Abstract
Bayesian statistics provide a flexible framework for improving maintenance and reliability strategies in mining operations where equipment performance degrades over time. This study applies a Bayesian Hierarchical Linear Model (HLM) to assess the reliability of SAG mill operations by modeling power consumption across different liner wear stages: initial, middle, and end. Key operational variables—such as RPM, throughput, ore hardness, coarse fraction, and copper grade—were incorporated into a probabilistic structure that captures performance shifts as liners degrade.
By treating liner phases as group-level effects, the Bayesian model quantifies uncertainty, identifies degradation patterns, and provides dynamic insights into the mill’s operational reliability. The results demonstrate that RPM’s impact on energy efficiency varies significantly across liner stages, with worn liners exhibiting reduced grinding performance and higher energy requirements. This approach supports predictive maintenance by facilitating earlier identification of efficiency losses and informing stage-specific operational adjustments.
Bayesian inference, using Markov Chain Monte Carlo (MCMC) sampling, enables robust estimation even with operational variability, integrating both historical performance trends and real-time process data. The hierarchical structure ensures that stage-specific variability is captured without overfitting, offering credible intervals around parameter estimates to reflect uncertainty.
In conclusion, Bayesian hierarchical modeling proves valuable for dynamically assessing equipment reliability in mining, supporting proactive maintenance strategies, improving uptime, and optimizing operational costs in SAG mill processes.