1、 Shaping Tomorrows Built Environment Today 2012 ASHRAE www.ashrae.org. This material may not be copied nor distributed in either paper or digital form without ASHRAEs permission. Requests for this report should be directed to the ASHRAE Manager of Research and Technical Services.How to Verify, Valid
2、ate, and Report Indoor Environment Modeling CFD Analyses ASHRAE RP-1133 Sponsored by ASHRAE TC 4.10, Indoor Environmental Modeling Final Report Submitted to ASHRAE 1791 Tullie Circle, NE Atlanta, GA 30329-2305 Qingyan (Yan) Chen, Ph.D.; Principal Investigator This work was completed when Chen was on
3、 leave at Welsh School of Architecture, Cardiff University Bute Building, King Edward VII Avenue Cardiff CF10 3NB, Wales, UK Jelena Srebric, Ph.D.; Assistant Professor Department of Architectural Engineering The Pennsylvania State University 222 Engineering Unit A University Park, PA 16802-1417 June
4、 29, 2001 ASHRAE 1133-RP Chen 1 TABLE OF CONTENTS SUMARY 21. INTRODUCTION 3 1.1 Background 3 1.2 Objective 4 2. MANUAL FOR CFD ANALYSIS OF INDOOR ENVIRONMENT MODELLING 5 2.1 Verification 7 2.1.1 Basic flow and heat transfer 8 2.1.2 Turbulence models 9 2.1.3 Auxiliary heat transfer and flow models 13
5、 2.1.4 Numerical methods 14 2.1.5 Assessing CFD predictions 16 2.2 Validation 17 2.2.1 Validation procedure 18 2.2.2 Validation criteria 19 2.3 Reporting of CFD Results 20 2.3.1 Experimental design 20 2.3.2 Turbulence models and auxiliary heat transfer and flow models 21 2.3.3 Boundary conditions 21
6、 2.3.4 Numerical methods 22 2.3.5 Assessing CFD predictions 23 2.3.6 Drawing conclusions 24 3. ILLUSTRATION OF APPLYING THE PROCESS DESCRIBED IN THE MANUAL 25 3.1 Example 1: An Office with Mechanical Ventilation 25 3.1.1 Verification 26 3.1.2 Validation and reporting of results 30 3.2 Example 2: An
7、Apartment Building with Natural Ventilation 38 3.2.1 Verification 38 3.2.2 Validation and reporting of results 45 4. CONCLUSIONS 51 REFERENCES 53 ASHRAE 1133-RP Chen 2 SUMMARY Computational fluid dynamics (CFD) has been used to help determine the fluid flow, heat transfer, and chemical species trans
8、port in the analysis of indoor environmental conditions as well as a wide range of other HVAC and (2) within each time step for transient physical phenomena. Criterion can be set to determine if a converged solution is reached, such as a specified absolute and relative residual tolerance. The residu
9、al is the unbalance of those variables solved, such as velocities, mass flow, energy, turbulence quantities, and species concentrations. For indoor environment modeling, a CFD solution has converged if: Residual for mass = The sum of the absolute residuals in each cell / the total mass inflow 0.1% R
10、esidual for energy = The sum of the absolute residuals in each cell / the total heat gains 1% Similar convergence criterion can be defined for other solved variables, such as species concentration and turbulence parameters. Note that for natural convection in a room, the net mass flow is zero. There
11、fore, one can conclude that a convergence has been reached if there is little change (no change in the 4thdigit) on the major dependent variables (temperature, velocities, and concentrations) within the last 100 iterations. However, a small relaxation factor can always give a false indication of con
12、vergence (Anderson et al. 1984). In order to obtain stable and converged results, the iteration procedure often uses relaxation factors for different variable solved, such as under-relaxation factors and false-time-steps. The under-relaxation factors differs a little bit from the false-time-steps, b
13、ut there is no substantial differences. 2.1.5 Assessing CFD predictions This section should provide in detail both qualitative and quantitative comparison of CFD results with data from experiment, analytical solutions, and direct numerical simulations. All the error analyses should be detailed in th
14、is section as well. The results present in this section should serve as a basis to judge whether the CFD code can be used for indoor environment modeling. Although this manual divides the verification into several parts, they are integrated in many cases. The turbulence model and numerical technique
15、 must work together in order to obtain a correct CFD prediction for the flow features being selected. However, it is necessary to break them down into individual items in some types of verifications, such as in the CFD code developments. Indoor environment designers often use commercial software. It
16、 is logical to assume that the codes have been verified during the code development. However, the verifications, if they are performed at all, may have used different flows that are irrelevant to indoor airflow. In addition, a user may not fully understand the functions of the CFD code. It is impera
17、tive for the user to “re-verify” the capabilities of a CFD code for indoor environment simulations. This will help the user become more familiar with the CFD code and eliminate human errors in using the code. ASHRAE 1133-RP Chen 17 In general, the cases used for verification are not company-propriet
18、ary or restricted for security reasons. These data are usually available from the literature. It is strongly recommended to report the verification. This is especially helpful in eliminating errors caused by the users, since most CFD codes may have been validated by those cases. There are many examp
19、les of failed CFD simulations due to the users mistakes. The verification should be done for the following parameters: G140G32 All the variables solved by the governing equations, such as velocity, temperature, species concentrations G140G32 Boundary conditions such as heat flux and mass inflow and
20、outflow rates With the verification described above, a CFD code should be able to correctly compute the airflow and heat transfer encountered in an indoor environment. The level of the accuracy depends on the criteria used in the verification. If the CFD code failed to compute correctly the flow, th
21、e problem may be: (1) the CFD code is not capable to solve the indoor airflows, (2) the CFD code has bugs, or (3) there are errors in the user input data that defines the problem to be solved. 2.2 Validation Validation is the demonstration of the coupled ability of the user and the CFD code to accur
22、ately predict representative indoor environmental applications for which some sort of reliable data is available. The validation estimates how accurately the user can apply the CFD code in simulating a full indoor environment problem in the real world. It gives the user the confidence to use the CFD
23、 code for further applications, such as a design tool. A CFD code may have solved the physical models that the user selects to describe the real world, however, the results may not be accurate because the selected models do not represent the physical reality. For example, an indoor environment may i
24、nvolve simultaneously conduction, convection, and radiation. A CFD user may misinterpret the problem as purely convection. The CFD prediction may be correct for the convection part, but fails in describing the complete physics involved in the case. It is obviously a problem on the users side, which
25、the validation process is also trying to eliminate. The fundamental strategy of validation is to identify suitable experimental data, to make sure that all the important physical phenomena in the problem of interest are correctly modeled, and to quantify the error and uncertainty in the CFD simulati
26、on. Since the primary role of CFD in indoor environment modeling is to serve as a high-fidelity tool for design and analysis, it is essential to have a systematic, rational, and affordable code validation process. Validation is focused on the G140G32 Confirmation of the capabilities of the turbulenc
27、e model and other auxiliary models in predicting all the important physical phenomena associated with an indoor environment, before applying the CFD model for design and evaluation of a similar indoor environment category ASHRAE 1133-RP Chen 18 G140G32 Confirmation of correctness of the discretizati
28、on method, grid resolution, and numerical algorithm for the flow simulation G140G32 Confirmation of the users knowledge on the CFD code and his/her understanding to the basic physics involved in the indoor environment analysis 2.2.1 Validation procedure Ideally, validation should be performed for a
29、complete indoor environment system that includes all the important airflow and heat transfer physics and a full geometric configuration. Experimental data for a complete system can be obtained from on-site measurements and the experiments in an environmental chamber. The data usually have a fairly h
30、igh degree of uncertainty and large errors. The data may contain little information about the initial and boundary conditions. Reasonable assumptions are needed to make a CFD simulation feasible. The definition of validation used in this manual sounds very similar to that of verification. There are
31、substantial differences. The validation is for a complete flow and heat transfer system or several subsystems that can altogether represent a complete system. However, verification is only for one of the flow aspects found in an indoor environment. The validation procedure is almost same between the
32、 verification and validation. The validation procedure involves: G140G32 The complete indoor environment system design for validation G140G32 Turbulence models G140G32 Auxiliary heat transfer and flow models G140G32 Numerical methods G140G32 Assessing CFD predictions G140G32 Drawing conclusions The
33、procedure will be the same as for verification, and, therefore, will not be repeated here. However, the validation requires analyzing the CFD results and experimental data in order to draw some conclusions for indoor environment analyses. Very often, experimental data may not be available for a comp
34、lete indoor environment system. It is acceptable to utilize validations for several subsystems or a less-than-complete system. A subsystem of indoor environment represents some of the flow features in an indoor environment to be analyzed. The overall effect of several subsystems is equivalent to a c
35、omplete system. For example, a complete indoor environment system consists of airflow and heat transfer in a room with occupants, furniture, and a forced air unit. If a user can correctly simulate several subsystems such as (1) airflow and heat transfer around a person, (2) airflow and heat transfer
36、 in a room with obstacles, and (3) airflow and heat transfer in a room with a forced air unit, the validation is acceptable. In the same example, a less-than-complete system for this environment can consist of airflow and heat transfer in a room with an occupant and a forced air unit. The furniture,
37、 although it affects the indoor environment, is not as important as the other components. Therefore, the validation with a less-than-complete system is acceptable. In either case, the key is that the validation should lead to a solid confirmation of the combined capabilities of the CFD user and code
38、. ASHRAE 1133-RP Chen 19 Although the validation is for a complete indoor environment system, it is not necessary to start with a very complicated case if the user has a number of alternatives. Reliability is better: G140G32 For a simple geometry, rather than a complicated one G140G32 For convection
39、, rather than combined convection, conduction, and radiation G140G32 For single-phase flows, rather than multi-phase flows G140G32 For chemically-inert materials, rather than chemically-reactive materials This is because, for complex physical phenomena in an indoor space, the input data for CFD anal
40、ysis may involve too much guesswork or imprecision. The available computer power may not be sufficient for high numerical accuracy, and the scientific knowledge base may be inadequate. The validation of a complete system should be broken down into several steps. The first step will be the setup of t
41、he building geometry, followed by placement of the inlets and outlets. The isothermal flow will give an indication of the airflow pattern. The second step will be adding heat transfer. Species concentration, particle trajectory, and others should be considered later. This progressive simulation proc
42、edure will not only build confidence in the user at performing the simulation, but will also discover some potential errors in the simulation. On the other hand, simple and popular models in a CFD code should be considered as the starting point for validation if a CFD code has multiple choices. The
43、starting point can be as basic as: G140G32 Standard k- model G140G32 No auxiliary flow and heat transfer models G140G32 Structured mesh system G140G32 Upwind scheme G140G32 SIMPLE algorithm The way of measuring the accuracy of the representation of the real world is to systematically compare CFD sim
44、ulations to the experimental data. The indoor environment systems used in validation are usually complicated, and the corresponding experimental data may contain biased errors and random errors. These errors should be reported as part of the validation. In case the errors are unknown, a report on th
45、e equipment used in the measurements will be helpful in assessing the quality of the data. Although desirable, it is expensive and time-consuming to obtain good quality data for a complete system. Therefore, reporting the CFD validation of the complete system cannot be overemphasized. 2.2.2 Validati
46、on criteria The criteria for accuracy when conducting a validation depend on the application. Very high accuracy, while desirable, is not essential since most design changes are incremental variations from a baseline. As long as the trends that are predicted are consistent, then less-than-perfect ac
47、curacy should be acceptable. The validation process should be flexible, allowing a varying ASHRAE 1133-RP Chen 20 level of accuracy, and be tolerant of incremental improvements as time and funding permit. The level of agreement achieved with the test data, taking into the account measurement uncerta
48、inties, should be reviewed in light of the CFD application requirements. For example, validation for modeling air temperature in a fire simulation requires a much lower accuracy than that for thermal comfort study for an indoor environment. If the validation cases are simple and represent a subsyste
49、m of a complex indoor airflow, the validation criteria should be more restrictive than those for the complete system. The criteria can also be selective. For example, if correct prediction of air velocity is more important, the criteria for heat transfer may be relaxed. Although the air velocity and temperature are inter-related, the impact of one parameter over the other may be of second order. This would allow the CFD user to use a fast and less detailed model, such as standard k- model, rat
