Overview

Unplanned grinding mill failures cost a leading gold producer USD 1.5M/day. Tech Mahindra's predictive analytics solution, built on 640+ historian files and FMECA-driven machine learning, predicted failures six months ahead with 85%+ accuracy, identified root causes, and delivered component-level remaining useful life and a system health index, transforming maintenance from reactive to predictive.

ai-optimized-comminution-circuit-thumb

Industry Challenge

A leading producer of gold and the by-products of copper, lead, zinc, and silver, operating across the mining value chain, including exploration, mining, mineral processing, and metal extraction, experienced unplanned shutdowns at its mineral processing plants, beginning with the grinding mill circuits. Key challenges include the criticality of processing technique, million-dollar losses per day, invisible mechanical failures and poor operational visibility.

Our Approach and Solution

To address the unplanned mill shutdowns, we deployed an AI-infused predictive maintenance solution built on four principles, data ingestion, processing, modeling, and validation. This step shifted the operation from reactive to predictive, insight-driven maintenance across gearboxes, pinions, and bearings in the grinding circuit.

Unified Data Foundation via Kepware

Ingested 640+ historian files through Kepware industrial connectivity software to build a reliable, time-sequenced data backbone for the comminution circuit and its critical rotating components.

Engineering-Led Pattern Mapping

Fused engineering insights, failure logs, and operational data to establish relationships between equipment behavioral patterns and specific failure modes across critical assets.

ML-Driven Forecast with FMECA Root Cause

Trained ML models to forecast failures, estimate remaining useful life (RUL), and map predicted root causes against the FMECA framework for severity, occurrence, and detectability.

System Health Index for Proactive Action

Generated a system health index map to prioritize maintenance interventions, enabling the maintenance teams to act before critical failures disrupted mill availability and production throughput.

Business and Community Impact

The engagement delivered greater visibility into the grinding mill circuit, converting reactive maintenance losses into predictable, planned interventions and translating into measurable business, operational, and workforce safety gains for the client.

  • Avoided losses of up to USD 1.5M/day from unplanned grinding mill shutdowns through 85%+ accurate failure prediction up to six months in advance.
  • Higher plant availability & throughput by shifting from reactive to predictive maintenance across gearboxes, pinions, and bearings.
  • 15–20% MRO inventory optimization and fewer emergency maintenance events, enabled by component-level RUL and health-index-led planning.
  • Improved worker safety by reducing exposure to unplanned failures and high-risk emergency repairs on rotating mill assets.