03 Michael Schaffer
-
Upload
angel-canales-alvarez -
Category
Documents
-
view
231 -
download
0
Transcript of 03 Michael Schaffer
-
7/27/2019 03 Michael Schaffer
1/27
Process
Plant Operational Excellence
through Advanced Process Control
By
Michael Schaffer, Robert Sloan,
Glenn Dobby and Andrea Festa
Adv anced Process Contro l Area
MinnovEX Technologies Inc.
Toronto, Canada
-
7/27/2019 03 Michael Schaffer
2/27
Agenda
Definitions Where are we today
Data
Technology
Integration
Case Studies Grinding
Flotation
Dewatering
Where are we going
Short term Long term
-
7/27/2019 03 Michael Schaffer
3/27
Definitions
Operational Excellence : the consistent application ofoperational best practice leading to increased quality coupled
with improved performance.
Advanced Process Control :
Advanced Control
Strategic as opposed to tactical
Concept development of a Global
model (Theoretical, Phenomenological,
Empirical, Stochastic, Heuristic)
Field Instrumentation
PLCs/Controllers
DCS / SCADA
MIS
APC
-
7/27/2019 03 Michael Schaffer
4/27
Introduction
In 1995 APC in the mineral processing industry moved from the
development phase into a practical industrial solution.
Now, in 2004 we can confidently state that advanced process
control has reached a mainstream level, achieving over 99%
availability with greater than 95% utilization and installed in
many of the major operating facilities in the world.
This success is due not only to the advances at the APC levelbut even more importantly at the regulatory, network and
hardware levels.
With the addition of new instrumentation, operations has been
able to move from a qualitative to a quantitative environment
facilitating a better understanding of the circuit state.Additionally, data driven and fundamental models have been
successfully applied as soft sensors and predictors.
The next step will be the morphing of these models into true
controllers.
-
7/27/2019 03 Michael Schaffer
5/27
Introduction (cont.)
The South American industry has positioned itself at the
vanguard of this control revolution. Recognising the potential
benefits and while managing risk, testing and implementing the
solutions.
Each advanced control solution has 3 components:
Data
Technology
Integration
-
7/27/2019 03 Michael Schaffer
6/27
The Data
The data us allows to make informed decisions. The data can be
measured directly or inferred.
Direct measurements are always better. Measurement instruments
are evolving fast enough to bring the industry closer to a fully defined
system and causing competition at a price level.
Examples: froth vision systems and cyclone overflow particle sizeanalysers.
Some characteristics of the operation on-line cannot be measured
and it is necessary to use average or typical values that will be
corrected through a sophisticated error model or filter. For example
a Neural Network model.
Research continues at both the micro and macro level, developing
and testing several techniques.
-
7/27/2019 03 Michael Schaffer
7/27
The DataExamples of these technologies :
Froth Vision Systems :
Originally : fibre optics, CCD cameras, separatelighting : USD20,000/camera + lights
Now : Wireless or Ethernet camera withincorporated lighting : USD4500/camera includinglights
Particle Size :
Originally : Physical contact : USD120,000
Now : Camera based, full distribution : USD40,000
-
7/27/2019 03 Michael Schaffer
8/27
The Data
Many key characteristics of the process can still not be measuredon-line resulting in compromises.
An example being ore hardness in a mill.
At present the industry adopts average or typical values in themodels with variability manifesting itself as an unmeasureddisturbance that is corrected or filtered using an error model orfilter (I.e. Kalmann filter or a Neural Network model).
300 m
800 m
0 200 SPI
-
7/27/2019 03 Michael Schaffer
9/27
The Data As the systems are better defined more models can be leveraged and
their extensibility exploited. Until that time many of the current systems (both human and
automated) rely on inferred data and this explains the success of
expert system technology that leverages a heuristic model.
Inferred data represent variables that are not directly measured but arerather indicated by some secondary or tertiary means. These
parameters are then predicted using various models some moresophisticated than others.
An example is load in a mill currently inferred by either bearing
pressure or weight (through load cells)
In summary it is essential that the type and quality of the data used tomake control decisions must be carefully considered. A well-designed solution actively participates in ensuring that the basis foractions are well supported
-
7/27/2019 03 Michael Schaffer
10/27
The Technology
Selection of the appropriate technology is critical.This applies both to the decision of whichfundamental approach to adopt as well as whichproduct to purchase.
-
7/27/2019 03 Michael Schaffer
11/27
-
7/27/2019 03 Michael Schaffer
12/27
The Technology - Platform
The selection of a platform is critical whenconsidering the long-term sustainability and
maintainability of the advanced control solution.
Special consideration must be given to the
resources available to administer the system. The operation must avoid a solution that
becomes obsolete without a clear path to
modernization such as custom-coded systems.
-
7/27/2019 03 Michael Schaffer
13/27
The Technology - Platform
The selection of the platform has as much to do with the concept
of commercial and technical risk as it does with functionality.
Clients must consider the long-term sustainability of the systems
due to the fast pace of progress within the Advanced Control field.
Many offerings have come and gone over the years it is difficult
to build a strong business model focusing on a single industry. Long term success has so far been the domain of vendors that
build multi industry platforms and count on integrators to
customise for a specific applications.
Such a broad based strategy ensures sufficient cash flow tosupport the research and development programs necessary to
meet the demands of industry and fully exploit the capabilities of
the systems as they become available.
One example is Gensym which still remains at the forefront of
this field after 16 years building expert system platforms.
-
7/27/2019 03 Michael Schaffer
14/27
The Technology - Platform Advanced Process Control solutions are a unique blend of control,
process and operations.
Users accessibility, intuitive diagnostic tools, self-documentation
and recording features are critical to this solutions as they grow
with the operation.
The ability and ease of interfacing with external environments isalso an important consideration
-
7/27/2019 03 Michael Schaffer
15/27
Integration
Integration is considered the most important component in the
successful deployment of an advanced process control
solution.
Regardless of the strength of the data or the technology being
deployed, if the package is not configured and tuned properly
and the solution is not accepted into the culture of theorganization it will fail.
Advanced process control empowers the operator to optimize
the plant.
Each implementation offers its own challenges and must be
considered on its own merits. Success can be accomplished
with the appropriate methodology and education program.
-
7/27/2019 03 Michael Schaffer
16/27
Comminution circuits
MET Solution Case Studies
SAG Mill Expert Systems
Advanced algorithms have been implemented where the expertsystem monitors the relationship between power and pressure
pushing the operation to the sweet spot by considering current,
past and future performance. Through logic the system self
regulates, adapting to changes in ore type and properties.
Ball Mill Expert Systems
Neural networks are a robust, data driven modeling technique thatleverages real-time plant data to continuously test and evaluate
itself, Neural nets are deployed to model the cyclone and predict itsperformance on a real-time basis.
-
7/27/2019 03 Michael Schaffer
17/27
Comminution circuits
A few considerations when using Neural networks are:
Re-training: to ensure that they represent the current
situation
Data preprocessing to make informed decisions
Extrapolation is not recommended
-
7/27/2019 03 Michael Schaffer
18/27
Froth flotation
MET Solution Case Studies
Froth Flotation Vision Systems
New vision-based technology for flotation cells coupled withheuristic flotation expert systems has permitted higher levels ofautomation and control.
The solution METCAM FS have simplified the application of the
technology through innovations such as integrated light andcamera systems and wireless options eliminating one of thehigher cost components of an installation.
Parameters monitored:
Froth Velocity (most important)
Bubble size
Froth stability
Color
Froth typing
-
7/27/2019 03 Michael Schaffer
19/27
Froth flotation
MET Solution Case Studies
MET Flotation Stabilizer (METFS)
It is a flow-surge flotation expert system.
It is currently employed at facilities in Chile and Argentina to
eliminate the need to manually deal with flow surges by
anticipating the onset of a surge before it actually arrives at
the circuit. Once a surge is detected, METFS moves the supervisory
control to standby and aggressively handles the surge by
controlling the instrumentation related.
METFS will return the control to the supervisory control once it
is satisfied that the surge is under control.
-
7/27/2019 03 Michael Schaffer
20/27
Froth flotation
MET Solution Case Studies
Circulating Loads and the MET Solution - Cleaner Circuit
The METCLEANER advisory expert system for a producer in Chileanticipates and controls both internal and external circulating
loads of a cleaner flotation circuit.
Once circulating loads are controlled the cleaner circuit can
achieve maximum performance for that given circuit and feed.
METCLEANER consists of soft sensors that will warn the operators
and suggest corrective actions for maintaining a stable cleaner
circuit .
METCLEANERhas been implemented without the use of cameras on
the first cleaner and cleaner scavengers however we believe it willbe advantageous to install cameras for a better control.
-
7/27/2019 03 Michael Schaffer
21/27
Froth flotation
MET Solution Case Studies
Open Channel Flow Measurements and the MET Solution
Open channel flow measurement, consists of a relatively
inexpensive ultrasonic level sensor and provides a continuousfeedback of the concentrate production. This method captures the
entire cell or row output so it is very reliable .
The open channel flow measurement of concentrate slurry has
been incorporated into the logic for the control of rougher flotation
at an Argentina operation.
The increased response speed and consistency of the expert has
improved stability in the rougher concentrate production in this
operation.
-
7/27/2019 03 Michael Schaffer
22/27
Froth flotation
MET Solution Case Studies
Rule for controlling Air and Level
-
7/27/2019 03 Michael Schaffer
23/27
Dewatering Facilities
The expert system monitors the plant process water tank levels and
thickener overflow capture pond levels.
The expert system automatically starts / stops "tranque" waterpumps to maintain adequate levels in the plant. Also, the expert
system can avoid overflows of the thickener capture ponds that
would otherwise lead to the loss of valuable process water.
Similar applications include CCDs which are increasing in
prominence.
-
7/27/2019 03 Michael Schaffer
24/27
Mini Expert Systems
Once the expert system platform has been purchased andinstalled, and an expert system for one area of the plant has beenintegrated and tested, further sub-modules can be developed
and run from the same platform.
These are opportunities where a new expert system would be
difficult to justify but applications that can clearly benefit frommore intelligent control.
For example a water consumption expert system has been
expanded to control freshwater valves to the concentrator relative
to KPIs for freshwater consumption (per tonne processed). This
aids the site by avoiding excessive water consumption.
-
7/27/2019 03 Michael Schaffer
25/27
A Look at the Near Future
Increased use of models with more, better and
reliable measurements of process variables for
soft sensing and prediction.
Increasingly holistic approaches considering
multi unit interactions and global optimisation
through the operation.
Increasing capabilities to understand rootcauses and address the problem source.
-
7/27/2019 03 Michael Schaffer
26/27
Further Out
Conversion of the models into
controllers.
Incorporation of economic objectivefunctions becomes possible optimising
the business drivers in the operation and
using those to define the process
requirements.
-
7/27/2019 03 Michael Schaffer
27/27
THE END