Hi

Im doing some airflow modeling around/through an auto chassis. One open issue is modeling the impact an engine cooling radiator. I can get reference data from the actual car being modeled (i.e. mass flow/pressure loss), but how does one represent a device like a cooling radiator in CFX? Would it make sense to just define a porous surface with the appropriate values? Another thought I had would be to just model this as a series of square tubes in a frame, with the geometry calibrated through some trial/error to approximate the reference data; this approach would keep the node count down relative to actually modeling cooling fins on tubes, etc.

Any help is appreciated-

Hi,

Use a momentum sink for the pressure drop and a heat source for the heat addition. Look in the documentation under sources and sinks.

Glenn Horrocks

Thanks for the insight on the momentum sink approach. Im assuming I should stick with the 'directional loss model' here since the needs are quite simple. In my model, the chassis/engine/radiator is modeled as a single assembly. I cant find a way to break the radiator out of the assembly for binding in a subdomain to permit application of the loss model to just this component. Im sure I am missing something, but any thoughts are appreciated.

Regards,

Hi,

What meshing software are you using? It is done different ways in different packages, but you want to define the region where the radiator will be as a separate block as the remaining mesh and then you can define it as a sub-domain.

If you are using CFX-Mesh have a look at the heating coil example. It is a multi-domain CHT example but the principle for generating the multiple domains is still the same for sub-domains.

Glenn Horrocks

Hi Glenn

Yes, I am using CFX Mesh at present. I have access to ICEM but have no experience with it. Maybe I would be better off investing some time with ICEM?

Ill check the heating coil example you mention- maybe it is possible to leverage that with my current model, but some of the difficulties Im having with node counts and convergence, I may need to investigate a better meshing solution.

Glenn/all

I cant figure out how to create multiple 3d regions within an assembly. Presumably, this is required to permit a subdomain treatment for my application (creating a momentum sink (in the form of a radiator) at the front of an engine bay of a car model).

In my test, I have a group of 3d solids comprising an automobile chassis. In design modeler, I setup an appropriate test tunnel around the frozen chassis components, and then use a body cut operation to remove the chassis from the tunnel. This operation is all or nothing; if I try two separate cut operations (chassis/engine, and radiator) it merges them both into a single solid body.

So far as I can tell, CFX Mesh will not permit the definition of multiple 3d regions- only 2d composite regions contained in a single 3d region. It appears to me that the sample you referenced, and the other info I have found on the matter, suggests that I need to build separate meshes and then join them later manually. This approach is simple for sequential components (i.e. simulation of an inline air filter), but it does not seem so simple when the component in question is contained within another 3d grid.

Not sure if my understanding of this is correct, but it would seem that there should be an easier way to define separate nested 3d regions for later 'subdomain' treatment.

Am I missing something here?

As always, thanks for the guidance.

Hi, I have not used CFX-mesh, but as you have access to ICEM, you can define n number of subdomains in ICEM just by picking up the required block and giving it a proper name. Its very simple if are conversent with multiblock structured meshing, just have a try. Bye Ram Dayal