- Email: [email protected]
Mill Wide Systems - The Way Ahead?
Co pyrig ht © I FAC Inst r u me ntat io n and Automat io n in the Paper, Rubber . Plast ics a nd Po lymerisat ion Ind ustries, Antwerp. Belgi um 1983
~ II LL- \\"ID E
MILL WIDE SYSTEMS -
SYSTH IS
THE WAY AHEAD?
P. Uronen l'lliversify of 011/11. Fill/al/d
ABSTRACT
It is to believe that interest towards millwide systems is still naininn rromentum and many new installations will be nut into oreration in the very near future esnecially if the economic situation of the Yorld's Pulp and Paper Industry will improve.
Durinn the last 10 years there has been an ever increasinr interest amonn the mananers and en~ineers in the Pulp and Paner Industry towards the so-called mill-wide systems. Even if the term mill-wide system is not well-defined, there are several such systems available in the market. About 50 mills have such inteorated computer systems in oneration, which most specialists could call willwide systems. These existinn systems, however, differ larnely in their wode o~ implementation and in their operations and functions. .
There are so~ commercial systems and solutions already available in the market and anain there exists a danner that the sunrliers will too much dictate the PI·I-bound solutions and the mills i.e. the users are not in the position to tailor or snecify these systems accordinn to their own needs and views. On the other hand this situation needs wuch more attention and work also from the users' side. It is not any more so much the auestion about know how but know what .
Some of these systems are usinn centralized HW, sowe others decentralized, distributed HI-!. Some are based on direct process connections via separate interfaces (datalo~~ers), others are hierarchically inte~rated with process control systems and are effectively utilizinn data-hinhway and local network technolooy .
1.
The relatively short history o~ comouterized process control in the Pulp and PaDer Industry can be seen in a nutshel in Finure 1 (Uronen, 1983)
However, the main emphasis must be on the SW-side - what does the system do and what kind of mill-wide tools are really needed for the manaoement at different levels of orpanization ' ?
.
1960-1970
INTRODUCTION
1970-1 980
1980 +
. . . - - - - - ---- - .- ~-----------...,.,.:;~.... - - - - - - - I
I
I
-I I I I
I
-
?rocess Control Sys terns
--
Process Control Application
I
I
!
I
Manag emen t Applicat ions
,_ - -
- -
I
Hi erarchical total mi ll i nf ormo t io n, , cont ro l and management I system
Ma nagemen t Sy st ems
- __ I
Figure 1. Three generations of computer system in pulp and paper industry. Characteristics: 1960-70, Centralized HW, "big" computers, in-house systems; 1970-80, minicomputers in Process Control, separate organizations and systems, packaged systems, mainframes in Management Applicati ons; 1980+, distributed HW, hierarchical structure, integrated systems. IS 7
P. Uronen
158
Today thprp is r~~i~ rlevelo~~ent in both process control syste~s and mananement tyne systems used in this industry. The major development and acvances in ~icroelectronics have resulted in dinital instrumentation system and other di~tributed ~~, which provide new opportunities for effective automation, coordination and opti~ization in all operations includin9 hinher-level schedulinr and planninn functions. The dyna~ics and complexity of the business and of the economy are continuously increasinn. Existing offline t ype methods of mana~ement will no lonner be able to CODe. There is also an increasina awareness that accurate real-ti~ information systems is a vital resource for the company and its business. Therefore, effective information, planninr and control systems are needed throunhout the ornanization i.e. a mill-wide system concent is now very popular . systems have been tODics of discussion amonn the syste~s anr. computer people in the Pulp and Paper Industry for more than 10 years now . ~owever, even a unique and well-accented definition of this concept is still missinn. The vendors, the users~ the researchers ~nd consultants all have and use a little different definition and meaninn of that term . Followinn neneral features can be listed :
~ill-wide
Mill-wide system collects and uses information from the whole mill i.e. not only from one department or even not from one area. (Definition of an area = pul~ mill, paper mill, enerny, etc.) The results have likewise importance and meanin~ for the whole mill.
Table 1.
It is a ~lanninn and coordination tool for the mill manane~nt anr. o~erational staff at di~ferent levels of ornanization . Most ¥unctions of the systems are on-line onen lOaD i.e. the final decisions of dlrect control actions are taken by onerators or mananement . Productivity and efficiency o¥ i.e. economics are key tasks.
o~eration
~\., realisation can be done in many r.if¥erent ways.
Effective reportinr and information nrocessinn are of vital importance (history, status, trenr., prediction for future, etc . ) Today there are ca. 50 such systems which minht be called mill-wide syste~s installer. (Edlund and Einarsson, 1982) in the Puln and Paper mills. These systems, however, differ larnely in their functions and in their ~~ anr. SY realisation . There have not been very much rocumented resu lts in direct sa vi nn finures (~ronen, LeiviskK, Poikela, 10 83), rrost imnortant beinn the increased nroduction ca~acity. It has also been stated (Edlund and Einarsson, 1982) that nreater nart of the total benefits of mill-wide control cannot be estimated in direct finures and therefore it should not even be tried. Maybe the situation is such that a user does not even know how important such a system is for him before he has installed and used it (Fadum, 1983). In a major investination concerninn the users' oninions and needs (Uronen, 1983) the list of most imnortant benefits of mill-wide systems was accordinn to Table 1.
Benefits of mill-wide production planning and control systems
Benefit
Percentage of answers
Better use of equipment and capacity Better control of cost and effectiveness Better decision making More accurate and timely information Increased production Decreased losses and risks Easier and more fle xible planning Savings in personnel Increased safety in operation Reduction of time loss Better customer service Better inventory management
;0 we see that the users who already have ;ome experience oive more credit to coordilation and planninn tasks than to direct ;avin(Js . It is then an other auestion,
61 58 55
47 45 37
18 16
11 3 3 3
how well these ideas will be acenter by mananers and suner-intendents of those mills which do not yet have any such s y ste~ but are perhaps nlanrinn samethinn like that .
Mill Wide Systems - The Way Ahead?
2.
HW AND SW FEATURES
Based on the above rounh definition of millwide systems it is evident that the ~W base for such a system must be some kind of local data network with sufficient memory and co~ putin9 capabilities. Further easy-to-use and interactive operator communication in form of colour video ter~inals and various reports is needed. So far followino basic approaches to this H~I solution have been presented : a. centralized HW usin~ data oatherin~ terminals, direct measurements, ~anual entries and star or point-to-point network.
159
Concerninn the existin0 packane-tyne unit process control systems there have further been two different tyoes of solutions; to include these in the hierarchy or to builr the mill-wide information and plannin~ syste~ fully se~arately from these. The former has as an advantane the possibility to use preprocessed information from the existina co~outer syste~s without needino duplicate rreasurements, the nroblem will arise in connectinn the systems of different vendors and different ane to work tonether i.e. the standardization nroblem is a difficult one. This oroblem does not exist if the ~ill decides to builr its mill-wide syster. separate1y anc. by us i n(1 the eaui ~rrent of one venror only. My opinion is that these systems will be more built with distributed ~\.I as hierarchies like that in Fioure 2. The communication problem can even now be solved technolonically but cheaDer solutions are needed in the future.
b. distributed HW usin9 data hiohway techniques to intenrate the functions and for data transmission. This can be further built in a hierarchy.
Level IV : Business management Management information system . decision support system
IVb
-------------Production scheduling
/
Pulp mill
1 0) o
Recovery
Energy
1 0 e
Environmen 1
Paper mill
IG
ICV
1
CD 0
0 0 CV
Levels I. 11 . 111. : Process contrOl
Figure 2. Process control hierarchy in an integrated paper mill. Levels I, 11 and IlIa comprise: 1, woodyard, Kamyr, washing and screening; 2, bleaching, bleach chemical preparation; 3, evaporator~, recovery boiler; 4, causticization and lime kiln; 5, power bOl1~r, turbine generator, water treatment; 5a, purchased power; 6, addltives, stock preparations; 7, on-machine processes; 8, roll handling and finishing; 9, laboratory; 10, wastewater treatments,. special measurements, monitoring and alarms. Level IIIb comprlses area control, detailed scheduling, coordination, quality control, optimization. Level IV comprises management information, production planning, decision support (Uronen, 1983).
160
P. Uronen
One important HI-' feature here is that in connection of mill-wide functions the absolute accuracy of the pri~ary information i.e. measureJrents is strinoent. In contrary to this in control systems the reliable trend in most cases is sufficient. Thus improvements in many measurements related to production rates and product quality in many departments of an intenrated paper miTI is needed. Concerninn the computers and memories anr other HW units used to build the hierarchy described its capacity, speed or memory are not any more a limitino factor; the problem is to do ri~ht thinns with these equipment. In SW side the problem of interlinkinn different systems was already mentioned. The hierarchical approach as denicted in FiGure 2 is now widely accepted (Uronen and Williams, 1978; Williams, 1982; Uronen, 1983; Brewster, 1983). Hierarchical control system as a concept is relatively simple but for buildinq a workinG computer based information, control and decision making hierarchy relatively exact MOdel s of the system to be controlled and also appropriate decision makinG alnorithms are needed. On the contrary, if the hierarchy is mainly used for explainino the behaviour of the system or desi~ninn new structures (for example determininq what decisions have to be made at each level) the models do not have to be so accurate. The existino theory seems to for~ sufficient ~round for desi~ninn of these systems (Findeisen, 1978; Lefkovitz and Cheliutskin, 1976) but better process models are still needed.
3. FUNCTIONS AND TOOLS OF A MILL-~IDE SYSTEM A mill-wide system can be analyzed by its tasks or by the workino tools it offers to the mill's operational staff. These tools typically include: distributed intellinence control elements, optimization capabilities, resource mananement techniques (e.n. order handlino, production schedulinn and trimmin0 programs), effective communication tools and simulation (Gallimore, 1981). The two main parts of functions in a mill wide system include: information handlinn part (gatherin0, pro~ramming, reportino, etc.) and calculation part (trimminn, scheduling, planning, simulation, control) (Leiviska, 1982). A third possibility is to relate these tasks to different areas of paper mill operations. This will then include three eleJrents : Production manaoeJrent, enerny mana0ement and quality mananeJrent (Fadum, 1983). These three ways to describe the functions and tools used in the existinn and olanned mill-wide systems are self-exolanatory and
a detailed discussion of these tasks is not needed here. The main idea is to build an infor~ation, coordination and planninn tool to plan (and if necessary flexibly and interactively reolan), coordinate and supervise the order-driven ~ill operation in an efficient way. The economy of ooeration is then in essence. I believe that the future development will follow the existino lines wi th more exact Il'i 11 model s and more i nteractive ooerator communication. ClosinG the loon i.e~ allowinG the syste~ operate ~he mill without human decision makino (i.e. the fully auto~ated oroduction) does not see~ to be a reality in short run (L1ronen, 1983) .
4. OPSTACKLES AND PARRIERS The benefits o~ mill-wide systems in the pul p and naper i ndus try \A'ere bri efl y discussed in introduction. Put the obstackles or barriers for wider acceptance and use of these systems is likewise a critical ["Joint. Table 2 nives the oninions of mill peoDle (Uronen, 1983). As we can see the most serious drawbacks and barriers to spreadinn these systell's further seem to be the development and imple~entation costs, lonn time to develon it, lack of necessary data and models and Questions related to staff and personnel (education, traininn, attitudes, suspicions, etc.). Minor problems nowadays seem to be the "technical ouestions" (I-I\.>' reliability, runninn costs, etc.). At this noint I would like to stress the importance of breakinG the institutional barriers by traininG and rinht information about these systems; what they are and what they can do and that these systems will come there to help peoole in their work not to replace them
5. RESEARCI-I NEEDS AND I-IINTS FOR
IMPLE~ENTATION
I"hat should then be the focal noint for research and develonment work related to mill-wide automatlo~? Tabel 3 will nive a list of most important R & D areas as seen by the users (Uronen, 1983) . The topics suonested Quire clearly indicate the followinn imoortant problem areas economics of these systems, ornanizational effects of mill-wide automation and standardization and easier rroorarrminr syste~s. \·'hen a company or mill is noinn to ill'plement a mill-wide automation system followinn should be imnortant : a. a total plan for the mill automation for the next 5-6 vears. Then also the possible chan~es in the operation, product mix and markets should be notified This olan woulr include a step-wise modular anrroach.
Mill Wide Systems - The Way Ahead?
Table 2.
Drawbacks of management applications
Drawback
Percentage of answers
Development costs Long time to develop Lack of necessary data Training and education Lack of special personnel Attitudes Updating and maintenance Lack of models and algorithms Lack of flexibility Software difficulties Poor documentation Running costs Unreliable hardware
Table 3.
28 28
25 14 11
3
Percentage of answers
Economics of the system Distribution of decision making and its effects on MIS Mill productivity models Production-scheduling and coordination algorithms Nonprocedural programming languages Standardization DSS in forest industry Fully automated mills Corporate-level hierarchy
b. the problem is different in an and in a new mill. trainin~
61 53 53 44 42 39 33
Future research topics
Topic
c. the
161
old mill
is utmost important.
d. a control or computer specialist is not necessarily the best person for project mana~er. A person havin9 production back~round and a broad experience is needed. Top manaoement involvement is also important. e. this kind of project needs a lot of resources commitment and oatience. The results can be seen much iater and mostly i ndi rectly.
6. CONCLUSIONS There is clearly remarkable interest, as well as progress, in the inte9ration of control, information, and plannin9 systems at various levels in the pulp and paper
39 39 36 36 33 33 28
19 14
industry. To a larne extent, this has been encouraned by the new oeneration of distributed hardware, which oermits a nreater flexibilitv in the construction of inteorated hierarchical systems. The anplication OT the new oeneration of these inteorated systems is now at its early staoes, without any standards or "package9" solutions . Therefore this would he an annronriate time to try to develop ouidelines and standards for these systems and their usaoe, accordinn to the oninions and wishes of the users, before the market becomes .too much influenced by the hardware and by the vendors of these systems, as has been the case sometimes in the short history of com~uters in industry. Some critical questions to be thorounhly studied when develooinn these systems are: how much complexity in the hierarchical structure and in the inteoration of the systems is reasonable at various levels of oroanization ?
P. Uronen
162
what is the optimum level of auto~ation and how much does it depend on local circumstances, mill or company size, product mix, etc . ?
Gallimore, J. (1981), ~ill-wide comnuter Automation Key Flement in Modernization Strateay, Pulp & Paper, September 1981, pp. 139-143.
how much standardization and how many ~eneral packaGes or modules is it feasible to develop for these kinds of inteGrated systems?
Lefkowitz, I. and Cheliutskin, A. (editors) (1976), Intenrated Systems Control in the Steel Industry, Conference Renort CP-76-13, IIASA, Laxenburn.
Economic payoff studies of different scenarios of inte~rated systems in a typical mill usina actual data would be important. What new models and al~orithms are needed in these systems? How applicable are existing models and alGorithms? What are the effects of these systems on oraanizations and on education of people at all levels of oraanization? It seems quite obvious that at mill level the intearation is feasible and advantageous. Most mills are planninn to implement this kind of system in the future. The completion of this intenration at division and corporate levels is not so certain and many companies do not see it as necessary. With this kind of development and by use of dioital instrumentation and other distributed HW and effective man-machine communications the work of personnel will chanGe significantly. Also the comoany's internal information system is becoming similar to those of some public info systems. Already today we have the technical canabilities to realize fully automated production where operators are only in one (or very few) information center(s). H~! develoflment seems to be far ahead of SI·' development and there are no established solutions or standards for inteGrated automation. Another big problem area will be the oroanizational and man-machine interface ouestion. The problem is not any more to know how but to know what?
7. REFERENCES Brewster, D.B. (1983), Newsprint mill-wide system desi~n considerations, ProceedinGS of the Third International Pulp and Paper Process Control Symposium, Vancouver, B.C . , May 2-4, pp . 183-194 . Edlund, S.G. and Einarsson, L. (1982), Mill-wide systems, state of the art, trends, opportunities, EUCEPA ~ymposium on Process Control Systems in Puln and Paper Industry, Stockholm, ~1ay 11-14. Fadum, O. (1983), Mill-wide automation Do you really need it?, Ibid, pn. 121129. Findeisen, W. ( 1978), Hierarchical Control Systems - An introduction, Professional Paper PP-78-1, IIASA, Laxenburo.
Uronen, P., Leiviska, K. and Poikela, T. (1983), Economic Justification of Production Control Systems, Proceedinas of the Third International Pulo and Paper Process Control Symnosium,. Vancouver, S.C., May 2-4, np. 1-4. Uronen, P. and williams, T.J. (1978), Hierarchical Comouter Control in the Pulp and Paller Industry, Report ~lo. 111, Purdue Laboratory for Aonlied Industrial Contro 1, Sentember 1978, y'es t La fayette. Uronen, P. (1983), Intenrated Comnuter Systems in the Pulp and Paper Industry, Research Report RR-83-6, IIASA, Laxenburo. T.J . (1982), Imnlications of Logical Models in Mill-wide Systems, EUCEPA Symposium. Proceedinos, ~ay 11-14, 1~82, Stockholm.
~illiams,
~ II LL- \\"ID E
MILL WIDE SYSTEMS -
SYSTH IS
THE WAY AHEAD?
P. Uronen l'lliversify of 011/11. Fill/al/d
ABSTRACT
It is to believe that interest towards millwide systems is still naininn rromentum and many new installations will be nut into oreration in the very near future esnecially if the economic situation of the Yorld's Pulp and Paper Industry will improve.
Durinn the last 10 years there has been an ever increasinr interest amonn the mananers and en~ineers in the Pulp and Paner Industry towards the so-called mill-wide systems. Even if the term mill-wide system is not well-defined, there are several such systems available in the market. About 50 mills have such inteorated computer systems in oneration, which most specialists could call willwide systems. These existinn systems, however, differ larnely in their wode o~ implementation and in their operations and functions. .
There are so~ commercial systems and solutions already available in the market and anain there exists a danner that the sunrliers will too much dictate the PI·I-bound solutions and the mills i.e. the users are not in the position to tailor or snecify these systems accordinn to their own needs and views. On the other hand this situation needs wuch more attention and work also from the users' side. It is not any more so much the auestion about know how but know what .
Some of these systems are usinn centralized HW, sowe others decentralized, distributed HI-!. Some are based on direct process connections via separate interfaces (datalo~~ers), others are hierarchically inte~rated with process control systems and are effectively utilizinn data-hinhway and local network technolooy .
1.
The relatively short history o~ comouterized process control in the Pulp and PaDer Industry can be seen in a nutshel in Finure 1 (Uronen, 1983)
However, the main emphasis must be on the SW-side - what does the system do and what kind of mill-wide tools are really needed for the manaoement at different levels of orpanization ' ?
.
1960-1970
INTRODUCTION
1970-1 980
1980 +
. . . - - - - - ---- - .- ~-----------...,.,.:;~.... - - - - - - - I
I
I
-I I I I
I
-
?rocess Control Sys terns
--
Process Control Application
I
I
!
I
Manag emen t Applicat ions
,_ - -
- -
I
Hi erarchical total mi ll i nf ormo t io n, , cont ro l and management I system
Ma nagemen t Sy st ems
- __ I
Figure 1. Three generations of computer system in pulp and paper industry. Characteristics: 1960-70, Centralized HW, "big" computers, in-house systems; 1970-80, minicomputers in Process Control, separate organizations and systems, packaged systems, mainframes in Management Applicati ons; 1980+, distributed HW, hierarchical structure, integrated systems. IS 7
P. Uronen
158
Today thprp is r~~i~ rlevelo~~ent in both process control syste~s and mananement tyne systems used in this industry. The major development and acvances in ~icroelectronics have resulted in dinital instrumentation system and other di~tributed ~~, which provide new opportunities for effective automation, coordination and opti~ization in all operations includin9 hinher-level schedulinr and planninn functions. The dyna~ics and complexity of the business and of the economy are continuously increasinn. Existing offline t ype methods of mana~ement will no lonner be able to CODe. There is also an increasina awareness that accurate real-ti~ information systems is a vital resource for the company and its business. Therefore, effective information, planninr and control systems are needed throunhout the ornanization i.e. a mill-wide system concent is now very popular . systems have been tODics of discussion amonn the syste~s anr. computer people in the Pulp and Paper Industry for more than 10 years now . ~owever, even a unique and well-accented definition of this concept is still missinn. The vendors, the users~ the researchers ~nd consultants all have and use a little different definition and meaninn of that term . Followinn neneral features can be listed :
~ill-wide
Mill-wide system collects and uses information from the whole mill i.e. not only from one department or even not from one area. (Definition of an area = pul~ mill, paper mill, enerny, etc.) The results have likewise importance and meanin~ for the whole mill.
Table 1.
It is a ~lanninn and coordination tool for the mill manane~nt anr. o~erational staff at di~ferent levels of ornanization . Most ¥unctions of the systems are on-line onen lOaD i.e. the final decisions of dlrect control actions are taken by onerators or mananement . Productivity and efficiency o¥ i.e. economics are key tasks.
o~eration
~\., realisation can be done in many r.if¥erent ways.
Effective reportinr and information nrocessinn are of vital importance (history, status, trenr., prediction for future, etc . ) Today there are ca. 50 such systems which minht be called mill-wide syste~s installer. (Edlund and Einarsson, 1982) in the Puln and Paper mills. These systems, however, differ larnely in their functions and in their ~~ anr. SY realisation . There have not been very much rocumented resu lts in direct sa vi nn finures (~ronen, LeiviskK, Poikela, 10 83), rrost imnortant beinn the increased nroduction ca~acity. It has also been stated (Edlund and Einarsson, 1982) that nreater nart of the total benefits of mill-wide control cannot be estimated in direct finures and therefore it should not even be tried. Maybe the situation is such that a user does not even know how important such a system is for him before he has installed and used it (Fadum, 1983). In a major investination concerninn the users' oninions and needs (Uronen, 1983) the list of most imnortant benefits of mill-wide systems was accordinn to Table 1.
Benefits of mill-wide production planning and control systems
Benefit
Percentage of answers
Better use of equipment and capacity Better control of cost and effectiveness Better decision making More accurate and timely information Increased production Decreased losses and risks Easier and more fle xible planning Savings in personnel Increased safety in operation Reduction of time loss Better customer service Better inventory management
;0 we see that the users who already have ;ome experience oive more credit to coordilation and planninn tasks than to direct ;avin(Js . It is then an other auestion,
61 58 55
47 45 37
18 16
11 3 3 3
how well these ideas will be acenter by mananers and suner-intendents of those mills which do not yet have any such s y ste~ but are perhaps nlanrinn samethinn like that .
Mill Wide Systems - The Way Ahead?
2.
HW AND SW FEATURES
Based on the above rounh definition of millwide systems it is evident that the ~W base for such a system must be some kind of local data network with sufficient memory and co~ putin9 capabilities. Further easy-to-use and interactive operator communication in form of colour video ter~inals and various reports is needed. So far followino basic approaches to this H~I solution have been presented : a. centralized HW usin~ data oatherin~ terminals, direct measurements, ~anual entries and star or point-to-point network.
159
Concerninn the existin0 packane-tyne unit process control systems there have further been two different tyoes of solutions; to include these in the hierarchy or to builr the mill-wide information and plannin~ syste~ fully se~arately from these. The former has as an advantane the possibility to use preprocessed information from the existina co~outer syste~s without needino duplicate rreasurements, the nroblem will arise in connectinn the systems of different vendors and different ane to work tonether i.e. the standardization nroblem is a difficult one. This oroblem does not exist if the ~ill decides to builr its mill-wide syster. separate1y anc. by us i n(1 the eaui ~rrent of one venror only. My opinion is that these systems will be more built with distributed ~\.I as hierarchies like that in Fioure 2. The communication problem can even now be solved technolonically but cheaDer solutions are needed in the future.
b. distributed HW usin9 data hiohway techniques to intenrate the functions and for data transmission. This can be further built in a hierarchy.
Level IV : Business management Management information system . decision support system
IVb
-------------Production scheduling
/
Pulp mill
1 0) o
Recovery
Energy
1 0 e
Environmen 1
Paper mill
IG
ICV
1
CD 0
0 0 CV
Levels I. 11 . 111. : Process contrOl
Figure 2. Process control hierarchy in an integrated paper mill. Levels I, 11 and IlIa comprise: 1, woodyard, Kamyr, washing and screening; 2, bleaching, bleach chemical preparation; 3, evaporator~, recovery boiler; 4, causticization and lime kiln; 5, power bOl1~r, turbine generator, water treatment; 5a, purchased power; 6, addltives, stock preparations; 7, on-machine processes; 8, roll handling and finishing; 9, laboratory; 10, wastewater treatments,. special measurements, monitoring and alarms. Level IIIb comprlses area control, detailed scheduling, coordination, quality control, optimization. Level IV comprises management information, production planning, decision support (Uronen, 1983).
160
P. Uronen
One important HI-' feature here is that in connection of mill-wide functions the absolute accuracy of the pri~ary information i.e. measureJrents is strinoent. In contrary to this in control systems the reliable trend in most cases is sufficient. Thus improvements in many measurements related to production rates and product quality in many departments of an intenrated paper miTI is needed. Concerninn the computers and memories anr other HW units used to build the hierarchy described its capacity, speed or memory are not any more a limitino factor; the problem is to do ri~ht thinns with these equipment. In SW side the problem of interlinkinn different systems was already mentioned. The hierarchical approach as denicted in FiGure 2 is now widely accepted (Uronen and Williams, 1978; Williams, 1982; Uronen, 1983; Brewster, 1983). Hierarchical control system as a concept is relatively simple but for buildinq a workinG computer based information, control and decision making hierarchy relatively exact MOdel s of the system to be controlled and also appropriate decision makinG alnorithms are needed. On the contrary, if the hierarchy is mainly used for explainino the behaviour of the system or desi~ninn new structures (for example determininq what decisions have to be made at each level) the models do not have to be so accurate. The existino theory seems to for~ sufficient ~round for desi~ninn of these systems (Findeisen, 1978; Lefkovitz and Cheliutskin, 1976) but better process models are still needed.
3. FUNCTIONS AND TOOLS OF A MILL-~IDE SYSTEM A mill-wide system can be analyzed by its tasks or by the workino tools it offers to the mill's operational staff. These tools typically include: distributed intellinence control elements, optimization capabilities, resource mananement techniques (e.n. order handlino, production schedulinn and trimmin0 programs), effective communication tools and simulation (Gallimore, 1981). The two main parts of functions in a mill wide system include: information handlinn part (gatherin0, pro~ramming, reportino, etc.) and calculation part (trimminn, scheduling, planning, simulation, control) (Leiviska, 1982). A third possibility is to relate these tasks to different areas of paper mill operations. This will then include three eleJrents : Production manaoeJrent, enerny mana0ement and quality mananeJrent (Fadum, 1983). These three ways to describe the functions and tools used in the existinn and olanned mill-wide systems are self-exolanatory and
a detailed discussion of these tasks is not needed here. The main idea is to build an infor~ation, coordination and planninn tool to plan (and if necessary flexibly and interactively reolan), coordinate and supervise the order-driven ~ill operation in an efficient way. The economy of ooeration is then in essence. I believe that the future development will follow the existino lines wi th more exact Il'i 11 model s and more i nteractive ooerator communication. ClosinG the loon i.e~ allowinG the syste~ operate ~he mill without human decision makino (i.e. the fully auto~ated oroduction) does not see~ to be a reality in short run (L1ronen, 1983) .
4. OPSTACKLES AND PARRIERS The benefits o~ mill-wide systems in the pul p and naper i ndus try \A'ere bri efl y discussed in introduction. Put the obstackles or barriers for wider acceptance and use of these systems is likewise a critical ["Joint. Table 2 nives the oninions of mill peoDle (Uronen, 1983). As we can see the most serious drawbacks and barriers to spreadinn these systell's further seem to be the development and imple~entation costs, lonn time to develon it, lack of necessary data and models and Questions related to staff and personnel (education, traininn, attitudes, suspicions, etc.). Minor problems nowadays seem to be the "technical ouestions" (I-I\.>' reliability, runninn costs, etc.). At this noint I would like to stress the importance of breakinG the institutional barriers by traininG and rinht information about these systems; what they are and what they can do and that these systems will come there to help peoole in their work not to replace them
5. RESEARCI-I NEEDS AND I-IINTS FOR
IMPLE~ENTATION
I"hat should then be the focal noint for research and develonment work related to mill-wide automatlo~? Tabel 3 will nive a list of most important R & D areas as seen by the users (Uronen, 1983) . The topics suonested Quire clearly indicate the followinn imoortant problem areas economics of these systems, ornanizational effects of mill-wide automation and standardization and easier rroorarrminr syste~s. \·'hen a company or mill is noinn to ill'plement a mill-wide automation system followinn should be imnortant : a. a total plan for the mill automation for the next 5-6 vears. Then also the possible chan~es in the operation, product mix and markets should be notified This olan woulr include a step-wise modular anrroach.
Mill Wide Systems - The Way Ahead?
Table 2.
Drawbacks of management applications
Drawback
Percentage of answers
Development costs Long time to develop Lack of necessary data Training and education Lack of special personnel Attitudes Updating and maintenance Lack of models and algorithms Lack of flexibility Software difficulties Poor documentation Running costs Unreliable hardware
Table 3.
28 28
25 14 11
3
Percentage of answers
Economics of the system Distribution of decision making and its effects on MIS Mill productivity models Production-scheduling and coordination algorithms Nonprocedural programming languages Standardization DSS in forest industry Fully automated mills Corporate-level hierarchy
b. the problem is different in an and in a new mill. trainin~
61 53 53 44 42 39 33
Future research topics
Topic
c. the
161
old mill
is utmost important.
d. a control or computer specialist is not necessarily the best person for project mana~er. A person havin9 production back~round and a broad experience is needed. Top manaoement involvement is also important. e. this kind of project needs a lot of resources commitment and oatience. The results can be seen much iater and mostly i ndi rectly.
6. CONCLUSIONS There is clearly remarkable interest, as well as progress, in the inte9ration of control, information, and plannin9 systems at various levels in the pulp and paper
39 39 36 36 33 33 28
19 14
industry. To a larne extent, this has been encouraned by the new oeneration of distributed hardware, which oermits a nreater flexibilitv in the construction of inteorated hierarchical systems. The anplication OT the new oeneration of these inteorated systems is now at its early staoes, without any standards or "package9" solutions . Therefore this would he an annronriate time to try to develop ouidelines and standards for these systems and their usaoe, accordinn to the oninions and wishes of the users, before the market becomes .too much influenced by the hardware and by the vendors of these systems, as has been the case sometimes in the short history of com~uters in industry. Some critical questions to be thorounhly studied when develooinn these systems are: how much complexity in the hierarchical structure and in the inteoration of the systems is reasonable at various levels of oroanization ?
P. Uronen
162
what is the optimum level of auto~ation and how much does it depend on local circumstances, mill or company size, product mix, etc . ?
Gallimore, J. (1981), ~ill-wide comnuter Automation Key Flement in Modernization Strateay, Pulp & Paper, September 1981, pp. 139-143.
how much standardization and how many ~eneral packaGes or modules is it feasible to develop for these kinds of inteGrated systems?
Lefkowitz, I. and Cheliutskin, A. (editors) (1976), Intenrated Systems Control in the Steel Industry, Conference Renort CP-76-13, IIASA, Laxenburn.
Economic payoff studies of different scenarios of inte~rated systems in a typical mill usina actual data would be important. What new models and al~orithms are needed in these systems? How applicable are existing models and alGorithms? What are the effects of these systems on oraanizations and on education of people at all levels of oraanization? It seems quite obvious that at mill level the intearation is feasible and advantageous. Most mills are planninn to implement this kind of system in the future. The completion of this intenration at division and corporate levels is not so certain and many companies do not see it as necessary. With this kind of development and by use of dioital instrumentation and other distributed HW and effective man-machine communications the work of personnel will chanGe significantly. Also the comoany's internal information system is becoming similar to those of some public info systems. Already today we have the technical canabilities to realize fully automated production where operators are only in one (or very few) information center(s). H~! develoflment seems to be far ahead of SI·' development and there are no established solutions or standards for inteGrated automation. Another big problem area will be the oroanizational and man-machine interface ouestion. The problem is not any more to know how but to know what?
7. REFERENCES Brewster, D.B. (1983), Newsprint mill-wide system desi~n considerations, ProceedinGS of the Third International Pulp and Paper Process Control Symposium, Vancouver, B.C . , May 2-4, pp . 183-194 . Edlund, S.G. and Einarsson, L. (1982), Mill-wide systems, state of the art, trends, opportunities, EUCEPA ~ymposium on Process Control Systems in Puln and Paper Industry, Stockholm, ~1ay 11-14. Fadum, O. (1983), Mill-wide automation Do you really need it?, Ibid, pn. 121129. Findeisen, W. ( 1978), Hierarchical Control Systems - An introduction, Professional Paper PP-78-1, IIASA, Laxenburo.
Uronen, P., Leiviska, K. and Poikela, T. (1983), Economic Justification of Production Control Systems, Proceedinas of the Third International Pulo and Paper Process Control Symnosium,. Vancouver, S.C., May 2-4, np. 1-4. Uronen, P. and williams, T.J. (1978), Hierarchical Comouter Control in the Pulp and Paller Industry, Report ~lo. 111, Purdue Laboratory for Aonlied Industrial Contro 1, Sentember 1978, y'es t La fayette. Uronen, P. (1983), Intenrated Comnuter Systems in the Pulp and Paper Industry, Research Report RR-83-6, IIASA, Laxenburo. T.J . (1982), Imnlications of Logical Models in Mill-wide Systems, EUCEPA Symposium. Proceedinos, ~ay 11-14, 1~82, Stockholm.
~illiams,