CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   running to mass flow/lift coefficient (

clifford bradford April 10, 2001 16:31

running to mass flow/lift coefficient
Hi, I have a simple 2-d program that I'd like to use mainly for internal flows and I'd like to modify it to run to mass flow. Also I'd be interest ed in running to lift coefficient for airfoils. I know that basically to run to mass flow (lift) that the back pressure (angle attack) is adjusted iteratively during the solution via a sort of control system. Can someone give a reference or any hints on how this is done for either case? Thanks.

John C. Chien April 13, 2001 13:23

Re: running to mass flow/lift coefficient
(1). You have several questions here, but I guess you are having problems with this simple 2-D program(a code I think). (2). A 2-D program can be anything. And I think, internal flows are largely 3-D. (3). Also, when you say "lift coefficient for airfoils", I will have to say that the problem is normally an external flow (external to the airfoil). (4)."via a sort of control system" you mean a hardware device? (5). I think, you will have to focus your question on one item at a time, in this way, readers will be able to guess at your need, and provide some hints. (6). What has this 2-D program(?) got to do with 3-D internal flows?

clifford bradford April 14, 2001 09:18

Re: running to mass flow/lift coefficient
Yes I suppose I should have focused my question better. I really just tacked on the the part about lift coefficient. As you know when running internal flow calculations in 2d or 3d (say a nozzle) it is typical to specify the inlet total pressure and total temperature and the flow angle and outlet static (back) pressure. Well the code I use at work allows you to run to a specified mass flow rate by adjusting the back pressure periodically during the calculation. The process works like a closed loop control system because the mass flow rate is not directly specified but is converged upon. In fact when you look at mass flow convergence plots the graphs looks like an underdamped oscillating system as it home in on the desired mass flow rate. Of course: (1) I can't use that code at home for my own private purposes (2) I don't know exactly how it works (3) I'd like to do some of my own coding. So I was asking if anyone out there had any information on this.

Similarly in airfoil calculations (external flow) it is possible to specify the desired lift coefficient and adjust the angle of attack periodically to attain the desired amount Cl.

Two separate questions but they had similar form so I mashed them together.

John C. Chien April 14, 2001 11:57

Re: running to mass flow/lift coefficient
(1). So, you are asking us to tell you how to implement the "control system to zoom in the mass flow" used in the code you are using at work. Since you are working for the company, could you just ask the person who developed the code for information? (2). If the exit pressure is changing all the time, the flow field will also change all the time. And the mass flow also will change all the time. (3). Actually, if you run a fixed exit pressure condition to convergence (assuming that it is a steady state calculation), then you will have the mass flow rate as part of your solutions. And we normally just run a couple of cases and determine the desired exit pressure from the target mass flow rate. After that, a third case is calculted, and in many cases, it should be fairly close to the desired mass flow condition. (4). The same principle applies to the airfoil case. Since one configuration (in your case, the angle of attack ) will give you at least one Cl. You could run a couple of cases and determine the desired parameter. (5). You could run into troubles in this kind of approach, because the relationship is sometimes not linear or unique. So, the oscillations can lead to divergence. I guess, it depends on the problem (configuration and conditions). (6). If you are trying to follow the code you are using at work, then it is easier to get the information from someone in the company, right? After all, we know nothing about the code you are using. Good luck.

clifford bradford April 15, 2001 00:36

Re: running to mass flow/lift coefficient
(1) I did and I got the answer but you know you have to be careful about these things. Such techniques even though they seem simple can be proprietary. (2)Plus having spoken to another *very* experienced CFD code developer at work who used to work at another company that the approach used at his old company was similar but different (and better) however, non-disclosure agreements make it impossible for him to tell me! or for that matter implement it in our codes! interesting huh.(3)so taking that into account I am loath to simply apply the technique that is used at my workplace because it's probably proprietary (I'd like to make the resultant code available to the public) and probably not the best.

To make along story short I'd like a technique that's good and publicly available.

I get your idea about linear interpolation, It'd quite likely work best for an unseparated airfoil considering that the Cl vs AOA curve is linear in that portion. Actually Jameson had a reference to how the Cl bit is done in one of his papers. The mass flow bit is more tricky is since of course the back pressure - mass flow relationship isn't linear. Moreover using some sort of controller actually doesn't slow down the convergence much. So it'd be a lot faster than iterating.

It's strange though John: I thought you'd have heard of stuff like this considering your experience.

John C. Chien April 16, 2001 01:31

Re: running to mass flow/lift coefficient
(1). It is likely that you are new to the company, so they are protecting themself. Or because you don't have the need to know. (2). In other word, your job does not require your knowledge of the code you are using. I think, every company has its own rules and regulations. (3). We don't want you to talk about the methods used in your work, I think, your job is more important than the forum.

All times are GMT -4. The time now is 11:27.