What causes variance between surveyed and reported stockpile tonnage

Every reconciliation period, someone puts the survey number next to the scale house number and they don't match. The question isn't whether there's a gap. There's almost always a gap. The question is whether it's a few points of measurement noise or something that needs explaining to the mine manager, the client, or the regulator.

Where the two numbers come from

The scale house tally is a running total built one truck or one conveyor pass at a time, using belt scale calibration or weighbridge readings, against an assumed moisture content and a bulk density figure that's often set once and left alone for a season. The survey number is a one-time snapshot: a volume, times a density, converted to tonnes. Different methods, different error sources, measured at different points in time. They were never going to land on the same figure exactly, so the real work is figuring out whether the gap you're looking at is explainable or whether it's flagging something wrong.

The usual suspects

Bulk density. This is the single biggest lever on a tonnage conversion and the one most often held constant when it shouldn't be. A coal pile that's been rained on for two weeks doesn't weigh what the spec sheet says. Run a volume-to-tonnage conversion with last quarter's density on this quarter's wet material and you'll manufacture a variance that has nothing to do with the pile itself.

Moisture. Related to density but worth calling out on its own, because moisture swings both the scale house side (if trucks are weighed wet and reported dry, or vice versa) and the survey side (wet material compacts differently and the surface sits lower than a dry equivalent would).

Survey method and timing. A walked RTK survey covers a sample of points across the pile and interpolates between them. On a pile with a clean, regular shape that works fine. On a pile with slumped faces, segregation (fines at the toe, coarse at the crest), or a working face that's been reshaped by the loader since the last shift, point interpolation misses the irregular bits. A photogrammetric surface, by contrast, captures the whole visible surface rather than a sample of it, which is part of why the method matters as much as the operator's care.

Timing mismatch. The scale house total accumulates continuously. The survey is a snapshot taken on one day. If there's a load-out or a dump between the survey flight and the reconciliation date, that's not variance, that's just two different dates being compared as if they were the same moment.

Void space and compaction. Loose-tipped material sits at a different density than material that's been tracked over by equipment or settled under its own weight for months. A pile that looks the same size as it did last quarter can hold meaningfully more or less tonnage if the compaction state changed.

Material loss. This is the one everyone is worried about when they start the reconciliation, and it's also the smallest category in practice. Wind loss on fine, dry material, drainage loss on stacked ore during wet weather, and genuine theft or misreporting all happen, but they're usually the last explanation to check, not the first, because density and method errors are so much more common and so much easier to fix.

Getting a number you can defend

None of this means the scale house is wrong or the survey crew made a mistake. It means tonnage reconciliation is comparing two measurement systems with different blind spots, and the audit trail needs to show which blind spot explains which part of the gap. That's easier to do when the survey side is a full surface model rather than a handful of RTK points, because you can actually see where the pile shape changed, not just infer it from a single volume figure.

It also helps when the survey method doesn't require a person walking the crest of a pile that might be sitting on a soft base or an active face. A surface model built from weekly drone or crewed-aircraft flights gives you that full-pile picture without putting a rover and a crew member on top of material that moved last week for reasons nobody wrote down.

If your reconciliation reports keep landing on "survey error" as the explanation by default, it might be worth checking whether the survey method itself is the thing introducing the gap.