Pressure Vessel FEA: When Closed-Form Codes Aren’t Enough

Closed-form code calculations in ASME Section VIII Division 1 cover the majority of pressure vessel geometries - cylinders, heads, nozzles with readily-defined reinforcement boundaries, supports under standard load cases. They are fast, well-understood, and accepted without question by inspectors. But they assume geometry that fits the formulae. When the geometry does not fit, the formulae are not just inconvenient - they are inapplicable, and applying them anyway is a code violation dressed up as a calculation.

The cases where we reach for FEA are predictable: non-circular geometries, nozzles placed close together or close to a head knuckle, attachments with complex load paths, thermal gradients that produce local stress concentrations the closed-form methods cannot capture, and fatigue-critical locations where the cycle life depends on a peak stress the code formulae do not compute.

“ASME VIII Div. 1 covers most geometries - but not all. Here is when we reach for FEA, what we model, and how we keep the design code-compliant under Div. 2.”

Under Div. 2, FEA is not an alternative to the code - it is a code-permitted design route, with its own set of rules. The stress categorisation, the linearisation path, the allowable stress limits and the fatigue assessment are all defined in the code. A compliant Div. 2 FEA is one where the model, the post-processing and the reporting follow those rules, not where the analyst applies general FEA best practice and hopes the result is conservative.

What we model, and to what level of detail, depends on the question. For a nozzle reinforcement check on a Div. 1 vessel where the geometry falls outside the closed-form limits, a sub-model of the nozzle-head region with a linear elastic analysis and a Div. 2-style linearisation is usually sufficient. For a fatigue-critical location, we model the full vessel or the relevant sub-structure, run the operating cycle as a transient thermal-mechanical coupled analysis, and extract the peak stress for the fatigue assessment.

The mesh is the part that usually determines whether the result is defensible. We run a mesh convergence study on the quantity of interest - typically the linearised membrane-plus-bending stress at the classification line, or the peak stress at the fatigue location - and report the convergence. A Div. 2 report without a mesh convergence study is, in our view, incomplete; the linearisation is path-dependent and the reported stress can move materially between mesh densities.

The deliverable is therefore not just a contour plot. It is a model description, a mesh convergence study, the linearisation paths and classification lines, the stress categorisation against the Div. 2 limits, and - where applicable - the fatigue assessment with the cycle count and the cumulative usage factor. That package is what the inspector, the AI and the client’s third-party reviewer can trace back to the code.