STRATEGIES FOR COUPLING ENERGY SIMULATION AND COMPUTATIONAL FLUID DYNAMICS PROGRAMS Zhiqiang Zhai and Qingyan Chen Massachusetts Institute of Technology Cambridge, MA 02139, USA Joseph H. Klems and Philip Haves Lawrence Berkeley National Laboratory Berkeley, CA 94720, USA ABSTRACT Energy simulation (ES) and computational fluid dynamics (CFD) can play important roles in building design by providing complementary information about the buildings’ environmental performance. However, separate applications of ES and CFD are usually unable to give an accurate prediction of building performance due to the assumptions involved in the separate calculations. Integration of ES and CFD eliminates many of these assumptions since the information provided by the models is complementary. Several different approaches to integrating ES and CFD are described. In order to bridge the discontinuities of time-scale, spatial resolution and computing speed between ES and CFD programs, a staged coupling strategy for different problems is proposed. The paper illustrates a typical dynamic coupling process by means of an example implemented using the EnergyPlus and MIT-CFD programs. Key words: energy simulation, computational fluid dynamics (CFD), integration, building design INTRODUCTION Energy simulation and computational fluid dynamics programs provide complementary information about the performance of buildings. ES programs, such as EnergyPlus (Crawley et al 2000), address the performance of the building envelope, as well as the heating, ventilating and air conditioning (HVAC) system, and provide the whole building energy analysis. Space-averaged indoor environmental conditions, cooling/heating loads, coil loads, and energy consumption can be obtained on an hourly or sub-hourly basis for periods of time ranging from a design day to a reference year or more. CFD programs, on the other hand, make detailed predictions of thermal comfort and indoor air quality (IAQ), including the distributions of air velocity, temperature, relative humidity and contaminant concentrations. The distributions can be used further to determine indices such as the predicted mean vote (PMV), the percentage of people dissatisfied (PPD) due to discomfort, the percentage dissatisfied (PD) due to draft, and ventilation effectiveness. With the information from both ES and CFD calculations, designers can design environmental control systems for buildings that satisfy multiple criteria. However, due to the complete mixing model used in ES, most ES programs cannot accurately predict energy for systems that produce non-uniform air temperature distributions in the occupied space, such as displacement ventilation systems. Moreover, the spatially averaged comfort information generated by the single node model of ES cannot satisfy advanced design requirements. The convective heat transfer coefficients used in ES programs are usually empirical and may not have general applicability, either. Furthermore, most ES programs are unable to provide information on the airflow entering a building, for example, by natural ventilation, while the ventilation rate information is very important for predicting room air temperature and (or) heating/ cooling load. CFD, on the other hand, can easily determine the temperature distribution and convective heat transfer coefficients, which ES needs. CFD is also a powerful tool for the simulation of natural ventilation driven by wind effect, stack effect, or both. At the same time, CFD also needs information from ES as inputs, such as air conditioning loads and surface temperatures. Otherwise, CFD has to compute results based on estimated boundary conditions. Therefore, coupling ES with CFD is very attractive and is the objective of the present investigation. Starting from the principles of ES and CFD, the paper describes possible approaches to ES and CFD coupling. The current study emphasizes the explicit coupling of individual ES and CFD programs by exchanging the inter-coupled boundary values.
Displacement ventilation may provide better indoor air quality than mixing ventilation. Proper design of displacement ventilation requires information concerning the air temperature difference between the head and foot level of a sedentary person and the ventilation effectiveness at the breathing level. This paper presents models to predict the air temperature difference and the ventilation effectiveness, based on a database of 56 cases with displacement ventilation. The database was generated by using a validated CFD program and covers four different types of US buildings: small offices, large offices with partitions, classrooms, and industrial workshops under different thermal and flow boundary conditions. Both the maximum cooling load that can be removed by displacement ventilation and the ventilation effectiveness are shown to depend on the heat source type and ventilation rate in a room.
Numerous steady Reynolds-averaged Navier-Stokes (SRANS) two-equation turbulence models have been applied to modeling urban airflow and pollutant dispersion. Their low accuracy has been attributed to SRANS ill-conditioning, the linear eddy viscosity hypothesis, and uncertainty contributed by empirical formulations and coefficients. Many studies have attempted to modify the two-equation models specifically for urban problems by correcting the model formulations and calibrating the coefficients. However, the models are not universally applicable to a variety of urban problems. To improve the generalizability of the models, this study introduced multiple correctors to the empirical formulations and coefficients of the k-ε model. The inherent shortcoming of SRANS ill-conditioning was managed by increasing the model’s flexibility in accommodating the model's flaws and prioritizing critical parameters. In particular, the transferability and robustness of the corrected model were considered by retaining the linear eddy viscosity closure and employing symbolic formulations. This investigation designed an integrated multiple expression programming framework to support simultaneous symbolic regression and coefficient calibration for corrector training. The generalizability of the corrected model was evaluated for airflow and dispersion around a single building, a building array, and a group of complex buildings. The corrected model performed consistently for the three flow types and exhibited generalizability.
The Smagorinsky subgrid-scale model, a dynamic subgrid-scale model, and a stimulated subgrid-scale model have been used in a large eddy simulation (LES) program to compute airflow in a room. A fast Fourier transformation (FFT) method and a conventional iteration method were used in solving the Poisson equation. The predicted distributions of indoor air velocity, temperature, and contaminant concentrations show that the three subgrid-scale models can produce acceptable results for indoor environment design. The dynamic and stimulated models performed slightly better than the Smagorinsky model. The use of FFT can significantly reduce the computing time. LES is a tool of the next generation of indoor air distribution design.
The designers of ventilation systems need to predict the air flow patterns in order to optimize design and to ensure a healthy interior. Numerical simula tion is a powerful tool to obtain the air flow patterns. In the present study, a computer program which solves the three-dimensional conservation equa tions of mass, momentum, energy, and contaminant concentration is used. The program is based on the κ-ε turbulence model with wall function expres sions for solid boundaries. Flow fields are computed for two gymnasia, of 24 × 12× 9 m3 and 44 × 23 × 10 m3, with variations in the ventilation rate, the arrangement of inlet and outlet, heating system, and the number of occupants. The simulation gives the field results for air velocity, temperature, contami nant concentration, percentage of dissatisfied people due to draught, and pre dicted percentage of dissatisfied due to thermal comfort.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)
