Current release of CFAST has verification cases that are out of tolerance

[mcgratta]

The current CFAST Validation lists verification cases that do not meet the tolerance

It appears that cfastbot is not flagging these errors. Are you familiar with these cases and can you guess why they are out of tolerance.

[mcgratta]

Below are plots from the verification case radiation_3.in. It is a vertical shaft filled with hot water vapor. There is a profile of heat flux to the ceiling (left) and side wall (right):

I would expect that the side wall CFAST result would at least be symmetric. The ceiling profile (left) should have some decrease near the corners to account for the increased influence of the cold walls.

[drjfloyd]

Does the target model in CFAST actually compute view factors to walls or is it just using the bulk layer emission plus point source radiation from a fire?

[mcgratta]

I believe that it computes view factors, and uses the distance from target to wall centroid for absorption calc. There are assumed to be 10 wall surfaces, floor, ceiling, upper/lower of each wall face. I'm tracking the printout backwards to see where there is an asymmetry. Right away, I see that even though we've chosen the "shaft" line layer option, CFAST is assuming a short lower layer. That might explain the asymmetry.