Looking for a "CFD for Designer"
 

Hello!,
I am looking for a CFD code targeted to be used for the CAD Designer running a Windows desktop computer, the typical user of 3-D mechanical CAD package like SOLID EDGE ST5, not a CFD expert.

The requirements are: easy to use, modern user interface with a good material database, with internal units management, windows native, not expensive at all, capable to solve conjugate fluid flow-thermal models simultaneously with conduction, convection & radiation, faster solver, a good & quality mesher, and report generation.

Please let me know you experts what options exist in the market -- thanks!.

Best regards,
Blas.

In german we call this the "eierlegende Wollmilchsau"

Apart from the constraint of being cheap, Ansys CFX might be the Software which is easiest to use for beginners.
But ease of use is exactly what makes commercial CFD software expensive.

@BlasMolero, I think there may be more to this than you realize...

But, first, I agree with flotus1. You get what you pay for.

Second, you need to address the physics, and the physics can get complex and painful for some simple geometries very quickly. For example, a cylinder. If the problem is to accurately calculate the drag over a cylinder for a range of Reynolds numbers, it will be costly in the sense of CPU, numerical methodologies, and experience. The actual process of gridding and code settings (assuming the code has the correct ones) is no problem.

I guess the question is, what sort of accuracy do you need?

If you are just looking for qualitative answers (colorful fluid dynamics), I believe there are packages out there.

Hello!,
Please understand me, things cost its value, simply I am not up-to-day with names of modern CFD codes, and this is the reason why you CFD experts knows the market very well.

I am looking for "intro" CFD codes in the range of <10,000€, I do not need "free surfaces", neither "rotating frames of reference" or "non-newtonian fluids", etc.. but a CFD code to compute pressure drop in a pipe with valve, or to solve the coupled thermal & flow problem on a PCB with a chip with natural convection, simply things like this ....

Best regards,
Blas.

That's the unfortunate thing, accurately calculating the pressure drop due to a valve is complex. It depends on Reynolds number, separated flow, turbulence, etc. It's not only about the geometry, but the physics too. What physics are you trying to capture?

If that's the sort of thing you're interested in, it is my opinion that you need a professional code (Fluent, etc.) and a lot of CPU power.

An example. Do you need LES (Large Eddy Simulation) or DES (Detached Eddy Simulation)? Or, is a RANS turbulence model enough?

Dear Martin,
I suppose that any modern CFD code of nowadays is able to arrive to similar solutions when solving not extremely complex problems, a CFD code using the Reynolds Averaged Navier-Stokes (RANS) to solve for turbulence, supporting a two-equation K_Epsilon well-know turbulent model is perfect to solve internal, steady state fluid flow problems. I like Ferrary, but for not possible for evident reasons...
Best regards,
Blas.

PD
In summary, not any suggestion for names??

There are OpenFOAM front ends such as Symscape (http://www.symscape.com/ using OpenCascade) and Khamsin (http://www.hibouscientificsoftware.com.au using Google Sketchup). There are Cartisean solvers such as Karalit http://www.karalit.com/ but I don't recommend that type of solver since I have not seen much Navier Stokes V&V on them. And, I assume, you are familiar with the main vendors such as ANSYS, Mentor, Metacomp, and CD-adapco.

That being said, the statement "I suppose that any modern CFD code of nowadays is able to arrive to similar solutions when solving not extremely complex problems" is not really true. That is why I mentioned the cylinder.

I'm looking forward to other responses to this thread!

Dear Martin,
Impressing the offerings that exist in the market!!, it cost, but finally we have some explicit contributions to "CFD for the Designer". I will take a look to them -- thank you!.

Best regards,
Blas.

There are a growing number of commercial CFD packages that are closely coupled to commercial CAD packages. They usually generate grids fully automatically removing this as a step the designer needs to consider. To facilitate this they tend to adopt one of several methods based on non-boundary conforming grids which tend to sacrifice solution accuracy for ease of producing solutions. Whether this is a good trade depends on what the simulations are being used to achieve.

Commercial packages can usually "automatically" produce reliable solution for linear problems like stress and heat transfer enabling engineers with a relatively weak grasp of the fundamentals to produce reliable and useful simulations. This is much less the case with CFD where significant non-linearity and large differences in the scales of contributing physics makes having someone in the loop with a good grasp of what is going on in the simulation much more important if useful and reliable solutions are to be produced. Part of this is the CFD supplier making the details of the assumptions used available to knowledgeable users and this is often not forthcoming for the "automatic" CFD codes closely coupled to CAD programs.

I do not know enough about your usage to state if the pros outweigh the cons for the cut-down CFD solvers. If CFD is to be a large part of what you do then the "full" CFD packages are likely to be better value in offering wider support and enabling you to improve the accuracy and scope of the simulations as your knowledge and ability grows. Having just followed the link to IBERISA (IBERDROLA once funded a visit to Spain for me many years ago) I think I may not have been answering the right question.

I would not reccomand it for anything serious, but if easiness of use is one of your major request, it might be worth a call to your CAD vendor. probably there are modules for very basic fluid dynamic simulations.

for example, I know for sure that solid-works have something like this...

it might be very simple, very basic, but cheaper than the big names.

BUT I would ask for verification and validation on your specific cases, compared against experimental results... and still I would have my doubts...