Framing
What analysis tools are and are not
CAD/CAM systems offer a considerable range of analysis capability: mass properties, the finite element method, thermal, fatigue, frequency, buckling, mould flow, fluid flow and motion simulation. Each implements a mathematical model. The crucial factor in using any of them is understanding how to set the problem up correctly, supply the right input parameters, and interpret the results.
That requirement is stronger than it sounds. Effective use of a stress analysis tool demands a sound understanding of the finite element method itself — not of the software's dialogue boxes. A designer who cannot say why a particular mesh, restraint set and load case represents the physical situation is not analysing anything; they are generating colourful pictures with a plausible legend.
Analysis software rarely refuses to produce an answer. Supply an over-constrained restraint set and it will return low stresses; supply a too-coarse mesh at a stress concentration and it will return low stresses; apply a load as a pressure when it is really a point contact and it will return low stresses. Every one of those errors flatters the design. Independent verification — hand calculation on a simplified model, or a mesh convergence study — is not optional.
Interoperability
Data exchange
Data exchange is routine in practice. A company that outsources manufacture must send models to subcontractors who may not use the same system, and every system stores models in its own proprietary native format, which other systems cannot open.
The solution is neutral file formats readable by all systems. Some are formal standards — STEP and IGES are the principal examples — and others are de facto standards, such as DXF and the ACIS kernel format. A separate group covers image and web representations: a model saved as a raster image, or as a web-viewable three-dimensional format.
What survives
Geometry, and depending on the format, topology and assembly structure. STEP generally preserves solid bodies and assembly hierarchy; IGES is surface-oriented and older.
What does not
Feature history, design intent, sketch relations, equations and configurations. A neutral file is a snapshot of the result, not the recipe.
Practical consequence
Imported geometry is dumb geometry. Editing it means direct face and body operations, not parameter changes — which is why round-tripping through neutral formats is a one-way trip in practice.
Oldest capability
Mass properties
Mass property calculation is the oldest application offered by CAD/CAM systems, and it rests on foundations from ordinary engineering mathematics. The quantities are volume, mass, centre of mass, first and second moments of inertia, and the principal moments and axes of inertia.
Volume follows from integrating over the solid; mass is volume multiplied by density, which is why material assignment matters. The centroid is the volume-weighted average position. Second moments of inertia about each axis, and the first moments (products of inertia) between pairs of axes, assemble into the inertia tensor. Diagonalising that tensor yields the principal moments and the principal axes — the orientation in which the products of inertia vanish.
Why these are computed numerically
The defining integrals are evaluated over the boundaries of a three-dimensional solid whose faces may be arbitrary surfaces. Closed-form evaluation is impossible in the general case, so the system integrates numerically over the boundary representation. This is where the normal vector introduced in Part 11 does its work: its direction distinguishes the inside of the solid, where material is, from the outside and from voids.
Mass properties are the fastest way to catch a modelling error. If a mild steel bracket reports 40 grams, something is wrong — a missing material assignment, a unit mismatch, or a body that failed to merge. Check mass and centre of gravity on every model before doing anything more sophisticated.
Behaviour over time
Motion analysis
Motion studies come in two forms. Time-based studies describe the response of an assembly to changes in its motion elements over time. Event-based studies are defined by a set of motion actions triggered by events, and the time sequence of element changes is derived by calculating the study rather than specified in advance.
Event-based motion is specified through combinations of sensors, times and prior events. It requires a set of tasks, which may run sequentially or overlap, each defined by a triggering event and an associated action that controls motion during that task. This is the appropriate model for machinery whose behaviour is conditional — a clamp that closes when a part arrives, an indexer that advances when a cycle completes.
Fluids
Flow simulation
Flow simulation analyses the movement of fluids — liquids, gases and steam, real gases, non-Newtonian liquids and compressible liquids — and systems typically ship with a library of predefined fluids that can be extended with user definitions. Mutual dissolution of fluids of the same type can be included.
The setup discipline is specific and unforgiving:
- Step 1 Establish a closed control volume. The solver requires one; an open geometry has no defined domain.
- Step 2 Define an inlet and an outlet for the fluid to enter and leave.
- Step 3 Cap those openings with lids so the volume is genuinely closed.
- Step 4 Set initial and boundary conditions — temperature, pressure, velocity and flow rate.
- Step 5 Confirm a driving gradient exists: inlet pressure must exceed outlet pressure, or nothing will move.
A simple and instructive first study is water flowing through a hose, with streamlines animated to visualise the movement. It is worth doing because it exercises every step above on a geometry where the expected answer is obvious.
The central method
The finite element method
The finite element method is a numerical method capable of solving almost any problem at almost any level of complexity. It comprises finite element modelling — building the discretised representation — and finite element analysis, the solution and interpretation.
Its elegance lies in a transformation. Many engineering problems are too complex for a closed-form solution of their governing equilibrium equations, which are differential. The method converts that differential equation into an integral one. Rather than solving over the entire continuum, it applies the integral form over many small regions called elements, then assembles the element equations into a set of simultaneous algebraic equations whose solution yields the answer across the whole domain.
Discretise
Divide the complex domain into elements. Elements are adjacent with no gaps between their sides and connect at nodes.
Choose element type
Element type, node count per element and the degrees of freedom at each node are among the most consequential decisions in the whole process.
Apply material and loads
Material properties, restraints and external loads — each an assertion about the physical situation.
Solve
Assemble and solve the global system of algebraic equations for the nodal degrees of freedom.
Interpret
Derive stresses and strains from the nodal solution, and judge whether they mean what they appear to mean.
Take a cantilever beam with a point load at its free end. Its governing equilibrium equation is the Euler–Bernoulli beam equation, a fourth-order differential equation in the deflection, with the fixed end supplying the boundary condition. Processed through the finite element formulation it becomes an integral expression combining the internal strain energy with the work done by the applied force. That integral is still difficult to solve in closed form over the whole beam — but applied element by element and assembled, it reduces to linear algebra. That reduction is the entire trick.
Tooling
Two tiers of simulation capability
| Capability | Entry-level module | Full module |
|---|---|---|
| Geometry | Parts only | Parts and assemblies, including contact and friction |
| Analysis type | Linear stress and displacement | Adds thermal, frequency, buckling, fatigue, non-linear and drop test |
| Loads | Uniform pressure and forces | Adds non-uniform pressure, torques and heat loading |
| Connectors | None | Springs, elastic foundations, pins, bolts, spot welds |
| Visualisation | Stress, displacement and factor of safety plots | Adds iso-surface plots and sectioned results |
| Reporting | HTML report, lightweight 3D file, video animation | Adds report customisation and image export in several formats |
The entry-level tool is best understood as a first-pass design verification aid: it allows a design to be tested and alternatives explored without manufacturing anything, and it is well suited to linear stress analysis of simple parts. Beyond that — anything with contact, non-linearity, assemblies or thermal coupling — requires the full module.
Interpretation
Von Mises stress and the yield criterion
Von Mises stress is a criterion for the failure of ductile materials. A ductile material is one exhibiting a yield stress; such materials are also described as isotropic and obey Hooke's law in the elastic range. On a stress–strain curve, steel and aluminium show a clear elastic region, a yield point, a region of plastic deformation, an ultimate stress and finally rupture. Brittle materials such as glass and cast iron show far less plastic region and rupture close to their ultimate stress.
The Von Mises stress is a scalar computed from the full three-dimensional stress state at a point. Its purpose is to reduce a six-component stress tensor to a single number that can be compared directly against the uniaxial yield stress obtained from a simple tensile test. When the Von Mises stress reaches the yield stress, the material is predicted to begin yielding at that point.
Factor of safety
Factor of safety is the yield stress of the material divided by the Von Mises stress at the point of interest. A value below one indicates predicted yielding. Most simulation tools plot factor of safety directly, which is generally more useful to read than raw stress because it carries the material property with it.
Two cautions. First, the criterion applies to ductile materials; using it for brittle materials, composites or anisotropic materials gives misleading answers, and different failure theories apply. Second, peak Von Mises stress at a sharp internal corner in a linear elastic model is a singularity, not a result — it will rise indefinitely as the mesh is refined. Model the actual fillet radius, or read the stress away from the corner.
Key takeaways
- Setting up the problem and interpreting the result are the difficult parts; running the solver is not.
- Neutral formats transfer geometry, not history — imported models are dumb geometry.
- Mass properties are computed numerically over the boundary representation and are the cheapest available sanity check on a model.
- Time-based motion describes response over time; event-based motion is driven by triggers and suits conditional machinery.
- Flow simulation needs a genuinely closed control volume with capped inlet and outlet and a real driving gradient.
- FEA converts a differential equilibrium equation into an integral one applied element by element and assembled into linear algebra.
- Von Mises stress reduces a stress tensor to a number comparable with uniaxial yield — valid for ductile materials, and meaningless at a modelled sharp corner.
Series
Continue the pathway
The SOLIDWORKS Design Approach series works through computer aided design as an engineering discipline, from first principles to manufacture.