Automated SAG Mill Speed Control Using

On-contact Vibration Sensors

Karl Gugel, Director Digital Control Lab & Professor University of Florida Karl.Gugel@DigitalControlLab.com

Luis J. Gutiérrez, Automation Specialist, TotalMetric, Chile Luis.Gutierrez@TotalMetric.cl

Juan Carlos Rodriguez, Process Control, Gold Corp Minera Peñasquito, México Juan.Rodriguez@GoldCorp.com

Slide 1

Abstract

• Low power RF vibration sensor has been placed directly on the SAG shell.

• Measure ball impact energy occurring at the liners in real-time.

• Fast Fourier Transforms (FFTs) show the vibration energy content and shape of the primary grinding region (toe). i.e. Polar Plot

• A new signal called Liner Damage Level (LDL) was created to automatically adjust mill speed.

• New rules for the expert system speed control: High LDL signal and low bearing pressure, reduce speed and prevent “metal on metal” operation.

• Two vibration systems installed and in operation at Gold Corp Peñasquitoon their two 2265 TPH SAG mils.

• Additions results/trends will be presented from other installations if time is remaining.

Slide 2

Process Description

• Gold Corp Peñasquito mine consists of two open pit mines, one concentrator plant with two grinding lines.

• Ore production: 120,000 DTPD/40,296,000 DTPY, 92% availability, 860,300 ounces of gold produced in 2015.

• Two grinding lines with two identical SAG mills: SAG1 and SAG2:

Slide 3

Size, [m (ft) (EGL) ] 11.6 x 6.1 / (38 x 20)

Operation: closed circuit w/screen classification and pebble crusher circulating load, [%]

23%

Percent of critical speed, operating range, [%] 73%

Ball loading, operating, [% v/v] 12-15%

Ball loading, max, [% v/v] 20%

Mill power transmission loss, estimated, [%] 5%

Feed slurry percent solids, [%] 65%

New feed rate operating average, [DTPH] 2265 DTPH

Maximum motor power, [kW] 19.000 kW

Process Control

• Mill Control: feed and speed. Feed is controlled via multiple bearing pressure load cells.

• Speed was initially tried to be controlled via array of microphones.

o The system was difficult to properly calibrate and remain tuned.

o The output information was difficult for the control room operators to understand and use.

o Low confidence in unit due to lack of correlation with other process signals.

• 60 day trial with vibration measurement system on SAG1 in mid 2015.

• Optimize mill speed and further understand the real-time operating conditions inside the mill.

Slide 4

Vibration System Components & Installation

Slide 5

Vibration System Installation Photos

Fixed Sensors(Inlet & Outlet)

Slide 6

Battery, Transmitter, T Sensor(top photo)

Main Unit (left)

T Sensor & Boom (right)

Operational Trends & Benefits

• New Information available to the operator & expert system: (3) fill level signals, Should/Toe Polar Plot & LDL Signal (auto-speed control).

• Available as 4-20 mA signals and as digital variables via standard Modbus.

Slide 7

Polar Plot energy plotted increasingtowards the center of the graph.

Mill rotating clockwise.

LDL is purple and at max value (100%).

Toe angle is in yellow.

Outlet fill level in blue, Inlet fill level inred and shell fill level is in green (0%).

Shell fill level was calibrated later.

Operational Trends (Start-up to 1.5 hours)

Slide 8

LDL = purple, Toe = yellow. Outlet fill level = blue, Inlet fill level = red, Shell fill level = green.

• Three plots show mill filling slowly over 1.5 hour period (left to right).

• Fill levels increase as expected and LDL begins to drop as mill becomes full.

• Toe is relatively constant due to no change in speed over this period of time.

Operational Trend: Control Room, 12 Hours

LDL - pink Shell FL - green Outlet FL - blue

Bearing Pressure (weight) - purple Speed - light brown

Slide 9

Manual Speed Reduction Based On LDL

Slide 10

2 Hour Trend:

LDL - pink

Shell FL - green

Outlet FL - blue

Weight - purple

Speed - light brown

Feed - White

Expert System LDL Integration

Slide 11

• Fuzzy Control: Three clusters with centers {17%, 40%, 60%}

• If weight < 2500 AND Density > 55% then the LDL is measured and allowed to reduce the speed. i.e. High condition: LDL > 35, Mid condition 15< LDL <35, Low condition LDL < 20. For a high condition, the speed is reduced proportionally from 0.05 to 0.1 RPMs every 4 minutes.

Expert System LDL Results for SAG1

Slide 12

SAG1 - April (No Speed Control) Results vs. September (LDL Speed Control) Results

• Shutdowns scheduled for end of March and August 2015 on SAG1. Above results are the month following each shut down.

• Average LDL signal decreases after auto-speed control is implemented in the expert system.

• Average speed for September is also significantly lower than that of April. However even with a lower speed monthly average, more TPH were generated in September with better Kwh/t results.

Summary & Future Work

• 2-3 minutes calibration time each for inlet, outlet and shell sensors.

• New correlations of vibration signals with traditional signals to observer how the the mill fills/empties and is affected by changes in speed.

• New expert system rules for minimizing and avoiding over-spilling events.

• MillSlicer Liner Damage Level (LDL) + mill weight (bearing pressure) to automatically reduce the mill speed in low weight/high LDL conditions.

• Short-term results with substantial savings in specific power reduction and reduced liner wear.

• Initial trial results on SAG1 led to a second system purchased for SAG2.

• Future studies to figure out how to use the fill level signals in an optimal manner and to predict liner wear based on ball trajectory (toe angle).

Slide 13

Additional Info – Teck Highland Valley Copper

New system installed February2016.

March data in Blue on the rightis the initial baseline LDL data.

Operators are informed of LDLsignal in April and LDL continuesto drop in April/May/June (notshown).

July LDL data in Pink on leftdistribution.

Process Engineers are workingon new Matlab auto controlrules presently to reduce millspeed using LDL.

Slide 14

Additional Information – Barrick Gold

Slide 15

Case 1. Empty Mill w/Excessive RPMs Case 2. Empty Mill w/Good RPMs Case 3. Full Mill & Cascade Grinding

mill rotation

135O

mill rotation

135O

mill rotation

135O

Ball trajectory

Bad Better Best!

liner strike energy liner strike energy

• Left plot shows the mill speed is too high resulting in “metal on metal” condition.• Center plot was the result of slowing the mill down but still unloaded. • Right most plot shows a full mill running at optimal speed (RPM). • Full mill shows a “compressed energy” profile in the shoulder region.

Additional Information – KGHM (RNMC)

Blockage Ends

Expert System increases mill speed to reduce the blockage.ES overshoots the end of the blockage and so mill empties.Speed should have been increased at the blockage start.

Blockage starts when Inlet and Outlet split.

Bright Blue – Feed Olive – RPM Pink – Pressure Green – Inlet Blue – Outlet Yellow - Shell

- At 10 min into the trend, inlet & outlet signals split, bearing pressure rises, blockage.

- 40 min later w/speed increase, blockage frees, inlet & outlet track once again.

Slide 16 - The End!