---
title: "Fourteen Stockpiles and a Six-Hour Walk: The Month-End Tonnage Variance Your Controller Keeps Writing Off"
description: "Six hours a month pacing toe lines still leaves a 3% variance. What nightly drone flights do to stockpile tonnage, belt scale drift and highwall checks."
canonical: https://callsphere.ai/blog/fourteen-stockpiles-and-a-six-hour-walk-the-month-end-tonnage-variance
category: "Industry Solutions"
tags: ["aggregates", "stockpile inventory", "autonomous drones", "belt scale", "quarry operations", "workplace examination"]
author: "CallSphere Team"
published: 2026-07-11T08:55:25.000Z
updated: 2026-07-25T23:18:52.373Z
---

# Fourteen Stockpiles and a Six-Hour Walk: The Month-End Tonnage Variance Your Controller Keeps Writing Off

> Six hours a month pacing toe lines still leaves a 3% variance. What nightly drone flights do to stockpile tonnage, belt scale drift and highwall checks.

## Fourteen piles, six hours, and a number nobody in the office believes

Fourteen stockpiles. Six hours. That is what a proper month-end count costs at a mid-sized crushed stone operation — the superintendent and a loader operator pacing off toe lines, arguing about whether the #57 pile has been robbed off the back, and converting cone volumes to tons with a density factor somebody wrote in a notebook in 2011.

Then the number goes to the controller, who compares it to the book quantity — belt scale production minus shipped tons off the tickets — and finds it off by two or three percent. Every month it gets an adjusting entry, because the alternative is admitting nobody knows within a few thousand tons what is in the yard.

An autonomous stockpile flight is a drone that lives in a weatherproof dock on your property, launches itself on a schedule, photographs the same piles and faces from the same positions every time, and hands your superintendent a tonnage-by-pile report and a list of physical changes before the first truck crosses the scale. That is the thing that stopped being a pilot project in 2026.

## Where the variance actually comes from — it is four things, not one

Belt scale drift is the big one. A belt scale that has quietly gone one and a half percent high does not announce itself; it manufactures phantom production month after month, lowers your reported cost per ton, flatters the plant's numbers, and builds an inventory on paper that is not on the ground. You find it at year end, when the auditor wants a physical count observed.

Then the loader operator's memory: small yard sales, two loads of screenings that went out Saturday under the wrong material code, fines blended into base and never sold as fines. Then moisture and degradation — a wet pile weighs differently than a dry one, and material that sits through a winter of freeze and thaw does not come back out with the gradation it went in with. Then genuine shrink: tons tracked off on tires, pushed into the haul road, or used on site and never ticketed.

None of those four are mysterious. They are just invisible unless you measure often enough to see them move. Measuring once a quarter tells you the total; measuring every week tells you which of the four it is.

None of that is fixed by the survey you already buy. A contractor drives out, sets ground control, flies the yard, and sends back volumes a week later, four times a year. Useful for the bank, the auditor and the reclamation bond; useless for running a plant, because the answer lands after the month has closed and after the crusher schedule that would have used it was set.

It also does nothing for the other thing that walk eats: the physical look at the highwall, the berms, the haul road, and the sediment ponds. Those get checked when somebody has time, which in July is never.

```mermaid
flowchart TD
  A["Dock launches the 5:10 a.m. flight"] --> B["Same photo route over 14 piles and the east face"]
  B --> C["Pile volumes converted to tons at your density factor"]
  B --> D["Highwall, berm and bench changes flagged against yesterday"]
  B --> E["Haul road condition and pond levels checked"]
  C --> F["Compared to belt scale tons and shipped tickets"]
  D --> G["Superintendent's 6:40 a.m. review list"]
  E --> G
  F --> G
```

## What changed in 2026: the aircraft stopped needing a person standing next to it

Two things moved this year. Autonomous inspection went from demonstration to production work — Hyundai put AI-guided autonomous drones to work inspecting its Georgia plant and cut inspection time by roughly ninety percent — and docked systems from Skydio and DJI got cheap and reliable enough that a quarry can own one instead of renting a pilot. Humanoid and wheeled robots also started showing up in real facilities rather than trade-show booths, which matters less to a pit today but tells you the direction.

The practical difference for you is repetition. A survey contractor flies four times a year at a price per visit. A dock on your property flies before shift every day at almost no extra cost, from the same positions in the same light, so it compares today's pile to yesterday's pile rather than to a different pilot's flight in different weather. Repetition turns a volume into a trend, and a trend is what catches a belt scale drifting.

Volume software such as Propeller, and the stockpile measurement services built for aggregates producers, does the boring part: it turns overlapping photos into a surface, subtracts the pad, applies your density factor per material, and gives tons by pile with a stated confidence.

## 6:40 a.m. on a Tuesday, with the flight already done

The superintendent opens his phone in the truck. Fourteen piles with tons and the change since yesterday. Washed #57 is down 2,240 tons, which matches the tickets. Manufactured sand is up 900 and the wash plant ran ten hours, so that reconciles. Base is down 1,600 and the tickets say 900 — the one line worth walking over to look at, because either somebody loaded the wrong pile or a material code is wrong on a dozen tickets.

Underneath that, three physical flags: a section of berm on the east bench is lower than it was Friday, new material sits at the toe of the highwall in the northwest corner, and the sediment pond is within a foot of the riser. The first two go on the list for the competent person doing today's workplace examination; the third goes to whoever holds your stormwater permit before the next rain, because a pond about to go over is a discharge you will be explaining later.

By the time the crusher starts, the superintendent has already decided what to run today based on tons actually on the ground, not on what the office thinks is there.

## The arithmetic: it is not the six hours, it is the belt scale

Suppose a pit ships 900,000 tons a year, carries about 140,000 tons on the ground at a production cost of $7.40 a ton, and counts by hand once a month. Illustrative numbers — use yours.

| **Line** | **Amount** |
| --- | --- |
| Hand count: 6 hours, superintendent plus loader operator, 12 times a year | 72 hours at about $58 loaded = $4,176 |
| Outside drone survey, four visits | about $5,600 |
| Owned dock, aircraft and volume software, per year | about $13,800 |
| Net additional cost | about $4,000 |
| Inventory carried on the ground | 140,000 tons = $1,036,000 |
| Typical month-end variance at 2.8% | 3,920 tons = $29,008 adjusted |
| Belt scale reading 1.5% high on 900,000 tons | 13,500 phantom tons = $99,900 of misstated cost |

The four thousand dollars is not the point. The last two rows are: you cannot catch a belt scale a percent and a half off by counting quarterly, and you can catch it in about six weeks by counting daily. If book quantity drifts up against measured tons in a straight line, that is not shrink — that is a scale that needs a certified test.

## What the drone cannot sign its name to

It does not do your workplace examination. Under the surface metal and nonmetal rules, a competent person examines each working place at least once a shift and records the location, the condition, and the date of correction. A photograph of a berm is a good prompt for that examination. It is not the examination, it is not the record, and an inspector will not accept it as one. The flight tells your examiner where to look first, nothing more.

It also does not fix density. Tons come out of cubic yards through a factor that changes with gradation, moisture, and how the pile was built, and if your factor is wrong every report is confidently wrong by the same percentage. Pull a truck-scale check against a measured pile a couple of times a year and correct the factor.

Then there is flying itself. A remote pilot in command is still required under the FAA's small-aircraft rules, flights beyond visual line of sight still need a waiver, and wind, fog, snow cover and standing water all degrade what the camera can measure. Piles under a conveyor or against a wall get measured badly. And no drone will tell you whether that rock at the toe came down last night or three weeks ago — a person who knows the face does that.

## Frequently asked questions

### Can I use these numbers for my financial statements?

Ask your accountant, but the direction of travel is clear: measured volumes with ground control and a stated accuracy are more defensible than a superintendent pacing a toe line, and auditors have accepted drone-measured inventory observations for years. What they want is a consistent method and a documented density factor.

### Do I need a full-time pilot on staff?

No, but you need someone on the payroll with a remote pilot certificate to be responsible for the flights, and it is usually the surveyor or a plant engineer who already wanted it. The dock does the flying; the certificate covers the responsibility.

### What about the piles at our satellite yards?

One dock covers one site. For sales yards forty miles out, the cheaper answer is a person with a hand-launched aircraft doing a monthly loop, using the same software so the numbers are comparable.

### Will it see a crack in the highwall?

It will see change — new material at the toe, a crack opening at the crest, a bench that has slumped — because it compares the same view day over day. Interpreting that change is engineering work. If you get a flag on a face, you call the engineer, not the software vendor.

## Where to start

Before you buy anything, fly the yard twice in one week with the survey contractor you already use and compare both numbers to your book quantity. If the flights agree with each other and disagree with your books by more than a percent or two, you have found the business case for about the cost of one repair on the secondary cone.

One footnote on the phone: better inventory numbers change what your salespeople can promise, and promises drive calls. When a spec runs short, the pit line lights up. [CallSphere](https://callsphere.ai) builds the voice and chat agents that answer those calls around the clock, tell a contractor honestly what is on the ground, and book the pickup or the callback. It measures nothing in the yard — that is the drone's job — but it makes sure nobody is sitting on hold while your superintendent is out looking at a pile.

---

Source: https://callsphere.ai/blog/fourteen-stockpiles-and-a-six-hour-walk-the-month-end-tonnage-variance