Skip to main content Skip to page footer

Understanding What Impacts a Mining Circuit’s Efficiency


The Challenge: 

A mining company needed to understand and resolve a change in the current efficiency of the electrowinning circuit at one of their mines. 

The Data Analysis Australia Approach: 

Through the use of a range of statistical tools, including breakpoint analysis and autoregressive linear models, the available data for the circuit were analysed.  After the changes in the current efficiency over time were understood, Data Analysis Australia investigated a range of potential causes. 

The Result: 

Breakpoint analysis demonstrated that the most recent change in efficiency was beyond previous trends and variability.  Investigation of the circuit’s efficiency leading up to this “break” identified an irregular event that preceded the efficiency changes. The client was then able to identify a particular sequence of events and take action to improve efficiency and prevent instability in the electrowinning circuit.

The Challenge

Understanding the nature of processes within a mineral processing plant in order to maintain efficiency is of great importance to mining operations.  Given the complexities of a mineral processing plant and the interdependent nature of the many factors that affect performance, obtaining insights through the analysis and interpretation of available data resources can greatly aid the operation.

Electrowinning (EW) is the recovery of metals, such as gold, silver and copper, from a solution by passing a current through the solution.   The current efficiency of an EW circuit describes the efficiency of the circuit in terms of how much metal is recovered for a given amount of electrical charge.  Data Analysis Australia engaged with a mining company that had recently observed changes in the current efficiency of the EW circuit at one of their mines to investigate the issue.  The initial aim was to confirm whether the observed change in efficiency was part of some longer-term trend or represented a discontinuity from past trends and patterns.

The Data Analysis Australia Approach

Data Analysis Australia was provided with several datasets containing daily observations of various measurements of both the EW circuit itself and the upstream solvent extraction circuit.  As well as observed current efficiency, these included measurements of the solvent’s composition at points across the circuit and various operational settings.  With a limited number of records readily available and many complex relationships, the data presented a number of challenges. 

Time was of the essence and Data Analysis Australia took a practical data-driven approach to the problem.  After conducting exploratory data analysis to quantify the characteristics of the datasets, more sophisticated statistical techniques were used to better understand the nature of the efficiency change.

After linking the solvent extraction and EW datasets sets, we generated a range of data visualisations to inspect and understand the relationships between the many variables.  From this we identified several variables of interest and used these to develop regression models to quantify the potential drivers of current efficiency.  

Diagnostics of these models revealed more complex relationships within the data.  This included identification of autocorrelation – correlation between observations separated by a period of time – in the current efficiency data.  To deal with this, we utilised more sophisticated models that include autoregressive terms which allow the model to incorporate the autocorrelated relationships in the data.  Failure to include such terms would have led to incorrect parameter estimates and misleading results.

Breakpoint analysis was also utilised to understand the timing of the changes in current efficiency.  This is a statistical technique used for determining whether there are any unexpected shifts in the data (generally in a series of observations made over time).  The breakpoint analysis enabled us to quantitatively assess the changes in the data, after accounting for known relationships, identifying points in time where the current efficiency had changed beyond what would be expected given the underlying trends and variability in the data.

The outcomes of the breakpoint analysis confirmed that the recent change in efficiency was not part of a longer-term trend or pattern but represented a clear departure from previous observations.  Previous breakpoints were also identified and further investigation suggested these were associated with an irregular event that impacted the circuit. 

To confirm this suggestion, the dates of all such events were obtained and the current efficiency following these events was evaluated.  Although changes in efficiency were observed after some of the events, steady efficiency values followed many others, suggesting that the relationship was more complex.  Further investigations revealed that it was only a particular combination of events that triggered the sudden changes in current efficiency.

This was critical for targeting aspects of the circuit that needed to be investigated further.  Once the causes of the change in current efficiency were understood, the client developed strategies to improve and maintain efficiency resulting in improved productivity.  

The Result

Throughout the project, Data Analysis Australia provided objective advice and recommendations using a statisticians’ understanding of the data.  We drew upon our extensive statistical expertise to choose the best statistical tool for each individual task rather than simply using a standard method with a limited toolset.  We worked collaboratively with the client in a compressed timeframe to investigate their data without preconceived ideas or hypotheses regarding the cause of the efficiency changes.  At the same time we acknowledge and exploited the specialist knowledge of the engineers and metallurgists familiar with the processing plant.  

This practical data-driven approach combined thorough exploratory data analysis with sophisticated statistical tools, such as autoregressive modelling and breakpoint analysis, leading to new insights regarding the EW circuit being provided and objective critiques of existing ideas.  This assisted the client to identify the key aspect of the circuit to focus on in further investigations.  Once implemented, the efficiency improvement strategies successfully resulted in higher levels of production and improved productivity valued in the order of $80,000 per day.

December 2015