Content uploaded by Carlos Agudelo
Author content
All content in this area was uploaded by Carlos Agudelo on Mar 26, 2018
Content may be subject to copyright.
Copyright 2004 by ISA – The Instrumentation, Systems and Automation Society.
Presented at the ISA 2004, 5-7 October 2004, Reliant Center Houston, Texas, www.isa.org
DEVELOPMENT OF AN OPERATOR TRAINING SYSTEM
(OTS) FOR A SOLVENT PRODUCTION PLANT AT A
COLOMBIAN REFINERY
Enrique Torres Ezequiel Acosta
Lead Process Control Engineer Process Control Engineer
ECOPETROL, R&D center ICP ECOPETROL, R&D center ICP
Bucaramanga, Colombia Bucaramanga, Colombia
Carlos Agudelo Alirio Acuña
Automation Engineer ECOPETROL, R&D center ICP
ECOPETROL, R&D center ICP Universidad Industrial de Stder.
Bucaramanga, Colombia Bucaramanga, Colombia
Enrique Mejia Jorge Prada
ECOPETROL, R&D center ICP Electronic Engineer
Universidad Industrial de Stder. Universidad Industrial de Stder.
Bucaramanga, Colombia Bucaramanga, Colombia
Ramón Sánchez Francisco Morant
Process Engineer Automation Professor
Universidad Industrial de Santander Univ. Politécnica de Valencia
Bucaramanga, Colombia Valencia, Spain
KEYWORDS
Operator Training Systems (OTS), Virtual Plant Model, Distributed Control Systems (DCS),
Emulation.
ABSTRACT
This document describes the design, implementation and testing of a simulation tool for training
operators at the Colombian Petroleum Company refinery located at the city of Barrancabermeja
(GCB). In this work, a rigorous model was developed for simulation of the solvent production
plant using commercial software (AspenTech’s Hysys™), and also it was developed with the
interfaces emulating the distributed control system (DCS) of the real plant. Some of the benefits
for this system are: To guarantee the skills an operator must develop in his/her job; faster plant
startups; checking of the Distributed Control System (DCS) configuration; pre-tuning of control
loops; modifications design for operation improvements; environment impact analysis; risk
assessment.
contents
Copyright 2004 by ISA – The Instrumentation, Systems and Automation Society.
Presented at the ISA 2004, 5-7 October 2004, Reliant Center Houston, Texas, www.isa.org
OPERATOR TRAINING SYSTEMS
The operator’s training is essential in any process unit and embraces from the basics of equipment
safety operation and DCS use, to the deepest knowledge of the process dynamics. The operator
learns working on the very own unit, facing sometimes risky situations with loss of valuable
products and sometimes injuries to operations staff and damages in process equipment. This costly
operation, due to off-specifications product that must be reprocessed, operator harm and equipment
damages, can be avoided using modern techniques for operator training, speeding up the learning
process and reducing risks, using a virtual plant instead of the real one (4, 5). Some of the benefits
in using virtual plant models for operator training are:
¾ To guarantee the skills an operator must develop for his/her duties.
¾ Quicker plant startups.
¾ DCS configuration checking and pre-tuning of regulatory control loops.
¾ To validate operation instructions manual and risk assessment.
¾ To design modifications in revamps.
¾ To verify the performance of new control strategies.
¾ To analyze process startup and shutdown environment impact.
The confidence in the simulation results, is of the major importance for its use as a training and
analysis tool. The virtual plant model must be accurate and robust, replicating the different steady
states of the process and the dynamics between them. The equipment must be modeled with high
fidelity from their datasheets. Likewise, chemical compositions and physical properties predictions
of the throughput and product streams, must replicate the real process values, and must be validated
against plant tests.
PROCESS MODEL
The methodology to adjust the process model’s parameters included the following steps:
1. Gathering design information of the process equipment: Pumps, heat exchangers, control
valves, drums, separation towers, etc.
2. Field checking of the equipment characteristics.
3. Compilation of recently process running reports and selection of the data for the base case to
configure in the simulation.
4. Configuration of the model with equipment information and operational conditions of the base
case.
5. Adjustments of the parameters in each of the process equipment to reproduce base case.
6. Tuning of control loops.
7. Checking the dynamics of the model against dynamic response of the real process recorded
during control loop tuning step tests.
8. Factory Acceptance Test (FAT) and Site Acceptance Test (SAT).
Copyright 2004 by ISA – The Instrumentation, Systems and Automation Society.
Presented at the ISA 2004, 5-7 October 2004, Reliant Center Houston, Texas, www.isa.org
SOLVENT PROCESS DESCRIPTION
The aliphatic solvents are volatile hydrocarbons, produced by the primary naphtha distillation or
byproducts of the aromatics plant. GCB´s plant has two separation columns installed by M.W.
Kellogg Co. in 1963. The de-isopentanizer column has fifty drilling trays and the stabilizer column
has twenty. Both can be operated in series or separately, according to the solvent grade being
produced.
PROCESS MODEL IMPLEMENTATION
The first step in the creation of the model was the specification of the input streams. For each one
of them, it was specified their temperature, pressure, volumetric flow, and chemical composition in
molar fractions of ethane, propane, i-butane, n-butane, i-pentane, n-pentane, n-hexane, etc. Then
the Process Flow Diagram (PFD) was built using the templates for the equipment’s models and
configuring its parameters (pressure drop, heat exchange coefficient, etc.).
FIG. 1- TOOL FOR RIGOROUS MODELING.
OTS ARCHITECTURE
INTRODUCTION
The purpose of training an operator, in a similar environment to the control room, wouldn’t be
satisfied without interfaces like the ones on the real plant DCS. A platform was designed to allow
the operator to manipulate the process’s variables in the model, through the same operation
interfaces in the control room of the unit, hiding the details of the model built in Hysys™ and
acting as a virtual plant.
Copyright 2004 by ISA – The Instrumentation, Systems and Automation Society.
Presented at the ISA 2004, 5-7 October 2004, Reliant Center Houston, Texas, www.isa.org
In the industry two approaches have been explored for training operation interfaces implementation
(3, 6): It can be using stations of the same manufacturer of the DCS, having a more realistic
simulation; or it can be emulation of the DCS consoles, through computer software to decrease the
costs of the implementation.
This work included the development and test of both approaches, and concluded that DCS
emulation had better performance in a real operator training environment for our particular
requirements. As it is discussed bellow, the first architecture proposed to use DCS consoles as
trainee interfaces. This architecture required the development of a data transfer application to
connect the model, acting as virtual plant, with DCS consoles. Testing the system we found that
this approach was difficult to adapt to our necessity of training various operators (more than one at
a time) with different levels of expertise requiring particular training protocols, because the lack of
adaptability and portability of the system. Then, one more portable and adaptable architecture was
propose, which allow us to set a system to train up to five operators at the same time and to emulate
displays of different DCS brands at a affordable cost. Following below, a description of each of the
two approaches is presented.
In the design, implementation and test of the first architecture, we developed a platform to read and
write information from/to the model, using Microsoft Excel™ spreadsheet. This spreadsheet ran a
process to send and receive data through a software written in C++ which was communicating to
the DCS, where the process displays were configured. The final result was that the trainee can
watch and control the model, using the same hardware and software of the real process DCS.
FIG. 2 - PRELIMINARY ARCHITECTURE FOR THE OTS.
On the DCS side, the operator had access to operation consoles similar to the ones in the control
room:
A
W - AP
CP
WP
FOXAPI-C INTERFACE
VIRTUAL BLOCKS
OPERATION DISPLAYS
Model Data
Valve positions and
Setpoints
Engineer/Model administrator Operator in training session
PC-Windows DCS (Foxboro I/A)
TCP/IP
HYSYS MODEL
EXCEL
C++ INTERFACE
Copyright 2004 by ISA – The Instrumentation, Systems and Automation Society.
Presented at the ISA 2004, 5-7 October 2004, Reliant Center Houston, Texas, www.isa.org
FIG. 3 - TYPICAL OPERATION DISPLAY.
The first architecture evolved to include a new application in a server station, which was intended
to run the simulations, allowing the instructor and operator applications to use all their resources in
their events processing and data exchange with the server application. In this server station was
built the database with records for each operator and instructor, performance information on each
training session, and disturbance routines for the simulation model. The emulation on the operator
application was almost identical to the DCS environment. The three applications of this new
architecture exchange data through the training room network with TCP/IP protocol.
FIG. 4 - OTS FINAL ARCHITECTURE.
OPERATOR STATION
The application running in the operator station makes the operator login and registry, data exchange
with the virtual plant model, and loads the DCS emulation.
INSTRUCTOR STATION
SERVER STATION
OPERATOR STATION
Copyright 2004 by ISA – The Instrumentation, Systems and Automation Society.
Presented at the ISA 2004, 5-7 October 2004, Reliant Center Houston, Texas, www.isa.org
FIG. 5 - OPERATION DISPLAY EMULATED ON THE OPERATOR STATION.
SERVER STATION
This station runs the application with the database of instructors, operators, and their evaluations. It
also runs Hysys™ as a subroutine using OLE (Microsoft's standard Object Linking and
Embedding) and the algorithms to read/write from/to the process model.
FIG. 6 - INSTRUCTORS AND OPERATORS LIST.
Copyright 2004 by ISA – The Instrumentation, Systems and Automation Society.
Presented at the ISA 2004, 5-7 October 2004, Reliant Center Houston, Texas, www.isa.org
INSTRUCTOR STATION
The instructor station allows direct communication with the model on the server station, and with
the operators in training session. The instructor can send disturbances to the virtual plant in order
to evaluate the operator's response. He/she also can send messages to the operator through a chat
tool, to exchange ideas and suggestions. Inside this application the instructor is able to monitor up
to five operators in training session.
FIG. 7 - MONITORING OPERATORS FROM THE INSTRUCTOR STATION.
PRACTICAL APPLICATION
Using this virtual plant simulation tool, the operators, supervisors and engineers were trained in the
different operational scenarios of the process. The simulations allowed them to analyze the
behavior of the performance indicators for each equipment and their restrictions. During a training
course using the virtual plant simulation at GCB´s refinery, fifty new operators learned how to use
the process degrees of freedom in order to keep the quality of the products and the process variables
in their targets, and to be well versed in the dynamics of the unit. The following tests were made
on the model:
¾ Throughput flow variation.
¾ Reboiler temperature variation.
¾ Top reflux flow variation.
¾ Towers top pressure variation.
For each adjustment it was recorded the equipment performance, the product quality and the
product yields.
Copyright 2004 by ISA – The Instrumentation, Systems and Automation Society.
Presented at the ISA 2004, 5-7 October 2004, Reliant Center Houston, Texas, www.isa.org
GCB´s refinery has setup a training room for continuous operator training, not only for the solvent
process but also for other units. In the near future it is planed to develop OTS for crude distillation
units, Fluid Catalytic Cracking, gasoline and fuel oil blending.
REFERENCES
1. AspenTech's Seminar. Dynamizing a plantwide Steady State Model. Instituto Colombiano del
Petróleo, Bucaramanga (Colombia), 2003.
2. Shadow Simulation. Hydrocarbon Engineering. January 2002.
3. Instructor Guidelines for Use of an Operator Training Simulator (OTS). A Compilation of
Experiences and Lessons Learned from OTS Users Group Meetings. EPRI, Palo Alto, CA:
2001.
4. Reinig, Gunter et al. Training Simulators: Engineering and Use. Chemical Engineering
Technology, 21 (1998).
5. Morgan, S.W. et al. Improve process training with dynamic simulation. Hydrocarbon
Processing, April 1994.
6. Sundararaman, C. Approach Towards Selection of Operator Training Simulators. ProtechSoft
System Pvt Ltd, Chennai, India.
7. Fuentes, Carmelo. Chemical process simulation. Universidad del Atlántico, Colombia.
8. Duque, Carlos y Tassone, Vince. Dynamic Simulation for Process Engineers. Colt
Engineering Co., Hyprotech Ltd.