Predictive maintenance

Improving early failure detection and reducing machine downtime

Mining, Natural Resources

In any capital-intensive industry, productivity depends on keeping the machines running. This is increasingly challenging in mining where operations are often located in remote areas.


Mechanical failures or unexpected maintenance can quickly bring an entire operation to a halt—and every hour of downtime translates to lost profits. A global mining firm was seeking to reduce the amount of time its equipment and machinery were inoperable due to maintenance and repair.

Insight and Action

QuantumBlack adopted a two-stage approach to analyse the reliability of the mining company’s dry-plant machinery. We used a proportional hazards model to calculate the probability of machinery failure using a range of factors, such as total number of operating hours, type of ore, load volume, and weather. Since modern machinery contains multiple sensors that generate data on performance, we then created analytical models capable of detecting incipient failures ,such as motor voltage, current, and temperature. Thanks to this combination of analyses, we uncovered opportunities for the company to use early detection to solve persistent failures on its conveyor belts.


  • 88Percentof failures detected within a seven-day window