process planning - Computer Aided Process Planning. (CAPP). Nevertheless ... can be planned; (b) programming and installation are comparatively simple; (c) ...
Knowledge Management in Process Planning 1
2
1
2
B. Denkena (1), M. Shpitalni (1), P. Kowalski , G. Molcho , Y. Zipori 1 Institut für Fertigungstechnik und Werkzeugmaschinen (IFW) Produktionstechnisches Zentrum Universität Hannover 2 Laboratory for CAD and LCE, Faculty of Mechanical Engineering Technion - Israel Institute of Technology 32000 Haifa, Israel
2
Abstract Considerable research and development efforts have been devoted to Computer Aided Process Planning (CAPP). Nevertheless, because the CAPP problem is complex and is characterized by many interdependent technical and business parameters and variables, no viable off-the-shelf solution is yet available that can be easily or widely implemented in industry. This paper presents an overview of the CAPP field and describes a holistic component manufacturing process planning model based on an integrated approach combining technological and business considerations. The model was derived based on available literature, an overview of the state-of-the-art in Digital Manufacturing, Product Lifecycle Management (PLM) and CAPP solution providers, and a survey of Small Medium Enterprise (SME) manufacturers. This model will form the basis for developing improved decision support and knowledge management capabilities to enhance available CAPP solutions. Keywords: CAPP, Abrasive processes, Knowledge management.
Hence, process planning has a major impact on manufacturing profitability. Emphasizing the business aspects of CAPP will enhance existing CAPP solutions that currently focus on technological feasibility and optimization. In addition, this paper presents PLM/CAD-CAPP solutions available in industry today from leading providers such as UGS, PTC, and Dassault Systemes, as well as their current development directions. Moreover, we discuss some additional complementary capabilities, identified during the literature review, market survey and industry survey, required to complete the holistic solution.
1 INTRODUCTION One of the most important steps in converting a design concept into a manufactured product is process planning. Such planning determines the manufacturing operations, operation sequence and resources required to manufacture a product based on an engineering drawing or a CAD model. A process plan elaborates the machines, setups, tool specifications, operation time estimates, etc. required to convert raw material into a part [1]. Traditionally, process planning was performed manually from scratch, hence requiring retrieval and manipulation of a great deal of information from many sources, including established standards, machinability data, machine capabilities, tooling inventories, stock availability and existing practice. Much research and development has been devoted to developing a computerized solution for process planning - Computer Aided Process Planning (CAPP). Nevertheless, because the CAPP problem is complex and characterized by many interdependent technical and business parameters and variables, no viable off-the-shelf solution can yet be easily or widely implemented in industry. Moreover, because expert process planners are becoming an expensive and rare resource in industry, their productivity, effectiveness and consistency must be enhanced through improved decision support tools and knowledge management capabilities. In this paper we have developed an ontology for the “extended" process planner environment and the process planner decision-making process model. These models are used as the basis for developing decision support and knowledge management templates and capabilities to support the process planning decision process for component manufacturing in the cutting and abrasive process domain. In these models a holistic approach was adopted that incorporates traditional technological aspects of process planning as well as complementary business aspects. Since the process plan is used in production scheduling as well as in machine control, it affects production efficiency, final cost and product quality [2].
Annals of the CIRP Vol. 56/1/2007
1.1 CAPP Overview There are two basic approaches to computer-aided process planning — variant and generative [3].
Variant CAPP was the first approach used to computerize planning techniques. It is based on the notion that similar parts will have similar process plans. Part coding and classification based on group technology are used to implement this concept. A “standard” plan is formulated and stored for each part family [4]. Variant CAPP has the following advantages: (a) once a standard plan has been written, a variety of components can be planned; (b) programming and installation are comparatively simple; (c) the system is understandable, and the planner has control over the final plan; and (d) it is easy to learn and use. Yet several problems are also associated with variant CAPP: (a) the components to be planned are limited to previously planned similar components, and process optimization is not included; (b) experienced process planners are still required to modify the standard plan for a specific component; (c) variant planning cannot be used in an entirely automated manufacturing system without additional process planning [2]. Generative CAPP envisions creation of a process plan from information available in a manufacturing database
-175-
doi:10.1016/j.cirp.2007.05.042
x x
without human intervention. Upon receiving the design model, the system is able to generate the required operations and operation sequence for the component [5]. A generative process-planning system comprises three main components: (a) part description, (b) manufacturing databases, and (c) decision-making logic and algorithms. Because the definition of generative process planning used in industry today is somewhat relaxed, any system containing some decision-making capabilities on process selection is called a generative system. Generative process planning is regarded as more advanced than variant process planning. Ideally, a generative process-planning system is a turnkey system with all the decision logic built in. However, due to differences among manufacturing shops, decision logics have to be customized for each shop. The generative process-planning approach has the following advantages: (a) process plans are generated rapidly; (b) new components can be planned without relying on previous similar parts; and (c) there is potential for integrating with automated manufacturing facilities to provide detailed objective control information [2][3]. Most research systems are of the generative CAPP type [6].
PLM KID management capability Interface to Enterprise Resource Planning (ERP) and Manufacturing Execution Systems (MES) systems Figure 1 presents the current system configuration, as well as the strategic aim of future systems. Current and future solutions differ mainly in the applicability of the PLM and Digital Manufacturing component as the KID backbone to support CAD-CAPP-CAM systems and the not-yet-existing feedback loop for improving design and process planning productivity and profitability. Current CAPP solutions are suited to mass production of specific product types, e.g., the power train market, the automotive industry, and, more recently, the aerospace industry. CAD
CAM
PLM (Digital Manufacturing component) x Product definition x Part Process Plan x Resource Data (tools, machines, fixtures, etc) x Resource Content (tool database, etc) x Manufacturing Knowledge
1.2 Digital Manufacturing Overview Historically, manufacturing process planning was manual and based primarily on the experience and knowledge of individual process planners, who typically developed manufacturing process plans after product planning. The failings of this sequential approach contributed to the advent of concurrent engineering, enabling simultaneous product and process planning. Most digital manufacturing technology suites are built around this core manufacturing process function. Digital manufacturing has become a key component of Product Lifecycle Management (PLM). PLM systems are the current solution for managing the integrated Knowledge Information and Data (KID) regarding product design, production process and production capabilities. Digital manufacturing seeks to define and manage manufacturing process information and support effective collaboration among engineering disciplines by using full digital product and plant definitions. It facilitates a holistic view of product and process design as integral to the product lifecycle and enables referencing to process constraints and capabilities during product design. Digital manufacturing supports product data release, engineering change management, factory modeling, visualization and collaboration, simulation of operations, and ergonomic and human factor analyses. The PLM platform is, therefore, the natural backbone for enabling technological KID management for CAPP.
Machine Code & Process Plan
Future Current
ERP/MES
Figure 1: Current and future process planning solutions. Figure 2 presents the most common knowledge-based methods used in CAx. Historically, computer aided design was the first to support knowledge in engineering processes and thus incorporates the largest number of knowledge management methods. Computer aided process planning and manufacturing have also introduced similar techniques, but knowledge generation and transfer must be carried out separately for each application and specific environmental conditions. Knowledge-based CAx
CAD
CAPP
CAM
x Macro programming x Element libraries x Parameters x Associativity x Features x User defined features x Model checks x Templates x Configuration tools
x Macro programming x Features x Templates x Knowledge-based search
x Macro programming x Parameters x Features x Templates x Simulation
Figure 2: Established knowledge tools in CAx systems. Generally, solution providers focus on CAD and CAM tools rather than on CAPP. However, available CAM solutions have been expanded to include some CAPP functionalities, e.g., feature recognition, process plan definition for compound features (including tool selection) and rule-based prioritization of machine operations. Because CAM systems are well accepted and implemented in the market, their expansion was appropriate. Moreover, cooperation between process planners and manufacturers—or shop floor—is more common than between process planner and designer. This trend of including CAPP capabilities in CAM systems is contrary to the expected benefits from collaboration between designers (CAD) and process planners (CAPP). Integrating product and manufacturing process design is vital, for it facilitates optimization of product cost. This integration also reduces product development lead time,
2 LEADING PLM/CAPP SOLUTIONS This research encompasses process planning of abrasive component manufacturing and therefore excludes additional CAPP capabilities such as assembly planning and shop floor planning. A field study collected information on process planning solutions available in industry and mapped the state-of-the-art in the field. Interviews with leading solution providers, among them UGS, PTC and Dassault Systemes, demonstrate the state-of-the-art in digital manufacturing. The companies employ similar strategies offering manufacturers an integrated solution for digital management of product and process KID. All of the proposed solutions include the following components: x x x
CAPP
CAD system Process planning capabilities CAM system
-176-
hence lowering R&D costs, and promises improved product quality and performance, thus enabling a more timely and profitable solution for industry [7].
3.3 Industrial Field Study Conclusions Process planning combines both technological and business considerations to select the optimal process to ensure part quality in accordance with specifications and at minimal cost. Nevertheless, our overview of SME manufacturers shows that:
In the future, solution providers will be able to interface CAD and CAM systems to a PLM backbone with the following digital manufacturing capabilities: x x x x x x
x
Rough cut process planning Resource management (tools, machines, fixtures) Knowledge management and rule base for process planning Process template management Change management of product and process data Part and process KID classification
x x x
2.1 Market solution discussion x
x x x
x
Currently, no complete solution is appropriate for SMEs with small batch production and vast product varieties (outsourcing market). Existing solutions suit in-house mass production manufacturers. CAPP solutions require significant investment to describe the specific environment of each variable. Existing solutions focus mainly on prismatic parts, with minimal support for freeform process planning. Existing CAPP solutions focus on detailed planning and finding the optimal solution for machine processing planning. They do not include business considerations that often determine process plan variables such as technology or machines selected.
3.4 Survey at IFW To determine the requirements and conditions of process planning in experimental situations, a survey was conducted at IFW. The goal was to identify relevant process data used for planning research experiments in an academic environment. Twenty researchers were questioned about necessary preparation and needed data while planning experiments. The results revealed that process parameters such as feed or cutting speed are often searched for. Yet frequently it is indirect and processed data sources such as diagrams or pictures that are searched. While such sources visualize parameter relations, they also make them more difficult to find and extract. The conclusion is that organizational knowledge tends to be converted into tacit knowledge and thus is difficult to extract.
3
SURVEY OF CAPP AND PROCESS PLANNING IN SME MANUFACTURING Experienced process planners from five SME manufacturing facilities and twenty expert researchers were interviewed to gather baseline information on process planning and CAPP. 3.1 SME Industrial Survey Four Israeli manufacturing facilities were selected initially. The interviews sought to understand decision-making in process planning as well as to collect available or missing decision support tools and to pinpoint decision processes where such tools would contribute most. The focus was on plants planning small batch production processes rather than mass production processes. Interviewees were production managers, process planners and CNC programmers, all experts in process planning. Analysis of the results of the first round of interviews yielded a preliminary decision model and an ontology structure for describing the process planners’ immediate environment (inputs and outputs of the decision process). Results were validated by a return visit to one of the plants for model verification as well as by model validation by interviewing a fifth process planner.
4
x x
x x x
MODELING THE ENVIRONMENT
PROCESS
PLANNER
4.1 Decision Process Model Whether a process planning decision procedure is completely manual or computer aided, it will inevitably include the following stages: x x x x x
Receiving order and design details. Selecting raw material and shape. Selecting process technology. Determining process order. Preparing processing plan, which includes: o Selection of machining operations o Sequencing of machining operations o Selection of cutting tools o Determination of setup requirements o Calculations of cutting parameters o Selection/design of jigs and fixtures o Planning tool path o Estimating processing setup costs and times. x Generating planning output (CNC program, documentation). Figure 3 presents the upper aggregate level of the process planning decision-making procedure. This model is a drill-down model where each activity is further explicated in a lower level model. Figure 4 demonstrates one such drill-down process for rough planning. For such applications, OPM provides a good modeling tool [8].
3.2 The Questionnaire x
Contrary to CAD/CAM, CAPP is not implemented in SMEs. Management of infrastructure knowledge (machining equipment and tools) is lacking. Production process knowledge is not managed, but rather only documentation of work performed. Difficulties exist in identifying similar jobs (e.g., in case of significant engineering change, job is considered entirely new), with reliance only on employee memory. Digital information is transferred from designer to manufacturer (i.e. STEP and PDF files), but there is no digital data collaboration for transferring feedback from manufacturer to designer.
Describe the environment, technologies, machines and existing CAD/CAM tools. Describe the information included in every new order (drawing, CAD model, routing, file formats, dimensions and tolerances). Describe the decision process for selecting a manufacturing technology, specific machines, tools, jigs, dividing operations into elements, parameters for each element, setups. Describe the interaction between product designer, process planner and manufacturer. Describe the relevant KID management capabilities. Define business parameters impacting process plan (e.g., priorities in manufacturing resource allocation).
-177-
process-planning ontology have been identified and verified, and the ontology is now being generated using the Protégé tool. Figure 5 presents the data base classes relevant to the facility data, and Figure 6 shows the database classes relevant to the order data. The rough plan and detailed plan are similarly described.
4.2 Ontology Structure The process planner environment was described by developing an ontology. For a particular domain, an ontology expresses the set of terms, entities, objects and classes and the relationships between them, and provides formal definitions and axioms that constrain the interpretation of these terms [9]. The ontology is often captured in some form of semantic network—a graph with nodes representing concepts or individual objects and arcs representing relationships or associations among the concepts [10].
5
THE HOLISTIC PLM/CAPP SOLUTION
In section 2, the current available solutions and current development strategies of the PLM market were discussed. In this section we identify and discuss a complementary set of tools and capabilities required for providing a holistic PLM/CAPP solution. Moreover, the added contribution to existing research directions is discussed.
The ontology’s aim is to make explicit the knowledge contained within software applications, as well as that within an organization and the business procedures in a particular domain. All the instances and classes of the
Rough plan: •Selected technology •Selected machines •Additional treatments •Finishing processes •Quality assurance •Sequencing
Facility Data
Rough cut planning: based on commercial, logistic and technical considerations
Output: •Machine code •List of tools •List of jigs •Process documentation
Detailed planning machine/ operation
Order Data
CAD
ERP/MES
PLM process KID manager
CAPP
CAM
Figure 3: A complete process plan.
Determine Technology
Review previous plans of similar products/orders CAD
Order data
Evaluate commercial value of the order: • Price • Profitability • Past experience • Future potential
•Engineering KID •Logistic and commercial KID
Facility Data Analyze resource availability ERP/MES
Allocate required sources: In-House or out-source
Analyze drawings & specifications: •Part dimensions •Tolerances •Raw material •Surface finish
Determine routing (machine type, and sequence)
* Rough plan
Figure 4: The drill down rough plan procedure.
-178-
FACILITY DATA
Resource availability
Variable commercial data
Implicit facility KID
Explicit facility KID
Machines
Variable work costs
Tools
Facility fixed costs
Raw material
Required profit margin
Manpower
Facility responsiveness
Subcontractors
Current backlog
Process documentation Resource templates
Previous experience Facility reputation
Process templates Available process standards CAD/CAM/CAPP tools
Jigs Transportation capabilities
Figure 5: The structure of the facility ontology ORDER DATA
Engineering data
Logistic/commercial kid Explicit KID
Product geometry
Implicit KID
3D model and cross sections
Order quantity
2D drawing and cross sections
Due date
Recognized features
Estimated cost
Customer prioritization Contribution to reputation Previous experience with customer
Product dimensions
Price
Customer flexibility
Tolerances
Payment terms
Perceived risk Previous similar tasks
Surface finish
Perceived future prospects
Additional mfg. processes Quality control requirements Media type CAD type Packing instructions
Figure 6: The structure of the work order ontology x
resource requirements from existing process plans and resource management databases, in particular recognition or retrieval of relevant fixturing solutions and tools by geometrical part recognition. Methodologies such as Case-Based Reasoning (CBR) can be applied for locating similar past problems to be used as the basis of the current process plan. x Improved Collaboration between Part Designer, Process Planner and Manufacturer. Process planning and product design are concurrent processes that require collaboration among all parties, both internal and external to the organization [13]. Such collaboration requires improving technology and business processes by using a more systematic and structured approach. In particular, tools, procedures and business culture must be developed that will enable systematic feedback to the product designer from the manufacturer/CNC programmer/process planner about required changes in product features, tolerances, etc., in order to improve product manufacturability/profitability. Solution providers recognize the need to improve collaboration between design and manufacturing. To this end, they are currently working in two different directions: (a) providing well-defined compound features, which are recognized by the CAPP system along with their recommended process plan, to be used as building blocks for the new product plan; and (b) implementing standards
Part recognition. Feature recognition has been considered a key technology to link design and manufacturing information. Development of process planning and cost evaluation systems considerably depends on such recognition [11]. Broadening this concept involves part recognition: o Part recognition by classification methodologies A part classification method should be developed within the PLM environment that incorporates all KID related to process planning. A generic part template must be developed to enable part family formulation and rapid access to similar parts, to be used as reference process plans. o Part recognition by geometrical similarity Part recognition by geometrical similarity is implemented by using similarity algorithms combined with an adequate index structure and similarity query mechanism to retrieve parts from the database [12]. Although existing PLM solution providers have begun implementation of user-friendly "resource managing" components to aid the process planner in classification and retrieval of resources (tools, jigs, machines), additional development efforts are required to develop complementary part recognition capabilities. Such capabilities for part recognition must be interfaced with the retrieval mechanisms for process templates and relevant
-179-
infrastructure for the planning problem.
for transferring manufacturing KID between the different CAD/CAPP/CAM/PLM systems (i.e. tolerances, thermal treatment, coating etc.), for example by incorporating this additional KID into the STEP standard. x Developing improved interface between the PLM and ERP system. An improved interface is required between the PLM and ERP systems that enables consideration of logistic and business parameters such as availability and cost of resources along with the technological requirements of process planning available in the PLM system. This interface should support the industrial environment of today that relies on outsourcing as well as in-house manufacturing. The solution therefore must go beyond the boundaries of the organization to include the ERP and PLM systems of the company’s’ subcontractors. x Improve reusability of tacit and explicit knowledge. Research and development are required to solve two problems with the rule-base mechanisms implemented in existing CAD-CAPP solutions: (a) the rule base is not comprehensible, so that existing knowledge bases currently cannot be reused in additional organizations, and (b) defining new rules requires significant programming skills and is not intuitively applicable for the standard process planner. x Freeform Process Planning. Existing CAPP solutions support prismatic parts but not freeform shaping. Freeform modeling shortens modeling processes by simplifying geometry description, especially within the tool, die and mold-making industry. Although some research on classifying and reusing geometrical freeform features has been conducted [14][15], few studies apply geometrical knowledge to machining operations during process planning and code generation. Such support for machining of freeform features can include automatic tool selection and suggestions of sequencing and type of machining operations for an autonomously determined region of a part based on the feature description provided by the user. Such a capability will facilitate reuse of previous process plan segments by adapting them to the new part or feature.
7 ACKNOWLEDGEMENTS This research is a collaborative project between the Technion and Hannover University, funded by the Niedersachsen Foundation, Grant 2005847, and supported in part by the Schlesinger Minerva Laboratory for Life Cycle Engineering. 8 REFERENCES [1] Wang, H.P., Li, J.K., 1991, Computer Aided Process Planning, Elsevier Science, Netherlands. [2] Chang, T.C., 1999, Tools for Manufacturing Process Planning, Modern Manufacturing, CRC Press LLC. [3] Boer, C.R, Petitti, M., Lombardi, F., Simon J.P., 1990, A CAPP/CAM expert system for a high productivity, high flexibility CNC turning center, Annals of the CIRP, 39/1:481-483. [4] Koenig, D.T., 1990, Computer Integrated Manufacturing: Theory and Practice, Taylor & Francis, Washington. [5] Rho, H.M., Geelink, R., van’t Erve, A.H., Kals, H.J.J., 1992, An integrated cutting tool selection and operation sequencing method, Annals of the CIRP, 41/1: 517-520. [6] Gologlu, C., 2004, A constraint-based operation sequencing for a knowledge-based process planning, Journal of intelligent manufacturing, 15:463-470 [7] Tichkiewitch, S., Brissaud, D., 2000, Co-ordination between product and process definitions in a concurrent engineering environment, Annals of the CIRP, 49/1:75-78. [8] Dori, D., Shpitalni, M., 2005, Mapping knowledge about Product Life Cycle Engineering for Ontology construction via Object-Process Methodology (OPM), Annals of the CIRP, 54/1:117-120. [9] Gomez-Perez, A., 1998, Knowledge Sharing and ReUse. In J. Liebowitz, (Ed.), The Handbook of Applied Expert Systems, Boca Raton, pp. 10:1-10:36). [10] Huhns, M. N. & Singh, M. P., 1997, Ontologies for Agents, IEEE-Internet Computing 1/6:81-83. [11] Woo, W., Wang, E., Kim, Y.S., Rho, H.M., 2005, A hybrid feature recognizer for machining process planning systems, Annals of the CIRP, 54/1:397. [12] Berchtold, S. and Kriegel, H., 1997, S3: Similarity search in CAD database systems, SIGMO. [13] Maropoulos, P.G., McKay, K.R., Bramall, D.G., 2002, Resource-aware aggregate planning for the distributed manufacturing enterprise, Annals of the CIRP, 51/1:363. [14] van den Berg, E., Bronsvoort W. F. and Vergeest J. S. M., 2002, Freeform feature modelling: concepts and prospects, Computers in Industry, 49/2:217-233. [15] Fontana, M., Giannini, F., and Meirana M., 1999, A freeform feature taxonomy, In: Brunet, P., Scopigno, R. (Eds.), Proceedings of Eurographics, Computer Graphics Forum, 18/3:107-118.
6 CONCLUSIONS The results of the literature review, market survey and industry survey emphasize the complexity of the process planning problem as well as the lack of mature holistic solutions for industry that include all relevant business, logistic and technological considerations. The findings of this comprehensive review of the CADCAPP market and of the available solutions provide the basis for developing an ontology for the process planner environment and a complementary decision process model. These models (a) provide the basis for knowledge management support and business process modeling, and (b) enable an improved understanding of the work of process planners by pinpointing opportunities for enhancing both the efficiency and the consistency of the process planner's decision making. The models are based on a holistic approach by zooming out on the CAPP environment for a broader view of the planning problem and by incorporating business considerations as part of the process planning solution. This comprehensive overview has highlighted the need for additional complementary capabilities required to provide a holistic solution to the process planning problem. Furthermore, it is evident that the most appropriate environment for developing the required support tools is PLM-ERP. These tools have reached maturity and are well developed and implemented in industry, providing an
-180-
