Austin SAG model

History

The SAG mill model by proposed by Leonard Austin (1990) was largely based on modifications of earlier tumbling mill models by Hogg & Fuerstenau and F. Bond. The model uses a kinetic-potential energy balance to describe the power draw of a mill charge. Many geometric components of the earlier models were fit to empirical relationships measured by Austin.

Model Form

The power draw model for a SAG mill cylinder of the following form:

$P = K D^{2.5} L ( 1 - AJ_{total}) \left [ (1-\epsilon_{B}) \left( \frac{\rho_{solids}}{w_{C}} \right ) J_{total} + 0.6 J_{balls} \left( \rho_{balls} - \frac{\rho_{solids}}{w_C} \right) \right ] \phi_{C} \left( 1 - \frac{0.1}{2^{9-10\phi_C}} \right)$

Where:

• P is the power evolved at the mill shell, kW
• K and A are empirical fitting factors (use 10.6 and 1.03, respectively)
• D is the mill effective diameter (inside the effective liner thickness), m
• L is the mill effective grinding length (also referred to as the 'belly length), m
• J_total is the mill total volumetric filling as a fraction (eg. 0.30 for 30%)
• ε_B is the porosity of the rock and ball load (use 0.3)
• w_C is the charge %solids, fraction by weight (use 0.80 ^{Doll, 2013})
• J_balls is the mill volumetric filling of balls as a fraction (eg. 0.10 for 10%)
• ρ_X is the density of component X, t/m³
• φ_C is the mill speed as a fraction of critical (eg. 0.75 for 75% of critical)

Mill cone angles

The model supports flat-ended mills (cone angle of 0°) or a cone angle of 15°. Any other value entered for the cone angle will be treated as 15°.

To account for cone ends of mills, an allowance of 5% is used ^{Doll, 2013} instead of the formula proposed by Austin.

Percent solids

The model is run with a fixed 80% solids by weight ^{Doll, 2013}. This is because the form of the equation proposed by Austin appears to have the %solids term in the wrong place (denominator of the expression rather than the numerator where other models put it). The calibration presented at IMPC 2016 suggests the fixed %solids term is reasonably valid over the range of 60% to 80% solids.

Model Inputs

the Austin model requests the following mill data:

• mill nominal diameter inside shell
• mill effective grinding length
• ball load as vol percentage (eg. 10)
• ball density
• total charge load as vol percentage (eg. 30)
• pulp percent solids
• mill speed
• liner effective thickness
• cone end angle (only values 0-5 or >5 matter, minor angle changes have no effect)

Motor characteristics are also requested:

• Motor rated power (output shaft)
• Mechanical efficiency of downstream drive (pinions, gearboxes)
• Motor efficiency and any other efficiency factors to the DCS measurement position in the network.
• Motor rated speed, in units of mill RPM (not motor RPM, must multiply by gear ratios).
• Quantity of pinions (and motors)

Model output

The Austin model outputs the mechanical power of the cylindrical section of a tumbling mill. The software adds an extra 5% for cone ends for situations where the mill has a cone angle greater than 1%. The default mill cone angle is 15%, so entering an angle of 0° will trigger the default of 15% (so use a fake cone angle of 0.5° for flat-ended mills).

Model calibration

Alex Doll did a model calibration check of several SAG mill power draw models (including the Austin model) in Procemin 2013 and IMPC 2016. The model deviation from measurements has a standard deviation of 9% and average/median of 1.1%/1.5% (meaning there is a minor bias high).

These works are available from:

Technical paper for Procemin 2013

Technical paper for IMPC 2016 (supersedes the Procemin 2013 paper)

Data set for IMPC 2016