Orebody Learning™ is a process that enables operations to learn about their orebody directly from operational experience. It is performed using PETRA’s proprietary digital twin technology: Data Fusion Ore Tracking & Machine Learning.
Orebody learning is a process of using machine learning algorithms to analyze geological data and gain insights into the mineral deposits within an orebody. An orebody is a natural concentration of valuable minerals that can be extracted economically from the Earth’s crust.
In the mining industry, orebody learning is used to optimize the mining process and maximize the extraction of valuable minerals. By analyzing geological data, such as geophysical surveys, drilling results, and geological maps, machine learning algorithms can identify patterns and relationships between different variables. These insights can then be used to build predictive models that can help mine operators make better decisions about where to mine and how to extract the minerals.
Orebody learning can be used to optimize many aspects of the mining process, including mine planning, mineral processing, and waste management. For example, by analyzing geological data, machine learning algorithms can identify the areas of the orebody that are most likely to contain the highest concentrations of valuable minerals. This information can then be used to develop more efficient mining plans that focus on these high-grade areas, reducing the need for excessive drilling and excavation.
Additionally, orebody learning can be used to optimize the processing of ore, which is the process of separating the valuable minerals from the surrounding rock. By analyzing the mineralogy of the orebody, machine learning algorithms can identify the most efficient processing methods, reducing the amount of energy and resources needed to extract the minerals.
Overall, orebody learning is an important application of machine learning in the mining industry, as it can help mine operators make more informed decisions about where and how to mine, reducing the environmental impact of mining and increasing the efficiency of the process.
Orebody learning provides:
- Site-specific models using a mine’s own data
- High granularity block-by-block predictions
- Updated block models as soon as new operational data has been provided
- File formats which are platform agnostic, designed to connect with in-house or third party data platforms
The problem it aims to solve
Previously, orebody knowledge was limited to:
- Grade & product quality
- Geotechnical characterisation
Many mining operations suffer from the inability to learn from experience, orebody learning using digital twin technology aims to bridge this gap.
Expanded orebody knowledge provides information about:
- What is the best way to process the ore
- What is the best way to blast the ore
- How to be more energy efficient
- How to predict material handling properties
- How to predict recovery
- How will certain geology blast
- How quickly will diggers get through certain geologies
- How will geology impact downstream performance
Orebody learning powers PETRA products and applications
Considered the heart of all PETRA solutions, orebody learning allows us to create decision support tools to predict, simulate & optimise using a mathematical model of how your mine works.
- Increase profit margin/ recover benefits faster
- Optimise the value chain for geological variability
- Support operational decisions based on data driven models
- Integrate with existing systems(software) and workflows
- Get more value from existing data
- Create a repository of geology linked to performance
- Create 10’s of thousands of data records
How does this compare to other solution providers?
- They rely upon 20th century empirical models for example Orica’s IES.
- Provide low granularity 3D models using spatially limited rock testing data
- Provide forward only predictions without a feedback loop
- Force mining companies to use their data platform
- Don’t use digital twin technology to improve orebody knowledge
PETRA products & services utilising orebody learning