Simscape Spatial Contact Force no contact signal when minimum distance is 0

I'm modelling the collision between a plate and the L shaped mesh shown in figure (aircraft winglet).

The plate is a brick solid of dimensions XYZ [0.7 0.64 0.005]m, with a mass of 2.5kg.

The L shape mesh is made of 2 distinct components assembled as follow:

The mesh files are 2 STLs, each one have already been prepared as convex hull in Blender.

As you can see the minim distance approaches 0 but the contact signal is never triggered.

This is where it gets weird: if you replace the brick solid with a STL of the plate from Blender with same dimensions and mass you still get no collision. So you change the geometry of the winglet with a unique convex hull, yet no progres. If you increase the volume of the winglet, such as extruding in the X direction, collisions start to get recognized, but only after a certain amount. Every other part of the helicopter works fine.

For what I could understand this issue does not depend on:

1) The geometry of either the plate or the winglet

2) The position in space of winglet, as I tried different configurations using rigid transforms

3) The integrator, I've used ODE15s, ODE23t, ODE45 and they all behave the same (I lowered the relative tolerance and set maximum step size of 1 ms).

4) Switching on and off frictional forces

5) Changing Spring stiffness and damping, except for one case in witch

The spatial contact force works perfectly in the same model with the same plate but different components, all connected to the same Follower signal line.

Final note: I've imported the two meshes in the steering wheel example, both the plate and winglet collide with the plane, but not with each other. Using point clouds don't change anything, and I need the contact signals for my analisys so it's not really an option. So I tried to run a Rigid Body simulation with Blender and they work fine, so the problem lies in Simscape.

[Riccardo Capellino]

5) Changing Spring stiffness and damping, except for one case with a very high stiffness (1e10 N/m) I can get the first contact, the plate bounces once, then gets through the object.

[Riccardo Capellino]

I just noticed that when the plate hits the tail rotor blades only 1 of the 3 blades (we'll call it blade C) resolves the collision, the other two just pass through the plate. The mesh loaded in the file solid block is the same, I've just rotated every blade by 120° with a rigid transform, yet only C works.

[Riccardo Capellino]

I think it has to do with the projected areas seen by the follower on base.

In the pic above the pink curved plate is dropped on a square plate of dimensions 1.12x1.12x0.01 m. No collision is detected and the plate gets through the square.

When I increase the size to of the square to 1.2x1.2.0.01 it works perfectly.

[Yifeng Tang]

If possible, I suggest that you post the actual model so the community can better help you diagnose.

[Riccardo Capellino]

Here's the model

[Yifeng Tang]

Hi @Riccardo Capellino,

It's missing a "AW169_EPP.008.stl" file.

But if I replace it with the part.stl file in the folder, the simulation looks OK to me.

While I've no idea what's being simulated, the red panel does seem to properly fall onto the grey wedge, bounce off and collide again, and then slide off.

And I'm not sure what that point cloud is supposed to do. Could you please check if the package is complete?

[Riccardo Capellino]

AW169_EPP.008.stl and part.stl are the same file, just renamed. The cloud point was there to test if it worked (it did not), I left it there because I didn't need it anymore but maybe someone could have found it useful.

Can you tell me what Matlab version are you on? I've tested this model in both my 2 laptops with 2023a and it never worked.

[Yifeng Tang]

MATLAB Version: 9.14.0.2674353 (R2023a) Update 7

The simulation works on my end with the ode23t solver you specified. If left to "auto", it'll choose ode45 and it works as well.

Answer 1

[Riccardo Capellino]

After a clean installation of Matlab 2024b the problem seems to be gone.