The separation of water from tailings (mineral processing waste product) is a critical process in mining, as it enables the recycled water to be reused within the processing plant. Sustaining a high density of tailings underflow increases the amount of water that can be reused. This reduces the operating cost for ongoing processing of the ore and improves the environmental sustainability of the operation.
The Humyn.ai community produced hundreds of predictive models. The best of these was deployed in a solution that predicts tailings density three hours in advance, allowing Newcrest to optimize the process. This has also made a significant contribution to the environmental sustainability of the operation - saving billions of litres of water.
"With Humyn.ai we are able to tap into a global network of data science talent. Together we've driven improvements such as reducing water usage, optimizing equipment, and use of core photography."
Co-Operative Bulk Handling (CBH) is Australia's largest exporter of grain. CBH conducts quality assurance inspections on grain to check for the overall quality and assist with grading of the grain. These visual inspections of grain samples check for indicators of different types of grain defects, including contaminants and weed seeds.
Visual inspection of grain is performed manually by a person. At the scale CBH operates, this manual inspection can be time-consuming, resource-intensive, and risky.
Through the use of computer vision machine learning, CBH increased the efficiency of this inspection process.
The Humyn.ai community produced multi-class classification solutions which accurately classify images of oat, wheat, barley, and weed grains. The solutions also automate the identification of grain defects.
“We quickly got hundreds of great machine vision models for classifying grain. It was easy to compare the results and select the best one. That process gave us confidence in the results we wouldn't have had if we got just one solution from a vendor. Plus, our team loved seeing talented people from around the world contribute to solving our problem. It was also powerful to be able to harness a trusted local provider in Humyn.ai who had impressive connections into a global data science network.”
Pressure changes indicate when the circuit breakers in electrical substations may need to be fixed. Predicting pressure changes in advance will enable better maintenance planning and reduce the impact to customers and the environment.
Western Power is an electrical utility which operates circuit breakers that use Sulphur Hexafluoride (SF6) gas to isolate and quench electrical arcing when the circuit breaker closes. The SF6 gas can leak over time, typically due to failing seals, to the point that the circuit breaker needs to be taken offline and the gas replaced.
Data scientists in the Humyn.ai community built machine learning models that predict when the SF6 gas pressure is likely to drop below critical threshold values with sufficient time (between 2 and 6 weeks!) to refill SF6 and repair the circuit breaker. A win for Western Power, and the environment.
"For the first time, I saw senior leaders and people in operational roles get excited about data! Working with Humyn.ai we got to connect with some of the best minds in data science. As a result, we're improving safety and predicting maintenance needs."
Because copper is a highly efficient conduit, it is used in renewable energy systems to generate power from solar, hydro, thermal and wind energy across the world. Copper helps reduce CO2 emissions and lowers the amount energy needed to produce electricity.
OZ Minerals has extensive exploration experience across the globe and is exploring for copper deposits across Peru. Peru is highly prospective for a range of economic minerals and metals. At present a great deal of Peru is under-explored and this presents a lot of opportunity to find new deposits.
A significant amount of country-scale satellite data has been collected and made publicly available. This data can be used to identify new regions in Peru that could host copper deposits.
The data scientists in the Humyn.ai community used this data to generate feature maps that highlight geological patterns that are now helping OZ Minerals in their exploration efforts.
“I have had the experience of running multiple data science crowd challenges with Humyn.ai. Their team helped me to access a large network of innovators in order to achieve my goals for creating new methods of building out datasets, proxying data from novel sources, and testing the feasibility of several completely out of the box ideas. The culture at Humyn.ai is very open to divergent thinking and their people are skilled at putting together a pragmatic approach to solving highly unusual problems.”
This project used data to understand if our body clocks, circadian rhythms and work patterns influence the likelihood of safety incidents occurring at work.
The Humyn.ai community built models that predict when incidents are more likely to occur, and provided detailed explanations of their findings and predictions in a report.
Data sources included both employee timesheet data and incident data from Western Power. Community members also were free to incorporate 3rd party data, such as weather data.
Downhole gauge pressure readings are important for monitoring and optimization of well operations. But due to the harsh working environment, the sensors can malfunction. Additionally, it is not feasible to equip all the wells with these sensors.
The Humyn.ai community built solutions that accurately predict downhole gauge pressure based on other parameters.
Autoclaves extract gold from rock. Optimizing the temperature in the autoclave keeps it running at peak efficiency over time, allowing Newcrest to reduce costs and emissions.
Existing data allowed operators to manage variables such as temperature, water content, oxygen, flow rate, and raw ore feed. However, factors outside of their control, such as grade and sulphide content can change the operating temperature and have a direct impact on operations. Could Newcrest achieve new levels of efficiency in the autoclave by implementing an ML solution to predict temperature?
The Humyn.ai community developed several production-ready ML models that accurately predict the temperature one hour in advance inside the autoclave, significantly reducing costs and emissions.
During regular operations, CSG wells can fail, rendering the well offline until a workover (repair) is performed to bring it back online and producing gas. The speed at which the scope of a workover is selected and performed is critical.
However, determining the scope of a workover, such as what parts need repair or replacing, is currently a lengthy, manual, human process that typically involves multiple stakeholders in energy businesses.
The team at Origin have wanted to use historical data of workovers to inform and even predict future workover scope.
The Humyn.ai community developed ML models that speed up the workover decision process to get a well back online faster. This increases efficiency and reduces costs of operations.