Metodología Bow Tie y redes bayesianas en gestión de seguridad para reducir accidentes por rocas en minera del sur peruano

Rockfalls are the leading cause of fatal accidents in Peru’s underground mining (27% of 494 fatalities, 2000–2019). This paper presents a safety-management framework that couples Bow-Tie analysis with Bayesian networks to prevent rockfall events in a southern Peruvian underground mine (Julcani). The approach integrates geomechanical characterization (RMR, Q, GSI), process mapping and IPERC-based hazard identification to define causes, top events, consequences, and critical controls. The Bow-Tie structure is converted into a Bayesian network to quantify prior likelihoods, update them with operational evidence (inspections, timeliness of support installation, ground-movement monitoring), and rank barriers by expected risk reduction. Validation contrasted initial and residual risk for high-exposure tasks (scaling, ground support, drift development). The framework reduced risk ratings from High to ALARP in the most critical tasks, improved compliance with critical controls, and established a repeatable monitoring and continuous-improvement cycle. Findings indicate that combining a transparent causal model with probabilistic updating strengthens decision-making under uncertainty and can be replicated in comparable underground operations.