The simulation was developed for and installed at Barrick’s Pueblo Viejo site. The simulation system was purchased through Lakeside Process Controls, who also provided project management support. The system was developed by Mynah Technologies
Barrick is a global company, with operations in North America, South America, the Australia-Pacific region, and Africa. Barrick currently has 25 operating mines, and employs about 20,000 people worldwide. In addition, Barrick has development projects and exploration activities in various parts of the world Barrick also has the largest reserves in the industry, with about 140 million ounces of proven and probable gold reserves, 6.5 billion pounds of copper reserves and 1.07 billion ounces of silver contained within gold reserves 1 as of December 31, 2010.
Who am I? My name is Paul Yaroshak. I was the project manager for the simulation-based training project at the processing facility that is the subject of this presentation. Why are we here? This presentation will attempt to relate to you our experience of implementing a dynamic process simulator at a new facility, and hopefully provide you with a little insight that you might be able to apply to your own facilities. Barrick is building a mine and mineral processing facility in the Dominican Republic that will produce gold, silver and copper. The capital cost of the project is approximately $3.5 billion with annual revenue expected to exceed $1 billion. The site is named Pueblo Viejo. Gold was first extracted from this area in the 15 th century, shortly after being discovered by Christoper Columbus. A major concern for this project was how to prepare and train plant operators for operation before the plant was actually operational. To address this concern, Barrick conceived the Operator’s Ready for Startup (RFSU) Training Process. The RFSU program prepares operators for safe and efficient plant startup.
The dynamic process simulator is one component of the multi-faceted Ready for Startup program being implemented at Pueblo Viejo The Ready for Startup program encompasses 3 key aspects of operator training, Their skills through field demonstrations and practice of tasks as per procedures Enhancing knowledge through the process fundamentals Computer Based Training (CBT). This is similar to textbook learning, however it is more interactive as the computer-based modules incorporate animations and immediate feedback Providing experience through the use of Simulator Based Training (SBT). Field demonstration and practice: where applicable, a field demonstration followed by hands-on practice by the employee is conducted.
The processing plant is complex as it is a series of several sub-processes that each present their own unique operational challenges. There are over 20 diverse sub-processes, ranging from crushing and grinding to high pressure oxidation and electrowinning.
Simulated areas are shown in yellow. Due to economic and time constraints, not all process areas were simulated. To determine which areas to simulate, we had to consider the area complexity, criticality and how representative it was of other processes. For example, the Limestone Crushing was simulated but the Primary Crusher was not, as Limestone Crushing is representative of a crushing process in general.
As mentioned in the introduction, Barrick determined that a dynamic process simulator should address many of the training concerns associated with the plant startup. Meeting aggressive startup / production deadlines Train a mixed labor pool with a wide range of processing experience levels Ensure safety of the employees. Mitigation of equipment damage risks (e.g. Certain severe service valves can only be stroked a few times, we need to minimize stroking valve unnecessarily) Qualify resources for a wide range of roles within the facility
Once we determined that a simulation was required, the next challenge was to select the appropriate simulator to meet our needs. There are several dynamic simulator systems available on the market. Barrick determined that a low to medium fidelity simulator was sufficient for training purposes (I’ll talk more about simulation fidelity in the next slide). We decided upon the Mimic dynamic process simulator from Mynah for the following reasons: Provides required levels of fidelity (complexity). Integrates seamlessly with DeltaV Mimic was already being used by the project engineering firms for checking IO. Easy to use and change so that our engineers can make modeling changes quickly and efficiently (avoids system becoming shelfware over time) Intuitive
An important aspect to consider when evaluating simulation packages is the simulation fidelity provided. Simulation professionals talk of 3 levels of simulation fidelity: low, medium and high. The Mimic system provides simple and robust low fidelity simulation, with Medium and High fidelity available as required. We implemented selective fidelity to provide a realistic operator experience on the training system, while at the same time managing time and cost constraints (higher fidelity = higher $) Studies have shown that, in most cases, increasing the complexity of the process models past medium fidelity do not result in more effective solutions. In addition, medium fidelity models provide a good compromise of realistic simulation, performance and maintainability. Low Fidelity Low Fidelity models utilizes basic I/O tiebacks with simple tuning where it has been deemed sufficient to provide responses for flow, temperature, etc. that are directionally correct for the given process unit. The tieback tuning will consist of a filter and gain whose values can be provided up front by a project team or will otherwise be set by MYNAH simulation developers based on their best judgment of what is appropriate. Precise material and energy balances are not necessary. Medium Fidelity Medium Fidelity models incorporate chemical engineering first principles of conservation of mass and energy in critical process areas to create a more accurate level of process response. Material and energy inventories (levels, pressures, and temperatures) are connected by mass and energy transfer equations in the form of driving force versus resistance. Material balances are tracked overall (gas, liquid, solid) and selective components may be individually tracked. Empirical relationships may be used where appropriate in place of first principles modeling techniques. High Fidelity A key difference between a high and medium fidelity model is in the formulation of mass and energy transfer equations. A high fidelity model will typically include rigorous solution of the various material and energy balances both at steady state and dynamically via differential or difference equations, while a medium fidelity model will attempt to combine and approximate the rigorous equations into a single or small set of simplified driving force / resistance equations. Very few if any empirical relationships are used. Individual components are tracked in main process streams. The model fidelity approach for the Autoclave simulation has been identified to be a combination of low and medium fidelity simulation with some elements of high fidelity models (e.g. reaction kinetics included for the autoclave compartment simulation
DeltaV Simulate: Simulates DeltaV controllers on a standard PC. All DeltaV Logic and Operator Graphics remain intact. Mimic Simulation: Simulates IO to the controllers in DeltaV Simulate. Process models within Mimic simulate the IO to provide a realistic process response. The DeltaV controllers are simulated within a standard PC by the DeltaV Simulate software The transmitters, final control elements, and the process itself are simulated by the MiMiC The combination of DeltaV Simulate, Mimic, and the actual DeltaV control logic and graphics creates a complete virtual plant environment for training and testing purposes. **Original Text** The operator graphics and DeltaV control logic are the same as in the production system. This way, logic and graphic design is also tested and evaluated via the simulation process. The dynamic, off-line simulator is built to provide a virtual control system and plant equivalent to the on-line control system and process in operation and response. In the real plant we have a unit operation, like a crusher or a mill. In order to operate the crusher safely and profitably we use a control system like DeltaV with transmitters and final control elements. In the virtual control system we use DeltaV Simulate to emulate the operator stations, engineering station, process controllers and higher level system functions. DeltaV Simulate provides an environment where the control system can run in an identical manner as in the actual plant. The transmitters, final control elements, and the process itself are simulated with MiMiC. MiMiC provides complete IO simulation and an environment where the development of complex, dynamic process models is quick and easy.
The Operator Training System architecture incorporates the DeltaV Simulate control system emulation software with MYNAH Technologies’ MiMiC simulation development software. The system architecture includes 4 independent DeltaV workstations and a MIMIC server machine. This architecture allows the instructor to launch 4 independent and simultaneous simulation training scenarios. The instructor can run the scenarios and introduce malfunctions from the MIMIC server machine.
Image of the DeltaV Graphic for the Limestone Crushing circuit in the simulation system. This graphic is identical to the graphic on the production system. It should be noted that the original graphic delivered to Pueblo Viejo by the engineering company has already been modified as a result of errors and enhancements revealed through the use of the simulator.
This screenshot is from the Mimic software itself. It is used by the trainer to monitor the process, modify model parameters and introduce malfunctions. In this example, the trainer can activate a pullcord on the conveyor, which triggers multiple interlocks that the operator must respond to. In our implementation, we are using the trainer screens are used to Introduce Ad-Hoc Failures Permit the trainer to view and adjust model parameters from one screen. This is much simpler than opening and modifying the model in the Mimic development environment. From this screen, it’s also possible to take snapshots of the process and launch the Operator Training Manager to automatically introduce malfunctions, however we aren’t currently using these features.
Challenges this project Structure / reality Multiple stakeholders Internal – Site representatives and Corporate External – 2 different EPCMs (Hatch and Fluor) responsible for different plant areas Distributed team We were in 5 Locations (2 offices in Toronto, Vancouver, St Louis, and the Dominican Republic) Technology ramp-up – Getting to know what and how to define requirements Learning the product capabilities and becoming familiar with the tools available. This was important to know how to best use the available toolset. Needed to ensure the system being built was aligned with the business requirements. Manage evolving requirements and scope creep
The challenge of developing a system with a geographically distributed team shouldn’t be underestimated. We demonstrated that, with proper management and the right collaboration tools, this can be successful. A few key face-to-face meetings are essential, such as the kick-off meeting and the Site Acceptance Test. We also had a single Factory Acceptance test midway through the project.
In the waterfall method of project execution, each of the phases has defined entry and exit criteria. This methodology works well when the requirements are frozen upfront and they are well documented without any ambiguity. The waterfall method was used to budget the project and develop preliminary schedules. This made creation of time lines and handoff points clean and easy. From Wikipedia: The waterfall model is a sequential design process, often used in software development processes, in which progress is seen as flowing steadily downwards (like a waterfall) through the phases of Conception, Initiation, Analysis, Design, Construction, Testing, Production/Implementation and Maintenance.
In reality because of the unavailability of key inputs, it became necessary to do one of two things: Hold the project until inputs were ready – shifting out timelines not good Re-organize our work to an iterative process to accommodated inputs In my (Kevin) mind I was thinking iterative as I knew it would driver ownership of the solution, force hands on time with the system, get the project team aligned Our partners were flexible and supported the change to iterative by accommodating the shifting. Reality in software projects is that you don’t know what you don’t know and you have to either document like crazy or build and refine in an iterative manner. Documents only go so far to expressing the solution. Message: With each circuit the number of circle backs decreased The form, accuracy, timeliness, and completeness of inputs improved over the project as both teams learned what was successful and understood how to communicate requirements Project deliverables took shape early when there was a lower cost to change direction (Not until after the 3 rd circuit were things stabilized ) We (Barrick) took ownership early on largely because Paul had the interest and need to review at a detail level to help the rest of the business review and understand what the system was to do.
We quickly moved to an iterative development approach since many of the required inputs weren’t ready upfront. We had a total of 10 areas to simulate, but we started with the simpler areas to familiarize everyone on the project team (Mynah, Lakeside, PVDC) with the simulation capabilities and complexities. We then incorporated what we learned from the early simulations in the subsequent, more complex simulations.
A model development process evolved early in the project that had the following steps: Develop an Excel model – this depends on having a metallurgist who is very familiar with the process and has an expert level of Excel skills. Provide DeltaV Configuration – ideally the final DeltaV configuration, unfortunately this was rarely the case Mimic Software: Perform Simulate Conversion and Simple Tie-Backs Medium and High Level Modeling Trainee Screen Development Initial Review Incorporate changes Final review
Early simulation work drives organizational focus on human factors Simulation is a key enabler to early identification and correction of system deficiencies. This contributes to improved utilization of time, as there is considerably more time to review and implement changes before commissioning than during commissioning. Helps identify basic problems with logic, such as interlocks with circular logic. Reveals control issues that are hard to detect on P&IDs, such as the behaviour of interacting controllers Reveals opportunities for adding new control strategy where previously not considered. Excellent tool for reviewing and diagnosing problems with HMI. A huge benefit was realized regarding the review Sequences Standard Operating Procedures can be developed and verified against the simulator, instead of waiting until startup to review the SOPs. Simulation is an indispensable tool for communication during plant development. It fosters creative discussions and communication between various stakeholders, such as engineers, control room operators, system integrators, operations supervisors, etc.
Low Level (tie-back) simulation of the complete DeltaV area should be considered, since breaking up a large area into smaller sections could actually cause more headache than it’s worth. * We have some displays that don’t work fully because of boundaries We must force some data via the simulator to suppress alarms We are cutting out many blocks from the exports to make things work Consideration for Alarm Management (elimination of Nuisance alarms) is required and is non trivial in the development of a Simulator to properly train users to respond to alarms* Developing good first principle models (excel) enables process experts not familiar with simulation software a chance to develop the solution and make informed decisions* Iterative design process is a must to deliver high value system. One must account for this in estimation and expectation management* Early engagement of the end users and key stakeholders (i.e. operators, supervisors, etc.) in the testing, reviews and project activities improves system and ownership.* Leveraging of the Simulation as a tool to support training and operational documentation (SOPs, Training Certification) development 8. We demonstrated that a complex simulation project can be completed with minimal face-to-face interaction between client and developer. In other words, simulation can be developed and implemented remotely to a large degree. In our situation this was necessary given the geographical separation of the project team members (Toronto, St. Louis, Dominican Republic). Internet-based collaboration tools such as Webex, email and instant
Source code version control of simulation logic, especially when developing in multi-developer environment, would be very beneficial. Snapshots are a great idea but don’t work well when DeltaV database is changing. Empirical data is often required for truly realistic simulations. However, implementing empirical data in Mimic is sometimes cumbersome. This is one area where Mynah could improve on their offering
Implementation of a Simulation-Based Operator Training System atBarrick Pueblo Viejo October, 2011
Company Background 3 Companies involved in this project Barrick Gold Corporation Pueblo Viejo Site Lakeside Process Controls Local Business Partner of Emerson Process Management Mynah Technologies Simulation Software Developer and System Integrator
Barrick Gold: A Global Company Pueblo Viejo Dominican Republic
Introduction Challenges New processing facility, not yet built. Need to prepare operators as early as possible. Objectives Safe Startup Efficient Startup Solution “Ready for Start Up” program (RFSU) Dynamic Simulation is a component of RFSU
Introduction The “Ready for Start Up” Program (RFSU) Prepares operators for safe and efficient plant startup 3 components: Training on Process Fundamentals using computer-based interactive training modules (textbook). Simulation-Based Training Field demonstration and practice
Project Background Pueblo Viejo will produce 1 million ounces of gold per year The plant will process 24,000 tons of ore /day at full production capacity. DCS platform is DeltaV Version 10.3, approximately 11,000 DST. Complex plant with many processes, requires a variety of skills and experience.
Pueblo Viejo Processing Site Iron 1 Thickener Autoclaves Acid Wash / Stripping Cyanide Destruction Limestone Copper Carbon In Crushing Recovery Leach
Business Need Prepare operations team for safe and efficient plant startup (RFSU) A dynamic process simulator should provide the following benefits: Meet aggressive startup / production deadlines Process understanding and familiarity with control system. Hands on practice of SOP Promote safe production / Mitigate risk of equipment damage Qualify resources for a wide range of roles within the facility Build organization capability to leverage the system for future training needs
Simulator Selection Several alternatives were evaluated. Low/medium level of fidelity (complexity) is sufficient for training simulation. The Mimic Dynamic Simulator from Mynah was selected Provides required levels of fidelity (complexity). Integrates seamlessly with DeltaV Project EPCMs already using Mimic Relatively easy to use and change (for implementing model changes and control system enhancements) Intuitive
Simulation Fidelity Select Simulation Fidelity (i.e. complexity) needed to achieve business results Low Fidelity – Basic I/O tiebacks with simple tuning Medium Fidelity – Engineering first principles, conservation of energy and mass High Fidelity – Mass / energy transfer equations, differential equations, empirical models, etc. Fidelity Breakdown on Pueblo Viejo Simulator Low Fidelity – 70% Medium Fidelity – 20% High Fidelity – 10%
Geographically Distributed Team BarrickBarrick Toronto, Canada(Pueblo Viejo)Dominican Republic Hatch Toronto, CanadaFluorVancouver, Canada Lakeside Mynah Toronto, Canada St Louis, MO
Proposed Project Design Process Waterfall Model: Sequential Development Process
Project Reality • Late changes • Control strategy errorsCircuit 1 • Graphic updates • Missing definitions • Model misalignment Circuit 2 Circuit 3
Why an Iterative Development Approachworks Lower risk; the waterfall is higher risk Early risk mitigation and discovery Accommodates and provokes early change; Manage complexity Confidence and satisfaction from early, repeated success Early partial product Relevant progress tracking; better predictability Higher quality; less defects in final product Final product better matches true client desires Early and regular process improvement Communication and engagement required “I’ll know it when I see it” required
Simulation Development Develop an Excel model Provide DeltaV Configuration DeltaV Simulate: Perform Simulate Conversion MiMiC: Create I/O Tie-Backs MiMiC: Implement Medium and High Level Models MiMiC: Develop Trainer Screens Initial Review Incorporate changes Final review
Business Results Achieved Expected Reduction is Commissioning Time Early identification of issues Enhanced familiarity with system complexities. Unanticipated benefits Early identification of usability and logic issues. Circular Interlocks, incorrect alarm limits Identification of HMI errors and issues Sequence development/testing Development of SOPs Improved training material Trainers have a working system to develop on and not via memory and documents alone
Lessons Learned Plan on Low Fidelity simulation of the complete DeltaV Areas Alarms need to be managed (to avoid false alarms). Develop first principle models in Excel. Consider iterative design process from the start. Early engagement of the end users (get buy-in) Distributed Project team can be successful Develop a plan early with system integrators to implement identified Control Strategy and HMI issues.
Opportunities for Improvement Version control of simulation system (model logic & trainer screens) Snapshots that better accommodate a changing DeltaV database Improved support for empirical data in Mimic e.g. data tables
Summary The project had unforeseen benefits of helping us Improve the final production control system Increase general understanding of our process. Approaching the project in an iterative fashion enabled our team to succeed Process simulation enabled the business to achieve the desired results Trained Users Developed Operations Documents Early adoption and entrenchment of the Process Control System
Special Thanks To Mynah Barrick Adisa Shaljani Benoit Bissonette Todd Anstine Lakeside Hatch Jorge Jimenez Will McCombe Andy McEwen