[Fahraynk]

Hi. I'm new and feel like a newb... anyway... I'm trying to model 2d couette flow.
Distance between plates = 0.254mm
Velocity of top plate (Ue)= 61 m/s
temperature is equal on both plates.
Reynolds number = 1104
Fluid = air

When I run a simulation (Comsol) with 0<Ue<1 m/s, it converges fast. under 20 seconds. If I increase it to 20-40 m/s, it takes 20 minutes. 50 m/s takes 3 hours. I am trying to simulate 61 m/s. I feel like something is wrong, because why would it take that much longer?

I just want to see the temp distribution and heat transfer to the walls. I think it should be VERY small.

Is it becoming turbulent? I think its supposed to be laminar with Reynolds < 2300.
Also how wide does it need to be for the walls to be infinite? I noticed when I make the wall length longer it seemed to converge faster. I had it at 2 meters in length.

Time going from 0 to 1 second, increments of 0.1 Is the time to solve supposed to grow this fast?

[FMDenaro]

But when you have low velocity, the Reynolds number is lower than 10^3 and the convergence is faster. Increasing the velocity, also the Re number increases and that gives approximately the order of magnitude of the non-dimensional time to reach a steady state. Eventually, at higher velocity, you can start to find oscillation in the solution.

[Fahraynk]

So you are saying that it sounds correct... So the time to solve should scale exponentially with RE in a 2d coutte flow? I guess it could be true I just wasn't expecting that! I was thinking I'd need a high mach number or turbulent flow for that.

Do you know anything I can study to help me understand how RE scales time to solve?

[FMDenaro]

Well, I can just suggest to read some texbook of basic fluid mechanics, the meaning of the Reynolds number has many aspects ...It not only gives the rate between convective and diffusive fluxes of the momentum but you can also see it as a ratio of two characteristic times (as well as lenght scale)