The SNNM Story: How a Formula Earned Regulatory Standing

Every discipline has a case study it should teach its newcomers, one story that shows how the field actually improves. For added resistance, that story unfolded in the open literature across roughly five years, and it ends inside the sea-trial standard itself. It is worth telling properly, because its moral applies to every method you will ever be asked to trust.

The gap

Part 1's methods landscape left the field with an open item: the workhorse semi-empirical corrections were valid in head or near-head seas, while ships sail, and trials are run, and routing decisions are made, in seas from every direction. What was needed was one formula family covering arbitrary headings, across the commercial fleet's range of hull forms, speeds and loadings, accurate enough to carry regulatory weight.

Development: regression over the evidence

The answer came from the seakeeping research community: a semi-empirical method built by regression over a large database of tank experiments, spanning ship types, proportions, speeds and, crucially, wave headings from dead ahead to dead astern. Its structure honours the two-mechanism physics from Part 1: separate expressions for the motion-induced component (with its resonance behaviour, its dependence on pitch mass distribution, fullness, trim and Froude number) and the reflection component (built from the bow and stern waterline geometry), summed, and scaling with amplitude squared as Part 1's opening law demands. The method is known as SNNM, after the four parties behind it: SHOPERA, the EU research project in which it began; the National Technical University of Athens, where it was developed; Nanyang Technological University in Singapore, where it was extended; and the Marine Design and Research Institute of China, which verified it.

Publication, though, is where such stories begin. A formula fitted by its own developers, however carefully, is a claim.

Validation: the adversarial phase

What happened next is the part worth memorising. Towing-tank institutions across Europe and Asia, ITTC members co-ordinated by its committee on ships in operation at sea, pooled their own basins' data and subjected SNNM to a validation whose analysis was done by laboratories independent of the developers, against roughly 1,500 experimental data points, specifically for the speed-trial application. The assessment was split into two heading regions, each judged against its own correlation threshold fixed in advance. The method reached 0.86 in both regions, against thresholds of 0.78 and 0.70: it cleared both, comfortably, and the results were published for anyone to check.

That design deserves admiration: pass criteria fixed in advance, the two heading regions of unequal difficulty judged separately rather than as one pooled score, the analysis in hands other than the developers'. It is the difference between "the developers say it works" and "the community tried to break it and failed."

Adoption, and extension

With validation on the record, adoption followed the ladder: into the ITTC's recommended procedure for speed/power trials (that link is the 2021 revision, Revision 06, that took SNNM in; the next lesson cites the 2024 revision, Revision 08), and from there into the 2025 generation of ISO 15016, the sea-trial standard, which took SNNM in and dropped two older wave-correction methods, STAWAVE-2 and the NMRI method, in the same revision. A journal formula had become, in the fullest sense, part of the rulebook. (Our ISO 15016 course tells the standard's side of this story.)

And the clock kept running. The validation campaign itself had flagged the method's edges: ships of extreme proportions (unusual length-to-beam and beam-to-draught ratios: think ro-pax and car carriers) sat at the boundary of the fitted population. A 2023 extension addressed exactly that, using numerical experiments with a 3D potential-flow panel code (not viscous CFD) to stretch the coverage: Part 1's methods ladder in miniature, with computation extending what the tanks had begun.

The morals

Three, all portable. Trust follows adversarial validation, when evaluating any vendor's "proprietary wave model", ask who tried to break it and where the results are published. The research-to-regulation lag is the validation: the years between paper and standard were not bureaucracy; they were the community doing its due diligence. And a method's identity includes its edition and envelope: analyses citing "the standard" from different generations can carry different wave machinery, and the fitted population's boundaries (the next lesson walks them) are part of what was validated.

So much for how the formula earned its place. What is actually inside it, which ship and wave properties drive the answer, and where its envelope fences lie, is where we go next.