I would like to simulate the combustion in internal combustion engine. What are the initial values (how do I find or calculate them) to set for k and epsilon for the k-epsilon turbolence model? Any information is apreciated

Choose k be using 3-5% of the mean velocity as the velocity fluctuation. You can then do

k = 3/2 uPrime^2

to get k.

For a length-scale, choose a fraction (say a quarter) of your inlet manifold diameter or a bit less of the cylinder diameter if you've got no manifolds - L = 1 cm is a decent size for a car engine. You can then estimate epsilon as:

epsilon = Cmu^0.75 k^1.5/L

Cmu is (of course) 0.09.

In reality it does not matter much what you choose. The whole setup is really a turbulence-making machine (you want good mixing, right?) so the initial field will be quickly replaced by "real" turbulence. When that happens, you can check that your estimated k and epsilon with some more realistic data.

Enjoy,

Hrv