The modelling of a turbulent premixed flame propagating in a constant volume vessel has been studied and compared with experiments. Three models including step by step the turbulent mixing, the flamelets propagation for wrinkled flames, and the flamelets stretching, interactions and thickening when the chemistry is not very fast, have been tested. The experiment considered here allows the knowledge of the turbulent kinetic energy and the turbulence scale at the time of the spark, and avoid any problem of spark fluctuations and residual gases, but displays again “cyclic variations” and the comparison concerns ensemble averages. The results of comparisons is that the agreement concerning the flame brush propagation is improved from the simplest to the largest model, when considering a domain in which in which
varies from 2. to .75 and
from 2. to 0.4 . The turbulence model itself (
k - ϵ or
k -
lt) has a small influence on the flame velocity but a big influence on the shape of the flame, especially through the compressibility term. Further studies and comparisons with local experiments results are needed for assess the components of the most convenient turbulence model.