I am investigating the feasibillity of using CFX to model the intake and compression stroke flows in a diesel engine.

The work will be in support of a colleague who is doing experimental work requiring the removal of material from the piston in order to change the compression ratio.

The modelling would not at this stage involve any combustion modelling, but will hopefully be extended into this complex phenomenon at a later stage.

My numerical background is more in Finite Element Analysis and I have only recently begun learning some of the principles of CFD.

Do any of the readers of this page have experience in this kind of work and have advice for me as I begin this project?

regards

Gareth

I have done some similar work in CFX-4 (although I only prepared the grid motion part) for a truck Diesel engine. CFX-5 now also has such moving mesh capability and should be easier to learn and implement. The best bet is to ask your vendor for some test case example he might have. They might put you in touch with their Uk and German offices where such work is definitely carried out.

Hi,
I don't believe CFX-5 has such capability implemented yet. However, I agree with you that moving grids in a unstructured mesh may be easier than dealing with a structured one. I also did similar work with CFX-4. However It was a Spark Ignition(SI) Engine. I had the piston motion but no valves. I used source terms in some regions trying to reproduce the presence of valves. I achieved no success at all in that occasion. Now, in my master thesis, I'm still working with a SI Engine with two valves but I'm using CFX-TASCflow. People might say that AEA doesn't have a specific code for these kind of application, as STAR-CD, KIVA, etc. However CFX-TASCflow can handle moving grids and I believe that the code can perfectly deal with this kind of problem.
Regards, cfd guy

CFX-5 does have moving grid, I'm fairly sure!

Hi Gareth,

About six months ago I submitted my PhD thesis on CFD modelling of 4 stroke, SI engine internal aerodynamics. I used CFX 4.3 to perform the simulations.

I performed two main sets of simulations in my thesis. The first was a validation of the code using a simplified engine geometry (that is a square piston, and only a very simple valve action) and compared to experimental results. The second set of simulations was modelling the air flows in the full engine.

The validation model gave excellent results, and demonstrated the importance of the turbulence model in engien modelling. The full engine model was also successful.

I used CFX 4.3 for the modelling. The validation model coded quite easily and worked well, but the full engine model was a bit of a nightmare. Modelling the motion of valves with a structured code is not for the feint-hearted!. CFX 5 and TASCFlow should be significantly easier to code using GGI (general grid interfaces) around the valves. However I don't think CFX 5.5 has artibrary grid motion yet, just multiple frames of reference (I think that is correct) so a full transient model of an engine is difficult at the moment.

Hopefully AEA will implement full arbitrary grid motion in CFX 5 for the next release. That will mean it has all the features of the traditional IC engine codes (KIVA, Star-CD), and with its extra features of a coupled solver and GGI it should be significantly better.

If you are interested in discussing any issues regarding this work feel free to contact me.

Regards, Glenn Horrocks

Thanks Glenn

Your comments are most encouraging. I have begun spending some time reading journel literature on the subject and had yet to find someone who has done significant engine internal modelling using CFX. It seems that STAR-CD, KIVA, FLUENT, RAMPANT and a few others are more common choices. Why did you use CFX?

I think i'll have to use CFX - it's the only code we have at University of Cape Town.

To which university did you submit you PhD? I assume that you would have had to write a number of journal papers on your research? Can you email the papers to me, or give me the references so I can read up on your "agonies and ecstasies"!!??

Why is a square piston simpler than a round one experimentally? I'll bet it was fun to machine the tolorances!

I can't wait to read the details of your work! What are your plans post PhD?

Regards Gareth