presentation to the mill optimisation summit 2011 presented by: paul wilson technology manager...
TRANSCRIPT
![Page 1: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/1.jpg)
Presentation to The Mill Optimisation Summit 2011
Presented by:Paul WilsonTechnology Manager Calibre Automation, Communications & Technology Group
Optimising SAG mill throughput:A case study in tuning
![Page 2: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/2.jpg)
Two SAG mills
1
![Page 3: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/3.jpg)
• Porgera Mine, PNG
• 4.5 Megawatt, variable speed drive
• About 500 tonnes per hour per mill
• Highly variable lithology with grinding factors from 6 to 18 kilowatt hours per tonne
The mills
2
![Page 4: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/4.jpg)
The mills
3
![Page 5: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/5.jpg)
The mills were often unstable, seen as oscillations in the feedrate trend graphs
Unstable behaviour
4
![Page 6: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/6.jpg)
Loss of production
7
As much as 15% on bad days
Up to 380 ounces of gold per day on bad days
At $425 US per ounce = $160,000 per day
You could hire a very good plant operator for that kind of money
![Page 7: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/7.jpg)
Natural instability
8
Poor tuning causes natural instability
![Page 8: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/8.jpg)
Operator-caused instability
8
Poor operator skills also forces instability
![Page 9: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/9.jpg)
The control system
9
Highest levelMinnovex expert system control
Optional top level controlConstraint control
Mid-levelClosed loop control
Bottom levelDelta-V distributed control system
![Page 10: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/10.jpg)
The Minnovex expert system
9
A fuzzy rules based artificial intelligence system
Running on a G2 expert system shell
Running on a Windows NT PC platform
Takes data from, & feeds setpoints to, the loop controllers on the Delta-V
![Page 11: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/11.jpg)
Performance comparison
9
Expert system control is far better than poor operator control
![Page 12: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/12.jpg)
Why mills go unstable (#1)
9
Mills stall (bog or centrifuge). The behaviour at maximum throughput is highly non-linear
![Page 13: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/13.jpg)
Why mills go unstable (#2)
9
Dynamic behaviour of a mill is type 1
Control engineers recognise that type 1 systems are more likely to be unstable than type 0 systems
Caused by the inherent integration in the mill transfer function
![Page 14: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/14.jpg)
Mill load (level) is the integral of the nett feedrate
level = ∫ (Qin – Qout).dt
This induces a -90o phase shift in the transfer function which leads to reduced stability
Why mills go unstable (#2)
10
![Page 15: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/15.jpg)
Why mills go unstable (#2)
9
The integration causes a phase shift
![Page 16: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/16.jpg)
Simulated integral response
11
![Page 17: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/17.jpg)
1. When a mill stalls it stops working. 2. The mill fills with unground material.3. It takes time to grind out the rock and get the
outflow going again
So: the control system / plant operator must be:
PATIENT
Why mills go unstable (#3)
10
![Page 18: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/18.jpg)
The result of operator impatience
13
![Page 19: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/19.jpg)
Mill under tuned expert system control
13
![Page 20: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/20.jpg)
Section A to B is the maximum speed of recovery to prevent stalling the mill again
Expert system recovers from a motor overheat event
13
![Page 21: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/21.jpg)
The result
30
• In 2004 / 2005 mill production rose from850,000 ounces to 1,000,000 ounces
• At $425 US per ounce that was $63.75 million US increase per year
Not possible without increasing SAG mill throughput
![Page 22: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/22.jpg)
Extra energy used
7
The difference in energy usage between the unstable zone and the stable zone is:
the unstable zone averages
25% more motor energy per tonne of product
than the stable zone and produces
15% less product
Which adds 10% to the energy costs for the remainder of the processing plant
![Page 23: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/23.jpg)
How was it done?
30
• Develop a mathematical model• Use trend plots and tests to characterise
the mill (find the characteristics of mill behaviour)
• Estimate fastest possible recovery times on the worst-case ores
• Retune the expert system rules for robust, always-stable behaviour
![Page 24: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/24.jpg)
How was it done?
30
Block diagram model of mill behaviour
![Page 25: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/25.jpg)
How was it done?
30
Transfer function of mill load to ore feedrate
![Page 26: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/26.jpg)
Additional development
30
The decision to secondary crush the harder ores.
A secondary crusher was installed. With a bit of clever mathematics we were able to estimate SAG mill grinding factor at the primary crusher. We used this to feed some of the hard ore (GF > 10 kWhr per tonne) through the secondary crusher thus increasing SAG mill throughput on the harder ore.
![Page 27: Presentation to The Mill Optimisation Summit 2011 Presented by: Paul Wilson Technology Manager Calibre Automation, Communications & Technology Group Optimising](https://reader035.vdocuments.us/reader035/viewer/2022062421/56649dc65503460f94abb1f1/html5/thumbnails/27.jpg)
Questions
31
Your questions are welcome
Plant characterisation & transfer function development is a complex process. I am happy to discuss some of the methods afterwards with anyone interested.