Platform Features

Five tabs.
Exactly as built.

Every demo window below uses the real platform design — same fonts, same colours, same components. What you see is what you get.

Chain Builder

Build your chain visually. No code.

A node-based graph editor. Add nodes from the category palette, connect them in the right order, set parameters per node. Connection validity is enforced against stream type rules — concentrate can't feed an electrorefinery without a smelter in between.

  • 20+ node types — Mining, Handling, Comminution, Separation, Downstream, Sale — each with category colour coding
  • Parameter panel — click any node to set nominal value and probability distribution (normal, lognormal, beta, triangular)
  • Performance alerts — define threshold triggers per metric; activation rates surface in simulation results
  • Bottleneck colouring — after a run, nodes turn red/amber by bottleneck confidence directly in the canvas
  • 5 starter templates — Simple Open Pit, Underground + Credits, Integrated Polymetallic, Gold CIL, Multi-Site Network
Chain Builder Underground + Flotation Chain
Dashboard
Chain Builder
Simulation
Optimiser
Actuals
Underground Mine
Mine Site A
1.2% Cu
208 tph nominal
Material Handling
Surface MHS
94% avail.
500 tph system cap
SAG Mill — Bottleneck
Primary Grinding
7,840 tph P50
58% of sim iterations
Flotation Circuit
Primary Flotation
87.3% recovery
28% conc. grade
Concentrate Sale
Revenue
$4.7M / day
$9,500/t · 96.65% payable
Flotation Circuit
normal
fixed
Performance Alert
recovery < 84% → warn
Stream Validation
Connection rules enforced
The backend node catalog defines valid input/output stream types for every node. Connections are validated server-side — the UI reflects errors before you run, preventing misconfigured chains.
Scenarios
Named chain configurations
Save chains as named scenarios — "Baseline FY26", "High Blend", "Optimised". Load any saved scenario to the canvas. The working canvas tracks unsaved edits and prompts before navigation away.
Templates
5 templates, one click
Simple Open Pit, Underground + Precious Metal Credits, Integrated Polymetallic Complex, Gold CIL, Multi-Site Network. Parameters pre-filled with industry-typical values for immediate simulation.
Monte Carlo Simulation

5,000 iterations. P10 / P50 / P90.

Click Run. ChainVision samples every uncertain parameter simultaneously across thousands of iterations. Results are probability distributions over a 30-day horizon — not point estimates.

  • Kuz-Ram blast model — fragment P80 from blast geometry; propagates to SAG feed size correction
  • Bond work index SAG — throughput = power / SPC; harder or coarser feed increases SPC
  • First-order kinetic flotation — R = R_max × (1 − exp(−k_f × t_r))
  • Sobol sensitivity — ranks which parameters drive the most EBITDA variance
  • Result tabs — Summary · Distributions · Fan Chart · Sensitivity · Waterfall · Logs
Simulation Results Baseline Scenario
Ready
Summary
Fan Chart
Sensitivity
Waterfall
Ready to simulate
P50 Outputs
Throughput P50
—
tpd
Recovery P50
—
%
EBITDA P50
—
USD/day
Bottleneck
—
Fan Chart — EBITDA · 30 days
— P10 — P50 — P90
Sensitivity — EBITDA drivers
Ore Hardness
Head Grade
SAG Avail.
Recovery
Distributions
Throughput · Recovery · EBITDA · Energy
Smoothed probability density curves for every output. P10/P50/P90 lines are the decision numbers. Wide distribution = high operational risk. Narrow = the model is confident in that range.
Waterfall
Revenue minus every cost
Revenue − Mining OPEX − Processing OPEX − Logistics − TC/RC − Royalties = EBITDA. Each component shown at P50. Tells you not just where value ends up but where it's consumed.
AI Insights
Executive summary after every run
After each run, ChainVision generates an executive summary, three ranked recommendations, and risk flags — grounded in your specific results: which node bottlenecked most, what P10 implies, where the highest-value opportunity lies.
Chain Intelligence

Flow diagram + data grid. Linked live.

The read-only management reporting layer. Auto-layout flow diagram on the left. Data grid on the right with P10/P50/P90 colour-coded vs plan. Click any period — both panels update together.

SELECT A PERIOD
Chain Intelligence Baseline · P50 · Jul-26
Jul-26 · Throughput (tph)
UNDERGROUND MINEMine Site A208 tph SAG MILL — BOTTLENECKGrinding Circuit7,840 tph FLOTATION CIRCUITPrimary Circuit87.3% CONCENTRATE SALERevenue$142K/day
Plan References
Budget · Forecast Rev 1 · Rev 2
Compare P50 simulation against any named plan reference. Colour coding: dark green = meeting plan, amber = 2–8% below, red = >8% below. Users define plan references from actuals uploads or manual entry.
Compare Mode
Two scenarios side by side
Toggle Compare Mode to add a second scenario selector. Each period shows two value columns and a delta — green if Scenario B improved vs A, red if worse. The flow diagram shows both values on each node.
Visibility Gate
Only after first completed run
Chain Intelligence only appears in navigation after the site has at least one completed simulation run. The sidebar link checks on every page navigation — no configuration needed.
Optimiser

SLSQP finds the optimal blend.

Two modes: Blend (ore source fractions) and Reagents (collector dosage). The solver runs against the full stochastic chain — not a simplified proxy — with 3 restart points to avoid local optima.

  • 5 objectives — Maximise EBITDA · Throughput · Recovery · Minimise Cost · Maximise CVaR
  • Constraints — max BWI (hardness cap to protect SAG), min head grade, supply availability caps
  • Confidence range — improvement shown as P10/P50/P90, not a single number
  • Reagent optimiser — collector dosage vs recovery trade-off for flotation
Optimiser
Blend
Reagents
Objective
Maximise EBITDA
Maximise Throughput
Maximise Recovery
Minimise Cost (OPEX)
Maximise CVaR
max BWI ≤ 14.8 kWh/t ✓
supply cap: 1,800 tpd ✓
Optimal Blend Fractions
Primary Sulphide (0.72% Cu)50%
Lower Grade (0.45% Cu)50%
EBITDA vs Equal Blend
Baseline P50 $142K/day
Optimised P50 $142K/day
Improvement —
CONFIDENCE RANGE
P10: — · P50: — · P90: —
Actuals & Calibration

Upload. Track. Calibrate.

Drag and drop a CSV of daily production actuals. ChainVision immediately compares each week against the simulation plan — node by node, metric by metric. Then calibrate model parameters to match your real plant data.

  • CSV format — date, node_label, metric, actual_value. Drag and drop with react-dropzone.
  • Variance tracking — actual vs plan, vs 3-month moving average, vs budget. Green/Amber/Red bands.
  • Model calibration — fits baseline_tph, availability, bwi_mean, recovery_baseline against actuals. Returns R² per metric.
  • Auto-tune — CMA-ES / differential evolution minimises weighted error across all metrics and periods
  • Calibrated flag — scenarios built from calibrated models are marked in results for full transparency
Actuals 30-day window
⬆
Drop actuals CSV or click to upload
date, node_label, metric, actual_value
Throughput (tpd) · 8 weeks — Plan: 3,200
Weekly Variance — Actuals vs Plan
Node / MetricPlanActualΔ%
Upload actuals to see variance

See it on your operation.

If you'd like to discuss a calibration against your own operation, get in touch.

Get in touch
This page contains commercially sensitive information about ChainVision and is provided to NDA-signed parties under section 3.7.4 of the Industry Growth Program Information Guide and applicable confidentiality agreements. Please do not redistribute.