UK - 01926 333777
ROI - 01 297 4440
01926 333777

# Idealisation Error Within SOLIDWORKS Simulation

Friday June 16, 2017 at 4:21pm
Idealisation Error Within SOLIDWORKS Simulation by Applications Engineer Tom McHale

Idealisation Error Within SOLIDWORKS Simulation

Introduction

It doesn’t matter how accurately a model is discretised or how good a solver is, if a mathematical model is poorly idealised, the solver will always converge on an incorrect solution.

Idealisation errors normally account for the largest source of error in any simulation and, ironically, they are introduced before Finite Element Analysis (FEA) is even applied to the model. The good news is that we typically engineer components with a Factor of Safety large enough to mitigate all three types of error within simulation. Nevertheless, reducing the magnitude of these errors will allow component design to be optimised to a greater degree of accuracy.

Generally, all simulation errors can be attributed to one of three categories and are introduced at different stages of the analysis process:

While idealisation errors occur before FEA is introduced, the creation and preparation of the mathematical model is still heavily reliant upon both the SOLIDWORKS CAD and SOLIDWORKS Simulation environments. Good engineering judgement is often required for appropriate idealisation of the mathematical model and also for interpretation of the results.

Types of Idealisation Error

Idealisation errors typically emerge when the geometry is created, the material model is defined and when boundary conditions are applied. All three areas control the degree of resemblance between the mathematical model and reality.

1.      Geometric Representation and Simplification

Geometric simplification is often unavoidable and is usually necessary for acceptable computation times. It is best practice to initially use an overly simplified model for any analysis in order to ensure that the model is behaving as expected. The degree of simplification should then be reduced: especially in areas of interest in order to get meaningful and representative results.

Typical simplifications include the removal of features not important to the analysis – defeaturing. Fillets are a good example of this. When removed, the mesh will be simpler and therefore will result in quicker computation times. However, a sharp edge will result in a singularity so the stresses in the immediate vicinity will be wrong. This is fine if the stresses in this area are not of interest.

It is also worth noting that defeaturing will often add/remove material which will affect the stiffness of the model. Again, this is acceptable if the stiffness in the area of interest is accurate.

Simplification doesn’t have to come in the form of defeaturing. One of the best ways to simplify a model is to take advantage of symmetry or to represent a 3D model as 2D. This can significantly reduce computation time due to the large reduction of elements needed to discretise the model.

2.      Material Model Definition

SOLIDWORKS has a material library containing a sample of typical materials. It is sometimes easy to take this for granted as many dedicated simulation packages do not provide this information. SOLIDWORKS materials will be more than sufficient for many simulations when used correctly. However, occasionally, the material model may need to be tweaked or a new material created from scratch using experimental data if an unusual material is required or it is used in a special situation. For example, consider the points below:

·         If a material is designed to operate below yield then a plasticity model is not needed
·         If a fatigue cycle is ‘zero based’ then an S-N curve for a ‘fully reversed’ cycle may not be representative
·         If a material is to operate in a high temperature environment then the stress-strain curve should be representative of that temperature
·         Material definitions are unlikely to account for imperfections

3.      Boundary Conditions

Boundary conditions include: fixtures, loads and contacts. It is normally the poor idealisation of fixtures that will cause the biggest error in any simulation. In reality, loads will always have some degree of eccentricity, fixtures are never rigid and contacts can often have some degree of slide and friction. Constraining too many degrees of freedom is easy to do and can result in over stiffening.

For example, we know from Euler that the type of fixtures applied to struts under a buckling condition can result in very different critical buckling loads. These loads can be calculated by hand in the two cases below: Fixed-Fixed and Pinned-Pinned. Note that the respective descriptions of each buckling condition do not correlate to an exact representation of fixtures applied; the degrees of freedom constrained are stated at each end of the strut and the respective differences shown in red.

The difference in critical buckling load is significantly different between the two buckling conditions, yet this difference is entirely due to how just one degree of freedom has been handled at each end of the beams. This is an extreme case but it is easy to see how important it is to consider every degree of freedom when constraining and loading your model.

Again, consider the thoughts below:

·         Is it appropriate to fix an entire face when it is the edges that are welded in place?

·         Does the applied load always act in the same direction or does it follow the deformation of the geometry?

·         Using remote mass / distributed mass is a great way to simplify a model for analysis but they will not account for the additional stiffness that would have been provided by the missing geometry – only mass.

Conclusion

The main point to take home from this blog is to always be perceptive when transitioning between reality and the mathematical model. It is easy to make quick decisions in the set-up of a study when it should really be the part of the analysis where most time is spent. By ensuring that simplification has been achieved intelligently, and that the material model and the boundary conditions are accurate, the degree of idealisation error will be minimised.

By Tom McHale

Elite Applications Engineer

## Related Blog Posts

SOLIDWORKS Simulation is an affordable 3D design simulation solution with powerful engineering capabilities that help you predict your product performance before you even consider prototyping.
In this short 4 minute run down, Jamie will explain the key differences between SOLIDWORKS 2019 Standard, Professional and Premium, helping you determine whether an upgrade is right for your business and which one. For a more in depth look at the dif....
Have you ever wondered whilst using SOLIDWORKS Flow Simulation - “How do I know where the minimum and maximum values are on my model?”If so, then you're in luck as this blog will explain how to easily find these extreme values.
Top