Chapter 30 · Dynamics and Integrated Case Studies

Material inventories, transport, and control

From a rate balance to a state balance

The working volume of a sump, the belt loading in a transfer circuit, and the accessible tonnes in a stockpile determine how disturbances move through a plant. Each stores material over time. The natural state is an amount vector M, containing water and dry solids resolved by component and, where needed, by size class. With t in hours and a feed vector F and withdrawal vectors P and E in t/h, the inventory balance over one accepted time interval is

M(t + Δt) − M(t) = ∫[t,t+Δt] (F − P − E) dt.

For a piecewise constant boundary rate, the amount delivered across a port is the rate times the segment duration. A reported average rate over a longer interval comes from adding the segment amounts and dividing by the whole interval. In a rate-based static flowsheet, conservation reduces to F = P + E at every inventory boundary. In a transient flowsheet it also includes initial stock, any unserved demand, and the final inventory. Unserved demand is a service statistic rather than a physical outlet. Overflow and rejected feed are physical outlets and must appear in the amount balance.

Apply the balance to water and each solid component and size class before reconstructing percent solids, grade, or percent passing. If inventory has 6 t of a 2.5 t/m³ solid and 3 t water at 1.0 t/m³, its phase volume is 5.4 m³ and slurry density is 1.667 t/m³. Averaging stream densities would give a different answer when their proportions vary. The time-step series should preserve both bounds and the material ledger. In a simultaneous convergence algorithm, repeated trial calculations within a step must leave its accepted inventory unchanged until the step is committed.

Mixed slurry storage and level breakthrough

Let V be liquid plus suspended solids phase volume, c the volume concentration of a conserved tracer, and Qᵢ, Qₚ and Qₑ the inlet, product and excess volumetric rates. An incompressible well-mixed model gives

dV/dt = Qᵢ − Qₚ − Qₑ
d(Vc)/dt = Qᵢcᵢ − (Qₚ + Qₑ)c
V dc/dt = Qᵢ(cᵢ − c).

The final line follows by using the volume balance in the product rule. It depends on every withdrawal having the tank concentration. Constant positive V and Qᵢ give c(t) = cᵢ + (c₀ − cᵢ) exp(−Qᵢt/V) for constant cᵢ. With constant nonzero net rate a = Qᵢ − Qₚ − Qₑ and V(t) = V₀ + at, integration instead gives c(t) = cᵢ + (c₀ − cᵢ)[V₀ / V(t)]^(Qᵢ/a). Piecewise expressions must stop at depletion, start of overflow, and any imposed change in flow. At zero inventory, a concentration inferred from volume is undefined until material is admitted.

A well-mixed 60 m³ tank receives 30 m³/h and discharges 30 m³/h. Its residence time is 2 h. After a tracer feed step from zero to concentration 0.10, its concentration at 2 h is 0.10(1 − e⁻¹) = 0.06321. Tracer inventory on a volumetric-fraction basis is then 3.793 m³ equivalent. Tracer feed was 6.000 m³ equivalent, so 2.207 m³ equivalent left in the product. The result explains why an instantaneous downstream grade change and a mixed-inventory response can produce the same long-horizon average while answering different operating questions.

A second sump starts at 80 m³, with capacity 100 m³, feed 40 m³/h, and available product flow 25 m³/h. It reaches capacity at (100 − 80)/(40 − 25) = 1.333 h. If those rates hold for 2 h, the overflow is 15 m³/h for the last 0.667 h. The 2 h volume ledger is 80 + 80 − 50 − 10 = 100 m³. Averaging excess over the whole interval gives 5 m³/h, while a downstream receiving vessel experiences 15 m³/h after breakthrough. Both numbers require the initial volume and event time.

Dry storage, reclaim order, and transport delay

Air voids contribute to a dry bed's occupied volume. With intrinsic particle volume Vₛ = ΣⱼMⱼ/ρⱼ and prescribed dry-bed void fraction ε, the volume law used in the conservative DPSIM bulk models is

Vbed = Vₛ/(1 − ε)
Voccupied = max(Vbed, Vₛ + Vw).

Water first fills the air voids. Layered storage evaluates this maximum separately for each layer and sums it, which preserves the occupied volume of spatially segregated wet and dry layers. For 90 t dry solids at 2.5 t/m³ and ε = 0.40, the grains occupy 36 m³, the dry bed occupies 60 m³, and its void volume is 24 m³. Retaining 18 m³ water leaves occupied volume at 60 m³. With 30 m³ water, occupied volume is 66 m³. This fixed-porosity calculation contains no consolidation or drainage.

Reclaim order matters even when every model closes the same mass balance. Start with 20 t dry ore at 1% metal and add 10 t dry ore at 4%. With 10 t dry reclaimed immediately, FIFO releases 0.10 t metal and ideal LIFO releases 0.40 t. A completely blended pile would release 0.20 t from the 30 t at 2% blended grade. The final inventories in the three cases contain 0.50, 0.20, and 0.40 t metal, respectively. Offtake positioned to favor one region can be represented by a layer policy when plant evidence supports this approximation. A separate geometric reclaim model would be needed to predict spatial mixing.

For a conveyor of length ℓ and uniform positive speed v, a parcel advances by vΔtₛ, where Δtₛ is in seconds. At constant feed, travel time is ℓ/v and mean belt inventory equals throughput multiplied by travel time. At wet loading λ kg/m, full-width throughput is 3.6λv t/h. A 120 m belt at 2 m/s with a maximum 25 kg/m wet loading can admit 180 t/h, holds 3 t when filled uniformly at that load, and has 60 s transit time. A 30 s belt stop displaces all remaining passage times by 30 s for a fixed return speed. A permitted upstream feed rate while stopped, and its destination, belong in the material boundary conditions.

DPSIM representationState and release mechanismImportant boundary behavior
Conservative Slurry SumpMixed solid and water amounts with additive phase volumesDischarge requested in m³/h, limited by available slurry. Excess has a material outlet.
FIFO Bulk HopperLayered bulk with a fixed void fractionOldest layers released first. Request specified as wet mass rate.
Layered Reclaim StockpileLayered storage with FIFO or ideal LIFOReclaim order and porosity affect the result. Ideal LIFO can remove newly admitted material before older layers.
Parcel Belt ConveyorPosition and material amounts in belt parcelsSpeed and loading define admission. Rejected feed has a material outlet. Stop preserves loaded material position.
Dynamic Process PileSingle blended holdupWet- or dry-basis discharge and a mass- or volume-based capacity convention.
Dynamic Conveyor BeltQueue of material amount snapshots advanced by a distance per stepRelease is discretized by the step. Verify the delay with steps shorter than belt transit time. The queued implementation has a separate branch for steps longer than travel time.
Dynamic Sump, Dynamic Overflowing Sump, PID Controlled SumpBuffered slurry with a volume-based product request, including a feedback-generated request in the PID variantOverflowing variant routes excess to a dedicated port. Track actuator bounds and the previous accepted level used for a controller update.

The model distinctions above apply to the dynamic calculation. Static treatment can pass the feed through while exposing a state variable or awaiting a dynamic transport step. For conservative models, the static feed-through result and zero excess do not forecast changes in storage. Compare the state trajectories under the intended execution mode. DPSIM dynamic modeling documentation

Conveyor power and transport delay

The steady Belt Conveyor Power representation evaluates a material-lifting and equivalent friction duty alongside an unchanged stream. Let L be belt length in m, h lift in m, v belt speed in m/s, f the equivalent friction factor, and q wet mass flow in kg/s. For an ascending conveyor the reference calculation is

H = max(0,h) + fL, P = q g H / 1,000 (kW).
τ = L/v.

Its capacity utilization is calculated from wet mass flow divided by the entered design rate, with both in the same units. At a wet rate of 450 t/h, length 180 m, lift 12 m, f = 0.03 and speed 2.5 m/s, the equivalent head is 17.4 m. Converting 450 t/h to 125 kg/s gives approximately 21.34 kW at g = 9.81 m/s². The travel time is 72 s. At a design limit of 500 wet t/h the nominal utilization is 90%. These are duty indicators under the chosen equivalent-friction convention. Motor, belt and empty-running losses, acceleration and route geometry require independent data for a detailed estimate.

An actual downhill belt can exchange useful work or need braking. The simplified nonnegative-lift calculation omits that behavior. At steady operation, 450 t/h distributed over 72 s corresponds to 9 t of material on the belt, a separate transport inventory. The dynamic belt representations use transit information to predict when a feed change reaches the next unit. The steady power operator calculates a travel-time indicator from the same length and speed even when it does not introduce that delay into the stream balance.

Feedback, permissives, and timing

For a pump manipulated to control sump level, define the error e = L − Lsp, with L as a fraction from zero to one. High level requires more discharge. An illustrative PI law is Qₚ,req = clip[Q₀ + Kₚe + Kᵢ∫e dt, 0, Qmax]. With time in hours, Kₚ has units m³/h per unit level fraction and Kᵢ m³/h² per unit level fraction. Gain values defined for percentage points differ by a factor of 100. A blocked discharge or an empty tank may leave the controller's request far from the achieved flow. Tracking or conditional integration limits windup during saturation. Model simulated response with the controller update interval, actual pump limits, and any latency in the level signal.

Suppose the preceding 100 m³ sump begins at 60%, feed suddenly rises to 40 m³/h, and actual discharge before response is 25 m³/h. A high-level action at 85% first becomes eligible after (85 − 60)/15 = 1.667 h, assuming level measures working volume. A continuous 12 min confirmation delays that action to 1.867 h, when volume is 88 m³. If the action instantly raises achievable discharge to 45 m³/h, the initial post-action slope is −5 m³/h. An 80% release threshold introduces hysteresis and prevents the rule from switching back as soon as level moves just below 85%. With the same flows, reaching that release threshold takes 1.6 h. The sequence changes if the action is a latch that requires a reset, a permissive for a discrete start, or a route selection that throttles another stream.

A process study should record separately the signal crossing, delay completion, action request, equipment transition, and resulting material change. DPSIM's interlock definitions expose threshold comparisons, hysteresis, on- and off-delays, and typed actions. A simultaneous pump trip and threshold-crossing event is a test of priority and causality. Check whether the model first commits the equipment state, recalculates the achieved discharge, or evaluates the timer. Those decisions govern the peak inventory. Simulated permissives and trips depend on the modeled signal and event sequence. DPSIM Developer documentation

Exercises with answers

  1. A well-mixed hold-up of 45 m³ has throughput 90 m³/h. How long to reach 95% of the final tracer response? Answer. −(45/90) ln(0.05) = 1.498 h.
  2. Two layers each contain 10 m³ grain volume and have ε = 0.50. The first holds 0 m³ water and the second holds 16 m³. Compare layer-wise and aggregate occupied volumes. Answer. The first layer occupies 20 m³ and the second 26 m³, for 46 m³. Lumped calculation gives max(40, 36) = 40 m³ because it lets the excess water in one layer use voids in the other.
  3. A 120 m conveyor normally runs at 2 m/s. A parcel is 40 m from discharge when speed falls to 1 m/s for 25 s, then returns to 2 m/s. How much later does it arrive? Answer. It travels 25 m during the slowdown and then travels 15 m in 7.5 s. Arrival in 32.5 s is 12.5 s later than the original 20 s.
  4. For the 100 m³ sump with net input 15 m³/h, a 1 min timing uncertainty near the high threshold converts to what uncertainty in level? Answer. 15/60 = 0.25 m³, or 0.25 percentage points of full working volume, while the net flow stays constant.