BCMMINING TECHNOLOGY
Home / Platforms / Misroute & Compliance Detection
Ore in the waste dump is value you already paid to drill, blast, load, and haul. It does not show up as a variance, it shows up as grade that never arrived, weeks later, with no way to tell which loads did it. This checks every completed dump against what your block model and your destination configuration actually say, and puts it in front of someone who can still act on it.
Detection
Comparing a truck's assigned destination against where it dumped catches the obvious cases and misses the expensive ones. Each layer below looks at a different point where material and plan can diverge, and each can be run independently while the rules are tuned.
Every load's material checked against what the sink destination is configured to accept, read live from Minestar's own Destination Editor rather than a spreadsheet kept somewhere else. Rules reload on every poll, so a change made in Minestar takes effect immediately with nothing to redeploy.
The load compared against the block model material for the block actually dumped on. This is what catches high grade landing on a low grade pad, and it becomes the authoritative check on destinations flagged to accept all materials for source-side reasons, where the destination configuration says nothing useful.
The load compared against the block it came out of, which catches material mis-tagged at the loading tool rather than at the dump. Deliberate contact digging is respected: a load matching the block's configured alternates passes without noise.
A sink destination with no acceptance rule at all is itself a finding, not something to skip quietly. Switchable, so you can run silent while the rule set is still being built and then start surfacing the gaps once it is not.
One rule overrides the rest. An ore-group block dug as waste-group material alerts even when the alternates list technically permits it, because that loss is unrecoverable and it is worth ten seconds of a supervisor's attention regardless of what the configuration allows.
Findings are keyed on the cycle so a load alerts once and never again, no matter how often the service polls. The source block check fails closed: if its supporting data cannot be read, that layer is skipped for the cycle rather than firing on incomplete information. An alerting system nobody trusts gets muted in a week.
Reading that card: a 152.8 tonne load tagged as waste, from a block whose model material is also waste, so nothing was mis-tagged at the loading tool. It went onto a high grade stockpile. Layer 01 fired because that destination does not accept waste. This is the dilution direction rather than the ore loss direction, and it costs you on mill feed grade instead of in the waste dump. The same service catches both, and the card names the block and the rule so the fix is obvious.
Screenshots from production systems. Personnel details have been altered and database schema identifiers are masked.
The alert
No new application, no per-seat licence, and no dashboard anyone has to remember to open. A finding arrives as a card in a Microsoft Teams or Cisco Webex channel, and you decide who is in that channel: dispatch, ore control, geology, the pit supervisor, the superintendent, or all of them.
Before you commit
The service has a replay mode that reads history instead of live cycles and reports what it would have flagged, without sending anything to anyone. Before a single alert reaches a channel, you get a list of the loads it would have caught over the last month, from your own data.
Tonnes of ore that went to waste, on named blocks, on named shifts. That is either a number worth acting on or it is not, and either way you find out before committing to anything.
Replay is also how the false positive rate gets driven down before go-live, by finding the destinations and alternates your configuration does not describe properly yet. Nobody's first day of alerts should be a hundred cards.
Why it pays
Reconciliation tells you the grade did not arrive. It does not tell you which loads, which blocks, or which shift, and by then nobody can do anything except argue about it. Detection at the time of the dump is a correction; detection at month end is a post mortem.
Most operations have no systematic detection at all. What they have is a periodic look at dispatch records and a general belief that it is probably fine. Every load checked, every shift, is a different thing.
Because the finding names the block, the material, and the configuration rule that fired, it points at process and setup as often as at people. Most findings turn out to be a destination configured years ago that nobody revisited.
Delivery
Next step
Tell us what you run for fleet management and how your destinations are configured, and the replay over your own history is the fastest way to find out whether this is worth doing.