I want to solve the combustor of dump type.

Geometry in combustion process has combustion instability in arbitrary equivalent ratio.

In steady state analysis with STAR-CD, mass residual value,of course, Not reduced .

In transient analysis approach,

case1) combustion model is EBU. Turbulent model is k-epsilon model.

case2) combustion model is PPDF. Turbulent model is k-epsilon model.

Problem of case2: 1. when I solve in the case2, Temperature increase with time step but In case1, Temperature Not change. I think this is because of ignition setting problem.

How does I determine ignition time step

2.Is it possible to see the Growth of a combustion instability following transient calculation with STAR-CD?

Kris,

I think dump combustors are usually premixed, which means the PPDF model is not an appropriate choice.

The approach you are describing is an unsteady RANS calculation. For premixed combustion in Star, you can choose from the flame front models (i.e., Weller and CFM) or ebu models. I believe that flame front models are more appropriate for premixed combustion, but they are also more difficult to use.

Getting successful numerical ignition often requires some experimentation. You should probably turn on ignition early in the calculation and leave it on for at least a residence time or some other characteristic timescale for your problem. Then you should run the unsteady analysis for at least several residence times beyond ignition. Typical ignition temperature (in Star) is 1800K and the burnt fraction might be set at 0.25.

Good luck - Mike

Thank you very much, Mike.

Actually, I used ummixed gas, inlet1(coannulus=fuel),inlet2(center=oxidant),in my geometry.

Recently, two kinds of model is used. those is PPDF and EBU.

The problem of EBU is ignition time step.

So, I first applied PPDF for my model. then that time in which combustion start was decided as starting ignition for EBU.

Of course, I condsidered characteristic timescale but combustion is NOT occured.

Now, I am going on calculation using PPDF.

For High order Temporal Discretization, I used Crank-Nicholson, applied for UD scheme execept density.

At this point there were another problem.

that is unstable. :<

actually, in this case, time step interval is a little long.

Original objective is to get Pressure oscillation