Different simulation time giving different results
94 views
Skip to first unread message
Joel Lederfind
Jan 5, 2021, 4:23:51 PM1/5/21
to CFAST
Hello,
I have encountered an issue with a particular model I am trying to run.
In running an example problem (the problem statement is attached), I get different answers depending on the total simulation time that I set in the program. With the default setting of 3600s, I get an ULT of 443.00 C at 300 seconds. If I change the time to 400s, my ULT becomes 348.45 C at 300 seconds. Up to 200 seconds, the 2 models are nearly identical, then the ULTs begin to diverge as the fire hits its peak HRR (a graph of the 2 ULTs over the fire's whole duration is attached).
I wondered if it may have had something to do with the simulation time effecting the number of time steps, and thus the resolution; but after changing the default maximum time step several times I still get roughly the same situation (the ULTs may change by less than a degree). I have yet to encounter this issue on any other models that I have run, so I wonder if something in my setup or input file is making the model unstable. I cant seem to wrap my head around why changing the total simulation time would affect the temperatures. The input files are also attached.
Any help is appreciated.
Thanks,
Joel Lederfind
I have encountered an issue with a particular model I am trying to run.
In running an example problem (the problem statement is attached), I get different answers depending on the total simulation time that I set in the program. With the default setting of 3600s, I get an ULT of 443.00 C at 300 seconds. If I change the time to 400s, my ULT becomes 348.45 C at 300 seconds. Up to 200 seconds, the 2 models are nearly identical, then the ULTs begin to diverge as the fire hits its peak HRR (a graph of the 2 ULTs over the fire's whole duration is attached).
I wondered if it may have had something to do with the simulation time effecting the number of time steps, and thus the resolution; but after changing the default maximum time step several times I still get roughly the same situation (the ULTs may change by less than a degree). I have yet to encounter this issue on any other models that I have run, so I wonder if something in my setup or input file is making the model unstable. I cant seem to wrap my head around why changing the total simulation time would affect the temperatures. The input files are also attached.
Any help is appreciated.
Thanks,
Joel Lederfind
Glenn Forney
Jan 11, 2021, 3:34:17 PM1/11/21
to CFAST
that's a bug. the solver should be generating the same output for the time interval the two cases have in common. something is different, we'll have to take a close look to see what that is
Joel Lederfind
Jan 27, 2021, 2:28:25 PM1/27/21
to CFAST
Has there been any success in characterizing the bugs relation to the program calculating the ULT, or any limitations the bug may have?
Or possibly any workarounds that may have been developed to avoid the issue? Just wondering, as this issue is holding up V&V efforts.
Thanks,
Joel
Glenn Forney
Jan 27, 2021, 2:57:50 PM1/27/21
to cf...@googlegroups.com
yes its been fixed. we should have a bug fix release in a few days.
--
You received this message because you are subscribed to a topic in the Google Groups "CFAST" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/cfast/uOqrQKQOBV0/unsubscribe.
To unsubscribe from this group and all its topics, send an email to cfast+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cfast/32a48da4-14f8-4368-86e3-ee7b3ee4923dn%40googlegroups.com.
Glenn Forney
Richard Peacock
Jan 28, 2021, 12:42:45 PM1/28/21
to CFAST
Here's a link to an updated version of CFAST. It should resolve the issue with the simulation time effecting the results. Since the solution was to choose a conservative value for the thermal penetration depth used to initialize wall conduction, the results may be a bit different from either of your two previous results. In our validation tests, the results changed by and average of less than 1 %. Let me know how it works and we'll create an official release.
Joel Lederfind
Jan 28, 2021, 4:57:51 PM1/28/21
to CFAST
Gave it a test and the ULT temperatures are now consistent over different total simulation times, seems to be fixed to me.
Thanks!
Joel
Reply all
Reply to author
Forward
0 new messages