Comments on: Reactive Flow Modeling- Two Dilemmas

By: Masha Petrova

Masha Petrova — Wed, 19 Aug 2009 22:37:05 +0000

Simon - you are absolutely right, chemical kinetic mechanisms are are problem for pretty much every one. I tried to demonstrate that in this presentation. Wouldn’t it be great if there was one, central place for mechanisms - where only truly tested and validate mechanisms were kept?

Lorenzo - thanks for the note on Google. I use Google products every chance I get, but youtube is now the second largest search engine, believe it or not, so I wanted to get the information out there for more people to find.

David, thank you for this great response! And yes, I believe we are saying the same thing. An expert is still essential for successful and accurate modeling work in reactive flows.

By: David Kelsall

David Kelsall — Wed, 12 Aug 2009 14:34:01 +0000

Hi Masha,

I’m not sure I totally agree :-). I don’t think engineering is always about perfection anyway. The challenges you discuss have been the province of chemical engineers for many years. There are some very powerful chemical process simulation packages around.

However I also believe you can’t get something for nothing. If chemical reactions are truly important to the processes you are trying to simulate, then I’m afraid there has to be some investment in finding out about them. At the simplest level, it might involve talking to a chemist - though much more research is likely to be involved :-). Even then computer modelling can only get you so far. It is very likely that some sort of validation (i.e. experimentation) is likely to be required - before the models can be used predictively. I guess this is why for some chemical companies the IPR for their processes is so closely guarded - because they have invested the time and money in procuring validated chemical kinetic data.

I don’t think computers always help. I don’t think they are yet powerful enough in general to solve chemical kinetic problems simultaneous with 3D flow simulation.

Fortunately nature helps a bit :-). The time- and length-scales for chemical kinetic processes are usually radically different from that of the flow processes, i.e. the former are much faster and shorter than the latter. Nevertheless the CPU intensity of a chemical kinetic scheme (covering just a few minutes of the process) can be of similar order to the simulation of the flow processes (covering maybe hours of the process). Depending on the degree of granularity required, computational effort could become excessive. (When I was a research student, I used to use a Cray supercomputer to calculate the chemical kinetic schemes I was trying to simulate. Few organisations could afford the sort of expense I used to incur :-))

I believe to a very good first approximation the two types of simulation can be done separately - and maybe one approximated in the other. For example, one might do a detailed chemical kinetic analysis to work out the rate of heat generation or absorption (for an endothermic reaction) and approximate it in the flow solution.

If you try to engage in simultaneous chemical and flow modelling - depending on how you do it - there are some pretty big holes you can fall down. My advice to someone with limited experience of chemical kinetics is they should talk to an expert. Don’t assume you can just flick a switch and it will work out of the box. But I guess this is what you are saying too :-).

Anybody else got any thoughts?

Best wishes,

David Kelsall

By: Lorenzo Bossi

Lorenzo Bossi — Wed, 12 Aug 2009 14:32:16 +0000

Interesting presentation Masha,

As there is no sound, you could also upload it on SlideShare or Google Presentation (Applications tab on LinkedIn), makes it easier to browse it at our own pace.

regards
Lorenzo

By: Simon Leefe

Simon Leefe — Wed, 12 Aug 2009 13:22:41 +0000

The thrust of your article appears to be combustion and I think many of your assuptions and conclusions come from this perspective. We have found that much can be learned about a wide variety of chemical process equipment from CFD, where often the focus is less on 2nd or 3rd significant figure accuracy, but rather on insight, to inform decisions e.g. about the comparative impact of alternative layout of internals, retrofit options for increased thoughput, behaviour on scale-up, etc. Here we are typically concerned with gaining an understanding of the interactions between geometry, physics and chemistry in order to characterise system behaviour and guide design decisions.

When chmistry modelling is important, we typically find that the main bottleneck is the absence of data on kinetics.