES•=DOE2.1A + =MEASURED

t

Figure 5. Weekly Whole-House Beating Loads

.90 and '1. zOe. Case 5 has Ro.'IS errors of from•60 to .aoe. Case 9 has the largest RMSerrors of from .40 to 1.60C.

Figure 7 shows the zone Z measured peakheating load and the peak heating loads pre-dicted by the three computer codes in cases 1through 9. The case 1 predictions of peakload are high by 36%-49%. The case 5 pre-dictions are high by 31% to 43%. The case 9

8

predictions are the most accurate and fallwithin ±5% of the measured peak load •

Figure 8 shows the peak load for thewhole house. The pattern is similar to thatobserved for zone Z with case 1 predictionsbeing least accurate and case 9 predictionsbeing most accurate.

Figure 9 shows the hourly temperatureprofile predicted by the DOE-Z.1A code in

SERI/TP-1953

.+--~---4---+--+--i---t---t---+--+---t,...~:II'-...+----+--+--4--+--;--+---11--+--1---1a-eo..J

"+--+---+---+--......,---t---+---+---+---+--"'1~

~

o+---+--+--+--+---1......-+--I--+--~-_1234 5 S 7 S

CASE NUMBERa =BLAST3.0 0 =SER IRES•=DOE2.1A + =MEASURED

Figure 8. Whole-Bouse Peak Load

:'

/\ f~~ ,r..1\

.'~",: "

fr\\ i \ :~ \ r-;...t:.', W rJ ~ ~ !'Y' ~.i'J'-' ' \. "

r- /\ /\ j\ J\ /\ .r-;r~

I~.~'

I,..IS 72 QII 120

TIME (HOURS).....P.Qi6,IA.... MEASURED_.llil.~.IA-l,!EASUREIL

A A :\ ~ .A;-",;:"" .:f hJ 'Ry' "t::1 \t;f \~i'J'-' ,........, "-,'

r- /\ /\' )\ /\ A/ ..../ / f-../

oo'".. 72 QII 120TIME (HOURS)

.....D.Q~JA.... MEASURED_llil.~.II1::l!{EASUR~IL

Figure 9. DOE-2.1A va. Measured Temperatures,Case I: Zone 1

Figure 10. DOE-2.IA va. Measured Temper-atures, Case 9: Zone I

relation to the measured temperature profilefor case I, zone 1. Zone 1 was primarily afree-floating zone during this time periodbecause temperatures remained above thethermostat set point from hour 36 tohour 168. Figure 9 shows that the predictedtemperature tends to overshoot measuredtemperature during the day and undershootmeasured temperature at night.

Figure 10 shows the same information forcase 9 as was shown for case 1 in Figure 9.In case 9 we see that predicted temperatures

overshoot by even more during the day than incase 1 and that they undershoot by less atnight than in case 1.

INrERPRETATION OF DATA

There is some apparent inconsistency inthe data with respect to the presumption thatcase 9 would always yield the most accuratepredictions. The most obvious is seen inFigure 6 where case 1 and case 5 had smallerRMS temperature errors than case 9. This

10

SERI/TP-1953

trend is the reverse of that seen inFigure 5. where. as expected, the mostaccurate load prediction was obtained incase 9. The most likely hypothesis now isthat (1) the amount of solar energy absorbedin the building is being overpredicted in allcases. and (2) the conductive losses throughwalls and roof are being overpredicted incases 1 and 5.

of accuracy in the case 9 peakpredictions is also consistent withexplanation. The code predictions wereaccurate when stored solar energy wasdepleted and envelope conduction wasdominant.

loadthismostmostmost

This conjecture is partially supportedby the large (approximately a factor of two)difference found between measured and assumedwall and roof resistances as shown inTable 5.

Table 5. Measured 'lS. Assumed Wall andCeiling Resistances

R-V~lue

W/m °c(Btu/h ft 2 OF)

CONCLUSIONS

This work is part of a multiyear,multilaboratory effort on the part of DOE toimprove calculational methods for buildingenergy analyses by collecting high qualitydetailed data and applying rigorousvalidation techniques. Although this work isfar from complete, several conclusions can bedrawn that should help guide futureactivities.

The smaller &~ temperature errors in cases 1and 5 could be explained, therefore, by theoffsetting effects of too high an envelopeconductance and the calculation of too I1Uchsolar radiation absorbed in the building.This explanation is consistent with thehourly temperature profiles seen in Figures 9and 10 where the case 1 predicted temperaturewas high in the day and low at night, whilethe case 9 predicted temperature was evenhigher during the day but not so low atnight. This also explains how the heatingloads in case 9 could be most accurate whilethe RMS temperature errors in case 9 weregreatest. The large RMS temperature errorswere caused primarily by the daytime over-prediction of temperature. The greateraccuracy in load prediction was still pos-sible because at night the performance of thebuilding was primarily governed by the con-ductive skin losses. The effect of the over-prediction of solar energy adsorbed resultedin the 10% to 17% underprediction of loads inFigure 1, case 9. Finally, the high degree

Average measured wallr-esistance

Assumed wall resistancef r on ASHRAE

Average measured ceilingresistance

Assumed ceiling resistancefrom ASHRAE

3.05(17.3)

1.56(8.83)

13 .19(75.03)

7.04(40.00)

• Input assumptions based on standardengineering references can cause pre-diction errors of approximately 60% evenwhen using measured meteorological data.

• Accurate temperature prediction does notguarantee accurate load prediction, nordoes it guarantee an accurate temper-ature prediction on the next buildingstudied. There is evidence of compen-sating errors giving a false sense ofconfidence. Any validation methodologymust account for the possibility ofhidden compensating errors.

• The heating load predictions for thethree codes for all cases were withinabout 7% of each other.

• Even when most input inaccuracies areeliminated using measured thermo-physical input data, prediction errorsranging from 10% to 17% have still beenfound. This can have a large impact onbuilding and HVAC system design options.

• A more detailed level of analysis andexperimentation will be necessary todetermine if these inaccuracies arecaused by unknown remaining external orinternal error sources. This additionalwork should include:

Corroborating the conductancesmeasured in the walls and ceilingwith an AS1M standard large sectionclamp-on guarded hot-box.

Installing a simpler window assemblyin the test house.

11

Developing a measurement technique todetermine the amount of solarradiation absorbed in the huilding.

Determining the sensitivity of outputaccuracy to isotropic versusanisotropic sky models.

• The methodological approach used in thiswork for skin load dominated buildingsshould be expanded to include the mech-anical systems in commercial buildings.

SERI/TP-1953

4. Judkoff, R. et al., Validation of Build-ing Energy Simulations, SERI/TR-254--1508, Solar Energy Research Institute,Golden, Colo. (forthcoming).

5. Wortman, D. et aI., The Implementationof an Analytical Verification Techniqueon Three Building Energy Analysis Codes:SUNCAT-2.4, DOE-2.1, and DEROB-III,SERI!TP-721-1008, Solar Energy ResearchInstitute, Golden, Colo. (1981).

2. Judkoff, R. et a1., A Comparative Studyof Four Building Energy Simulations:DOE, BLAST, SUNCAT-2.4, DEROB III,SERI!TP-721-837, Solar Energy ResearchInstitute, Golden, Colo. (1980).

1. Holtz, M., Program Area Plan for Per-formance Evaluation of Passive/HybridSolar Heating and Cooling Systems,SERI!PR-721-788, Solar Energy ResearchInstitute, Golden, Colo. (1980).

ACKNOWLEDGMENTS

This work was sponsoredDepartment of Energy, SystemsDevelopment Branch, Passive andHeat Technologies.

LITERATURE CITED

by the U.S.Research andHybrid Solar

6. Runn et al., Validation of Passive SolarAnalysis/Design Tools Using Class APerformance Evaluation Data, LA-UR-82--1732, Los Alamos National Laboratory,New Mexico (1982).

7. Burch, J. et al , , Site Handbook for theSERI Validation Test Residence, SolarEnergy Research Institute, Golden, Colo.(forthcoming).

8. Burch, J. et al., , Site Handbook for theSERI Two-Zone Validation Test Cell,Solar Energy Research Institute, Golden,Colo. (forthcoming).

9. Wortman, D. et a I , , A Multizone Infil-tration Monitoring System, SERI/TP-254-1638, Solar Energy Research Institute,Golden, Colo. (1982).

3. Judkoff, R. et al., A Comparative Studyof Four Building Energy Simulations,Phase II: DOE-2.1, BLAST-3.0, SUNCAT--2.4, and DEROB-4, SERI!TP 721-1326,Solar Energy Research Institute, Golden,Colo. (1981).

10. Wortman, D. et al., Empirical Validationof Building Energy Analysis Simulat ionsUsing Class A Data from the SERI Vali-dation Test House, SERI/TR 254-1840,Solar Energy Research Institute, Golden,Colo. (forthcoming).

12