Coding dataset for a structured review of automated layup constraints, manufacturing deviations and structural response in curvilinearly reinforced composite structures

Two independent layers are provided. The prevalence layer records six implementation indicators and five causal links for a corpus of 167 publications, of which full texts were obtained for 147; proportions are accompanied by assumption-free bounds rather than confidence intervals, because full-text availability was non-random.

The mechanism-and-effect layer is a purposively assembled matrix of 91 comparisons extracted from 44 publications, coded by effect class, reference state, direction of comparison, response level, deviation type and origin, fidelity level and the provenance of each reported value. The dataset also contains the screening log for 814 publications, the frozen analytical set of 476, the pilot validation blocks that established the accuracy of coding from bibliographic records, the claim ledger mapping every generalising statement onto the identifiers of the comparisons that support it, the discrepancy register, the metric harmonisation map, the comparison-direction adjudication, the gap-test log, a changelog of two corrections made after the data freeze, and an audit trail from which every numerical value reported in the article can be recomputed.