Inertia Relief with EIGRL method

Blake Berk

Hi,

Has anyone successfully used "PARAM,INREL,-2,<EIGRL>?" I am attempting to have a flexible model where I can swap configurations with a series of MPCs and SPCs, triggered depending on the load case. None of the loads are 100% balanced right now, so I am using inertia relief to 'constrain' the model. 

Historically, we have been running with inertia relief without issue. Now that we are adding the additional configurations, we cannot get the inertia relief to solve the same.

The EIGRL is successfully detecting >6 rigid modes, but the applied accelerations are not accurately balancing the forces. It is slightly asymmetric, but not flying away. So the loads appear to rotate the model, but not fail to balance loads. The relative model displacement looks promising, but the overall displacement is incorrect. 

I also did 2 simple beam models to test accuracy on a simple model. My method was to create 2 models - 1. with 1 beam and solve with INREL,-2; 2. with 2 beams, one with an SPC covering 5/6 DOFs and one unconstrained and the modeled solved with INREL,-2,<EIGRL>. All other parameters (properties, materials, loads, etc.) are the same for both models. However, when the model runs, the results are not the same for the 2 unconstrained beams. These .fem files are attached and the results with different displacements as an image.

Is there anything I am missing or just a small error difference? It seems to me that the displacement differences are within a small error tolerance and the overall displacements are a result of where the displacement is measured from in the overall system. My concern is the displacement in my bigger model is larger more concerning, thought the relative displacements are closer. Just want to make sure I can trust the results and my method is correct. 

Thanks,

Blake

PS - Documentation suggests EIGRL or EIGRA can be used as the method, but I get errors when using EIGRA. Unrelated to my main question though.

Beam Solutions.png

Double Bar Demo - INREL-2-EIGRL.fem

Single Bar Demo - INREL-2.fem

RobHoglund

Hi Blake,

A few thoughts on your question.  The displacements shown on the example are likely different because you're not preventing rigid body motion in the same place- with PARAM,INREL,-2 and METHOD you allow the solver to pick the locations where the rigid body motion will be fixed for each model.  If you want to control that, use PARAM,INREL,-1 and create SUPORT entries in a load collector like an SPC.

With inertial relief analysis, the displacement plots are relative displacements to where those support degrees of freedom are placed (automatically with PARAM,INREL,-2,METHOD) to prevent the rigid body motion. Whereas if you use SPC in static analysis you define an absolute displacement constraint.

If you've got more than 6 rigid body modes, it could be easier to just run a free-free modal analysis to pinpoint where the model will need to be connected to get down to 6 and then run inertia relief.

For the documentation there is an issue- only EIGRL is supported as the METHOD card.  That is why you see that error.  It is being fixed in a later version, hope that's helpful.  Thanks,

Rob