Stockpile ERP Reconciliation: Closing the Gap Between Measured Volume and Book Inventory

For operations that store coal, ore, aggregates, biomass, or clinker in open yards and bulk piles, the number in the ERP and the material actually on the ground are rarely the same. Stockpile ERP reconciliation is the structured process of comparing book inventory against an independent physical measurement, locating where the two diverge, and adjusting records so downstream planning works from reality rather than assumption. When the measurement source is a 3D LiDAR stockpile inventory system producing volume-derived tonnage on a defined cadence, reconciliation shifts from an occasional manual survey to a repeatable, auditable control. This article looks at where the gap comes from, how to run periodic and continuous reconciliation, how to wire measurement data into the ERP, and what the practice changes for dispatch and stock-take audits.

Where the inventory gap comes from

The difference between book and physical stock is not a single error; it accumulates from several structural sources.

The first is the input and output records themselves. Belt scales, truck weighbridges, and bucket counts each carry tolerances, and those tolerances stack over thousands of movements between counts. A scale reading slightly high on receipts and slightly low on dispatch produces a one-directional drift that book records cannot self-detect.

The second is unmeasured loss and gain. Wind erosion, moisture uptake, compaction, dusting, oxidation, and material left in dead corners all change the physical mass without generating a transaction. None of these appear in the ERP because no document was raised.

The third is timing and assignment. Material booked to the wrong pile, deliveries recorded in the wrong shift, or transfers logged after the cut-off all create apparent gaps that are really classification problems. These are common in multi-pile yards where one ERP material code maps to several physical heaps.

The combined effect is what operators informally call the inventory black hole: a balance that looks precise on paper but cannot be defended against a physical count. The point of reconciliation is not to assign blame for the gap but to size it, characterize it, and decide which portion is real loss versus record error.

What an independent measurement contributes

Reconciliation needs a reference that does not depend on the same transaction stream it is checking. A 3D LiDAR system supplies that reference by scanning the pile surface as a dense point cloud, computing volume against a known base plane or reference surface, and converting volume to tonnage using a configured density. Volumetric accuracy is typically in the range of 1 to 3 percent after calibration, and the achievable figure is model-dependent and influenced by pile geometry, surface conditions, and how well the assumed density matches the stored material.

Two characteristics make the measurement useful for reconciliation specifically. It is independent of the weighbridge and belt-scale chain, so it cross-checks rather than echoes those instruments. And it is repeatable on a schedule, so each reconciliation cycle compares like with like instead of relying on a one-off survey commissioned only when a discrepancy is already suspected.

A single system can cover up to 32 zones, which lets one ERP material code be reconciled at the level of individual physical piles rather than as one aggregate balance. That zone-level resolution is often where misassignment errors finally become visible.

Density is the honest caveat. Volume is measured directly; tonnage is inferred. Where moisture or compaction shifts density between cycles, the reported tonnage carries that uncertainty, and reconciliation should treat tonnage as a measured-volume figure times an assumed density rather than as a direct weight.

Periodic versus continuous reconciliation

There are two operating modes, and most sites use both for different purposes.

Periodic reconciliation runs on a fixed calendar: month-end, quarter-end, or before a financial stock-take. The LiDAR-derived tonnage at the cut-off moment becomes the physical figure, the ERP balance becomes the book figure, and the difference is investigated and posted as an adjustment. Because the measurement is captured the same way every cycle, the adjustment trend itself becomes a signal: a stable small variance suggests sound records, while a growing or oscillating variance points to a systematic instrument or process fault worth fixing at source.

Continuous reconciliation compares measured stock against book stock on a rolling basis, using update frequencies that can be configured daily, per shift, or in real time depending on the deployment. Instead of waiting for month-end, the system flags when measured and booked volumes diverge beyond a set threshold, so a miscoded delivery or a drifting belt scale surfaces within a shift rather than weeks later. This narrows the window in which decisions are made on bad numbers.

The two modes are complementary. Continuous monitoring catches drift early and keeps the gap small; periodic reconciliation formalizes the adjustment, creates the audit record, and validates that the continuous thresholds are set sensibly. Neither replaces the other.

Connecting measurement data to the ERP

Reconciliation only scales when the measured figure reaches the ERP without manual re-keying. The system exposes data through OPC UA, a REST API, MQTT, CSV export, and direct database export, which covers the common integration patterns found in industrial and enterprise environments.

For ERP and business-system integration, the REST API and database export are usually the natural fit. A scheduled job can pull current measured tonnage per zone and write it into a reconciliation table or a custom inventory-comparison object, where the ERP’s own logic compares it against book stock and raises a variance line. CSV export covers environments where a lightweight, file-based handoff to an existing import routine is preferred over a live connection.

For the operational and control layer, OPC UA and MQTT are the appropriate channels. OPC UA suits integration with SCADA, MES, and dispatch systems that already speak it, exposing measured volume and tonnage as structured, typed nodes. MQTT suits event-style distribution, where each new measurement is published to a topic that historians, dashboards, or a middleware layer subscribe to before forwarding curated values upstream.

A common and robust pattern is to route measurement data through MES or a middleware layer rather than wiring the sensor straight into the ERP. The operational layer consumes high-frequency OPC UA or MQTT data, applies validation and density assumptions, and then presents the ERP with a clean, periodic reconciliation figure over REST or database export. This keeps the cadence and data quality the ERP expects while preserving the granular feed for operations.

Whatever the channel, the integration should carry the measurement context alongside the number: timestamp, zone identifier, measured volume, the density assumption applied, and the resulting tonnage. Reconciliation that records only a final tonnage loses the ability to explain a variance later.

Value for dispatch and stock-take audits

For dispatch and scheduling, a reconciled stock figure changes what the planning system can rely on. Instead of dispatching against a book balance that may already be wrong, schedulers work from a measurement-backed figure refreshed on a known cadence. That reduces two failure modes: committing loads against material that is not physically there, and holding back dispatch because the recorded balance is conservatively understated. Zone-level data also lets dispatch target the correct physical pile when one material code spans several heaps.

For stock-take and financial audit, the practice produces something auditors value: an independent, timestamped, methodologically consistent measurement trail. Each periodic cycle yields a defensible physical figure, the book figure it was compared against, the variance, and the integration record showing how the number was derived and transmitted. The volume-to-tonnage method, including the density assumption, is explicit rather than buried in a manual survey note. This does not replace the auditor’s judgment, but it replaces an ad hoc annual survey with a standing, repeatable control whose history can be reviewed.

The structural benefit across both is the same: decisions move from a number that is assumed correct to a number that is checked. The gap does not vanish, but it becomes visible, bounded, and explainable.

Putting reconciliation into practice

A workable rollout starts narrow and proves the loop before scaling. Begin with one yard or a defined set of zones, establish the calibrated baseline, and run continuous monitoring alongside the existing periodic count for a few cycles to confirm that measured and adjusted-book figures track each other within expectation. Set variance thresholds from that observed behavior rather than from a guess, and document the density assumptions so every later tonnage figure can be traced back to its basis.

Once the comparison is trusted, formalize the integration: a defined channel into the ERP, a fixed cadence, and a reconciliation record that captures volume, density, tonnage, zone, and timestamp on every cycle. From there the same pattern extends to additional yards and material codes without rebuilding the logic each time.

To discuss how 3D LiDAR measurement could fit your existing ERP and dispatch stack, request an application review with the relevant material types, pile layout, and integration targets. For related material on volume measurement, multi-zone yard monitoring, and industrial data integration, see our technical insights library.