Worked example
Example: Wills Example 3.9 zinc, assays to 0.05 / 0.5 / 0.02 % Zn. The inputs, in the units selected above:
| Input | Value |
|---|---|
| Feed assay | 3.9300 % |
| Concentrate assay | 52.070 % |
| Tailings assay | 0.49000 % |
| Standard deviation of the feed assay (absolute) | 0.050000 % |
| Standard deviation of the concentrate assay | 0.50000 % |
| Standard deviation of the tailings assay | 0.020000 % |
Applying the formula below gives:
| Result | Value |
|---|---|
| Metal recovery | 88.363 % |
| Standard deviation of recovery | 0.50558 % |
| 95 % half-range of recovery | 0.99093 % |
| Share of variance from the feed assay | 10.032 % |
| Share from the concentrate assay | 0.025419 % |
| Share from the tailings assay | 89.942 % |
Open this example in the calculator and change any input.
Formula
\(V_R = \left(\frac{\partial R}{\partial f}\right)^2 V_f + \left(\frac{\partial R}{\partial c}\right)^2 V_c + \left(\frac{\partial R}{\partial t}\right)^2 V_t, \quad R = \frac{c(f - t)}{f(c - t)}\)
Variables
| Symbol | Meaning | SI unit |
|---|---|---|
| \(f\) | Feed assay | - |
| \(c\) | Concentrate assay | - |
| \(t\) | Tailings assay | - |
| \(\sigma_f\) | Standard deviation of the feed assay (absolute) | - |
| \(\sigma_c\) | Standard deviation of the concentrate assay | - |
| \(\sigma_t\) | Standard deviation of the tailings assay | - |
| \(R\) | Metal recovery | - |
| \(\sigma_R\) | Standard deviation of recovery | - |
| \(1.96\sigma_R\) | 95 % half-range of recovery | - |
| \(V_f\) | Share of variance from the feed assay | - |
| \(V_c\) | Share from the concentrate assay | - |
| \(V_t\) | Share from the tailings assay | - |
Method and standards
First-order error propagation.
Where it is used
- Planning assay precision for plant trials.
- Judging whether a change in recovery is real.
Common mistakes
- Entering relative instead of absolute standard deviations.
- Trusting the recovery of a poorly separated metal.
Assumptions and limits
- Independent, unbiased assay errors.
References
- Wills and Finch
- Wills' Mineral Processing Technology
- 8th ed. (Butterworth-Heinemann, 2016)
- ch. 3
- eqs. 3.30-3.33 and Example 3.9
Questions and answers
Which assay matters most?
Usually the tailings assay: its coefficient is largest because the recovery depends on the small difference between feed and tailings.
Why is copper so poorly defined in Wills's example?
Copper is barely separated in a zinc circuit (0.66 % against 0.14 %), so the denominator c − t is small and errors multiply.
What do the defaults represent?
Wills's Example 3.9 zinc assays with standard deviations of 0.05, 0.5 and 0.02 % Zn on feed, concentrate and tailings.
Method and article reviewed by FireCAD engineering on 17 November 2026. Every result is checked against an independent calculation on each release: how we verify our calculators.