Anda di halaman 1dari 8

298

IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 7, NO. 3, MAY 1999

Future Needs for Control Theory in Industry Report of the Control Technology Survey in Japanese Industry
Haruo Takatsu and Toshiaki Itoh
Abstract This paper reports the current situation and future directions of control theory and control technology which have been practically applied to Japanese industries. The control technology committee in the Japanese Society of Instrument and Control Engineering (SICE) conducted a control technology survey of Japanese industry in 1995. The committee sent 300 inquiry letters to Japanese manufacturing companies and received 110 effective answers which included 150 application examples in the practical industries. After the inquiry, the committee visited the leading companies and interviewed process and control engineers to discuss their opinions and comments regarding the inquiry. Index Terms Advanced control, control application, control theory, industrial needs, statistics of applications.

I. INTRODUCTION

N RECENT years Japanese industries have been on a rocky road because they are going through several major changes in their business environment. The rst one is a long recession after the bursting of the economic bubble. During this time, Japanese businesses suffered and were subsequently compelled to adjust their employment and investment. The second is the strong yen in the global business. The U.S. dollar changed from 130-yen level to 90-yen level during one year, which forced Japanese companies to cut their prices and reduce prots. To cope with these situations and to survive, industries are signicantly changing their management strategies. Companies are restructuring their systems by developing the following strategies: to simplify management and decentralization of decision making, to move one part of operation from Japan to other countries from the viewpoint of labor and market, to look to the future and gain a new market instead of the conventional one which may have reached its peak and so on. The utilization of information systems is also becoming more essential to adapt to change. The amount and kind of information available are now growing more and more. Computer network, Internet, multimedia devices are becoming cheaper and more popular, which will affect the way to do business.

From the viewpoint of control applications, control technology can assist the existing production system to operate at the maximum bounds of equipment to produce more prot. Furthermore, advances in the information devices make it possible to realize advanced control theory and control technology in the control production system. In order to investigate the effects of economic and society changes described above, the control technology committee in the Japanese Society of Instrument and Control Engineering (SICE) has carried out a control technology survey of Japanese industry. The same kind of investigation was done in 1989 and the result was announced in the CPC-IV annual meeting in 1991 with much concern as a result. This time the survey is focused on the practical issues of plant equipment investment: how to plan, how to design, how to implement, how to evaluate and so on. The differences between the previous investigation and future trends of control theory and techniques have been also investigated.

II. ACTIVITIES In October 1995, the committee sent 300 inquiry letters to Japanese manufacturing companies and got 110 effective answers which included 150 application examples used in industries. After the inquiry, the committee visited the leading companies plants and interviewed process and control engineers to discuss with them and get their opinions and comments regarding the inquiry. 1) Inquiry Items: Table I shows the inquiry items investigated. The items are classied into three groups: the target plants to be applied at each site, the applied control technology and the procedure of control system development, which are described in the following chapters. 2) Responding Companies and People: Tables IIIV show the statistical data as to respondents of the inquiry. Inquiry letters were sent to all industrial plants. But most of answers came from the raw material processing industries such as metal, renery, petrochemical, chemical, paper, power plant and so on. Answers from mechanical material industries were few in number. Respondents mainly belong to design, equipment management and production management departments. There exists little change during 6 yr in this respect. Answering engineers have experience of from 10 to 14 yr, or more than 20 yr, which shows this time statistical data is reliable.

Manuscript received January 21, 1997; revised December 29, 1997. Recommended by Editor-in-Chief, M. Spong. H. Takatsu is with the Control System Center, Yokogawa Electric Corporation, 2-9-32 Naka-cho, Musashino-shi, Tokyo 180, Japan. T. Itoh is with the Production Technology Division, Mitsubishi Chemical Corporation, 2-5-2 Marunouchi, Chiyoda-ku, Tokyo 100, Japan. Publisher Item Identier S 1063-6536(99)03995-0.

10636536/99$10.00 1999 IEEE

TAKATSU AND ITOH: FUTURE NEEDS FOR CONTROL THEORY IN INDUSTRY

299

LIST

OF

TABLE I INQUIRY ITEMS

TABLE II RESPONDS

AFFILIATION

TABLE III OF RESPONDING PEOPLE

TABLE IV EXPERIENCED YEARS OF RESPONDING PEOPLE

III. TARGET PROCESSES

TO BE

APPLIED

TABLE V APPLICATIONS IN TYPICAL PROCESSES

As to the actual applications at each site, the abstract of the controlled processes, applied control techniques, scale of the control loops and so on are listed and categorized. The major processes where recent control techniques are applied are written in Table V: mold level control in continuous casting equipment, steel board thickness control in rolling mills, component control of distillation columns in renery/petrochemical processes, paper property control in paper milling machines, combustion control in exhaust equipment and so on. From the viewpoint of control techniques actually operating, linear quadratic Gaussian (LQG)/H-innity control are used

300

IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 7, NO. 3, MAY 1999

TABLE VI CONTROL PROBLEMS

TABLE IX HARDWARE

TABLE VII NUMBER OF VARIABLES

TABLE VIII RESPONSE TIME

in steel and other metal industries. Model predictive control and fuzzy control are increasingly employed in renery and petrochemical industries. Advanced proportional integral derivative (PID) control such as dead time compensation and feedforward control are widely applied in other industries where conventional PID control are used. Objectives of these control strategies are mainly to cope with loop interaction, dead time and disturbances shown in Table VI. The increase in the proportion of interaction issue suggests that multivariable control has begun to be applied on many processes. Table VII shows the number of measured, controlled, and manipulated variables in applications, which indicates the processes should be regarded as multivariable system. Table VIII shows the response time of controlled processes. The response time in steel industry applications has generally less than 1 s and in renery and petrochemical more than 1 min. Table IX indicates the adopted hardware devices where each application is implemented. The ratio of dead time change (DCS) is large but remains at the same level as in 1989. The ratio of process computers, work stations, and personal computers has increased. The ratio of process computers is large in steel industry, but in petrochemical the ratio of DCS is large. Process computers and work stations are used to implement advanced control associated together with production information. In this case DCS is required to equip with backup control functions for emergency countermeasure to computers down.

IV. APPLIED CONTROL TECHNOLOGY In order to investigate the current application status of recent control techniques, the inquiry asked the number of applications, evaluation of results, technical possibility and its expectation, key factors for successful and failed applications, expected control system supporting environments and so on. The results were compared with those of 1989. The 18 investigated control techniques are classied as follows.

Advanced PID Type: I-PD and two degrees of freedom PID, decoupling PID, dead time compensation, gain scheduler, PID auto-tuning. Modern Control Theory Type: LOG regulator, observer, Kalman-lter, model predictive control (MPC), adaptive control, H-innity control/ analysis, repetitive control, sliding mode control, exact linearization and control, optimization control. FAN Type: (fuzzy, articial intelligence, neural network) fuzzy control, rule based control, neural-network control. 1) Number of Application: Fig. 1 shows the application status of above common techniques. Advanced PID type control is widely applied and about 30% of respondents have already used this type of control in their factories. MPC and fuzzy control are most widely used among modern control techniques and FAN techniques. Slightly less than 40% of the factories have applied these techniques. Application of modern control theory type such as LQG controller, observer, Kalmanlter, H-innity control/ -analysis is 10% of responding factories but the applications are increasing rapidly in a short period. Compared with the survey of 1989, applications of decoupling PID, dead time compensation, Kalman-lter, model predictive control, H-innity control/ -analysis, rule based control, fuzzy control, optimization are increasing more and more. 2) Evaluation: Fig. 2 shows the total satisfaction of each control technology. The satisfaction is increasing overall compared with 1989. Advanced PID is widely applied and the satisfaction is also high. Although modern control theory is not applied much except model predictive control, 6070% of respondents are satised with the results. The satisfaction of FAN is going up. The LQG control and adaptive control go down a little. Dead time compensation, MPC, and fuzzy control get high points. The effect of neural network and H-innity control indicate that satisfaction is increasing. 3) Technical Possibility and Expectation: The technical possibility and expectation of above 18 control techniques are quantitatively evaluated as indicated in Fig. 3. The technical possibility shows the possibility to be implemented on control systems and the expectation shows the expected effect obtained when applied. The starting point of each arrow is the mean value of whole respondents, the top point is that of leading companies respondents. Consequently the arrow shows the future trend of each control technique. In Fig. 3 the control techniques are found to be categorized into six groups. Advanced PID type control is widely put into use at plants with reasonable expectation. Model predictive control, fuzzy and neural network have promoted in use with large expectation. LQG control, H-innity control and adaptive

TAKATSU AND ITOH: FUTURE NEEDS FOR CONTROL THEORY IN INDUSTRY

301

Fig. 1. Current application of control technologies.

control have reached the level of practical use but expectation is small. Sliding mode control and nonlinear control do not reach the level of practical use but with some expectation. PID, auto-tuning, and rule-based control are already used but the results are not so good as are expected. Optimization control has reached the level of practical use with large expectation. 4) Key Factors for Successful and Failed Applications: Reasons for successful applications were asked in the questionnaire. As is shown in Table X, the top factors for successful application are the appropriate selection of control strategy and control devices. The next are appropriate control specications, exible system conguration, and appropriate emergency countermeasure. On the other hand, when the questionnaire asked if they have experienced in failed

applications, half of respondents selected yes. Table XI is the key factors for failed applications. The top three factors are the insufcient process analyzes, improper selection of sensors and improper countermeasure to disturbance. Respondents in steel and metal industries selected the sensor issues as the top of factors, but ber and cement industries selected the insufcient process analysis as the top factors. 5) Surrounding Environments: Table XII is the insufcient environments to support the introduction of control systems in industries. Respondents consider the process modeling, control performance evaluation, software productivity, and openness of information data/human-machine interface to be the major factors. 6) Application Issues: As common application issues to all control techniques, required engineering power is too high,

302

IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 7, NO. 3, MAY 1999

Fig. 2. Evaluation of each control technology.

there is no need to apply a new control in use, benet of a new technique is not clear obtained high points. As to application issues of each control, the modern control is pointed out to be difcult to apply. Advanced PID is pointed out not to be enough to cope with disturbances and perturbation of process characteristics. As for fuzzy control, the lack of proper design and simulation tools are pointed out and easy-to-use tools are desired. 7) Issues in the Future: Also asked were future issues of control technology as is shown in Table XIII. Many respondents say that new needs for control application should be developed and the education of new control theory and technique should be promoted. Improvement of adaptability of each control is also selected. The issue of control systems which were selected in the previous survey, are not pointed out because the control system performance have rapidly improved since the last survey.

V. CONTROL SYSTEM DEVELOPMENT FOR APPLICATION The following is the survey results for control system developments from the planning phase to the maintenance phase from the viewpoint of control systems life cycle. In order to obtain accurate, quantitative and effective data, the method to select one or two among the given items are applied, which results in a large amount of answers. 1) Planning Phase: In order to evaluate how much information and knowledge of target processes are required for control system developments, the inquiry asked who were the main planners of each development. The response showed that 45% of developments are planned in house, while 28% are by domestic vendors and 11% are by domestic engineering companies. Overseas vendors and engineering companies joined mainly in the oil renery and petrochemical industries consulting in 7% of the respondents. The respondents

TAKATSU AND ITOH: FUTURE NEEDS FOR CONTROL THEORY IN INDUSTRY

303

Fig. 3. Technical possibility and expectation. TABLE X SUCCESSFUL APPLICATION TABLE XI KEY FACTOR FOR FAILURE

KEY FACTOR

FOR

select the forecast of application benets and modeling as top two issues about which they have a hard time in the planning phase. 2) Design Phase: The design members of control projects are mainly their companies (51%) and domestic vendors (32%). Overseas engineering companies are utilized for only 3% of oil renery and petrochemical and chemistry control design projects (3%). Fig. 4 shows the proportion of designers

in each control category. Control engineers are combined with process engineers in order to investigate control designs. Control techniques are roughly categorized into the following: PID and advanced PID, FAN, model predictive control (MPC) and modern control theory (LQG, H-innity etc.). In PID and advanced PID and MPC, the ratio of control engineers and process engineers are almost the same, which shows MPC has been already matured in useful software packages

304

IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 7, NO. 3, MAY 1999

TABLE XII INSUFFICIENT CONTROL SYSTEM SUPPORTING FUNCTIONS

Fig. 5. Operator satisfaction with each control. TABLE XIII FUTURE ISSUES

Fig. 6. Required maintenance manpower in control category.

TABLE XIII(a) PAY BACK PERIOD VERSUS INVESTMENT AMOUNT

Fig. 4. Ratio of attended designer in control category.

with less engineering effort required for its application. On the other hand, the knowledge of plant operation is required in the design phase in which process engineers contribute signicantly.

3) Tuning Phase: Four weeks or more are required for tuning new control applications according to 45% of the respondents. 55% of commissioning is done in house and 30% is by domestic vendors. Fig. 5 shows the operators satisfaction degree classied according to control categories described above. FAN gets high percentage which suggests FAN can easily realize operators know-how of improving controllers in the same way they think. 4) Maintenance Phase: The proportion of projects requiring maintenance engineers after development is 88%, even though every site lacks maintenance engineers with advanced control background. This suggests an increased running time ratio is recognized to be important, which is described more at the evaluation phase. Fig. 6 shows the categorized result of maintenance manpower. The ratio of MPC and FAN is higher than PID and advanced PID, which means the former is expected to run continuously. The ratio of modern control theory is 71%, the lowest as was forecast. As a result of another question together with above one, MPC and FAN applications employ enough maintenance engineers to support a high running level. In addition maintenance engineers are expected to participate in the development mainly in the basic design phase. 5) Benets: With respect to the amount of investment in control system equipment including utilities in Table XIII, 200 million yen or more was reported for 30% of applications and 50 million yen or less was 29%. Others are large in order of from 50 to 100 million yen (20%), from 100 to 150 million yen (11%) and from 150 to 200 million yen (10%). Half of the respondents say the capital payback period was more than three years. The benets of investment on money base are shown in Fig. 7. The benets are in order of quality improvement, labor saving, production increase, energy saving

TAKATSU AND ITOH: FUTURE NEEDS FOR CONTROL THEORY IN INDUSTRY

305

Fig. 7. Realized benets.

and controller design have contributed to applications of many control techniques. Software tools on more upper levels such as scheduling, planning, and management are expected in the near future. 2) From the viewpoint of total production system development, appropriate understanding of the process to be controlled, keeping higher running time of the control systems, improvement of related technologies to control system including hardware and software and so on, are important and desired to be realized. Recent simulation technology with rigorous dynamic model will give one of solutions to these issues. 3) Education in new technologies, needs for control applications, benet evaluation are always expected for the promotion of control technology. How to transfer the technology to the next generation, how to keep the system improving consistently are desired earnestly in the industries. ACKNOWLEDGMENT The items discussed above were part of the survey activities. The survey report consists of the statistical data and analysis to predict the future needs of control technology in industry. These will contribute to academic research of control theory and promote the application of control theory in industries. The authors wish to thank all the companies and engineers for their cooperation to this survey.

Fig. 8. The ratio of systems running over 80% during one year after development.

REFERENCES
[1] The current status report of control technologies in Japanese industries, report of SICE (96PG0007), SICE (Society of Instrument and Control Engineers), 1996, (in Japanese). [2] The current status of advanced control systems, their need and future trend, report of JEMIMA, JEMIMA (Japan Electric Measuring Instrument Manufactories Association), 1990, (in Japanese). [3] S. Yamamoto and I. Hashimoto, Present status and future needs: The view from Japanese industry, Y. Arkun and W. H. Ray, Eds., CPC-IV, 1991, pp. 128.

and resource saving. The merit of quality improvement is larger especially in the steel industry. 6) Evaluation: The report says the number of systems which required a development time of 24 man-years and those requiring a development duration of 0.51 yr were both large. The top four technical reasons are in order of the proper selection of control method, proper selection of control system, easiness of tuning and proper setting of control specications. Fig. 8 shows the number of operating systems which are running over 80% in each control category. This is large in order of FAN, MPC, PID, and advanced PID and modern control theory. The fact that the running ratio in FAN and MPC is larger than that of advanced PID, will encourage companies who are planning to adopt the up-to-date advanced control technologies. On the other hand it was unexpected in advance that the ratio of PID and advanced PID is lower, which is the same level as modern control theory. The top three reasons of failure in the whole system development are listed as sensor issues, disturbance issues and lack of process analysis. In addition issues of robustness and nonlinearity are not a major concern. VI. CONCLUSION The committee has come to the following conclusion as the results of this survey. 1) The hierarchy of industrial automation has progressed steadily and continuously during the past decade. As one of the key technologies, advanced control techniques have been more widely applied to many plants than previous survey. Many software package tools for process modeling, simulation

Haruo Takatsu received the degree from Tokyo Institute of Technology in 1973. He joined Yokogawa Electric Corporation. He is a Control Systems Engineer and General Manager with the Application System Center, Yokogawa Electric Corporation, Tokyo, Japan. He has been engaged in the development of control system and control theory and implementation of control algorithm. Dr. Takatsu is a Member of SICE, IEEJ, and JSST in Japan.

Toshiaki Itoh was born on January 13, 1943. He received the degree from Kyoto University in 1967. He received the Ph.D. degree in process control technology from Kyoto University in 1997. He joined the Mitsubishi Chemical Corporation in 1967. He is an Associate Director in the R&D Coordination Department, Mitsubishi Chemical Corporation, Tokyo, Japan. He has been engaged in process control practice, manufacturing system design, and R&D planning. His main research interest is in systematic approach for process design and process control system design. Dr. Itoh won the Best Paper Award from ISCIE in 1992.

Anda mungkin juga menyukai