🎉Community Raffle - Win $25

An exclusive raffle opportunity for active members like you! Complete your profile, answer questions and get your first accepted badge to enter the raffle.

Join and Win

Disappearance of rigid body modes

Karthikeyan · 2025-05-16T13:54:15+00:00

There was an error rendering this rich post.

Karthikeyan

Hi,

I am trying to run a modal analysis of a two wheeler vehicle.

When i ran the modal analysis (through linear stability event) with only rigid bodies for suspension and frame i was able to see sprung modes like pitch, bounce, weave, wobble and the unsprung modes like hop.

However when i built a full vehicle model with all flexible bodies based on the FE mesh from the CAD surfaces for the suspension and various other parts that make up the sprung, the modes seen with the rigid body model have disappeared in the non zero frequency range.

Instead, these rigid body modes now got shifted to zero frequency with 100% damping. I am not able to understand how motionsolve is calculating these modes. With the flexible model, i can see upto 50 zero frequency modes with 100% damping.

Can you please offer some insights?

Find more posts tagged with

MotionView/MotionSolve

Accepted answers

All comments

Christopher_Fadanelli

Hi @Karthikeyan ,

Including some tutorial links: Link_1 , Link_2

One thing to verify is if the model is constrained properly.

Next, the A matrix can balanced. You can change the balancing in the solver parameters.

For certain models, the eigenvalues computed by MotionSolve can be extremely sensitive to small perturbations in the A matrix. This occurs because the eigenvector matrix, V , for such a model is highly ill-conditioned. As a result, even seemingly minor changes for such models may result in large changes in the eigenvalue solution.

K(V)=∥V∥^−1∥V∥ is the condition number of a matrix V . Consider the linear equation Vx=b.V=b . K(V) measures the rate at which the solution x will change due to a change in b. If K(V) is large, then even a small change in b can cause large changes in the solution x. In such as case, the matrix V is said to be ill-conditioned.

To obtain a robust eigenvalue solution in such cases, MotionSolve must balance the A matrix by finding a diagonal similarity transformation DAD^−1 such that the row and column norms of A are numerically close. By doing so, the eigenvector matrix becomes better conditioned and the eigenvalue solution is more robust to numerical noise in A.

You can also disable damping from all force elements from the eigenvalue solution, update the minimum and maximum frequency to filter the modes, change the cutoff_type from UNDAMPED to DAMPED.

For more reference on altering the Linear Solver Parameters, please refer to our help page.

Best,

Christopher Fadanelli

Karthikeyan

Hello Christopher,

We have left the balancing flag to be default in our simulations which is AUTO as per documentation.

Looks like MotionSolve is doing A matrix balancing based on the observations from the log file (shown below).

The model constraints all seem fine to us reflecting the vehicle design intent.

Do you have any other feedback/recommendation that can be done to avoid the issues related to the flex body modes that i mentioned in the original query? Appreciate your feedback, thanks!

In our rigid body model, i could see log file messages related to condition number of eigenvectors as below

From our flexbody model, we see the log file message as below. In this case, looks like even after auto balancing the condition number remains in the range of 1E+12.

Christopher_Fadanelli

Hi @Karthikeyan ,

I did some asking and you could try a Static+Linear event. You can add joints to keep the vehicle in place and disable them before linear analysis. Regarding the CMS Flexbodies, sometimes the composition of them can create problems.

Another potential solution would to also include dummy bodies between the flexbody connections.

Best,

Christopher Fadanelli