Physics-Informed Diagnostics for Construction Arbitration Assets in Kenya: Out-of-Distribution Fault Detection and Lifecycle Cost Trade-offs

Construction arbitration in Kenya increasingly relies on physical assets—hearing rooms, evidence servers, and recording systems—whose silent degradation can compromise procedural integrity. The central contribution is a formal proof that a hybrid residual combining physics-based prediction error with a statistical novelty score achieves strictly lower out-of-distribution false negative rates than either component alone, under a verifiable condition on the relative magnitude of model mismatch. The framework is developed through a state-space representation of asset degradation, a spectral characterisation of fault signatures, and a distributionally robust detection rule. A lifecycle cost model then demonstrates that the optimal inspection interval under a performance-based contract is shorter than the interval that minimises direct maintenance expenditure, revealing a trade-off that conventional cost minimisation obscures. The article concludes that Kenyan arbitration institutions should adopt physics-informed diagnostics not as a technological upgrade but as a governance mechanism that aligns asset stewardship with the evidentiary standards of arbitral practice.