BUILDING INTEGRATED PHOTOVOLTAICS IN A THERMAL BUILDINGSIMULATION TOOL

Senior Researcher, MSc, Kim B. WittchenDanish Building and Urban ResearchHørsholm - DK-2970 – DENMARK

ABSTRACTA simple approach for calculation of the electricalyield from building integrated PV (photovoltaic)systems has been implemented in BSim2002 - aprogram package for thermal simulation of energyand indoor climate conditions in buildings. Themodule – SimPv - takes advantage of the buildingmodel geometry and solar distribution routines ofBSim to simulate solar irradiation on a face fittedwith PV. The routines also handle shades from localand distant obstacles.

The electric yield is calculated in SimPv from a verylimited number of data on the PV system. Output issummarised from each individual face containingareas with PV.

SimPv is intended as a simple tool to help the engi-neer or architect at the early stage of the buildingdesign to evaluate the potential for implementing PVin the building envelope, allowing detailed analysesof the shading problem at any time of the year. Theinfluence from shadings can be visualised dynamic-ally and the impact on the electrical yield calculatedfor any period of the year. Thus, SimPv is not adetailed PV design tool, but a decision making de-sign phase tool.

Results from simulations with SimPv have beencompared with results from the PvSyst program.This paper discusses the calculation approach usedin SimPv and results from an inter- model compari-son exercise.

INTRODUCING BSIMBSim (Wittchen, Johnsen & Grau, 2000-2003) is acomputational design tool for analysis of indoorclimate, energy consumption, moisture and daylightperformance of buildings. BSim integrates differentcomputer models that make it possible to carry out acomplete thermal, moisture and daylight analysis ofa building. The analyses are made, considering typi-cal indoor use of the building, the influences fromthe outdoor environment, and various types of typi-

cal equipment, internal loads and controls for heat-ing, cooling and ventilation.

The core of the system is a common building datamodel shared by the design tools, and a commondatabase with typical building materials, construc-tions, windows and doors. Figure 1 shows a typicalscreen picture from BSim.

Figure 1. Screen picture from BSim. At the right,four views show the model graphically and at the lefta hierarchic tree structure shows the individualobjects of the model, e.g. the construction Concre-teFloor.

A more general description of the BSim2002 pack-age is given in (Grau, Wittchen & Grau, 2003).

PV SIMULATION APPROACHA very important problem to analyse when thebuilding designer considers integrating PV systemsin the building envelope, is the shading issue. Evenlimited partial shading can cause the electrical pro-duction from the system to approach zero. The re-duction due to shading is seldom proportional to theshaded area, e.g. if no special arrangement is made,shading of one solar cell in a 36-cell array, can re-duce the collective production from that array to 10% or less (Hagemann, 2002) & (Sick & Erge, 1996).

Eighth International IBPSA Conference Eindhoven, Netherlands

August 11-14, 2003

- 1409 -- 1417 -

In SimPV an estimate of the yield from a buildingintegrated PV system is calculated for a given loca-tion on the building model. Shadows striking thesolar cells, which will reduce the yield to almostzero, are taken into account. The yield is calculatedas:

pvpvpv

shddirdiftpv FAf

AA

1IIQ ⋅⋅⋅−⋅+⋅= ))((ε

where:

εt is the total efficiency (solar cells, converter,cable-losses etc.) in the solar cell system. Thesystem efficiency is defined in the materials partof the program database as shown in figure 2.

Idif is the diffuse solar power on the PV-array.

Idir is the direct solar power on the PV-array.

f is a factor indicating if a part of the area isshaded.f = 1 if Ashd ≤ 0;f = ShdEff if Ashd > 0 and PropShadRed = 0;f = 1 - Fpv if Ashd > 0 and PropShadRed = 1.

Apv is the total area of the solar cell system.

Ashd is the shaded area of the PV system.

Fpv gives the active fraction of the total PV area Apv.

ShdEff is a user-defined efficiency for the shadedarea of a PV system. In most PV systems, the effi-ciency will be in the magnitude of 10 %, but use ofamorphous solar cells or special attention wiring theindividual cells in the panels according to localshading patterns, can increase the number.

PropShadRed is a user given property flag indicatingthat the actual PV-array has proportional yield re-duction when hit by partial shading. Most PV-arrayswill have a dramatically reduced electrical output ifeven a small part of the active area is shaded. It isthough possible, by means of wiring or use of amor-phous solar cells to have a reduction proportional tothe shaded area.

Analogue to this, the nominal yield can be calculatedas if no shading hits the area as:

pvpvdirdiftpv FAIIQ ⋅⋅+⋅= )(ε

This result can be used to calculate the "performanceratio" which express how large a yield is possiblefrom an area compared to the actual yield taking intoaccount losses from shadings.

Figure 2. System definitions in the materials data-base.

Results from SimPv are given as monthly summa-tions of the electrical production from each surfacewith integrated PV. Calculations are though madeon time step basis (maximum ½ hour) and resultsare thus calculated with this resolution to take intoaccount the varying shading over the day. The re-sults calculated on hourly basis is saved in a resultfile and can be merged with the overall result file ofBSim. It is thus possible to compare the hourly elec-trical output from a PV system with other electricalrequirements elsewhere in the thermal buildingmodel, e.g. consumption of electricity for the enginesin the mechanical ventilation system.

Figure 3. User interface for SimPv simulation re-sults. The No shading reduction flag makes it possi-ble to turn shading on and off and easily calculatethe performance ratio.

Evaluation of the possible impact from shadows onthe electrical yield is essential to select suitablebuilding envelope areas for integration of PV. Theuser interface offers a flag for performing a simula-

- 1410 -- 1418 -

tion of the electrical production without taking intoaccount the effect from shading. This is helpful if theuser wants to find locations for PV or to analyse theinfluence of shadowing or the performance.

The monthly results are interesting for estimates ofthe total electrical production from a PV system, butnot for analyses of the daily variations. It is thusplanned to incorporate the results from SimPv in theresult analyses of tsbi5 (the thermal simulation toolof BSim2002), to be able to compare the PV-outputwith, for instance, the electricity need of the ventila-tion system hour-by-hour. This will make it possibleto make an approximate sizing of the PV-system thatnever creates more electricity than needed inside thebuilding at any time. Such analyses are interestingwhen the price of buying and selling electricity tothe public grid differs.

SHADING CALCULATIONSShadows on a PV area originates from varioussources as the building itself, neighbouring buildingsor obstacles and mounting system for PV-panels.The shadows from the building and distant obstaclesare dealt with, using the defined geometry in thebuilding model. Shadows from the mounting systemdo however have to be defined individually for eachPV-area.

Figure 4. Example sketch of mounting system forPV-panels and indication of shading edges.

When dealing with local shadows from the mountingsystem, the program assumes that it is symmetricalalong all edges of the PV-panel. Local shadows arethe defined by a frame distance and a recess. Theframe distance is the distance from the solar cellsclosest to the edge and to the shading edge (point aor b in figure 4). The recess is the distance perpen-dicular to the solar cells to the same shading point.

Besides local shadows, more distant geometricalobjects as overhangs and side fins, or other shadingobjects, e.g. surrounding buildings or trees, mayobscure PV-panels. The shading objects may betransparent in any degree, from non-transparent tocompletely transparent.

Shading calculations in the SimPv module are madeusing another build-in application, XSun. This isoriginally an implementation of polygon clipping fordetermination of shades on windows located in aplane surface, e.g. the facade of a building (Grau &Johnsen, 1995). PV-panels and the shading objectsare approximated by convex or concave polygons.Using polygon clipping, the fully or partly exposedregions of the PV-panels are calculated as regions ofpolygons, each polygon receiving a fraction of fullsolar radiation, dependent on the factor of transpar-ency of the shading objects. For each PV-panel, anoverall reduction factor, fsun is calculated as

∑=

=n

1kkk

PVsun fA

A1f

where:

fsun is the fraction of the direct solar radiationreaching the PV-panel, compared to the directradiation that would reach the PV-panels with-out any shading.

APV is the PV-panel area.

n the number of polygons in the calculated radi-ated region (the PV-panel).

Ak is the area of polygon k.

fk is the factor of transparency for polygon k.

XSun can be used in two different ways in BSim:

(1) For visual, animated analysis of shading condi-tions at any time of the year.

Figure 5. XSun visualisation of shadows from aflagpole on a PV panel. Direct solar radiationthrough two roof windows is at the same time shownas yellow sunspots on the floor and wall of thebuilding.

- 1411 -- 1419 -

(2) To be called as a calculation routine to generatea value of the reduction factor fsun for each PV-panel at any given time of the year. This routineis used both in SimPv and in the thermal simu-lation tool, tsbi5, which is part of the BSim2002program package.

In SimPv the fsun factor is used to calculate the Ashd

area as:

)( sunPVshd f1AA −=

SOLAR INCIDENCECalculation of incident solar radiation on PV-panelsand windows and includes the following tasks:

First, the position of the sun is calculated for a cer-tain day of the year at a given time. According to thesun's position, the shading objects are projected ontothe surface with the actual surface (PV-panel orwindow). As the sun is at an infinite distance, theprojection is a parallel projection in the direction ofthe sun. The projection gives the patch of the shad-ows on the surface.

Each sunlit surface is represented as a polygon re-ceiving full radiation. By clipping this polygonagainst all the projected shadow patches in turn, theresult becomes a region of polygons within the sur-face receiving full radiation, or a fraction of radia-tion dependent on the degree of transparency of theshading objects. For each surface, the overall reduc-tion factor, fsun is calculated as described earlier.

MODEL COMPARISONAn example was set up to compare the results fromthe SimPv module with the results from PvSyst. Themodel consists of three surfaces equipped with anarray of solar cells, a flat roof, a 45° sloping surfaceand a vertical face. Each of the faces has a PV sys-tem of 20 panels, measuring 0,66 • 1,32 m2 (figure6). The model was rotated for comparison of resultson different orientations (facade and tilted roof).

To make the comparison possible, the overall systemperformance used as input parameter in SimPvoriginates from a detailed simulation performed inPvSyst.

Results from the simplified calculation approachused in SimPv and BSim2002 have been comparedto the detailed simulations performed by PvSyst.

Annual differences in the calculated electrical pro-duction were found up to 15 % on selected surfaces.However the average discrepancy between the twotools was found as low as 2 %. A comparison of themonthly electrical production from a south facing45°-tilted surface is shown in figure 7. There is an

almost perfect match during the summer months anda somewhat larger difference during the months withless solar incidence and lower solar altitude.

Figure 6. Model used for inter-model-comparisonwith 20 solar panels on horizontal roof, 45° tiltedsurface and facade.

Roof 45°

0

50

100

150

200

250

1 2 3 4 5 6 7 8 9 10 11 12Month

Ele

ctric

al p

rodu

ctio

n [k

Wh]

BSim PvSyst

Figure 7. Comparison of calculated electrical pro-duction from PV-systems by BSim2002 and PvSyst.

LIMITATIONSPV system efficiency used in these simplified calcu-lations is a fixed number e.g. not changing due tothe cell temperature. This is an acceptable assump-tion when evaluating the possible electrical yield onannual basis, but not sufficiently accurate for PVsystem design. Even though information on systemefficiencies for some PV systems do exist in theBSim program database, the user must have somebasic knowledge of building integrated PV systemsto use and judge the results from the tool.

- 1412 -- 1420 -

The program are specifically designed to focus onthe shading problem of any building integrated PV-installation. SimPv is not a PV design tool and doesnot take into account losses if the geometry of thebuilding requires special wiring or other specialfeatures of the electrical part of the installation. It isthus not possible to analyse the influence of a poorsystem design, e.g. high quality solar cells connectedthrough long, thin wiring to an oversized AC/DCconverter.

For the time being there is no thermal connectionbetween the building model and the PV system inBSim. There is for instance not transfer of heat fromsolar cells integrated in windowpanes to the roominside the window. To evaluate this energy flow it isnecessary model the window using the optical andthermal properties of a window with integrated solarcells.

CONCLUSIONSA simple model has been developed for early designanalyses of the possible electrical yield from abuilding integrated PV-system. In the early designphase SimPv is a valuable tool for selecting appro-priate locations for integration of PV in the buildingskin.

The model has been implemented in a simple designtool, SimPv, which is integrated into a thermalbuilding simulation package, BSim2002. The ad-vantage is the reuse of the building model and thecalculation of shadows from the building itself andthe neighbouring buildings.

The simplified routines in SimPv have shown satis-factory accuracy, when compared to results from theadvanced PV simulation tool PvSyst. An averagediscrepancy of 2 % was found when comparing theelectrical output from PV systems integrated in sur-faces with different orientations and inclinations.

Due to the simplified routines in SimPv the toolmust not be considered a design tool, but only arough calculation of the yield to help the buildingowner decide to have building integrated solar cellsor not. When the decision is made, an advanced PVdesign tool must come into action.

ACKNOWLEDGEMENTSThe Danish Energy Authority via the EFP EnergyResearch Programme has supported the developmentof SimPv solar cell simulation module forBSim2002.

The work was conducted in collaboration betweenDanish Building and Urban Research(www.dbur.dk), Esbensen Consulting Engineers Ltd.(www.esbensen.dk) and the Danish TechnologicalInstitute (www.teknologisk.dk).

REFERENCESGautam NK & Kaushika ND (2002). „An efficientalgorithm to simulate the electrical performance ofsolar photovoltaic arrays” Energy, vol. 27, p 347-361.

Grau K, Wittchen KB & Grau C (2003) „Visualisa-tion of building models” In proceedings of IBPSABS2003, Eindhoven, The Netherlands.

Grau K & Johnsen K (1995) „General ShadingModel for Solar Building Design” In ASHRAEMeeting. San Diego, USA. ASHRAE Technical DataBulletin, 11(3), 100-112.

Hagemann I B (2002)„Gebäudeintegrierte Photo-voltaic. Architektonische Integration der Photovol-taik in die Gebäuehülle” Rudolf Müller GmbH &Co. KG, Köln, Gremany.

Sick F & Erge T (ed.) (1996) „Photovoltaics inbuildings. A Design Handbook for Architects andEngineers” IEA, James & James Ltd., London, UK.

Wittchen KB Johnsen K and Grau K (2000-2003)„BSim – User’s Guide“, Danish Building and UrbanResearch, Hørsholm, Denmark.

INTERNET SOURCESDanish Building and Urban Research: www.dbur.dk

Esbensen Consulting Engineers: www.esbensen.dk

Danish Technological Institute: www.teknologisk.dk

BSim2002: www.bsim.dk

PvSyst: www.pvsyst.com

- 1413 -- 1421 -

- 1414 -- 1422 -