The standard it is measured against
To judge whether a number is right, you need to know which framework it is measured against. None of these is a convention of our own.
See the sourcesF1, F2 and F3 factors, with their definition declared
Reconciliation factors are widely used across the industry, but their definitions vary between companies. Argus Mine declares its own — and they are not an analogy: they are database columns.
| Factor | Calculation | What it measures | Typical owner |
|---|---|---|---|
| F1 Mine vs. plan | mined / planned | Operational compliance: was the design mined, where it was designed? | Operations · Mine |
| F2 Plant vs. mine | plant / mined | Inventory control: did what left the mine reach the plant? Losses in stockpiles, haulage or misdirection. | Mine Control · Logistics |
| F3 Plant vs. plan | plant / planned | Block model reliability: does the resource estimate predict what actually reaches the plant? | Geology · Resources |
When comparing with the literature: in the classic formulation (Parker, 2012), F1 compares the grade-control model with the reserve model. Argus Mine's F1 measures spatial compliance with the design — what the literature calls mine-to-plan compliance (Otto and Musingwini, 2019) — and leaves block model reliability to F3. See the sources
3D partial block: the method that gets audited
Mining packages offer two conventions for intersecting a block model with a surface. Argus Mine uses the one that goes into an auditable reserves report.
Whole block (centroid)
The block is either fully in or fully out, depending on where its center falls. Fast, and valid for long-term planning where sub-block resolution means nothing.
3D partial block · Argus Mine
The fraction of the block's volume that falls inside is calculated, prorated in plan and in elevation. It is what is used when the number goes into an auditable report.
Why it matters: measured, not assumed
In a phase with a small footprint (about 3,000 m²) against 25×25×5 m blocks, the centroid test returned exactly the same 150 blocks in the first and last period, even though the design grew 15% in area: 11 of 12 periods reported 0 t. With partial blocks, the same phase detects 20,449 t of actual progress.
Our own measurement on a demonstration dataset, not on client data.
Reconciling the reconciliation
The principle of independent measurements also applies to the software that computes the reconciliation: the costliest defects are not found by reading code, but by recomputing the answer outside the engine.
Independent volumetrics
Recompute the period's cumulative total directly against the design geometry, bypassing the engine's cache. Catches order-of-magnitude divergences: double counting or a delta that swallowed the whole phase.
Volume from surfaces
Pure prismatic method over the TIN mesh, without touching the block model. The mean elevation per period proves whether the pit is actually getting deeper.
Telescoping identity
Results are stored as deltas per period: the sum of deltas up to period N must equal the actual cumulative total at N. If it doesn't close, something is double counted.
Measurability, before selling
Count how many blocks the design contributes in the first and last period of a phase. If the number doesn't change, progress falls below the model's resolution and the factors would read 1.000 for lack of measurement, not compliance.
All four checks run in SQL against the database, without touching the application. A Competent Person can run them on their own.
Never fabricate the missing assumption
A system that invents a number to avoid leaving a field empty gets caught only once — and never recovers. These are product rules, not implementation accidents.
No economic assumption, no NPV
If the discount rate or valuation date is missing, the NPV fields stay empty and flagged. Never zero.
Empty is not zero in measured grade
An element that was not assayed stays blank. An explicit 0.0 would be treated as a real measurement and wipe out the block's grade.
“Not verified” is not “verified and correct”
If there are no QAQC samples, or the real or nominal price basis is not declared, the result is flagged as not verifiable.
“Unknown” and “multiple” are valid causes
Assigning a cause the data doesn't support sends the wrong person to investigate and burns credibility faster than an imprecise number.
Directional severity
A favorable deviation calls no one: 0 mÂł of overbreak is the best possible result, not a critical alert.
The economic sign is never forced
Impact is derived from each block's actual net NSR. An unmined block of ore with negative NSR avoids a loss.
Argus Mine decides nothing. It produces evidence someone has to sign.
The argument every mine settles by hierarchy — is it an operations problem or a geological model problem? — becomes a number that can be arbitrated. Attribution is not cosmetic: it is the product.
| Who | What they sign or decide | Which signal calls them |
|---|---|---|
| Resource geologist | The block model and its estimation | Degraded F3 |
| Competent Person / QP | The resources and reserves statement | Sustained F3 by zone |
| Mine planning | The design and schedule of each phase | F1, schedule variance |
| Survey | The as-built survey and its moisture | Cause: survey control |
| Grade control | Blast-hole sampling | Cause: sample preparation |
| Laboratory | Assayed grade and its quality control | Cause: laboratory assay |
| Drill and blast | Overbreak and underbreak | Overbreak / underbreak |
| Mine operations | Mining what was designed, where it was designed | F1, misdirection |
| Metallurgist / plant | Independent feed measurement | F2 not interpretable |
| Geotechnical engineer | Slope stability and deformation | InSAR / ICESat-2 |
| Management control | Discount rate and real or nominal basis | NPV, price basis |
| Management and board | What gets prioritized and what is accepted | Materiality, value at risk |
We don't replace your software. We connect it.
Argus Mine acts as a horizontal integrator across the value chain: it consumes what your operation already produces and reconciles the result.
Learning is private to each site
The learned normal range and the history the system remembers are never mixed across sites or across clients.
History recall
Every reconciliation is indexed: the system finds periods with the same deviation pattern even if they happened long ago, and answers whether the previous mitigation worked.
Multi-site, multi-projection
Each site runs in its own UTM zone and satellite layers in WGS84, in a single corporate dashboard.
Every standard and study cited on this page has its full reference in Sources.
How much value is your mine losing today?
Every day a deviation goes undetected, unexplained or unresolved, the value at risk keeps growing. Start by seeing it.