A ball mill is a rotating steel drum, partly filled with steel balls, that grinds ore to a fine product. Larger mills draw more power and grind more ore per hour, and sizing a mill means determining the power needed to grind a given ore, at a given tonnage, from a given feed size to a given product size.
Fred Bond addressed this in the 1950s with a laboratory test and an equation. The test gives the ore's work index, a measure of its resistance to grinding, and the equation converts the work index to kilowatt-hours per tonne for a standard mill. Bond and later Carl Rowland added correction factors for mills that differ from the standard. The corrected energy multiplied by the tonnage gives the required power, from which the mill size is selected.
The Bond–Rowland method remains the most widely used means of sizing ball mills and of checking whether an existing mill is performing as expected. This article works through the method from work index to mill dimensions with a single example, explaining each efficiency factor. It builds on the articles on the Bond ball mill test and operating work index.
Bond specific energy
where
- specific grinding energy for Bond's reference mill: wet, closed circuit, 2.44 m diameter (kWh/t)
- Bond work index of the ore, metric tonne basis (kWh/t)
- ,
- 80% passing sizes of the mill feed and of the circuit product (µm)
- …
- Rowland efficiency factors for conditions different from the reference mill (1 when they do not apply)
- ,
- mill power at the pinion (kW) and dry solids feed rate (t/h)
Worked example: an ore with a ball mill work index of 14.3 kWh/t is fed to a ball mill at F80 = 8,000 µm from a closed-circuit crushing plant. The cyclone overflow target is P80 = 106 µm, and the mill runs wet in closed circuit with cyclones at 150 t/h.
W = 10 × 14.3 × (1/√106 − 1/√8,000) = 143 × (0.0971 − 0.0112) = 12.29 kWh/t.
This is the energy that Bond's reference mill would use, and it is corrected with Rowland's factors.
Rowland's efficiency factors
| Factor | Correction | Application | Value in the example |
|---|---|---|---|
| EF1 | Dry grinding | Dry grinding: 1.3 | 1.0 (wet) |
| EF2 | Open circuit | Open-circuit ball milling, from the table below | 1.0 (closed circuit) |
| EF3 | Mill diameter | All mills; mills larger than 2.44 m are more efficient | 0.914 |
| EF4 | Oversized feed | Feed coarser than the optimum feed size | 1.079 |
| EF5 | Fine grinding | P80 finer than 75 µm | 1.0 |
| EF6 | Rod mill reduction ratio | Rod mills only | n/a |
| EF7 | Low reduction ratio | Ball mills with F80 ÷ P80 below 6 | 1.0 |
| EF8 | Rod milling | Rod mills only | n/a |
where
- mill diameter inside the liners (m)
- work index on the short ton basis: metric value ÷ 1.102 (kWh/st)
- optimum feed size for a ball mill (µm); for a rod mill the constant is 16,000
- reduction ratio, F80 ÷ P80
EF2: open circuit
A ball mill in open circuit, without a classifier to return coarse particles, must over-grind to reduce its coarsest particles to the required size. The penalty depends on how closely the top size is controlled, expressed as the percentage passing a reference (control) size:
| % passing control size | 50 | 60 | 70 | 80 | 90 | 92 | 95 | 98 |
|---|---|---|---|---|---|---|---|---|
| EF2 | 1.035 | 1.05 | 1.10 | 1.20 | 1.40 | 1.46 | 1.57 | 1.70 |
EF3: diameter
Larger mills grind more efficiently, up to a limit. For a 3.0 m mill, EF3 = (2.44 ÷ 3.0)0.2 = 0.96. For mills of 3.81 m and larger, EF3 is fixed at 0.914. The mill in this example proves to be larger than 3.81 m, so EF3 = 0.914. Some designers do not take this credit and treat it as an undeclared safety margin; the design basis should state which approach is used.
EF4: oversized feed
A ball mill fed with coarse rock wastes energy breaking it with balls that are too small. Rowland's optimum ball mill feed size uses the work index on the short ton basis: 14.3 ÷ 1.102 = 12.97 kWh/st. The optimum feed size is Fo = 4,000 × √(13 ÷ 12.97) = 4,004 µm. The feed, at 8,000 µm, is coarser, so EF4 applies with Rr = 8,000 ÷ 106 = 75.5:
EF4 = [75.5 + (12.97 − 7) × (8,000 − 4,004) ÷ 4,004] ÷ 75.5 = 1.079.
EF5 and EF7: very fine grinding and low reduction ratio
Regrind mills often require both factors. For a regrind to P80 = 53 µm, EF5 = (53 + 10.3) ÷ (1.145 × 53) = 1.04. For a reduction ratio of 4, EF7 = [2 × (4 − 1.35) + 0.26] ÷ [2 × (4 − 1.35)] = 1.05.
EF6 and EF8: rod mills
Rod mills are sized in the same way, using the rod mill work index from Bond's rod mill test:
where
- closing sieve of the rod mill test (µm)
- rod mill grindability: net grams of undersize per revolution
- converts kWh per short ton to kWh per metric tonne
EF6 corrects for a reduction ratio far from the rod mill's optimum, Rro = 8 + 5 × (rod length ÷ mill diameter): EF6 = 1 + (Rr − Rro)² ÷ 150 when Rr differs from Rro by more than 2. EF8 is 1.4 for a rod mill alone fed from open-circuit crushing, 1.2 for a rod mill alone fed from closed-circuit crushing or for the rod mill in a rod-ball circuit fed from open-circuit crushing, and 1.0 otherwise.
Design energy and pinion power
Wdesign = 12.29 × 1.0 × 1.0 × 0.914 × 1.079 = 12.12 kWh/t.
P = 12.12 × 150 = 1,818 kW at the pinion.
The motor must also supply the drive-train losses in the gearbox, motor and any variable-speed drive. Pinion power is divided by the drive efficiency from the supplier, and the next standard motor size is selected with whatever margin the design criteria require.
Mill diameter and length
A mill power model relates power to mill size, charge filling and speed. Using the Hogg and Fuerstenau model in the mill power calculator, with 35% charge filling, 75% of critical speed, an apparent charge density of 5.4 t/m³ and a length of 1.5 diameters:
The mill that draws 1,818 kW is about 4.24 m in diameter and 6.36 m long, inside the liners. It is larger than 3.81 m, so the EF3 of 0.914 used above holds. Had the mill been smaller, EF3 would increase and the calculation would be repeated with the new value.
Suppliers build mills in standard sizes and apply their own power models, so the final dimensions come from the supplier. The Bond–Rowland calculation defines the requirement and provides a check on the supplier's offer.
Design checks
The mill should be sized for the hardness it will treat most of the time, with a check on the hardest ore type, rather than for the average of all samples. P80 is the circuit product, the cyclone overflow, not the mill discharge.
When a ball mill follows a SAG mill, the transfer size from the SAG mill is fine and variable. Methods such as Morrell's, in the SABC calculator, are designed for these circuits.
For an existing plant, the calculation is run in reverse: measured power and tonnage give the operating work index, which is compared with the laboratory value corrected by the same factors.
Glossary
- Work index, Wi
- A measure of how hard an ore is to grind, from Bond's laboratory test (kWh/t).
- F80, P80
- The sizes that 80% of the feed and of the product pass (µm).
- Efficiency factor
- A multiplier that corrects Bond's energy for conditions different from his reference mill.
- Pinion power
- The power delivered to the mill shell through the pinion and girth gear.
- Closed circuit
- A mill with a classifier, usually cyclones, that returns coarse particles for more grinding.
- Reduction ratio
- Feed size divided by product size, F80 ÷ P80.
- Charge filling
- The fraction of the mill volume occupied by the balls and the ore and water among them.
- Critical speed
- The speed at which the outer balls would stick to the shell; mills run at about 65 to 80% of it.
Frequently asked questions
How is ball mill power calculated from the work index?
Calculate Bond's energy W = 10·Wi·(1/√P80 − 1/√F80), multiply by the applicable efficiency factors, then multiply by the dry tonnes per hour to obtain kilowatts at the pinion.
What are Rowland's efficiency factors?
Eight multipliers, EF1 to EF8, that correct Bond's energy for dry grinding, open circuit, mill diameter, oversized feed, fine grinding, rod mill reduction ratio, low reduction ratio and rod milling.
Why does the diameter factor make large mills more efficient?
Bond's data showed larger mills grinding more efficiently per kilowatt. Rowland expressed this as (2.44/D)^0.2, capped at 0.914 for mills of 3.81 m and larger.
Should the metric or short ton work index be used?
Bond's energy equation uses the work index on the same basis as the result: a metric Wi gives kWh per metric tonne. The EF4 formula was written for the short ton value, so the metric work index is divided by 1.102 in that formula.
Discuss your plant with a metallurgist
Tiegel is a Perth metallurgy company. We help with testwork, plant optimisation, metal accounting and investigations, and we build the Tiegel Suite software.
Talk to a metallurgistReferences and further reading
- Bond, F.C. (1961). Crushing and grinding calculations. British Chemical Engineering, 6, 378–385 and 543–548.
- Rowland, C.A. and Kjos, D.M. (1980). Rod and ball mills. In Mular, A.L. and Bhappu, R.B. (eds), Mineral Processing Plant Design, 2nd edition. Society for Mining, Metallurgy and Exploration.
- Global Mining Guidelines Group (2021). Determining the Bond efficiency of industrial grinding circuits. Link
- Met Dynamics. Comminution circuit specific energy (Bond). Link
- Hogg, R. and Fuerstenau, D.W. (1972). Power relationships for tumbling mills. Transactions SME-AIME, 252.
Worked examples and charts marked illustrative use invented figures, not data from any operation.