- Email: [email protected]
The structuring of process knowledge: function, task, properties and state
Robotics & Computer-Integrated Manufacturing, Vol. 6, No. 2, pp. 101-107, 1989 Printed in Great Britain
•
0736-5845/89 $3.00 + 0.00 Pergamon Press plc
Paper THE STRUCTURING OF PROCESS KNOWLEDGE: FUNCTION, TASK, PROPERTIES AND STATE H U G O J. W . V L I E G E N a n d
H E R M A N H . VAN M A L
Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, The Netherlands
The structuring of process knowledge is based on the specification of the function, task, properties and state of the input and output of a process, and the production equipment involved as well as the relations among these specified factors. The above conception is the outcome of a scientific approach to the design process. The mapping of the process knowledge is visualized by six coupled matrices in a so-called "process knowledge map". The structuring of process knowledge is of vital importance for improving manufacturing productivity. The authors consider the approach a basis for improving decision-making among designers and all other parties involved in a design project, for integrating CAD and CAM, and for the support of quality control.
INTRODUCTION The chosen level of control of processes determines manufacturing productivity. Inadequate control of manufacturing processes is often caused by problems concerning the availability of the right knowledge at the right time and in the right place. Lack of knowledge necessary for controlling manufacturing processes can result in unacceptable product quality, late delivery of requested products and high manufacturing costs. Process knowledge is the basis of engineering activities. These engineering activities involve many disciplines, concerning the design of the product, the processes and the production equipment. A lack of a systematic approach for structuring process knowledge hinders the synthesis of design decisions in the disciplines. A lack of a systematic approach also limits the accessibility of existing process knowledge and the analysis of manufacturing problems. Suh 11 states that without a scientific base for design, academicians lack pedagogical vehicles to teach their students, and industrial engineers cannot store their knowledge effectively for others to use. The purpose of this paper is to describe a systematic approach for structuring process knowledge. This approach can support the analysis of existing manufacturing processes and the design of new products including process and equipment design. Van Mal et
al. 1 2 - x 4 took the first step towards the development of this approach in an analysis of manufacturing processes of components. To improve this approach for structuring process knowledge we benefited greatly from the many past contributions to the establishment of a scientific base for design. In particular the work of Paynter 8 and Thoma 1° about bond graphs and the work of Feekes 4 about the hierarchy of energy systems confirmed our opinion that all systems can be seen from the viewpoint of energy or power transport and transformation. Hansen 6 and Eekels 3 broadened our insight into the possibility of describing a design in terms of related subfunctions, the tasks derived from such subfunctions, its properties and its physical state. In this paper we will start to define the terms, function, task, properties and state, before elaborating the approach for structuring process knowledge. The process knowledge map will be introduced as an important document in which process knowledge can be laid down systematically.
DESIGN PROCESS Human beings use physical objects, called products, to fulfil their needs. The design process enables us to create these objects. For the purpose of design, these needs can be treated as problems to be solved, specified as desired changes of state. For example, most men want to have their beard removed every morning. This means that they want to have their rough face transformed into a smooth state. The first step in design is to specify the desired change of state which is often considered as a set of
Acknowledgements--The authors wish to thank Jan Stoker of the Dutch Philips Corporation and Jan Vosmer of TNO for their contribution to this article. Paul H. Tobias proposed valuable modifications concerning the exposition.
101
Robotics & Computer-Integrated Manufacturing • Volume 6, Number 2, 1989
102
functional requirements. The second step of the design process is to search for physical objects with which the desired change of state can be realized. Persons who want to have their rough face transformed into a smooth state can choose from a number of different physical objects or sets of objects to reach their goal. These physical objects are each characterized by their geometry, dimensions, material composition, structure and surface finish. This means that the state of a physical object can be specified. In conclusion, the design process involves the creation of a physical object state with which a desired change of state can be realized (see Fig. 1). The search for suitable states or physical objects is a trial and error process. Engineers try to increase the chances o f f i n d i n g a suitable solution without too many iterations by using available scientific methods. These scientific methods involve structured knowledge about phenomena and enable engineers to predict the suitability of an idea. Engineers try to discover the process conditions necessary for realizing the desired change of state or function (see Fig. 1). In the case of removing hair by means of shaving, the engineer has to specify the feed, the depth of cut and cutting speed. These sets of process conditions define the task of the shaver being designed and these conditions are in the first instance determined by the properties of the hair and the time in which the beard has to be removed. The task to be performed or the set of process conditions to be generated by the shaver has to be transformed into a set of properties of the shaver to be designed (see Fig. 1). This set of properties concerns the possibility of transporting and transforming power necessary for performing the specified task and concerns the properties like required operating power, strength, etc. The designer tries to translate the required set of properties into a state or physical object characterized by shape, dimensions, material composition, structure and surface finishing. D E F I N I T I O N O F F U N C T I O N , TASK, P R O P E R T I E S AND STATE In this section we will define the terms: • function (F) or generation of desired change of state (AS),
f"~E
1
STING~'~.
• task (T) or generation of process conditions (C1, C2,..., C.), • properties P1, P 2 . . . . . Pn and • state $1, $2 . . . . . S.. The function of a physical object is its potential ability to realize the desired change of state in the environment. The desired change of state or function (AS) can involve changes of geometry, dimensions, material composition, structure and/or surface finishing (of course, a change of state can also concern mental changes). The change of state (AS) can be obtained by exposing the ingoing state to a certain set of process conditions over a specified period of time. These process conditions can be considered as the task (T) to be performed for obtaining the desired change of state. The generation of a set of process conditions (C1, C2 . . . . . C,) or task (T), concerning a series of changes taking place during the process, determines the energy flow or power necessary to obtain the desired change of state. This means that the required power is a function of process conditions: Power = f ( C 1, C 2. . . . .
Cn)
(Nm/s)
(1)
where C~, C2 . . . . . C. is a set of process conditions, and f specifies the amount of power due to the process conditions. The process conditions can be adapted during the transformation of an input state into a desired state. Example of Eq. (1): Cutting power = f(feed, depth of cut, cutting speed) (Nm/s). The feed, depth of cut and cutting speed are the process conditions required for cutting material. In other words, knowledge of these process conditions is necessary to solve the problem of changing a present state of a product into a required state. The amount of power needed depends on the properties of the input state. A property determines the power needed to change a certain input state, exposed to certain process conditions during a certain period of time. For irreversible processes this means that, Power = f ( C 1 , C2 . . . . . C.) = P i * A S / A T
(Nm/s) (2)
where Pi specifies a certain property (i properties can be involved), t refers to the time interval, C~, C2 . . . . . C. are a set of process conditions, f specifies the amount of power provided by the process conditions, and AS the desired permanent change of state. Example of Eq. (2) : Cutting power = K*A*V (Nm/s)
SPECIFYING DESIRED CHANGEOF STATE (FUNCTION/ PROEILEM}
[ I
"
[
SPECIFYING STATEOF OBJECT (SOLUTION OF PROBLEM} i
Fig. 1. Design process.
where K is the specific cutting force, A the chip cross-sectional area and V the cutting speed. The specific cutting force is a property (P) which specifies the amount of energy needed to remove by cutting a quantity of material during a period of time. Designing means creating a physical object which is able to perform or to generate process conditions necessary for changing a present state (input state)
Structuring of process knowledge • H. J. W. VLIEGENand H. H. VANMAL into a required state. In order to generate these process conditions a physical object must possess certain properties (P1, P2,--., P,)- The state of an ideal physical object should change as flexibly as possible while performing its task (i.e. no permanent distortion). The properties, e.g. strength, should make the transformation of power possible. For determining the required properties it is sufficient to specify the maximum or ultimate load and allowed reversible change of state which the object should be able to sustain. For simulating reversible processes this means that, Max(load) = P'Max(AS)
(N) or (Nm)
is often called a material property. In our opinion it is better to call E a material parameter with which the structure and the composition of a material are characterized. These are both part of the state. Knowledge required for process control can be structured systematically by means of the terms function (desired change of state), task (generation of process conditions), properties, state and their mutual relations as specified by Eqs (1)-(4). ASSEMBLED P R O D U C T S , E N E R G Y F L O W S Products often consist of several connected physical objects, called component parts, which make up the state of an assembled product. These component parts have to make the transport and transformation of power feasible with maximum reversible change of state of the parts (function). In order to do this, each component part must perform its own task made possible by its properties which are realized by its chosen state. Figure 2 shows an example of the energy flows among components of a turning lathe. We will elaborate this example in detail in a subsequent paper. Before determining the role of a component part, the designer first has to specify the function of each part by specifying the maximum load together with the maximum allowed reversible change of state each part should be able to sustain and/or the power each part should be able to transport and transform. A set of connected component parts can he considered as a black box, a subassembly, with its own function(s), task(s), properties and state(s).
(3)
where Max(load) is the maximum load (force or moment of force) to be sustained by the object, P the desired property of the object and Max(AS) the maximum allowed change of state of the physical object. Example of Eq. (3) : F = R * A L
(N)
where F is the tensile force, R the required strength and AL the maximum allowed elongation of an object. Knowing the required properties (P1, P2 . . . . . P,), the designer has to choose a state which possesses these properties. This means that, Pi = gi(S1, $2 . . . . . Sn)
(Nm per defined change of state)
(4)
where P is a desired property, ($1, $2 . . . . . S,) characterize the state of a physical object and 9i specifies the relation between the state and property. The physical object state is characterized by geometry, dimensions, material, composition, structure and surface finish. Example of Eq. (4) : R = A * E / L
C O U P L E D PROCESSES Several successive process steps or manufacturing operations are generally required for producing component parts or semi-manufactured articles. These process steps involve one or more physical and chemical processes, which we will call technical processes. During the successive process steps or technical
(Nm/m 2)
where R is the strength of a bar, A the cross-sectional area of the bar, E the modulus of elasticity of the material and L the length of the bar. A, E and L describe the state of the bar. The modulus of elasticity
Mp : MOMENT OF ROTAT1ON (WORKPECE)
=,
O)p : ANGULAR VELOCITY (WOR~EC.E) M= : MOMENTOFROTATION (FEED)
c p cop I
.
.
.
.
.
.
i
.
.
.
.
.
.
.
.
.
.
TOOL I
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
wo.,,.,.0. HU--I
I
(0= : ANGULAR VELOCITY (FEED) E= : CUTTING FORCE
' V¢ : CUTTING SPEED F= ; FORCE (FEED) V
,,Vo F,,V, I cU'mNG FLUID ~" ~I l F,:V,.0
I TOOL"OLO"
F,, V.
103
: FEED
Ft : RESPONSIVE FORCE
~. ,v. F, ',V,- 0 M'.m. TOOL POST
t~ : (RESPONSIVE) SPEED(-O) (]~ : HE~TTR/~ISFER
I
LATHE
Fig. 2. Energy flows among components of a lathe.
104
Robotics & Computer-IntegratedManufacturing • Volume6, Number2, 1989
processes the state of the raw material is changed into final products with a required state. The different process steps are coupled because the output of a certain process must be suitable for serving as an input for a successive step. The object of process control is to guarantee the required function, task, properties and state of the output of each process step. For this, knowledge about these coupled processes is needed. This knowledge must give insight into the relations between the ingoing material, production equipment and outgoing semi-manufactured product. In the next section we will introduce a model of technical processes based on the terms function, task, properties and state and their relations, with which knowledge required for controlling a process can be structured. MODEL OF TECHNICAL PROCESSES Figure 3 shows a model of technical processes. The physical and chemical processes controlled by man result in changes of state (AS). To accomplish this a designer chooses production equipment which generates the right process conditions (T(m)) required for realizing the desired change of state (function of the production equipment). The change of state (AS) of S(i) into S(o) causes the transformation of the properties P(i) into P(o). The properties P(o) make possible the generation of process conditions and thereby the performance of task T(o) by the product. The possibility of performing task T(o) is required for establishing the desired product function. The change of state (AS) of S(i) into S(o) occurs through the change in the process conditions generated through task T(m) performed by the production equipment. S(m) can be partly fixed and partly adjustable. The adjustable part determines the possibility of changing the process conditions and thereby task T(m) of the production equipment.
Vital for controlling manufacturing operations effectively is knowing how the adjustment of the state of the production equipment relates to the task of fulfilling the required function of the outgoing product. This demands the specification of the relations between the desired function, task, properties and state of the outgoing product and the function, task, properties and adjustable state of the production equipment involved.
MAPPING OF PROCESS KNOWLEDGE The knowledge necessary for controlling technical processes can be structured and laid down systematically in six coupled matrices. The process knowledge map (Fig. 4) puts into visual form these linked matrices modified from Refs 12-14. The matrices specify the relations between the task, properties and state of the outgoing product and the task, properties and state of the production equipment. Due to the nature of Eqs (1)-(4), these relations represent not only linear transformations but also nonlinear transformations, which are the rule rather than the exception in actual systems. The matrices are as follows.
1. Relations [T(o),P(o)] This matrix specifies the relations between the product task (to generate process conditions) and the product properties. The product must possess properties to perform its task. Specifying this matrix is the speciality of product design in co-operation with marketing, development, service, sales, etc.
2. Relations [P(o),S(o)] This matrix specifies the relations between the product properties and the product state. Development has to select a product state which guarantees that the product possesses the required properties. Specifying
T(o) : TASKOF OUTGOING PRODUCT P(O) : PROPERTIESOF OUTGOING PRODUCT S(O) : STATE OF OUTGOING PRODUCT
S(m)
T(m) : TASKOF PRODUCTIONEQUIPMENT P(m) : PROPERTIESOF PRODUCTIONEQUIPMENT S(m) : STATE OF pRODUCTION EQUIPMENT
M
I
T(i) : TASKOF INGOINGPRODUCT P(i) : PROPERTIESOF INGOINGPRODUCT S(i) : STATEOF INGOINGPRODUCT
P(rn)
T(i) P(i) S(i)=l
-I ......... ...........
T}m)
.........
IS(o) P(o) T(o)
M
: PRODUCTION EQUIPMENT
I
M'
: PROCESS
A S : CHANGEOF STATE
.~AS4- . . . . . . . . . .
TECHNICALPROCESS Fig. 3. Modelof technicalprocesses.
Structuring of process knowledge • H. J. W. VLIEGENand H. H. VANMAL
P(m)
T(m)~
~(m)
105
T(o) : TASK OF OUTGOING PRODUCT P(o) : PROPERTIES OF OUTGOING PRODUCT S(o) : STATE OF OUTGOING PRODUCT T(m) : TASK OF PRODUCTION EQUIPMENT P(m) : PROPERTIES OF PRODUCTION EQUIPMENT S(m): STATE OF PRODUCTION EQUIPMENT
:STATE OF INGOING PRODUCT I :RELATION [ T(o), P(o) ] II :RELATION [P(o), S(o)] III :REL:ATION [ S(o), T(m) ] S(i)
IV V VI
P(o)
:RELATION [T(m), P(m)] : RELATION [ P(m), S(m) ] :RELATION [S(m), T(o)]
Fig.4. Processknowledgemap. this matrix is the speciality of product development in co-operation with sales, engineering, process planning, etc. 3. Relations [S(o),T(m)]
This matrix specifies the relations between the product state and the process conditions (production equipment task) imposed on a well-determined ingoing product state. Process conditions are necessary to transform an ingoing product state into the desired outgoing product state. Specifying this matrix is the speciality of process development in co-operation with product development, process planning, manufacturing, etc. 4. Relations [T(m),P(m)]
This matrix specifies the relations between the production equipment task (to generate process conditions) and the production equipment properties. The production equipment must possess the properties to perform its task. Specifying this matrix is the speciality of production equipment design in co-operation with. process development, engineering, tool design, etc. 5. Relations [P(m),S(m)]
This matrix specifies the relation between the production equipment properties and the production equipment state. Engineering has to choose an equipment state which guarantees that the production equipment possesses the required properties. Specifying this matrix is the speciality of the engineering and tool design in co-operation with process planning, manufacturing, etc. 6. Relations [S(m),T(o)]
This matrix specifies the relations between the production equipment state and the task of the outgoing product. This matrix is the final outcome of all preceding matrices and specifies the instructions for manufacturing the products which are able to perform the required task, and so realizes the function requested by the customer.
In contrast to Suh and others 3'6'~1 who focus only on the transformation of the functional domain to the physical domain, the process knowledge map includes per process step as well as all intermediate transformations concerning the product in relation to the production equipment. In other words the process conditions and properties, the scientific base for transformations, are also considered for each process step. Hauser and Clausing 7 who do use some linked matrices, do not distinguish functions, tasks, properties and state systematically. They also do not consider the decomposition of products into parts and the coupling of manufacturing processes. This hampers the decision-making of designers and other parties involved in trying to specify the product, component parts, processes, and production equipment. Figure 5 shows the process knowledge map for manufacturing an axle by turning. 1,2,5 We will elaborate on this example in detail in a future paper.
IMPORTANCE We are convinced that the approach for structuring process knowledge as presented here will be useful, because: • it can be used for analyzing existing manufacturing processes, which is the first step in improving product quality. This makes it an important tool in quality; • it is necessary to make process knowledge explicit if the automation of manufacturing processes is pursued; • it is supplementary to existing methods like SADT 9 for describing the information flow between activities or processes. The main difference is that methods like SADT consider processes as black boxes and do not give an insight into the knowledge necessary for controlling the processes within a black box. The approach of structuring process knowledge makes it possible to open black boxes and to determine nested
106
Robotics & Computer-Integrated Manufacturing • Volume 6, Number 2, 1989
T(o)
S(m)
TA5K
~TOFBENDN IG RADIAL BEARING
P(o)
FORCE
TURNING OF
AN
ON
A
AXLE LATHE
~.,~ ~ ~~ O.\, A ~' ~% ,:~~TL.,.[K % m ~ CU~ TIN \G TDEPTH I P(
(o) i i i t
"Fj.,'tOPFA-I".k'I"A~--.~-~c"~,c_.__zol Rs l
J
Tim)
I l
",,
I
I
PURCHASING MATERIAL
IS (i)
Fig. 5. Example of the process knowledge map.
process specifications; it makes the synthesis of design decisions made by diverse engineering disciplines involved in a design project possible. This means that the approach can be used as a communication aid a m o n g designers and other interested parties for improving decision-making and supporting the management of design projects; it can play an important role in the integration of CAD and CAM. Of vital importance for the integration of CAD and CAM is the automatic matching a newly developed product with the already available production facilities of a company. This requires a knowledge base which contains all known suitable relations between product features and features of available production equipment. We think that the structure of any knowledge base for C A D / C A M can be derived from the approach for structuring process knowledge as presented in this paper. CONCLUSIONS The process knowledge m a p is a tool that helps to better specify the relations between the product function and manufacturing processes. It increases the chances of designers finding suitable solutions without too m a n y iterations. The essence of the method is the scientific approach, which includes process conditions
and properties of the product and the related production equipment per process step. The utility of the approach has not yet been proved extensively. Further research will be conducted in the specification phase of new products in particular.
REFERENCES 1. Creamer, R. H.: Machine Design. Reading, AddisonWesley 1968. 2. Doyle, L. E.: Manufacturing Processes and Materials for Engineers. Englewood Cliffs, Prentice-Hall, 1969. 3. Eekels, J., Roozenburg, N.: Ontwerpmethodologie. Delft, Delft University of Technology, 1981. 4. Feekes, G. B.: The Hierarchy of Energy Systems, from Atom to Society. Oxford, Pergamon Press, 1986. 5. Gaskin, F., McArthur, G. M.: Mechanical Engineering Science in SI Units. London, Edward Arnold, 1970. 6. Hansen, F.: Konstruktionswissenschaft, Grundlagen und Methoden. Berlin, VEB Verlag Technik, 1974. 7. Hauser, J. R., Clausing, D.: The house of quality. Harvard Business Review 3: 63-73, 1988. 8. Paynter, H. M.: Analysis and Design of Engineering Systems. Cambridge, MIT Press, 1961. 9. Ross, D. T.: Structured analysis (SA): a language for communicating ideas. 1EEE Trans. Software Engng SE3(1): 16-34, 1977. 10. Thomas, J.: Bond graphs for thermal energy transport and entropy flow. J. Franklin Inst. 292: 109-120, 1971. 11. Suh, N. P.: Development of the science base for the manufacturing field through the axiomatic approach. Robotics Comput.-Integr. Mfg 1(3/4): 397-415, 1984.
Structuring of process knowledge • H. J. W. VLIEGENand H. H. VANMAL 12. Van Mal, H. H., Kools, F., Hekma, E. J.: Technisehe processen, procesbeheersing, ontwerpkaart. Eindhoven, Eindhoven University of Technology, 1983. 13. Van Mal, H. H., Kools, F., Hekma, E. J.: !nformatieordening voor procesbeheersing, een modelmatioe benaderin 9. Eindhoven, Dutch Philips Corporation, 1985.
107
14. Van Mal, H. H., Van Someren, W. S. M., Pans, R. F. M.: Towards the integration of CAD and CAM, carpet manufacturing in the future. In: Computer Application in Production and Engineering, CAPE '86, Bo, K., Estensen, L., Falster, P., Warman, E. A. (Eds). Amsterdam, North-Holland, 1987, 879-890.
•
0736-5845/89 $3.00 + 0.00 Pergamon Press plc
Paper THE STRUCTURING OF PROCESS KNOWLEDGE: FUNCTION, TASK, PROPERTIES AND STATE H U G O J. W . V L I E G E N a n d
H E R M A N H . VAN M A L
Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, The Netherlands
The structuring of process knowledge is based on the specification of the function, task, properties and state of the input and output of a process, and the production equipment involved as well as the relations among these specified factors. The above conception is the outcome of a scientific approach to the design process. The mapping of the process knowledge is visualized by six coupled matrices in a so-called "process knowledge map". The structuring of process knowledge is of vital importance for improving manufacturing productivity. The authors consider the approach a basis for improving decision-making among designers and all other parties involved in a design project, for integrating CAD and CAM, and for the support of quality control.
INTRODUCTION The chosen level of control of processes determines manufacturing productivity. Inadequate control of manufacturing processes is often caused by problems concerning the availability of the right knowledge at the right time and in the right place. Lack of knowledge necessary for controlling manufacturing processes can result in unacceptable product quality, late delivery of requested products and high manufacturing costs. Process knowledge is the basis of engineering activities. These engineering activities involve many disciplines, concerning the design of the product, the processes and the production equipment. A lack of a systematic approach for structuring process knowledge hinders the synthesis of design decisions in the disciplines. A lack of a systematic approach also limits the accessibility of existing process knowledge and the analysis of manufacturing problems. Suh 11 states that without a scientific base for design, academicians lack pedagogical vehicles to teach their students, and industrial engineers cannot store their knowledge effectively for others to use. The purpose of this paper is to describe a systematic approach for structuring process knowledge. This approach can support the analysis of existing manufacturing processes and the design of new products including process and equipment design. Van Mal et
al. 1 2 - x 4 took the first step towards the development of this approach in an analysis of manufacturing processes of components. To improve this approach for structuring process knowledge we benefited greatly from the many past contributions to the establishment of a scientific base for design. In particular the work of Paynter 8 and Thoma 1° about bond graphs and the work of Feekes 4 about the hierarchy of energy systems confirmed our opinion that all systems can be seen from the viewpoint of energy or power transport and transformation. Hansen 6 and Eekels 3 broadened our insight into the possibility of describing a design in terms of related subfunctions, the tasks derived from such subfunctions, its properties and its physical state. In this paper we will start to define the terms, function, task, properties and state, before elaborating the approach for structuring process knowledge. The process knowledge map will be introduced as an important document in which process knowledge can be laid down systematically.
DESIGN PROCESS Human beings use physical objects, called products, to fulfil their needs. The design process enables us to create these objects. For the purpose of design, these needs can be treated as problems to be solved, specified as desired changes of state. For example, most men want to have their beard removed every morning. This means that they want to have their rough face transformed into a smooth state. The first step in design is to specify the desired change of state which is often considered as a set of
Acknowledgements--The authors wish to thank Jan Stoker of the Dutch Philips Corporation and Jan Vosmer of TNO for their contribution to this article. Paul H. Tobias proposed valuable modifications concerning the exposition.
101
Robotics & Computer-Integrated Manufacturing • Volume 6, Number 2, 1989
102
functional requirements. The second step of the design process is to search for physical objects with which the desired change of state can be realized. Persons who want to have their rough face transformed into a smooth state can choose from a number of different physical objects or sets of objects to reach their goal. These physical objects are each characterized by their geometry, dimensions, material composition, structure and surface finish. This means that the state of a physical object can be specified. In conclusion, the design process involves the creation of a physical object state with which a desired change of state can be realized (see Fig. 1). The search for suitable states or physical objects is a trial and error process. Engineers try to increase the chances o f f i n d i n g a suitable solution without too many iterations by using available scientific methods. These scientific methods involve structured knowledge about phenomena and enable engineers to predict the suitability of an idea. Engineers try to discover the process conditions necessary for realizing the desired change of state or function (see Fig. 1). In the case of removing hair by means of shaving, the engineer has to specify the feed, the depth of cut and cutting speed. These sets of process conditions define the task of the shaver being designed and these conditions are in the first instance determined by the properties of the hair and the time in which the beard has to be removed. The task to be performed or the set of process conditions to be generated by the shaver has to be transformed into a set of properties of the shaver to be designed (see Fig. 1). This set of properties concerns the possibility of transporting and transforming power necessary for performing the specified task and concerns the properties like required operating power, strength, etc. The designer tries to translate the required set of properties into a state or physical object characterized by shape, dimensions, material composition, structure and surface finishing. D E F I N I T I O N O F F U N C T I O N , TASK, P R O P E R T I E S AND STATE In this section we will define the terms: • function (F) or generation of desired change of state (AS),
f"~E
1
STING~'~.
• task (T) or generation of process conditions (C1, C2,..., C.), • properties P1, P 2 . . . . . Pn and • state $1, $2 . . . . . S.. The function of a physical object is its potential ability to realize the desired change of state in the environment. The desired change of state or function (AS) can involve changes of geometry, dimensions, material composition, structure and/or surface finishing (of course, a change of state can also concern mental changes). The change of state (AS) can be obtained by exposing the ingoing state to a certain set of process conditions over a specified period of time. These process conditions can be considered as the task (T) to be performed for obtaining the desired change of state. The generation of a set of process conditions (C1, C2 . . . . . C,) or task (T), concerning a series of changes taking place during the process, determines the energy flow or power necessary to obtain the desired change of state. This means that the required power is a function of process conditions: Power = f ( C 1, C 2. . . . .
Cn)
(Nm/s)
(1)
where C~, C2 . . . . . C. is a set of process conditions, and f specifies the amount of power due to the process conditions. The process conditions can be adapted during the transformation of an input state into a desired state. Example of Eq. (1): Cutting power = f(feed, depth of cut, cutting speed) (Nm/s). The feed, depth of cut and cutting speed are the process conditions required for cutting material. In other words, knowledge of these process conditions is necessary to solve the problem of changing a present state of a product into a required state. The amount of power needed depends on the properties of the input state. A property determines the power needed to change a certain input state, exposed to certain process conditions during a certain period of time. For irreversible processes this means that, Power = f ( C 1 , C2 . . . . . C.) = P i * A S / A T
(Nm/s) (2)
where Pi specifies a certain property (i properties can be involved), t refers to the time interval, C~, C2 . . . . . C. are a set of process conditions, f specifies the amount of power provided by the process conditions, and AS the desired permanent change of state. Example of Eq. (2) : Cutting power = K*A*V (Nm/s)
SPECIFYING DESIRED CHANGEOF STATE (FUNCTION/ PROEILEM}
[ I
"
[
SPECIFYING STATEOF OBJECT (SOLUTION OF PROBLEM} i
Fig. 1. Design process.
where K is the specific cutting force, A the chip cross-sectional area and V the cutting speed. The specific cutting force is a property (P) which specifies the amount of energy needed to remove by cutting a quantity of material during a period of time. Designing means creating a physical object which is able to perform or to generate process conditions necessary for changing a present state (input state)
Structuring of process knowledge • H. J. W. VLIEGENand H. H. VANMAL into a required state. In order to generate these process conditions a physical object must possess certain properties (P1, P2,--., P,)- The state of an ideal physical object should change as flexibly as possible while performing its task (i.e. no permanent distortion). The properties, e.g. strength, should make the transformation of power possible. For determining the required properties it is sufficient to specify the maximum or ultimate load and allowed reversible change of state which the object should be able to sustain. For simulating reversible processes this means that, Max(load) = P'Max(AS)
(N) or (Nm)
is often called a material property. In our opinion it is better to call E a material parameter with which the structure and the composition of a material are characterized. These are both part of the state. Knowledge required for process control can be structured systematically by means of the terms function (desired change of state), task (generation of process conditions), properties, state and their mutual relations as specified by Eqs (1)-(4). ASSEMBLED P R O D U C T S , E N E R G Y F L O W S Products often consist of several connected physical objects, called component parts, which make up the state of an assembled product. These component parts have to make the transport and transformation of power feasible with maximum reversible change of state of the parts (function). In order to do this, each component part must perform its own task made possible by its properties which are realized by its chosen state. Figure 2 shows an example of the energy flows among components of a turning lathe. We will elaborate this example in detail in a subsequent paper. Before determining the role of a component part, the designer first has to specify the function of each part by specifying the maximum load together with the maximum allowed reversible change of state each part should be able to sustain and/or the power each part should be able to transport and transform. A set of connected component parts can he considered as a black box, a subassembly, with its own function(s), task(s), properties and state(s).
(3)
where Max(load) is the maximum load (force or moment of force) to be sustained by the object, P the desired property of the object and Max(AS) the maximum allowed change of state of the physical object. Example of Eq. (3) : F = R * A L
(N)
where F is the tensile force, R the required strength and AL the maximum allowed elongation of an object. Knowing the required properties (P1, P2 . . . . . P,), the designer has to choose a state which possesses these properties. This means that, Pi = gi(S1, $2 . . . . . Sn)
(Nm per defined change of state)
(4)
where P is a desired property, ($1, $2 . . . . . S,) characterize the state of a physical object and 9i specifies the relation between the state and property. The physical object state is characterized by geometry, dimensions, material, composition, structure and surface finish. Example of Eq. (4) : R = A * E / L
C O U P L E D PROCESSES Several successive process steps or manufacturing operations are generally required for producing component parts or semi-manufactured articles. These process steps involve one or more physical and chemical processes, which we will call technical processes. During the successive process steps or technical
(Nm/m 2)
where R is the strength of a bar, A the cross-sectional area of the bar, E the modulus of elasticity of the material and L the length of the bar. A, E and L describe the state of the bar. The modulus of elasticity
Mp : MOMENT OF ROTAT1ON (WORKPECE)
=,
O)p : ANGULAR VELOCITY (WOR~EC.E) M= : MOMENTOFROTATION (FEED)
c p cop I
.
.
.
.
.
.
i
.
.
.
.
.
.
.
.
.
.
TOOL I
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
wo.,,.,.0. HU--I
I
(0= : ANGULAR VELOCITY (FEED) E= : CUTTING FORCE
' V¢ : CUTTING SPEED F= ; FORCE (FEED) V
,,Vo F,,V, I cU'mNG FLUID ~" ~I l F,:V,.0
I TOOL"OLO"
F,, V.
103
: FEED
Ft : RESPONSIVE FORCE
~. ,v. F, ',V,- 0 M'.m. TOOL POST
t~ : (RESPONSIVE) SPEED(-O) (]~ : HE~TTR/~ISFER
I
LATHE
Fig. 2. Energy flows among components of a lathe.
104
Robotics & Computer-IntegratedManufacturing • Volume6, Number2, 1989
processes the state of the raw material is changed into final products with a required state. The different process steps are coupled because the output of a certain process must be suitable for serving as an input for a successive step. The object of process control is to guarantee the required function, task, properties and state of the output of each process step. For this, knowledge about these coupled processes is needed. This knowledge must give insight into the relations between the ingoing material, production equipment and outgoing semi-manufactured product. In the next section we will introduce a model of technical processes based on the terms function, task, properties and state and their relations, with which knowledge required for controlling a process can be structured. MODEL OF TECHNICAL PROCESSES Figure 3 shows a model of technical processes. The physical and chemical processes controlled by man result in changes of state (AS). To accomplish this a designer chooses production equipment which generates the right process conditions (T(m)) required for realizing the desired change of state (function of the production equipment). The change of state (AS) of S(i) into S(o) causes the transformation of the properties P(i) into P(o). The properties P(o) make possible the generation of process conditions and thereby the performance of task T(o) by the product. The possibility of performing task T(o) is required for establishing the desired product function. The change of state (AS) of S(i) into S(o) occurs through the change in the process conditions generated through task T(m) performed by the production equipment. S(m) can be partly fixed and partly adjustable. The adjustable part determines the possibility of changing the process conditions and thereby task T(m) of the production equipment.
Vital for controlling manufacturing operations effectively is knowing how the adjustment of the state of the production equipment relates to the task of fulfilling the required function of the outgoing product. This demands the specification of the relations between the desired function, task, properties and state of the outgoing product and the function, task, properties and adjustable state of the production equipment involved.
MAPPING OF PROCESS KNOWLEDGE The knowledge necessary for controlling technical processes can be structured and laid down systematically in six coupled matrices. The process knowledge map (Fig. 4) puts into visual form these linked matrices modified from Refs 12-14. The matrices specify the relations between the task, properties and state of the outgoing product and the task, properties and state of the production equipment. Due to the nature of Eqs (1)-(4), these relations represent not only linear transformations but also nonlinear transformations, which are the rule rather than the exception in actual systems. The matrices are as follows.
1. Relations [T(o),P(o)] This matrix specifies the relations between the product task (to generate process conditions) and the product properties. The product must possess properties to perform its task. Specifying this matrix is the speciality of product design in co-operation with marketing, development, service, sales, etc.
2. Relations [P(o),S(o)] This matrix specifies the relations between the product properties and the product state. Development has to select a product state which guarantees that the product possesses the required properties. Specifying
T(o) : TASKOF OUTGOING PRODUCT P(O) : PROPERTIESOF OUTGOING PRODUCT S(O) : STATE OF OUTGOING PRODUCT
S(m)
T(m) : TASKOF PRODUCTIONEQUIPMENT P(m) : PROPERTIESOF PRODUCTIONEQUIPMENT S(m) : STATE OF pRODUCTION EQUIPMENT
M
I
T(i) : TASKOF INGOINGPRODUCT P(i) : PROPERTIESOF INGOINGPRODUCT S(i) : STATEOF INGOINGPRODUCT
P(rn)
T(i) P(i) S(i)=l
-I ......... ...........
T}m)
.........
IS(o) P(o) T(o)
M
: PRODUCTION EQUIPMENT
I
M'
: PROCESS
A S : CHANGEOF STATE
.~AS4- . . . . . . . . . .
TECHNICALPROCESS Fig. 3. Modelof technicalprocesses.
Structuring of process knowledge • H. J. W. VLIEGENand H. H. VANMAL
P(m)
T(m)~
~(m)
105
T(o) : TASK OF OUTGOING PRODUCT P(o) : PROPERTIES OF OUTGOING PRODUCT S(o) : STATE OF OUTGOING PRODUCT T(m) : TASK OF PRODUCTION EQUIPMENT P(m) : PROPERTIES OF PRODUCTION EQUIPMENT S(m): STATE OF PRODUCTION EQUIPMENT
:STATE OF INGOING PRODUCT I :RELATION [ T(o), P(o) ] II :RELATION [P(o), S(o)] III :REL:ATION [ S(o), T(m) ] S(i)
IV V VI
P(o)
:RELATION [T(m), P(m)] : RELATION [ P(m), S(m) ] :RELATION [S(m), T(o)]
Fig.4. Processknowledgemap. this matrix is the speciality of product development in co-operation with sales, engineering, process planning, etc. 3. Relations [S(o),T(m)]
This matrix specifies the relations between the product state and the process conditions (production equipment task) imposed on a well-determined ingoing product state. Process conditions are necessary to transform an ingoing product state into the desired outgoing product state. Specifying this matrix is the speciality of process development in co-operation with product development, process planning, manufacturing, etc. 4. Relations [T(m),P(m)]
This matrix specifies the relations between the production equipment task (to generate process conditions) and the production equipment properties. The production equipment must possess the properties to perform its task. Specifying this matrix is the speciality of production equipment design in co-operation with. process development, engineering, tool design, etc. 5. Relations [P(m),S(m)]
This matrix specifies the relation between the production equipment properties and the production equipment state. Engineering has to choose an equipment state which guarantees that the production equipment possesses the required properties. Specifying this matrix is the speciality of the engineering and tool design in co-operation with process planning, manufacturing, etc. 6. Relations [S(m),T(o)]
This matrix specifies the relations between the production equipment state and the task of the outgoing product. This matrix is the final outcome of all preceding matrices and specifies the instructions for manufacturing the products which are able to perform the required task, and so realizes the function requested by the customer.
In contrast to Suh and others 3'6'~1 who focus only on the transformation of the functional domain to the physical domain, the process knowledge map includes per process step as well as all intermediate transformations concerning the product in relation to the production equipment. In other words the process conditions and properties, the scientific base for transformations, are also considered for each process step. Hauser and Clausing 7 who do use some linked matrices, do not distinguish functions, tasks, properties and state systematically. They also do not consider the decomposition of products into parts and the coupling of manufacturing processes. This hampers the decision-making of designers and other parties involved in trying to specify the product, component parts, processes, and production equipment. Figure 5 shows the process knowledge map for manufacturing an axle by turning. 1,2,5 We will elaborate on this example in detail in a future paper.
IMPORTANCE We are convinced that the approach for structuring process knowledge as presented here will be useful, because: • it can be used for analyzing existing manufacturing processes, which is the first step in improving product quality. This makes it an important tool in quality; • it is necessary to make process knowledge explicit if the automation of manufacturing processes is pursued; • it is supplementary to existing methods like SADT 9 for describing the information flow between activities or processes. The main difference is that methods like SADT consider processes as black boxes and do not give an insight into the knowledge necessary for controlling the processes within a black box. The approach of structuring process knowledge makes it possible to open black boxes and to determine nested
106
Robotics & Computer-Integrated Manufacturing • Volume 6, Number 2, 1989
T(o)
S(m)
TA5K
~TOFBENDN IG RADIAL BEARING
P(o)
FORCE
TURNING OF
AN
ON
A
AXLE LATHE
~.,~ ~ ~~ O.\, A ~' ~% ,:~~TL.,.[K % m ~ CU~ TIN \G TDEPTH I P(
(o) i i i t
"Fj.,'tOPFA-I".k'I"A~--.~-~c"~,c_.__zol Rs l
J
Tim)
I l
",,
I
I
PURCHASING MATERIAL
IS (i)
Fig. 5. Example of the process knowledge map.
process specifications; it makes the synthesis of design decisions made by diverse engineering disciplines involved in a design project possible. This means that the approach can be used as a communication aid a m o n g designers and other interested parties for improving decision-making and supporting the management of design projects; it can play an important role in the integration of CAD and CAM. Of vital importance for the integration of CAD and CAM is the automatic matching a newly developed product with the already available production facilities of a company. This requires a knowledge base which contains all known suitable relations between product features and features of available production equipment. We think that the structure of any knowledge base for C A D / C A M can be derived from the approach for structuring process knowledge as presented in this paper. CONCLUSIONS The process knowledge m a p is a tool that helps to better specify the relations between the product function and manufacturing processes. It increases the chances of designers finding suitable solutions without too m a n y iterations. The essence of the method is the scientific approach, which includes process conditions
and properties of the product and the related production equipment per process step. The utility of the approach has not yet been proved extensively. Further research will be conducted in the specification phase of new products in particular.
REFERENCES 1. Creamer, R. H.: Machine Design. Reading, AddisonWesley 1968. 2. Doyle, L. E.: Manufacturing Processes and Materials for Engineers. Englewood Cliffs, Prentice-Hall, 1969. 3. Eekels, J., Roozenburg, N.: Ontwerpmethodologie. Delft, Delft University of Technology, 1981. 4. Feekes, G. B.: The Hierarchy of Energy Systems, from Atom to Society. Oxford, Pergamon Press, 1986. 5. Gaskin, F., McArthur, G. M.: Mechanical Engineering Science in SI Units. London, Edward Arnold, 1970. 6. Hansen, F.: Konstruktionswissenschaft, Grundlagen und Methoden. Berlin, VEB Verlag Technik, 1974. 7. Hauser, J. R., Clausing, D.: The house of quality. Harvard Business Review 3: 63-73, 1988. 8. Paynter, H. M.: Analysis and Design of Engineering Systems. Cambridge, MIT Press, 1961. 9. Ross, D. T.: Structured analysis (SA): a language for communicating ideas. 1EEE Trans. Software Engng SE3(1): 16-34, 1977. 10. Thomas, J.: Bond graphs for thermal energy transport and entropy flow. J. Franklin Inst. 292: 109-120, 1971. 11. Suh, N. P.: Development of the science base for the manufacturing field through the axiomatic approach. Robotics Comput.-Integr. Mfg 1(3/4): 397-415, 1984.
Structuring of process knowledge • H. J. W. VLIEGENand H. H. VANMAL 12. Van Mal, H. H., Kools, F., Hekma, E. J.: Technisehe processen, procesbeheersing, ontwerpkaart. Eindhoven, Eindhoven University of Technology, 1983. 13. Van Mal, H. H., Kools, F., Hekma, E. J.: !nformatieordening voor procesbeheersing, een modelmatioe benaderin 9. Eindhoven, Dutch Philips Corporation, 1985.
107
14. Van Mal, H. H., Van Someren, W. S. M., Pans, R. F. M.: Towards the integration of CAD and CAM, carpet manufacturing in the future. In: Computer Application in Production and Engineering, CAPE '86, Bo, K., Estensen, L., Falster, P., Warman, E. A. (Eds). Amsterdam, North-Holland, 1987, 879-890.