Reduced Order Flexible Solid

I’ve been trying to create a flexible multibody system using the new SimScape block: “Reduced Order Flexible Solid”. However, I have a few questions regarding the inner workings of this block, since there is very little information available online. I have got the block working for two different systems. I used Craig Bampton reduced mass and stiffness matrices and a damping matrix according to: “sm_flex_body_lib_fei_form_def_matrices.m”. The interface frames are the same as the boundary nodes in my reduced matrices and no internal modes have been used for simplicity. I use the daessc solver with variable time step.

Now I’ve noticed that the Reduced order flexible solid splits the body in the same number of parts as interface frames, but how does it decide in what way the body gets split. When I tried to model a flexible rod with 3 interface frames, one on each end and one in its center of mass, it appears that the body is split in 4 parts, where 2 are rigidly assigned to the middle frame. This seems rather logical to me. However, when I tried to model a symmetrical plate with 5 interface frames (2 on each long side) something strange happens, as shown in the figure below.

As expected, the plate is broken into 5 parts. However, I expected 4 of those parts to be identical, and only the middle section to be different. But during the partitioning of the part, it appears that a hole was created, as shown in the encircled part on the right. This brings me to these questions:

• How does the Reduced Order Flexible Solid block make the partitioning?
• How are the masses and inertias distributed? In the technical paper “Modeling Flexible Bodies with Simscape Multibody Software” 3 different methods are proposed on pg. 23. Does the Reduced Order Flexible Solid block use the third method or something else?
• If I wanted to make a comparison between the Reduced Order Flexible solid and the “traditional” method using the state space matrices, how should the mass and the rotational inertias be distributed over the inertia frames?

The old method for modeling flexible bodies uses the state space matrices according to the scheme shown in the figure below. In this scheme, there is a filter in all flex subsystems.

• Is this filter also present in the new Reduced Order Flexible Solid block? Because all simulations with this block take very long due to noise.

Thanks in advance!

[Steve Miller]

Hi Vincent - can you share this model?

--Steve

Answer 1

[Steve Miller]

Hi Vincent,

Your screenshots indicate very unusual behavior. We would need to look at the files directly to understand what is happening with your files and to answer your other questions. If you can upload them, we can look into it further.

--Steve