Extracting Value from Data

Miners have understood the importance of recording their activities since antiquity with Sumerian scribes marking clay tablets to record mine depths and shipments. The technology has changed but miners still use tablets to collect data on their operations.

Given current volatility in commodity markets, the mining industry is creating additional value in existing operations by reducing costs and increasing production. This involves coordinated systems that allow data from across the operation to be quickly interrogated as required. These data resources can include everything from observations collected by automated sensors through to reports submitted by maintenance teams. Utilising statistical tools to engage with the data allows each component of the operation to be understood at the global level. Key challenges are then focused on and productivity is enhanced through the resolution of bottlenecks, wastage and underperformance across the myriad activities that take place from “mine to market”.

There are challenges to overcome. The data resources for a single operation are often held across many entrenched systems built for the specific needs of each business unit. These systems must be consolidated, breaking down the “data silos” that segregate an operation, distilling key information from each data source and developing effective data linkage algorithms.

However, the mere aggregation of masses of data is not enough.  Big data does not identify opportunities by itself and harvesting data does not spontaneously create value.  There is a risk that analytical results are assumed to be useful and correct simply because the analysis incorporates big data.  This has been described as “automated arrogance” and is a pitfall that must be avoided amongst the ongoing hype - the value of a data warehouse is not the amount of data it holds but the new insights it delivers.

A solid understanding of the core concepts of statistics is fundamental to any big data analysis.  The value of big data is optimised when it is combined with more proven forms of information collection and innovative statistical methods. This not only brings relevant focus to the analysis but militates against misleading results and unproductive decisions, such as confusing correlation with causation.  The critical eyes of a statistician see past the data traps to the valuable information hidden amongst big data.

Predicting when equipment will fail is one area that greatly benefits from statistical analysis.  With the aid of sensors that send data back in real-time, statisticians can develop models that predict when equipment will be in need of repair.  Such models draw on many sources of information to make accurate predictions including utilisation rates, energy consumption and environmental conditions.  Even text mining of maintenance reports can be used to identify failure indicators.  By using big data resources to make these predictions, mining operations not only improve efficiency but prevent costly and potentially dangerous catastrophic failures. 

Despite some common misconceptions, mining involves much more than just digging holes.  By leveraging the data sources held across a mining operation’s value chain, we can achieve productivity increases, waste reduction and safer workplaces.  

For further information, please contact daa(at)daa.com.au, or phone 08 9468 2533.

December 2015