Re: transient counterflow flames

[Ray Speth]

Dirk,

You are correct, Cantera's 1D solver only solves the steady-state problem. 

As it happens, I have written a program, Ember, for simulating time-dependent 1D flames. I recently made it available under an open source (MIT) license. You can download the source code from Github:

    https://github.com/speth/ember

And view the documentation here:

    http://speth.github.io/ember-doc/sphinx/html/index.html

The program solves a variety of 1D flame configurations: strained or unstrained, premixed or diffusion, planar or curved (tubular). It is built on top of Cantera, and uses Cantera to do all the low-level kinetics and thermodynamics bookkeeping. The solver itself is quite robust;  In my experience, it is able to find solutions to many problems where Cantera's solver does not converge. The main reason I haven't publicized it before is because it doesn't currently have a lot of good examples to get users started, though there are a few included that I think you could adapt to the problem you're interested in. Please let me know if you find Ember to be useful.

Regards,

Ray

On Monday, May 20, 2013 6:03:48 PM UTC-4, dirk.g...@googlemail.com wrote:

Hi,
I would like to compute the ignition process of fuel versus heated air. 

Is it possible to do this with Cantera by observing the transient ignition processes within the counterflow flame ?
I.e. the changes in composition and temperature during the ignition process.

I did not find any example for that on the web and fear that it may not be possible to do this with Cantera.

Dirk

[Ray Speth]

Hi Dirk,

(1) It should be possible to simulate this configuration, except that as you note there's no way to currently define the composition of the hot gas directly. I'll look into adding a way of specifying this information; it shouldn't be too difficult.

(2) Ember can include the Soret diffusion fluxes using the thermal diffusion coefficients computed from the Multicomponent model. However, the fluxes due to concentration gradients are currently always computed using the mixture-averaged properties, since in this case the problem becomes independent for each species. In principle, the full multicomponent equations could be solved, but I think this would come at a significant performance penalty, and would take a fair bit of work to implement.

(3) The strain rate in Ember is the strain rate in the unburned mixture. If you want to think of the strained flame in terms of the typical experimental realization, Ember's model is effectively the limiting case where the separation between the jets approaches infinity. I'm not sure how to apply the definition of the "global strain rate" in this case.

Regards,

Ray

On Saturday, June 1, 2013 5:36:39 AM UTC-4, dirk.g...@googlemail.com wrote:

Ray,

I was looking over the examples (looks like I might give it a try soon).

Three questions:
1.) Is it possible it possible to run a case with cold prexmixed fuel versus hot products (autoignition) ? I did not find an example showing a way to define hot products.

2.) Is there a way to switch between the transport models ? (mixture averaged, multicomponent, multicomponent+soret) . We see in our data for the steady cases that the multicomponent+soret settings match the experiments best

3.) The strain rate settings: is it the global srain rate which is used here (Sehadri/Williams for example) or the strain rate in front of the thermal layer. Sorry for this last question, my background is on the experimental side.

Greetings,
Dirk

 

On Wednesday, May 22, 2013 2:38:41 PM UTC+2, dirk.g...@googlemail.com wrote:

Ray,

this sounds very interesting. I will see if I can find a student to run some tests with that. Too time consuming for me during the term, but cetrainly worth to spent some time in during the term break.

Dirk