Chapter 4 · Material, Measurements, and Balances

Process balances and stream constraints

A flowsheet with steady solids throughput still needs distinct balances for dry solids, water and selected constituents. For a boundary around any unit, steady operation gives zero accumulation of a conserved quantity. In dynamic simulation, the difference of the incoming and outgoing flows is the rate of change of inventory. Apply these equations to absolute amounts. A concentration is a ratio and cannot be summed across streams without its denominator.

dM/dt = Σ qᵢₙ − Σ qₒᵤₜ + g.

The term g accounts for any genuine source or sink of the conserved quantity specified within the boundary. For a nonreactive mineral circuit with all physical streams counted, g is zero. Added dilution water is an input. Evaporation is an exit when it is significant. Fresh solids can also appear inside a chosen accounting boundary if the upstream stockpile is excluded and its reclaim is treated as a measured feed. State the boundary before explaining a residual.

Dry solids and water bookkeeping

Let S and W be dry solids and water rates, and ρₛ and ρw the respective densities on consistent mass and volume units. With additive particle and water volumes, the following quantities are available directly.

Q = S/ρₛ + W/ρw, ρₚ = (S + W)/Q.
wₛ = S/(S + W), φₛ = (S/ρₛ)/Q.

The first solids fraction is by mass and the second by volume. At S = 120 t/h, W = 80 t/h, ρₛ = 3.2 t/m³ and ρw = 1.0 t/m³, Q = 117.5 m³/h and ρₚ = 1.7021 t/m³. Solids are 60.0% by mass and 31.9% by volume. If 3.2 were used as bulk stockpile density, porosity and its water content would be counted incorrectly. When soluble salts, gas entrainment or significant nonadditive volumes matter, derive an appropriate density closure or use measured density with its conditions.

For an exclusive mineral mixture, determine true particle density using specific volumes. With proportion wᵢ for mineral i and density ρᵢ, specific volume is Σ wᵢ/ρᵢ. A mixture of 20% mineral of density 5.0 t/m³ and 80% gangue of density 2.65 t/m³ has density 1/(0.20/5.0 + 0.80/2.65) ≈ 2.925 t/m³. Direct arithmetic averaging of the densities gives 3.12 t/m³. This difference propagates into slurry volume and volumetric flows used for hydraulic normalization.

A feed that can only add water

Given solids S and a target mass fraction w, the total water corresponding to the target is W = S(1 − w)/w. A dilution unit adds max(0, W − W) when w is positive and attainable. It predicts a shortfall if actual feed is already more dilute than the target. In a feed of 180 t/h solids and 120 t/h water, reducing solids from 60% to 45% requires 180 × 0.55/0.45 − 120 = 100 t/h extra water. Tightening the target to 65% calls for only 96.92 t/h total water, which would require removal of 23.08 t/h. A physically constrained dilution model leaves the water unchanged in that case. A controller that accumulates error against such an unreachable set point needs saturation handling or a different actuator.

DPSIM's feed abstractions cover measured product streams, pure water, varying distribution, batch composition, imposed flow changes and water addition toward a solids target. The flow-rate transition is useful at a measured interface where upstream composition determines the downstream composition but measured downstream throughput sets scale. If linked to a physical continuous plant, the difference between the two dry solids flow rates must be understood as an external input, an exit, accumulation or measurement discrepancy. Re-scaling a stream to the newer flow is an imposed boundary operation. Any physical storage or loss behind it has to be described or counted explicitly.

Mixers and splitters

The grade of a blended stream is the sum of tracer rates divided by combined total dry rate. For feed A of 80 t/h at 2.0% and feed B of 120 t/h at 0.5%, the blended rate is 200 t/h and its tracer rate is 1.6 + 0.6 = 2.2 t/h. The grade is 1.10%. If their passing fractions at a common aperture are 0.25 and 0.80, the combined passing fraction is (80 × 0.25 + 120 × 0.80)/200 = 0.58. The direct arithmetic average 0.525 would describe an equal-solids blend. In a material model with multiple components, mix class inventories directly and reconstruct conditional grades in each class.

An ideal stream splitter sends a common fraction to each outlet for both solids and water and for every component–size cell. A three-way distribution with factors 0.40, 0.25 and 0.35 conserves both phases and preserves every normalized property at all three outputs. Factor sum must be unity. A volume-based target on an ideal splitter uses the common ratio Qtarget/Qin where that target is achievable. If a boundary is designed to send different solids fractions and water fractions, it represents an operation with selective phase transfer or connected dilution or extraction. Its products have their own concentrations.

A negative complement is the easiest indication of an invalid requested split. A less obvious limit arises when a model fixes output size and component grade together. If an output of 50 t/h requires 30 t/h of coarse particles and the feed carries only 20 t/h coarse, a true splitter cannot supply the request. An imposed PSD output can still be useful in a test circuit when size transformation is external or inferred. Carry a separate record of the assumed transform. If the splitter is physical, enforce 0 ≤ y[c,i] ≤ f[c,i] and compute the complement from the actual feed. A two-output prescribed PSD with a physically conserved residual has one fewer free choice: for every cell the second output is fixed by difference. If the model fixes product PSD while preserving bulk constituent grade, it conserves that constituent globally while it can redistribute its association with size. In a size-selective separation, validate that association before reusing it.

The two-product balance and its sensitivity

A feed rate F is separated to concentrate C and tailings T. Let g_f, g_c and g_t be grades of a conserved tracer on a common dry mass basis.

F = C + T, F g_f = C g_c + T g_t.
C/F = (g_f − g_t)/(g_c − g_t).
R꜀ = (C/F)(g_c/g_f).

For F = 200 t/h, grades of 1.10%, 11.0% and 0.30%, estimate yield 0.80/10.70 = 0.074766, concentrate rate 14.953 t/h, and valuable recovery 0.74766. The tracer rates are 1.6449 t/h in concentrate and 0.5551 t/h in tailings. Their sum is 2.2000 t/h. The tail grade carries leverage because tailings mass is much larger than concentrate mass. If the measured tail grade changes from 0.30% to 0.35% while feed and concentrate grades are fixed, yield becomes 0.75/10.65 = 0.07042 and recovery 70.42%, a decrease of 4.34 percentage points.

The formula is ill conditioned when concentrate and tailings grades approach one another. For 1.10% feed with 1.13% and 1.07% products, yield before noise is 0.5. An assay change of just 0.03 percentage points to 1.13% in the feed suggests all mass went to concentrate. This agrees algebraically and is unreliable physically given the weak separation signal. A calibrated belt scale or a second, more selective tracer can add independent information.

Mixing and retained inventory

The dry mass entering a stockpile during an interval and phase-specific rates out may differ substantially. Closing a boundary around a buffer during a scheduled transition requires subtracting inventory change. Over two hours with 150 t/h in and 120 t/h reclaimed, the dry stock increases by 60 t. If the incoming composite is coarse and the initial stockpile is fine, transition timing of the product depends on reclaim geometry. A fully mixed stockpile updates bulk grades immediately upon addition. A chronological reclaim model can show a delayed response. Both remain consistent with the same aggregate balance.

Time averages of grades are meaningful only with their weights. Over an hour at 100 t/h and 1.8%, followed by an hour at 200 t/h and 0.6%, combined grade is (1.8 + 1.2)/300 = 1.0%. Time weighting alone gives 1.2%. Flow-weighted composites and sampling clocks synchronized to surge and residence time are essential if the measurements are to compare with instantaneous or campaign model outputs.

Exercises with solutions

Exercise 1. A blend has 40 t/h of solids at true density 4.0 t/m³ and 60 t/h at 2.5 t/m³, with 100 t/h water of density 1.0. Find the mixture particle density, slurry volumetric flow and solids volume fraction.

Solids volumes are 10 and 24 m³/h. Particle density is 100/34 = 2.9412 t/m³. Slurry volume is 134 m³/h and solids volume fraction 34/134 = 25.37%. Mass fraction is 50%.

Exercise 2. A splitter with 200 m³/h slurry receives a demand for 55 m³/h at one outlet. Incoming solids and water are 150 and 140 t/h. Compare the phase flows if the unit is an ideal splitter and if the specified product in fact carries 10 t/h solids.

An ideal split uses fraction 0.275, producing 41.25 t/h solids and 38.50 t/h water. A product of the same volume carrying 10 t/h solids is physically viable only with a different water fraction and a separation or additional water balance. At solids density 2.5 t/m³ and water density 1.0, the target carries 4 m³/h solids and 51 m³/h water. The total volume constraint by itself does not determine the phases.

Exercise 3. A measured feed is 90 t/h at 2.0% valuable mineral. An imposed downstream rate is 120 t/h at the same grade. Find the implied external input if operation is steady, all measured rates are correct and there is no reaction.

An uncounted 30 t/h dry solids must enter with 0.60 t/h valuable mineral, consistent with an external 2.0% source. If an independent source has very different grade, one of the hypotheses about source or matching composition has failed.