design and implementation of advanced automatic control ... · "design and implementation of...
TRANSCRIPT
"Design and Implementation of Advanced Automatic Control Strategy Based on Dynamic Models for High
Capacity SAG Mill"
Iván Yutronic (Collahuasi)Rodrigo Toro (Honeywell Chile)
Who are we?• Collahuasi is operated by a joint
venture Xstrata (44%), Anglo American (44%), and Mitsui & Co. Ltd (12%).
• Collahuasi is located in the Andean plateau of northern Chile's Tarapacá Region.
• 210 km southeast of the city of Iquique average altitude of 4300 m.a.s.l.
• Reserves – 5115 Mt at 0.81% Cu y 0.023% Mo.
• Employees Approximately 3,600 people.
• 400,000 t.p.y. of copper in concentrates, 60,000 t.p.y. of copper cathode .
CollahuasiCollahuasi
CS - 1011
CS - 1012
PP- 1023 PP- 1024
PP- 1021 PP- 1022
BM 3
SAG 1
BM 1013
SAG 1011
BM 4SAG 2
Line #1
Line #2
Line #3
TO FLOTATION
DI 01
BM 1012
DI 1051
DI 02
Fe - 003
Fe - 004
Fe - 005
Fe - 006
Fe - 007
Fe - 008
Fe - 1031
Fe - 1032
Fe - 1033
Fe - 1034
Fe - 1035
Fe - 1036
Fe - 1037 Fe - 1038
CV - 05
CV - 06
SU - 01
SU - 02
CS - 001
CS - 002
CS - 003
CS - 004
SU - 1021
CS -1013
CS - 1014
CV - 1036
CV - 1038
CV - 1037
CV - 1035
PP- 03 PP- 04
PP- 05 PP- 06
CV - 07
CV - 09
CV - 08
CV - 10
Collahuasi Grinding Circuit
• Three SAG Mill lines and produces copper and molybdenum concentrate.
• The total production of the concentrator plant is 140000 [tpd].
• SAG Mill 1011 production is 60% of total production.
The Challenge
• High Capacity SAG Mill– Size: 40x24 [ft]
– Power: 21500 [kW]
– Max throughput: 5300 [ton/h]
Manual Operation Automatic Operation
The Team
GSOGSOOperationsOperations
HoneywellHoneywell
Search of advanced control solutions at the market
The choice of available alternatives.
The formulation and implementation of advanced control strategies.
*GSO: Operational Services Management
• Honeywell Chile– APC(Control/Process Engineers)
• Collahuasi:– Operations(Process Engineers)
– GSO*(DCS, Automation Engineers)
Targets
• Implement an advanced control solution able to:
– Govern the SAG mill.– Stabilize the main variables.
– Handle process constraints.
• Optimize the operation – Keeping the mill weight within the
operational range (940-1020 [ton]).
– Maximizing the fresh feed rate.
– Reducing the impact of process disturbances (variations in feed particle size and recycle of pebbles).
Advanced Process Control
• Increase the profit by:
– Carry out the process to an optimal state
– An increase of plant operational efficiency
– Give a measurement of plant performance
– Coordination between the different process units
Variable
TimePoor control
Constraint Limit
Good RegulatoryControl
Advanced Control
$$
MPC Control Strategy
• Like a chess master– A set of (optimal) movements is
calculated (based in a prediction) in order to reach the objective.
– Optimal movements are computed at each control interval in order to handle changes in the “game conditions”.
• MPC: Model based Predictive Control– A well-established industrial control technology which dates back over 30
years.– 2000+ documented industrial applications*.
• ~100 applications in mining process with Honeywell Technology since 1996.
– Refining and Petrochemical applications are typically dominant but MPC is being rapidly adopted in other markets.
* ref: Control System Design (G. C. Goodwin et al. 2001)
Honeywell’s MPC: Profit Controller
• Profit Controller (RMPCT*):– RCA: Range Control Algorithm.
– Economical optimization (e.g. minimize power consumption).
– Robust control technology.
Known values Optimal response
Unforced prediction
Past Present Future
*RMPCT: Roboust Multivariable Predictive Control Technology
The Solution: ProfitSAG
• ProfitSAG is an MPC solution for SAG Mills– Objective function designed to accomplish the goals (maximize fresh feed rate)
– Fault tolerant policies (anti-windup integration with regulatory control level)
– Fully integrated with measured disturbances
SAG 1011
• Fresh Feed• Mill Speed• Solids
• Mill’s Weight• Mill’s Power• Mill’s Noise• Mill’s Torque• Produced Pebbles
• Returned pebble• Particle size
Process Process ValueValue PredictionsPredictions
MVs
DVs
CVs
ProfitSAG Dynamical ModelsFinal Trials
CV1 -WEIGHT
CV2 -NOISE
CV3 -POWER
CV4 -PEBBLES
CV5 -TORQUE
MV1 -ORE FEED [TPH]
MV2 -MILL SPEED [RPM]
MV3 -SOLIDS [%]
DV1 -RETURNED PEBBLE [TPH]
DV2 -FEED PARTICLE SIZE (<1”) [%]
G(s) = .01951
3s + 1e
-2s0 5 10 15 20
G(s) = -16.71
4s^2 + 4.18s + 1e
-1.67s0 5.25 10.5 15.7 21
G(s) = .07051
5.85s^2 + 5.33s + 1e
-0s0 4.5 9 13.5 18
G(s) = -4.121
38.4s + 1e
-6.08s0 39.9 79.8 120 160
G(s) = -.0195-.231s + 1
14s^2 + 5.96s + 1e
-0s0 5 10 15 20
G(s) = 6.011.85s + 1
.459s^2 + 3.89s + 1e
-0s0 3.17 6.33 9.5 12.7
G(s) = -.07171
5.76s^2 + 5.98s + 1e
-0s0 5.12 10.2 15.4 20.5
G(s) = 2.121
2.56s + 1e
-3.25s0 3.37 6.75 10.1 13.5
G(s) = 6631
.49s + 1e
-0s0 1.37 2.75 4.12 5.5
G(s) = 90.11.23s + 1
.368s^2 + 1.21s + 1e
-0s0 2 4 6 8
G(s) = -36.81
1.57s + 1e
-.75s0 3 6 9 12
G(s) = 4.211
.00949s^2 + .423s + 1e
-.0833s0 .625 1.25 1.87 2.5
ls MV1 -ORE FEED [TPH]
MV2 -MILL SPEED [RPM]
G(s) = .01951
3s + 1e
-2s0 5 10 15 20
G(s) = -16.71
4s^2 + 4.18s + 1e
-1.67s0 5.25 10.5 15.7 21
G(s) = -.0195-.231s + 1
14s^2 + 5.96s + 1e
-0s0 5 10 15 20
G(s) = 6.011.85s + 1
.459s^2 + 3.89s + 1e
-0s0 3.17 6.33 9.5 12.7
Evaluation
Scenarios:
1. Unconstrained Process • Sending produced pebbles (~8-10%) to
crushing plant.
2. Constrained Process• Recycling pebbles to SAG mill.
Performance Index:
1. Feed rate [tph]2. Specific energy consumption [kW/tph]
Scenario 1: without pebbles recycling
3 3.5 4 4.5 5 5.5 6 6.5 7 7.50
2
4
6
8
10
Specif ic Energy Consumption [kW/TPH]
Per
cent
age
of O
ccur
renc
e
Histogram of Specif ic Energy Consumption
Prof itSAG ONMedia: 4.9639Std: 0.44711Prof itSAG OFFMedia: 5.0755Std: 0.46949
2000 2500 3000 3500 4000 4500 50000
2
4
6
8
10
12
14
Fresh Feed [TPH]
Per
cent
age
of O
ccur
renc
e
Histogram of Fresh Feed
Prof itSAG ONMedia: 3649.9041std: 347.8902Prof itSAG OFFMedia: 3567.0156std: 361.0295
• The average has been reduced by 2.2% (0.1 [kW/tph]) and its standard deviation has been reduced by 4.8% (0.02 [kW/tph]).
• The fresh feed rate has been increased by 2.3% (82 [tph]) and its standard deviation has been reduced by 3.6% (14 [tph]) .
Scenario 2: recycling pebbles
2000 2500 3000 3500 4000 45000
4
8
12
16
20
Fresh Feed [TPH]
Per
cent
age
of O
ccur
renc
e
Histogram of Fresh Feed
Prof itSAG ONMedia: 3074.3594Std: 290.5276Prof itSAG OFFMedia: 2792.3525Std: 295.7731
4 4.5 5 5.5 6 6.5 7 7.5 80
2
4
6
8
9
Specif ic Energy Consumption [KW/TPH]
Per
cent
age
of O
ccur
renc
e
Histogram of Specif ic Energy Consumption
Prof itSAG ONMedia: 5.5155Std: 0.4567Prof itSAG OFFMedia: 5.7726Std: 0.60116
• The Specific Energy Consumption average has been reduced by 4.5% (0.24 [kW/tph]) and its standard deviation has been reduced by 24% (0.15 [kW/tph]) .
• The Fresh feed rate has been increased by 10.1% (282 [tph]) and its standard deviation has been reduced by 1.7% (12 [tph]) .
Conclusions
• Successfully integration between Collahuasi Operations, GSO and Honeywell APC Chile.
• Implementation of a Fully Automatic Solution in record time (1 month).
• Excellent initial utilization 90%.
• High confidence of operators and supervision in the controller actions.
• ProfitSAG shows good disturbance rejection and handling of constraints (e.g. recycle Pebbles).
• Fresh feed rate, has been increased in a range of 2.3 to 10 % (82 to 282 [tph]) and reduced variations by 3.6 % (14 [tph]), depending on the operational conditions.
• The Specific Energy Consumption, has been decreased in a range of 2.2 to 4.5% (0.1 to 0.24 [kW/tph]) and its standard deviation has been reduced by 24%, depending of the operational conditions.