Hi dear scholars:

I am running a Compressible Thin Layer N.S. code for the the flow in Backward facing step. THis flow contains reciculationg zones. And my questions are:

1) are the governing equations, i.e. TLNS, valid for this geometry. If not, any clue how much should i expect my results to be off. suppose the flow is laminar flow, say Re=400 (Armaly 1983)

2) And more importantly, if the TLNS is not valid in reciculationg zones, then why people are using TNLS for the Shock boundary layer interaction or compression corners which both of them contain flow separeation.

I thank you for your contributions.

Mohammad

(1). In those days, NASA/Ames was one of the leading research laboratory in CFD. (2). Many researchers followed their approaches, and in many cases, the codes were available. (3). Nowadays, the issue is not whether it is coming from NASA/Ames. The main issue is "validation". (4). "validation" of the solution ,which one is trying to use. (5). So, don't be confused with the "validation" of the code, which is related to the programming errors. (6). A program free of bugs, is in no way related to its ability to produce a useful solution or accurate solution. (7). In the name of the CFD simulation, the exact equations and methods used is not the issue. It is whether the specific feature of the solution is useful in the process of his design. (8). So, we rarely ask whether the method is right or wrong, whether a code is appropriate or not. We are mainly concerned about the reliability of the solution, and the accuracy of the solution consistent with the equations and the assumptions used. (9). So, if a code can not produce a reliable solution, then it is useless. (10). No one is going to use an Inviscid solution to address the viscous loss problem. (11). I think, both NASA and the Air Force have done some studies about the accuracy of the solution based on the thin layer and full Navier-Stokes equations. I have seen those a few times, somewhere in AIAA journal?

tlns is probably not good for this problem. tlns as you probably know ignores the gradients in the streamwise direction. it is ok for slightly separated flow like what you might see on the trailing edge of an airfoil or other streamlined body. it's probably ok for comp corner and (weak) shock-bl interaction. for bluff bodies like you bfs i'd say it's not ok. you may still see papers using tlns code especially TLNS3D developed by NASA for several reasons (1) computational cost is less (2)the code is very popular and optimised for certain vector machines (Crays) (3)inertia - people will use a code even after it's obsolete because they've spent much time and energy learning the code and installing systems optimised to use this code. case in point: in the 80s NASA developed many implicit codes optimised for big vector machines, but by turn of the 90s parallel machines became the architecture of choice and explicit codes were preferred. as a result the money and time invested in vector codes seems wasted. there are still groups at NASA working on parallelising these old codes since they can't just throw them away.