#Best practice for modeling condensed solid aerosol fire suppression in FDS (PART vs SPEC)

[zepatos]

Hello FDS Community,

I am currently working on an FDS-based study to simulate and visualize fire suppression using a KNO₃-based condensed solid aerosol extinguishing agent, and I would greatly appreciate your expert guidance on the most appropriate modeling approach.

Background and Objective

The goal of this work is to numerically simulate and visualize the fire suppression process of a condensed solid aerosol system, including:

• Aerosol discharge from different heights and angles
• Interaction with various fire grades (A, B, etc)
• Visualization of:
  • Aerosol cloud formation and movement
  • Jet-induced vortices and circulation
  • Mixing, oxygen displacement, and suppression behavior
• Tracking key quantities such as:
  • Temperature evolution
  • Pressure buildup due to fire and discharge
  • Gas species (O₂, CO, CO₂)
  • Flow direction and discharge rate

This simulation is intended to support and verify physical extinguishing tests, especially since many internal quantities (temperature fields, pressure transients, species distributions) are difficult to measure experimentally and are required for engineering analysis and validation support.

Approach 1: Aerosol defined as particles (PART)

I first modeled the condensed aerosol as Lagrangian particles using &PART, coupled with an aerosol nozzle surface.
However, I encountered the following issues:

1. To obtain a visually meaningful aerosol cloud, a relatively high NPPC was required.
2. Once the aerosol discharge begins, the simulation time step collapses significantly.
3. As a result, the total computation time becomes impractically long (in my case, estimated to exceed 1800 hours).
4. In some cases, the simulation becomes unstable or crashes
   This makes the particle-based approach difficult to use for parametric studies (height, angle, fire size).

Approach 2: Aerosol defined as a gas species (SPEC)

I then modeled the aerosol as an Eulerian gas-phase species using &SPEC, injected through a nozzle via &SURF and &VENT.
This approach is numerically stable and computationally efficient, but I encountered different limitations:

1. The visualization of the aerosol cloud is not intuitive and does not resemble particle-like behavior.
2. It is difficult to clearly visualize:
   • Cloud boundaries
   • Vortex-driven motion
   • Local concentration build-up near the fire
3. The suppression effect appears less obvious compared to the particle-based model, making interpretation and presentation more challenging.

Main Questions

I would appreciate guidance on the following points:

1. Modeling strategy

• For condensed solid aerosol agents, is it more appropriate to use:
  • PART (Lagrangian particles),
  • SPEC (Eulerian gas-phase surrogate),
  • or a hybrid/phenomenological approach?
• What is the recommended practice within the FDS framework?

2. Particle-based modeling (PART)

• Which parameters are most critical to control computational cost while retaining physical meaning?
  • NPPC
  • AGE
  • particle size / density
  • injection duration
• Are there recommended upper bounds or scaling strategies for suppression simulations?

3. Species-based modeling (SPEC)

• How can visualization of aerosol “cloud-like” behavior be improved?
  • Choice of MW
  • Slice quantities (mass fraction vs volume fraction)
  • Effective use of velocity, vorticity, or extinction coefficient
• Is it reasonable to treat condensed aerosol primarily as an inerting / diluting agent in FDS?

4. Physical realism

• Given that FDS does not explicitly model aerosol chemistry (e.g., K-radical inhibition), what assumptions are commonly accepted when using FDS to approximate condensed aerosol suppression?
• How close can such simulations reasonably be expected to match real extinguishing tests?

5. Performance vs accuracy

• What strategies are recommended to:
  • Minimize simulation time
  • Maintain numerical stability
  • Achieve clear, defensible visualization results suitable for engineering analysis?

I will attach my FDS input files and parameter settings below for reference.

Any guidance, references, or example cases would be greatly appreciated.
Thank you very much for your time and support.

Best regards,

FDS_aerosol_particles.docx
FDS_aerosol_species.docx

[drjfloyd]

KNO3 has combined thermal and chemical effects on extinction. You are not going to capture the thermal effects without somehow accounting for the particle decomposition and the particle interactions when they settle on a surface. The later is not something that FDS fully does currently. The chemical effect is not something you can do without accounting for free radicals which at the moment needs some level of detailed chemistry and DNS. This is not a routine engineering application of FDS. This is research.

[zepatos]

Thank you @drjfloyd for the clarification.

Given the current limitations of FDS regarding solid aerosol decomposition and detailed chemical (radical) suppression mechanisms, what would you consider the most appropriate and defensible way to use FDS at this stage?

To be more specific:

• Which physical effects (e.g., flow field, jet dynamics, thermal dilution, oxygen displacement, aerosol/gas transport) can be modeled with reasonable confidence using standard FDS capabilities?
• How would you recommend representing a condensed solid aerosol agent in FDS (e.g., as a gas species, tracer, or hybrid approach) to maximize physical realism while keeping computation practical?
• Are there any best-practice assumptions, surrogate models, or simplifications commonly used in research to study aerosol fire suppression with FDS without over-interpreting the results?
  My goal is to utilize FDS within its valid engineering domain to support qualitative understanding and experimental verification, rather than to claim full chemical extinction modeling.

I appreciate any guidance on how to proceed in the most scientifically sound way.

Thank you.

[drjfloyd]

I am not aware of this being done by anyone for an engineering assessment; this is a research project. If this is a real world engineering application with real human life on the line, your uncertainties may be too high for it to be of practical use. We have investigated this at DNS scale with detailed, finite rate chemistry for sodium bicarbonate powder in a cup burner. It is very challenging even for such a simple configuration and at the moment not something that can be scaled up to an engineering scale simulation. At best I can give you some high level steps. I cannot give you detailed steps.

I think you will need to:

1. Develop and validate the time dependent boundary condition for the aerosol generator:
   -flow rates of gas and particles as a function of time (I think your best bet is to represent the particles - recognizing that FDS currently does not do adhesion/liftoff calculations for solid particles)
   -particle size distribution
   -temperature

2. Develop and validate decomposition kinetics for the particles

3. Devevlop and validate some metric based on particle concentration, decomposed gas concentrations, local oxygen concentrations, local temperature, etc. to assess exintction likelihood

[zepatos]

Thank you very much for your time and clarification! I really appreciate it.