AF1: A Claim-Bounded Numerical Audit of Graphene-Modified Marine Coatings: From Formulation-Space Screening to Experimental Handoff

Research question

Which coating candidates survive sequential numerical controls and can be prioritized for paired physical tests without conflating release kinetics, biological efficacy, a hydrodynamic proxy and measured material performance?

MODEL SCOPE

Public model description

A staged, dimensionless phenomenological chain connects undisclosed composition descriptors, a coarse graphene-distribution field, mechanical ageing, surface state, release/exposure, fouling occupancy and a finite coating budget. A later monotonic uncertain mapping defines a hydrodynamic proxy. Paired no-graphene controls must recompute dependent priors. Biological admission and hydrodynamic-target admission are different endpoints; neither maps uniquely to manufactured coating performance. T2 supplies methodological context without transferring physical validation.

PROTOCOL

Declared evidence chain

AF1.1–AF1.12 preserve corrective detector tests, matched-control reconstruction, common-exposure mechanics, orientation-sensitive surface checks, release and biological controls, cycling/resource extensions, inverse search and non-overlapping forward batches. All 21 selected survivors reach the intermediate grid; only eight promotions plus two separate references are tested at the final grid. Seed ancestry and finite output stability qualify that transfer. ET-01 proposes separate integrity, water-exposure and attachment-removal tests, with a distinct species-specific module; it contributes no completed physical observations.

Series: AF1.1–AF1.12 · corrective audits · paired 0G controls · forward confirmation · grid transfer · experimental handoff

• AF1.1–AF1.12
• corrective audits
• paired 0G controls
• forward confirmation
• grid transfer
• experimental handoff

The labels below preserve the public protocol identity. They are an index to the publication, not a substitute for its full methods or an open reproducibility bundle.

EVIDENCE

Results inside the model

1. The initial provisional screen admits 225,876 of 500,000 candidates (45.1752%). This is a sampling- and gate-dependent formulation-space fraction, not a manufactured-coating efficacy estimate. Detector, comparator and exposure-time corrections remain part of the evidence history.
2. AF1.6–AF1.8 do not establish strong robust biological suppression under the inherited criterion. The tested near-surface exposure extension does not rescue the active branch. Useful temporal release flattening remains a transport result; later proxy admission does not reverse the biological verdict.
3. Non-overlapping forward batches evaluate 6, 18 and 104 candidates and retain 2, 4 and 15 full hydrodynamic-target survivors: 21 of the selected 128-candidate queue. This does not identify a global optimum or a population success rate.
4. All 21 survivors retain admission at the intermediate grid. Eight preselected promotions and two separately labelled references all pass the final-grid cascade. The other eleven intermediate-grid survivors were not tested at that final grid.
5. Across the ten finalists, worst-scenario median hydrodynamic-proxy reduction is approximately 1.096–1.531%. Its relative intermediate-to-final change ranges from −1.19% to +0.46%. These are selected model outputs, not measured ship drag, fuel savings or a continuum-convergence proof.
6. The handoff selects five role-diverse candidates, five paired no-graphene controls and five reserve finalists. Physical realization, laboratory acceptance and measured economics remain pending; the proposed integrity, water-exposure and attachment-removal tests are not completed experiments.

BOUNDARY

CLAIM BOUNDARY

Numerical evidence is not physical validation. Transfer to a material, device, organism, environment, or operational service requires a separate validation chain whenever such a transfer is relevant.

Publication record