Sprinkler Modelling in CFD Projects
- davebrigly0
- Jun 7
- 1 min read
As software develops, the ability to simulate the effects of sprinklers in models is improving. But is this progress leading to assumptions that can impact on the robustness of the overall fire safety package?
If we are simulating a fire size to assess the performance of a smoke control system, such as a firefighting lobby, corridor extract system or atrium how far do we go with introducing lower fire sizes and sprinkler cooling. Should we ignore sprinklers all together and assume worst case scenario?
Of course, this should always be project specific, but there appears to be little guidance on the practical application of introducing sprinkler performance into real projects.
How do approvers assess? Is the dependency on sprinklers in the design, explicitly passed onto occupiers or even the Fire Service?
What do you think?
I am not in favor of modeling sprinklers in CFD using FDS, as the suppression physics are not accurately captured. In FDS, suppression agents are modeled as Lagrangian particles, and the cooling effect lacks sufficient validation. While other CFD tools like ANSYS or OpenFOAM offer more advanced suppression modeling capabilities, they are computationally intensive.
A more conservative and practical approach is to rely on full-scale experimental data and model the fire without incorporating decay—treating it as a steady-state fire at peak heat release rate (HRR). However, for sprinkler-controlled fire scenarios, simplified methods such as the equations developed by Alpert and Heskestad can be used to cap the HRR at the time of sprinkler activation, with an appropriate safety margin, and…
I believe most will agree that omitting spray cooling gives a more conservative and defensible result. If you cap the HRR and apply cooling sprays, you're taking credit for two suppression mechanisms: a lower HRR and the cooling of combustion gases—without physically modelling the fire's response to suppression (e.g., Heat from combustion is capped, but still removed by cooling, leading to double-counting the effect). This approach risks overstating system effectiveness unless supported by empirical justification.