CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   Automatic 2D optimization (

René Heise February 16, 2001 05:38

Automatic 2D optimization

I am currently engaged in the design of air intakes for a helicopter. I am looking for a 2D CFD code that can automaticaly optimise the duct geometry. Can anybody help me on this one or provide alternative sugestions?

Thank you

Greg Perkins February 22, 2001 03:10

Re: Automatic 2D optimization
I'm not aware of any CFD codes with inbuilt optimization - though some presumably exist.

In general it seems that optimization is performed by a separate optimizer that can automatically run/analyse/generate cases for the cfd code.

One program I am aware of is Ensight, which I believe can be used with any general analysis program - fea, cfd etc. I have seen some published results with Ensight used together with Fluent, to optimize cooling of brake designs/linings. As stated I believe it can be used with other programs too.

I haven't used it, though several years ago we did have a license and looking at the manuals it appears a very comprehensive, and thus complex program in its self.

I suspect that much of the work is in establishing suitable interfaces between the progams such that they can understand the problem you want to solve automatically.


Greg Perkins February 22, 2001 03:21

Re: Automatic 2D optimization
Have you tried automatic parameterization??

By this I mean write a simple program that can automatically generate cases to solve. Then write another program to evaluate each of the cases based on your design criteria.

Get the programs to generate a simple text file with the inputs vs the results then plot on a spreadshett etc.

While not automatic optimization this may tell you a lot about the process sensitivity etc.

Following from this, you could then decide to write a program which connects the processes in a closed loop to find the optimum. Perhaps use Hill Climbing etc., to find a near optimum solution, use it to generate new case files and extract some simple object function from the cfd code.

On the other hand implementing this second stage might be a lot more difficult than my quick conceptual model of the process..who knows? But the first stage should be pretty straightforward.


John C. Chien February 22, 2001 13:34

Re: Automatic 2D optimization
(1). Inlet duct has to perform a couple of functions. (2). One is to slow down the flow, so that it will enter the engine face (fan or compressor) at a low Mach number to reduce the loss. This is the function of the diffuser portion of the inlet.(area increase)(3). The other is to handle the low flight Mach number condition, when the flow has to come from all directions into the inlet, to avoid the flow separation. (including the take off case) In this case, a thick lip of 2:1 ellipse shape or something like that is commonly used. (4). To reduce the weight, you would like to have a short inlet and diffuser. But the short diffuser flow also tend to separate easier if the area ratio is large. (5). Ideally, you would like to size the inlet such that you can minimize the external drag in flight.

Jan Rusås February 27, 2001 13:33

Re: Automatic 2D optimization

I while ago I posted a similar question in the CFX forum, and was given the following link look under papers there are some about multiobjective optimization using FRONTIER.

Regards Jan

All times are GMT -4. The time now is 06:57.