is this simulation possible?
 

Hello,

I would like to know if it is possible to do the following simulation with OF tools, how much complexity would be involved, and if so what are the recommended steps ?

Simulate all the dynamic physical process involved with a piston-type air compressor including valves over the course of many cycles with changing BCs. This would include heat transfer between gas and all solid components, the temperature distribution of all solid components, the stress distribution in all solid components, leakage flow at all crevice clearances including the piston annular clearance.

thanks!

First, anything is possible with OpenFOAM. All the necessary tools needed exist in it's framework to tackle such a problem.

The practical answer is that it would be challenging.

Assuming you want to include solid stresses due not only to thermal expansion but due to the fluid forces, it would be a three-way physics problem: fluid dynamics, conjugate heat transfer and solid mechanics. I don't believe such a solver currently exists so you would have to write one. There are fluid-solid solvers (search for FSI) and fluid-thermal (search for chtmultiregion) but not all three. Actually now that I think about it, you might even need to model a fourth physics: kinematics, which includes both rigid-body motion and deformation, if you wanted to model stresses due to accelerations of the bodies. You might also want to search for Philip Cardiff's solidMechanics package.

You would need to use the moving and deforming mesh algorithms to handle the piston motion and possible body deformation. There are existing methods to handle this but it adds another level of complexity.

Time varying BCs whereby the type may change. Not sure if this has been done in OpenFOAM. If not, likely doable with swak4Foam. Another level of complexity

You mention resolving leakage flows. To do so would require a fine mesh resolution and most likely a large mesh. Another level of complexity.

Since the flow is likely compressible, you would either have to use an existing compressible solver in OpenFOAM as a basis, use another one that might have been published or write your own. Another level of complexity.

Oh ya. The flow is transient which means you'll likely want to run it parallel. Running OpenFOAM in parallel is generally straightforward but most would agree running parallel often adds another level of complexity.

So as you can see, OpenFOAM "technically" is able to do this problem _but_ it would be very complex. Unless you have a lot of experience running OpenFOAM or have no deadline to finish such a project, you might think about reducing the complexity of the problem or consider another package. I suspect this problem would even be challenging in one of the big commercial packages.

I don't mean to be pessimistic but hold no illusions - this would take time and determination even for a seasoned OpenFOAM developer.

Thank you very much for responding with detailed information! As for varying boundary conditions I mean to study the effects of varying manifold conditions, inlet temp and restriction, and outlet restriction on efficiency and output. I don't think I need to model the solid deformation interaction with the fluid flow, unless I try reed valving. But I am interested in optimization of machines for performance and durability etc.

Could you explain what you mean by running it parallel?

I will be researching related information for similar problems for a while but no urgent deadline, I see it as more of an expansion of a hobby inventor's capabilities with some applications in mind. So, If I could get it working in 2-3 months that would seem worthwhile.

Again thank you very much for your suggestions.

steve

hello Steve,

he ment that due to the compelxibility and the fact that such a simulation is transient you will run it in parallel mode. Means use more than one core of your cpu or using a cluster to do a fast calculation. Otherwise if you take all the mentioned topics into account, your simulation would run a very long time on "one" core - even its the latest one (: