FDS input file crashing

[2000-lang]

Below is my FDS input file. No matter how much I change the mesh combinations its crashing in the HPC. Also, please suggest me the best mesh division for my geometry and also I uploaded my SLURM script please suggest me changes if required.

&HEAD CHID='street_canyon_treefire2',
TITLE='Street canyon with three burning particle trees2' /

!------------------ MESHES ------------------
&MESH IJK=126,63,45, XB=0.0,252.0, 0.0,126.0, 0.0,90.0, MULT_ID='COARSE_MESH' /

&MULT ID='COARSE_MESH',
DX=252.0, DY=126.0, DZ=90.0,
I_LOWER=0, I_UPPER=1,
J_LOWER=0, J_UPPER=1,
K_LOWER=0, K_UPPER=0 /

! Fine core (high resolution around canyon and trees)
&MESH ID='FINE_MESH',
XB = 88.0,148.0,
16.0,110.0,
0.0,20.0,
IJK = 60,94,20,
CHECK_MESH_ALIGNMENT = .TRUE. /

&MISC TMPA=293.0 /

&SURF ID='INERT' /

&TIME T_BEGIN=0., T_END=15.0 /

!----------------- WIND AND OPEN BOUNDARIES -----------------
&WIND SPEED=6., Z_0=0.01, L=-100., Z_REF=2., DIRECTION=180.0 /

&RADI RADIATION=F /

&VENT PBX=0.0, SURF_ID='OPEN' /
&VENT PBX=252.0, SURF_ID='OPEN' /
&VENT PBY=0.0, SURF_ID='OPEN' /
&VENT PBY=126.0, SURF_ID='OPEN' /
&VENT PBZ=90.0, SURF_ID='OPEN' /

!----------------- SPECIES (from single_tree_fire) -----------------
&SPEC ID='WOOD', FORMULA='CH1.7O0.74N0.002' /
&SPEC ID='OXYGEN', LUMPED_COMPONENT_ONLY=T /
&SPEC ID='NITROGEN', LUMPED_COMPONENT_ONLY=T /
&SPEC ID='WATER VAPOR', LUMPED_COMPONENT_ONLY=T /
&SPEC ID='CARBON MONOXIDE', LUMPED_COMPONENT_ONLY=T /
&SPEC ID='CARBON DIOXIDE', LUMPED_COMPONENT_ONLY=T /
&SPEC ID='SOOT', LUMPED_COMPONENT_ONLY=T /

&SPEC ID='AIR',
SPEC_ID(1)='OXYGEN', VOLUME_FRACTION(1)=1.,
SPEC_ID(2)='NITROGEN', VOLUME_FRACTION(2)=3.76,
BACKGROUND=T /

&SPEC ID='PRODUCTS_2',
SPEC_ID(1)='CARBON DIOXIDE', VOLUME_FRACTION(1)=0.96438,
SPEC_ID(2)='CARBON MONOXIDE',VOLUME_FRACTION(2)=0.00365,
SPEC_ID(3)='WATER VAPOR', VOLUME_FRACTION(3)=0.85,
SPEC_ID(4)='SOOT', VOLUME_FRACTION(4)=0.03196,
SPEC_ID(5)='NITROGEN', VOLUME_FRACTION(5)=3.84076 /

&REAC ID='WOOD_FIRE',
FUEL='WOOD',
HEAT_OF_COMBUSTION=16400.,
SPEC_ID_NU='WOOD','AIR','PRODUCTS_2',
NU=-1.,-1.02121,1. /

!----------------- MATERIALS (same as single_tree_fire) -----------------
&MATL ID='MOISTURE',
DENSITY=1000.,
CONDUCTIVITY=0.6,
SPECIFIC_HEAT=4.19 /

&MATL ID='DRY_PINE',
DENSITY=500.,
CONDUCTIVITY=0.1,
SPECIFIC_HEAT=1.0 /

! Trunk burning surface
&SURF ID='TRUNK_WOOD',
MATL_ID(1,1)='DRY_PINE',
MATL_MASS_FRACTION(1,1)=1.0,
THICKNESS(1)=0.20,
HRRPUA=150.0,
SPEC_ID='WOOD',
COLOR='BROWN' /

! Crown needle surface (particle fuel, identical to single_tree_fire)
&SURF ID='needle surface',
MATL_ID(1,1)='DRY_PINE',
MATL_MASS_FRACTION(1,1)=1.0,
MOISTURE_FRACTION=0.25,
SURFACE_VOLUME_RATIO=8000.,
LENGTH=0.10,
GEOMETRY='CYLINDRICAL',
HRRPUA=150.0,
SPEC_ID='WOOD',
BURN_AWAY=.TRUE.,
COLOR='FOREST GREEN' /

&PART ID='needles',
SURF_ID='needle surface',
DRAG_COEFFICIENT=2.8,
STATIC=.TRUE.,
COLOR='FOREST GREEN' /

&PROP ID='needle image',
SMOKEVIEW_ID='TUBE',
SMOKEVIEW_PARAMETERS='L=0.1,D=0.0005' /

!----------------- BUILDINGS -----------------
&OBST XB=90.0,108.0, 18.0,108.0, 0.0,18.0, SURF_ID='INERT' /
&OBST XB=126.0,144.0, 18.0,108.0, 0.0,18.0, SURF_ID='INERT' /

!----------------- TREE TRUNKS (keep your original 0–4 m) -----------------
&OBST ID='TREE1_TRUNK',
XB=118.5,119.5, 39.5,40.5, 0.0,4.0,
SURF_ID='TRUNK_WOOD' /

&OBST ID='TREE2_TRUNK',
XB=118.5,119.5, 63.5,64.5, 0.0,4.0,
SURF_ID='TRUNK_WOOD' /

&OBST ID='TREE3_TRUNK',
XB=118.5,119.5, 87.5,88.5, 0.0,4.0,
SURF_ID='TRUNK_WOOD' /

!----------------- TREE CROWNS (your 4–8 m cones) -----------------
&INIT ID='TREE1_CROWN',
PART_ID='needles',
SHAPE='CONE',
XYZ=119.0,40.0,0.0,
CROWN_BASE_WIDTH=3.0,
CROWN_BASE_HEIGHT=4.0,
TREE_HEIGHT=8.0,
N_PARTICLES_PER_CELL=3,
MASS_PER_VOLUME=8.0,
DRY=.TRUE. /

&INIT ID='TREE2_CROWN',
PART_ID='needles',
SHAPE='CONE',
XYZ=119.0,64.0,0.0,
CROWN_BASE_WIDTH=3.0,
CROWN_BASE_HEIGHT=4.0,
TREE_HEIGHT=8.0,
N_PARTICLES_PER_CELL=3,
MASS_PER_VOLUME=8.0,
DRY=.TRUE. /

&INIT ID='TREE3_CROWN',
PART_ID='needles',
SHAPE='CONE',
XYZ=119.0,88.0,0.0,
CROWN_BASE_WIDTH=3.0,
CROWN_BASE_HEIGHT=4.0,
TREE_HEIGHT=8.0,
N_PARTICLES_PER_CELL=3,
MASS_PER_VOLUME=8.0,
DRY=.TRUE. /

!----------------- OUTPUT -----------------
&DUMP DT_HRR=5.0, DT_SMOKE3D=15.0,
DT_SL3D=5.0, DT_PL3D=5.0 /

&SLCF QUANTITY='TEMPERATURE' /

&TAIL /

#!/bin/bash
#SBATCH --job-name=simple_test
#SBATCH --output=simple_test.out
#SBATCH --error=simple_test.err
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=8
#SBATCH --time=24:00:00
#SBATCH --partition=long

module purge

module load tbb

module load compiler-rt/2024.2.1

module load compiler-intel-llvm/2024.2.1

module load mpi/2021.13

srun ./fds_impi_intel_linux practise.fds > fds_output.log

[drjfloyd]

You have TMPA=293 degrees C; I suspect you mean TMPA=20 C.

Your species inputs are more complicated than they need to be. You don't need any of those SPEC inputs. Your fuel only has C,H,O, and N, so you could just use:

&REAC FUEL='WOOD',CH1.7,O=0.74,N=0.002,HEAT_OF_COMBUSTION=16400,SOOT_YIELD=your soot yield, CO_YIELD=your CO yield.

FDS will compute the product species.

Don't redefine the INERT surface (delete &SURF ID='INERT' /)

Look at your outputs with Smokeview (this is something you should always be doing every time you make a new input). You have solid walls for two of your boundaries. You have not correctly computed the max X and Y boundaries. I suspect this is the cause of your problems as this will prevent the wind field from forming correctly.

Do not overlap meshes like this. Meshes only share information at their boundary. Your fine mesh will pull data from the coarse at its boundaries; however, the coarse mesh will not see anything that is happening inside the fine mesh.

Your fine mesh should be larger. You are not providing any room to resolve the flow at higher detail as it moves over the buildings.

[2000-lang]

Thank you for the valuable inputs. I wanted your way of defining the mesh for my requirement. I am looking to have a fine mesh (0.2 m) resolution surrounding the buildings and 2m for rest of the area.

Please write down the best way of defining this type of mesh.

[drjfloyd]

There is basic guidance on meshing in 6.3 of the User's Guide.

[2000-lang]

If I follow as per user guide, I am getting around 33million cells. Is that practical ? and also I have uploaded my slurm script above, do you think I have to change anything?

[drjfloyd]

Pratical is going to depend on your hardware and time constraints. 33 million cells is possible with enough memory but you are going to want to have a large number of meshes if you want it to run in a reasonable period of time.

The syntax of the SBATCH portions of your slurm script look ok; however, you want to make sure the number of cpus you request is the same as the number of meshes. I suggest you make a simple test case with multiple meshes and test your script. The non-SBATCH portion is going to have some dependence on your specific system. Right now you are saying that the executable for fds is in the same folder as your input. Would not suggest this. Have your input file someplace other than where you installed fds, run the script from where the input file is, and in the script give the path to where fds is.