Confined blast propagation and geometry-induced pressure redistribution in rectangular tunnel networks: A validation-bounded loss-prevention assessment

Underground industrial service tunnels, protective access passages and utility corridors can transmit explosion-generated pressure waves after an external or near-portal event. This study integrates reduced-scale rectangular tunnel evidence, archived LS-DYNA ALE modelling information and tunnel-network geometry observations from an archived source programme into a validation-bounded loss-prevention assessment. The source programme included a 30 cm x 30 cm steel tunnel, C4 charges of 100-350 g, a 60 cm external stand-off and internal gauge locations at L/D = 0.07, 1.00, 3.00 and 5.67. The archived numerical workflow used C4 JWL modelling, a null-material air domain and Eulerian-Lagrangian coupling. The new contribution is not a new blast-test series or a newly executed numerical campaign. It is an evidence-controlled reanalysis that (i) defines four interpretation tiers, (ii) extracts a case-specific T-junction branch-to-main local peak-overpressure indicator, (iii) consolidates geometry mechanisms into a screening map, and (iv) converts these findings into an action-oriented loss-prevention workflow. Because complete raw pressure-time histories, model decks and repeat-test statistics are not available in a unified archive, the manuscript does not claim design-grade validation. The evidence tiers are calibration check, trend-level support, ratio-level warning and geometry-risk flag. The reanalysis identifies where simplified straight-tunnel attenuation assumptions are insufficient. A recovered equal-section T-junction dataset gives a branch-to-main local peak-overpressure indicator of approximately 0.53. L-bends, open chambers and closed chambers are treated respectively as reflection-concentration zones, expansion/discharge features and recompression-risk spaces, but without universal design coefficients. Three engineering deductions follow from the reanalysis: branch-ratio dispersion must be retained rather than collapsed into a deterministic reduction coefficient; sequential features can compound branch loading and recompression risk; and the assigned evidence tier determines whether preliminary screening, rerun, pressure-time validation or layout redesign is required. The workflow helps engineers prioritise CFD/ALE reruns, pressure-time validation and layout redesign in underground loss-prevention planning while clearly separating screening-level interpretation from design-grade extrapolation.