search

firemodels / fds

Fire Dynamics Simulator
https://pages.nist.gov/fds-smv/
Other
650 stars 618 forks source link

EVACUATION_ONLY=T for WIND case #5905

Closed rmcdermo closed 6 years ago

rmcdermo commented 6 years ago

The attached case fails because EVACUATION_ONLY(NM)=T for a given MESH, which does not ALLOCATE MEAN_FORCING_CELL per lines 2129 in read.f90 (READ_WIND). The code fails when MEAN_FORCING_CELL tries to initialize in init.f90

molly:alexi rmcdermo$ fds_db wind_example.fds 
 EVAC: Emesh 1 PF has 1 door flow fields
forrtl: severe (408): fort: (8): Attempt to fetch from allocatable variable MEAN_FORCING_CELL when it is not allocated

Image              PC                Routine            Line        Source             
fds_mpi_intel_osx  0000000112DD8446  Unknown               Unknown  Unknown
fds_mpi_intel_osx  0000000112778171  _init_mp_initiali         370  init.f90
fds_mpi_intel_osx  000000011292264E  _MAIN__                   163  main.f90
fds_mpi_intel_osx  000000010EA542FE  Unknown               Unknown  Unknown

wind_example.fds.txt

rmcdermo commented 6 years ago

@tkorhon1 please use this to generate a WIND / EVAC verification test. Thanks.

mcgratta commented 6 years ago

PR #5907 cleans up obvious bugs with EVAC and MEAN_FORCING. However, I still get the errors:

ERROR: Memory allocation failed for Evac ZZ in the routine MAIN (CHID: wind_example)

ERROR: Memory allocation failed for Evac RHO in the routine MAIN (CHID: wind_example)

ERROR: Memory allocation failed for Evac RSUM in the routine MAIN (CHID: wind_example)

ERROR: Memory allocation failed for Evac TMP in the routine MAIN (CHID: wind_example)

ERROR: Memory allocation failed for Evac UII in the routine MAIN (CHID: wind_example)
tkorhon1 commented 6 years ago

Well, I just did git remote update and also downloaded the wind_example. So, I'll try to see this also.

TimoK

tkorhon1 commented 6 years ago

Well, I did not get an error message, when using debug mode (mpi_intel_linux_64_db).

Fire Dynamics Simulator

Current Date : January 26, 2018 16:29:57 Version : FDS 6.6.0 Revision : FDS6.6.0-792-gf72ca19 Revision Date : Thu Jan 25 20:36:38 2018 -0500 Compiler : Intel ifort 17.0.0 Compilation Date : Jan 26, 2018 16:27:25

MPI Enabled; Number of MPI Processes: 9 OpenMP Enabled; Number of OpenMP Threads: 4

MPI version: 3.0 MPI library version: Open MPI v1.8.3, package: Open MPI tstopi@espvm5m030.ad.vt .fi Distribution, ident: 1.8.3, Sep 25, 2014

Job TITLE : Test of z-profile mean velocity ramp for one-way WRF couplin with FDS Job ID string : wind_example_new

Time Step: -49, Evacuation Initialization Time: -0.490 s Time Step: 1, Simulation Time: 0.54 s Time Step: 2, Simulation Time: 1.08 s Time Step: 3, Simulation Time: 1.62 s Time Step: 4, Simulation Time: 2.16 s Time Step: 5, Simulation Time: 2.70 s Time Step: 6, Simulation Time: 3.24 s Time Step: 7, Simulation Time: 3.78 s Time Step: 8, Simulation Time: 4.33 s Time Step: 9, Simulation Time: 4.87 s

And still running. I did compile and run 4 days ago and I got the same thing, i.e., no error messages. (I ran it for the total 60s = T_END)

Timo

mcgratta commented 6 years ago

My mistake. The case posted above works when I designate 9 MPI processes, but fails with the memory allocation error if I designate 8 MPI process. There are 8 fire meshes and 1 evac mesh. I assume that this is the way Evac cases are to be run. If not, open up the issue again.