I am trying to do some advanced combustion modeling which apparently requires me to use both SORSCA and SORENT subroutines. The combustor model is a simple premixed dump combustor design using methane and air. Before "turning on' SORSCA and SORENT, I have first acheived a reasonable solution scheme using a simple 2 step methane-air reaction using the normal combustion module. I then use this solution field (T, p, scalar mf's, etc.) as my restart point when I "turn on" SORSCA and SORENT. This is where the problems begin.

My values of S1P and S2P being generated in both SORSCA and SORENT seem very reasonable. However, the flame slowly goes out and the temperature values everywhere eventually reach the inlet temperature of 298K. This is true for both CHEMICO and Thermal, SIMPLE or PISO, small levels of underrelaxation, and various discretisation schemes. It seems that something more fundamental must be incorrect.

Does anyone have any pointers on using SORSCA/SORENT to model a combustion process? If so, I would be eternally grateful.

Could you provide us with your geometry, boundaries and the combustion scheme you are using?

Thanks for your offer to help. As far as the geometry and boundaries are concerned, I would be more than happy to send those to you via the most convenient method. The combustion scheme is a very long and involved subroutine (SORSCA) that computes molar production rates for nine chemical species. Unfortunately, I am not sure I am in a position to provide this code. However, based on the work of others, as well as the results I see coming out of the subroutine, I feel fairly confident that it is behaving satisfactorily. If you feel it is absolutely necessary, I can possibly send you a portion of the code where the molar production rates are linearised for all the scalars, as well as the SORENT routine.

What do you think? Incidentally, I am running v3.10 on a Linux box. Should I email the .mdl file to you directly?

Thanks again.

Kevin, No, I don't need any confidential info. But to be able to help we need some information to understand what's going on. Any way, this is a simple question: how do you ignite your mixture??

Thanks for your continuing help and your understanding regarding my situation with the combustion scheme.

You raise a very good question about how I ignite my mixture. I have been working under the (mistaken?) assumption that since I am restarting from a converged solution (again, with results computed based upon a standard methane/air 2 step reaction scheme) I did not need to worry about igniting my mixture. Before restarting, the max temperature of my "converged" solution is about 1750 K, with inlet temperatures of 298K. If it would be helpful, I would be happy to send you my .mdl file along with the .pst file which I use for my restart.

One question: my cell and vertex numbering seems rather random. Could this be causing my problems when I turn on SORSCA and SORENT?

Thanks again.

Keven, The cells and verticies are OK. What I'm worried about is the temperature, 1750 K seems to be lower than what it should be. I myself was facing the same problem, this temperature looks like as if it's just a heat transfer process after ignition and not a combustion. Methane/air combustion should produce at least 2300 K when you realy have combustion. Try to change the ignition process to get this temperature and see if it will continue.

Hi Kevin,

Ahmed raises a valid question about the ignition temperature, but I have a more basic question. If you are introducing an enthalpy source into the domain using SORENT, this MUST increase the enthalpy. So the temperature cannot go back to 298 unless your source term is also going to zero (or the equations are not converging).

Here's what I would check. Fix the enthalpy source terms in SORENT at their initial value after the restart(i.e. do not let them vary with anything after the first time step) and see if you can sustain the high temperature. If not, your source terms are wrong, or are not being applied (you can also check the net source in the .info file, under the item QSOR). If they do, then you should check your combustion model to see if an incorrect dependance on the T or concentrations is driving the enthalpy source terms to zero, extinguishing the flame (perhaps because of insufficient ignition temperature).

Ahmed:

Thanks for your continuting help. I have been busy trying your suggestions. I have changed my model slightly, and my max temperature now (using the standard combustion scheme) is about 2160K. With this change, my problem now is NOT the flame apparently going out when I restart with my chemistry. In fact, the max. temperature climbs steadily, in some cases up to 5000K before it bombs. I have tried different combinations of under-relaxation factors, and now have these set to about .001 for all scalars and the temperature. At the risk of sounding ignorant, this brings up two questions:

1. In the SORENT routine, the code says the enthalpy source is = S1P-S2P*T. In both my SORENT and SORSCA routines, I have tried to follow the scheme laid out by Patankar on pages 145-146 (Numerical Heat Transfer). Thus, in SORSCA, I have been dividing my S2P by the current value of the species mass fraction. Likewise, in SORENT, I am dividing my value of S2P (where S2P is the negative energy component of the enthalpy source for that cell) by the cell's current temperature. However, it seems like I should be dividing by the current enthalpy value of the cell rather than the temperature. Am I way off base? If I should use enthalpy, do I use specific enthalpy (J/kg) or just enthalpy (J) of the cell?

2. I also seem to be having trouble with the sum of my mass fractions deviating significantly from 1 in a number of the cells (in spite of very small values for under-relaxation). Is there a fundamental reason why Star sweeps across every cell for a specific species, then goes to the next species and sweeps over all the cells for this species, etc? It seems to me that it would be more stable to update all the scalars in one cell, then go to the next cell, etc. Again, am I way off base?

Thanks again. Let me know what you think.

Kevin

Sreenadh:

Thanks for your input. I responded to Ahmed's comments, but they also relate to your input. You might want to review my comments to him.

Thanks.

Kevin