Desktop Engineering Blog

FEA Model Checking - Part 1

Posted by Andy Woodward on 04-Jan-2017 15:00:00

I’ve been an FEA user since 1992 and providing first line technical support for the last ten years. A surprising number of models sent to us over the years with ‘problem’ behaviour have had fundamental mistakes in the way they were constructed that could have been easily identified through systematic model checking. 

In this two-part blog we will discuss some recommended model checks that can help identify basic problems.  In the first one we’ll look at mass and stiffness checking.

FEA model checking - mass check on vehicle body

Mass Check

Chances are your model geometry came from a CAD system.  You simplified and discretised this into a finite element mesh and applied properties to it.  With all those approximations it should still have roughly the same mass as the CAD BOM says it should. This said, I have seen models with vastly different weights, including my own personal mistake where a model created in inches had the thickness defined in millimetres resulting in a 25X too high mass.  Given that in a static analysis it is not necessary to enter density for a material if there are no inertial loads this is an easy thing to get wrong. 

All pre-processors and solvers will give the mass output, for example:

model checking image 1.png

FEA model checking image


Mass/Stiffness Check

The easiest way to check this is via modal analysis, with the bonus result of also checking your connectivity and grounding. You should run two normal modes analyses on your model, one unconstrained and one grounded. 

The unconstrained modes should show exactly six nominal zero Hz modes – the rigid body modes of the model. If it has less than six then you have messed up connecting the model, probably via springs and/or MPC’s and the first non-zero mode will give you some indication of where/how that is.  If you have more than six modes then it suggests that you have not connected some part of your structure.  I’ve seen complex assemblies, such as a dashboard, with dozens of zero Hz modes where little sub-components were not connected in correctly resulting in static analyses failing due to singularities.  You should also look to see that there is a clear gap between the highest rigid body frequency and the first flexible mode, 3 or 4 orders of magnitude is suggested.

FEAmodel checking image mode values

The constrained modes will tell you if you have fixed the model down adequately for a static analysis.  You should have no nominal zero Hz modes indicating that the model is correctly grounded.

Real Eigenvalues FEA model checking image

It is also worth considering the frequency of the first few modes – does it sound right?  If you were to bang your structure with a hammer, would it ring like a (high frequency) bell or give a (low frequency) clunk. If the frequency seems very wrong it is worth checking the material entries to make sure your stiffness and density have magnitudes consistent with each other and the scale of the model.


In the next article we’ll look at static analysis tests.

Having an experienced partner to deliver your technical support is a great reason to choose DTE for your FEA requirements.  DTE is backed by MSC’s local support office which has over 200 years’ experience with the tools.

Topics: Various - CAD CAM FEA PLM, Finite Element Analysis (FEA)