Modelling Enclosed Fires
Fires are a major hazard on submarines. However, the lack of a detailed deterministic analysis of the heat flux and smoke movement within a submarine during and after a compartment fire has previously hampered a full understanding.
This work looked at methods of modelling a fire within the aft compartments of a submarine, and the way in which mitigation effects were modelled. Two models were used, a zone model, (CFAST), and a field model or Computational Fluid Dynamics (CFD) code, (FDS). Both these codes have been produced by the US National Institute of Standards and Technology (NIST) for modelling fires within buildings, and are used extensively by fire modelling engineers.
CFAST (Consolidated Fire And Smoke Transport) is a fast running code that solves a collection of Ordinary Differential equations03to model fires within a compartment and the movement of smoke between compartments. In each compartment there is an upper smoke layer and a lower air layer. FDS (Fire Dynamics Simulator) takes a longer time to produce results than CFAST, because it predicts detailed information on the flow of air and smoke, and their mixing. Both codes include models of water sprays, with CFAST using a simple correlation to model the spray's effects, whereas FDS models the behaviour of water droplets within the compartments.
Models of a submarine engine room were developed using both CFAST and FDS. The engine room has a complicated geometry, and so was modelled as consisting of a number of interconnecting compartments using CFAST. For FDS the engine room was modelled as a single compartment, but with a number of obstacles placed within the model representing structures within the room. An oil fire at the bottom of the engine room was modelled and results compared between the codes. Both codes included the effects of a water spray. The vulnerability of electrical cable bundles to a fire within the engine room was also investigated.
This work demonstrated that it was possible to model fires within complex geometries using both zone and CFD models, and that mitigation methods could be modelled successfully. It also demonstrated that modelling complex geometries is not necessarily straightforward, and that users need to be experienced to produce realistic results.