Recent Changes - Search:




edit SideBar

Modelling of Hydrogen Explosions

The description of the modelling is not limited to hydrogen but it deals with premixed combustion, regardless of the fuel. Paragraphs on numerical simulations are instead specifically dedicated to hydrogen explosions. The emphasis of the paragraphs on modelling is on turbulent combustion because most of the explosions in real applications are turbulent flames. Because of that reason, laminar flames and detonations have been neglected in the description of the modelling. Some numerical simulations of hydrogen detonations are described in the paragraph “Simulations of Detonations and DDT”. A more comprehensive treatment of numerical combustion may be found in the excellent book by Poinsot and Veynant (2005) from which relevant material has been extracted in order to write these paragraphs. Relevant reviews on turbulent combustion can be found in Veynante and Vervisch (2002), in Lipatnikov and Chomiak (2002, 2005), and in the NEA report on flame acceleration and DDT (Breitung et al., NEA/CSNI/R(2000)7). A review document dedicated to CFD modelling of gas-explosions was written by Lea and Ledin (2002).

CFD (Computational Fluid Dynamics) simulations can be performed using 3 main fundamental approaches: Reynolds Averaged Navier-Stokes (RANS), Large Eddy Simulation (LES) and Direct Numerical Simulations (DNS). In DNS, the mesh resolution is so fine that all turbulent length and time scales are captured by means of the mesh itself. Therefore no modelling is required for the turbulence. The requirements for the computer resources are so high that DNS applications are limited to simple academic problems of flows with low/moderate Reynolds number. Nevertheless DNS is extremely useful in providing a large amount of data for a better understanding of the basic mechanisms, for combustion modelling and for model validation. Being DNS applications restricted to simple academic problems, this approach has been neglected hereafter. In the RANS approach, only time-averaged quantities are calculated and no turbulent scales are captured by the means of the mesh. Therefore this approach requires turbulence modelling. Thanks to its lower requirements in terms of computational resources, RANS is still the most widely used computational approach for real-scale engineering problems.

In the LES approach, only the large turbulent scales are resolved by means of the mesh while the small turbulent scales are modelled using sub-grid closure rules. The computer requirements are intermediate between DNS and LES. This approach is now developing rapidly even for industrial applications (Westbrook et al, 2005)

<< Combustion of inhomogeneous mixture in closed vessel and pressure build up | Content | Simulations of Hydrogen Deflagrations >>

Edit - History - Print - Recent Changes - Search
Page last modified on February 17, 2009, at 02:58 PM