Hi I'm relatively new in CFX.I'm working in a Internal Combustion Engine simulation using CFX 10.I have reach the movement of the piston but I have a lot of problems with the Valve motion. The main problem is when i try to simulate the valve motion. I use a CEL expression which describes the motion. But when it's closing, mesh elements near valvehead are destructed because of the proximity between the valve and the wall. It's written : ERROR #002100012 has occurred in subroutine cVolSec. Message: A negative element volume has been detected.

What can i do? 1)Is it possible to simulate the valve motion with a mesh mapping using JUNCTION BOX ? 2)Is it possible to create new mesh elements during a simulation? Is it possible with ICEM? 3)Is it impossible to do this, only with a CEL expression?

Thanks you for hearing me

You cant do that.

1)Is it possible to simulate the valve motion with a mesh mapping using JUNCTION BOX ? No, you cant change the mesh topology.

2)Is it possible to create new mesh elements during a simulation? No

Hi,

It is possible to change the mesh topology by stopping the run, changing the topoplogy and restarting with interpolated initial conditions. Very messy, but possible.

An alternative method is to use a GGI interface at the valve curtain. You can slide this until it no longer intersects and therefore closes the valve. You will probably have to make some small changes in your geometry to do this.

This is one area where Fluent and StarCD has a clear advantage over CFX as I understand both these codes support this feature. Hopefully in coming releases some of this functionality can be ported into CFX and mesh addition and deletion can be done inside a run.

Regards, Glenn

Hmmm GGI ... didnt think of that

Hi,

And I thought of another way last night:

You could move the valve to almost closed, then set a momentum sink in the small gap left to stop the flow, or at least put so much resistance in it that is effectively stopped. Not sure if this will be numerically robust but hopefully with a little fiddling it should work. This would probably be the easiest way to do it as there is no topology changes and as long as you are careful with your meshing you should be able to move the valve to very close to shut.

For the record: My PhD thesis, (submitted about 5 years ago) was modelling an internal combustion engine with CFX4. You can have a look at it on the following link if you are interested.

Regards, Glenn Horrocks

Your suggestion does sound like numerical instabilities wet dream. However, it could be usefull for my particular application which is more like a butterfly valve action rather than a linear automotive valve. The GGI idea wouldnt be feasable for me but this appraoch could be. I will investigate this in due course.

Moddeling butterfly valve-like valves in CFX is going to be real pain in the ass.

I have downloaded your PhD (thanks for the link) ... I now have 6 PhD theses in my reading inbox

I have been trying for a time with the GGI interfaces, but no results. What is the region that i have to descrive like an interface? When i try to simulate the closing of the valve, i always find the same error: Negative volume element... i don´t know if i´m making a really bad mesh or it`s another type of error. But it appears really far from the absolute close of the valve. What can i do? thank you for everything