Benchmarking: Specific Energy Consumption Models

Benchmarking: Amelunxen SGI - Agnico Eagle Laronde

Starkey, J., Robitaille, J., Cousin, P., Jordan, J. and Kosick, G., Design of the Agnico-Eagle Laronde Division SAG mill. Proceedings of SAG 2001, pages III-165 to III-178.

• Survey conducted for six months after start-up
• Specific energy values corrected to "mill shell" basis (motor input basis given in the reference)

Result for default Amelunxen SGI SAB model conditions:

	Esag	Epeb	Ebm	Etotal	t/h
Predicted	6.5	-	6.9	13.4	196
Survey	6.1	-	6.4	12.5	210
Difference	0.4	-	0.5	0.9	14
Difference	model 7% high	-	model 8% high	model 7% high	model 7% low

Show details of benchmarking

Benchmarking: Bond/Barratt - Cadia East (underground)

Engelhardt, D., Robertson, J., Lane, G., Powwel, M.S. and Griffin, P., Cadia Expansion - From open pit to block cave and beyond. Proceedings of MetSoc 2012.

• Design criteria and plant trial of underground Cadia East ore
• Ore was blasted underground and had F₈₀ = 80 mm on surface
• Work index appears to have been determined on a non-standard rod mill apparatus. The rod mill Wi is probably invalid.
• This ore breaks the Bond/Barratt models, the combination of work index values is not a good fit for Barratt's original calibration of his model.

Result for default Optimized Bond/Barratt SABC model conditions:

	ESAG	Eball	Etotal	t/h
Model	12.2 kWh/t	15.1 kWh/t	27.7 kWh/t	1294 t/h
Measured	10.6 kWh/t	13.2 kWh/t	23.8 kWh/t	1482 t/h
Difference	1.6 kWh/t	1.9 kWh/t	3.9 kWh/t	-188 t/h
Difference	15%	14%	16%	-13%

Show details of benchmarking

Benchmarking: Bond/Barratt - Cadia

Dunne, R., Morrell, S., Lane, G., Valery, W. and Hart, S., Design of the 40 foot diameter SAG mill installed at the Cadia gold copper mine. Proceedings of SAG 2001, pages I-43 to I-58.

Lane, G., Foggiatto, B. and Bueno, M, Power-based comminution calculations using Ausgrind. Proceedings of Procemin 2013, Chapter 2, paper 2.

• Survey conducted shortly after start-up

Result for default Optimized Bond/Barratt SABC model conditions:

	Esag	Epeb	Ebm	Etotal	t/h
Predicted	8.6	0.3	8.5	17.4	2000
Survey	8.6	-	8	16.6	2065
Difference	0	0.3	0.5	0.8	-65
Difference	0%	-	6%	5%	-3%

Show details of benchmarking

Benchmarking: Bond/Barratt SABC Circuit Specific Energy Consumption - Copper Mountain

• Morrison, R., Current Plant Conditions at Copper Mountain. Presentation to the BC/Yukon Branch Canadian Mineral Processors, November 29, 2012; Vancouver, Canada.
• van de Vijfeijken, M., Filidore, A., Walbert, M. and Marks, A., Copper Mountain: Overview on the Grinding Mills and their Dual Pinion Mill Drives. Proceedings of the SAG 2011 Conference, September 25-28, 2011; Vancouver, Canada.
• Marks, A., Sams, C. and Major, K., Grinding Circuit Design for Similco Mines. Proceedings of the SAG 2011 Conference, September 25-28, 2011; Vancouver, Canada.

Result for default Optimized Bond/Barratt model conditions:

	Tonnage
Model	1455 t/h
Measured	1600 t/h
Difference	145 t/h
Difference	9.5%

Show details of benchmarking

Benchmarking: Bond/Barratt SAG Mill Specific Energy Consumption - Detour Lake

• J. Torrealba-Vargas, J.-F. Dupont, J. McMullen, A. Allaire and R. Welyhorsky, The successful development of the detour lake grinding circuit: from testwork to production. Proceedings of the SAG 2015 Conference, September 2015, Vancouver, Canada, Paper 38.

The mill is in the final stages of ramp-up, and has not reached its ultimate capacity. If we use the April/May 2015 values from Figure 5, the results look like:

• Etotal : model 7% low
• t/h : model 13% high

The throughput is likely to continue to increase as the ramp-up continues, so the difference between the model and the actual plant is likely to reduce.

Show details of benchmarking

Benchmarking: Bond/Barratt SAG Mill Specific Energy Consumption - Esperanza

• Villanueva, F. and Soto, L., SEC and the impact on a mills selection for DMC project. Proceedings of the XXVII International Mineral Processing Congress, October 2014, Santiago, Chile. C14-23.

Paper describes modelling of two expansion cases (Esperanza Sur and Encuentro). Some operating data for the current Esperanza pit & plant are provided as they were used to tune the comminution models used for the expansion cases.

	Average ore
Measured SAG specific energy consumption, kWh/t	5.5
Predicted SAG specific energy consumption, kWh/t	5.9
Difference, kWh/t	0.4
Difference, %	model predicts 7% high

Show details of benchmarking

Benchmarking: Bond/Barratt SABC Circuit Specific Energy Consumption - Fort Knox

• Magnuson, R.; Hallow, J.; Mosher, J.; Major, K., The Fort Knox Mill: Design, Commissioning and Operation. Proceedings of the SAG 2001 Conference, Vancouver, Canada.

Optimized Bond/Barratt SABC circuit (10% Essbm calibration factor).

Result for default model conditions:

	Etotal	WiO		Tonnage
Model	11.43	15.90	kWh/t	1,607	t/h
Measured	10.50	14.61	kWh/t	1,733	t/h
Difference	0.93	1.29	kWh/t	126	t/h
Difference	8.9%	8.8%		7.3%

Model predicts 8.9% harder than survey resulting in predicted throughput 7.3% lower than survey.

Show details of benchmarking

Benchmarking: Bond/Barratt SABC Circuit Specific Energy Consumption - Kanowna Belle

Lunt, D.J., Thompson, A. and Ritchie, I. The Design and Operation of the Kanowna Belle Milling Circuit, SAG 1996, Pages 81-96.

	Survey	Optimized Model	Difference
ESAG	12.55 kWh/t	13.27 kWh/t	+5.7%
Eball	9.70 kWh/t	10.24 kWh/t	+5.6%
Epeb	0.99 kWh/t	1.04 kWh/t	+5.2%
Etotal	23.24 kWh/t	24.55 kWh/t	+5.6%

Two possible circuit models were tested, and the Optimized Bond/Barratt SABC model better matches the survey than the Raw Bond/Barratt model.

Show details of benchmarking

Benchmarking: Bond/Barratt SABC Circuit Specific Energy Consumption - Meadowbank

• Muteb, P. & Allaire, J., Meadowbank Mine Process Plant Throughput Increase, Proceedings of the Canadian Mineral Processors Annual General Meeting, Ottawa, Canada, January 2013.

Paper describes a "sick" SAG mill and the changes made to "bring it to health". The "healthy" mill conditions benchmark as follows:

• Actual SAG/ball motor powers (at shell): 3,168 kW / 4,105 kW
• Actual daily average throughput: 500 tonnes/hour
• Predicted SAG/ball motor powers (at shell): 3,096 kW / 4,182 kW
• Predicted nominal throughput: 500 tonnes/hour (0% difference)

	SAG	Ball Mill	total
Measured specific energy consumption, kWh/t	6.34	8.21	14.55
Predicted specific energy consumption, kWh/t	6.19	8.36	14.55
Difference, kWh/t	-0.15	0.15	0.00
Difference, %	-2.3%	1.8%	0.0%

Show details of benchmarking

Benchmarking: Amelunxen SGI SABC Circuit Specific Energy Consumption - Macraes

Barns, K., Lane, G., Osten, K. & Scagliotta, N., Benchmarking Energy Efficiency - A Case Study at Macraes Gold Mine, Proceedings of the AusIMM MetPlant conference, Perth, Australia, September 2004.

		Survey 5	Survey 4	Survey 3	Survey 2	Survey 1	Overall Average
SAG specific power		7.1%	7.6%	23.4%	5.0%	9.4%	10.5%
ball mill specific power		7.1%	7.7%	16.6%	4.9%	11.4%	9.6%
total specific power		7.1%	7.7%	19.0%	5.0%	10.8%	9.9%
CFball		-17.5%	16.9%	0.5%	10.6%	8.3%	3.8%

The overall specific energy consumption predictions of the model are conservative by about 10%. The CF_ball predictions are within 4% overall, but can be wildly different on any particular sample. Circuit feed sizes were varied during the surveys, ranging between 51 and 107 mm; the SGI equation doesn't have a explicit F₈₀ term, so feed sizes may be confusing the method.

Show details of benchmarking

Benchmarking: Bond/Barratt SABC Circuit Specific Energy Consumption - Macraes

Barns, K., Lane, G., Osten, K. & Scagliotta, N., Benchmarking Energy Efficiency - A Case Study at Macraes Gold Mine, Proceedings of the AusIMM MetPlant conference, Perth, Australia, September 2004.

Difference between model results and plant surveys:

		Survey 5	Survey 4	Survey 3	Survey 2	Survey 1	Overall Average
SAG specific power		-5.3%	-5.0%	6.3%	0.5%	2.2%	-0.2%
ball mill specific power		-6.3%	-4.7%	3.6%	1.9%	3.2%	-0.5%
total specific power		-6.8%	-3.8%	5.4%	1.4%	2.9%	-0.2%

Show details of benchmarking

Benchmarking: Bond/Barratt ABC Circuit Specific Energy Consumption - Santa Rita

Latchireddi, S. & Faria, E., Achievement of High Energy Efficiency in Grinding Mills at Santa Rita, Proceedings of the Canadian Mineral Processors Annual General Meeting, Ottawa, Canada, January 2013.

Faria, E. & Latchireddi, S., Commissioning and Operation of Milling Circuit at Santa Rita Nickel Operation, Paper #137: Proceedings of the International Autogenous Grinding, Semiautogenous Grinding and High Pressure Grinding Roll Technology Conference, Vancouver, Canada, September 2011.

	FAG	BM	Pebble Crusher	total
Measured specific energy consumption, kWh/t	9.55	7.28	0.34	17.18
Predicted specific energy consumption, kWh/t	9.78	8.33	0.39	18.50
Difference, kWh/t	0.23	1.05	0.05	1.32
Difference, %	2.4%	14.4%	14.7%	7.7%

Show details of benchmarking

Benchmarking: Bond/Barratt SAB Circuit Specific Energy Consumption - Selbaie

Duval, L. and Wood, K., Testing, Design and Operation of SAG Circuit at Les Mines Selbaie, Proceedings of the SAG 1989 Conference, Vancouver, Canada, September 1989.

Wood, K. and Duval, L., Mill Expansion at Les Mines Selbaie, Proceedings of the 25th Anniversary Issue of the Canadian Mineral Processors, Ottawa, Canada (Paper № 8), January 1987.

• SAB circuit
• SAG feed F₈₀ = 117 mm
• cyclone overflow, P₈₀ = 45 µm

	SAG	BM	total
Predicted specific energy consumption, kWh/t	11.99	12.37	24.37
Predicted operating work index, metric	20.3	14.2	16.7
Measured operating work index, metric	20.0	13.8	16.0
Difference, metric Wi units	0.3	0.4	0.6
Difference, %	2%	3%	4%

Show details of benchmarking

Benchmarking: Bond/Rowland SSBM Circuit Specific Energy Consumption - Boddington

The Bond/Rowland SSBM model was fit to the observed operation of the Boddington HPGR circuit. The fitting parameters are:

• E_ssbm calibration factor: (-0.13)
• Mechanical efficiency of HPGR crushers: 0.28.

Because the model is specifically fit to the Boddington data, it doesn't make any meaningful throughput predictions, but the following predictions are available:

• Ball mill operating work index reduction versus laboratory: 5% (microcracking/phantom cyclone effect)
• Secondary crusher WiO is 18.1 kWh/tonne (versus laboratory determination 27.7 kWh/t).

Show details of benchmarking

Benchmarking: Bond/Rowland SSBM Circuit Specific Energy Consumption - Tropicana

The Bond/Rowland SSBM model was fit to the observed operation of the Tropicana HPGR circuit. The fitting parameters are:

• E_ssbm calibration factor: (0%)
• Mechanical efficiency of HPGR crushers: 0.44.

Show details of benchmarking

Benchmarking: Morrell Mih model Circuit Specific Energy Consumption - Tropicana

The Morrell Mi model (SMC test & recalibrated ball mill work index) was compared to the observed operation of the Tropicana HPGR circuit:

	Model Prediction	Survey Actual	Difference
EsecCr	0.52 kWh/t	0.40 kWh/t	model 30% high
Ehpgr	2.83 kWh/t	2.60 kWh/t	model 9% high
Eball	18.36 kWh/t	16.77 kWh/t	model 9% high
Etotal	21.70 kWh/t	19.77 kWh/t	model 10% high

The Morrell model is about 10% conservative versus the survey results.

The throughput determined by the ball mill treatment rate is 710 t/h, versus observed 734 t/h (model is 3% conservative).

Show details of benchmarking

Benchmarking: Bond/Barratt Single Stage SAG model Circuit Specific Energy Consumption - Yanacocha

Two surveys were published, the first survey was noted that "the circuit was not operating efficiently", so the model predictions are expected to be optimistic.

	Survey Actual	Model Prediction	Difference	Comment
Survey 1 EASAG	19.8 kWh/t	16.4 kWh/t	model 20% low	Ignore survey, mill operating poorly
Survey 2 EASAG	18.0 kWh/t	16.3 kWh/t	model 10% low	Survey probably OK

The first survey has significantly higher energy consumption than the model predicts. The author of the reference paper says that the mill was not operating well, so it is reasonable to ignore this survey and not use it as a valid "benchmark" as we are only interested in benchmarking against circuits that are operating "well".

The second survey is operating better, and can be assumed valid basis for benchmarking.

Show details of benchmarking

Benchmarking: Morrell Mi Single Stage SAG model Circuit Specific Energy Consumption - Yanacocha

The same two surveys as the Bond/Barratt model (above) give the following:

	Survey Actual	Model Prediction	Difference	Comment
Survey 1 EASAG	19.8 kWh/t	12.7 kWh/t	model 56% low	Ignore survey, mill operating poorly
Survey 2 EASAG	18.0 kWh/t	12.5 kWh/t	model 43% low	Significant difference

Both these predictions are significantly different from the surveys. Either the circuit is performing much worse than the Bond/Barratt model predicts or else there is something in Yanacocha that is confusing the Mi method.

Show details of benchmarking

Benchmarking: More Single-stage SAG surveys

• See the Procemin 2017 paper Power-based modelling of single-stage AG and SAG mill circuits, A. Doll & M. Becerra PDF download

	Barratt model	Morrell model	Amelunxen model	El Soldado model
Qty of parameters:	3	2	2	1
El Soldado	5%	-1%	-4%	-5%
Palabora	3%	-28%	-21%	-34%
Tarkwa	-1%	6%	11%	125%
Degrussa	ESAG: 28% Etotal: 17%	ESAG: 4% Etotal: 2%	ESAG: 7% Etotal: 15%	ESAG: 13%
Yanacocha	-18%	-33%	-25%	-40%

Benchmarking: Private surveys

Private surveys that do not include details, but the difference between plant performance and model predictions are provided below:

• Project 0082, using Optimized Bond-Barratt model.

1. SAB survey: E_total: actual = 6.72 kWh/t, predicted = 6.78 kWh/t, difference 0.9%
2. ABC survey: E_total: actual = 7.75 kWh/t, predicted = 8.14 kWh/t, difference 4.8%

• Project 0104, SABC-A survey

1. Optimized Bond-Barratt model : E_total: actual = 20.8 kWh/t, predicted = 20.7 kWh/t, difference 0.0%
2. Morrell Mi (SMC) model : E_total: actual = 20.8 kWh/t, predicted = 20.2 kWh/t, difference 2.9%

• Project 0149, SAB survey (BC Cu porphyry, without Josefin correction)

1. Optimized Bond-Barratt model : E_total: actual = 6.42 kWh/t, predicted = 7.04 kWh/t, difference 9.7%
2. Morrell Mi (SMC) model : E_total: actual = 6.42 kWh/t, predicted = 7.27 kWh/t, difference 13.2%