One dimensional flame simulation

[hille]

Hello!

I am having a problem simulating a burner stabilized flat flame from
Matlab using the Cantera toolbox. I am using a reaction mechanism from
MIT which is very large (over 6000 reactions). I have done a lot of
simulations with this reaction mechanism and never had any problems
with it (except for long simulation times) but suddenly all of my
simulations crash and I get the following error message:


???

************************************************
Cantera Error!
************************************************


Procedure: OneDim::timeStep
Error: Time integration failed.

Error in ==> @Stack\private\stack_methods at 13
v = ctmethods(90, n, job, a, b);

Error in ==> Stack.solve at 4
stack_methods(s.stack_id, 104, loglevel, refine_grid);

Error in ==> MIT_30cm_syngas at 94
solve(fl, loglevel, refine_grid);


I have used the flame1.m program as a base for my own program and haven
´t changed it that much. Why is it suddenly not working? Does anyone
have a clue? I would really appreciate the help ;)


Helena Svensson, PhD Student
Department of Chemical Engineering
Lund University

[Harry]

Hi Helena,

I'm not sure why your program is failing. Could you provide me with
more details. What version of Cantera did you have success with? What
version of Cantera is your code failing with? Do you have intermediate
output concerning the convergence behavior?

There hasn't been changes in the 1D code algorithms for several years
now. So, the code base has been fairly stable. However, I've certainly
noticed that the current algorithm isn't stable with respect to
numerical round-off error. Going forward, we are going to be looking
at this and possibly change the discretization method and regridding
algorithms.

As part of getting release 1.8.0, I've just today found an array
overwrite error in Sim1D.cpp and have checked a fix into the svn
googleCode repository today. But, I doubt it's relevant to your issue.

It appears the code failed to convergence at some point.
If we progress on this, you should probably send to me a sample
problem, so that we can explore within a debugger why it failed to
converge.

Best wishes,
Harry

[divyatman khare]

Hi,

I to have experienced similar error but with GRI 3.0 mechanism while using Cantera toolbox from MATLAB although the code with same parameters when called from Python gives the results. Most of the times i could get through this by changing initial grid. 

--
Divyatman Khare
B.Tech III Year
Indian Institute Of Technology Roorkee