Stockpile Blend Planner

Work out the blended grade of a ROM stockpile from its lots — mass-weighted, on a wet or dry basis, with the grade coverage shown next to every result so a partly-assayed pile never reads like a measured one.

100% FreeRuns in your browserWorks offlineNothing storedProfessional

What this does

Enter your lots and it gives you the blended grade of the pile — mass-weighted, on a wet or dry basis, with the coverage shown next to every result so a partly-assayed stockpile never reads like a measured one. It also draws the pile down pro rata for a reclaim, and solves how much of a source you can add before a grade target is breached.

It does not yet read tonnes and grades off a dispatch screen by itself. The screen capture test further down is how that gets built: run it against a real display for an hour and it records exactly what is legible there, which is what a reader has to be written against.

Blend check

Mass-weighted grades over the lots you enter. An unassayed lot adds its tonnes without touching that attribute's average — it lowers the coverage instead, and the coverage is shown next to every grade.

Lots, their tonnages and their assays
LotWet tMoisture %Dry tLi₂O %Fe %Row

Leave an assay blank when there is no result. A blank is not a zero, and a measured 0 should be typed as 0.

Blended result

Total dry tonnes
810
Lots
3

Li₂O %

1.12estimate from 92.6% of the tonnes

750 t of 810 t carry a result. The unassayed tonnes are in the total but not in this average, so the real blend could sit either side of it.

No result: Lot C.

Fe %

0.4296whole blend assayed

Reclaim and targets

Optional. Reclaim draws tonnes off every lot in proportion, which is the honest model for a pile that was never homogeneous — so the modelled grades do not move, only the tonnes.

Enter the source tonnes and at least one target to see how much can be added.

Checked one attribute at a time against the blend as shown above. A limit computed from a partly-assayed blend inherits that blend's coverage, so read it with the coverage figures.

Screen capture test

Point this at the dispatcher display, leave it running, then export what it saw. It reads the whole frame and records every word it finds with its position, rather than looking in places someone has guessed at — which is what decides how the rest of this tool has to work.

Nothing is uploaded. The recogniser is served from this site and runs in this tab; frames are drawn to a canvas and discarded. The export file is the only thing that leaves this machine, and you choose when that happens.

Written into the export so two runs can be told apart.

Can be changed while it runs. Faster is not better: recognition takes a second or two, and the gaps are recorded either way.

Labels on a map are usually small, so 2x is the one to start with. Enlarging costs speed but finds text that is otherwise invisible to the recogniser; 3x is the slowest.

Tip: Enter any known values to calculate the remaining results.

Designed to run locally in your browser. Tool inputs are not intentionally stored by Zerdly unless stated otherwise.