MODEL PREDICTIVE CONTROL APPLICATION AT LUMWANA COPPER CONCENTRATOR

Similar documents
MODEL PREDICTIVE CONTROL TO OPTIMIZE MILLING CIRCUITS USING NKOMATI NICKEL MINE AS A CASE STUDY

Increasing Grinding Circuit Robustness with Advanced Process Control

MILLING CONTROL & OPTIMISATION

SAG Page 2. Page 1

FLOTATION CONTROL & OPTIMISATION

A Holistic Approach to Control and Optimisation of an Industrial Crushing Circuit

Operational Excellence on the plant floor and in the control room: A disciplined approach to understanding the opportunities to maximise the

SAGwise total process control The key to efficient SAG mill operation

High pressure grinding rolls applications for the platinum industry

20 June Pat Turner, President FLSmidth Krebs. Pumps and Cyclones. Capital Market Day 2017

SAGDesign TM Using Open Technology for Mill Design and Performance Assessments

SAG Page 1. Page 2

The implications of ore hardness variability on comminution circuit energy efficiency (and some other thoughts)

APPLICATION AND BENEFITS OF ADVANCED CONTROL TO ALUMINA REFINING

FEATURES AND BENEFITS

Increasing Productivity Without Capital Expenditure

Process optimization using real time tracking of coarse material in individual cyclone overflow streams

Goulds XHD. Lenntech. Extra Heavy Duty World-Class Lined Slurry Pumps. Tel Fax.

Development and Implementation of the new 2240 kw (3000 HP) VERTIMILL Grinding Mill for Newcrest

JME Journal of Mining & Environment, Vol.2, No.1, 2011,

Abstract. 1 Introduction. Golder Associates Ltd, Canada

Crushing. Comminution

KnowledgeScape! EfficiencyExpert! Increase Profit. Reduce loss.

Minerals. WARMAN Centrifugal Slurry Pumps WBV Vertical Cantilever Slurry Sump Pump Series

New project in Russia with semiautogenous mill selected according to the results of SAGDesign testing

SAG Mill Control at Northparkes Mines (Not So Hard After All)

ADVANCED PROCESS CONTROL FOR YARA UREUM PLANT BRUNSBÜTTEL

Sukari revise Gold production in Egypt

Cadia Expansion The Impact of Installing High Pressure Grinding Rolls Prior to a Semi-Autogenous Grinding Mill

Comminution Circuit Design vs. Feed Size or Mine-to-Mill 2.0

COMMINUTION CIRCUIT OPTIMISATION

Practise of large scale hematite ore beneficiation

Resource Efficient Mining Processes of Tomorrow

Hydrocyclone classification optimization using realtime coarse particles detection in the overflow stream

Communition Abstract. Fine grind attritional mills; can they or should they go coarser. D Capstick and B Currie

STACK SIZER TM. Patented Technology. The highest capacity, most efficient, fine wet screening machine in the world

GE Intelligent Platforms

NEW APPROACH ON FLOTATION TAILINGS DEWATERING

TAKING CONTROL OF THE MILL FEED: CASE STUDY - PARTIAL SECONDARY CRUSHING MT RAWDON. Brian Putland, Bernie Siddall and Andrew Gunstone

ACHIEVEMENT OF HIGH ENERGY EFFICIENCY IN GRINDING MILLS AT SANTA RITA. *S. latchireddi 1 and, E. Faria 2

Daniel A. Silva Minera Los Pelambres, Salamanca, Chile. Matt F. McGarry ANDRITZ AUTOMATION, Bellingham WA, USA

Page 1 of 5. Sonar Technology: More than Just a Flow Meter - Leveraging Dual Measurement Capability for Enhanced Value in Industrial Processes

HRC : Taking HPGR efficiency to the next level by reducing edge effect

Comminution 2014 ABSTRACT

We are building NEW AITIK. New Aitik. 36 million tonnes

Two-stage classification in one unit

GROSS POWER CALCULATOR

IPS Inclined plate settlers. Maximum process recovery, minimum plant footprint

PAPER # 1882 SYMPOSIUM: MINERAL PROCESSING Mineral Processing of Complex Sulphides

Smarter Washing Solutions AGGST RM 150

Flotation cplant: an Optimum Modular Approach

CT425 Optimize Your Process to Make Profits

Hannah Yan. Mobile/Skype/WhatsAPP: Wechat:yhr MSP HEAVY DUTY SUMP PUMPS

OUTOTEC FLOTATION TECHNOLOGIES

ABB Cement Fingerprint Holistic Approach for Cement Industries

For cleaner coal at lower cost. Modular Coal Preparation Plant. Modular Coal Preparation Plant. Classification, cleaning and dewatering

LSA Pump Series - Low maintenance, abrasion resistant for heavy duty service

Development of an Innovative Copper Flowsheet at Phu Kham. M F Young and I Crnkovic

Institute of Quarrying Australia

Global Product Group Cement, CPM Cement. cpmplus Expert Optimizer Designed by process experts, customized to your needs

Industrial IT : Optimization for the Cement Industry

The Increased Recovery Project at the Phu Kham Copper-Gold Operation, Laos PDR. Authors: A Hoyle, D Bennett, and P Walker

The TMAC Resources Inc. Hope Bay Project An Innovative Approach. Gerhard Bezuidenhout

A CLOSER LOOK AT INCREASING HPGR EFFICIENCY VIA REDUCTIONS IN EDGE EFFECT

Energy-Mass-Size balance model for dynamic control of a wet open circuit grinding mill: Part II Simulation tests

New trends in Process Automation for the cement industry

ABSTRACT INTRODUCTION

WET MIX MACADAM PLANT

HPC-30 Specifications

AUTOMATIC CONTROL OF A HIGH TENSION ROLL SEPARATOR

MOUNT CARLTON COMMINUTION CIRCUIT DESIGN, START-UP AND OPTIMISATION

Minto Mine. Mill Operations Plan

FLSmidth Dorr-Oliver Eimco Flotation Technology. Superior Metallurgy Higher Availability

ENGINEERING SERVICES

GRAVITY CIRCUIT Knelson Performance, Maintenance, Back to Basics. Drissa ARAMA, Plant Manager

IMPROVEMENTS IN SAG MILL THROUGHPUT FROM FINER FEED SIZE AT THE NEWMONT AHAFO OPERATION

From copper ore to pure copper with the power of TAKRAF and ABB equipment

The EDEM BulkSim Solution:

Gravity Gold Recovery at President Steyn Mine. P. F. Van den Steen

VXP2500 STIRRED MILL OPTIMIZATION AT CASMYN MINING TURK MINE

The Impact Wear Behavior of Large Rocks on Slurry Pump Materials and Equipment.

Developing Process Control Standards for Optimal Plant Performance at PanAust Limited

Asset Performance Management for Mining

OCEANAGOLD PROVIDES AN UPDATE TO HAILE COMMISSIONING AND REVISED COMPANY GUIDANCE

POLYCOM high-pressure grinding roll.

PIN MILLS POWDER PROCESSING TECHNOLOGY: THE STURTEVANT SOLUTION.

IN-PIT CRUSHING AND CONVEYING MINE PLANNING AND OPERATIONS Skillings Mining Review, June 1, 1985 by: Independent Mining Consultants, Inc.

City of San Mateo WWTP Filter Cake System

Evaluation of Grade Engineering using Enterprise Optimization. Michael Scott, Nick Redwood

Metso SmartTag The Next Generation and Beyond

sepro product catalogue mineral systems mining mineral processing aggregates milling screening scrubbing gravity specialists

A PROTOCOL FOR CONDUCTING AND ANALYSING PLANT TRIALS: TESTING OF HIGH-CHROME GRINDING MEDIA FOR IMPROVED METALLURGY

Minerals Enduron HPGR High Pressure Grinding Roll

SHEDDING LIGHT ON SECONDARY CRUSHING

ABC of Mine-to-Mill and Metal Price Cycles

T S X : R M X N Y S E. M K T : R B Y

Energy Efficiency Comparison in Fine Grinding in the Mining Industry

Performance Audit of a Semi-autogenous Grinding Mill Circuit

Mining company uses Operations Optimization, powered by Predix, to improve throughput by more than 5.5%

Transcription:

MODEL PREDICTIVE CONTROL APPLICATION AT LUMWANA COPPER CONCENTRATOR Mick Rogers Process Manager, Lumwana Mining Company, Zambia David G Almond Global Product Manager Minerals Process Control, FLSmidth Inc., Automation Division, USA Tejas D. Maru Development Engineer, FLSmidth Inc., Automation Division, USA King Becerra Product Manager, High Level Process Control, FLSmidth Inc., Automation Division, USA ABSTRACT Equinox Minerals has constructed and is operating a new mine; Lumwana Mining Company located close to Solwezi in the North Western province of the Republic of Zambia. The plant employs a conventional copper processing circuit whereby low grade, primary crushed copper ore is processed through a SAG/Ball mill circuit and flotation to produce concentrate. Plant nameplate design capacity is 20 million dry metric tons per annum. To optimize the operation and ramp up production, Lumwana Mining Company signed a contract with FLSmidth Automation to implement FLSmidth s advanced process control solution, the ECS/ProcessExpert System. This paper presents the implementation results and performance data recorded during system evaluation. The results conclusively demonstrate benefits during circuit operation with the system ON. INTRODUCTION The Lumwana processing plant grinding circuit is of conventional design, consisting of a SAG/Ball milling circuit closed by cyclones. Powered by Siemens Gearless Mill Drives, the mills were supplied by FLSmidth and cyclones and cyclone feed pumps by FLSmidth Krebs. Low grade copper ore from the primary crushed ore stockpile is reclaimed via two variable speed apron feeders onto the primary mill feed belt. Primary grinding is achieved in an 18MW, 11.58m diameter x 5.49m EGL SAG mill discharging over a steel trommel into a sump shared with the secondary ball mill. Trommel oversize is recycled back to the SAG mill feed chute via two conveyors. Primary ground slurry is pumped by twin variable speed MF600 MillMax pumps to twin cyclone packs of eight

838mm GMax cyclones each. Cyclone underflow gravitates to the 16MW, 7.92m diameter x 12.19m EGL secondary ball mill discharging into the common sump. Cyclone overflow flows to the flotation circuit. The circuit described together with available measurements and actuators is shown in Figure 1 below. Figure 1: Lumwana circuit flow sheet Large mineral deposits such as that processed at Lumwana Mining Company (LMC), typically comprise ore of significantly varying characteristics and these create feed conditions that make stable operation of the circuit challenging. Furthermore, the mine development program has resulted in a significant period of tonnage ramp up and occasions when it has been necessary to operate the plant in a significantly turned down condition. To facilitate ramp up to full production LMC decided to install an advanced process control (APC) system. Since FLSmidth was the supplier of the grinding circuit equipment they were selected to supply the APC. ECS/ProcessExpert The system implemented was developed based on FLSmidth s ECS/ProcessExpert System (PxP), an advanced process control solution used to stabilize and then optimize key plant processes. The system was supplied as a complete solution for custom modeling and control application. For this application, advanced techniques (model predictive control and fuzzy-logic) were used in the application modules to enable hybrid control schemes that meet the specific process control requirements. These modules perform complex and continuous evaluations of process

conditions and execute control actions on a more frequent and reliable basis than human operators. Being an open toolbox, PxP allows the solution to be specifically tailored to the needs of the plant by incorporating the best control knowledge. Application objectives The objectives of the PxP grinding circuit application are: Stabilize SAG mill operation and prevent over filling of the SAG mill. Minimize circuit stoppages and allow controlled operation at maximum and significantly reduced feed rates. Optimize control and balance loads between the SAG and ball mills. Protect the SAG mill against undesirable liner impacts (monitored by a specialized instrument called an Impactmeter). Avoid sump spillage and maintain cyclone inlet pressure to set points. Maintain constant flow to the flotation plant. Keep particle size analysis (manual value) on target. Improve grind and energy efficiency. The PxP graphical operator interfaces for the primary and secondary grinding circuits are shown in Figures 2 and 3 below. Both user interfaces display the controller mode, control status (indices), online measurements, actuators, main mimic and a trend display. The user interface was designed to easily identify the actions being performed by the controller and enable the operator to partially enable/disable the control strategy whenever an instrumentation or process problem occurs. Information regarding the utilization of the system is also displayed. Figure 2: ECS/ProcessExpert operator station graphical interface primary grinding circuit

Figure 3: ECS/ProcessExpert operator station graphical interface secondary grinding circuit METHODOLOGY An eight-step methodology was followed to assure the success of the ECS/ProcessExpert Grinding Circuit application. This methodology comprises all the activities and steps that are required for an efficient development of advanced process control solution, aligning customer expectations with the application goals for the successful commissioning of the system. In summary, the eight-step methodology comprises the following activities: 1. Project planning 2. Kick-off and process interviews 3. Application Design 4. Primary commissioning 5. ECS/ProcessExpert training 6. Remote monitoring and tuning 7. Follow-up visit 8. Long-term support To evaluate the project success a test protocol defining the main guidelines to measure operational benefits was agreed between LMC and the supplier. The process control techniques, implementation timeframe, control strategy and results are presented below. Techniques The advanced process control solution was developed based on FLSmidth s ECS/ProcessExpert System. At this application has sparse number of measurements and actuators and involve a

multi-time scale delayed process, model predictive control (MPC) was identified as the best technique for this application complemented by fuzzy logic rules. The model predictive control technology uses a mathematical model of the process to predict future process behavior and pre-plan actions in the future to attain desired targets. The controller output is a manipulated variable (MV) applied to the inputs of the process and the process model, and is a part of every MPC controller. The process model computes a predicted trajectory of the controlled variable (CV) that is the process output. After correction of this trajectory for any mismatch between the predicted value and an actual measured value of the controlled variable, the predicted trajectory is subtracted from the future trajectory of the set point to form an error vector (Figure 4) [1]. CV Setpoint trajectory Predicted errors CV prediction MV Future calculated moves Past Future time Figure 4: Graphic illustration of the operation of a MPC controller System implementation The system was prepared at the suppliers premises and installed at the plant in August 2009. Establishing communications with the main process control system was a simple matter due to the compatible drivers installed in the PxP. Checking of signals and tags was completed within 5 working days thereafter the system was left gathering data and under remote monitoring from the USA for a period of 12 weeks. Commissioning of the system was complete after a further 6 weeks on site permitted control strategies and setup to be incorporated into the ECS/ProcessExpert. Final tuning of the system was completed via a remote connection using the connection established during installation by the supplier. General control strategy As part of the application design, the key process measurements required for control were identified between LMC and FLSmidth. It was decided to split the application into two main modules or components; the primary grinding circuit comprising the SAG Mill and raw ore feed and the secondary grinding circuit comprising the ball mill, sump and cyclones. The SAG mill control strategy was developed with the following main goals in mind: SAG mill load stability SAG mill throughput optimization

Reduction of charge generated undesirable impacts The above goals were attained through maintaining SAG mill load to set points; the SAG mill load is measured using the bearing pressure and controlled by speed and feed together with signals from an FLSmidth Impactmeter. The primary control technique used in the structure of expert system is MPC based. Combination of Fuzzy Logic and MPC is used in the construction of the SAG mill centrifuge identifier and SAG mill weight optimizer. The expert System has an option to use the MPC based on transfer function as well as state space. The MPC technique used in this application is based on transfer function. Table 1 illustrates the process measurements used as inputs and the outputs of the MPC controller. Table 1: Inputs and outputs of the MPC MPC Inputs SAG Mill Power SAG Mill Weight SAG Mill Solid % (SAG Density) Critical Impacts Cyclone Inlet Pressure Cyclone Flow Sump Level Ball Mill Bearing Pressure MPC Outputs SAG Mill Feed SAG Mill Speed SAG Mill Water Flow Sump Water Flow Cyclone Pump Speed Ball Mill Water Flow The process models were obtained to understand the effects of actuators and disturbances on the process measurement. The majority of the models were developed from plant data based on the response of the process measurements by manipulating the actuators. Some of the models were also obtained from the design data of the equipment and then were compared with the real time models obtained from plant data to compare the quality of models. Figure 5 illustrates the three different modes to control the SAG mill with the ECS/ProcessExpert System: startup control mode, normal control mode and upset control mode. Upset Control Mode Normal Control Mode Startup Control Mode Figure 5: ECS/ProcessExpert control modes

The SAG mill upset control mode is designed to handle unexpected and sudden disturbances to the SAG mill. This upset control mode will bring the process back to normal operation as soon as the process disturbances disappear. Inclusion of impact monitoring via signals from the Impactmeter have also been built into the load control strategy and monitors and prevents the mill from being run at excessive speed while maintaining the load to ensure good ore reduction, but avoiding critical impacts to protect the mill liners. A SAG mill weight target optimizer was developed to automatically establish the SAG mill weight target, manage mill critical impacts and adjust the SAG mill load according to the ore condition. This function optimizes the SAG mill weight target by correlation of the historic data of mill weight, current mill weight target, mill critical impacts and mill speed whenever the SAG mill experiences critical impacts during normal operation. Within the secondary grinding circuit the mills'discharge sump level control and cyclone classification control strategies were also implemented using model predictive control techniques. Stability of operation of the downstream flotation circuit is essential to maximising recovery, thus an important criteria for the PxP was to reduce fluctuations in flow and grinding circuit product size. Stable flow and constant size distribution is dependent on good cyclone operation which requires stable cyclone pressure and density control as well as smooth change over when the number of cyclones operating in each of the two cyclone banks is varied. The absence of density meters prevented density control by the PxP so focus was on maintaining constant pressure and flows through management of pumps speed and cyclones selection. RESULTS AND DISCUSSIONS A test protocol to assess potential operational and economic benefits resulting from the implementation of the PxP was agreed between LMC and FLSmidth. The test period was defined to run for 2 weeks and included the main variables which influence operational and economical benefits: tonnes per hour (tph), specific power (kwh/t), cyclone inlet pressure (kpa), SAG mill weight (t) and particle size analysis (in microns). The test protocol definition included running the mill for 15 hours (single period) with the PxP ON and OFF in a randomized sequence. The test protocol was executed during February 2010 and eight periods were identified as valid for analysis and measurement of operational benefits (four periods with the PxP ON and four with it OFF). After gathering and comparing results, main achievements of the implementation of the PxP on the Grinding Circuit of LMC are: Circuit stability on startup was attained twice as quickly with the PxP ON as than when it was OFF (within 60 minutes). Reduction of 3.4% in cyclone feed pump energy consumption. Reduction of 64% in SAG mill weight variability. Reduction of 14% in mill undesirable critical impacts occurrences. Although the impact meter is used to prevent critical impacts during ON and OFF test, the PxP was more efficient at reducing undesirable critical impacts. Through improved stability and control of the pumps and cyclone pressures, cyclone operation within the optimum pressure range specified by the supplier was achieved for

twice as much operational time with the PxP ON than with it OFF, keeping product grind on target. Pump 1 speed: 42% process variability reduction. Pump 2 speed: 44% process variability reduction. Cyclopack 1 Pressure: 54% process variability reduction. Cyclopack 2 Pressure: 65% process variability reduction. Better representation of the stability results are illustrated in Figures 6 and 7. Figure 6 illustrates the SAG mill weight standard deviation for each period with and without the PxP and reflects how the SAG mill weight process variability was reduced in average 64%, while maintaining the required target. Mill Weight Standard Deviaton Standard Deviation 35 30 25 20 15 10 5 6 15 8 11 10 30 5 24 0 1st period 2nd period 3rd period 4th period Period PXP ON PXP OFF Figure 6: Mill weight standard deviation by period Figure 7 illustrates the improved stability in cyclones pressure control. Cyclones operation within the optimum pressure was maintained for twice as much operational time with the PxP ON than with it OFF. CycloPack Pressures 70% 60% 50% 56.7% % Time 40% 30% 20% 10% 0% 33.5% 28.8% 21.8% 15.7% 13.6% 13.3% 8.1% 8.4% 0.1% >94 92-94 84-92 80-84 <80 KPa PXP ON PXP OFF Figure 7: Percentage of time cyclopack pressures was within each operational range

Improvements in consistency and stability of grinding circuit control is expected to yield benefit in terms of ultimate circuit capacity and overall circuit power efficiency when the PxP system is allowed to run "open" without ore supply constraint. Similarly, the improvement in stability of the grinding circuit in terms of product quality and flow rate output translates directly to improved stability in the flotation circuit, which is expected to result in some improvement in recovery of copper. The reduction in pump speed variation resulted in an increase in pump power efficiency, which coupled with the expected concomitant wet end wear reduction translates to significant financial benefit. Additionally, improvement in pressure and flow control is expected to yield wear benefit beyond that associated with higher power efficiency. Fluctuating pump speed generates pressure surges in the pump casing which can force slurry down the front and back of the impeller running clearances. This will accelerate wear on the suction liner and wear ring due to an increase in solids being dewatered by the higher pressure as it passes through the wear ring clearance. Pressure surges at the back impeller clearance can also exceed the inlet gland water pressure and result in spurts of slurry exiting the gland. The solids in the slurry can impregnate the soft gland packing and result in wearing the shaft sleeve which could result in the pump having to be stripped for overhaul prematurely. An accelerated flow in the pump will increase wear at the eye of the impeller and at the pump casing cutwater. CONCLUSIONS The ECS/ProcessExpert System has been successfully implemented on the grinding circuit at Lumwana Mining Company. The recorded results demonstrate benefits during circuit operation with the PxP ON. Circuit stability and improved equipment performance were significantly better than during manual control. An ECS/ProcessExpert model predictive control module has been used as the main process control technique, surpassing limitations of traditional control techniques, such as PIDs. The model predictive control technology has proven its ability to effectively control the grinding circuit. At the time of preparing this paper the system utilization was 98%. System utilization is an indirect measure of success of advanced process control projects. This utilization is very high and is attributed to the robust nature of the installed system platform together with a high degree of operator confidence. The test protocol was executed as a randomised block trial under similar process and operational conditions to assure the representativeness of data. While issues with availability of ore supply to the grinding circuit and plant reliability limited the number of valid data points obtained during the trial period, the PxP system demonstrated the ability to improve performance of the Lumwana grinding circuit. In summary, pump power and wear savings were demonstrated, circuit stability on startup was attained twice as quickly, cyclone feed flow variation was significantly reduced and cyclones were operated within the optimum pressure range for twice more operational time with the PxP ON than with it OFF. Although not tested, it is expected that increased average ultimate grinding circuit capacity will result from the improved stability of operation with PxP ON, as well as improved recovery due to reduction of frequency and magnitude of surging in the flotation circuit.

To assure longevity of the system and continued efficiency, remote monitoring services for troubleshooting and tuning are available through a secure internet connection which enables the supplier s technical experts to effectively support the system remotely. Despite this, the system has proved itself to be extremely reliable and no additional tuning has been necessary since it was handed over to the plant. ACKNOWLEDGEMENTS The authors acknowledge and thank the following: The management and staff of Lumwana Mining Company for their support in gathering and analyzing evaluation data and for approval to present these results. Pat Turner and John Frater of FLSmidth Krebs for assistance provided in analyzing cyclone and cyclone feed pump performance. The management of FLSmidth for approval to present performance evaluation of the ECS/ProcessExpert System. NOMENCLATURE SAG Semi-Autogenous Grinding LMC Lumwana Mining Company APC Advanced Process Control ECS Expert Control & Supervisory PxP ECS/ProcessExpert System EGL Effective Grinding Length MPC Model predictive Control MV Manipulated Variable CV Controlled variable PSM Particle Size Measurement REFERENCES Blevins, T.L, McMillan, G.K., Wojsznis, W.K. & Brown. M.W (2003) Advanced Control Unleashed. ISA-The Instrumentation, Systems, and Automation Society, pp. 309 [1]