Appendix B · Appendices

Calculation laboratories

The following studies use illustrative data. They can be solved with a calculator, a script, or any computation environment that preserves complete table precision. Each study poses a decision, gives its input dataset and develops an answer that can be independently checked.

Laboratory 1. A screen cut with uncertain fines

A screen treats 180 dry t/h with 75 t/h water. All size classes are given in descending order. The measured oversize partition is judged stable for the first comparison. An interlocked downstream load limit requires no more than 100 dry t/h in oversize. Process engineering proposes using the feed's reported d₈₀ as the only material descriptor for a range of planned feed blends. Test the adequacy of that basis.

Size range, mmBaseline dry fractionAlternative dry fractionOversize partition
> 80.120.120.98
4–80.240.240.88
2–40.240.310.55
1–20.150.160.14
≤ 10.250.170.06

Both PSDs have 88% passing 8 mm and 64% passing 4 mm. Under a log-size interpolation, d₈₀ is 4 × 2^[(0.80 − 0.64)/(0.88 − 0.64)] ≈ 6.350 mm. The value is identical before any screen calculation. The two PSDs distribute dry mass differently around the working cut region.

For the baseline blend, oversize flow contributions are 21.168, 38.016, 23.760, 3.780 and 2.700 t/h. The corresponding sum is 89.424 t/h. The undersize is 90.576 t/h. In the alternative blend, the first two class flows are identical and the remainder becomes 30.690, 4.032 and 1.836 t/h. Oversize is 95.742 t/h and undersize is 84.258 t/h. Both meet the nominal load limit, but free capacity declines from 10.576 to 4.258 t/h at fixed fresh feed. The effect is entirely within classes smaller than 4 mm.

If the oversize water fraction is 0.17, both oversize products take 12.75 t/h water. For the baseline and changed solids distributions, oversize solids fractions are about 87.52% and 88.25%. The water fraction was imposed from a survey for this comparison. A change in screen operation associated with the load could also change this split. At unchanged partition, the maximum alternative-blend fresh rate compatible with the dry oversize limit is 100/(95.742/180) ≈ 188.01 t/h. With 180 t/h fresh input, a proportional increase of 5% gives 189 t/h and nominally exceeds the oversize limit.

Now test a wet-condition change that raises partition in the 2–4 mm class from 0.55 to 0.65 because the deck retains more near-size material. It adds 0.10 × 180 × 0.31 = 5.58 t/h to oversize on the changed blend. Oversize becomes 101.322 t/h. The same d₈₀ and fresh feed rate now accompany an overload. This assessment separates the impact of the feed distribution from an independent shift in separator behavior. Data in both regions are required to establish which schedule remains feasible.

Extension. Calculate the marginal effect of redistributing one percentage point of the feed from the pan class to 2–4 mm under baseline operating conditions. The change is 180 × 0.01 × (0.55 − 0.06) = 0.882 t/h extra oversize. This derivative conserves dry feed and can be multiplied by a small uncertainty in the measured redistribution. It should be re-evaluated if the loading itself changes the partitions.

Laboratory 2. Mill breakage and selective return

The plant receives two size classes coarse and fine. The fresh flow vector is (72, 48) t/h. The fraction of mill feed coarse material converted to fine per pass is β. A perfect binary transfer leaves the remaining coarse in its class and leaves incoming fine material as fine. The coarse and fine partitions to classifier return are r꜀ = 0.78 and rꜰ = 0.18. Use β = 0.42 initially. Determine whether an increase in per-pass breakage from 0.42 to 0.50 is worth pursuing when the mill circulating-solids limit is 200 t/h total mill feed.

Let X꜀ and Xꜰ be mill feed rates. Coarse recirculation is r꜀(1 − β)X꜀. Fine recirculation is rꜰ(Xꜰ + βX꜀). The exact closed solutions are

X꜀ = 72/[1 − r꜀(1 − β)].
Xꜰ = (48 + rꜰ β X꜀)/(1 − rꜰ).

For β = 0.42, X꜀ is approximately 131.48 and Xꜰ 70.66 t/h, giving 202.14 t/h mill feed. The mill discharge components are (1 − β)X꜀ ≈ 76.26 t/h coarse and the complementary 125.88 t/h fine. Final coarse is approximately 16.78 t/h, and the final fine rate is 103.22 t/h. Sum to 120 t/h fresh feed and check that (X꜀ + Xꜰ) − 120 equals the sum of the recycled coarse and fine rates, about 82.14 t/h.

At β = 0.50, the solution is X꜀ ≈ 118.03 and Xꜰ ≈ 71.49 t/h, giving 189.52 t/h mill feed. Final coarse falls to about 12.98 t/h. Under fixed classifier partitions this setting achieves the mill feed limit with 10.48 t/h margin. Some of the extra breakage effect is consumed by reduced internal loop load. The per-pass β required to reach the improved state may need increased power or a change in holdup. Test such changes under the actual equipment constraint before assuming they are achievable.

Quantityβ = 0.42β = 0.50
Fresh dry solids, t/h120.00120.00
Mill feed dry solids, t/h202.14189.52
Circulating dry solids, t/h82.1469.52
Final coarse dry solids, t/h16.7812.98
Final fine dry solids, t/h103.22107.02

If a proposed change also increases rꜰ from 0.18 to 0.28 at β = 0.50, hold X꜀ fixed in the same equation and solve for Xꜰ again. It becomes (48 + 0.28 × 0.50 × 118.03)/0.72 ≈ 89.62 t/h. Total mill feed is about 207.65 t/h. Fines misplacement can erase the capacity benefit of improved breakage. This interaction warrants checking classifier water split and feed concentration together with mill power.

Extension. In the baseline model, assume a 5% decline in the dry fresh throughput with unchanged fresh size fractions, β and classifier partitions. All rates scale to 0.95 of baseline, so mill feed becomes roughly 192.03 t/h. The 5% reduction is below the limiting throughput and preserves product size fractions. Any change in solids concentration that changes classifier behavior invalidates exact proportional scaling.

Laboratory 3. Entrainment, grade and an observed water response

A rougher circuit processes 90 dry t/h comprising 9 t/h valuable and 81 t/h gangue. The initially fitted fractions recovered by attachment are 0.78 and 0.035 for valuable and gangue, respectively. Of the unfloated fraction, a water entrainment approximation transfers e Rw where e is 0.10 for valuable and 0.30 for gangue. Feed water is 140 t/h. Compare the concentrate and tailings for water recovery Rw = 0.18 and 0.26. Under the exercise conditions the true flotation fractions are independent of Rw.

R꜀ = Rattach + (1 − Rattach) e Rw.

For Rw = 0.18, the valuable recovery is 0.78396 and gangue recovery is 0.08711. Concentrate solids are 7.05564 t/h valuable and 7.05591 t/h gangue, almost exactly equal. Overall grade is approximately 50.00%. Concentrate water is 25.2 t/h. For Rw = 0.26, valuable recovery is 0.78572 and gangue recovery is 0.11027. The concentrate contains 7.07148 t/h valuable and 8.93187 t/h gangue. Grade falls to about 44.19% and concentrate water rises to 36.4 t/h.

Valuable mineral recovery rises by 0.176 percentage points. The dry concentrate contains about 1.876 t/h extra gangue and only 0.01584 t/h extra valuable mineral. This small incremental ratio can justify a concentrate-capacity or quality analysis independent of changes in pipeline or filter loading. The experiment assumes entrainment and attachment are sequential mechanisms on the feed fraction. A model using joint transfer probabilities may need recalibration to give the same observed grade and yield.

QuantityRw = 0.18Rw = 0.26
Concentrate solids, t/h14.111616.0034
Concentrate grade, %50.0044.19
Valuable recovery, %78.39678.572
Concentrate water, t/h25.2036.40
Tailings solids, t/h75.888473.9967
Tailings water, t/h114.80103.60

Survey the water rates and gangue or valuable assays over independent, comparable water settings to estimate the entrainment response. A single fitted concentrate grade does not identify the two mechanisms. Algebraically, for fixed Rw and observed overall recovery r, solving for attachment gives (r − e Rw)/(1 − e Rw). A range of e values can satisfy the one observation. A second water setting can provide independent information only if attachment and material properties were held sufficiently comparable.

Extension. If the tailings thickener receives the complete tailings and overflow solids loss is negligible, predict the underflow water at 58% solids by mass. For the 18% water recovery case it is 75.8884 × 0.42/0.58 ≈ 54.95 t/h, leaving about 59.85 t/h overflow water. For the 26% case it is approximately 53.58 t/h, leaving about 50.02 t/h overflow. Raising froth water recovery can reduce tailings overflow water available for return even though dry throughput is unchanged.

Laboratory 4. Sump protection and restart inventory

A sump has 36 m³ of usable inventory between operating level and high trip. A total inlet of 216 m³/h arrives during stable operation. A temporary drive limit reduces discharge from 216 to 144 m³/h. The high-level interlock is confirmed after the measured level exceeds its threshold for 20 seconds. An alarm and corrective action already trigger when there are 9 m³ of headroom left. A sample-based control signal is evaluated at 10-second intervals. Treat the flows as exactly constant between their specified changes.

The net filling rate is 72 m³/h = 1.2 m³/min. It takes 27/1.2 = 22.5 minutes to reach the alarm and 36/1.2 = 30 minutes to reach the high trip. Over the 20-second confirmation interval, another 0.4 m³ enters. Phase routing to overflow or further safe storage needs that amount of capacity if inflow and outflow remain unchanged. Sample timing can add up to one 10-second interval to a threshold detection under the stated clock, corresponding to a further 0.2 m³.

A relief routing activated at the alarm diverts 54 m³/h from the inlet, reducing net filling rate to 18 m³/h. The time from alarm to high trip lengthens from 7.5 minutes to 30 minutes. A diversion of exactly 72 m³/h holds the model level constant, giving no recovery margin for sensor bias or fluctuating inflow. A diversion of 90 m³/h draws down at 18 m³/h. If the pump recovers to 216 m³/h while that 90 m³/h diversion remains active, sump inventory shrinks at 90 m³/h. Releasing the diversion on recovery can prevent a new low-level condition.

The ability to route 90 m³/h elsewhere introduces another boundary with capacity and phase conditions. A flow path opens physically only if the receiving equipment and operating logic allow it. A simulation can test delayed corrective action by tracking each control command and its actual achieved flow, together with measured level and accumulated overflow.

Extension. With the 54 m³/h diversion in place, a one-minute delay to detect the alarm spends 1.2 m³ of the remaining 9 m³ headroom before diversion begins. At 18 m³/h thereafter, the remaining 7.8 m³ lasts 26 minutes. Total post-alarm headroom time becomes 27 minutes, three minutes less than instantaneous diversion.

Laboratory 5. Thickening and filter-water competition

A plant water-recovery train receives 78 t/h dry tailings and 112 t/h water. The thickener captures 98% of dry solids to underflow, with an underflow mass fraction of solids of 0.59. Assume dissolved material is negligible and phase volumes are additive. The underflow feeds a filter that captures 99.5% of its dry solids in cake. Cake moisture is 17% wet basis. Both recovered liquid streams carry their computed residual solids. Estimate return quantities and determine which can be sent directly to a branch with a 1.0% by mass solids limit.

QuantityThickener overflowFilter filtrate
Dry solids, t/h1.56000.3822
Water, t/h58.880737.5412
Solids content by mass, %2.5811.008

The thickener underflow carries 76.44 t/h solids. At the specified concentration its water rate is 76.44 × 0.41/0.59 = 53.12 t/h. The cake carries 76.0578 t/h solids and approximately 15.5781 t/h water. The filtrate contains the overflow complement of the filter solids, 0.3822 t/h, and the remaining 37.5412 t/h water. Summing final solids gives 1.56 + 0.3822 + 76.0578 = 78 t/h. Water closure gives 58.88 + 37.5412 + 15.5781 = 112 t/h. Additional decimal places are useful for the quality check because the filter result lies close to the threshold.

Neither recovered stream meets the nominal 1.0% mass limit unaided. For water-grade treatment of the 37.5412 t/h filtrate, its solids would need to be at most W × 0.01/0.99 ≈ 0.3792 t/h, a difference of around 3 kg/h from the illustrative prediction. A formal acceptance decision therefore depends strongly on measurement tolerance and process variation. Combining the two streams makes a return of 96.4219 t/h water and 1.9422 t/h solids at approximately 1.975% by mass. If perfectly clean water could be added, dilution of the combined stream to 1.0% would require about 95.86 t/h added water. This consumes more clean water than the combined recovered rate. Classification of acceptable uses by their quality tolerances or clarification of part of the return provides different options for a whole-plant water model.

For a quality target Cw in mass fraction and returned clean-water rate A, the constraint is S ≤ Cw(W + A)/(1 − Cw). The same algebra can estimate a minimum fines removal fraction. It needs revision if dissolved constituents or particle-species specific restrictions control the destination. At every proposed split, check where removed solids and reject-water mass report.

Extension. If cake moisture declines from 17% to 15% wet basis at unchanged solids capture, cake water becomes 76.0578 × 0.15/0.85 ≈ 13.4220 t/h. Filter water return rises by about 2.1562 t/h. The resulting filtrate solids content is about 0.954% by mass, which meets the nominal criterion in this deterministic example. Verifying the filter capacity and cake moisture at the same feed is a separate test.