a convenient coupled simulation method for thermal ...palencaivc2007\volume2... · been...

$: A convenient coupled simulation method for thermal ...PalencAIVC2007\Volume2... · been investigated and validated with full CFD simu-lation (Wang and Wong, 2006). Multi-zone scenari-os$
10972nd PALENC Conference and 28th AIVC Conference on Building Low Energy Cooling and Advanced Ventilation Technologies in the 21st Century, September 2007, Crete island, Greece

A convenient coupled simulation method for thermal environment prediction in naturally ventilated buildings

L. Wang, N. H. WongWongNational University of Singapore, Singapore

prediction can be simplified. However, the airflow net-work method for airflow estimation in building simula-tion cannot accurately predict indoor airflow. Therefore, how to effectively integrate the BS and CFD simulation to improve the accuracy to assess the performance of natural ventilation becomes a primary objective. In this paper, coupling procedures and a coupling pro-gram interface are introduced and the developed cou-pling program is further validated with two methods: full CFD simulation (both indoor and outdoor airflow are calculated simultaneously) and field measurement results, which are presented as follows.

2. COUPLING PROCEDURES

Commercial software FLUENT is used as the external CFD program to couple with ESP-r (Clarke and McLean, (Clarke and McLean,Clarke and McLean,, 1988). A one-way coupling is adopted for wind-drivencoupling is adopted for wind-driven is adopted for wind-driven natural ventilation from ESP-r to FLUENT simulation. From building simulation, the inside surface tempera-tures for each wall, detailed boundary conditions for the openings including velocity boundary conditions (or pressure boundary conditions), airflow direction, air tem-perature are provided for detailed indoor CFD simula-tion. The predicted indoor average temperature by CFD simulation will be compared with zone temperature from building simulation results. A text-mode data exchange interface has been programmed to provide the conven-ient data passage between the two programs to auto-matically implement the data exchange between the two programs. Detailed thermal comfort prediction model for indoor environment is finally provided for evaluation in the data interface. Coupled simulations can be done au-tomatically using the program interface. In the follow-ing section, coupled simulation results are to be validated with full CFD simulations and field measurement studies.

3. FULL CFD VALIDATION

In this full CFD validation, a multi-zone scenario was investigated for full CFD validation. Two cases (case 1: 2nd Jan 18:30 and case 2: 1st Jan 12:30), with wind di-rection 0°(N)and wind direction 45°(NE), selected from different weather conditions in Singapore weather file 2001 are simulated in the scenario and the climatic data

ABSTRACT

As both CFD and building simulation have their own limita-tions in the thermal prediction of natural ventilation. ExternalExternal coupled simulation method between CFD (FLUENT) and building simulation (ESP-r) has been put forward for ther-) has been put forward for ther- been put forward for ther-mal environment prediction for naturally ventilated building design. The coupling mechanism and coupling program in-The coupling mechanism and coupling program in- coupling mechanism and coupling program in-terface are introduced in the paper. Full CFD simulation andFull CFD simulation andull CFD simulation and on site field measurements are done for the validation of the coupled simulation method. The results of coupled simula-simulation method. The results of coupled simula- method. The results of coupled simula-The results of coupled simula-he results of coupled simula-tions show that the integration of CFD and building simula-tion can accurately and efficiently provide indoor thermal environment and can be used as a convenient tool for fa-convenient tool for fa- tool for fa-fa-cade designs for naturally ventilated residential buildings. designs for naturally ventilated residential buildings.naturally ventilated residential buildings. ventilated residential buildings.Key words: Coupled simulation, CFD, Building simulation

1. INTRODUCTION

The concept of natural ventilation is well accepted and welcomed by people and designers in the world, espe-cially when the benefits of natural ventilation, including reducing operation cost, improving indoor air quality and providing satisfactory thermal comfort in certain climates, are being acknowledged. Even in hot-humid climates, where air-conditioners are common in both residential and commercial buildings, HDB (Housing & Development Board) residential buildings, where about 86% of people in Singapore live, are designed to be naturally ventilated.CFD simulation is one of the important methods for natural ventilation study. It can provide detailed air tem-perature, air velocity, contaminant concentration within the building or outdoor spaces. It has become a reli-able tool for the evaluation of thermal environment and contaminant information. However, thermal comfort prediction of naturally ventilated buildings with CFD simulation only is not an easy task. Building simula-tion (BS) is another popular method for investigating naturally ventilated building design. Thermal simula-tion and airflow network are two fundamental modules in building simulation tool. With the aids of building simulations (BS), which provide rapid prediction of fa-cade thermal behaviours through solving the heat and mass transfer and airflow network in the building sys-tems in one dimension, the task for natural ventilation

PALENC 2007 - Vol 2.indd 1097 7/9/2007 1:26:06 µµ

1098 2nd PALENC Conference and 28th AIVC Conference on Building Low Energy Cooling and Advanced Ventilation Technologies in the 21st Century, September 2007, Crete island, Greece

for both cases are shown in Table 1.Table 1: Climatic data

Case Wind Direction

(0°N; 90°E)

Wind Speed (m/s)

Temperature(oC)

Relative Humidity

(%)

1 0 1.7 25 922 45 5.8 29 76

Single zone scenarios with coupling program have been investigated and validated with full CFD simu-lation (Wang and Wong, 2006). Multi-zone scenari-os are more complicated than single zone scenarios since one of the opening in one zone as outlet (inlet) has been used to connect with another zone as inlet (outlet). Therefore, the boundary conditions for the internal opening are significant but difficult to pre-dict. The comparison results of single zone scenarios show that taking pressure as opening boundary con-ditions is more accurate than taking velocity as open-ing boundary conditions (Wang and Wong, 2006). Therefore, for the multi-zone scenarios, only pres-sure boundary conditions are taken into consideration. There are three ways to treat internal openings, which connect two nodes (rooms). The first one is called the “pre” method. In this method, indoor CFD simulation is carried out in one of the two rooms that we are interest-ed in. The internal opening boundary condition is taken from the neighbouring node pressure (the pressure of the other room) as internal opening boundary condition in the “pre” method. The second method is called the “pre-ave” method. Similar to the “pre” method, indoor CFD simulation is carried out in one particular room, but the pressure boundary conditions for internal openings are taken as the average value of two connected nodes. The third method is called “pre-room” method. In this method, the whole connected zones are in the compu-tation domain for indoor CFD. Therefore, the internal openings are treated in CFD as default openings and no pressure values need to be assigned to them beforehand. The inaccuracy originated from the estimation of inter-nal opening pressure could be eliminated using the third strategy. The three methods are tested and compared us-ing coupled simulation results and full CFD simulation results for a three zone scenario in the following sections.It is a three-zone naturally ventilated room including zone 1: north living, zone 2: south kitchen and zone 3: block. The dimension of each zone is 3mx5mx3m (Lx-WxH). In the aspect of airflow paths, north living zone and south kitchen are connected by an internal door and the block zone is not connected with other zones. The layout of the multi-zone is shown in Figure 1. Coupled1. Coupled. Coupled ESP-r and FLUENT simulations are processed within two connected zones. The empirical value of the dis-

charge coefficient 0.65 is adopted in the ESP-r simula-tion. The full CFD computations for both indoor and outdoor airflow simulations are simulated for valida-tion of coupled simulations. The velocity inlet bound-ary condition for full CFD simulation uses the 1/4 shear flow profile to model the atmospheric boundary layer. Pressure-outlet boundary condition is used for the full CFD simulation domain. Symmetry boundary condition is adopted at the top of the computational domain.The airflow network in scenario 1 comprises four nodes (north, south, and living and kitchen) in the building simulation ESP-r. ‘north’ and ‘south’ are the names of the wind-induced boundary nodes representing outdoor node in north direction and south direction respectively. The nodes, ‘living’ and ‘kitchen’, indicate the windward zone and connected passive zone, respectively.

Figure 1 A three-zone room with two opposite windows layout1 A three-zone room with two opposite windows layoutA three-zone room with two opposite windows layout (Scenario 1)1))

The ESP-r simulation results provide boundary condi-tions for indoor CFD simulation. The area weighted ve-locity magnitude results at the height of 1.5 m shown inshown in Table 2 are compared among full CFD simulation, cou-2 are compared among full CFD simulation, cou- are compared among full CFD simulation, cou-pled ESP-r and indoor CFD simulation in a particular zone (pre, pre-ave, pre-room), ESP-r simulation only. The area-weighted air velocity for full CFD simulation is the indoor averaged air velocity at a certain level, and for ESP-r is defined by Equation 1

AmVρ

= (1)

where V (m/s) represents the area-weighted indoor air velocity, which is unique in multi-zone airflow simula-tion with well-mixed assumption, m (kg/s) is the total mass flow rate, ρ (kg/m3)is the air density from the zone where the air flows into, and A (m2)is the section area perpendicular to the airflow.The detailed area weighted velocity magnitude results for case 1 and case 2 are compared in Figures 2 and 32 and 3

��

��



along height and length for living zone. It can be seen. It can be seen It can be seen from comparison results that with proper coupling be-tween the building simulation and CFD program with pre-ave and pre-room methods, the indoor airflow pro-file can be well predicted for natural ventilation. The results show that the coupling program results are consistent with the full CFD simulation results. The coupling program can better predict the average ve-locity in the room than using building simulation pro-gram alone. However, coupled simulations may not be able to accurately predict the velocity profile near the inlet. In addition, the pre method shows the larg-est errors between full CFD and coupled simulations. Table 2: Result comparison2: Result comparison Result comparison

Area Weighted

room velocity(1.5m

Height)

Full CFD

simula-tion

(m/s)

Indoor CFD with pressure

inlet (m/s)

Indoor CFD

with pres-pres-sure –ave–aveave

(m/s)

Indoor CFD with pressure-

room (m/s)

ESP-r (m/s)

Case 1: 2ase 1: 2nd Jan 18:30 0.17048 0.2633 0.2214 0.1954 0.091

Case 2: 1ase 2: 1ST Jan 12:30 12:30 0.56018 0.6999 0.5740 0.64039 0.286

Figure 2 Area_weighted velocity results comparison along verti-2 Area_weighted velocity results comparison along verti-Area_weighted velocity results comparison along verti-cal (z) direction and length (y) direction for living room in case 1 among full CFD simulation, indoor CFD simulation with average pressure boundary condition, indoor CFD simulation with pressure boundary condition, and indoor CFD simulation for the whole room.

Figure 3 Area_weighted velocity results comparison along verti-3 Area_weighted velocity results comparison along verti- Area_weighted velocity results comparison along verti-cal (z) direction and length (y) direction for living room in case 2 among full CFD simulation, indoor CFD simulation with average pressure boundary condition, indoor CFD simulation with pressure boundary condition, and indoor CFD simulation for the whole room.

4. VALIDATION WITH FIELD MEASUREMENT

The HDB block601 is located in Clementi, West coast. It is facing 18 degree northwest and the height of the building is 38.7 meters. A sixth level 4-room unit has been used for this field measurement. The digital multi-channel data logger of BABUC has been used to record indoor thermal parameters, including dry bulb tem-perature, indoor velocity, relative humidity, globe tem-perature and wet bulb temperature. The measurement was taken at the centre point of the living room at 20 minute interval. Eight thermal coupling wires (T type) have been used to measure the surface temperature both inside and outside of external wall in living room and kitchen room at 10 minute interval..There are three bedrooms in the unit (Figure 4). For the4). For the). For the purpose of validation, the doors of bedroom connected with the living room are closed. The internal doors are inhe internal doors are in good airtightness. Therefore, the airflow network in the unitairtightness. Therefore, the airflow network in the unitness. Therefore, the airflow network in the unit comprises two boundary nodes (north and south nodes) and two internal nodes (living room node and kitchen node), as illustrated in Figure 4. The airflow route of the unit throughhe airflow route of the unit through the living room and kitchen room is North->living room ->kitchen ->South or South->kitchen->living room->North.



The weather data were obtained from NUS weather station, located on the rooftop of building E2 (Faculty of Engineer-ing) of the National University of Singapore, Kent Ridge campus. The geographical coordinates are approximately: 1 deg18 min N (latitude) and 103 deg 46 min E (longi-tude). The vertical height of the weather station above the ground is 25.3 m (at the cup anemometer position).The simplified equation (2) is used to describe the rela-tionship between building height wind speed and wind speed measured from Meteorological station at a stand-ard height of 10m.

al

l kzvv

=10 (2) lv ---building height wind speed (m/s)

v10---wind speed measured from Meteorological station at a standard height of 10m (m/s)

lz ----building height (m)k, a ----constants dependent on terrain. Here, we use the terrain coefficients for urban, k=0.35 and a=0.25.

Because the NUS weather station is located on a small ridge, however, the positioning of the instruments does not conform to WMO guidelines (WMO,1983) for weather stations in open terrain or recommendations for urban areas. We use the modified wind speed equation (3) for the building height wind speed.

a

nm

l

nm

l

zz

vv

=

(3)vnm--- wind speed measured from NUS weather station (m/s)znm ----NUS weather station height (25.3 m)

Figure 4 the layout of the four-room HDB unit 4 the layout of the four-room HDB unit the layout of the four-room HDB unit

4.1 ESP-r simulationA HDB building has been modelled in ESP-r (a 14-sto-rey residential building). The subject of the ESP-r simu-lation is the living room of a HDB unit on the sixth floor. The purpose of the simulation is to compare the simu-lation results with field measurements. There are alto-

gether 22 zones in the building model. Each neighbour-ing level of the sixth level, such as the fourth level, fifth level, seventh level and eighth level, are divided into two zones, including left zone and right zone. The sixth level includes 10 zones for each specific unit layout. The other distant levels have to be combined together due to the limitation of maximum-zones and geomet-ric model in ESP-r. The windows in living room and in kitchen zone are adopted half area (1.44m2) as airflow network components. The building material setting is consistent with the specific materials in the real build-ing. The solar absorptance of external walls is set to 0.3 (white color surface) and the emissivity is set to 0.9. The simulation results of ESP-r compared with field measurement have been shown in Figures 5-9. The in-s 5-9. The in- 5-9. The in-5-9. The in--9. The in-9. The in-. The in-ternal surface temperature and external surface temper-ature calculated with ESP-r have been compared with those measured surface temperature (shown in Figuress 5 and 6). The comparison results indicate that the inter- and 6). The comparison results indicate that the inter-and 6). The comparison results indicate that the inter-). The comparison results indicate that the inter-nal surface temperature with building simulation agrees well with the measuring data and the external surface temperature shows fluctuations around the measured external surface temperature. The relative humidity simulation results are well fitted with measuring results (Figure 7). The comparison results for dry bulb temper-7). The comparison results for dry bulb temper-). The comparison results for dry bulb temper-ature for the living room are shown in Figure 8. From8. From. From the results, it can be seen that building simulation tends to predict slightly higher zone temperature than field measurement results. Figure 9 shows the indoor air ve-9 shows the indoor air ve- shows the indoor air ve- the indoor air ve-the indoor air ve-locity results between field measurement and building simulation. As we can see from the results, the predicted indoor air flow is much lower than the measuring data.

Figure 5 Internal surface temperature of living room comparison5 Internal surface temperature of living room comparison Internal surface temperature of living room comparison between measurement and building simulation

Figure 6 External surface temperature of living room comparison6 External surface temperature of living room comparison External surface temperature of living room comparison between measurement and building simulation

v10

kz

vnm



Figure 7 Relative Humidity result comparison between measure-7 Relative Humidity result comparison between measure- Relative Humidity result comparison between measure-ment and building simulation

Figure 8 Dry bulb temperature result comparison between mea-8 Dry bulb temperature result comparison between mea- Dry bulb temperature result comparison between mea-surement and building simulation

Figure 9 Indoor air velocity result comparison between measure-9 Indoor air velocity result comparison between measure- Indoor air velocity result comparison between measure-ment and building simulation

4.2 Coupled simulation resultsIt has been noticed that the multi-zone model in ESP-r could not accurately predict the indoor air velocity, which would probably affect the prediction of other indoor thermal parameters. Therefore, to improve the accuracy of natural ventilation prediction, the cou-pling program between ESP-r and CFD are adopted. The predicted indoor environment (indoor air velocity and dry bulb temperature) with coupling program be-tween building simulation (ESP-r) and computationalSP-r) and computational-r) and computational fluid dynamics(CFD), has been compared with the field measuring data as shown in Figure 10 and Figure 11.10 and Figure 11. and Figure 11.11..Figure 10 shows the comparison of indoor air veloc-10 shows the comparison of indoor air veloc- shows the comparison of indoor air veloc-ity among field measured data, ESP-r simulation results, ESP-r simulation results ESP-r simulation resultsSP-r simulation results-r simulation results and coupling program results from Dec 26, 2005 to Dec 29, 2005. The field measurement results are taken at the level of 1.5m above the floor near the centre of the room.

The predicted area-weighted velocity at 1.5m above the floor and the predicted indoor air velocity at the specific measuring point are obtained by using BS-CFD coupling program for a time series simulation. Compared with theCompared with the field measurement results, the predicted indoor air veloci-ties by ESP-r are quite low, although the results show good correlation between the predicted velocity by ESP-r and measured data. These results confirm that building simu-These results confirm that building simu-lation (airflow network) cannot accurately predict mass airflow rate and indoor air velocity. This can be attributed to simply using a set of empirical equations governedsimply using a set of empirical equations governed empirical equations governed by pressure differences for mass flow rate prediction of different kinds of components connecting airflow in the building simulation. Therefore, mass flow rate passingTherefore, mass flow rate passingass flow rate passing by each component cannot be accurately estimated. Thee underestimation becomes more obvious especially for cross ventilation conditions, under which air velocities around openings and indoor air flow rates can be highly increased. In the airflow network simulation, mass flow rates are calculated based on empirical equations relateds are calculated based on empirical equations related are calculated based on empirical equations related with the types of components and only mass conservations and only mass conservation and only mass conservation equation is considered, but momentum equation and tur-bulence equation are not reflected in the airflow network. It can be seen from the comparison results (Figure 10) that from the comparison results (Figure 10) that with the use of coupling program, the indoor air veloc-coupling program, the indoor air veloc-ity prediction can be largely improved and the predictedan be largely improved and the predicted be largely improved and the predicted air velocity with BS-CFD coupling program can providean provideprovide the designers more reliable indoor air velocity results.Figure 11 shows the comparison results of indoor air dry11 shows the comparison results of indoor air dry shows the comparison results of indoor air dry bulb temperature among field measurement results and ESP-r simulation results and coupling program resultsSP-r simulation results and coupling program results-r simulation results and coupling program results from Dec 26 –Dec 29, 2005. The area-weight indoor air dry bulb temperature and the dry bulb temperature at the specific measuring point are obtained by using BS-CFD coupling program. From the figure, it can be seen that the predicted dry bulb temperature by ESP-r is quite closeSP-r is quite close-r is quite close to that predicted by coupling program, which are mainly controlled by outdoor ambient temperature.From the comparison results, it can be concluded that quasi-steady coupling program could largely improve the accuracy of indoor air velocity prediction and fur-ther improve thermal comfort evaluation.

Figure 10 Indoor air velocity comparison among Field measure-10 Indoor air velocity comparison among Field measure- Indoor air velocity comparison among Field measure-ment, ESP-r simulation only and coupled ESP-r-CFD simulationSP-r simulation only and coupled ESP-r-CFD simulation-r simulation only and coupled ESP-r-CFD simulationonly and coupled ESP-r-CFD simulationand coupled ESP-r-CFD simulation coupled ESP-r-CFD simulationESP-r-CFD simulationSP-r-CFD simulation-r-CFD simulation



Figure 11IndoorairtemperaturecomparisonamongFieldmeasure-11IndoorairtemperaturecomparisonamongFieldmeasure- Indoor air temperaturecomparisonamongFieldmeasure- comparisonamongFieldmeasure- among Field measure-ment, ESP-r simulation only and coupled ESP-r-CFD simulation.SP-r simulation only and coupled ESP-r-CFD simulation.-r simulation only and coupled ESP-r-CFD simulation.coupled ESP-r-CFD simulation.ESP-r-CFD simulation.SP-r-CFD simulation.-r-CFD simulation..

5. CONCLUSION

A coupling program between ESP-r and FLUENT and coupling program between ESP-r and FLUENT and an automatically running text-mode interface was devel-oped. The validation results by full CFD simulation andThe validation results by full CFD simulation and validation results by full CFD simulation and field measurements indicate that the coupling programhe coupling program can improve indoor thermal environment predictionan improve indoor thermal environment predictionimprove indoor thermal environment prediction for better prediction of natural ventilation taking wind as the major force. Comparing with full CFD simula-tion (both indoor and outdoor airflow simulation), the coupling program can largely reduce the computationalcan largely reduce the computational largely reduce the computational cost and the impacts of solar radiation on facade heat gains could be easily considered. On the other hand, it can improve the accuracy airflow prediction for wind improve the accuracy airflow prediction for wind driven natural ventilation based on building simulation program. The coupling program between BS and CFDThe coupling program between BS and CFD simulation can improve the accuracy and efficiency for predicting indoor thermal environment in detail and canin detail and canand can be used as a convenient tool for facade design evalua-convenient tool for facade design evalua- tool for facade design evalua-facade design evalua- design evalua-tion of naturally ventilated buildings.

REFERENCES

Clarke JA, and McLean D.(1988). ESP-A building and plant energy simulation system. Strathclyde: Energy Simulation Re-search Unit. University of Strathclyde.FLUENT user’ s guide version 6.2. FLUENT Inc.’ s guide version 6.2. FLUENT Inc. s guide version 6.2. FLUENT Inc. Wang, L. and Wong N.H. (2006). Coupling between CFD and building simulation for better prediction of natural ventilation. 2nd INTA conference, Jogjakarta, Indonesia, 3-5 April,2006.WMO (1983). Guide to Meteorological Instruments and Meth-ods of Observation. World Meteorological Organization No.8, 5th edition, Geneva Switzerland.


a convenient coupled simulation method for thermal ...palencaivc2007\volume2... · been...

Documents