Installation Testing

[Bowenaro]

Is there a file that contains all of the expected results for installation testing? How were the data points for each test chosen to determine a pass/fail? To give some context, I am working with a windows 10 OS and do not have access to python at the moment, so I am looking for other avenues to verify the install was done correctly for Software Quality Assurance purposes. Thanks!

[drjfloyd]

See App B of the Verification Guide.

[Bowenaro]

Thanks for your reply!
I used App. B to help me determine which models to test. Using ashrae_7_exp.csv. as an example. Do I use the values from every time interval to determine pass/fail? If all of the FDS output values are within tolerance except one value in the set, it is outside the error tolerance, then the whole test fails?

[drjfloyd]

[drjfloyd]

End of simulation.

[mcgratta]

Open this file with Excel or similar. Find the row coinciding with the name of your case (it should be in alphabetic order). Details of the comparison are found in the row. Towards the end of the row, you will find the words "Absolute Error" or "Relative Error", along with a value, like 0.01, say, and another word like "end" or "mean". Somewhere earlier in the row you will find which files, and the columns within these files, that are being compared. There are other values nearby that indicate the range over which the comparison is made. Typically, each row corresponds to one comparison for each column of data being compared. This wiki explains the meaning of the columns in the dataplot input file.

[Bowenaro]

Thanks for the help! Just out of curiosity, how do points get chosen as reference points that dictate pass/fail criteria? Is it random between every test case?

[drjfloyd]

Verification cases are designed to test specific aspects of FDS. Part of that design process is determining what outputs to use for comparison and what the values are to compare against.

[mcgratta]

I would not use the word "random," but rather "case by case." For some cases, we expect FDS to converge to an analytical solution with decreasing spatial and time steps. In such cases, we might pick a single point, let's say the surface temperature of a solid, compute the exact steady-state temperature, and then run progressive FDS simulations demonstrating the order of convergence. In such a case, the specific point of comparison is not that important, but rather the demonstration of convergence. However, in other cases, the point chosen might be the one that challenges the algorithm the most, or is of greatest consequence vis a vis fire safety.