[qrie]

Hi,

I am simulating the development of a pipe flow. At the inlet, uniform velocity is imposed and it reaches a parabolic profile somewhere downstream and becomes fully developed. My problem, however, is that, the location at which it reaches the parabolic profile is too close to the inlet (on comparing with results from journal papers). Any suggestions on why this might be happening?

Thanks in advance,
Qrie

[FMDenaro]

Originally Posted by qrie

Hi,

I am simulating the development of a pipe flow. At the inlet, uniform velocity is imposed and it reaches a parabolic profile somewhere downstream and becomes fully developed. My problem, however, is that, the location at which it reaches the parabolic profile is too close to the inlet (on comparing with results from journal papers). Any suggestions on why this might be happening?

Thanks in advance,
Qrie

what about your Reynolds number?

[qrie]

I have tried it for Re 10 and 100. Encountered the same problem in both cases.

[FMDenaro]

Originally Posted by qrie

I have tried it for Re 10 and 100. Encountered the same problem in both cases.

you get the same lenght x/D both at Re=10 and Re=100? Are you sure that the solution corresponds to the analytical parabolic velocity profile?

[qrie]

I meant I get fully developed profiles much earlier in both cases compared to the respective values in the journal paper. Yes, the solution matches the Poiseuille parabolic profile.

[qrie]

And I checked now, and turns out - yes for both I get it at the same x/D!

[FMDenaro]

Originally Posted by qrie

And I checked now, and turns out - yes for both I get it at the same x/D!

of course something wrong must be in the code... check in the code the actual value used for the Re number, try also if Re=500 gives you again the same x/D value ... what about the outflow BC?

[qrie]

Ok, will do that. I have given convective outflow bcs.

[FMDenaro]

Originally Posted by qrie

Ok, will do that. I have given convective outflow bcs.

Assuming that the code has no bug, what about the grid sizes and discretization of the convective terms? If your grid is too coarse and you are using first order upwind, maybe you have so much artificial viscosity that overcome the real one... but I am more for some bug ...

[qrie]

The code uses second order central differences, and isn't coarse. I have refined it to the same size as in the paper. Its a direct solver, no artificial viscosity is used. Thanks, I will check for any bugs. Have already done that actually. Is there anything else that might be causing this?

[FMDenaro]

Originally Posted by qrie

The code uses second order central differences, and isn't coarse. I have refined it to the same size as in the paper. Its a direct solver, no artificial viscosity is used. Thanks, I will check for any bugs. Have already done that actually. Is there anything else that might be causing this?

I have no other idea than checking for some bug in the input data... maybe somehow the value of the Re number is fixed to a small value...check also for the divergence of the velocity if is zero everywhere

[qrie]

Yeah velocity is not diverging, however if I increase the pipe length too much it does.