Predictive System for Performance and Detection of Anomalies in Fuel Consumption of High Tonnage Mining Trucks

摘要

It is estimated that there is a 10 % increase in fuel consumption due to operational practices in heavydutymining trucks (excluding autonomous trucks) involving between 65,000 and 100,000 liters ofdiesel per year per truck. These practices refer to the change of gears, acceleration, route geometry,transported cargo and road conditions. The aforementioned impacts on an overall cost between USD40,000 and USD 60,000 per year, which increases the cost of transporting minerals and impacts thephenomenon of global warming due to CO2 generationCurrently CAEX manufacturers have not included technology to measure fuel consumption rate[liters/hour], only very wide consumption ranges due to process variability. The existing solutions inthe market consist mainly of hardware (sensors) that require a high initial investment, fieldinstallation personnel and maintenance due to component failure.Therefore, a system for predicting the performance and detection of fuel consumption anomalies ofhigh tonnage mining trucks is developed based on Advanced Analytics, which allows reducing thecost of fuel and CO2 emissions without installing hardware. This system orients the focus on detectinganomalous transactions, predicting in real time the level of the fuel tank, measuringoverconsumption due to practical operations and generating feedback with alerts to users of differentoperational areas to take immediate actions: dispatchers, operator instructors, maintainers, costmanagement engineers and contract administrators.Its pilot implementation in The Pelambres mining has reduced the total consumption rate by 3.6 %,and reduced the amount of anomalous fuel transactions by 22,000 liters per month, guiding feedbackto operators with greater opportunities for improvement in their driving.