WRC 537 vs WRC 297: Finally Making Sense of Nozzle Stress Calculations
Why these bulletins matter
Specifications still lean on Welding Research Council bulletins whenever nozzle loads come up, and the references are often a generation out of date. The job of these methods is to estimate local stresses at a nozzle or attachment from external piping loads, in a way a reviewer can follow and a code margin can be applied to. Pick the wrong bulletin, or claim more from it than it actually delivers, and the review drags: questions multiply, someone reruns the numbers, and the schedule wears it. Pick the right one and state its limits up front, and the conversation gets short. This post sets out what WRC 537 and WRC 297 each actually cover — from the bulletins themselves, not from folklore.
WRC 107 became WRC 537 — same method, better data
WRC 107 was published in 1965 and revised several times through 1979. It packaged Professor P. P. Bijlaard’s shell solutions into non-dimensional design curves: read the curve, take a membrane force and a bending moment, combine them into surface stresses at the attachment-to-shell junction. Generations of engineers learned nozzle loads from it, and plenty of specifications still cite it.
In 2010 WRC 537 superseded it. WRC 537 is not a new theory. It is the WRC 107 method reissued with corrected data and precision curve-fit equations — every figure now carries polynomial coefficients, so software (and careful spreadsheet users) evaluate the fit instead of eyeballing a log-scale chart. The known errata from the 107 revision history, such as the reversed curve labels documented in its Appendix A, are folded in. If your specification says WRC 107, treat that as legacy wording and run WRC 537. The answer is the same method with the housekeeping done.
What WRC 537 actually gives you
WRC 537’s own statement of its limits is blunt, and it applies to the whole bulletin: the procedure yields “stresses in the shell, but not in the nozzle” (§3.5.1, repeated for cylinders in §4.5.3). It computes surface stresses in the vessel wall at eight points around the attachment junction — four positions, inner and outer surface. What it assumes about the attachment differs between its two halves, and the difference is worth knowing.
Cylindrical shells (Section 4). The curves treat the attachment as a loaded footprint on an unpenetrated shell — there is no hole and no nozzle wall in the model. WRC 297’s Appendix C says exactly that when comparing the two bulletins. Round attachments enter through an equivalent square: the attachment parameter is taken as 0.875·r₀/Rm (§4.2.2.1). Square and rectangular solid footprints are handled directly, rectangles via multiplication factors on the square case. There is no hollow-attachment case for cylinders: nozzle wall thickness never appears in the Section 4 parameters, so a thin flexible nozzle and a solid trunnion of the same outside radius produce identical shell stresses.
Spherical shells (Section 3). Here the bulletin does distinguish a rigid solid insert (§3.2.2.1) from a hollow nozzle (§3.2.2.2), and the nozzle wall enters the solution through the parameters rm/t and T/t. There is even a hollow square case for box-section attachments (§3.2.2.3). So the sphere curves account for nozzle-wall flexibility when computing the shell stresses — but the output is still shell stresses. Section 3.5.1 says so in terms, and Appendix A quotes the underlying assessment: solutions that give only shell stresses “may seriously underestimate the peak stress” when loads come in through a thin-walled nozzle. Both halves of the bulletin carry the same warning — the nozzle wall is most likely to govern when the opening is unreinforced, or when the reinforcement sits on the vessel rather than the nozzle.
Ellipsoidal heads. WRC 537 §3.5.3 allows the spherical method on ellipsoidal heads with “reasonable accuracy” if you use the local mean radius of the head at the attachment. For a nozzle in the crown of a 2:1 semi-ellipsoidal head that local spherical radius is about 0.9D — not the nominal head radius and not a mean of anything. The idealisation is at its best in the crown, where curvature is nearly constant; toward the knuckle the meridional radius changes rapidly and treating the spot as a sphere becomes progressively harder to defend.
What WRC 297 adds
WRC 297 (1984, revised 1987) is written as a supplement to WRC 107, and it is deliberately narrow: cylindrical nozzles radially attached to cylindrical vessels, nothing else. Within that geometry it does two things the older bulletin does not.
First, it models the junction rather than a footprint. The method is C. R. Steele’s thin-shell solution of two intersecting cylinders: the opening is real and the nozzle is a flexible shell, not a rigid plug. That is what lets 297 reach configurations 107/537 never covered — thinner nozzles, larger d/D, vessel D/T well beyond the Bijlaard curves.
Second, it reports stresses on both sides of the junction. The foreword states it plainly: stresses in both the nozzle and the vessel can be determined. Section 3 gives vessel surface stresses and nozzle axial stresses at the junction; nozzle circumferential membrane stress is set equal to the shell’s, and nozzle circumferential bending is neglected as insignificant.
The stated applicability limits (§3.2) are worth pinning to the wall: d/t ≥ 20, D/T ≥ 20 and d/T ≥ 5 for the thin-shell theory to hold, D/T ≤ 2500 as a recommended ceiling, curves plotted to d/t ≤ 100, and d/D up to about 0.5 depending on D/T. A thick nozzle acts as a nearly rigid insert, so vessel stresses stay reasonable at low d/t even though the nozzle-stress side of the method loses validity. The nozzle must be isolated (roughly 2√(DT) from any other discontinuity), must not protrude inside the vessel, and must be attached by a through-penetration weld.
Scope at a glance
This table is the part to keep. It replaces the folklore versions — “537 is shells, 297 is nozzles” — with what the bulletins actually say.
| Method | Geometry covered | Attachment as modelled | What stress you get, and where |
|---|---|---|---|
| WRC 537, Section 4 | Cylindrical shell, radial attachment | Solid footprint on an unpenetrated shell; round attachments via the 0.875·r₀ equivalent square; square and rectangular solid shapes directly; no hollow case, nozzle wall ignored | Shell stresses only, at 8 points around the junction (§4.5.3) |
| WRC 537, Section 3 | Spherical shell; ellipsoidal head treated as a local sphere using the crown radius at the attachment (§3.5.3) | Rigid solid insert, or hollow round nozzle (rm/t, T/t), or hollow square section — nozzle-wall flexibility enters the shell solution | Shell stresses only, at 8 points (§3.5.1); explicit warning that the nozzle wall may govern |
| WRC 297 | Radial cylindrical nozzle on a cylindrical vessel, that geometry only | Flexible thin-shell nozzle intersecting the vessel (Steele); opening modelled; d/t ≥ 20, 20 ≤ D/T ≤ 2500, d/T ≥ 5, d/D ≲ 0.5 | Shell and nozzle stresses at the junction — vessel surface stresses plus nozzle axial and circumferential membrane |
So which one do you run?
For the classic radial cylinder-on-cylinder nozzle, common practice is to run both: 537 for the vessel-side check reviewers grew up with, 297 for the nozzle wall and for geometry 537’s cylinder curves cannot reach. The publisher takes the same view — WRC sells 107/297 “as an integral set.” But be clear about what that pairing is: a practice preference, not a scope boundary. WRC 297 is not merely a nozzle-stress annex; it produces its own vessel stresses, from a model that includes the opening, where 537’s cylinder curves assume an unpenetrated shell. The two therefore give different shell stresses for the same loads — 297’s Appendix C compares them and finds consistent trends but real differences either side, depending on nozzle-neck thickness. Which set to trust, and when the difference matters, is an argument reviewers genuinely have; it deserves its own post and will get one.
For spheres, dished heads, trunnions, and rectangular attachments, 537 is the only one of the pair that applies. Take its shell stresses, use the local crown radius on ellipsoidal heads, and deal with the nozzle wall separately — the bulletin itself tells you it has not checked it for you.
When the bulletins are not enough
The envelope is the whole deal. Inside it — single, isolated, radial attachment; geometry within the stated ratios — the bulletin methods are fast, repeatable and defensible, and there is no engineering reason to reach past them. Outside it — hillside and lateral nozzles, closely spaced openings interacting, reinforcing pads (which neither bulletin’s curves model), parameters off the end of the charts — forcing a bulletin to fit produces numbers nobody should defend. That is the point to escalate to a proper local analysis, and to say so in the calc rather than bury it.
Key takeaways
WRC 537 has replaced WRC 107; cite and run 537. It delivers shell stresses only — its own limitation sections say so for both cylinders and spheres — and its cylinder curves never see the nozzle wall at all, while its sphere curves at least feel the nozzle’s flexibility. WRC 297 covers one geometry, the radial cylinder-on-cylinder junction, and there it gives both shell and nozzle stresses from a flexible-nozzle model with hard applicability limits. Running 537 and 297 together on that geometry is sound practice, provided you understand you are looking at two different models of the same junction, not two halves of one answer.
A worked example is coming: a full WRC 537 cylinder calculation, all eight stress locations by hand, checked against a finite-element run of the same junction.
Run WRC 537 and 297 with PV Cloud
PV Cloud runs WRC 537 and WRC 297 side by side, with the geometry limits checked up front and the curve-fit equations evaluated exactly — no chart squinting, no spreadsheet archaeology.