Computer-aided process planning in virtual one-of ... - Semantic Scholar

1 downloads 12667 Views 272KB Size Report
This paper reports a framework for computer-aided process planning CAPP in virtual ... manufacturing company or a manufacturing network normally consists of ...
Computers in Industry 41 Ž2000. 99–110 www.elsevier.nlrlocatercompind

Computer-aided process planning in virtual one-of-a-kind production Yiliu Tu

a,)

, Xulin Chu b, Wenyu Yang

c

a

c

Department of Mechanical Engineering, UniÕersity of Canterbury, PriÕate Bag 4800, Christchurch, New Zealand b CIMS Institute, Hefei UniÕersity of Technology, Hefei, Anhui, China School of Mechanical Science and Engineering, Huazhong UniÕersity of Science and Technology, Wuhan, Hubei, China Received 28 February 1997; accepted 9 November 1998

Abstract This paper reports a framework for computer-aided process planning ŽCAPP. in virtual One-of-a-Kind Production ŽOKP.. The emphasis will be placed on the particular problems of developing a CAPP system in virtual OKP company. A virtual manufacturing company or a manufacturing network normally consists of geographically dispersed master company’s branches, sub-contractors, joint ventures, and partners. In this paper, we will address issues raised from highly customised products, the concurrent approach of product development and production through a manufacturing network, incomplete product and production data in OKP, the need for quickly capturing marketing opportunities and early response to the customer’s demands, continuous customer influence upon production, the optimal selection of partners, and partner synthesis in virtual manufacturing. The CAPP framework proposed in this paper includes a reference architecture for structuring a CAPP system in virtual OKP, a new CAPP method which is named the ‘incremental process planning ŽIPP.’, and an optimalrrational cost analysis model. It also includes three data models for data modelling and structuring of the product, the process, and manufacturing plants used in virtual OKP. q 2000 Elsevier Science B.V. All rights reserved. Keywords: Computer-aided process planning ŽCAPP.; One-of-a-kind production ŽOKP.; Virtual manufacturing ŽVM.

1. Introduction Global competition is forcing manufacturing companies to develop the ability to quickly produce customised or even One-of-a-Kind Production ŽOKP.

)

Corresponding author. E-mail: [email protected]

products of high quality at competitive prices. The features and characteristics of OKP has been welldiscussed by a number of authors. A comprehensive literature review on these discussions and a clear definition of OKP were given by Tu w1x. In practice, an OKP company could be loosely understood as an advanced jobbing shop which provides different services Žor customised products. in a certain manufacturing domain Ži.e., ‘Kind’ in OKP., e.g., a shipbuilding manufacturer, a steel frame fabricator, a

0166-3615r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 6 - 3 6 1 5 Ž 9 9 . 0 0 0 0 6 - 8

100

Y. Tu et al.r Computers in Industry 41 (2000) 99–110

mouldrtool manufacturer, a metal-cutting machine shop, etc. The keen global competition has also caused many manufacturing companies to cut ‘fats’ Žor costs. through strategically moving their manufacturing bases and facilities from developed areas, e.g., Hong Kong, Western Europe, and the USA, to developing countries and areas which provide favourable conditions, such as lower labour costs and other cheaper resources. This movement is leading to a new manufacturing mode called virtual manufacturing. A virtual manufacturing company is a global manufacturing network where the nodes are branches, subcontractors, joint ventures, or partners of the master company w2x. These nodes have been called ‘ virtual cells’ by Tu w1x. The virtual cells of a virtual manufacturing company are geographically dispersed. Virtual manufacturing companies are seen as a new generation of companies involved in agile w6x and global manufacturing. In order to be able to control and synthesise production in such a global manufacturing network, researchers have concentrated on developing a reference architecture and control principles for this network. Virtual OKP companies which are the focus of this research, belong to this new generation. A virtual OKP company aims to produce one-of-a-kind products through a global manufacturing network. With the support of developing technologies, a virtual OKP company can flexibly capture opportunities in the market and quickly respond to the customers’ varied demands on design and manufacturing processes. As discovered by the authors through a literature review and industrial visits, the virtual OKP is likely to become a promising and important part of the manufacturing philosophy in the world. However, there are a number of issues raised from the special requirements andror problems in the virtual manufacturing, particularly in the virtual OKP. These requirements andror problems include highly customised products, the concurrent approach of product development and production, incomplete product and production data, the need for quickly capturing marketing opportunities and early response to the customer’s demands, continuous customer influence upon production, and the optimal selection of partners and partner synthesis in virtual manufacturing.

Furthermore, in virtual manufacturing, product development and the synthesis of partnership for product realisation are two important activities. Process planning will play a key role in linking these two activities. Current CAPP systems have been developed based on conventional process planning theory. In OKP, continuous customer influence on production runs concurrently with product development and production. Process planning based on conventional CAPP systems becomes very difficult in this situation. Conventional process planning theory is a linear process model. It can only run in series, not concurrently. Hence, the concepts, methods, and theories for CAPP in virtual OKP need to be developed to support the production planning and control in virtual OKP. This paper will propose a framework for CAPP in virtual OKP. The emphasis will be placed on the particular problems of developing a CAPP system in virtual OKP as mentioned above. The proposed CAPP in virtual OKP will be of particular help to the manufacturing companies which want to gain agility w6x and, consequently, their competitive edge in the marketplace. The proposed CAPP framework includes a reference architecture for structuring a CAPP system in virtual OKP, a novel CAPP method which is named the ‘incremental process planning’, and an optimalrrational cost analysis model. It also includes three data models for data modelling and structuring of the product, the process, and partner manufacturing plants used in virtual OKP. To avoid confusing readers, we would like to define the computer-aided process planning system in virtual OKP in the following according to the conventional definitions of process planning w3–5x and the particular needs of virtual OKP.

CAPP in Õirtual OKP is a computer-aided decision making support and planning tool which aims to generate the methods and cost estimates to economically and competitiÕely conÕert the technical specifications of customer’s requirements into a wanted product. The technical specifications of customer’s requirements are understood as the design of a product which is modelled by a proper product modelling tool.

Y. Tu et al.r Computers in Industry 41 (2000) 99–110

2. Literature review As described by Kidd w6x, the core feature of an agile manufacturing ŽAM. company is virtual. Recently, agile and virtual are alternatively used to name and describe a new type of manufacturing company which is referred as virtual manufacturing company in this paper. In fact, the ‘agile’ description focuses on the technical functions of virtual manufacturing companies, whereas the ‘ virtual’ description focuses on the managerial and strategic formats of the companies. One-of-a-Kind production ŽOKP., as predicated by scholars like Rolstadas ˚ w7x, Wortmann w8x, and Hirsch w9x, could be a promising manufacturing model for the factory of the future. At the same time, OKP poses some challenges to the factory of the future. Tu w1x characterised the OKP philosophy as: Ž1. high customisation, Ž2. successful product development and production in one go, Ž3. optimal or rational utilisation of technologies and resources, Ž4. adaptive production planning and control, Ž5. continuous customer influence throughout the production, Ž6. incremental process planning, Ž7. distributed control and inter-organisational autonomy, and Ž8. virtual company structure and global manufacturing. Process planning has been defined as ‘‘the subsystem responsible for the conversion of design data to work instruction’’ w3x. A more specific definition of process planning is given as, ‘‘The function within a manufacturing facility that establishes the processes and process parameters to be used Žas well as those machines capable of performing these processes. in order to convert a piece-part from its initial form to a final form which is predetermined on a detailed engineering drawing’’ w4x. According to the surveys conducted by Alting and Zhang w5x, Sood et al. w10x, Maropoulos w11x, and Kiritsis w12 x, the approaches to developing computer-aided process planning systems for various part types are based on one or more of three basic methods: variant, semi-generative, and generative. The variant approach is sometimes referred to as a data retrieving method. Through the use of a coding method which is developed according to group technology ŽGT. and consequently called GT code, a standard process plan can be stored in a database. A process plan for a new part is created by recalling,

101

identifying, and retrieving an existing plan for a similar part which is called a master part w5x. The generative approach can be defined as a system that synthesises process information in order to create a process plan for a new part automatically w4x. The semi-generative approach is intermediate between two other approaches. The variant method has advantages such as being close to the human expert’s way of process planning. Hence, it is easily implemented. However, its application has been limited by its lower automation level and flexibility. The generative method has a higher automation level and flexibility. Due to some unresolved technical and theoretical problems, however, the CAPP system developed by the generative method cannot efficiently create high-quality process plans. According to a survey conducted by Maropoulos w11x during a travel study project to several university research centres in the USA, the UK, the Netherlands, Germany, Sweden, and Denmark, it seems that most research in CAPP is based on the generative approach. The problems which are currently solved using the generative approach may be seen from four aspects. They are tooling technology and management Že.g., wear balancing, intelligent tool selection., process modelling Žmodelling tools and methods, integrated modelling of products and processes, automatic feature recognition., process planning Že.g., interface between design and scheduling, distributed process planning, feature-based design and process planning, geometric reasoning., and the integration of CADrCAPPr CAM. All the problems that exist in the research and development of CAPP systems would be the problems of developing a CAPP system in virtual OKP. In addition to these common problems in the development of CAPP systems, however, the development of a CAPP system in virtual OKP will be further faced with the special problems such as partner synthesis in virtual manufacturing, manufacturing process planning Žor estimating. and testing in different partner plants, special needs of data modelling in virtual OKP, selecting partners and building a manufacturing network for producing an OKP product through an integrated and concurrent approach of product design, manufacturing process planning, shop floor scheduling, and cost analysis. These special problems of developing a CAPP system in virtual

102

Y. Tu et al.r Computers in Industry 41 (2000) 99–110

OKP have not been particularly attacked and meaningfully covered with those researches and developments in the area of so-called computer-aided process planning. Hence, this paper does not particularly place its emphases on the common problems in the area of CAPP. Instead it focuses on the special problems in the development of a CAPP system in virtual OKP. It is worthwhile to point out that we will and we have to solve those common problems through our research project by our own methods or adopting the findings of other research projects in the area of CAPP. In the following, we would like to further summarise several interesting developments which are relevant to the work to be presented in this paper. A number of ESPRIT projects have been carried out with regard to CADrCAPPrCAM integration. IMPPACT ŽESPRIT 2165, Integrated Manufacturing of Products and Processes using Advanced Computer Technologies. has proposed several reference models, such as product model, process model, and factory model, for the integration of CADrCAPPr CAM w13x. In the process model, manufacturing activities were placed in four layers including process group, process, operation, and pass. The factory model specified the components of production systems or production mechanisms and the physical and logical relations between the mechanisms. The production mechanisms included shop floors, workcentres, workstations, tools, tool shops, tool adaptors, workers, means of transport, and carrying supports. Park and Khoshnevis w14x and Khoshnevis et al. w15x proposed a real time computer-aided process planning ŽRTCAPP. system as a support tool for economic product design. The RTCAPP system was designed as a manufacturing cost estimator to support the design of a part or a product. After a new feature has been added, the RTCAPP system will suggest a process plan and estimate the consequent manufacturing cost. The RTCAPP system is assumed to interactively work with a design system in real time. Bernardi w16x presented a skeletal plan-based methodology for emulating the human expert’s process planning procedure. In this method, a skeletal plan Žabstracted plan or fragments of a plan. is associated with manufacturing features. The generation of a process plan is consequently subject to this

sequence: abstraction, selection, and refinement. The geometrical andror technological representation of a part in Bernardi’s method allows the process planning system to recognise the relevant features. Then the skeletal plans which are related to these features will be selected, merged, and refined by the process planning system until a complete process plan is created. Cho et al. w17x presented an integrated system of CAPP and shop floor control. In their presentation, the authors considered that a CIM architecture would generally consist of three hierarchical levels: shop floor, workstation, and equipment. The CAPP consists of machining feature identification, definition, classification, representation, and reasoning, provided through a CAD model of a product. The design of a part is converted into a feature graph. The feature graph will be converted by the CAPP into a task graph which is a process plan with alternative feasible processes. Finally, the task graph will be hierarchically decomposed into operation schedules at the three CIM architecture hierarchical levels. To find an optimal shop floor schedule with a process plan consisting of alternative processes, Chen and Khoshnevis w18x have proposed a heuristic method consisting of a search method and a concurrent assignment algorithm. The STEP w19x has defined an international standard for the representation and exchange of computerised product information throughout the life cycle of a product. This standard is independent of particular computer systems, and enables consistent implementations across multiple applications and systems. Candadai w20x analysed the information necessary in agile manufacturing and defined three types of information for product design and production. These are product data, process data, and manufacturing plant data. Chu and Holm w21x developed a control system for product manufacturability control in concurrent engineering environments. A hierarchy of manufacturability criteria was established for measuring the manufacturability of a product. Methods for calculating and setting manufacturability criteria were also developed. Tu w1x proposed a reference architecture for an agile OKP. An agile OKP was viewed as a flexible firm which consists of marketing networks, manufac-

Y. Tu et al.r Computers in Industry 41 (2000) 99–110

turing networks, and several centralised intelligent agents. The networks were seen to formulate the flexible extensions of the firm. The intelligent agents constitute the core of the firm. These intelligent agents included an information processing agent, an engineering agent, a managerial agent, and manufacturing agents. Zhang w22x proposed several methods of product data modelling and partner company data modelling for the development of a conceptual database model to support partner synthesis in virtual manufacturing. This research was carried out through a case study conducted in an electronics company in Hong Kong which produces customised products in its distributed manufacturing network. The contribution of this research is that it combined the variant of products and the variant of production resources and capacities in the partner companies through the data modelling of products and partner companies.

103

3. Reference architecture for the CAPP system in agile OKP According to our studies and observations on various manufacturing companies in the world today, virtual manufacturing companies already existed w1,22x. However, most companies have no agility to cope with market changes. Tu w1x proposed a reference architecture which focuses on the production planning and control problems in a virtual OKP company. Based on that reference architecture, and the characteristics of the virtual OKP, a reference architecture of a CAPP system for a virtual OKP company is proposed in Fig. 1. As shown in Fig. 1, once the technical specifications of the customer’s requirements have been received by the Project Manager, it will be decomposed into or realised by a number of parts or sub-assemblies according to previous technological

Fig. 1. A reference architecture of the CAPP system in a virtual OKP company.

104

Y. Tu et al.r Computers in Industry 41 (2000) 99–110

experiences and a cost estimate made by the cost analysis model Žto be discussed in Section 5.. The manufacturing features, such as sculptured surfaces of a mould or a bended metal plate in sub-assemblies of a steel structure, can be identified. Then the process planner can use the Searching Model to search for these features or similar features Žcalled primitiÕes. from the STEP-Based Product Database. If a primitive is found, the process planner can further search for a process plan for the primitive from the GT code index Process Base by using the group technology ŽGT. code. The process plans for all these existing features or similar features are gathered together to formulate a process plan for producing the product. This process plan is named here as primitiÕe plan which will be used through the rest of this paper without any further explanation. If a feature or primitive cannot be found, the process planner will leave it empty in the primitive plan. It shall be mentioned here that a primitive plan is only a rough process plan. This rough process plan may include a lot of alternative manufacturing processes for creating the same existing or similar feature and empties or uncertainties of process for the new features or possible changes of product design. Hence, the process planner needs to use the Incremental Process Planning Model and Cost Analysis Model to further refine the primitive plan until a feasible and rational process plan is developed. Since the searching model, as shown in Fig. 1, can easily be found from several existing variant or generative CAPP systems w5,10,12x, our research focuses mainly on the development of the incremental process planning model and the cost analysis model, and the particular problems for data modelling and structures in agile OKP.

4. Incremental process planning As summarised by many authors, including Hirsch w9x, Rolstadas ˚ w7x, Kuhlmann w23x, and Tu w1x, the production of an OKP product is started with incomplete product data and is carried out under continuous customer influence. The development and manufacture of an OKP product is concurrently approached in an OKP company. The product data become clear through production and a consistent

collaboration between the manufacturer and the customer. This evolutionary and concurrent product development and production makes it difficult for an OKP company to gain the agility to change the process plan and respond to the customer’s input. To cope with this problem, a novel process planning method called incremental process planning ŽIPP. and a cost analysis model are developed as the core of the CAPP system in virtual OKP. A primitive plan, as mentioned in Section 3 of this paper, is merely a skeletal process plan. A primitive plan is either provided by the searching model, or created by the process planner according to some basic and critical features Že.g., base faces, functional surfaces, etc.. of a product. According to the experience gained by Khoshnevis et al. w15x and Park and Khoshnevis w14x, a primitive plan for developing a product can be created in the earlier stage of the product design. Hence, the primitive plan is robust with respect to evolutionary Žor frequent. changes of product design in OKP. The IPP method is used to extend or modify the primitive plan according to the new features which are identified from a product design Ži.e., technical specifications of customer’s requirements. until no more new features can be found. The IPP model is illustrated in Fig. 2. According to the shape type, size accuracy, surface roughness and overall structure of an identified feature, as well as material, weight, and production volume of a part, a series of processes are either searched from the GT code index Process Base or designed by a process planner. The process planner can design the processes according to his experiences or through some industrial tests, e.g., to cut an uncertain sculptured surface of a mould, or to make a mock-up for a steel structure. The processes for a new feature need either to be added to or inserted into the primitive plan by the computer system. To add the processes to the primitive plan is relatively simple. However, in inserting the processes into the primitive plan, the difficulty is to find the insertion point and to correspondingly modify the process plan following the insertion point. To solve this problem, a heuristic search method has been developed by Khoshnevis et al. w15x. In order to evaluate the manufacturability of a part, a hierarchy of manufacturability criteria ŽMC.

Y. Tu et al.r Computers in Industry 41 (2000) 99–110

Fig. 2. The flow chart of incremental process planning.

is developed. The MC are divided into two categories w21x. They are economic criteria ŽEC. and technical criteria ŽTC.. The EC is covered by the cost analysis model which will be further discussed in Section 5. Within the IPP method, only the TC is used to assess the manufacturability. The TC includes standardisation of the part shape, manufacturing processes to be used, material and tolerance of the part, rationality of the machining dimension pattern, machinability, and quality of a partner plant. Finally, it should be pointed out that a complete process plan generated by the IPP model may include alternative processes, too. This means that a part can be processed by alternative machines in alternative sequences in a partner plant. The part can also be produced in Žor sub-contracted to. alternative partner plants. This process plan needs to be further refined by the cost analysis model under considerations of cost and lead time Žor economic criteria..

5. Cost analysis model In view of the requirements of virtual OKP, the cost analysis ŽCA. model of the CAPP system should be able to generate an optimalrrational Žor mini-

105

mumreconomic. cost estimate for an OKP product according to a complete process plan generated from the IPP model. Consequently, it refines the complete process plan so that the final process plan does not include the alternative processes. Hence, the CA model determines the partner plants for the production of an OKP product, and suggests the production resources to be used in each selected partner plant. For the optimalrrational cost analysis, a heuristic method is developed by dynamic programming techniques. The costs which are analysed by the CA model include the manufacturing costs Žset-up q material handlingq machiningq tool changeq tool costs., operation cost Žcounted by the make-span in a virtual cell., and transportation cost Žcost for transporting a part or parts between virtual cells.. To simplify the problem, the transportation cost is assumed to be a constant. This assumption is reasonable since the transportation fee between a partner plant and the final assembly plant can, in practice, be pre-determined. The dynamic programming networks for the optimalrrational cost analysis are illustrated in Fig. 3. Using the dynamic programming network as shown in Fig. 3Ža., the optimal manufacturing cost and the optimal operation time Žor operation cost. can be, respectively, derived. However, a compromise needs to be struck between the optimal manufacturing costs and the operation cost. Hence, the Pareto Optimality and trade-off curves w24x are used to find the compromising optimal solution in accordance with different strategies Žfocusing on cost or lead time. of the master company. Once the compromising optimal result has been found, the result together with the transportation cost, will be used as the weights for the network as shown in Fig. 3Žb.. Through further dynamic programming, a partner plant can be selected by the CA model. Based on the above calculations, the estimated optimalrrational cost and lead time for the OKP products to be fabricated by a group of selected partner plants can be generated by the CA model. If the cost is within the budget and the lead time does not exceed the due date, a refined process plan, which does not include any alternative processes, will be issued to the partner plants. Otherwise, detailed feedback on the cost and lead time will be reported to the Project Manager Žsee Fig. 1..

106

Y. Tu et al.r Computers in Industry 41 (2000) 99–110

Fig. 3. Dynamic programming networks for optimalrrational cost analysis.

It should be mentioned that, by applying probabilistic dynamic programming techniques, the production of OKP products can be simulated Žor estimated.. Consequently, the minimum, maximum, and most likely value of the cost and the lead time for each of the parts can be simulated. Furthermore, the production can be controlled by using the program evaluation and review technique ŽPERT.. It shall be mentioned that the IPP model and CA model will be iteratively used during the production planning period until a satisfactory agreement on product design, process planning, sub-contracts, and a whole production schedule can be made between the project manager, customers, and all partner manufacturing plants. In fact, as mentioned in Section 3, the CA model is first used to help the generation of a primitive plan. During the production, the IPP model and the CA model are used to integrated control product design, process planning, and partner synthesis in virtual manufacturing.

6. Data modelling and structure in virtual OKP

briefly describe these three data models in the following. 6.1. Product model As shown in Fig. 1, the emerging STEP w19x standard which aims at meeting the need for rigorous and complete representation of product data is employed for product data modelling. A STEP-based product reference model is developed. The primary use of this model is to describe all attributes of a product from its design to its manufacture. This model consists of four sub-models: geometric feature model, shape model, tolerance model, and administration model. The geometric feature model is used to describe portions of the skin of a part that conforms to some machining methods and the relationships among these manufacturing features. The shape model uses a CSGrB-rep data structure to represent the part shape. The tolerance model describes the geometric and shape position dimension tolerance of a part. The administration model includes product material, overall shape dimension, batch size, predicted number, and GT code. 6.2. Process model

To create the databases as shown in Fig. 1, three data models, i.e., product model, process model, and manufacturing plant model, are developed. We

It is a reference model of manufacturing processes. In this model, manufacturing processes are

Y. Tu et al.r Computers in Industry 41 (2000) 99–110

classified as primary processes, primaryrsecondary processes, and tertiary processes w20x. Primary processes are main-shape generating processes such as casting and forging. Primaryrsecondary processes generate both the main-shape of the part and the manufacturing features. Tertiary processes do not affect the geometry and comprise mainly grinding or finishing operations. We use a set of data to describe the capabilities of manufacturing processes, which include process name, type, economic production quantity, machining accuracy and surface roughness, and suitable material shape.

6.3. Manufacturing plant model As shown in Fig. 1, a diverse amount of data about partner plants is needed for the IPP and CA models. Hence, a manufacturing plant model is developed. The data in this model include production resources, production states, links with the master company, operational data, quality standards, and plant capabilities. Methods for developing the conceptual data model have been proposed by Zhang w22x to support partner synthesis in virtual manufacturing. They have been adopted to develop the manufacturing plant data model.

107

7. Implementation Fig. 4 shows a part of the steel frame of a rail station in Hong Kong. The steel frame construction project was contracted to Argos Engineering and Heavy Industries which is a well-known and experienced steel frame construction company. Argos has its own manufacturing plants in south part of China and somewhere else in Southeast Asia. It also often sub-contracts the parts of its projects Žor products. to some other companies in China and Southeast Asia. In short, it is a typical virtual OKP company. The design of the overall steel frame was done by an engineering design company in Hong Kong. The project manager in the Argos used the system as shown in Fig. 1 to plan the necessary manufacturing processes and sub-contractors. According to the design, the overall frame structure was decomposed into a number of sub-assemblies. One of these subassemblies is shown in Fig. 4. The critical problem here to the company is to identify those engineering features, such as bended metal plate, joints, holes, etc. In combination with the materials to be used, these engineering features determine the way Žor processes. to be used to produce it. In practice, most of these features are ‘old features’ which have been done by the company through other projects, and the

Fig. 4. A sub-assembly in the railway station steel structure.

108

Y. Tu et al.r Computers in Industry 41 (2000) 99–110

process plan can be searched from the company’s process knowledge base. The processes for those new features are planned through the incremental process planning model. The planning procedure not only includes those modifications or retrievings made on the CADrCAM system or knowledge base, but also evolves a number of discussions between the project manager, the design company, and those manufacturing plants in south part of China. To support the decisions on the processes to be used to formulate these new features, sometimes mock-ups have to be made to test the new process designs or processes are simulated by computer simulation systems. To meet the cost considerations and manufacturing capacities of the partner plants Žor sub-contractors., some changes have to be made to product design. Most of these design changes will need negotiations and agreements between the manufacturer Žor project manager. and the customer. The geometric model Žor shape. of a sub-assembly, for instance, has to be changed due to the limited capacity of the bending equipment to bend a large metal plate. To get an optimalrrational cost of the whole project, the overall project has to be properly decomposed into these sub-assemblies. The sub-assemblies are then sub-contracted to the company’s own plantŽs. and some other partner manufacturing plants through calculations made by the cost analysis model as presented in this paper. To be more realistic and accurate, all the sub-contractors’ plants need to be further modelled by computer simulation systems based on the data from the Plant Database of Partners and the production schedules will be tested on the simulation systems. The feedbacks will possibly lead changes of product design and process plan. A simulation model for the plant of a sub-contractor who produces the sub-assemblies like the one illustrated in Fig. 4 was developed based on a production simulation package called Pro-Model. The detailed manufacturing processes and production schedule were tested on the simulation model w25x. Finally, it shall be mentioned that all the sub-assemblies were further transported from these sub-contractors to Hong Kong, and assembled into the steel frame of the railway station on the site. In addition to this steel construction case briefly described here, Tu et al. w26x also implemented the methods and principles presented in this paper for

concurrently designing and manufacturing plastic injection mouldsrtools.

8. Conclusions and future developments In this paper, a reference framework of the CAPP system for a virtual OKP company has been proposed. An industrial test is also briefly described in Section 7 of this paper. From this paper, it can be seen that a virtual OKP company, supported by the reference framework and the relevant methods presented in this paper, can be expected to concurrently conduct its product design, process planning, cost analysis, and production scheduling through involving its customers and its all functional departments Žor teams. through the whole product development cycle. In the paper, the proposed IPP method is a feasible and useful process planning method in OKP to deal with the continuous customer influences through the production. By the IPP method, the knowledge and skills of human process planning experts can be well-incorporated with computer’s computation and memory capacities to generate a proper process plan for a customised product Žor OKP product.. The CA model in the proposed CAPP system is a development in CAPP. As presented in this paper, this model is used as an optimisation tool to select production resources among the partner plants and finally refine the process plan.

References w1x Y.L. Tu, A framework for production planning and control in a virtual OKP company, Technical Papers of North American Manufacturing Research Institution of SME 1996, SME, pp. 121–126. w2x A. Rolstadas, ˚ Beyond year 2000—production management in the virtual company, production management methods, IFIP Transactions B 19 Ž1994. 3–10. w3x C.H. Link, CAPP-CAM-I Automated Process Planning System, Proceedings of the 1976 NC conference. w4x T.C. Chang, R.A. Wysk, An Introduction to Computer-Aided Process Planning Systems, Prentice Hall, NJ, 1985.

Y. Tu et al.r Computers in Industry 41 (2000) 99–110 w5x L. Alting, H. Zhang, Computer aided process planning: the state-of-the art survey, International Journal of Production Research 27 Ž4. Ž1989. 553–585. w6x P.L. Kidd, Agile Manufacturing, Forging New Frontiers, Addison-Wesley Publishers, 1994. w7x A. Rolstadas, ˚ ESPRIT basic research action no. 3143—FOF production theory, Computers in Industry 16 Ž1991. 129–139. w8x J.C. Wortmann, Towards One-of-a-Kind Production: the future of European industry, in: E. Eloranta ŽEd.., Advances in Production Management Systems, Elsevier, North-Holland, 1991, pp. 41–49. w9x B.E. Hirsch, Future research in One-of-a-Kind Production, in: B.E. Hirsch, K.-D. Thoben ŽEds.., One-of-a-Kind Production: New Approaches, Elsevier, North-Holland, 1992, pp. 87–94. w10x S. Sood, P.K. Wright, J. MacFarlane, Process planning: a review, DE-Vol. 66, Intelligent Concurrent Design: Fundamentals, Methodology, Modelling and Practice, ASME, 1993, pp. 45–54. w11x P.G. Maropoulos, Review of research in tooling technology, process modelling and process planning: I. Tooling and process modelling, II. Process planning, Computer Integrated Manufacturing Systems 8 Ž1. Ž1995. 5–20. w12x D. Kiritsis, A review of knowledge-based expert systems for process planning, methods and problems, Advanced Manufacturing Technology 10 Ž1995. 240–262. w13x O. Bjorke, O. Myklebust, IMPPACT: Integrated Modelling of Products and Processes using Advanced Computer Technologies, Tapir, Trondheim, Norway, 1992. w14x J.Y. Park, B. Khoshnevis, A real-time computer-aided process planning system as a support tool for economic product design, Journal of Manufacturing System 12 Ž2. Ž1993. 181– 193. w15x B. Khoshnevis, J.Y. Park, D. Sormaz, A cost based system for concurrent part and process design, The Engineering Economist 40 Ž1. Ž1994. 101–124. w16x A. Bernardi, PIM-skeletal plan-based CAPP, Computers in Industry 23 Ž1993. 87–97. w17x H. Cho, A. Derebail, T. Hale, R.A. Wysk, A formal approach to integrating computer-aided process planning and shop floor control, Journal of Engineering for Industry 116 Ž1994. 108–116. w18x Q. Chen, B. Khoshnevis, Scheduling with flexible process plans, Production Planning and Control 4 Ž4. Ž1993. 333–343. w19x ISO, TC 184rsc4 n81 ISO DIS 10303-1 Product Data Representation and Exchange: I. Overview and Fundamental Principles, 1993. w20x A. Candadai, Information Needs In Agile Manufacturing, Engineering Data Management: Integrating the Engineering Enterprise, 1994, pp. 101–110. w21x X. Chu, H. Holm, Product manufacturability control for concurrent engineering, Computers in Industry 24 Ž1. Ž1994. 29–39. w22x W.J. Zhang, On methodology of developing a conceptual database model to support partner synthesis in virtual enterprises, Proceedings of the 3rd CIRP Workshop on Design

w23x

w24x w25x

w26x

109

and Implementation of Intelligent Manufacturing Systems, Tokyo, Japan, 1996, pp. 137–144. T. Kuhlmann, A Revolving Planning and Control System, Proceedings of New Approaches towards One-of-a-Kind Production, Bremen, 1991, pp. 95–112. W.L. Winston, Operations Research, Applications and Algorithms, Duxbury Press, Belmont, California, USA, 1994. Y.L. Tu, Z.B. Jiang, F.C.K. Lau, A Shop Floor Simulation Platform for a Virtual One-of-a-Kind Production, Proceedings of 1st International Conference on Engineering Design and Automation, Bangkok, Thailand, March 18–21, 1997, pp. 397–400. Y.L. Tu, W.Y. Yang, Y.L. Xiong, A concurrent manufacturing strategy for producing OKP products with complicated sculptured surface, The International Journal of Advanced Manufacturing Technologies 14 Ž1998. 93–98.

Yiliu Tu received a BSc in Electronic Engineering and an MSc in Mechanical Engineering both from Huazhong University of Science and Technology ŽHUST., P.R. China, respectively, in 1982 and 1985. From 1985 to 1990, he was a lecturer at the Department of Mechanical Engineering Ž1. of HUST. In 1993, he received his PhD from Aalborg University of Denmark, and then worked at the Department of Production as a post-doctoral research fellow from 1993 to 1995. From 1995 to 1997, he was an assistant professor at the Department of Manufacturing Engineering and Engineering Management, City University of Hong Kong. Since 1997, he has been a lecturer at the University of Canterbury, New Zealand. His present research interests are virtual product design and development in One-of-a-Kind Production, IDSS ŽIntelligent Decision Support Systems. in maintenance, and IT ŽInformation Technology.. He is a senior member of SME and CASArSME, and a member of Institution of Professional Engineers New Zealand ŽIPENZ..

Xuening Chu received his BSc and MSc in manufacturing Engineering from Beijing University of Aeronautics and Astronautics ŽBUAA., respectively, in 1983 and 1986. During 1992–1993, he was a visiting scholar at Aalborg University, Denmark. Since 1997, he has been a professor at Hefei University of Technology, where his main research subjects are Computer-Aided Process Planning ŽCAPP., Product Modelling, Concurrent Engineering and Agile Manufacturing.

110

Y. Tu et al.r Computers in Industry 41 (2000) 99–110 Wenyu Yang received his PhD in Mechanical Engineering from Huazhong University of Science and Technology, P.R. China, where he is currently a post-doctoral research fellow. His research work has centred on sculptured surface CADrCAM, design for manufacture, CNC machining collision avoiding algorithm, and development of 3D visualisation tool for detecting the machining collisions.