Volume Methodology

How a scan becomes a volume you can report, and what moves that number.

What gets measured

The platform measures the volume of material sitting above the zone’s floor, within the footprint you drew. From that volume it derives tonnage and fill level, if the zone is set up for them.

Volume is the material above the floor line, within the footprint. Fixed structures inside the footprint are carved out of it.

Anatomy of a measurement

Every scan flows through the same fixed pipeline. Walk through it below — each stage shows the same pile at that point in the measurement.

Loading the interactive measurement walk-through…

The point cloud

A scan produces a point cloud: hundreds of thousands of individual 3D points, each one a spot the sensor measured on a real surface — the top of the pile, the bare floor, a wall. Which way is “up” is recorded at capture and travels with the scan, so ground-level and overhead sensors both produce a correctly oriented measurement with no manual step.

The zone footprint

Only points inside the footprint you drew are considered. Points outside it — a neighboring pile, a passing truck, the far wall — are set aside before any volume is computed. Draw it slightly wider than the material: the ring of bare floor around the pile is used as evidence in the next step.

The floor baseline

To measure how tall the material is, the platform first needs to know where the floor is. This is the single largest factor in accuracy: a floor set too high or too low shifts every reading by that same error, everywhere in the zone. There are two ways to establish it, labeled Ground Detection on the zone form:

• Dynamic — the zone you drew is wider than the material, so there is bare floor inside it. The platform models the floor from that visible ground, at every measurement.
• Empty scan — the exact contour of the floor and any surrounding structures, captured once with the material cleared out.

Floor Modes & Empty ScansWhat each option does, how dynamic detection weighs candidate floors, and when to capture an empty scan.

From material to volume

With the floor established, the platform works out which returns inside the zone are genuinely material and which are noise. The detected material is then turned into a continuous surface, with gaps filled where the sensor could not reach, and the volume is the space between the floor and that surface.

The surface is built on a grid whose cell size is chosen dynamically per zone — there is nothing to set.

Structures and walls

Fixed structures inside the footprint — support columns, poles, dividers, wall returns — can be subtracted out, so they are not counted as material. The carve-out is by area, not height, so a low curb removes as much footprint as a full-height pillar.

Better still, capture an empty scan: fixed structures inside the zone are then in the baseline and in every live scan, so they cancel out with nothing to maintain.

The roof, or maximum height

Indoors, a scan can catch the roof, rafters, or the tops of walls, and those returns would otherwise read as very tall material. The zone’s maximum height rules them out, so set it below the ceiling wherever you can. An empty scan also records the real headroom. Ceilings the platform infers on its own are rejected — a declared or captured height is what it works from.

When the number moves

The published volume holds while the pile is at rest and moves when the pile actually changes. It is not a rolling average creeping toward a new value, so a flat line is the expected shape for an untouched pile.

Obstruction and occlusion

Two different things get between the sensor and a clean measurement, and the platform handles both.

An obstruction is inside the zone: a loader working the face, a truck backing in, a person walking through. None of it is inventory, but all of it returns points inside the footprint, and a bucket over the pile would measure as a very tall column of material. Obstruction-aware volume is on by default. It holds the last volume while something that is not material is in the zone, and shows Obstructed while it is there and Stabilizing… while the view settles afterwards. The spike never reaches the chart.

It also captures evidence of what happened in the zone — a delivery, a person, a piece of equipment — which you can see in the cloud portal and keep as a record of activity on the zone.

An occlusion is between the sensor and the zone: something parked in the way, a structure, a sheet of material. If it blocks the sensor so the zone stops receiving returns, that is detected and flagged too.

Turning obstruction handling off is rarely right. A zone with it off publishes whatever passed through the view as though it were product.

Tonnage, fill level, and trend

Volume is the raw measurement. The figures you report are derived from it, and each needs one piece of setup.

• Tonnage = volume × density, so density is what drives it. We can help: we hold density references for common materials, and where a scale is available we can combine it with the system to work out an average density for your pile. Talk to your provider.
• Fill level = volume ÷ capacity, as a percentage. An empty scan gives us the bay’s real dimensions, and we will work with you to define what “full” means for that zone, so percentages and alerts mean something.
• Trend and projections compare volume over time. Once there is enough consumption to observe, that becomes a rate and a days-remaining figure.

Moisture content and void fractions can be configured as transformations — dry tonnage, for example — during commissioning. Talk to your provider about the values for your material.

Volume is reported in whichever unit you choose for the zone — cubic meters, cubic feet, tons, or bushels — from the same underlying measurement.

What affects accuracy

• A correct floor — the biggest single lever by a wide margin. A floor off by a few centimeters shifts every reading in the zone by that amount and still looks plausible. An empty scan removes the question.
• Sensor coverage — how much of the zone the sensors actually reach. Work with your provider on sensor count and placement to get good coverage.
• A well-drawn zone. Drawn too tight it clips real material; drawn with no surrounding floor, dynamic detection has less bare ground to model from. Slightly wider than the material is the sweet spot.
• Occlusion. Anything between the sensor and the pile hides surface.
• Correct product and capacity settings. They don’t change the measured volume, but they drive tonnage and fill level.

Material reflectivity — very dark, wet, or dusty surfaces — is a low risk in practice; it returns fewer points but rarely changes the number.

Internally, each scan is scored against previous baselines, and that score decides whether the reading is promoted. A scan that is clearly off is thrown out rather than published.

Floor Modes & Empty ScansThe two floor options and how to choose one.

TroubleshootingA zone that stops updating, reads zero, or looks stuck.