Your circuit, connected to your plant data, balanced automatically — then kept true, stress-tested, reconciled and signed off.
Data gathered by hand, every period. One typo or broken link and the month is wrong.
Numbers get nudged until it balances. Nobody can defend how.
Metal moves into tanks and stockpiles. Recovery swings for no real reason.
The flowsheet you draw is the simulator, the control model and the accounting engine. Nothing to keep in sync. The rest of this story follows these five steps:
Real unit ops, your minerals and elements, your units. Solids, water and every metal balanced together — through every recycle.
No copy-paste. Tonnages, assays and levels come straight from your historian and lab — for whichever shift, day or month you pick. Every number in the model knows where it came from:
Simulation, calibration and accounting all see the same window.
The circuit re-solves on streaming historian data — a real-time mass balance.
Predict the unmeasured, feed it straight into the balance.
Every stream, every recycle, in the blink of an eye — fast enough to run hundreds of times, which is what makes everything that follows possible.
Circulating loads close. Impossible setups get a clear error pointing at the unit — not a wrong answer.
Too much or too little information on a unit is flagged before you solve.
What balanced, what was adjusted, what to check — visible after every run.
Recovery isn't a fixed number — it depends on residence time. Cells follow Kelsall fast/slow kinetics at the time the slurry actually spends in them, so slowing the bank down or pulling harder does what it does on the plant.
Type the constants, or fit them to lab timed-recovery tests — with live curve previews.
A scale-up factor bridges lab kinetics to plant reality.
Tests at different head grades blend automatically to the grade the bank actually sees.
Cells share a design; the bank accumulates residence time. Change the design once — the whole bank moves.
Today's feed grade and tonnage drive the model automatically — it always starts from what the plant is actually doing.
"Hold conc grade at 24.5%." Pick the knob — the model finds the setting that achieves it, through the whole circuit.
Pin a stream's tonnage, %solids or volume — pump and pipe constraints the balance must respect.
A target the circuit can't physically meet is reported, never fudged.
Plants drift — recoveries change, kinetics slow. Each solve nudges your chosen parameters to match what the plant actually produced, so last month's model is still right this month.
Finds the smallest change that matches measured reality — and shows before → after for every parameter.
Guess. If your data can't tell two parameters apart, it says so instead of inventing an answer.
Take a real shift. Change one thing — feed grade, tonnage, mass pull. See what the plant would have done, side-by-side with what it actually did.
| Final concentrate | Base | Scenario | Δ | Δ% |
|---|---|---|---|---|
| Solids (t/h) | 15.6 | 18.4 | +2.8 | +17.9% |
| Cu grade (%) | 24.6 | 24.9 | +0.3 | +1.2% |
| Cu metal (t/h) | 3.84 | 4.58 | +0.74 | +19.3% |
In, out, what stayed behind in tanks and stockpiles — and the gap between them, reported honestly and graded A/B/C, period after period.
| Stream | Solids (t) | Cu grade (%) | Cu metal (t) |
|---|---|---|---|
| Plant feed | 182,280 | 2.24 | 4,083.1 |
| Final concentrate | 3,772 | 24.61 | 928.3 |
| Final tails | 178,508 | 0.19 | 339.2 |
| ΔInventory (tanks, thickeners, stockpile) | +1,890 | — | +47.9 |
| Unaccounted (in − out − ΔInv) | — | — | −32.3 (0.8%) |
Tell it how much you trust each instrument. It makes the smallest adjustments that balance the books — and shows every single one.
| Measured input | σ (trust) | Measured | Reconciled | Move |
|---|---|---|---|---|
| Feed weightometer (t/h) | ± 2.0% | 252.5 | 249.7 | −1.1% (0.6σ) |
| Feed Cu assay (%) | ± 3.0% | 2.24 | 2.27 | +1.3% (0.4σ) |
| Conc Cu assay (%) | ± 1.5% | 24.61 | 24.55 | −0.2% (0.2σ) |
| Tails Cu assay (%) | ± 5.0% | 0.190 | 0.196 | +3.2% (0.6σ) |
| CCD inventory holdup (t) | ± 8.0% | 1,890 | 1,962 | +3.8% (0.5σ) |
Every run keeps its period, its inputs and the exact flowsheet it ran on. Nothing is overwritten, ever.
P754 — the international Code of Practice for Metal Accounting — is the standard auditors, boards and due-diligence teams ask about. Metaltech follows its principles by design, not by promise.
The statement comes from a full mass balance of the circuit — not from recovery factors.
Every instrument carries a stated uncertainty, and every adjustment is judged against it.
Every number traces back to its source, its run and the exact flowsheet version.
Gaps are reported and rated, period after period — never quietly absorbed.
The P754 discipline, built into the tool — not bolted onto a spreadsheet.
| Excel today | Metaltech | |
|---|---|---|
| The balance | A web of linked sheets one person understands | Your real circuit — recycles, water, kinetics |
| Plant data | Copy-pasted monthly | Straight from historian and lab — live |
| Recovery factors | Fixed assumptions, set years ago | Re-calibrated to the plant every period |
| Asking "what if?" | Gut feel | Real period data, answered in seconds |
| Closing the balance | Numbers nudged by hand | σ-weighted, every move shown |
| Auditability | One file, overwritten | Versioned runs, draft → approved |
| AMIRA P754 | Hard to evidence | The workflow follows it by design |
| Effort per period | Days | Minutes |