Anda di halaman 1dari 4

248 IEEE REGION 8 SIBIRCON 2008

Use of Simulation for Development of Process


Control System
Victor Okol'nishnikov, Alexander Zenzin
Design Technological Institute of Digital Techniques
of Siberian Branch of Russian Academy of Sciences,
Novosibirsk, Russia, okoln@kti.nsc.ru

Abstract— The use of simulation at different stages of life simulation, visual interactive simulation, real time simula-
cycle of process control system is described. The application tion.
of this approach to the development of distributed process There are various implementations of the simulation
control system of North-Muisk railway tunnel (Baikal-Amur system Mera. The sequential implementations are realized
Railroad, East Siberia, Russia) is considered. The develop- for Windows [2] and in Java. The distributed implementa-
ment of distributed process control system was carried out tions are realized for Supercomputer MBC 1000/M (Rus-
with the help of a simulator. The simulator contains the sia) [3] and for a network of workstations.
following programs and models: an operator interface, con-
trol programs, a model of processing equipment, a model of The models in Mera can be executed at real time or
railway traffic, a model of the outside world, and a model of faster. A scale of acceleration is defined in model settings.
microclimate inside the tunnel. The simulator was used for There is a capability to set different scales for the different
following purposes: debugging of control programs in test periods of simulation time in Mera.
bed, optimization in test bed, training of operating person- The complex of models uses a part of software of the
nel, and testing of control programs in real-world system. process control system, in particular, the operator work-
The models were executed within the framework of distrib- station for interaction of the operator with model and
uted simulation system Mera. Mera is the C++ based proc- visualization of results of simulation.
ess-oriented discrete simulation system. There are various
implementations of the simulation system Mera. The se- II. REAL-WORLD SYSTEM
quential implementations are realized for Windows and in
Java. The distributed implementations are realized for Su- North-Mujsk railway tunnel is located in East Siberia in
percomputer MBC 1000/M (Russia). A new implementation Russia. It is a part of the Baikal-Amur Railroad. The
of the simulation system Mera in Java – transportation si- length of North-Mujsk railway tunnel is 15.3 km. It is the
mulation system also is described. longest railway tunnel in Russia. The tunnel has two por-
tals and four ventilating shafts.
The process control system should control the work of a
I. INTRODUCTION set of units of processing equipment (heaters, centrifugal
One of problems of development of process control sys- and axial fans, valves, gates, pumps, and switches). The
tems is the impossibility of complete debugging and test- total quantity of controlled and inspected processing
ing of control programs neither with the help of instru- equipment is more than 450 units. The process control
mental tools nor on the real-world system. system handles more than 2500 input and output parame-
ters. The process control system should support a condi-
The most suitable method of this problem solution is
tion of a microclimate inside the tunnel within certain lim-
the simulation. A means of this problem solution is a
its and ensure safe passage of trains through the tunnel.
complex of models integrated into process control system.
By this is meant that models can use interfaces of subsys- The process control system has two levels: the top lev-
tems of process control system and models can be exe- el, and the bottom level. There are Pentium II (III) work-
cuted as components of process control system [1]. stations (Operator workstation, Signaling workstation,
Engineer workstation, and Server) at the top level. The top
The development of distributed process control system
level works under Windows NT operating system. The
of North-Muisk railway tunnel is described in this paper.
following functions are realized on workstations at the top
The development of distributed process control system
level: data collection, data storage, and data visualization.
was carried out together with development of a complex
of models. The models were used for debugging and test- There are programmable logic controllers (Octagon
ing of control programs for distributed process control Systems, Fastwel), input/output modules (Grayhill) con-
system of North-Muisk railway tunnel. Now the distrib- nected to sensors and actuators at the bottom level. The
uted process control system of North-Muisk railway tun- bottom level works under QNX 4 operating system. Con-
nel is put in operation. trol programs are carried out on the programmable logic
controllers.
The models were executed within the framework of dis-
tributed simulation system Mera. Mera is the C++ based There are two modes of control in the process control
process-oriented discrete simulation system. The simula- system: the operator remote control, and the automatic
tion system Mera provides following capabilities: interac- control. In the case of the operator remote control the op-
tion of processes between themselves with the help of erator chooses a single unit of processing equipment and
message passing, building of hierarchical models, dy- sends a control instruction to it. For example, there are
namical change of the structure of the model, distributed two instructions for a centrifugal fan: turn on and turn off.

978-1-4244-2134-3/08/$25.00 ©2008 IEEE


VICTOR OKOL’NISHNIKOV AND ALEXANDER ZENZIN: USE OF SIMULATION FOR DEVELOPMENT 249

It is required to fulfill a sequence of similar operations in speed of air flow) in relation to states of processing
order to turn on a group of fans. equipment, environmental conditions, and railway traffic
In the case of the automatic control the truth of some The certain values of these parameters initiate auto-
conditional expression initiates automatic start of the con- matic start of control programs. Control programs change
trol program that performs a group operation (for exam- states of processing equipment. New states lead to the
ple, turning on a group of fans). Operands of the condi- change of parameters of tunnel's microclimate. The
tional expression are current or average values of in- scheme of simulator is shown in Fig. 1.
spected parameters of the process control system (tem-
perature of outside air, contents of radon in air inside the Operator Model of
tunnel and so on). outside world

III. SIMULATOR FOR PROCESS CONTROL SYSTEM


A simulator was realized for process control system of Control Model of Model of
programs tunnel railway
North-Muisk railway tunnel. The simulator contains the i li ffi
following programs and models: an operator interface,
control programs, a model of processing equipment, a
model of railway traffic, a model of the outside world, and Model of processing equipment
a model of microclimate inside the tunnel.
Control programs are developed with the help of the Fig. 1. The scheme of simulator.
program constructor [4]. The program constructor is a The simulator was used for following purposes:
PLC programmable system. The program constructor is
intended for development of control programs not in the Debugging of control programs in test bed. The
programming language but in the simple specification model of the outside world and the model of railway traf-
language with the help of the special interface. fic were executed with different input data for comprehen-
sive debugging of control programs.
The program constructor contains the following com-
ponents: a simple friendly window interface, an archive Optimization in test bed. Control programs were mod-
for storing ready control programs and subroutines, a code ified for problem solving of optimization using a closed
generator for automatic development control programs circuit (Control programs, model of processing equip-
and conditions of their automatic start in C++, and a report ment, and model of tunnel's microclimate).
generator for automatic generation of the text document in Training of operating personnel. The operator inter-
Microsoft Word. face was used for a training of operating personnel of
In the first version of the program constructor the pos- process control system.
sibility of generation of the control program for fulfillment Testing of control programs in real-world system.
on one computer was realized. It was required to divide Testing of process control system in real-world system
control programs into off-line parts and to distribute these was realized under conditions of partial availability of
parts on various controllers for saving partial capacity for processing equipment. A part of processing equipment
work of the process control system in case of a failure of a was real. Another part was replaced with the model of
local network and for decrease of loading of the local processing equipment.
network. This possibility is realized in the second version A use of the model of tunnel's microclimate in operat-
of the program constructor. ing circuit of process control system is supposed later. The
The parts of a control program are executed on different model will be used for prediction values of inspected pa-
controllers. The user develops the control program as the rameters depending on external events and actions (or
whole. The program constructor automatically divides a inaction) of operator.
control program into parts and arranges them on control-
lers. IV. SIMULATION OF TRANSPORTATION SYSTEMS
The operator interface allows using the operator work- This section describes a new implementation of the si-
station for interaction of the operator with the simulator mulation system Mera in Java – transportation simulation
and for visualization of the run of simulation. Control system (TSS). There is a possibility of development of
programs in the simulator are the same ones as in the models both in the input language, and with the help of
process control system, but they are fulfilled in simulation the special interface in the system. This interface is in-
environment. tended for development and execution models of complex
The model of processing equipment is a set of models transportation systems. The interface is a part of a visual
of processing units of equipment (heaters, centrifugal and interactive transportation simulation system. An applica-
axial fans, valves, gates, pumps, and switches). The model tion for logistics simulation in a warehouse distribution
simulates state of these units and performs control instruc- center is developed with the help of transportation simula-
tions from control programs to these units. tion system.
The model of railway traffic implements real-world The main window of applications of TSS contains from
railway timetable through the tunnel. The model of the the next components:
outside world simulates change of surrounding tempera- • Play Scene. Play Scene is a region of application
ture, pressure, and wind's speed in winter. main window where model animation is shown. Sys-
The model of microclimate inside the tunnel simulates tem builder can interact with Play Scene by selecting
radon evolution. The model calculates values of parame- and moving elements of model.
ters inside the tunnel (temperature, pressure, direction and
250 IEEE REGION 8 SIBIRCON 2008

• Toolbox. Toolbox contains buttons. These buttons set using a scale that system builder defines. Play Scene
can be used to select any available simulation ele- can scroll using two scrolls or by dragging the Play Scene.
ments for further addition of correspondent elements An active element will be referred below as a subject.
of model to the Play Scene. Some subjects can present transportation facilities (cars,
• Details panel. Details panel is a panel that contains a trains, trolleys, and so on). They have graphical images
table of type “name”-“value” filled with the parame- and can move in Play Scene. System builder can send
ters of the selected element on Play Scene or with commands to the subject.
Play Scene settings. Sets of values are object- The subject can represent an object that can be sched-
specific and values can be complex, managed by the uled for the execution by the simulation system. (For ex-
special dialogs. ample in logistics simulation subject is a transportation
• Action panels. Action panels are: simulation watch, facility (trolley). Also subjects in a logistics simulation
disposition panel, and status string. are: task sources (so called pickup elements) and task ac-
Usually model can contain elements of various types. ceptors (so called deposit elements). Pickup will produce
These types are defined in a library that can be loaded tasks using some, configured for it, schedule. Trolley is a
separately. After library loaded: each library object type is facility to move these tasks to the deposit objects.)
presented by a button in toolbox, and by one string in Each subject of the model has own process and is in-
“Add” menu element. Therefore any number of libraries cluded in the simulation system. Subjects execute in speci-
can be developed for this TSS framework. fied model time moments. Subjects can delay, schedule or
Elements of model are divided into passive and active reschedule its executions in model time. Simulation sys-
ones. A passive element will be referred below as a block. tem is responsible for the model time advance, synchroni-
The block represents a part of the transport network. It can zation, and subject's execution.
be a part of a road net in the automobile transportation Originally subjects are terms of Mera. By Mera terms
model. It can be a railway in a junction model. It can be a subjects can be connected to other subjects or blocks by
flight strip in an airport model and so on. There are the means of ports: each subject can have a number of input
following types of blocks in the library: and output ports. Two subjects (a sender and a receiver)
can be connected in such a manner that output port of the
• Straight block. This is a simple straight part of the
receiver is connected to input port of the sender.
way. Straight blocks are drawn using two points.
There is a possibility to change the angle and length The sender can send a signal (any data structure) to its
of the straight block by dragging one of the points. output port and the receiver will obtain this signal. Ports
are used to simulate any real-world connections and
• Curved block. This type defines blocks in building communications between any two subjects.
the curve road that can be rounded using three
points. Also TSS has the abstraction of the Disposition Man-
ager. The disposition manager is a special subject. The
• Merge block. This is a connection of two ways in disposition manager has a special presentation in the main
one. This type defines block that contain three window (disposition panel) and control of the disposition
straight or curved blocks. Only one transportation of model.
facility can be in merge block at any time. When
two transportation facilities arrive in merge blocks The disposition manager is connected to each subject it
simultaneously one transportation facility will re- needs to manage: by separate port. It examines states of
ceive a ticket to move and the other transportation elements and handles them. There are three levels for
facility will engage stop. handling of a model disposition:
1. Distributed logic. The logic is placed in subject proc-
• Diverge block. This type defines blocks that contain
three straight or curved blocks. Diverge block will esses.
allow to split one road into two roads. When a trans- 2. User logic. System builder interact with model with
portation facility arrives in diverge block its further the help of the disposition panel.
movement depends on current active exit block. 3. High-level managing logic. The logic is placed in
Current active exit block can be switched from the disposition manager process. The disposition manager
disposition of model. defines and resolves problems with movement of more
• Intersection block. This type defines blocks that con- than one transportation facility. It implements some sce-
tain four straight or curved blocks. The intersection nario of model execution (timetable for example).
block permits to two roads to be intersected. If one A user can construct a model using such tools as of
of roads is passive then transportation facility on it adding, removing or editing of the model elements.
stop. The road can be done active or passive from The user can start a model. Simulation system will se-
the disposition of model or by Switch button. lect subjects that are registered for the time moment equal
• Pickup block. This type defines blocks that simulate to the start time, and will run them.
appearance of a new transportation facility in a A model run will be completed in the following three
model. Transportation facilities will be created upon cases: a following event is not scheduled for any subject,
scenarios in given time intervals. time limit is exceeded, and the model is stopped. The user
• Service block. This type defines blocks that delay a can stop a model. Simulation run can be combined with
transportation facility for a certain time or for a simulation build. The model can be changed only after it
queuing. is completed.
Blocks can be connected. The set of connected blocks is The model runs in real time or faster. A scale of accel-
a layout of the transport network. A size of each block is eration is defined in model settings.
VICTOR OKOL’NISHNIKOV AND ALEXANDER ZENZIN: USE OF SIMULATION FOR DEVELOPMENT 251

The user can save or restore a model. The model can be The development of control programs was carried out
saved only when a model is not run. A model can be together with development of a complex of models. The
rolled back to its latest saved state. A saved state is the models were used for debugging and testing of control
state that corresponds to last-restored model if the latest programs. In future the development of automatic control
action was to restore model, and to last-saved model if the programs is supposed for a case of a fire or an earthquake.
latest action was to save model. TSS is a result of luck to have some more reliable way
TSS also allows to render current layer using 3 - dimen- to work with Mera than just write code in programming
sional (3D) style. This is the additional feature needed to language like Java or C++.
improve visualization and made it more friendly. System TSS allows building libraries of elements (logistics si-
builder will be able not only to create the static 3D screen, mulation is an example) that allows creating models
but also to see model moves. System builder can select its graphically, and model will be at the same time: built by
own view like in action game or like the bird flying. the terms of Mera (subjects, systems, ports, subject con-
Java 3D technology is used. Major is still the 2 - dimen- nections, message passing and receiving etc). That brings
sional animation, since it is the only way to build and re- Mera from academic research to something that can be
build the model, and tune its parameters. contributed by some manufacturing applications.
2D animation is faster, although 3D uses the capabili-
ties of the graphics accelerators (widely used today) to do REFERENCES
fast affine transforms. But overhead is not so much (1.5 - [1] V.V. Okol'nishnikov, and S.W. Rudometov. Distributed simula-
2 times comparing to the 2D speed), and it still allows tion system integrated into control system. Proceedings of the 13h
using this animation in real-time speed. European Simulation Multiconference, Warsaw, 2: 510-512, 1999.
[2] V.V. Okol'nishnikov, and D.A. Iakimovitch. Visual Interactive
An application for logistics simulation in a warehouse Industrial Simulation Environment. Proceedings of the 15th
distribution center is developed with the help of transpor- IMACS World Congress on Scientific Computation, Modeling
tation simulation system [5]. and Applied Mathematics, Berlin, 5:391-395, 1997.
[3] V.V. Okol'nishnikov, and S.W. Rudometov. Development of
V. CONCLUSION distributed simulation system. Lecture Notes Comput. Sci.
2763:524-527, 2003.
With usage of the program constructor automatic con- [4] D.V. Chernakov, and V.V. Okol'nishnikov. Control Program
trol programs for the process control system of North- Development System. Proceedings of the Second IASTED Inter-
Musk railway tunnel was developed. These programs pro- national Multi-Conference AUTOMATION, CONTROL, AND
vide necessary conditions for exploitation of North-Musk APPLICATIONS (ACIT-ACA). Novosibirsk, 142-145, 2005.
railway tunnel. At present the process control system of [5] V.V. Okol'nishnikov, and S.W. Rudometov. Simulation of Com-
North-Musk railway tunnel puts into operation. plex Transportation Systems. Proceedings of the Second IASTED
International Multi-Conference SOFTWARE ENGINEERING
(ACIT-SE). Novosibirsk, 60-64, 2005.

Anda mungkin juga menyukai