Answers are grounded in this paper and cite their sources.
Automotive radiators are critical heat-rejection components whose corrosion compromises leak-tightness, heat-transfer efficiency and reliability. Their compact aluminium-based, multi-interface architecture promotes coupled internal and external degradation, with signatures often hidden, indirect or confounded by fouling and residues, limiting conventional inspection. This review integrates radiator architecture, exposure conditions, degradation mechanisms, observable signatures, diagnostic approaches and industrial transferability within a unified corrosion-diagnosis framework. A PRISMA-guided workflow screened Scopus and Web of Science records published from January 2016 to 15 April 2026 using eligibility screening and full-text assessment. The synthesis evaluated corrosion mechanisms, diagnostic targets, data modalities, validation evidence and deployment constraints. The evidence indicates a shift from corrosion description toward monitoring, sensing, AI-assisted assessment and multimodal diagnosis. Corrosion initiation and observability are governed mainly by radiator geometry, brazed interfaces, sacrificial layers, coolant chemistry, wet-dry salt exposure and flow heterogeneity. Internal degradation produces coolant, pressure or thermal disturbances, whereas external corrosion yields visual, thermal or blockage-related signatures. Failure-relevant signals remain poorly accessible in tube interiors, headers and buried joints. Conventional inspection is deployable but surface-biased, while NDT, sensing, thermo-hydraulic analysis and computer vision improve observability but remain target-specific and confounded. Validation is limited by scarce ground truth, non-representative datasets, domain shift and deployment constraints. Reliable diagnosis requires linking mechanisms, signatures, measurements, models and decisions. Progress depends on multimodal, physics-informed inference, realistic validation, standardized acquisition and uncertainty-aware deployment.
No takes yet. Share an insight, caveat, or question.
Zarhyoun et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: