Sanity Checks for AEC Simulation Results
Before drawing final conclusions from any results in Simulation CFD, they must first be validated. Even if physical test results are not available for benchmarking, the following checklist of validation considerations will provide increased confidence in the solution.
A common issue with simulations, even veteran users, when in a hurry, can miss a decimal point, put in wrong units, or make a mistake in assigning materials. Input items to check include:
- CAD Geometry
- Material assignments
- Boundary conditions
- Solver settings
In the Simulation CFD Setup tab, the Design Study Bar is a convenient means of quickly scanning all of the materials and boundary conditions.
Another technique of validating correct inputs is by checking the results to make sure the solver has interpreted your intentions correctly. Here are just a few examples:
- A small AEC space has a total of 160 Watts of heat generation assigned to both fluid and solid volumes. After solving, if lines 4 and 7 of the Energy Balance section of the Summary File do not add up to 160, then there are missing or incorrect heat loads assigned.
- External flow simulated for a building in Colorado at an altitude of 5,000 feet and ambient temperature of 32F should result in a density of 1.073 Kg/m^3 according to published air tables. In Results, set the Global Results to Density to verify that the air density is correct.
- Flow rates are specified for an HVAC system at 100 CFM at each duct inlet to a large factory. Add a Plane, orient it to normal with the flow path in the supplies and use the Bulk Calculator to verify the assigned flow rate.
TIP: The Summary File is very useful in spotting potential issues since it provides a very concise listing of the operating conditions of the model.
A poor quality mesh can provide misleading results. If there are any doubts about the mesh, clone the scenario, add mesh refinement and run again to verify that the results do not change appreciably. More information about meshing can be found here.
During the early stages of convergence, the results change substantially while the solution works itself out. Interpreting results too early and terminating the solution prematurely can lead to incorrect conclusions. Check the convergence monitor in the Output bar to verify that the results are not changing more than 3% over the last 10-20% of the iteration steps. If there is any doubt, restart the simulation and run for another 100 or 200 steps and check again. Simulations with poor convergence behavior as seen here usually indicate poor mesh quality, improper boundary conditions, or less than ideal solver settings.
Although closed form solutions are typically limited by underlying assumptions, they are still a very useful means of validating overall simulation results. It is highly recommended to have fluid flow and heat transfer references available, such as Fluid Mechanics by Frank White, as a benchmark.
At times, correct simulation results may not appear intuitive at first. Even after going through the previous check boxes, the results cannot yet be fully explained. In these cases, separate test models may prove beneficial. For example, a “sub-model,” which isolates the area in questions, can be quickly simulated to confirm the results.