Just above cut‑off: how certain is the ore/waste decision?
Worked example · assumed grades · no site data
A worked example: a mined parcel estimated at 1.02 g/t gold, a cut‑off of 1.00, and four true grades it may have, each equally likely. We weigh milling it against dumping it, and what each wrong call would cost.
What we found
- A coin toss, still worth processingThe parcel has only a 50 % chance of being above its cut‑off, and processing it is still the better call.
- How far, not how oftenEach call is wrong in half the cases. Averaged over the four grades, processing's wrong calls cost 0.0625 g/t and dumping's 0.0825 g/t.
- The average grade decides, hereWith every g/t above or below the cut‑off worth the same, the parcel goes by the average of its four grades, here the 1.02 estimate. The chance of being ore never enters it.
- What sampling can addThe expected cost of the wrong calls caps what better grade information can be worth: 25 times more here than for a parcel with the same estimate and a tighter spread.
Method
Four assumed grades, one cut‑off
- AssumedFour true grades a parcelEach 25 % likely, a cut‑off of 1.00 g/t, and a value of processing over dumping that is a straight line in grade. Not from any site.
- EstablishedStandard decision rulesA cut‑off set by the cost difference between the two destinations, sunk costs left out; routing by expected grade; the expected cost of the wrong calls as the value of perfect information.
- CheckedExact arithmeticEach parcel's four grades give every value, by hand or in a spreadsheet. Our short script recomputes them and looks for each key statement on the page.
The references are under Sources, at the end of the page.
01 · The parcel
Just above cut‑off, on a coin toss
A parcel has already been mined, so its mining cost is spent either way: it goes to the mill or to the waste dump. The cut‑off, 1.00 g/t, is where the gold recovered just pays the extra cost of milling over dumping. The parcel is estimated at 1.02, and grade control leaves four true grades equally likely: 0.80, 0.95, 1.10 and 1.23 g/t.
Hypothetical parcel · g/t gold
Each pin is 25 % likely; its height only keeps the labels apart. Filled: above the cut‑off, worth processing. Hollow: below it. The grades are assumed for the example.
Two grades above, two below: the chance that this parcel is ore is 50 %. The estimate says ore; the odds say a coin toss.
02 · Both calls
Process or dump: each is wrong half the time
Take the value of processing over dumping as a straight line in grade: every g/t above the cut‑off earns the same, every g/t below it costs the same. A wrong call then costs the distance between the true grade and the cut‑off.
Costs in g/t of grade against the cut‑off; 0 where the call is right. Multiply by the gold price in dollars per gram and the recovery as a fraction for dollars per tonne.
Processing loses 0.0625 g/t on average to its wrong calls; dumping would lose 0.0825 g/t. The difference, +0.02 g/t, is the expected margin: the estimate minus the cut‑off.
The estimate decides
With straight-line value, the expected margin is the expected grade minus the cut‑off. Sending the parcel by its expected grade is the best call on average; the 50 % never enters it.
How far, not how often
Count every wrong call as equally bad and this parcel is a tie at 50 %. Processing misses by 0.05 or 0.20; dumping would miss by 0.10 or 0.23. The size of each miss decides.
03 · Same estimate
The same 1.02, three different parcels
Each parcel here averages 1.02 g/t, so processing has the same expected margin, +0.02 g/t, for all three. The chance of being ore runs from 25 % to 75 %. Pick a parcel and watch what moves.
Each pin is 25 % likely; heights only keep the labels apart.
3 of 4 grades above cut‑off: a 75 % chance of being ore. The expected margin from processing is still +0.02 g/t, so processing is still the better call. What changes is the cost of its wrong calls: 0.0025 g/t on average.
04 · Where it counts
Where the 50 % does count
With straight-line value, the chance of a wrong call does not change where the parcel goes. Knowing the true grade before deciding would remove the wrong calls, so their expected cost caps what better information can add.
The parcels of section 03. Any sample can add at most this much, before its own cost.
The same 1.02 and the same call, yet the most that better information could add is 25 times higher for the first parcel than for the tight one. The chance of being ore alone does not rank them: the parcel at 25 % has a lower cap than the one at 50 %. What one test adds depends on which grades it tells apart.
When value is not a straight line
Recovery that changes with grade, a full mill whose next tonne would displace better ore, or a stockpile that keeps the decision open can bend the line. Then the spread can count, not only the mean.
A parcel is not an assay
The four grades describe what the whole parcel may average, given limited samples of it. Uncertainty reported for one assay is not the parcel's; check what sampling and preparation it covers. The JORC Code keeps sampling, analytical and estimation errors apart.
A grade-control model that reports a distribution for each parcel, not one number, gives both: the best call, and what being wrong may cost.
Sources
- Dimitrakopoulos, R. and Godoy, M. (2014). Grade control based on economic ore/waste classification functions and stochastic simulations: examples, comparisons and applications. Mining Technology, 123(2), 90–106. doi:10.1179/1743286314Y.0000000062, on weighing each misclassification by its economic cost.
- Rendu, J.-M. (2008). An Introduction to Cut‑off Grade Estimation. Society for Mining, Metallurgy, and Exploration, Littleton, Colorado. Chapter 3, on the ore/waste cut‑off from destination costs, variable recovery and stockpiles.
- Lane, K. F. (2015). The Economic Definition of Ore: Cut‑off Grades in Theory and Practice, 4th edition. COMET Strategy, Brisbane. Chapter 2, on capacity constraints.
- Velasquez, H. G. and Deutsch, C. V. (2021). Decision making in the presence of geological uncertainty. Geostatistics Lessons. geostatisticslessons.com, on choosing an estimate and choosing an action as different problems.
- Phillips, J., Newman, A. M. and Walls, M. R. (2009). Utilizing a value of information framework to improve ore collection and classification procedures. The Engineering Economist, 54(1), 50–74. doi:10.1080/00137910802711883.
- Joint Ore Reserves Committee (2012). Australasian Code for Reporting of Exploration Results, Mineral Resources and Ore Reserves (the JORC Code), 2012 edition. PDF.