CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Pulsating flow with non-reflective inlet boundary (http://www.cfd-online.com/Forums/cfx/63971-pulsating-flow-non-reflective-inlet-boundary.html)

cc1000 April 25, 2009 06:43

Pulsating flow with non-reflective inlet boundary
 
I am wondering if someone could give me some advice concerning how to implement a non-reflective boundary.

I am trying to model a turbocharger with a transient moving mesh rotor fed by a stationary inlet duct (volute). The total pressure inlet boundary condition is an unsteady, time-dependent pulse similar to an exhaust pulse. Ultimately, I am trying to model conditions that I have measured experimentally. I have measured the pulsating pressure at the inlet to the real turbocharger and I want to be able to apply this pressure to the model and obtain a mass flow.

However,

1. If I select a standard inlet pressure boundary condition (non-reflective feature NOT enabled), from what I understand, this standard inlet BC does not allow pressure wave reflections from within the domain to travel back out through the boundary (as they would in the real case). The applied pressure pulse is of high enough amplitude that a failure to model acoustic reflections poses a potentionally significant innaccuracy in the model.

2. If I select a non-reflective (transmissive) inlet boundary condition and apply the pressure that I have measured in the laboratory at this location, the resultant values of pressures are totally different to my measured data. This is due to the fact that the specified BC actually must represent a 'far-field' value, and thus the solver does not strictly enforce the value that is applied.

So, the way I see it, in order to use the non-reflective boundary I must be able to find (or calculate?) the appropriate 'far field' value so that when applied, the resulting pressure at the inlet BC represents the experimentally measured values. So the obvious question is - how do I do this ? Is there a way to iterate at each time step so that the appropriate far field value can be calculated in order to reproduce the measured pressure at the inlet BC ?

Any advice would be appreciated. Especially good would be any literature someone could point me to where this sort of problem has been dealt with.

Many thanks.

Luk_Fiz April 26, 2009 14:50

Dear Sir,
I am not sure if understand You well: You mean that You have measured pressure time-profile that inculdes "primary" (upcoming) and "secondary" (reflected) waves? The problem is to subtract the primary component of these to put it as inlet BC?

Luk

ghorrocks April 26, 2009 18:53

Hi,

I think there is a new beta feature in V12 to implement non-reflecting boundary conditions. I have never used is so know nothing about it. Talk to your CFX support person for more information.

Glenn Horrocks

cc1000 April 27, 2009 07:05

Yes, I have measured the pressure using a standard pressure transducer in the inlet pipe. This will, as you say Luk, measure the summation of all the acoustic action (incoming, returning, etc). However, in talking to the ansys support, they indicate that the new non-reflective boundary condition feature requires applying a 'far field' pressure pulse (ie, no influence from acoustic reflections). This is, in reality, hard to measure.

So I am trying to figure out a way i can calculate the required pressure to apply such that the resultant pressure matches the measured profile.

Luk_Fiz April 27, 2009 07:30

At first, I must say that I dont know the answer for Your question. In my opinion You are right that it is rather physics not a CFD problem.
In my opinion a solution (if exist) can be difficult cause the measured pressure profile mixes impinging and reflected waves. The reflected wave clearly depends on Your arrangement (the device You want to characterize with CFD). The only solution I see at the moment is:
- try to measure pulses with something like "time of flight" technique, I mean in 2 distances upwind Your arrangement, with good time resolution. This can help You to establish the shape of incoming pulse.
- try to fit somehow. For example assume a sin(t) pulse and see the pressure profile after calculation does it seem similar, and again....

In my opinion it is difficult question and I would be happy if someone can add something useful.

Luk

cc1000 April 27, 2009 07:57

Thanks Luk, for your replies.

I can see what you are saying that it is a physics problem, but it must be a situation that is (or will be) encountered often when trying to use this non-reflective boundary in CFD. I know of one other work where this was an issue - but using Star-CD. I don't have access to details, but i believe an iterative technique was used whereby the difference the applied pressure and the resultant pressure was used to reapply a different pressure. I am unclear how to do this in CFX or really how this would work. Thus, my post was just in the hope that someone else might have had to deal with a similar problem.

I take your point that i could somehow use a "time of flight" technique, but I would rather find a numerical solution just because of experimental logistical issues.

Thanks for the posts anyway.

Luk_Fiz April 27, 2009 08:10

Well, I agree that it should be a problem oftenly faced by peoples involved in pressure impulses simulation but personally I have no such simulation in my portfolio:))))
From Your post it is also unclear for me, how these technique from STAR CD team should work, but maybe there is some trick.

Anyway, If You have some idea post Your solution here.

Luk


All times are GMT -4. The time now is 04:17.