Hi dear friends,

There has always been a question between the fluid solvers?

Which one of the following approaches are more suitable to solve fluid equations?

1) To directly solve the fluid equations in physical domain by finite volume approach or

2) to transform the fluid equations into a generalized coordinates system (into a computational domain) and solve in this domain.

There are some very well know benefits of (1) as they are very flexible and can easily occupy the domain of compuation. Are there any good reference out there that discusses one against the other or can any one explain on this issue.

Thanks.

M

If I am correct, the latter case is using so-called bfc system. I learned cfd by (1) but more than 10 years I have been using (2) which is more general though less intuitive at first (but the results are usually expressed by u,v,w,p and scalars in Cartesian which we can easily understand). So, personally, I prefer (2) because it's been working well for numerous models with curved surfaces. In terms of accuracy of the results, I have no idea since there are so many factors changing the results.

Selina

Hi I have a question on the subject of your question in the following way: What is our final result by doing strictly CFD and beeing appart of the physics behind, do we go far from theoretical and most analytical discussions when we go into CFD codes and methods. As an example the case of dense gas dynamics and the usage of time coupled p-version least square finite element method on the governing equations (NS, Enskog,...)

God bless you Reza

(1). If you have not studied the boundary layer theory, then it is time to do so. (2). The boundary layer theory book by Schlichting is a good reference and text book. (3). After you finished reading the book, let's know your conclusion.