[husker]

Hi,

In the previous posts I read about XFLOW software.

I don't know or imagine how physical equations can be solved numerically without the finite element approach?

Could any expert write some about this method? How does it work, what are its advantages and disadvantages?

[cfdnewbie]

Yes, would be nice if somebody in the know could tell us the basics...basics. advantages, disadvantages, efficiency, limitations...

no propaganda, just the plain facts.

would be very much appreciated!

[Saket.Chaudhary]

Guys,

I can tell you my experience and try not to sound like sales,
I have experience with multiple CFD software like CFD-ACE, CCM+, CFX, CFdesign etc used either in school or professional career both as a user and working for CFD companies.

The biggest time sink/pain was to get the geometry in the CFD software, Clean up(Either in a CFD software or if i am lucky in CAD software), then meshing ..... this represented 80% of the time in the simulation cycle. Exponential increase in computation power has led to simulations getting bigger and complex in size in today's world and while meshing technology are making good stides this still requires many man hours which are becoming more expensive than computer hardware.

You can probably see the advantages now
- No meshing, step completely eliminated
- Very lenient to dirt CAD ( intersections, duplicate surfaces etc)
- Complex motion problems ( no remeshing )
- Linearly scalable for parallel simulation

As with any new technology I am sure there would be some areas of improvement. We are everyday working with customers and running real life problems. The feedback has been tremendous right now....not hard to imagine that the first reaction is WOW and then they want to see their own problem in XFlow. With the real life problems we running everyday the performance has been exceptional in terms of time savings for customers, speed, accuracy.

@ Husker: Obviously I can share the technical details but if you can Google it you will see the amount of research on this with Messless and LES really the next step in CFD.

if you guys want to have a deep dive into XFlow send me a note saket.chaudhary@mscsoftware.com

[cfdnewbie]

Saket,
thank you for your comments. I agree that pre-processing can take up a large amount of your computational time, but it also offers a huge chance of getting to the "right" result. I guess we both agree that a bad mesh can lead to very wrong results, if you are way off, even grid refinement won't tell you sth is wrong.

So here's my question:
On what physical/mathematical principle is your code based? Surely you must ensure some kind of conservation?

Secondly, efficiency and speed is not all in CFD.... how does your code compare in terms of accuracy per DOF? There must be some research papers your work is based on in journals like JCP or such... could you point us to them?

thank you!

[husker]

Thank you Saket.

I will be reading about this subject through academical papers. This topic is very interesting to me while it has some basic differences. However, I actually feel shamed because of lack of my knowledge about this technique.

[gama]

Has anybody found the physical/mathematical principles and the CFD formulation of the XFLOW yet? (I am interested in a deeper level than what XFLOW can do or what is its advantages). Can you post a link to a paper or documentation which talks about XFLOW formulation?

[Saket.Chaudhary]

@ Gama. Send me an e-mail at saket.chaudhary@mscsoftware.com.

[cfdnewbie]

Originally Posted by gama

Has anybody found the physical/mathematical principles and the CFD formulation of the XFLOW yet? (I am interested in a deeper level than what XFLOW can do or what is its advantages). Can you post a link to a paper or documentation which talks about XFLOW formulation?

no, sadly, that's kind of under wraps or by invitation only... would be interested in the scientific basis as well! sounds too good to be true, and the videos they posted are very impressive. Just let us look under the hood

[Jing_min]

Originally Posted by Saket.Chaudhary

Guys,

I can tell you my experience and try not to sound like sales,
I have experience with multiple CFD software like CFD-ACE, CCM+, CFX, CFdesign etc used either in school or professional career both as a user and working for CFD companies.

The biggest time sink/pain was to get the geometry in the CFD software, Clean up(Either in a CFD software or if i am lucky in CAD software), then meshing ..... this represented 80% of the time in the simulation cycle. Exponential increase in computation power has led to simulations getting bigger and complex in size in today's world and while meshing technology are making good stides this still requires many man hours which are becoming more expensive than computer hardware.

You can probably see the advantages now
- No meshing, step completely eliminated
- Very lenient to dirt CAD ( intersections, duplicate surfaces etc)
- Complex motion problems ( no remeshing )
- Linearly scalable for parallel simulation

As with any new technology I am sure there would be some areas of improvement. We are everyday working with customers and running real life problems. The feedback has been tremendous right now....not hard to imagine that the first reaction is WOW and then they want to see their own problem in XFlow. With the real life problems we running everyday the performance has been exceptional in terms of time savings for customers, speed, accuracy.

@ Husker: Obviously I can share the technical details but if you can Google it you will see the amount of research on this with Messless and LES really the next step in CFD.

if you guys want to have a deep dive into XFlow send me a note saket.chaudhary@mscsoftware.com

Hi
What are the advatages of LBM over SPH?
Can you compare them?