The downstream boundary of the mill
Two particles of the same size may experience the same breakage environment and have different chances of leaving if they are in different transport regions. Conversely, a particle can stay in the same size class across an abrasion event and still pass through a discharge opening. A mill model needs explicit rules for both destinations. The breakage rule tells us how material enters size classes. The discharge rule tells us whether those classes contribute to inventory, the product or a separate discharge stream.
At steady state, let Dᵍᵢ and Dᵖᵢ be discharge rates to a grate and to a pebble port, both in h⁻¹. The total first-order discharge rate is Dᵢ = Dᵍᵢ + Dᵖᵢ. For a single completely mixed inventory Mᵢ, the two outlet rates are
When Dᵢ > 0, the pebble share of that class's discharged mass is Dᵖᵢ/Dᵢ. This conditional share is distinct from Pᵖᵢ/Fᵢ, which can include gains from coarser classes. A useful check is that the calculated products sum to DᵢMᵢ for every class and component. A more stringent check reconciles the inferred total inventory against a measured charge filling or independently estimated hold-up. A family of models can reproduce the first check and disagree substantially on the second.
SAG mills with grate and pebble discharge
The variable-rates SAG representation assigns streams to a normal grate and, when used, a pebble opening. Within each opening's transition interval, a size x has fractional access
Access is one at and below x𝒻 and zero at and above xₐ. The separate opening fractions scale the corresponding rates. All sizes in the logarithm must use the same units. Sizes far above the grate opening can leave through pebbles if their pebble access remains positive. Even among particles below an opening, class discharge changes continuously across its access interval.
Breakage uses distinct impact and abrasion distributions. If w is the impact share, each daughter probability combines w bᶦᵐᵖᵢⱼ + (1−w)bᵃᵇʳᵢⱼ. This preserves the daughter column sum when both columns are individually normalized. A branch that returns mass to the parent class cancels the corresponding part of gross selection when solving for inventory. Mill geometry, total filling, charge media, water and solids establish an inventory closure. A calibrated throughput can therefore be sensitive to a grate-access change even if the breakage parameters are held fixed.
The simplified population-balance SAG form uses geometry and charge to estimate net power and divides selection between contributions from balls, rocks and self-breakage. It applies a component-specific grinding transformation and a grate discharge relation with internal return. The variable-rates form exposes two solids discharge routes and fits selection and discharge rates on an inventory basis. Use separate pebble sampling to identify the latter model's opening terms. If a single mass measurement combines both discharges, changes in the two rate terms can compensate, leaving the combined product nearly unchanged while changing calculated pebble load.
Worked example with separate routes
One coarse class spans 5,000 to 3,000 µm and is represented by x = √(5000 × 3000) = 3,872.98 µm. Feed consists of 100 t/h in this class at solids density 2.60 t/m³ and 80 t/h of water. Let full access begin at 843 µm, with grate and pebble upper limits of 1,858 and 75,000 µm. The coarse grate access is zero. Pebble access is approximately ln(75,000/3,872.98)/ln(75,000/843) = 0.66027. Multiplying by the pebble opening fraction 0.85 and the fitted maximum discharge rate 28.7479 h⁻¹ gives a coarse discharge rate of 16.1341 h⁻¹.
Gross coarse breakage is 4 h⁻¹. Half of the breakage is impact sending all selected coarse mass to the finest class. The remaining half is abrasion that returns the mass to the same modeled coarse interval. Net coarse transfer is 2 h⁻¹. The coarse inventory is 100/(16.1341 + 2) = 5.514 t, the pebble route carries 88.971 t/h of coarse solids, and 11.029 t/h enters the fine class. In this case the fines leave through the grate. The fine inventory is about 11.029/28.7479 = 0.384 t. Total ore inventory is about 5.898 t.
With 2% of water assigned to pebbles, the water rates are 1.60 t/h in pebbles and 78.40 t/h at the grate. The water inventory at the fitted maximum rate is 80/28.7479 = 2.783 t. The reference geometry and charge inputs for this worked case give chamber volume 212.750 m³ and media inventory 139.394 t. The discharge audit closes at 100 t/h of ore and 80 t/h of water. Both the computed pebble load and the charge inventory can be checked against independent observations. A coarse sample from the pebble route constrains the accessibility term more directly than the total mill product does.
Rod-mill stage scaling
The calibrated Lynch rod-mill representation applies a sequence of breakage stages. A convenient calibration is an effective stage count N_ref at a reference solids rate Q_ref. For actual rate Q, the scaled count is
Each full stage breaks all flow available in a successive parent interval. The fractional last stage breaks only its fraction of the available parent. At stage j, letting N be the effective count capped by the number of intervals from the first occupied parent to the pan, the stage fraction is
Subtract fⱼMᵢ from the current parent flow Mᵢ and place it among its finer daughters. Subsequent stages include any new material in their parent interval. The process preserves mass when the daughter fractions are normalized. It models how the exposure varies with solids throughput, with no explicit water hold-up equation. Its fitted stage count can absorb changes in other operating conditions.
Take Q_ref = 100 t/h, N_ref = 1 and actual Q = 200 t/h. Feed all 200 t/h in 8,000–4,000 µm. The half-stage retains 100 t/h in the parent and transforms the other 100 t/h. The Broadbent–Callcott cumulative probability B(x|y) = [1−exp(−x/y)]/[1−exp(−1)] uses the lower parent boundary y = 4,000 µm. At 2,000 and 1,000 µm the cumulative fractions are 0.622459 and 0.349932. Differencing these values and collecting the final remainder gives 37.754 t/h in 4,000–2,000 µm, 27.253 t/h in 2,000–1,000 µm and 34.993 t/h in the pan below 1,000 µm. With the retained 100 t/h, product closes at 200 t/h. If there is 6 t/h of feed water it reports to the product unchanged.
Vertical stirred mills and retained coarse charge
A vertical stirred chamber has space occupied by media and slurry. With diameter D, length L, media bed height H and media solids fraction fₘ, a simple geometric closure is
Divide slurry volume by slurry throughput to obtain a nominal residence time. A rate-ratio formulation can use this nominal time for all discharged size classes. A rate-and-discharge formulation can instead retain coarse material that cannot access slurry discharge. This retained coarse charge occupies part of Vₛₗ. The remainder determines a liquid discharge rate and an inventory consistent with slurry transport.
A chamber of diameter 4.10 m and length 4.80 m has V꜀ = 63.372 m³. A 1.92 m media bed with solids fraction 0.60 occupies 15.209 m³, leaving 48.163 m³ for slurry. Feed of 238.565 t/h solids at 2.80 t/m³ and 128.458 t/h water gives Qₛₗ = 213.660 m³/h when water density is 1.00 t/m³. The nominal residence time is 48.163/213.660 = 0.2254 h or 13.53 min.
For a feed entirely in one coarse class, with net coarse breakage 3 h⁻¹ and all coarse discharge accessible, the ratio of breakage to discharge is 3 × 0.2254 = 0.6763. The two-class steady product is 238.565/(1+0.6763) = 142.320 t/h coarse and 96.245 t/h fine. Now consider a different parameterization, with no coarse discharge because representative size 424 µm is above a 400 µm cutoff, and net coarse breakage rate 20 h⁻¹. The coarse inventory must be 238.565/20 = 11.928 t, occupying 4.260 m³. Available flowing slurry space is 48.163−4.260 = 43.903 m³. Water discharge rate becomes 213.660/43.903 = 4.867 h⁻¹ and fine solids carry the entire 238.565 t/h product. Total solids and water inventories are about 60.949 t and 26.396 t. These examples deliberately change both transport and breakage. They illustrate what each closure computes and cannot isolate an effect attributable solely to one parameter.
Exercises and solutions
1. Rate and outlet checks
A single coarse inventory receives 50 t/h. Breakage rate is 3 h⁻¹, half of the selected mass remains in the same class, and discharge rates are 5 h⁻¹ to pebbles and 0.5 h⁻¹ to the grate. Calculate coarse lead-through to both outlets and the rate of fine formation.
Solution. Net breakage is 3(1−0.5) = 1.5 h⁻¹. M = 50/(1.5+5+0.5) = 7.1429 t. Pebble, grate and fine formation are 35.714, 3.571 and 10.714 t/h respectively. Round only after summing.
2. Space available for transport
A stirred mill has 30 m³ of slurry space, including 5 m³ of coarse retained solids. Slurry throughput is 150 m³/h. Compare the nominal discharge rate with the rate based on available flowing space.
Solution. Rates are 150/30 = 5 h⁻¹ and 150/25 = 6 h⁻¹. At the same throughput, accounting for retained solids reduces the space assigned to flowing inventory.
3. Calibration strategy
A calibrated rod-mill model matches one product curve. A second test at higher rate gives a much finer product than stage scaling predicts. Which observation should be checked before fitting a second arbitrary stage count?
Solution. Check incoming size and ore changes, mill power and filling, and whether the test was at steady state. Use both surveys to test inverse-throughput scaling. A stage-count fit at each rate will reproduce the measured products without demonstrating a transferable throughput relation.
