Advanced data analytics

A new competitive advantage to increase energy efficiency in surface mines

Dr Ali Soofastaei: Postdoctoral Research Fellow, The University of Queensland, Mining3, Australia.

It’s no secret that data analytics is leading to major changes to data-intensive businesses across the world. It is the science of examining raw data to gain new insights that will improve decision making about business processes or products.

The value gained through data analytics is cost reductions, faster and better decision making and improved design and operation of new products and services[1, 2]. The ability to analytically simulate a large range of production or operational scenarios improves understanding and reduces risks and costs.

The uses for data analytics are many, often in areas that may not have been considered before. One sector that sees lots of potential in data analytics is the mining industry[3]. For an industry generating trillions of dollars every year, data analytic is no longer a luxury but viewed as a necessity.

There are many phases of the mining process where data analytics can be put to valuable use. Four key application areas where large operational costs and business value are created are:

  • Extraction of ore
  • Mining and handling mined materials
  • Grinding materials in preparation for processing
  • Separating and concentrating the usable components into saleable products.

In the current financial climate, a particular focus for many companies is identifying how best to make efficiency improvements in the second phase[4]. Without using data analytics to explore what most influences the efficiency of this phase, there is little likelihood operators will be able to identify critical operating parameters and optimise the process and use of their high asset value equipment.

As an application of data analytics in handling mine materials, an investigation into the use of Artificial Intelligence methods, to minimise energy consumption by haul trucks, has been completed by a group of researchers from The University of Queensland and CRCMining.

Mining is one of the most energy-intensive industries in Australia consuming a vast amount of energy in every stage of its operations, across the main exploration, exploitation and minerals processing phases. These operations can be optimised to create more energy efficient processes, particularly through using better management strategies. The mining system employed – the mining method and haulage operations – primarily determines the type of energy sources required on the mine site. Haulage operations are of particular interest, where haul trucks and excavators/loaders work in tandem to meet production goals and schedules. These haul trucks generally use diesel fuel as an energy source and as a result, the mining industry is constantly seeking ways to make them more fuel efficient and cost effective. A large number of factors contribute to the fuel consumption of haul trucks such as: haul road properties and condition; truck operation; truck speed; rolling resistance and environmental conditions (See Figure 1).

Truck effeciency Sml

Figure 1: A schematic diagram of a typical haul truck and effective key factors on truck performance[4]

The data collected from the Truck’s Vital Information Management System (VIMS) at various surface mines confirm that Truck Speed (S), Gross Vehicle Weight (GVW) and Total Resistance (TR) are the most significant factors and must be considered when analysing mine productivity, diesel energy consumption, greenhouse gas emissions and associated costs. The project aimed to investigate the energy efficiency opportunities for truck haulage operations associated with these parameters in surface mines using advanced analytic models.

In this project Fuel Consumption index (FCIndex) has been examined for the selected haul truck used in a large open-cut coal mine in Arizona, operating on an eight-hour shift roster. This index calculates the amount of diesel fuel consumed by the truck to move one tonne of mined material through one kilometre of haul road. The data collected by a special onboard computer device installed on a CAT 785D haul truck for a duration of 12 months. These data were analysed by an innovative computer model working based on a complex Artificial Neural Network (ANN). The developed advanced data analytic model simulated the correlation between FCIndex, TR, S and GVW (See Figure 2).

Truck effeciency graph

Figure 2: Correlation between GVW, S, TR and FCIndex based on the developed ANN model for (CAT 785D), Arizona, USA July 2015

In order to generate the proposed ANN model, 1,000,000 data points were randomly selected from the collected real datasets from the mine site. In order to test the neural network accuracy and validate the model, 1,000,000 independent samples were used again. The results show good agreement between the actual and estimated values of fuel consumption.

In summary, the results of this project show that data analytic can play a valuable role in minimising haul truck fuel consumption. By identifying the best values for the key control parameters the energy efficiency in haulage operations in surface mines can be increased.

Further, the relationships between truck fuel consumption and the influencing parameters are not time stationary. Engines wear, roads need grading, driver performance changes. The analysis has demonstrated that changes in these variables will influence fuel consumption, changing it from the initial mine design assumptions.

Periodic assessment of the changes in these control variables will enable the ANN models to be updated, to ensure optimised performance can be maintained. The research team’s current work includes analysing how best to manage these parameters in a holistic manner, so that the mine’s overall operation is better tuned to create optimised performance along the end-to-end mining value chain.

  • Davenport, T.H., J.G. Harris, and R. Morison, Analytics at work: Smarter decisions, better results. 3, ed. 2. Vol. 2. 2015, Boston Harvard Business Press.
  • Eckerson, W.W., Predictive Analytics. Extending the Value of Your Data Warehousing Investment. TDWI Best Practices Report, 2007. 1: p. 1-36Soofastaei, A., et al., Development of a multi-layer perceptron artificial neural network model to determine haul trucks energy consumption. International Journal of Mining Science and Technology, 2016. 2(1): p. 156-172.
  • Soofastaei, A., et al., Development of a multi-layer perceptron artificial neural network model to determine haul trucks energy consumption. International Journal of Mining Science and Technology, 2016. 2(1): p. 156-172.
  • Soofastaei, A., et al., Simulation of Payload Variance Effects on Truck Bunching to Minimise Energy Consumption and Greenhouse Gas Emissions. Coal conference, Wollongong University, 2015. 2(1): p. 255-262.(/span>