# Run a structural analysis

Use Adam to set up a structural study, inspect an existing one, or compare design revisions. Name the solver if you have a requirement; otherwise ask Adam to check an available formulation appropriate to the problem.

## Choose the right physical model

A small-displacement linear study can answer some stiffness questions. Contact separation, snap fits, large deformation, impact, or material yielding may require different physics. A modal shape shows a vibration mode with arbitrary normalization; its plotted amplitude is not an operating displacement prediction.

Provide geometry and units, sourced material properties, physical attachment, loads, and the quantity you need. Distinguish an applied force from a prescribed displacement: the latter can be used to calculate the reaction needed to impose that motion.

## Start with a controlled example

```text
For an illustrative study, model a rectangular cantilever 120 mm long, 20 mm wide, and 4 mm thick. Use an ideal linear elastic material with E = 70 GPa and Poisson ratio 0.33. Fix the end at x = 0 and distribute a total 10 N load downward over the opposite end face, bending through the 4 mm thickness. Check small-deflection assumptions. Report tip displacement and support reaction, compare displacement with a beam estimate, and examine mesh sensitivity. Label the idealized fixture and material assumptions.
```

This is a numerical learning example, not a qualification of a real bracket or joint.

### Compare with the beam estimate

For the idealized end load above, the elementary bending estimate is `deflection = F L³ / (3 E I)`, with `I = width × thickness³ / 12`. In consistent N and mm units, `E = 70,000 N/mm²` and `I = 106.67 mm⁴`.

| Quantity | Reference value |
| --- | --- |
| Tip deflection magnitude | Approximately 0.771 mm |
| Vertical support reaction magnitude | 10 N |
| Support bending moment magnitude | 1,200 N·mm |

These are beam-theory reference values. A three-dimensional finite-element model can differ because of its formulation and end constraints. Ask Adam to explain the difference and mesh sensitivity. A factor-of-1,000 discrepancy is a reason to check units before changing the physics.

## Inspect setup and results

**Verify the physical inputs**
Read back material assignments, units, load direction and total, support locations, and contact assumptions.
**Run and preserve evidence**
Keep the input deck or native study, mesh, solver version, logs, and achieved load or time step.
**Check numerical behavior**
Review termination, finite results, equilibrium where available, and refinement of the requested response.
**Interpret the engineering result**
Compare against the supplied requirement and explain the effect of assumptions, singularities, or missing checks.

## Compare a design change

```text
Compare revisions A and B under the same documented load case. Rebuild geometry selections and remesh each revision. Report mass, displacement at [location], and stress using the same component and averaging convention. Keep material and support assumptions fixed. Include solver completion, reaction balance where available, and mesh sensitivity before recommending a revision.
```

A stress peak at an idealized constraint may not be a meaningful failure criterion. A factor of safety needs the appropriate material limit and failure mode; displacement alone does not establish strength.

For native studies, see [Native CAE setup](/integrations/native-cae). For deliverables, see [Results and comparison](/cae/results).