First-Hand:Origin of Toshiba Computer Software Product Line COPOS and PODIA for Power-Generation Plant and its induction into the Software Product Line Hall of Fame at Carnegie Mellon University
Origin of Toshiba Computer Software Product Line “COPOS and PODIA” for Power-Generation Plant and its induction into the Software Product Line Hall of Fame at Carnegie Mellon University (CMU)
June 14, 2009
IEEE Life Fellow(LF-06538995)
Toshiba "Computer Software Product Line" for Power-Generation Plant control and operation has been one of the most successful and the historically earliest large-scale real-time software for industrial plants.
This "Computer Software Product Line" has been inducted into the Software Product Line Hall of Fame at Carnegie Mellon University (CMU) Software Engineering Institute (SEI) in 2008 (The 12th International Software Product Line Conference (SPLC2008)) .
- This "Computer Software Product Line" was named COPOS (Computerized Optimum Plant Operation System) first applied for fossil power plants and PODIA (Plant Operation by Displayed Information and Automation) first applied for nuclear power plants, respectively. Although it was called by a different name for a different type of power-generation plants, a unified standard asset of software has been developed and maintained since late 1970s as a standard product family.
- Later in the part of ' 1. Pre -History' the author will describe the background and necessity to develop the productive method of software manufacturing and to improve the software quality (bug free) in software manufacturing and testing or during power plant commissioning.
- As a main inventor of this "software system" for power plant control and operation, the author describes here the origin of its software development including pre-history, and the original systems concepts, which might be informative as well as useful to the engineers and/or researchers in the other domain of technologies.
Pre-history of COPOS and PODIA
Since 1960s to early 70s at first in USA ambitious plant automation using real-time process computer for fossil power generation plants in order to transfer from manual operation to automatic computer control had been developed to solve the lack of skillful plant operators because it took long time to train them and also the needs for skillful operators had been increasing due to rapid increase of power plants for economic growth.
All of these initial developments in USA, however, failed because the computer hardware and its software performance had been far lower than required.
In Japan, the first introduction of process computer into power plant was that of Takigawa No.2 Unit (75MW) of Hokkaido Electric Co. in 1961. The computer system introduced was manufactured by Toshiba and major functions provided for were scan and alarm, and performance calculations.
As the author with colleagues and customers had seen the above mentioned experiences in USA, they had initiated highly advanced development projects to accumulate the experiences of computer automation from partial scale to full-scale from 1967 jointly with Tokyo, Tohoku and Chubu Electric Power Company as customers and Toshiba as a manufacturer .
In 1968, the first successful DDC (Direct Digital Control) operation of turbine run-up control was realized at Yokohama No.6 Unit (350MW) and the first successful boiler DDC and SSC (Supervisory Control) was realized at Hachinohe No.3 Unit (250MW), which was regarded as partial plant automation.
In high economic growth era, those electric companies had been building more than ten electric power plants in one year in Japan to meet the economic demand.
As enough skillful operators were not always available and therefore the electric power companies came to the conclusion to develop automation system of power plant operation. Majority of these system developments were ordered to Toshiba, since they were playing the leading role for these advanced developments in those days.
The author and colleagues of Toshiba studied the technological experiences of plant automation in the United States and analyzed the reasons of failures of that plant automation. They made all kinds of efforts to develop the method that could realize the required plant automation under the limited performance of then-available process computers.
Customers' needs for comprehensive plant automation, or full-scale plant automation was advocated then to realize such functions as reliable computer operations in stead of operators, optimum plant schedule calculations, easy-to-use human machine interfaces and re-start operation from plant disturbances and the like.
In order to achieve these functions it was really necessary to develop both reliable computer hardware and sophisticated software.
Main memory was typically 32 K words, and auxiliary memory was 160 K words, with no back up system configuration. Reliability and availability of sensors and actuators were also technical issues to be overcome for successful plant automation.
In 1974, COPOS was developed for a fossil power generating plant, Sodegaura No.1 unit (600MW) of Tokyo Electric Power Company (TEPCO).
This system development was revolutionary in nature from design concept point of view and was patented in Japan (#1063709 applied in 1974), the U.S. (#4328556) and Canada (#1137167).
(1) At first, they developed "event-oriented" concept, which has such a feature that software programs are triggered and executed only when certain plant conditions are fulfilled. The CPU loading was much reduced and saved in this architecture, since the software is executed only when plant changes its status or its conditions like on-off positions of the switches in control room and open-close positions of valves in the fields.
They had noticed that the fundamental algorithms of “action” could be unified by introducing the event-oriented approach. Regardless of the differences of either plant specific information ( i.e. various plant equipments and plant types such as boiler and turbine), judgment, or corresponding operation by plant operators, this information processing could be generalized as an action-chain uniformly in the world of event-oriented software.
By separating the data or database, which varies plant by plant, and the algorithm or application software, which is related to monitoring, judgment and plant operation, plant automation software was then generalized successfully in spite of the differences of plant specifications.
If the contents of above database were expressed in a “Fill-in-the-blanks” format, which can be easily understood without software expertise, plant engineers or even plant operators could describe the automation logic, and thus work split between computer software engineers and plant engineers could be made in parallel. TEPCO engineers had contributed to this innovative leap of concept for intelligent tables in the basic patents of plant automation.
On the other hand, it was getting almost impossible to organize appropriate project teams for Toshiba since more than hundred engineers were necessary for each project in order to develop relevant plant automation software.
In order to solve these difficult issues such as the lack of software engineers and the huge number of software bugs during the projects development, they had put their heart and soul into innovative efforts days and nights without compromise.
One of the key concepts concerning this invention of plant automaton, which was born out of the distress, was to adopt a mainframe off-line computer to manufacture (generate) the plant automation software for the target real-time computer both automatically and in off-line mode.
If those tables, written in domain specific language, were read by the main frame (off-line) computer for software manufacturing and compiled into the automation software for the target real-time computer, then that automation software would be loaded to the target real-time computer and run directly on it.
This invention prepared also for the database and the automation logic algorithm in a table format to display and modify on CRT display with keyboard of the target computer interactively during the stages of software test run in a factory and during the plant field commissioning.
This so called maintenance function eliminates the necessity of program logic-changes nor re-compilation by the main frame (off-line) computer, even if real-time automation software changes were needed. Since this maintenance was so easy and fast that it could eliminate human errors and programming works at plant site, the computer-based plant automation was verified and completed as a mile stone in the power-plant automation history.
(2) The second design concept leap, which was evaluated as a major breakthrough, was realized as a "computer software" system named PODIA which was developed for a nuclear power generating plant, Fukushimadaini No.3 unit (1100MW) of TEPCO in 1982.
After many commercially-based experiences applying COPOS, large-scale system enhancement was made to introduce innovative human-machine interface (HMI) functions. That was to utilize emerging display devices, i.e. color graphic CRTs to their full extents, like conventional instruments.
The development which realized the operation without control panels and bench boards was evaluated as “extraordinary” as a breakthrough. This system development was again patented in Japan (# 964086 applied in 1972).
As a summary statement, the author would like to stress that COPOS and PODIA software product family were so successful to advance the plant automation technology not only applied in the power plant industry in the historically earliest stage but also applied later on in other field of industries.
System Concepts of COPOS and PODIA
The COPOS System:
(1) Unique Decision Table named "Plant Table"
This describes relations among operational actions of plant equipment, parameters to trigger the actions, data for supervising the plant condition, and messages & recorded data, and the like.
This Plant Table is a kind of domain specific table for plant automation. Each operational action has its own triggering information in it and the triggering chain of actions can be edited automatically by trigger link functions. Thus, complex plant operational procedures could be built without any mismatch.
(2) To separate the algorithms from data which are described in Plant Table
An algorithm that interprets this data in Plant Table and initiates an action for operation of plant equipment (e.g. a start of a pump) is described in advance and standardized.
The mechanism of this data made it possible to express the various differences of plant specifications, such as type of power, fossil, hydro, nuclear and etc., as well as those of operational procedures without implementing software codes.
Accordingly, mass customization of plant automation was realized just by preparing these plant specific data in table forms and target real-time computer software generated from the standard asset of software.
(3) Automatic Software Manufacturing (Generation)
Software programs were compiled and generated automatically on a main frame (off-line) computer, using standardized algorithms and the data in Plant Table that have been read in advance: punched cards were used first, then a punched tape, and then OCR (optical character recognition) which made automatic reading possible.
Therefore, the human involvement previously required to write a program and remove its bugs have been basically eliminated, and the productivity of software manufacturing was significantly improved, often by more than hundred times better.
In other words, manpower for software development resource for plant automation was reduced from several hundred people to just a few people who prepared the required plant specific data.
This concept was built-in into the Software Factory of reference (6).
(4) Easy to change data for test and commissioning
To test and adjust the parameters for the plant automation in a manufacturer's factory and during field commissioning, a CRT display was used to show the Plant Table and its data. The data can be changed through the CRT even when the plant is in operation. It is not necessary to change software program in order to change these parameters. Simulation tools were also provided as a part of support software to verify the automation software before actual field testing.
Although the COPOS concept was established more than thirty years ago, it is utilized even today, regardless of the fact that information & control systems have been experiencing such disruptive changes from process and super-mini computers to client-server and distributed real-time embedded systems.
The PODIA System:
(5) An Advanced Displayed Control Room
This system made a major breakthrough by reducing the updating cycle of CRTs from seconds to milliseconds, and replaced the conventional meters and recorders on BTG (Boiler Turbine Generator) control panel boards to a real-time CRT displays, using advanced software technology.
Based on COPOS development, similar approach was adopted to realize most advanced human machine interface systems. Clear separation of data from software algorithms was designed in.
Separation of data from algorithm was further advanced in PODIA system. One class of data was that of fixed-data to define plant structure, while the other class was that of the variable-data to indicate updated process values. Thus, the data volume to process in high frequencies was reduced down to the appropriate level of the system-bus traffic.
The conventional large control panel of a power plant equipped with many meters, switches and recorders in a power plant was replaced by a few displays with touch-sensitive screens.
In case of the failure of conventional instruments due to single component failure, the loss of information is just limited to the related indicator only. All the information, however, will be lost in case of the failure of digital processor due to similar single component failure. So, the function of the redundant system software contributed to the embodiment of the requirements of the PODIA system.
Nowadays an advanced control rooms for any industries including Airplane cockpits are equipped with such displayed systems.
Our invention of PODIA system in early 1970s was really the origin of these modern control rooms.
- Concept of Plant Table in (1) of Section 2 above: Mr. Karashima and Mr. Takeuchi, his supervisor, of Tokyo Electric Company (TEPCO) were the original inventors. Mr. Abe, Mr. Maeomichi of Toshiba cooperated with them.
*Software Concept and Basic Design of COPOS described in (2) to (4) of Section 2 above was invented and designed by Mr. Kawahara of Toshiba. Mr. Kogure and Mr.Minoura of Toshiba developed the computer software program to implement the Basic Design of the Concept.
- Software Concept and Basic Design of PODIA in (5) of Section 2 above was invented and designed by Mr. Kawahara. Mr. Neda developed the computer software program to implement the Basic Design of the Concept.
- After these two revolutionary Concepts and Software of COPOS and PODIA, many related technical concepts were patented by our colleagues including Mr. Kawai et al (reference (7) below) as significant incremental improvements to contribute the completeness of the COPOS and PODIA product family.
- Mr.Kawahara was elected to the grade of Fellow by IEEE in January 1999 for these contributions.
More than 150 systems were delivered to power plants worldwide as shown in http://www.sei.cmu.edu/productlines/plp_hof.html , the Web site of CMU.
(1) T. Nakamoto, H. Kawahara, et al., "Development of 1975 year model computer control system for power generating plant 'COPOS' , Toshiba Review, October 1974.
(2) Y. Kogure, T. Minoura, et al., "New Software System for Power Generation Plant Computer Control (COPOS)", IEEE PICA Conference 1975.
(3) H. Kawahara, T. Neda, "Recent Computerized Power Generation Plant Automation and Advanced Man-Machine Interface System", IFAC 1977.
(4) M. Itho, H. Kawahara, et al., "Application of Process Computer and Color CRT Display in BWR Plant Control Room," IAEA 1982.
(5) K.Fujii, T.Neda, S.Tanaka,O.Suto, Y.Ikeda, and H.Hayakawa," BWR PLANT ADVANCED CENTRAL CONTROL PANEL---PODIA", IEEE Tr. Nuclear Science, Vol. NS-30, No.1, Feb.1983, pp.833-837.
(6) Y. Matsumoto.”A Software Factory, An Overall Approach to Software Production,” Software Reusability ed. by P. Freeman, IEEE Computer Society, March 1987.
(7) K.Kawai, “Knowledge engineering in power-plant control and operation”, Control Engineering Practice 4, 1996.