I am working on a structural adapter and I’m experimenting it with some of the tutorials. When simulating something like the vertical flap with SU2 euler flow, I obtain something like the video below, which at least qualitatively looks physically reasonable:
Going into the data of the simulation, I recovered the data from the interface mesh and imported them together with the fluid flow. During some timesteps find situations like the one below:
Pressure is about 1 bar around the flap, which is 1m tall, divided into 40 segments. Sometimes the x component of the interface nodal forse is as plotted: positive on the right and negative on the left, which seems odd. Besides the nodal force is much lower than what I would expect (approx 10e5/40). Am I missing something?
is this potentially not a visualization issue where the the value is referenced to some normal direction? Which structural solver is this? What could be happening in the structural solver is the value is given in some direction relative to the face normal. If this is the case, when the normal is facing outwards from the structure, and the force is going inwards, it could be taken as negative and that could be what is being plotted. Therefore the force is negative to the face normal, but still in the correct direction.
To check if the actual forces being sent from preCICE are correct, you could plot the values in the x,y, z direction and check that the direction of the forces during the simulation runtime are actually correct? If the structure is moving correctly (from initial looks), that is a good sign.
thank you for your observations. I’ll try to answer to some of your questions.
The problem is 2D. The force data are given in x,y component. I know that the structure is bending, but not so much. I believe that the components should remain aligned with the axes. I am not able to bend the interface mesh as SU2 works with displacement deltas, I should integrate them to obtain the deformed structure.
As to the structural solver, I am using MBDyn. Anyway I am looking at the data coming from SU2 to the PreCICE mesh (which is what I was plotting). I believe that all this happens before the data are passed to the structural solver.
The PreCICE interface meshes, which I have at each timestep, converge to very close values both for forces and for displacement-deltas. The structure bends in the “correct” direction. Fluid is coming from the left, in fact pressure is higher on the left face of the structure and lower on the right side. What puzzles me is the magnitude and direction of forces in x direction (the flow direction) viewed at the interface mesh. I am just wondering if the simulation has some physical flaw. I have been looking at the same thing in the SU2-CalculiX tutorial, adding <export-vtk> at each solver. Maybe in this case the structure bends “too much” and the mesh is too coarse to understand, but the situation I am describing is visible there too.
So to try and understand, in the figure in your first post, you are plotting the fluid pressure, which seems correct (larger on the left of the structure), and you are also plotting the force on the interface of the fluid solver (which seems like it is pushing away from the structure)?