# Strange epsilon

 Register Blogs Members List Search Today's Posts Mark Forums Read

 September 13, 2006, 03:23 I am running a simulation with #1 Member   chris book Join Date: Mar 2009 Posts: 85 Rep Power: 17 I am running a simulation with the standart k-epsilon model. I have specified an initial condition for k=9.5 and epsilon=1750. However the volume averaged epsilon after the first time step is = 3700! Can anybody shed some light into it why the volume averaged epsilon is some much different form the initial condition?

 September 13, 2006, 05:53 That's because your turbelence #2 Senior Member   Hrvoje Jasak Join Date: Mar 2009 Location: London, England Posts: 1,906 Rep Power: 33 That's because your turbelence is huge! Your velocity FLUCTUATION is 3.5 m/s - this is the speed I have trouble to walk on. Assuming turbulence intensity of 1%, your mean velocity would be the speed of sound (with a lot of turbulence). I would revisit my initial condition, it does not look right. Hrv __________________ Hrvoje Jasak Providing commercial FOAM/OpenFOAM and CFD Consulting: http://wikki.co.uk

 September 13, 2006, 06:12 Hrv, I think a velocity flu #3 Member   chris book Join Date: Mar 2009 Posts: 85 Rep Power: 17 Hrv, I think a velocity fluctuation of 3.5 m/s corresponds not to k=9.4 m^2/s^2, instead a fluctuation of 2.52 m/s is right. Do you agree ? The turbulence data (velocity fluctuation, integral lengthscale) comes from a experimental turbulence generator (multihole plate) with an integral lenghtscale of 3mm. In the experiment I want to simulate there is almost no mean flow but there is a sort of background turbulence created by the turbulence generator. So do you still think there is a incorrect initial condition?

 September 13, 2006, 06:17 Ah, OK: if you have a turbulen #4 Senior Member   Hrvoje Jasak Join Date: Mar 2009 Location: London, England Posts: 1,906 Rep Power: 33 Ah, OK: if you have a turbulence generator and experiments, this may well be right. In that case, you should just let the flow develop and the model should do its job. As for k, I probably made a mistake: k = 3/2 uPrime^2 I should have said 2.5 m/s, i.e walking speed :-) Hrv __________________ Hrvoje Jasak Providing commercial FOAM/OpenFOAM and CFD Consulting: http://wikki.co.uk

 September 13, 2006, 07:58 Hi, Epsilon usually scales #5 Member   Ola Widlund Join Date: Mar 2009 Location: Sweden Posts: 87 Rep Power: 17 Hi, Epsilon usually scales something like K^1.5/(0.3*L). In your case this is 3e4. In that light the value after one time step is not very high, rather your initial value is very small. /Ola

 September 13, 2006, 09:03 ok but the same case running w #6 Member   chris book Join Date: Mar 2009 Posts: 85 Rep Power: 17 ok but the same case running with star give me averaged epsilons from 2000..6000! So I don't know where the big difference gomes into the game

 October 24, 2006, 09:05 What is resonable choice for e #7 Member   chris book Join Date: Mar 2009 Posts: 85 Rep Power: 17 What is resonable choice for epsilon (I use the compressible RNG-k-epilon model) in the case of no initial turbulence (i.e. k=1e-8)?