A few years ago, there were some interests in powerflow (I came across some powerflow sales men who claimed powerflow will take over the CFD world). Now I search the database, the last time powerflow was mentioned was in 1999.

Could anyone tell me whether the company which provides powerflow is dead now or still alive ?

Yes

It very much alive, Visit Exa's site

Apurva

Was it lincoln who said something like the one below:-

you can fool some of the people all the time and all of the people some of the time. You cann't fool all the people all the time. So it's only a matter of time......

"...I came across some powerflow sales men who claimed powerflow will take over the CFD world..."

If I had a dollar for everytime a salesman made such a claim to me, I'd be in Margaritaville right now.

A few years ago, they DID claim that, not only their sales men, but the whole company believed so. Suddenly they found the life in CFD is not SIMPLE. I belived there was a discussion in this forum about 2 years ago, and someone from Exa said people made this claim had left the company long time ago!

I believe the sales men from other companies, to a less extent, behave the same anyway.

The method used by PowerFlow has its attractiveness, and PowerFlow seems to be very competitive in simulating external aerodynamic flow.

John,

I agree with you.

Apurva

There was some discussion a few months back concerning powerflow and exa with some links to powerflow vs. lattice boltzmann vs. finite difference for the flow over an aerofoil. Powerflow did not perform well. However, I get the impression (but may well be wrong) that they have quietly dropped much of their lattice gas stuff and moved towards a lattice boltzmann method.

Their initial predictions of flows over car-shapes did appear superior to RANS predictions but they were not open about what they were doing near walls. However, the predictions did not seem superior to equivalent LES predictions and, I suspect, the flexibility of LES would win in most circumstances.

Hei,

I am working on Lattice Boltzmann methods for a few years. During the work I came in touch with Powerflow (who had a quite bad marketing!).

AFAIK Powerflow still uses a LES type turbulence model. You should keep in mind that each computational method can predict turbulent flow structures as good as the underlying turbulence model.

If you have weak points in your turbulence model, they are present regardless the method you are using. So, Powerflow can perform only as good as the (Smagorsinky ?-)LES model they are using. And YES, even these "gas kinetic based" schemes need wall functions for turbulent flows as long as the boundary layer is not resolved (which requires quite a lot of cells as in every other CFD method).

Regards, F&S

>>> AFAIK Powerflow still uses a LES type turbulence model. You should keep in mind that each computational method can predict turbulent flow structures as good as the underlying turbulence model.

Although true for RANS flow (with signficant turbulent stresses) this is misleading for LES. For almost all the solution region of most engineering flows (but not everywhere unfortunately) the accuracy of the turbulence model is irrelevant because the terms are very small and do not influence anything significant. What matters is the accuracy of the convection terms and here "traditional" LES can win signficantly over a lattice boltzmann approach. Also at low Mach numbers the "standard" lattice boltzmann approachs (if such things can be said to exist) are explicit, limited by the symmetry of the grid, limited by how the process of diffusion is represented,... These are all real world problems that can be addressed straightforwardly in "traditional" LES.

My point about wall functions did not concern their use (wise in many cases) but that exa would not tell me what they were doing near the walls. (Strictly that ought to read exa promotional people - I never tried to access the technical people). Now nobody can perform reliable simulations without knowing what assumptions are being used by the model. This requirement to use the code as a black box is inevitably going to prevent able users from doing anything serious with the code.

As an ex-insider, my opinion is that Exa's days are numbered because:

1. Exa convinced Venture Capitalists (VC)with claims such as John mentioned (taking over the CFD world etc etc) and managed to borrow millions of dollars from VC to start the company.

2. After over 10 years, exa is still not able to take over the CFD world. As far as I know, they only have a handful of clients (most of them are from car exterior aerodynamics). Now VC want their money back. However there are not a lot of option available to VC. VC tried to sell exa, nobody want to buy it (unless it cost next to nothing).

3. If exa cann't take over the cfd world soon, VC will pull the plug (Vc are not long term investors, 10 years are already too long for them).

This is why exa's days are numbered and the lessons here is: you cann't build fundation with lies, sooner or later, people will find out the truth......

I've done some study on turbulence modelling before

I'd agree with F&S, if exa still use turbulence models, then it should have similar problems facing the traditional codes. I am not an expert on Lattice gas/Boltzmann, however it seemed to me all the big talks about powerflow being the 21st century tool because it uses lattice gas/boltzmann is more of marketing gimmick if it still relies on the GOOD OLD TURBULENCE MODELS.