NTWIST has witnessed the challenges that come with a mine's use of outdated assumptions for material flow in stockpile tracking. One of the most expensive mistakes we've seen occurring in this space is the lack of visibility over the piles as it leads to value loss, incorrect routing, and overall reconciliation problems.
In this piece, we'll discuss why the inaccuracy with stockpile information occurs, what it incurs in terms of cost, and how AI-based tracking is helping mines eliminate this guesswork in its entirety.
Stockpiles are meant to act as a buffer between the mining operations and the processing plant, but in cases where material is dumped without high confidence in its grade or during reclaim are executed without the right contextual information, the variability in the orebody that is being processed runs the risk of being invisible.
The uncertainty around the stockpile's content is driven down to two main points;
disjointed data systems among the dispatch, geological and survey functions,
and
The delay in measurement methods that are used may take days or weeks to validate the content of a stockpile.
When this information is fed into the hands of a planner, the stockpile may well have different properties than what is measured. With that information being fed into a precision-sensitive flotation circuit or blend model, the cost of this uncertainty escalates.
With recent developments in AI, drones and digital modeling, mining operations are in a unique position to track ore movement with near real-time accuracy, which not only enables visibility on how a stockpile's volumes change but also what has been delivered to it or taken from it.
Drones fitted with LiDAR or photogrammetry scanners can be used to frequently scan active stockpile areas, while corresponding AI models transform this information into digital elevation models or volume comparisons to provide instant feedback on stockpile movements (Toll Uncrewed Systems, 2024).
By providing this information alongside real-time dispatch information, AI models can go a step further and verify that what has been delivered to a stockpile is in fact coming from the desired source. This enables mine planners and metallurgists to take proactive measures to avoid any deviation in the content of a stockpile, reclaim feeder or blend circuit, which directly contributes to increased performance.
An open pit mine takes advantage of weekly drone passes over three stockpiles that are currently active in the operation. AI tracking systems transform the point of cloud data into volume estimates, which then get compared against the volume of material delivered to the stockpile as per the dispatch system. In one of the stockpiles, it is revealed that several trucks have been incorrectly routed to contribute to its overall volume.
The AI model flags several trucks that have contributed to the growth of this stockpile as coming from Block 27, which has marginal Cu-Au content, having been mistakenly routed to the high-grade stockpile. In response to this discovery, mine planners are now able to update the routing and dispatch systems so that any trucks that should no longer be contributing to this stockpile are avoided until such time as the mistake can be rectified on the ground.
In comparison to a mine that does not have this level of visibility into their stockpile movements, this mistake would not have been caught until discrepancies started to appear in their mill to feed grades.
More accurate feed forecasting and increased blend or recovery prediction - The reconciliation process that informs these values is greatly assisted through increased visibility and the ability to apply this information in real-time to blend circuits or reclaim feeders.
Less rehandling spend - Trust in the contents of a stockpile reduces the amount of time trucks or equipment to spend rehandling material across the mine site.
More informed metallurgists and planners - With increased accuracy and visibility, metallurgists and mine planners are operating from a position of increased knowledge and control.
More accountability across operations, shovelers and dispatch - Everyone has a role to play in the maintenance of a connected material flow chain.
If you can't see it, you can't manage it. At every level, visibility into the contents of a stockpile enables informed decision making, which is key to managing risk in stockpile management.
Drones + AI = advantage. Taking advantage of survey technology and AI to gather, analyze, and apply this information is crucial to reaping the rewards of increased visibility. Dispatch data completes the triad.
Fix misrouting before it becomes a loss. The earlier these issues can be flagged and resolved, the less risk there is that they will affect overall operations. In many cases, these issues can be resolved in the short-term without having any long-term consequences for the mine.
NTWIST enables mines to move from reactive stockpile management to proactive decision making by leveraging AI to enhance operations and planning. With new models that incorporate dispatch data, reconciliation information and AI capabilities, the guesswork involved in stockpile management can be eliminated for good.
After all, as we've seen, the risks involved with incorrect stockpile management can be costly, and with the power of the right technology, unnecessary.
AI-powered stockpile tracking is a method of using drones, sensors and AI to monitor stockpile volume, movement and material flow in near real time.
By enabling real-time reconciliation, AI is able to identify discrepancies, misrouting, and overall stockpile changes with greater speed and accuracy.
With an integrated approach to stockpile data and geological and dispatch information, AI can vastly improve grade visibility and reconciliation.
Through the use of LiDAR or photogrammetry scanners, drones can capture information about stockpile volume and movement.
Improved feed forecasting reduced rehandling costs, improved blending capabilities, and informed decision making.
Pramodh Chandrashekar is an industry and technology expert and Senior Researcher at NTWIST. His expertise includes industrial AI, mining and manufacturing technology, digital transformation, and operational optimization.
He focuses on applying AI and data-driven technologies to improve operational efficiency and business performance.
LinkedIn: Pramodh Chandrashekar
Explore Mine-to-Mill Optimization
ReferencesToll Uncrewed Systems. (2024). How Drones in Mining Boosts Efficiency. Retrieved from https://www.tollgroup.com/news-and-media/2024/how-drones-in-mining-boosts-efficiency
Mine Australia. (2023). The Impact of Drone Technology on the Mining Sector. Retrieved from https://www.mining-technology.com/features/the-impact-of-drone-technology-on-the-mining-sector/