The question was asked, can we build a tool that uses live production data to dynamically predict the optimum date for a crusher shutdown?
Nearly all forms of mining involve rock crushing machinery, with some mines crushing and re-crushing in multiple stages, using various types of crusher. As large rocks are broken into smaller rocks, the internal crusher walls are subjected to enormous amounts of pressure and friction.
To prevent damage, the inside walls are lined with hard wearing abrasive resistant plates call wear plates. When the wear plates have worn thin the mine has to shut down its crushing circuit to replace them.
The shape and location of rock crushing machines often means it is impossible, or just too dangerous to inspect the condition of wear plates before the shut down. The hostile environment inside the crushers would destroy any measuring devices.
In mining, lost time equals lost production, so the timing of a shut down is crucial. Shutdown too early and you waste money by replacing wear plates that still have plenty of life left in them. Worse still, shut down too late and you risk damaging the crushing machine.
To help predict the optimum shut down date, a simple linear regression calculation of ore throughput wasn’t enough. The moisture and quartz content (quartz is more abrasive than most ores) need to be included into the calculation. Then as mining continues the latest production data is continually added and the model outputted a dynamic prediction of the wear plate condition. The tool benefits the operations and maintenance teams in the enormous task of planning and executing a shutdown.