Page 1 of 1

Question about constant source term in periodic channel

Posted: Thu Dec 06, 2018 8:41 pm
by Jundi He
Hi Code Saturne team:

I have a periodic channel with ribs on the wall, the mesh is shown in the photo below. The top and bottom face are periodic, and the flow is downward. I was using a constant source term to drive the flow, which is a typical method to simulate periodic flow. But I realised that near the ribs, the boundary layer of the flow is separated, which means the friction loss is changing in that area, and the pressure loss should be different in different locations, so how can we use constant source term to drive the flow if the pressure loss is changing locally. It means that if we use constant source term to simulate such flow, it's not physical. Am I saying correctly? Thanks!

Regards!
Jundi

Re: Question about constant source term in periodic channel

Posted: Fri Dec 07, 2018 2:19 pm
by Yvan Fournier
Hello,

Yes, I had the same doubts over this, and discussing it with a colleague, it seems there is actually no issue if you consider the constant source term as equivalent to a change of variable between the pressure P and "P + a.y" (assuming y is your periodic direction and a adjusted so as to make P periodic). I am not 100% sure, as I have only had the explanation in a short discussion, without a white-board or means of writing it cleanly at hand, an may have gotten this wrong, so I'll suggest to Martin to check this post and correct me if necessary...

Best regards,

Yvan

Re: Question about constant source term in periodic channel

Posted: Mon Dec 10, 2018 6:20 pm
by Jundi He
Hi Yvan:

Yes, if the streamwise pressure gradient is linear, using a constant source term to simulate the periodic flow is physical enough, as the source term is the local force (pressure drop) that drives the flow, there are lots of such typical flows: straight channel, pipe flows. But for a flow like my first post, the section area is changing in streamwise direction (Bernoulli effect), the boundary layer is broken after the ribs (friction loss changed), the pressure drop in streamwise direction is no longer linear, so is it still physical to use a constant source term to simulate such periodic flows? Thanks!

Regards!
Jundi