torsdag 10 juli 2014

Breaking the Spell of Prandtl: From Impossible to Possible CFD

When I accepted to receive the Prandtl Medal I stated as a precondition that my views on Prandtl's legacy as the Father of Modern Fluid Mechanics should be made clear when the award is presented at the opening ceremony and then be combined by an open scientific discussion. I will in the mini-session Advanced Methods in CFD I following the opening ceremony present evidence of the unfortunate dominating influence of Prandtl as the Father of Modern Aerodynamics, which has effectively blocked progress for 100 years, under the title:
  • Breaking the Spell of Prandtl: From Impossible to Possible CFD (slides)
with the following abstract:

I was surprised of recieving the message that I had been awarded the ECCOMAS 2014 Prandtl Medal, because my work in CFD since 20 years together with Johan Hoffman and his group at KTH gives evidence that Prandtl's main contribution as Father of  Modern Fluid Mechanics, his boundary layer theory initiated in his famous 1904 article On Motion of Fluids Flow with Very Little Viscosity claiming that separation, lift and drag in slightly viscous incompressible bluff body flow (Reynolds number larger than $10^6$) all originate from thin no-slip boundary layers, is not in accordance with observations and thus is incorrect as scientific theory (listen to Prandtl expressing his discovery here with perspectives here and here and here, and comment by Euler here).

The evidence consists of computational solutions of the incompressible Navier-Stokes equations with slip boundary condition, which does not generate any no-slip boundary layers and is motivated by the observation that skin friction is small in slightly viscous flow, do agree with observations with the accuracy increasing with increasing Reynolds number. The evidence shows that separation, lift and drag in slightly viscous flow do not originate from no-slip boundary layers. 

I have thus been awarded the Prandtl Medal while my main contribution contradicts Father Prandtl himself, a somewhat unusual happening.

Our work breaks the spell of Prandtl asking for computational resolution of thin boundary layers which is impossible on any forseeable computer, which has paralyzed CFD since start, by showing that direct computational simulation of turbulent slightly viscous flow is today possible on a supercomputer and tomorrow on a laptop, simply by solving the Navier-Stokes equations with slip boundary conditions using a stabilized finite element method, without any user specified turbulence modeling. This makes CFD of slightly viscous flow possible today for a wide range of users and thus opens many new roads. 

As an example of what can be achieved, we have developed a new theory of flight exhibiting the actual fluid mechanics of flying, which has hitherto been hidden. Another is a characterization of slightly viscous incompressible bluff body flow as potential flow modified by 3d rotational slip separation. 

More complete presentations are listed under Presentations by CJ on this blog.

To get perspective on the Prandtl Medal (including Medals given to Prandtl), I advise to read and contemplate:
The net result is that I can only accept to receive the Prandtl Medal under the condition that my view that Prandtl was wrong, in several respects, is made clear together with the award.

Here is the statement I have asked to be read by me or the person presenting my work at the award ceremony:

The famous Danish physicist Niels Bohr said: “How wonderful that we have met with a paradox. Now we have some hope of making progress.” And yes, to give the Prandtl Medal to me is a paradox, or a contradiction, and as such a starting point for progress.

The contradiction is that my work together with Johan Hoffman has shown that Prandtl’s boundary layer theory, Prandtl’s main contribution as the named Father of Modern Fluid Mechanics, is not in accordance with observations and thus incorrect as scientific theory. Our evidence consists of computational solutions of the incompressible Navier-Stokes equations as the basic model for slightly viscous turbulent flow, combined with slip boundary condition as a model of observed small skin friction, which does not generate any no-slip boundary layers but nevertheless
agree with observation of separation, drag and lift for a wide range of problems.

We conclude that separation, drag and lift  in slightly viscous flow do not originate from thin no-slip boundary layers, in direct contradiction to Father Prandtl. 

Our work breaks the spell of Prandtl asking for impossible computational resolution of thin boundary layers beyond the capacity of thinkable computers, and thus opens a wide range of new possibilities for many users of CFD. I will present some of these possibilities in the session Advanced Methods in CFD I following this award ceremony.  

Inga kommentarer:

Skicka en kommentar