Upright Sprinkler Modelling

[danish2016]

Dear FDS Users,

I am trying to model an upright sprinkler located 225 mm below the ceiling. To represent the upright configuration, I set the ORIENTATION to 1. To observe the spray behavior, I placed the sprinkler bulb slightly below the ceiling level.

From the particle visualization, the overall spray pattern appears generally reasonable. However, I observed that water particles at the center of the sprinkler are moving vertically upward. In reality, an upright sprinkler has a deflector plate above the orifice, which redirects the water radially and prevents the jet from moving upward.

My question is: How can the sprinkler deflector plate be modeled in FDS to properly redirect the spray and avoid the vertical upward flow at the center?

Please refer to the figure below for illustration.

Thank you.

sprinkler_new.fds
Generated by PyroSim 2025.1.0728
6 Mar 2026, 17:40:44

&HEAD CHID='sprinkler_new'/
&TIME T_END=20.0/
&DUMP DT_RESTART=300.0, DT_SL3D=0.25/

&MESH ID='Mesh01', IJK=70,59,60, XB=1.8,8.8,2.1,8.0,0.0,6.0/

&SPEC ID='WATER VAPOR'/

&PART ID='Water',
SPEC_ID='WATER VAPOR',
DIAMETER=500.0,
DISTRIBUTION='ROSIN-RAMMLER',
AGE=60.0,
SAMPLING_FACTOR=1/

&PROP ID='Generic Commercial Spray',
PART_ID='Water',
FLOW_RATE=56.02,
PARTICLE_VELOCITY=6.0,
SPRAY_ANGLE=30.0,80.0/
&DEVC ID='SPRK', PROP_ID='Generic Commercial Spray', XYZ=5.4,5.0,4.2, ORIENTATION=0.0,0.0,1.0, QUANTITY='TIME', SETPOINT=1.0/
&VENT ID='Mesh Vent: Mesh01 [XMAX]', SURF_ID='OPEN', XB=8.8,8.8,2.1,8.0,0.0,6.0/
&VENT ID='Mesh Vent: Mesh01 [XMIN]', SURF_ID='OPEN', XB=1.8,1.8,2.1,8.0,0.0,6.0/
&VENT ID='Mesh Vent: Mesh01 [YMAX]', SURF_ID='OPEN', XB=1.8,8.8,8.0,8.0,0.0,6.0/
&VENT ID='Mesh Vent: Mesh01 [YMIN]', SURF_ID='OPEN', XB=1.8,8.8,2.1,2.1,0.0,6.0/

&TAIL /

[mcgratta]

I think that you are misinterpreting the role of ORIENTATION. It is assumed that the water droplets emanate from a single point corresponding to location of the strike plate. The ORIENTATION and SPRAY__ANGLEs are tied to this location.

[danish2016]

Thanks for the clarification.

I have now defined ORIENTATION = -1, and specified the spray angle and offset distance based on the deflector position from the ceiling.

Currently, we are running sample test cases, but we are encountering the following error once the sprinkler activates:

Segmentation fault: address not mapped to object at address 0xfffffffc0b5f9200

I am trying to debug the issue and will update you once the test cases are completed.

Based on our FDS-smv group discussion regarding sprinkler modelling, we found that mesh size and jet stream offset could be a potential cause of this error, due to a large number of particles being generated within a single mesh.

At present, we are using:

Mesh size: 0.1 m

Default jet stream offset: 0.05 m

[mcgratta]

Are you using the latest nightly build? If not, try it. Also, can you post a simple case that fails.

[danish2016]

Dear Kevin,

I trust this message finds you well.

For the sprinkler modelling, we tested four cases using FDS 6.9.0:

a) Sprinkler_Simple_case – A single sprinkler was modelled with a single mesh, and the sprinkler was activated based on time for quick verification. The simulation completed successfully.

b) ACP-Summer-sprinkler_Case13 – Multiple sprinklers were modelled with multiple meshes, and the sprinklers were activated based on time for quick verification. The simulation completed successfully.

c) ACP-Summer-sprinkler_Case14 – Obstructions and multiple sprinklers were included, and a burner (fire source) was introduced. In this case, the simulation stopped just before sprinkler activation.

d) ACP-Summer-sprinkler_Case15 – This case is similar to Case 14, but the fire source was removed. The simulation ran successfully.

From these observations, it appears that whenever the model includes both a sprinkler and a fire source, the simulation stops before sprinkler activation.

Could you please advise on the possible reason for this? All the above cases run successfully in FDS 6.8.0, but we need to run the model using FDS 6.9.0 or 6.9.1 as we are using S-Pyro modelling.

Please find the attached zip file (including all four above cases FDS files) for your reference.

Thanks for your support.

Regards,
Wali
For FDS.zip

[mcgratta]

I am going to run Case 14.

[mcgratta]

Approximately when does Case 14 fail?

[danish2016]

Hi Kevin, the simulation stops at 9.9. seconds, when I ran using the 6.9.0 version.

[mcgratta]

With the latest source code, Case 14 has run out to 20 s and is still running.

[danish2016]

Dear Kevin, Could you please advise which version you use to run the model, because we tested for both 6.9.0 and 6.9.1, and the simulation stop. And for 6.10.0 i checked with our HPC service provider, they mentioned that the machine is not compatible with the latest code, therefore, I did not get a chance to run the model using 6.10.0

Regards
Wali

[drjfloyd]

Kevin used the current source which is not yet an official release. Try removing the ADIABATIC surface from your burner. In older versions of FDS that was a source of instabilities with droplets as you cannot meaningfully get a water cooling value for such a surface. In current versions we skip that calculation for ADIABATIC surfaces but I don't remember if that fix was before or after 6.9.

[mcgratta]

This case is still running with

 Fire Dynamics Simulator

 Current Date     : March 13, 2026  09:18:17
 Revision         : FDS-6.10.1-2138-g87b916e-master
 Revision Date    : Wed Mar 11 17:42:24 2026 -0400
 Compiler         : Intel(R) Fortran Compiler for applications running on Intel(R) 64, Version 2025.3.2 Build 20260112
 Compilation Date : Mar 12, 2026 09:50:34

 Number of MPI Processes:  6

 MPI version: 4.1
 MPI library version: Intel(R) MPI Library 2021.17 for Linux* OS

 Hypre library version: v3.0.0
 Sundials library version: v7.5.0

[mcgratta]

Question -- why is your computer not compatible with the current release?

[danish2016]

Dear Kevin, I appreciate your kind support. The simulation is successfully running. And regarding the 6.10.0 compatibility, I got an email from the service that they require the Rocky 8.10 version, while the current 6.10.0 require newer operating system.