Measuring a Stockpile While Loaders Are Still On It

Most of what is written about drone stockpile measurement quietly assumes the pile is sitting still. In a working aggregate yard, a sand and gravel operation, a coal or salt storage pad or a grain or woodchip yard, it usually is not. Loaders are feeding a crusher, trucks are being loaded out, a conveyor is building the pile from one end while a loader takes from the other, and nobody wants to stop any of it so a drone can go up.

That is a workable situation. It just has to be handled deliberately, because an actively worked pile fails in several directions at once and only one of those directions is the one people worry about.

What actually goes wrong, in order of size

The concern everybody raises first is material leaving the pile during the flight. That is usually the smallest error in the set. A capture over a single pile is a matter of minutes, and a loader working at a normal rate moves a quantity in those minutes that is small against a pile of any size. Worth noting in the report, rarely the thing that breaks the number.

The problems that actually break it, roughly in order:

• Equipment on or against the pile. A loader parked on the stockpile is reconstructed as part of the surface. So is a truck at the toe, a stacker, a conveyor gantry and a light plant. Photogrammetry has no concept of a machine; it returns the top of what reflected light.
• A toe that is not where it was. Working a pile pushes the toe out, pulls it back and smears it into the pad. The toe boundary is the single most sensitive input into a stockpile volume, and on a worked pile it is genuinely ambiguous rather than merely hard to see.
• A base surface that is changing. If the loader is cutting into the pad, or the pad has been scraped, graded or built up since the base surface was captured, the volume is being computed against a surface that no longer exists underneath the material.
• Two piles that have become one. Working adjacent piles merges their toes. Where one product ends and the next begins becomes an operational judgement, and it has to be made the same way every period or the per-product numbers wander even when the total is right.
• Dust and the atmosphere. A crusher running in a dry yard produces a dust plume that degrades image matching across part of the pile, which shows as noise or holes in the surface rather than as an obvious failure.
• Moving people and traffic, which is a flight planning and safety problem before it is a measurement problem, and which in practice is what limits where the crew can operate from.

The conventions that make the number defensible anyway

A volume on a working pile can be perfectly defensible. What makes it defensible is not precision, it is convention: a small set of rules stated in the report and applied identically every single period.

• One stated cutoff instant. The volume is as at a time, not as at a day. Pick the start of the capture, write it down, and reconcile the production and loadout records to the same instant rather than to the shift.
• One stated base surface, by name and date. Either a surveyed pad surface captured when the pad was empty, or an interpolated base beneath the current toe, and the report has to say which. These two choices routinely differ by more than every other error in the job combined.
• One stated toe boundary method. Break of slope, a surveyed pad edge, a fixed polygon reused each period, or a stated offset from a structure. The method is the thing that gets reused, not the polygon from a day the pile happened to be a certain shape.
• A fixed polygon per product where piles merge. If two products share a toe, draw the dividing line once, in agreement with the yard, and keep it. A moving line produces a transfer of volume between products that looks like a measurement error and is not.
• Equipment handling stated explicitly. Either the machine moved before the flight, or it was masked and the gap was filled by interpolation from the surrounding surface, with the approximate footprint noted. Silently reconstructing a loader into the tonnage is the single worst outcome here.
• The same flight parameters every period. Altitude, overlap, control and processing template held constant, so that a change between periods is material movement rather than method.

None of that is exotic. It is the same discipline that keeps a measurement series comparable over years, applied to a pile that will not hold still.

A working face is a different problem

There is an important distinction between a pile being drawn down and a face being cut. On a stockpile, the material sits on a pad that is not being changed, and the uncertainty lives in the toe and in what is parked on top. On a working face in a pit, the ground itself is being removed, which means the base surface is not a fixed reference at all and the comparison is surface against surface between two dates.

Both are measurable. They are not the same job, and a quote for one is not a quote for the other. A face needs the earlier surface to have been captured in the same frame, and it needs the survey's own uncertainty to be well inside the quantity being argued about, because there is no static base to anchor against.

When to take the twenty minute stand-down

Yards resist stopping, and for routine internal tracking they are right to. For some purposes, stopping is clearly the cheaper option, and this is the judgement call worth making in advance rather than on the day:

• Take the stand-down when the number goes to a third party: a month end or year end inventory, an auditor, a lender, a royalty or levy calculation, a dispute, or a reconciliation against scale-house tonnage that somebody is going to be asked to explain.
• Take the stand-down on the first capture of a pile, because that is when the base surface, toe boundary and product boundaries get established, and every later period inherits those choices.
• Fly it working for routine production tracking, a weekly or monthly trend, a sanity check against expectation, or anything internal where a moderately wider uncertainty changes nobody's decision.
• Fly it working when the alternative is not flying, which it often is. A number with a stated method and a stated uncertainty beats a guess from the pad, every time.

Stand-down in practice means clearing machines off and away from the pile and holding traffic for the capture window over that pile. It does not mean shutting the operation for a morning, and framing it that way to a yard foreman is usually the reason the request gets refused.

How to report it

A volume on a worked pile should arrive with its conditions attached, because the person reading it in six months will not remember them:

• The quantity, with the base surface named and dated and the toe method stated.
• The cutoff instant, not the date.
• Whether the pile was being worked during the capture, and what was moving.
• Any equipment masked, with its approximate footprint and how the gap was filled.
• Product boundary polygons as files, so the next period uses the same ones.
• An uncertainty that is honestly wider than a static pile would carry, with the reason given rather than a single number asserted.

That last point is the one people skip, and it is the one that prevents the argument. Two captures of the same pile can legitimately differ, and when they do the question that gets asked is what the method was. If the answer is written down, there is no argument. We have covered what to do when two surveys of the same pile disagree, and nearly every case in that article traces back to a convention that was never stated.