ASHRAE LV-11-C075-2011 CFD Simulation of Cross-Ventilation Using Fluctuating Pressure Boundary Conditions.pdf

《ASHRAE LV-11-C075-2011 CFD Simulation of Cross-Ventilation Using Fluctuating Pressure Boundary Conditions.pdf》由会员分享，可在线阅读，更多相关《ASHRAE LV-11-C075-2011 CFD Simulation of Cross-Ventilation Using Fluctuating Pressure Boundary Conditions.pdf（8页珍藏版）》请在麦多课文档分享上搜索。

1、& L. James Lo is a doctoral student in the Department of Civil, Architectural and Environmental Engineering, the University of Texas at Austin, Austin, TX. Atila Novoselac is an assistant professor in the Department of Civil, Architectural and Environmental Engineering, the University of Texas at Au

2、stin, Austin, TX. CFD Simulation of Cross-Ventilation Using Fluctuating Pressure Boundary Conditions L. James Lo Atila Novoselac, PhD Student Member ASHRAE Member ASHRAE ABSTRACT As much of the world gears toward developing energy efficient and healthy buildings, natural ventilation is becoming an i

3、mportant aspect of building design. While ventilation driven by the stack effect is relatively simple to predict for many different building geometries, the prediction of wind-driven natural ventilation is far more complex due to ever-changing wind speed and direction. Currently, building designs ut

4、ilizing wind-driven ventilation can only rely on roughly estimated airflow behaviors. In this study, we used full-scale experimental data and computational fluid dynamics (CFD) with the two equation turbulence model (k - RNG) and fluctuating pressure boundary condition s to determine the accuracy of

5、 this standard turbulence model in analyses of cross-ventilation airflow. Experimental data for wind induced airflow was obtained by consistently measuring airflow characteristics inside and outside of a test house. The measured facade pressures were then used as the unsteady boundary conditions for

6、 the indoor CFD airflow model. The CFD results were then compared with the experimental data to determine whether the prediction was accurate. The preliminary finding of this study suggests that even with unsteady winds, time-varying pressure boundary conditions can be used to model and predict comp

7、lex wind driven indoor airflow characteristics using CFD. This method can be helpful to engineers in assessing potential for natural ventilation as well as in designing an appropriate system for a proposed building. INTRODUCTION Energy efficient building is synonymous with reducing building energy c

8、onsumption, and the most effective way is by reducing energy used for buildings heating, ventilation and air conditioning needs. Designers today have tried to design buildings embracing the natural ventilation idea used by many cultures for thousands of years. However, with the size of buildings, co

9、mplex geometry and density of buildings in urban society , natural ventilation, especially wind driven cross- ventilation, is difficult to design. A primitive community can easily use prevailing wind due to the simplistic dwelling design and the lack of manmade landscapes. Compared to these pioneers

10、, designers today must acquire more information in addition to typical weather data to incorporate cross-ventilation in their much more challenging building designs. However, details related to cross-ventilation designs remain scarce at the present time. Researchers have attempted to fill this gap b

11、y studying cross-ventilation with different opening configurations. Kato et al. (1992) explored whether Large Eddy Simulation (LES) can be used to simulate the interaction between outdoor and indoor air flow and found it suitable. However, Kato et al (1997) also followed with a “chained analysis” co

12、mbining CFD with wind tunnel testing, citing the difficulty to simulate both indoor and outdoor airflow practically. The concept of only using CFD for indoor airflow and obtaining the CFD boundary conditions from wind tunnel testing was novel at the time, but a literature review yields no further de

13、velopment of such techniques. There were, however, many later literatures which tried to address the cross-ventilation issue by taking a different route. LV-11-C075 2011 ASHRAE 621 2011. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in AS

14、HRAE Transactions, Volume 117, Part 1. For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAES prior written permission.Ohba et al (2001) used a wind tunnel study to show cross-ventilation was dominated by outdoor

15、 eddies and related less to the incident angle of the prevailing wind. Like Kato, Jiang and Chen (2002) and Jiang et al. (2003) studied the LES simulation of outdoor eddies using experimentally obtained boundary conditions with results agreeing with the experimental data. Tan and Glicksman (2005) in

16、tegrated the multi-zone model with CFD providing a simpler way for determining natural ventilation in buildings but without experimental validation. Evola and Popov (2006) showed CFD with the Renormalization Group (RNG) turbulence model yields greater accuracy in determining natural ventilation flow

17、 rates compared to the standard K- model. Karava et al (2007) showed the opening discharge coefficient is very important for estimating the natural ventilation rate. Also, Chu et al (2009) found the mean air flow rate can be predicted by knowing the external pressure distribution. All these studies

18、further the understanding of cross-ventilation but there is still a need for further improvement especially in modeling of cross ventilation with fluctuating wind. While many researchers have attempted to use LES to capture wind turbulence and couple outdoor and indoor air flow, LES is limited to la

19、rge research institutes due to the high computational cost (Chen 2009). Furthermore, recent studies such as Chu and Wang (2010) explored opening loss factor and its relationsh ip with faade pressure in wind-driven natural ventilation scenarios. On the CFD front, Nore et al. (2010) examined the limit

20、ation of Reynolds-averaged Navier-Stokes (RANS) modeling and concluded RANS produced questionable results when the Reynolds number is in the transitional regime. This paper covers the first phase of a larger study which hopes to determine whether it is possible to model the cross- ventilation flow s

21、imply by knowing the faade pressures of a building. Faade pressures are readily available to the designers of large complex buildings due to the wind tunnel testing requirements for structural and facades systems. If one could use faade pressure as inlet boundary conditions for a CFD model and still

22、 achieve moderate accuracy, this approach could be instrumental for designers with much lower computational costs and data availability. The drawback of using pressure based boundary conditions for CFD simulation is that the momentum properties carried by the outdoor flow/eddies are lost, as discuss

23、ed in Katos work in 1992. However the above mentioned studies are for large openings when compared to the size of the building. For designers who are interested in cross-ventilation with multiple smaller openings such as partly open sliding windows, the effect of eddies could be small and a pressure

24、 based boundary condition could be an ef fective mean s for estimating the ventilation flow. Experimental outdoor and indoor airflow data, including ventilation flow rate and faade pressures were collected at a full scale test house located in Austin, TX. The faade pressures then were used as bounda

25、ry conditions for a two opening CFD model identical to the test house. The resulting flow rate from CFD was then compared to the experimental flow rate. The experimental and CFD modeling methodology are explained below. For this first phase, this paper only addresses the specific two openings (shown

26、 in Figure 1) scenario meant to be a preliminary proof of concept for the later analysis. Further studies are planned to be conducted in addition to the results discussed in this paper. EXPERIMENTAL SETUP A 13m x 8m x 3m test house was used for the full scale cross-ventilation experiment. For the pu

27、rpose of this study, a configuration of two openings was used. With the prevailing south wind, the south facing window was selected as the inlet opening while the furthest Northwest window was used as the outlet. Figure 1 shows the orientation and locations of the openings as well as instrumentation

28、s. Instrumentation The most important data for this study are the ventilation flow rates through the windows. These data were captured by using two pressure based flow meters mounted on the inlet and outlet windows. Each of the air volume flow meters consists of 6 interconnected pitot tubes that mea

29、sure dynamic pressures at multiple locations at each of the window openings. Additional experiments with a blower door were conducted to calibrate the flow meter by converting the averaged dynamic pressures to the flow rate. Using these flow meters, flow data were collected at 5 Hz, and the pressure

30、 transducer used has a measurement uncertainty of 1% reading or 0.15 Pa. In addition to the flow rate measurement, a CO 2decay test was also 622 ASHRAE Transactionsconducted simultaneously to determine the overall air exchange rate during the data collection period. Seven CO 2sensors (one in each ro

31、om) were used inside the test house and air exchange rates for each room were determined. The air exchange rate of the house was then determined by using a volume weighted average of each room. To gather the faade pressure data around the test house, pressure taps were installed in all four walls an

32、d connected to identical pressure sensors used for the instantaneous flow measurements. These faade pressures are also recorded at 5 Hz so both the flow and pressure data can be correlated. There were a total of 10 faade pressure taps with four on the south facing wall, three on the east wall, two o

33、n the west wall, and one on the south wall. The locations of the faade pressure taps are also noted in Figure 1. In addition to the faade pressure, static pressure inside of the house was also recorded and can be used to estimate infiltration flow not measured at the two flow stations. Finally, to p

34、roperly characterize any internal flow driven by buoyancy, the surface temperature of the indoor walls were measured at the beginning of the each run using an infrared thermometer with measurement error of 1C. Due to the thermal property of the wall materials, it was assumed the internal walls retai

35、ned their temperature during a test (with test duration of roughly 20 minutes). Figure 1 Schematics of the experimental setup Data collection Each of the experiments was conducted when the local weather satisfied two criteria. First, the prevailing wind must be significant enough for the pressure ga

36、uge to measure accurately. An average wind speed of 2 m/s was selected as the cut off point. Secondly, the indoor and outdoor air temperature must be within the 1C (the error of the air temperature instrument) before a run. Prior to the run, the test house was sealed and CO 2tracer gas was injected

37、at seven locations (one in each room) to 2011 ASHRAE 623promote proper mixing. Many fans were used to enhance the mixing and when all CO2 sensors registered 3500 ppm, the injection and mixing stopped. All data recording devices were turned on at that time and the inlet and outlet windows were opened

38、 to of its total size (inlet=71cmx36xm, outlet=86cmx25xm), signaling start of a run. Data collection was terminated when the CO 2concentration reduced to a level of background. Based on the wind intensity, direction and the corresponding flow rate, each of the experiments lasted 15 to 25 minutes. CF

39、D MODEL The CFD simulation used Reynolds Averaged Navier-Stokes Equations (RANS) with Re-Normalisation Group (RNG) k- turbulence model (Chen 2009). The values of k and used at the inlet openings were determined by experimentally obtained wind characteristics and the length scales of the test house.

40、A coarse mesh (20 cm boundary cell size 15000 cells) and a finer mesh (10 cm boundary cell size and 150000 cells) were used to verify grid independency using calculated transient ventilation rates. The comparison yields very little difference ( 2%) in flow rates between the coarser and the finer mes

41、h. The reason why such a coarse mesh can yield comparable results can be explained by the act of using pressure boundary conditions, which did not preserve the momentum characteristics of the inflow wind. One would expect if velocity and direction boundary conditions were used, a more refined grid w

42、ould be required to capture this information and the grid independency limit would be raised significantly. However, since the internal flow distribution was not of interest for this study, the grid independency comparison is valid. The coarse mesh was used for the remainder of the study for the ben

43、efit of lesser computation cost. Boundary Conditions The key boundary condition in this study is the fluctuating pressure at inlet and outlet ca used by the external wind driven flow. The window faade pressures measured closest to the openings from the experiment were fed into the CFD model as the c

44、hanging boundary conditions at the openings. To ensure good results during the transient simulations, the very first recorded time step pressure boundary conditions were used in a steady state simulation in order to establish the correct internal flow field. By doing so, the subsequent time steps du

45、ring the transient simulation had better starting points and required much less time to reach convergence. The exact time steps and iterations used are parts of the sensitivity analysis, discussed in the next section. To account for buoyancy driven indoor flow caused by the temperature difference be

46、tween the air and the internal wall surfaces, thermal boundary conditions based on experimentally obtained internal surface temperatures, were used in the model. The exact effect of the internal buoyancy driven flow is also a part of the sensitivity analysis for the CFD model, discussed in the next

47、section. An additional factor to the temperature difference that can affect the ventilation rate is the effect of infiltration. The test house used here is a real building, not a tightly sealed box inside a chamber. In this study, the infiltration effect in the test house was captured by using a sta

48、ndard blower door test at 50 Pa and 25 Pa then interpolated back to a given static pressure difference between inside and out side of the test house, for every time step measured. The estimated infiltration flow rates were then used to adjust the experimentally obtained ventilation flow rates, measu

49、red at the inlet and outlet of the test house. While the measures were taken to account for infiltration, the significant difference between the infiltration flow path and the CFD simulated flow could introduce another source of uncertainty into the results comparison. Parametric Analysis on Sensitivity To determine whether the approach stated in this study can be extrapolated to everyday scenarios, the CFD models sensitivity to several criteria must be understood. Three specific factors listed below were examined: 1. Transient simulation time steps: It is impo