8/18/2019 IBPSA2013

1/8

353

1

PROCEDURE FOR BUILDINGS’ ENERGY MODELING SUITED FORINTEGRATED CONTROL SIMULATION

ABSTRACT Numerical simulations are becoming more and morea key step in designing an integrated building +energy system. The use of a detailed numericalmodel and an accurate calculation algorithm permitsnot only to study the influence of design parameters, but also to evaluate the building energy demand andindoor comfort conditions.The computational effort in modelling the building behaviour might become a limit in terms ofsimulation time, in particular in the case of complexsystems. This is evident when an integrated controlof building and thermal energy system is performed, because of the need of a short time step (1 to 5minutes) for control purposes. In this sense, a clearedge between architectural and energy modelling hasto be drawn.In this paper, a methodology for simplifying a

detailed building model has been presented, bydefining phases, quantitative figures, limits anduncertainty of the results. Such procedure isreplicable and could be useful during any numericalmodelling process.With reference to a three-story apartment building,the influence of (1) the algorithm for short- and long-wave radiation, (2) building and surroundingcontext’s shadings, (3) different levels of geometricsurface information have been investigated. For each phase, building’s energy balance and simulationruntime has been reported. Additionally, thereliability and adherence of simulation outputs have been inspected by comparing the numerical model

response with a whole year of monitoring data.INTRODUCTIONBuilding models are more and more used forstudying the influence of design parameters and forthe evaluation of building energy demand and indoorcomfort conditions. Accurate building modelingmight require computational efforts duringsimulations. The developing of a good buildingmodel requires to focus on most important building’sfeatures (weather file, building size, energy loads…),

to minimize the number of thermal zones, to properlycharacterize HVAC and controls (IBPSA-USA bis,2012). Building models are commonly referred to predict the energy consumption, and their accuracy isrelated to the phase of the design process (IBPSA-USA, 2012). Less importance is given to the designand operation of integrated building energy andcontrol systems model. When the interaction betweenthe energy plant and building model is investigated, astrong reduction of computational effort is required.The model needs to be as complex as needed toachieve its purpose. A good work is made when a balance between accuracy and model complexity isfound. For this reason, it is important to define priorities and to individuate the features which havegreater impact on performances.Main aim of this work is the elaboration of asimplified building model to be used in the study ofan integrated control between building and energy

supply system. To this end, (1) a detailed model has been created; (2) the calibration of the ventilationmass flow rate and infiltration rate has been carriedout for a better agreement between model and realcase; (3) a semplification procedure for the reductionof the computational effort of the detailed model has been developed. A satisfing approximation of theheating demand between the detailed and thesimplified model has been reached with a strongreduction of the simulation runtime.

CASE STUDYThe building under investigation is located inBronzolo, Italy, and it has been built by IPES (a local

social housing institution), in 2006, according to the“CasaClima A Plus” standard (Direttiva Casaclima,2011). It is a residential building with 8 apartmentsfor a total of 577 m2 of conditioned living areadistributed on three storeys. The building is orientedalong the axis North-West to South-East, with afaçade oriented to the South.Domestic Hot Water (DHW) and heating demand arecovered by a 15 kW pellet boiler. The hot water isstored in a tank-in-tank puffer of 800 L and thendistributed to each apartment. A recirculation water

Chiara Dipasquale1,2, Matteo D’Antoni2, Roberto Fedrizzi2, Michaël Kummert3, LuigiMarletta1

1University of Catania, Catania, Italy, chiara.dipasquale@eurac.edu2 Eurac Research, Bolzano, Italy, matteo.dantoni@eurac.edu

3 École Polytechnique, Montréal, Canada, michael.kummert@polymtl.ca

8/18/2019 IBPSA2013

2/8

353

2system is also used in order to provide DHW during peak hours. For the supply air, a forced ventilationsystem with a heat recovery unit is used. External airis pre-heated by geothermal probes, to avoid thefreezing of the ventilation fan; an AHU acts as heatrecovery from the exhaust air to the fresh air and thesupply air is then divided in three ducts, for the

distribution on the three floors. A post-heating ineach apartment is then provided through coils fed bya pellet boiler.

Fig. 1 Ventilation and Domestic Hot Water (DHW)system (IPES, 2007)

A monitoring system has been installed and data ofinternal temperature, relative humidity and CO2 levels, external temperature, relative humidity andsolar radiation, electrical and thermal consumptionhave been collected for a whole year.Geometrical and physical characteristics from designhave been employed to develop a detailed modelwhich reproduces with high accuracy the real building. The large amount of monitored data have been used to define several boundary conditions andto calibrate the model.

DETAILED MODELThe detailed model of the building has been made ina previous work (Ecker, M., 2011) using GoogleSketchUp and Trnsys 3D plugin (Ellis, P., 2009). Walls and floors have been defined according to thereal geometry and orientation; for windows, a predefined window has been used, whosecharacteristics reproduce the original ones. Theenvelope characteristics are reported in Table 1.

Table 1U-values of the building envelope (design values)

Wall type U-value [W/m2K]Exterior walls 0.14Roof 0.08Cellar ceiling 0.15Entrance door 0.7Windows 0.86

Fig. 2 Zone partitioning of basement (BS), ground floor(GF), first (F1) and second (F2) floor

Due to calculation modes’ requirements for runningthe detailed model, only convex zones have beenaccepted. Apartments 1, 2, 3, 5, 6 and 8 have a L-shape, so they have been divided in two zones. Fig. 2shows the result of zone partitioning and the labelsused to indicate the zones. Staircase has beenmodeled as a single zone with 4 stacked air-node,one for each storey. This zone is the only one notheated. In the 3D building model, self-buildingshadings and shadings due to the surrounding have been modeled with several shading groups. In Fig. 3,a picture of the real case and a view of the SketchUpmodel are shown.

Fig. 3 Picture of the real case (above) and view ofSketchUp model (below)

BSGF

F2 F1

8/18/2019 IBPSA2013

3/8

353

3Simulation boundary conditionsSimulations have been run with TRNSYS 17 (KleinS.A., 2009) and the following boundary conditionshave been set:- Weather: monitored weather file has been used.Data have been collected for one year with a timestepof 1 hour. The file format is an *.epw (Energy Plusformat) and it has been read by Type 15.- Infiltration: a yearly fixed value for each apartmenthas been defined after performing a building modelcalibration. It takes into account wind-drivenleakages and average-users behavior.- Ventilation: a fresh air supply flow rate has beendefined for each apartment during the building modelcalibration. A heat recovery of the 85% has beenapplied between the exhaust and the fresh air.- Internal gains: monitored data of electricalconsumption have been used to model internal gainsdue to electrical devices, lighting and cooking. Themeasured value is split into radiative (40%) and

convective (60%) part. Monitored data have beencollected during a whole year with a timestep of 1hour. Monitored data for users’ occupancy are notavailable, so a schedule based on standard EN ISO7730 (EN ISO 7730, 2007) has been used. Theoccupancy profile is assumed to be in accordancewith power consumption profile; different user’sactivities during the day are also taken into account(Ecker M., 2011).- Heating: an indoor air setpoint of 21°C is defined;the heating season is fixed from October to April;- Cooling: an indoor air setpoint of 26°C is defined;the cooling season is fixed from May to September.

DETAILED MODEL CALIBRATIONThe detailed model has been developed following all building characteristics available and monitoringdata. Nevertheless, ventilation mass flow rate andinfiltration rate needed to be further fixed through amodel calibration because not directly measurable.The calibration process consits in an iterative processto match observed and simulated behaviors.

Ventilation The coils’ numerical model has been calibratedconsidering as boundaries the coil itself and usingmonitored data collected for one year with a timestep of 1 hour: temperature and relative humidity ofthe supply air, air speed, water flow rate, inlet andoutlet water temperature have been employed to thisend. For each coil, an UA coefficient has beendefined (Incropera et all., 2007). Rated values areavailable for fixed inlet air temperature (17°C), waterflow-rate (180 m3/h) and inlet water temperature(70°C). Because of off-rated conditions of the airmass and fluid flow rate, the following expressioncan be adopted:

8.0

design

water 160coil m

m1UAUA

(1)Coils have been modeled as a counter-flow heatexchanger (Type 5) with water in the source side andair in the load side. The total air flow rate has been

calculated from monitored air velocity in thechannels; then the guess air flow rate of eachapartment has been defined multiplying the total airflow rate for design fractions (case VEN_1).The discrepancy of the exchanged heat in each coil, between monitored consumption and VEN_1, is quitehigh in some apartments rather than in others. Inapartments 1, 3, 6 and 8 the difference is less than10%, whereas in apartments 4 and 7 (the smallerones), the difference between the monitored and thesimulated exchanged heat in the coils achieves the50%. This discrepancy is due to a wrong position ofthe sensor for the measurement of the air velocity. Tosolve such a discrepancy, the calibration of the airflow rate has been made with the process explainedabove. A corrective value of the guess rate (designedwith Y) has been iteratively calculated up to obtain amatch of simulated and monitored heat in each coil.The correct air flow rate has been calculated for eachtime step as follows:

)1(,, Y mm guessair calibrated air (2)

A constant yearly corrective value for each apartmenthas been individuated; in particular, the mostfrequent value has been chosen rather than anaverage value because not affected by measurements

in transient conditions.The yearly heating demand has been re-calculatedwith the corrected air fractions (case VEN_2) and asatisfactory approximation of the exchanged heat inthe coils has been reached (Fig. 4).

Fig. 4 Comparison of ventilation rate frommonitoring (MONITORING), before (VEN_1) and

after calibration (VEN_2) in terms of exchanged heat

8/18/2019 IBPSA2013

4/8

353

4An increase of 22% of the measured air mass flow ismade, nevertheless, the total air flow still remainswithin the design value of 1440 m3/h (Castagna M.,2009).

Infiltration From monitoring, measurements on the instantaneousinfiltration rate were not available, so the n50 valuefrom the blower-door test has been assumed (caseINFIL_1) as first attempt. In this case, the n50 value isequal to 0.8 h-1(IPES, 2006). Commonly, to calculatethe infiltration rate from the n50 value, the Shermanequation is used (Sherman, 1987):

04.020

50nn (3)

where nreal [h-1] is the infiltration rate.The calculated airtightness value gives an idea of theinfiltration rate, but it does not take into account the building exposure to the wind direction, theventilation strategy and occupant’s behaviour. Forthis reason a survey on the infiltration rate has beenfurther done, in order to fix an overall value for eachapartment.In the staircase, the infiltration rate has been assumedto be 0.35 h-1 (Diamond R.C. et all, 1996) forconsidering the opening of the doors in the basement,in the ground floor and an air intake on the thirdfloor. No calibration for the staircase infiltration ratehas been done, because monitored data were notavailable.For the definition of yearly fixed infiltration rate, thesame procedure used for the ventilation rate has beenapplied. The calibration has been made by comparingsimulated and monitored heating demand in eachapartment. Monitored data have been collectedduring a year with a time step of 1 hour. The iterative process individuated a corrective infiltration value(designed with X) for each time step up to obtain thematch between simulated and monitored heatingdemand.

)1( X nn calibrated (4)

Also in this case, the most frequent X value has beenchosen for the definition of the constant yearlyinfiltration rate for each apartment. In particular, for

apartments 3, 4 and 7, an infiltration rate of 0.34 h-1

has been individuated, while 0.44 h-1 has been takenfor apartments 1, 2, 5, 6 and 8. These values arerelated to the apartment’s size or to the occupancy. New infiltration rate values have been set in thedetailed model (case INFIL_2) and heating demandshave been compared with case INFIL_1 and with themeasured consumption.The discrepancy in the heating demand, for the whole building, between monitored and calibrated case(INFIL_2) has been reduced to 1% (see Fig. 5). A

slight difference between monitoring and simulationstill remains due to the fact that (1) the occupancy profile is assumed (Ecker M., 2011); (2) thethermostat set point temperatures have been deducedfrom monitored indoor air temperature; (3) theinfiltration rate value has been defined to be constantduring the year. As a consequence, higher or lower

heating demand with respect to the monitoredconsumption might occur.

Fig. 5 Specific h eating demand calculated in eachapartment from monitoring (MONIT), before (INFIL_1)

and after (INFIL_2) calibration

SIMPLIFICATION PROCEDUREThe detailed model has been simulated withTRNSYS 17 (Klein S.A., 2009) for a whole year(8760 hours) with a time-step of 5 minutes. Althoughthe real case has been reproduced with high accuracy,on the other hand, the complete simulation required2.7 hours to run. For studying the energy buildingmodel integrated with the supply energy system, a

simplified model is therefore necessary. For thisreason, a procedure has been individuated (see Table2) starting from a detailed model and movingtowards a simplified one. This process consists ofthree steps that analyze the main issues influencingthe energy balance and the simulation runtime in a building model developed with TRNSYS. Theimpact of 1) radiation mode, 2) geometry mode and3) shadings elements on building performance andsimulation runtime have been analyzed. Finally, amodel, with reduced computational efforts whichmaintains sufficient accuracy, has been created. Theanalyzed cases are reported in Table 2 where the usedradiation mode, geometry mode and shading

elements are specified.For each step of Table 2, the incoming (denoted as positive terms) and the out-coming (denoted asnegative terms) energy fluxes of each zone have beenstudied. Yearly energy losses and gains have beenthen calculated for the whole building according tothe boundary conditions presented above.

8/18/2019 IBPSA2013

5/8

353

5Table 2

Steps of the simplification process

CASE Radiationmode ShadingsGeometry

modeRAD_1 Detailed Shad group 3D dataRAD_2 Standard Shad group 3D data

SHD_1 Standard Shad Factor 3D dataSHD_2 Type 34 3D dataGEO_1 Standard Type 34 3D dataGEO_2 Manual

Radiation modeIn TRNSYS 17, two different modes for thedistribution of the radiation within a zone areavailable, the standard and the detailed mode. In thecase of the standard mode, the direct short-waveradiation, is proportionally divided fixing a constantuser-defined fraction (GEOSURF values), while thedetailed mode calculates the distribution of the

entering direct radiation according to shading andinsolation matrices generated by an auxiliary program called TRNSHD (Hiller et all., 2000). Thestandard mode bases the diffuse radiation distributionon absorption-transmission weighted area ratios forall surfaces of a zone, while the detailed modeconsiders the multi-reflection too, thanks to the useof view factor matrix generated by an auxiliary program known as TRNVFM. Finally, the standardmode treats the long-wave radiation with the star-node approach (Seem, J.E., 1987). This approachtakes into account no user defined emissivity ofinside surfaces nor radiation exchange over morethan one air-node. This aspect will be important forfurther considerations. For the treatment of long-wave radiation, the detailed radiation mode considersthe multi-reflection too, using the view factor matrixgenerated by TRNVFM.Comparing the energy balance for cases RAD_1 andRAD_2 (see Table 3), a difference in infiltrationlosses (QINF), ventilation gains (QVENT), transmissionlosses (QTRANS), heating (QHEAT) and cooling demand(QCOOL) is shown.

Table 3Yearly energy gains and losses for the whole building

in RAD_1 and RAD_2 cases

RAD_1 RAD_2Q HEAT [kWh/m2y] 37.6 34.1QCOOL [kWh/m

2y] -4.7 -6.9Q INF [kWh/m

2y] -35.0 -35.4QTRANS [kWh/m

2y] -41.8 -34.8QGINT [kWh/m

2y] 41.9 41.9QSOL [kWh/m

2y] 23.8 23.7QVENT [kWh/m

2y] -21.8 -22.6

In order to figure out the reason of this discrepancy,the radiation absorbed by all the apartments’ wallshas been analyzed. The total radiation absorbed (andtransmitted) at all inside (QABSI) and outside (QABSO)surfaces has been investigated. For the sake ofclarity, the control volume is assumed to be the zone,so the term “inside” is referred to the radiation

coming from the zone, while the term “outside”concerns the radiation coming from outside the zone(Klein S.A. et all, 2009). In cases RAD_1 andRAD_2, the QABSI differs less than 2%, while higherdifference is verified in the QABSO (see Table 4).Labels “EXT”, “BND”, “ADJ” are referred to thesurface’s categories, external, boundary or adjacent,respectively. The radiation mode is referred to theinner radiation distribution, in fact the total externalradiation absorbed at external surfaces (QABSO_EXT) isthe same in both cases. For the detailed mode,negative values indicate the absorption of radiationon a surface, whereas positive heat flux means a netemission. The main differences between standard anddetailed mode have been pointed out by adjacentsurfaces. In fact, all apartments border with anunconditioned multi air-node zone (the staircase andthe lift), that influences the external absorbedradiation for adjacent walls. In particular, lowerapartments are more affected by the exchange withstaircase zone than higher.

Table 4 Absorbed radiation on external walls

CASE Q ABSO [kWh] TOT EXT BND ADJ

A P

_ 1 RAD_1 42495 42927 0 -432

RAD_2 44004 42927 0 1077

A P

_ 2 RAD_1 31353 32098 0 -745

RAD_2 33164 32098 0 1066

A P

_ 3 RAD_1 21297 21990 - -693

RAD_2 23691 21990 - 1701

A P

_ 4 RAD_1 27831 27505 - 326

RAD_2 28497 27505 - 993

A P

_ 5 RAD_1 24763 25098 - -335

RAD_2 26902 25098 - 1804

A P

_ 6 RAD_1 79724 79693 - 32

RAD_2 80535 79693 - 843

A P

_ 7 RAD_1 61442 61261 - 181 RAD_2 61865 61261 - 604

A P

_ 8 RAD_1 113303 113091 - 213

RAD_2 114060 113091 - 969

Maintaining all the boundary conditions unvaried andchanging only the distribution of the radiation, theeffect of the different heat transfer with multi air-node zones can be observed with the Mean Radiant

8/18/2019 IBPSA2013

6/8

353

6Temperature (TMR ) within the zone. This measuretakes into account the area weighted meantemperature of all walls of the zone, so a variation inthe air-node temperature might come out analyzingthe different TMR .The cumulative frequency of the Mean RadiantTemperature for apartments 2 (situated in the GF)

and 8 (situated in the F2) is shown in Fig. 6. Thegraph represents the yearly frequency distribution ofthe TMR in the zones for cases RAD_1 and RAD_2.In apartment 2, the TMR differs of about 0.7°C for the40% of the year, while in apartment 8, there is a goodoverlapping of the two curves.

Fig. 6 Cumulative distribution during a year of the Mean Radiant Temperature in apartment 2 and 8.

A different distribution of the radiation within thezone influences the ventilation load, the heating andcooling supply energy and the infiltration load. In particular, a discrepancy of about the 9% is observedon the whole building heating load.The radiation mode does not influences only internalair temperature within the zones, but also thesimulation runtime. In fact, computational effort arestrongly affected and a reduction of the 86% ofruntime passing from the detailed to the standardradiation mode is observed.In light of this, the choice of the radiation mode isvery important for the reduction of computationaleffort, but attention should be paid when multi air-node zones are adjacent to single air-node zones.

Shading elementsExternal or internal shading elements may be definedfor any transparent surface of a zone. To implementexternal shaders and self-shading of the building inthe SketchUp building model, several “shading

groups” have been used. At the beginning of thesimulation, TRNBuild generates a shading matrixwhich takes into account the presence of shadingelements or surrounding structures. During thesimulation, Type 56 determines the actual sunlitfraction of surfaces thanks to the use of the shadingmatrix file with respect to the sun’s current positionfor each time step. If no shading group is present inthe building model, external shadings might bemodeled in TRNBuild itself. For an external window,the user can select an internal and/or external shading

device, specifying its shading factor, which indicatesthe shaded fraction of the window element.In this work, the effect of two different externalshaders and self-shading of the building model have been investigated and compared with the case withthe shading groups (RAD_2). A building with noshading elements has been modeled and shadings’

effect has been reproduced by External ShadingFactors. An external file with shading factor obtained by the detailed model (case SHD_1) and a TRNSYScomponent, Type 34, (case SHD_2) have been usedto determine the External Shading Factor inputs.The use of both shading factors and Type 34 produces a difference of the yearly heating demandof the whole building, QHEAT, of 1% with the case ofshading groups (case RAD_2). The use of Type 34increases the cooling demand of the building, QCOOL,of around the 8%.In order to understand the different cooling demandneeded in the two cases, single apartments have beenanalyzed. As already seen, the detailed radiationmode does not consider irradiation from shade projected on exterior surfaces other than windows.For this reason, the incident radiation on windowsonly is taken into account.Case SHD_1 has been considered to find out thedifference of the simulation runtime when an externalfile with the sunlit portion area is used. Regarding theenergy building balance, as the external file here usedcorresponds to the external shading factor of theRAD_2 case, no difference is shown.The accuracy of the shadings modeled with Type 34influences the agreement between cases SHD_2 andRAD_2. In Table 5, the incident radiation on allapartments’ windows for cases RAD_2 and SHD_2

is reported. The third column indicates the difference,in percentage, of incident radiation between the twocases.

Table 5 Incident radiation on all apartments’ windows in

RAD_2 and SHD_2 cases

CASE RAD_2 SHD_2 RAD 2 , S HD 2 [MWh] [MWh] [MWh]

AP_1 6.2 6.2 -0.3%AP_2 5.9 6.7 -12.8%AP_3 7.2 6.7 7.6%AP_4 3.6 4.9 -30.8%AP_5 5.9 5.7 4.7%AP_6 8.5 7.4 13.2%AP_7 4.0 3.6 10.2%AP_8 6.0 6.2 -3.6%

Type 34 has been set to model roof’s overhangs and balconies, while surrounding contribution and building’s shadings (wall thickness, balconies on theadjacent sides) have not been reproduced. A higheffect of this modeling is shown in apartment 4,where the difference with case RAD_2 is 31%, or in

8/18/2019 IBPSA2013

7/8

353

7that apartments oriented to South-East, where thedifference amounts to the 8-13%.Regarding the simulation runtime, the use of externalfile for the external shading factor (case SHD_1)reduces the simulation runtime of 6%, while the useof Type 34 (SHD_2) increases it of about 4%. Theresult of case SHD_2 also depends on the number of

Type 34 units used into the model.

Geometry modeFor each zone, TRNBuild supports different levels ofgeometric surface information, known as “manual”and “3D data” mode. In the manual mode, thegeometry of the building is individuated according tothe definition of walls and floors and their boundaryconditions. The advantage to model directly inTRNBuild is that no detailed shape definition isrequested. Defining the area and the boundaryconditions for each surface, the softwareautomatically calculates the interactions between thesurfaces and the zone. Walls’ categories used in thiscase are external, boundary and adjacent. “External”is referred to walls which border to outside,“boundary” is a wall in which boundary conditions ofthe first type can be specified and “adjacent” is a wallwhich borders another air-node.The “3D data” mode provides, for all surfaces of thezone, three dimensional coordinates. Geometrymodel is designed in Google SketchUp with theTrnsys3D plugin and then an *.idf file is imported tothe TRNBuild environment. If the detailed radiationmode is used, radiative zones must be convex polyhedrons.Looking at yearly energy gains and losses in cases

GEO_1 and GEO_2, no differences have beenobserved. For both cases, the simulation runtime isalso unvaried.For the calculation of shadings effect, the manualgeometry mode does not use shadings matrix. In fact,even if shading elements are inserted in the buildingmodel, the geometry mode does not consider theseinformation. For this reason, if this mode is used,external inputs for modeling the shadings arerequired.

SIMPLIFIED MODELThe steps analyzed above have been used as a guidefor the definition of a simplified building model. Thisnew model has been conceived following the criteriaof reducing the efforts during the numerical modelingdesign phase and the computational efforts:- manual geometry mode has been applied in order to

create zones in a flexible way because walls’geometry and category are specified directly in theTRNBuild environment;

- standard radiation mode has been used to run themodel, in order to reduce the simulation runtime

and to draw the zones with any shape notnecessarily convex;

- the use of Type 34 has allowed to model shadingelements without any previous calculation for thedefinition of the sunlit portion area of the windows.

Building energy gains and losses of the simplifiedmodel have been compared with the reference case in

order to verify the accuracy of the results.With the use of standard radiation mode and manualgeometry mode, a single zone for each apartment has been created. The staircase zone has been created asin the detailed model. All the characteristics and boundary conditions of the detailed mode have beenset in the simplified model, too. The only exceptionhas been the definition of windows because the totalamount of glazed surface for each external wall has been taken into account.The difference of heating demand of the whole building between reference case and the simplifiedcase is around the 8% (Fig. 7). Higher differences areobserved in those apartments in which the constantinfiltration value approximates with lower accuracythe real infiltration rate or in which the incident solarradiation differs more from the reference case (seeTable 5).Comparing the simulation runtime in the two cases, astrongly reduction of the 89% is observed decreasingfrom 2.7 hours to 20 minutes. This result is due to thefact that the standard radiation mode has been usedand 9 zones (apartments zones + staircase zone)versus the 15 of the Reference case have beenmodeled.

Fig. 7 Yearly specific heating demand in thereference case (RAD_1) and in the 8 zones case

CONCLUSIONSBuilding modelling integrated in supply energysystems are more and more challenging for theinterdisciplinary of the treated aspects and thecomplexity of the model itself. A simplification procedure of a detailed building model has been presented in this work. The starting point has been adetailed model which has been calibrated withmonitored data to obtain an accurate model.Theimpact of radiation mode, geometry mode andshadings elements on building performance and

8/18/2019 IBPSA2013

8/8

353

8simulation runtime have been then analyzed. Amodel with sufficient accuracy and reducedcomputational effort has been developed.Simulations have been run in TRNSYS 17 and thefollowing considerations have been made inaccording to the calcolation modes used by thissimulation tool.

The use of the manual or 3D data geometry mode,makes no changes in both simulation runtime and building energy response. An important reduction ofcomputational efforts is instead observed in movingfrom a detailed radiation mode to a standard mode(around 85%). This simplification leads to a changein the air-node temperature in those zones which border with unconditioned multi air-node zones.When the standard mode is used, the building heatingand cooling demand differs of about 9% with respectto the detailed mode.The modeling of External Shading Factor withexternal inputs does not reduce significantly thesimulation runtime, which might varies of around ±5%. The influence on energy gains and losses usingType 34 depends on the apartment’s orientation.Type 34 models shadings taking into account onlyoverhangs or wing-walls of the building and not theshading effect of the surrounding.The use of standard radiation mode leads asignificant reduction of simulation runtimemaintaining the same solar radiation arriving onwalls and windows, but neglecting the effect ofmulti-reflection within the zone. Differences might be observed in that cases in which conditioned zones border with unconditioned multi air-node zones. Ifdetailed radiation mode is used and the shadingsshape is not complex to be modeled, there are no

advantages in using external inputs for the definitionof the sunlit portion area. Indeed, if the geometrycharacteristics of the building are defined directly onTRNBuild and the manual geometry mode is set, theuse of inputs for the External Shading Factor isrequested. In particular, better results can be achievedwith an external file which defines the sunlit portionarea considering all the shadings effects (for exampleobtained by a shader program (Nathaniel L. et all,2006). If these information are not available a goodapproximation of shading effect on windows can bemade by Type 34.A unique solution for the reduction of computationalefforts does not exist because it is strictly related to

the aim of the simulation and the parameters whichhave to be analyzed. The main aim of the work here presented is the definition of a final building modelwhich reproduces as better as possible the real building behavior, reducing computational effortsand the time consuming during the design phase.For this purpose, standard radiation mode, manualgeometry mode and Types 34 have been selected forthe simplified model.

The final simplified building model runs in 20minutes with a reduction of the simulation runtime ofthe 89%. A discrepancy of the heating demand of 9%has been found and, for the aim of the work, it might be considered suitable.

REFERENCESCastagna, M. 2009. Studio di una Casa Passiva con

un programma di simulazione dinamica.Universitá di Trento.

Diamond, R. C., Feustel, H. E. and Dickerhoff, D. J.1996. Ventilation and Infiltration in High-RiseApartment Buildings. Technical report,Lawrence Berkley Laboratory, University ofCalifornia, Berkley, California.

Ecker, M. 2011. Modelling and Parametric Analysisfor a Small-Scale Solar Heating and CoolingSystem. Master thesis, FachhochschuleTechnikum, Wien.

Ellis, P. 2009. OpenStudio plugin for GoogleSketchUp3D, Version 1.0, January.http://sourceforge.net/projects/openstudio

EN ISO 13786:2007. Thermal performance of building components – Dynamic thermalcharacteristics – Calculation methods.International Organization for Standardization.

IBPSA-USA. Energy Modeling of Buildings.IBPSA_US.http://energymodeling.pbworks.com

IBPSA-USA bis. 2012. Modeling Best Practice.Building Energy Modelling. IBPSA-USA,Rocky Mountain Institute, ASHRAE.http://bembook.ibpsa.us/

Gebhart, B. 1971. Heat Transfer, McGraw-Hill, NewYork.

Hiller, M.D.E., Beckman, W.A, Mitchell, J.W. 2000.TRNSHD-a program for shading and insolationcalculations, Building and Environment, Vol. 35.

Incropera, F., Dewitt, D., Bergman T., Lavine A.2007. Fundamentals of heat and mass transfer.John Wiley & sons, United States of America.

IPES. 2006. La casa passiva Bronzolo. Istituto perl’edilizia sociale della Provincia Autonoma diBolzano, Bolzano.

Klein, S.A. et al. 2009. Trnsys 17. A transient

simulation program. Solar Energy Laboratory,Univeristy of Wisconsin, Madison. Nathaniel, L.J., Donald, P. 2012. Hardware

accelerated computation of direct solar radiationthrough transparent shades and screens. Programof Computer Graphics, Cornell University,Ithaca, New York, USA.

Seem, J.E. 1987. Modeling of Heat in Buildings. PhDthesis, Solar Energy Laboratory, University ofWisconsin Madison.