Heap-leach intelligence

One pad. Two futures.

Same ore. Same pad. Same ninety days. The only difference is whether you can see inside the heap — and whether an AI that learns your pad is watching it with you.

oz on pad 41,220 PLS 0.31 g/t cells on curve 14/16
Scroll

The two futures

The same quarter, told twice.

2.5 million tonnes stacked, one 90-day leach cycle. In one future the pad runs on spreadsheets. In the other, LixiTrack is watching every parcel. Toggle between them.

0 oz off the books this quarter
  1. Day 12

    PLS grade sags. The pad can't say why.

    The pregnant solution blends every fast and slow path in the heap, so the assay hides the cause. The ore gets blamed; irrigation carries on watering the problem.

    −620 oz drifting off forecast

    The heap answers in one screen.

    CELL C-07 · PLS −28% vs forecast → parcels L2-P14…P17 under-percolating. Rest irrigation area IA-3, rip and re-drip 1.2 ha.

    Those parcels were stacked in the wet week and surface-compacted. Nine days after the fix, they're back on curve.

    +540 oz back on forecast
  2. Day 34

    Haul trucks shortcut across the fresh lift.

    Nobody connects today's route to next year's recovery. Compacted agglomerate leaches in years, not weeks — the write-off lands long after the tyre tracks fade.

    −480 oz of future recovery destroyed

    The stacking plan knows what's underneath.

    STACKING PLAN · Active parcels beneath proposed route. Divert to the L3 east ramp — zero traffic on wetted ore.

    Every metre of pad knows what it's carrying.

    +480 oz protected
  3. Day 61

    Mine and plant disagree by 900 oz.

    Month-end reconciliation becomes an argument between spreadsheets. The gap gets "plugged" and rolls forward, waiting for the year-end audit to find it.

    −900 oz plugged, unexplained

    The twin engines catch the drift early.

    TWIN ENGINES · Kinetic vs AI forecast diverging 4% · drift localized to Cell B-03.

    The gap resolves into 120 oz of real solution hold-up and one slow parcel cluster — adjusted parcel-by-parcel, on the record.

    +780 oz explained and booked
  4. Day 90

    Quarter-end: "the ounces are on the pad… somewhere."

    The auditor can't sign geography they can't see. Conservatism does the accounting, and the provision comes out of your ounces-on-pad.

    −400 oz written down

    Inventory that survives an audit.

    STACKING LEDGER · Every stacked tonne addressed. Ounces-on-pad by parcel, each with its own curve, ready to sign.

    A number defended by geometry, not by confidence.

    +100 oz — and a number you can defend

Illustrative cycle — 2.5 Mt at 0.50 g/t on an oxide heap. Your numbers will differ. The physics won't.

Platform

One heap. One truth.

Spatial truth and financial truth, fused. Everything feeds the ledger; the ledger feeds the forecast — and the plan.

3D Heap Builder

Build the pad as it really is.

Survey-driven lifts stacked in true benched geometry — flat tops, terrain-conformed bases. X-ray any lift, project any parcel, from the views your operators actually use.

Stacking Ledger

Every parcel carries its own books.

Tonnes, grade, stacking date, contained and recoverable ounces — tracked from truck to ledger at parcel level. Defendable metal accounting, not tribal knowledge.

Cells & Variance

Give under-performance an address.

Fixed drainage cells tie solution chemistry back to collection points. When PLS grade sags, you know which part of the heap to interrogate — not just that "recovery is down."

Twin Forecast Engines

Kinetics + AI, reconciled.

Every parcel is forecast twice — deterministic leach-kinetics curves, and an AI model trained on your pad's actuals — then reconciled against what the plant actually pours.

Production Planning

Plan the pad before the trucks roll.

Stacking sequences, ramp routes and irrigation schedules planned on the real geometry. Every planned parcel carries its forecast before the first truck dumps.

Scenario Forecasts

Ask what-if before the pad does.

Branch the plan: change the blend, the stacking order or the irrigation strategy, and compare recovered ounces side by side — before committing tonnes.

Multi-Site Forecast

Every pad, one portfolio.

Forecast every operation from one place — recoverable ounces per site, rolled up to a portfolio view for corporate technical services and the board.

Reporting

Month-end, ready to sign.

The Monthly Reconciliation Pack and more — parcel ledgers, recovery curves and variance, exportable to PDF and Excel for month-end and audit.

Data Governance

Per-user roles and audit trail.

Everyone works from the same heap, but sees exactly what their role allows. Per-user permissions and a full audit trail — every change logged, on your servers.

The living pad

Irrigation areas rotate. The heap remembers every wetting, every rest, every hour of application.

Solution percolates down each zone's own column — and reports to the PLS pond with its paper trail.

When the kinetic and AI forecasts drift apart, the gap is traced to a cell before month-end.

How it works

The spine: Pad → Lift → Parcel

Every tonne gets a permanent spatial identity the day it's stacked. Leach stages aren't declared by operators — they're derived from position in the stack and the solution path.

Engineering cross-section of a LixiTrack heap Three 5-metre lifts at level elevations on a graded liner. Gold solution percolates from irrigation area IA-3 down through the stack, along the liner and out to the PLS pond. L3 is in active primary leach, L2 in secondary leach, L1 draining. Cell C-07 drains to a collection point. EL +15.0 m EL +10.0 m EL +5.0 m 5.0 m lift L1 · DRAINING L2 · SECONDARY LEACH L3 · ACTIVE · PRIMARY LEACH IA-3 · on PARCEL L3-P07 CELL C-07 → collection point graded liner · ~2% fall → toe ditch · PLS pond
Section A–A′ · 5 m lifts at level elevations · base conforms to the graded liner · stages derived from position + solution path, never declared

Forecast intelligence

Two engines.
One number you can defend.

Every parcel is forecast twice — once by metallurgy, once by machine learning — and both are reconciled daily against what the plant actually pours.

Deterministic engine

Leach-kinetics recovery curves built from your column tests, stacking plan and irrigation schedule. Transparent, auditable, tied out to the ounce — the number your auditor signs.

Column-test scale-up Flux-based curves Fully auditable

AI engine

A model that learns each pad's real behaviour from flux, grade and solution-chemistry actuals — and catches the channelling, compaction and drift that first-principles models can't see. Every forecast ships with confidence bands.

Learns from actuals Confidence bands Drift alerts

Anatomy of a forecast

The kinetic curve is the baseline: recovery against cumulative leach ratio — flux-based, because irrigation schedules change and calendar days lie. The AI learns the residual: where your pad actually runs above or below its metallurgy, and why.

Kinetic recovery curve with AI correction Recovery versus cumulative leach ratio. The orange kinetic baseline plateaus near 76 percent. Plant actuals pull the cyan AI-corrected forecast slightly below it after drift is detected at Cell B-03, with a widening confidence band. 0% 50% 100% 0 1.5 3.0 CUMULATIVE LEACH RATIO · m³/t RECOVERY % drift detected → CELL B-03 KINETIC BASELINE · column tests, scaled AI CORRECTION · learned from actuals · ± band plant actuals
Flux-based, not calendar-based — cumulative leach ratio is the most predictive single variable a heap has.
Kinetic inputs
ore type & P80 column recovery curves lift height & stack date irrigation schedule → cumulative flux
The AI corrects with
daily flux actuals PLS grade & chemistry cell drainage variance parcel age & burial depth

When the two engines disagree, the gap itself is information — LixiTrack traces it back to the parcels that caused it.

The heap is the ledger.

Why it holds up

100% of tonnes traced to a named parcel
±4% kinetic vs AI agreement, reconciled daily
30-day rolling recoverable-ounce horizon

When the auditor asks where the ounces are, you don't open five spreadsheets. You open the heap.

Built on an operating pad, not in a lab.

LixiTrack comes from the day shift — born on a working heap-leach operation where the ounces had to be found in a spreadsheet every month-end. It's built for the people who sign that number.

Your spreadsheet stays

Import the stacking sheet you already keep, as-is. LixiTrack reads it — you don't rebuild it.

One version of the story

Built for operators, metallurgists, managers and executives alike — everyone reads the same heap, the same ledger, the same number.

It never silently guesses

Every gap and assumption is visible and labelled. Audit-grade means showing your work.

Your data stays on your servers

LixiTrack runs on your infrastructure. Every record remains inside your network — no cloud requirement, no third parties.

Start anywhere, in 30 minutes

Begin at any stage of the heap's life, with whatever history you have — day one or year ten. Initial implementation takes about 30 minutes.

Bring your pad
into focus.

LixiTrack is in private pilot with operating heap-leach sites. Implementation takes about 30 minutes — start at any stage, with any history, and all data stays on your servers.

We reply within two working days. Your data never leaves your servers.