Model-based localisation of water-use inefficiencies in deep-level mines

Abstract

The South African deep-level gold mining sector is facing profitability challenges due to increasing operating costs and declining production trends. Electricity costs can account for over R16 billion of the annual cost of operating a deep-level mine. This is a concerning factor and has led to the early closure of some South African mining companies. Integrated water reticulation systems, which consist of refrigeration, water distribution, and dewatering, contribute to around 28% of the overall electricity consumption of deep-level mines. Research has shown that improving energy efficiency can contribute to reducing operating costs of companies, leading to better profits. An area presenting the potential to lower operating costs in deep-level gold mines is by addressing inefficiencies in the water reticulation system. These inefficiencies are represented in the water reticulation system in all three phases (refrigeration, pumping and water distribution). By reducing the wastage in the water reticulation system, refrigeration systems can deliver chilled water successfully for underground use and the volume of water required to be dewatered by pumping systems is reduced. This results in more chilled water for mining and cooling, better production rates, and a safer working environment. Harsh underground conditions and flow rates of around 300 L/s result in poor conditions of pipe networks used for water distribution. The water distribution network presents an opportunity to improve efficiency of the water reticulation system. This is achievable by reducing water inefficiencies in the complex distribution network. However, these water distribution networks currently require manual labour to walk up to 20 km along the underground pipelines to find and report on the found water inefficiencies. It is evident that a need exists to develop a model to assist with the localisation of water inefficiencies in deep-level mines using the available infrastructure and instrumentation. A method was developed to identify anomalies, classify the type of water-use inefficiency, and address these water inefficiencies in the water distribution network. The method incorporated guidelines and procedures for effectively locating water inefficiencies. For context, the process of limiting inefficiencies to a particular area is referred to as localisation. The localisation of water inefficiencies assists in reducing the need to walk along each pipe in the water reticulation system during underground audits. Mine A was subjected to the developed method. Water inefficiencies could be detected from surface monitoring of available data. Validation of these results was obtained by evaluating the accuracy of the model by comparing located inefficiencies with false alarms. The model had a precision rate of 93%, thus identified, classified, and assisted with locating 93% of the 68 detected flow anomalies. The investigation implemented on Mine A indicated that the average water distribution network inefficiency resulted in a flow increase of approximately 11 L/s when open ends occurred. This increase in flow effectively led to a reduction of chilled water available to perform mining activities. Chilled water that was wasted before could be used and an additional electrical cost saving of R8.6 million per annum was realised after corrective action was taken.