SIMULATION APPLICATIONS FOR ALUMINA REFINING

Freeman N1*, Pandya M2 and De Roos R3*

Honeywell Process Solutions, Perth, WA, Australia
1

2
Rio Tinto Alcan Engineering, Brisbane, QLD, Australia
3
Honeywell Process Solutions, Brisbane, QLD, Australia
Abstract
Strong demand and environmental considerations are increasing the pressure for alumina producers to operate safely,
reliably and efficiently. Traditional operational methodologies are being tested and compared to best practices in other
industries for improvement opportunities. Operator training simulation is one technology, used extensively for many years
throughout the hydrocarbon processing industries, which is now being deployed successfully within alumina refining,
especially for greenfield and brownfield expansions..
Operator training simulators utilise dynamic simulation to provide a high fidelity model of the process dynamics associated
with an alumina refinery. This has been proven to provide benefits through increasing availability (operators better able to
handle upsets), safety and operator competency certification. Additionally it enables process and control design validation
minimising design shortfalls that would negatively impact throughput, quality and commissioning times.
This paper provides an overview of this technology and the benefits with reference to an alumina refining case study – the
Gove double digester simulation project.

1. Introduction 2. Simulator Applications

The traditional view that minerals processing operations are all 2.1 Training Process Operators
about shovels and conveyors is changing. There is an increasing
awareness of the complexity and demanding requirements for 2.1.1 Using Custom Models
the efficient operation and maintenance of mineral processing A custom simulation model of the process, control and logic
plants. Coupling this with occupational health and safety places systems can be delivered which would include malfunctions,
a heavy onus of training on owners. As a result extensive use of process disturbances, training performance assessment, event
dynamic process simulators is becoming the norm for greenfield recording, third party applications and training exercises. This
facilities in metals and mining, including alumina refineries. simulation model can then be used during start-up and shut down,
Many leading companies are benefiting from the use of dynamic during normal operations and when new technologies either
simulation for greenfield projects and control upgrades because control or process are trialed. It can be specifically used to:
the inherent risk in new process plant design and the complexity
of controls in the Bayer cycle. Interactive and accurate simulation • Train operators, maintenance and engineering personnel
provides the mechanism for enhanced operator training, and for how to run the plant;
validating and improving new process and control designs. • Train new operators or replacement personnel;
• Certify operators to meet specific standards;
A correctly built dynamic simulation is a tool that can be • Maintain and upgrade operator skills on an on-going basis;
used throughout an operations lifecycle, from process and • Develop operating, maintenance and emergency
control engineering through operator training to operational procedures;
debottlenecking. In order to fully realise the benefits it is necessary • Develop and test training and operations documentation;
that a full high fidelity model of the process be built. This must be • Develop optimum operating strategies.
based on engineering first principles with full energy and mass
balance capabilities. Considerable work is required to configure 2.1.2 Using Standard Models
such a simulator making full use of kinetic, thermodynamic and
phase equilibrium data, however the paybacks can be enormous. Some vendors offer standard models which are “off-the-shelf”
A high fidelity process simulation model will provide realistic examples of typical processes which meet industry standards
hands-on experience without risk of damaging the plant. That and specifications. Although generic in nature, standard models
is the simulator has the look (displays) and feel (operations) as are realistic in how they operate and respond to changes.
the real plant. They include process specific malfunctions and disturbances,
selected user functions, and training exercises. They may utilise
For a project, the major uses of dynamic simulation can be a Distributed Control System (DCS) operator console (which
categorised as: requires an interface program), or the colour monitor of the
simulation computer.
Process engineering
Control engineering The standard models are a useful alternative when:
Operations training
• Only general process knowledge is required;
Each of these areas will provide for significant benefits, however • The simulation budget is limited;
in order to avoid three separate simulators to cover all of these • The engineering data will not be available in time to develop
the simulator chosen should provide for all these functions. custom models;
• Short delivery time is required.

52 Proceedings of the 8th International Alumina Quality Workshop • 2008

AQW2008_Proceedings_FINAL-1-120.52 52 4/08/2008 11:35:18 AM

 2.2 Process Engineering Functionality
A custom dynamic simulation model can be used during the The system must be flexible such that it spans a wide range of
engineering design phase of the process and control systems to applications, accessed through the graphical user interfaces.
vet designs and uncover potential start-up problems. It can also Each interface representing a potential user of such as engineers,
be used to assist in evaluating modifications to existing plants. instructors, and trainees. These interfaces must be able to
run the simulator, set specific exercises, trigger malfunctions
It can be specifically used to: and upsets, monitor the simulator and document operator
• Test design philosophy; performance. Ideally they would also be used to build, tune and
• Locate design flaws; update the simulation model, perform diagnostic actions, and
• Identify constraints in existing or proposed equipment; control access to the simulator. The result is a simulator that is
• Test alternate designs; effective for training, testing and operations applications.
• Develop start-up and operations plans and procedures; 4. Benefits
• Conduct HAZOP studies.
There are many benefits from dynamic simulation including:
2.3 Control Engineering
- Improved plant start up times;
Some dynamic simulators offer full emulation of control
- Reduced off-specification product during start up;
algorithms for DCS systems. This allows for the validation of
- Competent trained operators;
control strategies off-line. As well, the simulation model can
- Design check and verification;
be connected to the operator console and the control devices
- Fault finding of control code;
to completely check out the implementation of the DCS control
- Controller tuning prior to start up;
database, safety systems, graphics and alarming systems.
- Operating procedure validation;
The simulation model can be specifically used to: - Testing and implementation of control schemes, logic and
graphics;
• Test the DCS configuration; - Control System Optimisation.
• Test the regulatory control strategies;
• Test the Emergency Shutdown System (ESD) and other logic One of the major benefits is for improving operator effectiveness.
systems; In most operating alumina refineries, the equipment is operated
• Test and tune advanced control and optimization programs; continuously and many operators are not well practiced in
• Develop and optimise the control strategies. running under startup, shut down, or emergency conditions.
Similarly, in new installations, operators may have even less skills
3. Simulator Requirements in managing the process and the knowledge of the equipment
There are four main requirements for any simulator: limits, even under normal operating conditions. Fundamentally,
the essential mechanism in learning is comprehension and
Realism repetition. Dynamic simulation provides these essential learning
The system must closely represent the control room environment. mechanisms through allowing operators to practice their
It achieves this by connecting directly to an actual DCS operator operational skills without the possibility of damaging the plant.
console similar to those used in the plant control room, and by
The need for better operator training to reduce incidents is further
utilising the actual plant DCS control and display configurations.
exemplified by the Chemical Manufactures Association’s study of
The operator therefore becomes fully familiar with the keyboard,
six sites regarding the causes of incidents and found that:
screens, process displays, tag names, alarms, logic, and control
schemes before using the DCS in the plant. Causes of Incidents:
Fidelity People and work context 35 – 58%
The system must closely represent the actual performance of a
process, unit operation, area or plant. The software must enable Equipment 30 – 40%
building of a high fidelity model of the process which simulates
Process 3 – 35%
its operation and the actions of the associated regulatory,
sequential and supervisory control algorithms. The software The people and work context can be further broken down as
must have an extensive physical properties database, and illustrated (Figure 1).
perform mathematical calculations to accurately represent the
dynamic behaviour of the process or control system. The model
must be based on actual process and control information, and
include malfunctions and process disturbances. The operator
must be able to start up the model, run it in real time, shut it
down, change production grades, and experience emergencies
and alarm conditions.

Maintenance
The system must be easily maintained. This is aided if the
software used to build the simulation model is modular. The
nature of the software must allow the model to be changed as
the process changes. It should contain modules that represent
the various components of process equipment, such as heat
exchangers, distillation columns, compressors, turbines, pumps,
valves, etc. A graphical model building program will speed up *Impacted by dynamic simulation when used for training and developing procedures.
the configuration process and make the system easy to use. The
Figure 1. Causes of Incidents
system should be an open system that can be updated on an on-
going basis and easily maintained.

Proceedings of the 8th International Alumina Quality Workshop • 2008 53

AQW2008_Proceedings_FINAL-1-120.53 53 4/08/2008 2:55:08 AM

 The benefits of dynamic simulations include improved safety steps to start up, run or shut down the plant. The operator learns
and reduce environmental incidents. Tangible financial benefits how to react to alarms, emergency shutdowns and other lesser
include better response to upsets that regain quality or crises. The operator can also be taught how to optimise the
production, identify malfunctions or design errors that minimize process and make best use of the control facilities available to
downtime and production losses. Reductions in commissioning them.
time by as much as 25% have been reported.
A separate console is used by the trainee for operation of those
5. Case Study – Gove Digestion field operated devices (FODs) which are not available through the
DCS operator console, but require manipulation to allow effective
The Rio Tinto Alcan Gove bauxite mine and alumina refinery is
operation of the plant. Those activities are normally taken by a
located at Nhulunbuy on the Gove Peninsula in the east Arnhem
field operator in concert with the DCS operator. The combination
Land region of Australia’s Northern Territory. Rio Tinto Alcan Gove
of an FOD station and the DCS console allows the trainee to fully
is 100% owned by Rio Tinto a global leader in aluminium and
operate the simulated process without outside help.
packaging and one of the world’s largest alumina producers.

The Rio Tinto Alcan Gove alumina refinery has just completed a Instructors can control the simulation through the instructor
(US) $3 billion expansion (formerly known as the G3 Expansion console, which is also the FOD console in this case, and conduct
training exercises, examine and change values in the simulation
Project) which when fully commissioned will increase alumina
and create new exercises. They may execute such capabilities
production from 2 million to 3.8 million tonnes per annum. The
as malfunctions, freeze/unfreeze, backtrack, record/replay, slow
expansion represents a significant investment, delivering long
down, speed up, trainee performance monitoring, etc. Trainees
term security and economic benefits to the Gove region, and
can also access the system and perform training sessions
allowing the refinery to operate independently of its local bauxite
without instructors if desired.
reserves.

One of the major benefits from this expansion is predicted to be 5.2 The process model
a 10% improvement in recovery of alumina from bauxite and a The UniSim software contains a library of modules which
25% reduction in residue per tonne of alumina produced through mathematically represent the behaviour of process equipment,
improved digestion technology. This technology, known as double logic and control components under dynamic conditions.
digestion, uses both low temperature digestion for removal of The modules include heat and material balances, operating
trihydrate alumina followed by high temperature digestion for equations, thermodynamics and physical property calculations.
the monohydrate alumina. Gove had not previously used this These modules are used as building blocks to create a realistic
technology which therefore represented significant operational representation of a specific process, area or plant.
risk. In order to mitigate this risk it was decided to develop an
The digestion process model includes 135 tank modules, 85
operator training simulator of the digestion process using the
pumps, 1037 control valves and approximately 158 other pieces
Honeywell UniSim Operations Suite software package.
of unit process equipment such as heat exchangers. There are
The simulator was justified on the basis that it could save 386 field operated devices, mainly manually operated valves,
approximately 1 month from commissioning, from two main and 7370 control points are simulated. Training features include
areas 1242 malfunctions. The process model takes about 0.2 cpu
second to run on a personal computer and the model runs every
1. Operator Training; Rio Tinto Alcan Gove recognised that 2 seconds, which is more than sufficient to realistically simulate
there was no previous experience of running the double the process dynamics.
digestion process. Hence without a simulator, it would be
difficult to effectively train operations personnel without 5.3 Implementation
adversely affecting plant operations. The project had a very aggressive schedule meaning that a
2. Control System Commissioning; The control system standard project delivery method would not have been successful.
architecture involved interfacing between multiple DCS Hence the project was set-up as a partnership between all the
platforms which had to be managed to minimise the impact stakeholders (design, operations, control and simulation) as an
due to mismatch of data. Additionally the control system code integrated team. This enabled transparent information exchange
was developed off site (due to labour constraints and EHS which was managed through a closed loop change management
considerations). The simulator allowed for comprehensive process.
code testing before transfer to site.
This project required considerable engineering hours and inorder
In addition there would be significant benefits from: to meet the tight delivery schedule up to 12 simulation personnel
were resourced from Brisbane, Perth and Pune. These teams
- Improved Environmental, health and safety;
developed sub-models, according to predetermined dynamic
- Ongoing competency based training;
simulation standards, which were merged to form the final
- Control scheme validation and tuning;
models. Once process models were completed DCS integration
- Operating procedure development;
and testing for the training process models was done before
- Plant control optimisation.
factory acceptance testing. All of this was coordinated by the
5.1 Simulator Description Brisbane simulation personnel who worked in the Rio Tinto
Alcan Engineering offices throughout the project. This maximised
The UniSim Operations system is a direct connect, full replica,
communication with Rio Tinto Alcan process engineering, Rio
dynamic process simulator. It is an integrated collection of
Tinto Alcan management and the control system engineers.
computer and control system hardware and software which
allows a high fidelity model of the process to run in real time The project was delivered in 2 phases to coincide with plant
and appear from the DCS console as though a real plant is being commissioning. Each phase had simulation factory acceptance
controlled. In addition other features are provided which facilitate testing with the control system in Rio Tinto Alcan’s Brisbane
the use of the system as a training and instruction tool. office before commissioning on site at Gove. The low temperature
digestion section was delivered for site training 3 months before
The operator trainee sits at the DCS console with the actual
commissioning. Finally the complete digestion simulator,
process control schematics and can proceed through the various

54 Proceedings of the 8th International Alumina Quality Workshop • 2008

AQW2008_Proceedings_FINAL-1-120.54 54 4/08/2008 2:55:08 AM

 including high temperature digestion, was delivered 6 months The Gove simulator proved to be very successful resulting in a
before high temperature digestion commissioning. large number of control system improvements prior to start up
and justified the prediction of a month improvement in achieving
5.4 Benefits full production.
The simulator project was delivered to enable many months
of operator training prior to plant start up. This resulted in the In order to operate safely, reliably and efficiently it is necessary to
operators being knowledgeable on the process and the control have well trained competent operators. It is very difficult, especially
system, but perhaps more importantly on how to control the for a new operation, to obtain this without a direct connect high
process. Thus the operators became a valuable part of the fidelity simulation that replicates plant conditions and problems.
commissioning team and were able to retain the lessons learnt This has been recognised for many years throughout the
from commissioning. aerospace industries and within the high risk process industries
such as for oil platforms. There is now recognition within the
In order to enable operator training several months in advance alumina refining industry that this technology can significantly
of plant commissioning the control system had to be developed improve commissioning times and incident avoidance.
6 months earlier than would have been necessary without
the simulator. Whilst this placed an additional burden on the
control system engineers, the result was that the control system
was able to be tested and commissioned on the simulator well
in advance of plant commissioning. This resulted in over 240
improvement suggestions ranging from nice to have to critical,
examples being:

• Feed should trip on LTD trip;

• Numerous controllers tuned;
• Numerous control system graphic fixes;
• Trip setting improvements;
• Highlighted need for major high rate decanter control system
changes before the commissioning. These improvements
were made just in time;
• Enabled changes to allow use of high rate decanter flocculant
control system before commissioning.

In effect this meant that the control system was fully commissioned
prior to actual plant commissioning, allowing the commissioning
engineers to just focus on the process and equipment.

6. Conclusions
Rio Tinto Alcan Gove are in the final stages of commissioning
their upgrade project aimed at doubling the plant capacity. This
is making use of a Rio Tinto Alcan technology, double digestion
which uses both high and low temperature digestion to maximise
alumina extraction. This was a new technology for Gove and a
Honeywell UniSim simulator was constructed connected to the
distributed control system to mitigate the risk by training the
operators and testing out the control system configuration prior
to plant commissioning.

The dynamic simulation, based on engineering first principles

with full heat and mass balancing, was used throughout the
project lifecycle providing benefits in:

• Health and safety – reducing the need for people on site;

• Operator training – training on a facility with the same
look and feel as the real plant under numerous operating
conditions;
• Control System Design – Effectively pre-commissioning
before site delivery.

References
1. Don K. Lorenzo, “A Managers Guide to Reducing Human Error”; (American Chemistry Council, Washington, DC, 1990, p. 11).

Proceedings of the 8th International Alumina Quality Workshop • 2008 55

AQW2008_Proceedings_FINAL-1-120.55 55 4/08/2008 2:55:08 AM