Key Takeaways
- Shovel grade sensing with MineSense ShovelSense mounts high-speed XRF sensors on shovel or excavator buckets to scan material for grade in real time, bucket by bucket.
- The system’s ML models are trained per ore body, so grade estimates adapt to your deposit’s mineralogy instead of using a generic calibration.
- MineSense BeltSense is the conveyor-belt equivalent: the same sensing concept applied downstream of the shovel.
- Real-time grade data feeds automated truck-diversion decisions, routing ore to mill, stockpile or waste without waiting for lab assays.
- This is a mature, commercially deployed technology, not an experiment.
TL;DR
Fit your shovel buckets with high-speed XRF sensors that scan every dig pass. Train the grade model on your own assay data, then connect its output to your dispatch system. Dispatch then routes trucks by measured grade, not by the block-model estimate alone.
What Is Real-Time Ore Grade Sensing?
Your block model tells you the expected grade of a mining block before you dig it. However, real grade varies inside that block. Some buckets carry rich ore, and others carry waste that the model averaged in. Lab assays from physical samples take hours or days. By then, the ore has often been dug, hauled and dumped.
Real-time ore grade sensing measures the material itself during mining. Sensors read the rock during loading, and software turns those readings into a grade estimate within seconds. That estimate arrives in time to change where each truck goes. The goal is simple: send ore to the mill and waste to the dump, based on what is actually in the bucket.
How Do I Monitor Ore Grade at the Shovel With AI?
The approach has three parts: sense the bucket, translate the signal into grade and act on the result.
How shovel grade sensing works
MineSense’s ShovelSense mounts a high-speed XRF sensor array directly on the shovel or excavator. It scans the bucket’s contents as the machine swings. That happens on every pass, not just on a periodic sample.
Training the grade model on your ore body
XRF readings alone are not grade. They are a raw elemental signal, and the ML layer translates it into a grade estimate. MineSense trains its classification and regression models per ore body. It uses your site’s own assay and mineralogy data as ground truth. Therefore, the system learns your deposit’s specific link between XRF signal and true grade. That per-site step matters, because a model trained on one deposit’s gangue will not transfer cleanly to a geochemically different one.
Routing trucks by measured grade
The payoff is automated truck diversion. Instead of routing trucks by block-model estimate and hoping, the real-time reading goes straight to the fleet dispatch system. Each truckload then goes to the mill, a low-grade stockpile or waste based on the measured grade. For the control-room side, see how AI supports mine dispatch.
How Shovel Grade Sensing Fits the Rest of the Mine
BeltSense applies the same technology on the conveyor belt. It gives you a second real-time grade checkpoint after trucks have loaded and hauled the material. That helps catch misrouted material and validate the shovel reading. Used together, the two systems confirm and refine the routing decision at two points in the flow. For the conveyor side in detail, see BeltSense.
Grade data at the shovel also improves planning upstream and downstream. Comparing shovel readings with the mining block model shows where the model overestimates or underestimates grade. Meanwhile, better routing helps keep plant feed steady. This is one of the more mature AI deployments on this site, running commercially at multiple operating mines.
FAQs
How accurate is shovel grade sensing?
Accuracy depends on the element, the ore type and how well the model is trained on local assay data. Because MineSense calibrates the ML models per ore body, a well-trained system tracks your deposit far better than a generic calibration would. Ongoing reconciliation against assays keeps it honest.
What does shovel grade sensing measure?
It uses XRF sensors to measure elemental signals from material in the bucket. Machine learning converts those signals into a grade estimate for the target metal. The system then classifies each bucket or truckload as ore, low grade or waste.
Does shovel grade sensing replace the block model?
No. The block model still guides where and what to mine. Shovel data refines the routing decision for each load and shows where the block model needs updating.
What is the difference between ShovelSense and BeltSense?
ShovelSense measures material in the shovel bucket at the dig face. BeltSense measures material on a conveyor belt further downstream. Many operations use both, so the second reading can confirm the first.
Try It With Geocluster
Turning a stream of shovel-mounted sensor readings into a trustworthy, deposit-specific grade model requires careful reconciliation against your assay database. That is the kind of applied research Geocluster is built to support. If you are evaluating real-time grade sensing, Geocluster can help you validate the model against your own data before you trust it in production. You can also read how the Geocluster Research Harness works.