before the synthesis gas is sent to the ammonia synthesis loop where it is converted into ammonia. The CO2 separated in the CO2 removal section is sent to the compression section of the urea plant and then to the synthesis reactor together with ammonia. Downstream the urea synthesis, the decomposition and relevant recovery of unconverted chemical reagents is carried out in three subsequent steps: High Pressure Decomposition, Medium Pressure Decomposition, and Low Pressure Decomposition. Urea is then concentrated in the Vacuum Concentration section before being fed to the granulation unit. The process condensate is treated in the waste water treatment section and is sent to battery limit.
What is important to consider for operator training is that the process contains many fast-moving unit operations (ie compressors) coupled with interactions between the process and the steam which is generated from the excess heat of the process. Every action taken by the operator will have multiple effects in different parts of the process. Some equipment failures, such as an air compressor trip, can be recovered by quick operator action, thus saving a plant trip and a much greater amount of lost production.
System Architecture
The Profertil OTS consists of the following major components (figure 5):
- Simulation Computer
- Dynamic Process Models
- Instructor Station
- Emulated Operator Stations
- Field Operator Station
The hardware consists of a Simulation Computer (Compaq Workstation) and Instructor Station (monitor, keyboard, mouse), while the Emulated Operator Stations, and Field Operator Station are standard PC’s networked to the Simulation Computer. A printer provides hardcopy documentation of the training session with various printouts such as files from instructor sessions with process reports, alarms logging, and a log of events that occurred during the training session, both instructor and operator actions.
The Compaq Simulation Computer is the heart of the system where the simulation executive, simulation models, instructor’s software, and the operator station interface or emulation runs.
The instructor can control the session, monitor all the process variables, insert malfunctions, override instrument signals, change battery limit conditions, measure the student's performance, and record the results of the training session. A user-friendly interface minimizes the need of instructor’s computer knowledge and allows the instructor to be focused only on the training session. In addition the instructor can change the time scale from 0.1 times to 10 times the real time. With this training feature the instructor can run faster operations that have been already examined or operations that need long time (filling of tanks, heating and pressurizing phases) or he can run the process slower than the real time to allow the student to carefully observe critical situations. More sophisticated instructor functions include creation of training scenarios, sending text or audio or video files to the student, and management of the student’s training. Some typical windows and menus of the instructor station are shown in figure 6.
The Emulated Operator Stations closely resemble the Foxboro I/A Series DCS including monitors, keyboards, and touchscreen / pointing devices. The actual keyboards can be used as well as console cabinetry to maximize “control room realism”. The man-machine-interface is faithfully represented for all operator related activities including operating groups, trends, alarms, graphics, keyboard functions, touch-targets, etc. The graphic displays are exactly duplicated by using a translator which uses the actual Foxboro configurations and integrates them into the Emulated Foxboro I/A format with all the functional features such as windows, control targets, etc. In addition, a significant part of the emulator includes simulation of the many and various control algorithms and functions available in the Foxboro Control Processor (CP). Lastly, the emulation software includes a set of utilities which permits the client to change the DCS configurations and control soft-wiring in a similar manner to how they are changed using the engineering configuration tools at the real DCS.
The overall result is a very high-fidelity representation of the operator station found in the control room. This product allows both the operator training to occur with maximum realism as well as engineering applications since the control algorithms are simulated in detail. The cost savings from using this approach are significant, both in the initial cost for the OTS as well as for the ongoing maintenance of the system. Figures 7a thru 7c show emulated operator stations used for several different OTS projects, and it is difficult if not impossible to discern which are emulated and which are interfaced to actual DCS. More information about specific benefits are contained later in this paper.
The Field Operator Station (FOS) was configured to include the remote or field operated equipment not available from the main DCS operator station and also the ESD/Interlock functions using graphics instead of the hardwired panel displays in the control room. Field equipment includes block and bypass valves, motor starters, local relay and switch controls and other types of digital and analog devices. The FOS was physically located near the Emulated Foxboro I/A operator stations to provide the operator more complete control of the simulated plant. Since the FOS is a subset of the instructor station the same functionality is also available at the instructor station so that the presence of a field operator is not necessary to conduct a training session.
Process Model Overview
The single most important component in an OTS is the mathematical model that should accurately simulate the dynamic behavior of the process.. The overall fidelity of the model should ensure that operators can be trained to observe and respond correctly to a variety of operating conditions.
The “high fidelity” model should be enough accurate to reproduce not only plant responses due to disturbances around the normal operating conditions, but also the dynamic behavior for non-design operations including cold start-up, process upsets and emergency conditions.
Models are derived from the First Principles of Chemical Engineering (conservation of mass and energy), thermodynamic properties, and equipment performance/design data to ensure accurate responses in all the possible conditions. Differential and algebraic (linear and non-linear) equations are developed for all plant equipment, which ensures that individual unit responses as well as complex unit interactions are preserved. Process models are constructed using the algorithm library of equipment modules. Modules that simulate unit operations include compressors, distillation sections, drivers, heat exchangers, pumps, valves, and vessels. Distillation sections available in the simulator are trayed sections and draw trays. These modules perform VLE calculations on feed streams to determine the distribution of the components in the vapor and liquid streams exiting each tray. Other modules provide instrumentation and logic functionality such as control processors, level transmitter and compressor surge controller. Reactors or specific units, for instance where a reaction can occur or there are special mass-transfers occuring, are modeled with custom algorithms and the kinetic equations are sometimes rewritten to satisfy the need for faster than real-time computation.
TRIDENT’s unit operation algorithm library, developed over many years and projects, has nearly every kind of equipment necessary for developing a custom process model.
For Profertil’s OTS project, the urea and ammonia process and the steam network sections were simulated rigorously and in detail, while other auxiliary systems such as lube & seal oil systems, cooling water and regeneration of catalyst were simulated in less detail since these areas were not the major focus of operator training.
Malfunctions and their effects were included in the model such as heat exchanger fouling, plugging, pump failure, electric power failure, cooling water supply failure, steam supply failure, natural gas supply to primary reformer failure, air to secondary reformer supply failure. All battery limit conditions such as ambient temperature, cooling water temperature, fuel gas heating value as well as catalyst activity can be modified by the instructor to enhance the training session.
The main functionality and effects of multivariable process control implemented in the real plant have been included in the model although in less detail. The goal was to provide realistic scenarios for building operator confidence in using the APC application and to enable monitoring of the main effects of the APC. Also the DCS interface to the APC modules has been replicated so that the operator functionality (i.e. switching on and off the APC modules) is exactly reproduced.
In summary, the quality and accuracy of the process model depended primarily on the experience of the simulation engineers implementing the dynamic model as well as the involvement of the process licensor during testing of the OTS.
Case study - Urea start-up
Start-up is one of the most difficult and critical operations of the plant, and training in the field could not allow new operators to witness or practice these operations. The only feasible way to build the operator’s confidence is by intensive and repetitive “hands-on” training that could be performed on an OTS with realistic, high-fidelity models.
Particular attention was paid in the development and testing of the models in these areas of operation. Close cooperation between Snamprogetti’s process engineers with deep experience of the plant start-up and the Trident/ApcTech simulation engineers was the key point for obtaining of a high-fidelity model which would respond faithfully over a wide range of conditions. All the steps of a cold start-up are reproduced, with the only assumption that all the pre-start operations have been completed, such as testing, washing and purging of lines and equipment.
To perform start-up from empty and ambient conditions, first the medium pressure section, isolated from the rest of the plant, is pressurized. This is done charging ammonia and then heating and evaporating it. The same operation has to be done in the low-pressure section. The reaction zone is then heated with steam to 160°C before introducing process fluid and then pressurized with ammonia. During the ammoniation period, the CO2 compressor, driven by a steam turbine had to be put in service. Once all these conditions are reached it is possible to feed NH3 and CO2 to the reactor. From the moment the overflow of the reactor occurs the operator has to be skillful to quickly stabilize the plant condition and reach the design operating setpoints.
All these operations should be done within the ESD/logic sequence and bypassing some interlocks, so it is very important that the model exactly reproduces also the logic and that the hardwired panel functions are available. Figure 8 shows a screen display with some operations of the start-up in high-pressure section of the urea plant are shown.
This section is heated and pressurized acting on the steam and condensate local valves (from the FOS) and then drained. The DCS operator should follow and guide the field operator through the measurement on DCS of the reactor temperature.
Case study – Disturbance from normal operating condition
Operators needed to be trained to react quickly and properly on possible malfunctions or equipment failures. It was important to define all the possible scenarios to be taken into account and simulated in the model. This was done during the development of the Functional Design Specification between Snamprogetti’s process engineers and Trident / APC Tech simulation engineers.
An example of a possible disturbance that occurs in a urea plant is condenser fouling. The presence of CO2 in the upper section of the M.P. absorber is very dangerous because it reacts with ammonia to form ammonium carbamate and as a result the downstream ammonia condenser will be plugged.
The instructor can monitor the CO2 content since he has the full visibility of all the model variables, while the operator does not have any indication of the overhead composition. When the CO2 content reaches the high critical value the instructor can decide to plug the ammonia condenser. This effect was not automatically included in the model since the instructor has more flexibility to enable this malfunction.
In order to avoid plugging it is very important to control the temperature at the top of the absorber to inhibit the reaction; this variable is the only key variable that the operator can see as an index of CO2 content.
A possible exercise to test the operator’s ability to react fast to possible disturbance is to increase the fouling of the M.P. condenser upstream the absorber or to increase the cooling water supply temperature. The temperature at the top of the absorber increases and the operator should increase the reflux of fresh ammonia from the ammonia receiver to maintain the desired value. The DCS display with the absorber top temperature trend is shown in figure 9 for this kind of test where the operator increases the reflux flow rate to control the temperature.
Cost Benefits & Discussion of Emulated Operator Stations
Profertil had a choice of operator stations for the OTS: using emulated Foxboro I/A operator stations or having the OTS interfaced to real Foxboro I/A DCS equipment. Emulated Operator Stations were chosen for a number of reasons, including:
- Very realistic representation of the Foxboro MMI and controls,
- Lower initial investment cost and reduced longer-term maintenance costs,
- More functionality (multi-user capability, engineering applications)
The only way to truly evaluate if the emulated approach has sufficient realism for the OTS is to actually see the product. The keyboard and pointing device usage, screen displays, etc. should be nearly identical to the actual DCS. The emulation should preferably be configurable with utilities for both translating the graphics from the real DCS configurations as well as entering and modifying the basic DCS database. Of equal importance and often overlooked, the control algorithms should be included in as much detail as possible.
While realism and robustness of the operator station is important, the amount of training on the DCS usage represents only 20-30% of the total training on the OTS, with a far greater amount of training focused on the process model (ie startup, shutdown, emergency scenarios, etc.). From a cost / benefit basis, it makes sense to use emulated operator stations and invest the cost savings into the dynamic model.
The cost savings from use of emulated operator stations is significant, depending on which DCS and the availability of interface products. For example, a typical set of Foxboro I/A DCS hardware costs about $100,000.USD plus (depending on the specific interface approach used) an additional $100,000 to $200,000 for communication software and software CP’s. The total cost for two (2) Emulated Operator Stations is about $30,000 to $40,000. so the cost comparison is easy for Foxboro I/A as well as for most other major DCS’s.
Occasionally buyers will have concerns regarding future updates of the DCS and incorporating the same into the OTS. Ironically, the emulation has far less to be updated primarily because so few of the DCS releases actually impact the operator interface. On the other hand, an update to the DCS version / software could require extensive software updates to the interface software.
Conclusions
Many benefits were achieved with this OTS project which included both operator training and engineering.
The operator training simulator was successfully utilized to teach operators how to operate in critical conditions and to run the plant in a safe manner. It was also used to continuously test the operator performances in critical plant conditions, such as plant start-up, normal and unscheduled shut-down that otherwise with traditional on the field training are seldom tested. The results achieved were significant and could only be obtained through the intensive use of the OTS. As part of the training program, operators were trained in 8-hour shifts over a 24-hour period to simulate actual plant operations and shift changes.
Plant engineers practiced and witnessed many different plant scenarios, conducting what if analysis and monitoring the dynamic process behavior.
While it is difficult to quantify the exact savings, typically one unscheduled shutdown, trip, or plant upset can be avoided each year. On this basis, the OTS essentially pays for itself each year. If a serious event or equipment-damaging accident can be avoided, the simulator pays for itself many, many times over.
Results from several OTS users show that simulator training greatly accelerates the “real-plant” experience levels of new operators. Several studies show that new operators gain the equivalent of six (6) or more years of process experience from OTS combined with O-T-J. Snamprogetti also achieved benefits from their involvement in the OTS development. Changes were made to the DCS configuration before the DCS commissioning thanks to the use of the OTS. Engineers discovered, through the DCS emulation, minor deficiencies or inefficiencies in the DCS that were not identified during extensive testing carried out at the DCS Factory Acceptance Test. This experience was also important for Snamprogetti Licensor know-how and familiarization with dynamic simulation tools.
Future Developments
As the cost / benefit ratio for OTS continues to improve, there will be additional emphasis on OTS, both from cost-justification to safety to government regulation. Instructors will have more tools to conduct training and objectively measure operator performance. Models developed for “training” will be used for multiple applications including expanded engineering applications, control studies, and plant design. In addition, actual plant data can now be used to initialize the OTS to previous plant conditions for training.
Additional data on training benefits, resultant cost savings and improved plant performance criteria from all OTS users would greatly help others to evaluate and quantify the specific benefits which can be obtained from an OTS.
References:
- Chemical Manufacturer’s Association, “A Manager’s Guide to Reducing Human Error”, July 1990
- Hydrocarbon Processing, “Major Fires and Explosions Analyzed for a 30-Year Period”, W.G. Garrison, September 1988
- Hydrocarbon Processing, “Loss Prevention – What it is and isn’t”, September 1997 (ref. “Large Property Damage Losses in the HCI – A 30-Year Review”, J&H Marsh & McLennan, 1997)
- Power Engineering, “Simulator Reduces Cogen Plant’s Start-up and Training Costs”, J.H. Brewer & C.T. Gaines, Westinghouse Electric Corp, May 1992
- NPRA Computer Conference, “Dynamic Model Supports Control, Engineering, and Training for Crude Unit Operations”, Gary Reiley, et al, November 1995, Raytheon Engineering & Constructors
