APPLICATION OF ROUTINE

HIT ORE HARDNESS TESTING

AT THE GRINDING CIRCUIT,
MEADOWBANK GOLD MINE

Karl LEETMAA, Senior

Metallurgist AEM
Yanick BERGERON, Project
Metallurgist AEM
Toni KOJOVIC, Managing
Director SimSAGe P/L
Project Metallurgist
AusIMM MetSoc
AGNICOPlant Operators'
EAGLE MINES
Forum, 28th August, 2018
Summary of Presentation

• Introduction

• Meadowbank Process Plant

• Hardness Index Tester (HIT)

• HIT Evaluation Program

• Results

2
Introduction

• Located in the Kivalliq region of Nunavut in Canada

• Owner: Agnico Eagle Mines Ltd. (AEM)
• Throughput: 11,000 tpd

3
Process Plant Description

445 – 515 tph throughput depending on ore type, P80 75-110 microns
95 % Availability with 90.5 to 95.5 % recovery
4
What is the HIT?

• HIT means Hardness Index Tester

• Portable device performing rapid rock hardness determination

• HIT generates quantitative estimates of Axb and BWi

• Allows operational, on the spot determination of rock hardness variability

• Ideal for routine testing of mill feed samples to monitor performance

/adjust mill settings for optimum performance or geometallurgical program

5
Hardness Index Tester Highlights

• Initially developed for rapid low cost Axb

estimation

• Industrial trials demonstrated the

possibility of using HIT for Bond BWi
estimation

• Calibration against standard tests may

be required, especially Bond BMWi

• Quality control protocols included

• Online software access for Axb and

Bond BWi calculations Prototype V2
Patent Pending
62/241,852, PCT/I B2016/001591

6
HIT Evaluation at Meadowbank

• Program initiated by AEM Meadowbank in late 2017

• Can routine ore hardness testing, using SAG feed samples, assist in
forecasting the SAB grinding circuit performance?

• Daily samples were collected over a two-month period

• Submitted for HIT testing at CSD laboratory to provide estimates of both the
competency (Axb) and grindability (Bond BWi) hardness.

• Program extended until end of March 2018, with a further 107 days of
sampling

• Meadowbank deposit is almost depleted and AEM is planning to open

Amaruq mine and process ore in the Meadowbank plant.

• A geometallurgical program, including hardness characterization

using the HIT, is ongoing.

7
Meadowbank and Amaruq Projects

8
Data Analysis Approach

• AEM provided a summary of the triplicate HIT Axb and duplicate HIT
Bond Wi results in Excel (for each apron feeder sample collected)
• Daily operating information from Meadowbank
Ore Zone Crusher shut Mill Down time Availability (%)
Vault (%) down (hrs) (hrs)
Throughput Tonnage SAG Power Draw BM Power Draw
(TPH) (tpd) (kW) (kW)
SAG F80 (inch) SAG %-1inch SAG T80 Product P80
WipFrag WipFrag (if reliable) (cyc. OF)

• Multiple least-squares regression analysis of SAG and BM performance

• SAG specific energy S.E. kWh/t and grind size P80 (microns)
• Inputs included HIT hardness parameters and operating variables
expected to affect the mill performance.

9
Challenges

• HIT testing and feed sizing was conducted on material from a ~10kg
sample of SAG feed (mix of grab samples and composites).
• Inconsistent sample taking due to cross shift effect.
• Manual sampling of coarse material challenging, despite lower F80.

• Operating data reflect the daily average performance/statistics, which

may not align with the feed sample collected for HIT testing and size
analysis.

• Daily performance may not reflect the optimum/maximum performance,

for example, packing issues in the SAG due to new lifter liners.

• HIT Bond Wi estimates are derived from a default calibration model,

which may need to be updated to reflect more accurately the grindability
of Meadowbank ore types.

10
Initial Modelling Results

• Strong correlation confirmed the well-known expectation that feed size

and hardness does affect the SAB grinding performance
• Improved sampling strategy may improve overall accuracy
SAG S.E. ‐ 7Oct to 30Nov17
8.5

8.0

7.5
SAG S.E.  (kWh/t)

7.0

6.5

6.0
Calc
5.5
Actual
BM P80 ‐ 7Oct to 30Nov17
135
5.0
10/02/2017 10/12/2017 10/22/2017 11/01/2017 11/11/2017 11/21/2017
130 12/01/2017 12/11/2017

125
BM P80 (microns)

120

115

110

105
Calc
100
Actual
11 95
10/02/2017 10/12/2017 10/22/2017 11/01/2017 11/11/2017 11/21/2017 12/01/2017 12/11/2017
Validation Results

• Use model developed from initial 53 days of monitoring to forecast

performance for remaining 107 days not used in model calibration.
• Clearly not as accurate as the calibration, but still within 5% relative error.
• Results very encouraging, the model tracking a major change in SAG
Specific Power and BM P80 in mid-February 2018
BM P80 ‐
SAG S.E. ‐7Oct17 to 31Mar18
7Oct17 to 31Mar18
135
10.0
130 Calc
9.5 Calc Actual
125
9.0 Actual
120
BM P80 (microns)

8.5
(kWh/t)

115
8.0
110
7.5
SAG S.E. 

105
7.0
100
6.5
95
90
6.0
85
5.5
calibration forecast
80
5.0
09/22/2017
09/22/2017 10/12/2017
10/12/2017 11/01/2017
11/01/2017 11/21/2017
11/21/2017 12/11/2017
12/11/2017 12/31/2017
12/31/2017 01/20/2018
01/20/2018 02/09/2018
02/09/2018 03/01/2018
03/01/2018 03/21/2018
03/21/2018 04/10/2018
04/10/2018

12
Conclusions

• Variation in Axb was more significant than Bond Wi, 27 to 45 for Axb and
12.9 to 15.4 kWh/t for Bond Wi. – values align with range of standard test
results from Meadowbank deposit.

• Observed variation in the grinding circuit performance is strongly

correlated to the HIT ore hardness parameters.

• Other variables affecting the SAG and Ball Mill circuit performance are
feed size, specifically the %-1", and ore SG as expected.

• Data from the HIT test was used to justify throughput issues in Q1 2018
• Site has capacity to conduct BWi tests, but no information on Axb
• Meadowbank wants to re-work the models to predict the TPH for a given
target P80.
• More adapted to the needs: maximizing TPH for a set grind target
• Method to verify if the grinding circuit is performing optimally

13
Future Work

• Meadowbank has decided to extend HIT evaluation until end of 2018 as a

tool to benchmark the performance of the grinding circuit.

• SimSAGe working to update the HIT Bond Wi calibration model once

sufficient actual Bond Wi test results are available on the same samples.
• Should provide a more reliable model in both cases, calibration and forecast.
• Validation using plant performance data around the time of sampling

• Use the new model to verify the actual TPH versus the predicted value as a
measure of the circuit efficiency.
• Justification of performance when faced with different feed types
• Indicator that circuit performance is not optimal, requiring investigation
• Implement use of the HIT test to map the hardness variation in the future
Amaruq deposit and develop an ore blending strategy based on the results.

14
 agnico-eagle.com

15 simsage.com.au