Learning Forum

[Georgiy Tanasov]

Hello everyone, Im studying an airflow inside a finned tubular heat exchanger at various inlet velocities in Fluent with real gas model. Starting at 10 ms inlet velocity it runs smoothly and converges (and i get right results) but with increasing velocity it starts to diverge (at 12,5ms) and i just cant explain why. Ive tried many different things: k-epsilon and omega SST models, enhanced wall treatment or without, different inlet turbulence intensities and settings (initial pressure), restricted backflow reversal, lower relaxation factors, etc.. Im using NIST real gas model of air. My initial guess was that it has to do with the mesh quality and y+ value, but after doubling the number of cells i still get the same results. Ideal gas model works fine at even larger velocities. Analyzing the results, my turbulent Reynolds number just skyrockets at the end of the domain and the inlet pressure is way too low (around -20kPa when I have 0Pa outlet condition doesnt make any sense, it should be around 5kPa). Furthermore if i put mass flow inlet corresponding to the approximate value for the same velocity, the simulation converges normally. Is there a problem with NIST real gas model and velocity inlet? And why does it happen at a such specific velocity value? I would appreciate any help .

[Rob]

Is there a reason for using a real gas model? If the cell quality is good, how well resolved is the mesh? What boundary conditions are you using? What are you trying to find out? Some images of the model may help. 

[Georgiy Tanasov]

There are some points in the domain where local velocity exceeds 100m/s and i wanted to account for compressibilty and just for good quality results, of course i could use another model but im just wondering why it happens. Inlet temp is 313K and wall temperatures are 299K. Amount of cells started wit 420k and i increased it gradually to 1,2M and no change of the behaviour, it was always the same point of 12,5ms where it diverged (12,4 still converged). Im studying pressure drop and hea transfer rate

[Rob]

You could be switching to a different flow pattern, ie separation which gives a significant acceleration in the flow too. 

[Georgiy Tanasov]

How can it be explained then, when i put mass flow inlet which coresponds to 15 m/s inlet velocity it converges normally?

[Rob]

What initial conditions did you use in each case? Pressure in & out means the mass flow is part of the solution, and as density also varies you've got to get a good initial solution. 

[Georgiy Tanasov]

So ive tried different initial conditions, different inlet velocity (10-17ms) and every time after 12,5 it wont converge, then ive put mass flow inlet with a value from const air parameters study and it resulted in a inlet velocity of 14 ms and converged normally