Discussion Forum

Can you provide a simple example? There should not be any difference.

Hello Henrik,

Thanks for your reply. I am making one example and will get back to you soon.

Thanks Pratik

Hi Henrik,

Here is a sample mpdel I created in COMSOL (attached) where water is flowing over a membrane. The model is created as a fluid-structure interaction model by using 'fluid-membrane interaction' module. I have created the 'intop1' function as seen in another comsol tutorial (screenshot attached) to calculate lift and drag and I used this technique for several models without any problem including fluid-solid interaction or CFD models. But somehow when you use fluid-shell interaction or fluid-membrane interaction, then it is not coming through which I have seen for every models that has shell or membrane involved. For this model attached, I am getting NaN for drag force (screenshot attached).

I was thinking about the reason behind this and one thing that came to my mind was, when we used fluid-solid interaction or CFD models, then the surfaces on which you want to measure the drag were to be specified and each of those surfaces had some sort of thickness depending on which you could select upper or lower surfaces. However, here for the membrane, since it has no thickness, therefore, when I selected the face of the membrane as the surface of interest where I need to calculate the drag, there was only one selection of face and I am not sure if it selected upper or lower surface of the membrane. Could this NaN value of drag be because of this problem (I don't think so)?

As you mentioed earlier, the technique should work similary for all models, but here for some reason it is not working. If you could put some insight on this, I will greatly appreciate.

I have cleared the solutions to reduce the file size, but if you want you can run it. It just takes 40 seconds to run.

Thanks Pratik

When you have a membrane that is embedded in the fluid, there will be two different tractions on the upside and the downside. The ordinary traction variable is not defined there.

The expressions you need to integrate are spf.T_stress_dx+spf.T_stress_ux etc. Here 'd' and 'u' stands for 'down' and 'up' respectively.

Hello Henrik,

Thank you very much for your feedback.

Our COMSOL software was unsinstalled for the last few days due to some maintenance and we were not able to run the model as per your feedback. However, when we tried it for the earlier few days before maintenance, it was giving an error and we were not able to diagnose it further as the sytem was down. I will be running it from today and at first, I will try to fix the previous errors. If the error comes consistently, then I will get back yo you.

I appreciate your time behind this.

Thanks Pratik

Hello Henrik, We did some simulations as per your instructions to calculate lift and drag for the FSI model that I sent you in the previous email (please refer to that model). However, we are confused about some other things here. In the previous simulation, our direction of velocity at the inlet was horizontal, i.e., parallel to the shell. But this time, we need to change the direction of velocity at an angle (angle of attack, AoA). I know that for this case, I have two options to do it. One is to rotate the entire geometry at the desired AoA with the x axis or another is to use the two components of the inclined velocity without rotating the geometry (attachment ‘velocity components’ for a nonzero AoA). I prefer the second method and therefore, I think the expression of the velocity components something like the one mentioned in the attachment ‘velocity components will serve the purpose. In this regard, I have three questions: 1. According to the guidance that you provided in the past, we are using the ‘intop1’ function as below for a 0 degree AoA:

a) For lift: intop1(spf.T_stress_dy+spf.T_stress_uy) b) For drag: intop1(spf.T_stress_dx+spf.T_stress_ux)

Will this expression remain same for an inlet velocity with a nonzero AoA? If not, how this expression will be changed?

2. From my fundamental understanding, I know that when there is an external flow over a body, lift and drag is calculated by using the normal component N and the axial component A (parallel to the chord or plate) as shown in the attachment ‘formula for lift and drag’. Does COMSOL also use the calculation of N and A that way? I mean, by using the ‘intop1’ function, if we are able to get two forces, then will those two forces be N and A anyway from the ‘intop1’ function as mentioned in question 1 and if yes, then do we need to calculate the lift and drag as per the formula in the attachment ‘formula for lift and drag’ manually?

3. We tried to use the application gallery of COMSOL to find a suitable tutorial to get some help for calculating the lift and drag regarding an inclined plate but did not find any appropriate source there that used a nonzero AoA. The closest thing we found is the tutorial ‘Flow Around an Inclined NACA 0012 Airfoil’. However, in that tutorial, they used line integration instead of ‘intop1’ function and it was used in the results section, not in the ‘definitions’ section where the ‘intop1’ function was used. I understand that they used line integration because it was a 2d body instead of a 3d body, but I was wondering what’s wrong with the ‘intop1’ function in this case? I tried to use that way to find the lift force, but it was giving me a very small value. For example, in that tutorial, the lift coefficient found for an 8 degree AoA was around 0.866 (attachment ‘airfoil tutorial’) and then the lift force should be 0.51.20(1.81)50^2*0.866 (please refer to the parameter section of the tutorial for values) which shows the lift force as 2338 N. However, if we use the same command as mentioned in the first expression in question 1, it gives us too low a value like 10^-23 N range. I am guessing that I am missing something here. Do you have any thoughts on that?

For our case, we need to run the 3d simulation anyway and if we need to use the command that the airfoil tutorial used (screenshot ‘airfoil tutorial lift coefficient expression’), then could you please help us write the appropriate command for a 3d body for this FSI model with shell or membrane having 2 sides? I can see that the tutorial used a pressure coefficient p in the expression, but I am not sure where that variable is located in the model.

FYI, the attachment ‘velocity components’ mentioned in question 1 is a screenshot taken from the airfoil tutorial model referred to question 3.

Any help is greatly appreciated.

Thanks Pratik