Critiquing FEA Results Responsibly

Introduction: Why checking FEA results matters

students, finite element analysis, or FEA, is a powerful tool for predicting how structures behave under load 🏗️. Engineers use it to estimate stress, strain, deflection, and safety before a real product is built. But a computer result is not automatically a correct result. The quality of an FEA answer depends on the model, the assumptions, and the way the results are interpreted.

The main idea of this lesson is simple: do not trust a contour plot just because it looks precise. A colourful stress image may hide a poor mesh, wrong boundary conditions, unrealistic material data, or a misunderstanding of what the model can actually predict. Critiquing FEA responsibly means asking smart questions about the model and the output before drawing conclusions.

By the end of this lesson, you should be able to explain key terms such as discretisation, mesh quality, and boundary conditions; check whether an FEA result is believable; and connect these checks to the wider process of solid mechanics analysis. You will also see how these ideas fit into real engineering decisions, such as judging whether a bracket, beam, or machine part is safe to use.

What FEA is really doing

FEA works by breaking a complicated object into many small pieces called elements. This process is called discretisation. Instead of solving a difficult problem for the whole structure at once, the computer solves simpler equations for each element and joins them together at the nodes. The result is an approximate solution to the original continuum problem.

This approximation is powerful, but it has limits. The answer depends on how fine the mesh is, what element type is used, and what assumptions were made about loading and support. If a beam is represented by too few elements, the bending shape may be too rough. If the mesh is too coarse near a hole or corner, the stress may be poorly captured. So when critiquing FEA, the first question is not “What is the answer?” but “How was the answer produced?”

A useful way to think about FEA is like a photo made from pixels 📸. A low-resolution image can show the main shape, but small details may be missed. A finer mesh gives more detail, but it also increases computation time and may still not fix a bad modelling assumption. Good criticism means checking both resolution and model logic.

Mesh quality and convergence: is the result stable?

One of the most important checks in FEA is mesh convergence. If refining the mesh changes the result a lot, the model may not yet be reliable. If the result changes only a little as the mesh becomes finer, the solution is likely approaching a stable value.

Suppose a student models a cantilever beam and finds a tip deflection of $8.0\,\text{mm}$ with one mesh, then $8.6\,\text{mm}$ with a finer mesh, and $8.7\,\text{mm}$ after further refinement. This suggests the solution is converging. If the values jump between $6\,\text{mm}$ and $11\,\text{mm}$, the model needs more investigation.

Mesh quality is not only about the number of elements. Shape matters too. Very stretched or distorted elements can produce inaccurate results. A mesh should be finer where stress gradients are high, such as near holes, fillets, contact zones, and load application points. In low-stress regions, a coarser mesh is often acceptable.

A responsible critique asks:

Was a convergence study done?
Were critical areas meshed finely enough?
Are the elements well-shaped?
Did the key output change much when the mesh was refined?

A common mistake is to trust the highest stress value in the whole model without checking whether it comes from a single distorted element or a sharp corner singularity. The highest number is not always the most meaningful number.

Boundary conditions: are the supports and loads realistic?

Boundary conditions tell the model how the structure is held and how it is loaded. In solid mechanics, these choices strongly affect the answer. A model can look mathematically correct and still represent the real world badly if the supports or loads are unrealistic.

For example, if a bolt joint is modelled as perfectly fixed everywhere, the structure may appear stiffer than it really is. If a force is applied at a single node, the model may create unrealistically high local stress. In reality, a load is usually spread over an area, and supports have some flexibility.

This is why it is important to compare the model setup with the physical situation. If a beam is welded at one end, a fixed constraint may be reasonable. If a part rests on rollers, a sliding or pinned condition may be more appropriate. If a load is applied through a pad, it should probably be distributed, not concentrated at a single point.

Good critique also checks whether the boundary conditions match the symmetry of the problem. Symmetry can reduce the size of the model, but only if the actual loading and geometry are symmetric too. Using symmetry when the real situation is not symmetric can give misleading results.

Ask yourself:

Do the supports represent the real test or service condition?
Is the load applied in a physically realistic way?
Could the constraints be making the structure too stiff?
Has symmetry been used correctly?

Stress results: what numbers can and cannot tell you

Stress plots are often the first thing people look at in FEA, but they need careful interpretation. Stress is not one single number everywhere. Different stress measures exist, including normal stress and shear stress, and in many solid mechanics problems engineers also examine equivalent stress such as von Mises stress.

A high von Mises stress can indicate a region where yielding may begin in ductile materials. However, it does not directly tell the whole story. Material failure depends on the material model, loading type, safety factor, fatigue history, and whether the stress is truly converged.

One important issue is stress concentration. Sharp corners, holes, notches, and re-entrant geometries can produce high local stress. This is not necessarily a mistake; it may be a real effect. But if the geometry contains a mathematically sharp corner, the stress may increase without bound as the mesh is refined. That is called a stress singularity. In such cases, the exact peak stress is not meaningful, and the engineer should instead evaluate a more realistic geometry or use another measure such as averaged stress or a stress away from the singular point.

Consider a plate with a circular hole under tension. The stress near the hole edge will be higher than in the far field. This is expected and useful. But if the mesh shows one extremely high value at a point load or perfectly sharp corner, students should be cautious. The value may be dominated by modelling idealisation rather than real behaviour.

A responsible critique separates three ideas:

physically meaningful high stress due to geometry or loading,
numerical artefacts caused by mesh or element issues,
singular stresses that do not represent a real limit.

Checking displacements, reactions, and equilibrium

Stress is only one part of the picture. Displacements and reaction forces are just as important. A structure can show reasonable stress but unrealistic displacement, or the opposite. Displacement tells you how much the part moves, which matters for fit, vibration, comfort, and function.

Reaction forces provide a powerful check on overall equilibrium. If a model is correct, the sum of reaction forces should balance the applied loads, within numerical tolerance. For example, if a total downward load of $10\,\text{kN}$ is applied, the support reactions should add up to about $10\,\text{kN}$ upward. If they do not, something may be wrong with the constraints or load definition.

Energy checks can also help. In a stable linear analysis, the internal response should be consistent with the applied work and the deformation pattern. While students do not always calculate energy manually, the principle is useful: the model should obey basic mechanics, not just produce colorful contours.

A good habit is to compare FEA results with hand calculations whenever possible. For a beam, simple formulas from solid mechanics can estimate deflection or bending stress. If the FEA answer is wildly different from the hand estimate, that difference needs explanation. Sometimes FEA is more detailed than a simple formula. Sometimes the FEA model is wrong.

For example, a cantilever beam with an end load has a tip deflection that can be estimated using standard beam theory. If the FEA result is close, that is reassuring. If it is much smaller, the model may be too stiff because of over-constraint. If it is much larger, the material properties, geometry, or boundary conditions may be incorrect.

Limitations, assumptions, and when to be cautious

Responsible critique always includes model limitations. FEA results are only as good as the assumptions used. Linear elastic analysis assumes small deformation and no permanent material change. If the part is actually yielding, buckling, or experiencing contact separation, a linear model may not be enough.

Important assumptions to check include:

linear or nonlinear material behavior,
small or large deformation,
static or dynamic loading,
perfect or imperfect contact,
idealised geometry versus real manufacturing features.

Real engineering parts often contain welds, fasteners, clearance gaps, rough surfaces, and residual stresses. These details may matter a lot. A model that ignores them is not automatically useless, but its conclusions must be limited to what the model actually represents.

This is why engineers often use the phrase “valid for this model” rather than “true in every case.” The role of critique is to understand the scope of the result. students should never assume that one FEA plot proves safety in every situation. Instead, the plot should be treated as evidence that must be checked against mechanics, experience, and, when possible, experimental data.

Conclusion

Critiquing FEA results responsibly means checking whether the model, the mesh, the boundary conditions, and the output all make sense together. In Solid Mechanics 2, this skill is essential because the method is only useful when its limitations are understood. A good FEA result is not just a number on a screen; it is a solution that has been tested against physics, verified by sensible modelling choices, and interpreted with care.

When you analyse an FEA result, remember the key questions: Is the mesh good enough? Are the supports and loads realistic? Do the stresses and displacements agree with mechanics? Are there singularities or modelling artefacts? If you can answer these questions clearly, you are using FEA as a careful engineering tool rather than a guess machine ⚙️.

Study Notes

FEA is an approximate method that solves a structure by dividing it into small elements through discretisation.
Mesh convergence checks whether the main results stay nearly the same as the mesh is refined.
Poor element quality, especially distorted or stretched elements, can reduce accuracy.
Boundary conditions strongly affect stiffness, stress, and displacement results.
Loads should be applied in a physically realistic way, not just at a single point unless that is appropriate.
High stress values near sharp corners, holes, or point loads may be real stress concentrations or numerical singularities.
Reaction forces should balance applied loads in a correct static model.
Displacements are important for checking whether the part still functions properly.
Hand calculations are useful for checking whether FEA results are reasonable.
Always interpret FEA results in the context of the model assumptions and limitations.