View more...


  • Rating

  • Date

    December 1969
  • Size

  • Views

  • Categories

Preview only show first 6 pages with water mark for full document please download


COMPUTER-AIDED PROCESS PLANNING (CAPP) SECOND EDITION SEPTEMBER 1991 ARCHITECTURE TECHNOLOGY CORPORATION SPECIALISTS IN COMPUTER ARCHITECTURE P.O. BOX 24344 · MINNEAPOLIS. MINNESOTA 55424 · (612) 935-2035 ELSEVIER ADVANCED TECHNOLOGY DISTRIBUTED OUTSIDE THE USA/CANADA BY: ELSEVIER ADVANCED TECHNOLOGY MAYFIELD HOUSE 256 BANBURY ROAD OXFORD OX2 7DH UNITED KINGDOM ® Copyright 1991 Architecture Technology Corporation. All rights reserved. No part of this publication may be reproduced, photocopied, stored on a retrieval system, or transmitted without the express prior written consent of the publisher. COMPUTER-AIDED PROCESS PLANNING (CAPP) SECOND EDITION SEPTEMBER 1991 ARCHITECTURE TECHNOLOGY CORPORATION SPECIALISTS IN COMPUTER ARCHITECTURE P.O. BOX 24344 · MINNEAPOLIS. MINNESOTA 55424 · (612) 935-2035 ELSEVIER ADVANCED TECHNOLOGY DISTRIBUTED OUTSIDE THE USA/CANADA BY: ELSEVIER ADVANCED TECHNOLOGY MAYFIELD HOUSE 256 BANBURY ROAD OXFORD OX2 7DH UNITED KINGDOM ® Copyright 1991 Architecture Technology Corporation. All rights reserved. No part of this publication may be reproduced, photocopied, stored on a retrieval system, or transmitted without the express prior written consent of the publisher. DISCLAIMER Architecture Technology Corporation makes no representations or warranties with respect to the contents hereof and specifically disclaims any implied warranties of merchantability of fitness for any particular purpose. Further, reasonable care has been taken to ensure the accuracy of this report, but errors and omissions could have occurred. Architecture Technology assumes no responsibility for any incidental or consequential damages caused thereby. Further, Architecture Technology Corporation reserves the right to revise this guide and to make changes from time to time in the content thereof without obligation to notify any person or organization of such revision or changes. This disclaimer applies to all parts of this document. FOREWORD Process planning involves creating detailed plans of the manufacturing steps and equipment necessary to produce a finished part. Workpiece requirements call for detailed analyses and accurate descriptions prior to the actual manufacturing process. A large assortment of machines and operations, as well as many different workers with a variety of skills, may be involved in the production of a specific part. The computer is a vital part of the process planning function, which includes two different approaches. O n e is called t h e variant (similar part) method of process planning and the other is generative (expert systembased). Both will produce similar process plans. Most computer applications, however, are of the variant type, because the software is easier to develop and new process plans are based o n previous ones. T h e variant method had its beginnings with the G T concept, along with parts classification and coding systems. G T is a manufacturing philosophy based on the idea that similarities occur in the design and manufacture of component parts. These parts can be classified into groups, or families, if the basic configurations and attributes are identified. A reduction in expenses can b e achieved through the structured classification and grouping of parts into families based upon engineering design and manufacturing similarities. Using the variant method, C A P P groups families of parts by a structured classification and coding plan. All previously processed parts are coded using this method. T h e parts are then divided into part families, such as cylindrical or prismatic, based on general configuration. A standard order of operations or sequences is stored on the computer for each part family. W h e n a new part is ready for planning, the classification or G T code for the new part is used to compare and retrieve the standard process plan for that part family. Editing capabilities further enable the process planner to alter the standard order of operations for final refinement. T h e completed process plan is then stored on the computer database by part number. In generative process planning, the parts are again broken into part families, and a detailed analysis is made for each part family to determine individual part operations. This type of system develops the actual operation sequence based o n the part geometry, usage requirements, material size and configuration, and available equipment. T h e generative approach creates process planning logic for the part family groupings. The logic is then stored internally as a decision model. As new workpieces require process planning, analyses must be conducted to determine and compare the features incorporated in the decision model with those on the actual part. T h e family part decision model is then retrieved, and a routing sheet is generated by processing the decision model with t h e new workpiece attributes. A generative system must b e driven by much m o r e elaborate and powerful software than a variant system. Development and optimization work continues on the variant and generative approaches to process planning. Both systems, however, build the needed decision-making logic and planning ability into the computer rather than rely o n a decreasing experience level in the process planning work force. Computer-Aided Process P l a n n i n g (CAPP) Figure List Figure 1: Production of a Simple Part 2 Figure 2: Operations Overview 5 5 Figure 3: Part Family Matrix File 5 Figure 4: Process Plan 6 Figure 5: Standard Plan Structure 6 Figure 6: Basic System M e n u and Sub-program M e n u 8 Figure 7: Outline of an In-house Classification and Coding System 10 Figure 8: Structure of Variant Module 12 Figure 9: Structure of the Semi-generative Module 14 Figure 10: System Structure of t h e Proposed G T Based C A P P System 15 Figure 11: Architectural Overview 31 Figure 12: Process Planning External Architecture 36 Figure 13: ICAPPS System 38 Figure 14: Extracting Generated Surfaces 40 Figure 15: Flow Chart of Process Planning 42 Figure 16: Functional View of C M P P 46 Figure 17: Major C M P P C o m p o n e n t s 46 Figure 18: Data Base System 47 Figure 19: Part Input System 48 Figure 20: Process Planning System 48 Figure 2 1 : Typical Steps in the Assembly of a P/CB 51 Figure 22: A C I M Architecture for the Production of Electronic Systems 54 Figure 23: T h e CAPP-Postprocessor Relationship 56 Figure 24: Sample Product R o u t e List 60 Figure 25: Master R o u t e Flow 60 Figure 26: M R P View of the Master R o u t e 61 Figure 27: W F C View of the Master R o u t e 61 Figure 28: F D C View of the R o u t e Master Flow 62 Figure 29: Data Relationship Between the Process Master and the R o u t e View 64 Figure 30: Physical Operations, Typical Manufacturing Facility 68 Figure 3 1 : Traditional Building Control Focus 69 Figure 32: Integrated Manufacturing Facility 71 ν Computer-Aided Process Planning (CAPP) 1. Introduction 1.1 System Organization Given the engineering design of an item which has to be manufactured, process planning is the act of generating an ordered sequence of the manufacturing operations necessary to produce that part within the available manufacturing facility. For maximum efficiency, the developed process plan should produce the part at the lowest cost consistent with acceptable quality standards and using established procedures. In achieving these aims, it is essential that process plans for parts which are basically similar should be standardized. Traditionally, process planning has been regarded as a manual operation, usually carried o u t by a manufacturing engineer (process planner, process engineer, etc.) who is often a qualified and experienced machinist or tool maker. T h e success rate achieved by an individual planner is largely dependent upon individual skill and aptitude for the planning task, knowledge of manufacturing processes, equipment, materials and methods in general, and those available in a production facility in particular. T h e rapid evolution of increasingly sophisticated production methods, the ever increasing trend towards the use of flexible manufacturing systems, and the need for improved manufacturing efficiency has necessitated a requirement for improved process planning. T h e use of computers to achieve this is now well established, and new computerized methods are continually being sought. 1.1.1 Traditional Process P l a n n i n g Traditional process planning methods usually allocate planning jobs to the next available planner. Each task is therefore treated as a new process plan with usually limited attention being given to process plans for similar parts already in existence. As a result, each process plan will reflect primarily the skills and experience of the individual planner concerned; hence, it is extremely difficult to maintain consistency and discipline in the planning process. Process planning also requires the use of many complex disciplines including sequencing, machine selection, time and motion study, programming, and material flow to n a m e some of the most common. All of these skills are needed in an increasingly complex manufacturing environment and are making it even m o r e difficult to produce efficient process plans. Several studies have been conducted to investigate the efficiency of manual process planning methods. O n e , conducted by Cincinnati Milacron, reports the results of a study into the process planning for the production of a family of spur gears. F o r a sample of 425 relatively simple spur gears, there were 377 different process plans (operation sequence and machine groups), 54 different types of machines required, and 15 different materials utilized. T h e plans gave 252 unique combinations of machine tools for successive operations. T h e inefficiency arising from this situation must be obvious to anybody with manufacturing experience. A perfect example of the problem concerns the production of the simple part shown in Figure 1, together with the processes proposed by four different process planners. T h e four different methods of producing the 18 m m hole are readily apparent and obviously reflect the past experience of the individual planners concerned. Attempts to standardize and rationalize the process planning task revolve around the maintenance of card index systems which relate to previous plans. Such systems quickly become outdated or are, at best, very cumbersome to use. T h e use of a computer for storing such data was an obvious first step and the first 1 Computer-Aided Process Planning (CAPP) CAPP systems were basically just computerized card index systems. 1 2 3 4 5 6 7 8 9 10 Turn 30 Face Cut Drill 1 5 ° Countersink Turn 30 Drill 1 8 ° Drill 17.5° Thread Cut Turn 30 Drill 16° Bore 18° Thread Cut Bore 18° Thread Cut Face Cut Face Cut CutOff CutOff Second Face Cut Second Face Cut CutOff Second Face Cut Face Cut Turn 30 Countersink Drill 16° Bore 18° Chamfer CutOff Second Face CutChamfer Thread Cut Figure 1: Production of a Simple Part 1.1.2 Classification and Coding T h e number of different part numbers which flowed through most manufacturing facilities were such that they made the initial card index approach only a very marginal improvement on the manual systems which they superseded. T h e development of standardized process plans required the identification of part families where such families can be distinguished by the sequence of manufacturing operations required. Several part families may share common machine tool routings but a unique production part family has the same, or nearly t h e same, list of operations. T o define part families, it is necessary to have some form of parts classification and coding system. Classification and coding systems are often considered to be a part of the G r o u p Technology (GT) production concept. However, classification and coding systems have advantages quite separate and distinct from their use with G T . F o r example, a well designed classification and coding system can provide significant advantages in the basic engineering design process by reducing the number of drawings and different part numbers which commonly occur over a period of time through the duplication of part designs which perform the same or similar function. 2 Computer-Aided Process Planning (CAPP) When used with G T , the benefits of classification and coding can be summarized thusly: • T h e formation of part families and machine groups • Quick retrieval of designs, drawings and production plans • Design rationalization and reduction of design costs • Secure reliable workpiece statistics • Accurate estimation of machine tool requirements, rationalized machine loading, and optimized capital expenditure • Rationalization of tooling setups and the reduction of set-up times and overall production times • Rationalization and improvement of tool design and the reduction of tool design time and cost as well as tool fabrication time and cost • Rationalization of production planning procedures and scheduling • Accurate cost accounting and cost estimating • Better utilization of machine tools, work holding devices, and manpower • Improvement of Numerical Control (N/C) programming and the m o r e effective use of N/C machines. Many of these benefits can be achieved without using them in conjunction with the G T cell concept. T h e r e are a large number of proprietary classification and coding systems available ranging from the "look and see" approach to large numerically coded systems. T h e most well known systems include among others, the O P I T Z system developed in Germany, the MICLASS system developed in Holland, and the B R I S C H B I R N system developed in Sweden. In practice, most proprietary systems have been found to be inadequate for particular company applications and most successful classification and coding systems have been developed by adapting a basic proprietary system to suit the needs of the particular facility concerned. A basic problem in using any classification and coding system is the h u m a n problem of applying it to particular parts; i.e., different people using the same coding system will arrive at different codes for the same part. Most applications nowadays therefore, use computer assistance in order to obtain standardized coding. O n e such aid is the D C L A S S system developed at Brigham Y o u n g University, which can be utilized in conjunction with any classification and coding system to obtain consistent results. This system is an interactive system which enables a planner seated at a C R T to obtain a classification code by responding to a series of system produced queries. T h e basic starting point of a C A P P system is the implementation of a good classification and coding system. However, work put into classification and coding has many other useful applications in the manufacturing and design environments, such that implementation costs can b e easily recouped from efficiency improvements. 3 Computer-Aided Process Planning (CAPP) 1.1.3 CAM-I CAPP O n e of the earliest C A P P systems was developed under contract to C o m p u t e r Aided ManufacturingInternational, Inc. (CAM-I) in the USA. CAM-I is a not-for-profit organization, formed in the U.S. in the early 1970s by a number of major American manufacturing companies, to provide leadership in the development of computer aids to manufacturing industry. CAM-I is now an internationally recognized body with member companies in the U.S., E u r o p e , and the Pacific Rim. T h e C A P P program has since been used and adapted by CAM-I m e m b e r companies for their own use and thus forms the basis for many of the well-known company sponsored systems. Two versions of C A P P are now available for 16 bit and 32 bit computers, respectively. Both versions are identical in operation and effect. C A P P is based on the use of a classification and coding system to group parts into part families. Each part family has a common, or nearly common, process plan, which is stored in the computer as a standard plan for that part family. Each standard plan is a sequential set of instructions that includes general processing requirements, jig and tool data, machine data, and detailed operating instructions. In the CAPP system, these standard plan details are called work elements and work element parameters. T h e system allows for these items to be referred to in any format which will suit the normal facility terminology. In addition, the C A P P system permits the use of any classification and coding system up to a maximum of 36 digits. C A P P is an interactive system which permits a planner to sit at a terminal and locate process plans, modify existing plans, or carry out any other normal process planning operations without any need to know computer programming. C A P P is written in standard ANSI F O R T R A N - 7 7 ; however, two or three of the sub-routines involved are machine or terminal dependent and may require modification to suit particular installations. In general, however, the use of F O R T R A N - 7 7 does mean that user modifications can easily be built into the basic C A P P program. CAM-I only aims to produce prototype software and a particular implementation of C A P P requires some debugging effort on the part of the user. 1.1.4 The CAPP System Basically C A P P requires a data structure of six main files: part family matrix file, standard sequence file, operation code table, operation plan file, part family code table, operation plan file, part family set-up file and a process plan store file. T h e flow diagram in Figure 2 gives an overview of the sequence of operations. System operation is menu-driven. T h e particular operation to be carried out can be called by invoking the particular m e n u concerned and responding with coded commands to carry out the operation required. Process plan data, process plan header data, part numbering, and other data are all inserted using normal facility nomenclature and formats. Although this data is specified during system setup, it can easily be altered later as facility requirements dictate. T h e part family matrix file contains the part family codes in the form of a matrix as shown in Figure 3. T h e C A P P system allows for a 35 digit coding format with up to 16 permissible characters for each digit. T h e system also allows for part attributes to use up to four digits for coding and for multiple characters to b e used in each digit. T h e part family matrix file, therefore, allows great flexibility for the user to select a classification and coding system which gives him the greatest possible benefits. T h e C A P P system requires the user to develop and store a standard process plan for each family. Each process plan contains header data which can be initialized by the user for the types of information needed on the printed process plan and for any language desired. Approximately 100 items are available as shown in Figure 4. A user specified header menu is built up when the system is initialized and appropriate data is inserted into the menu when a new process plan is generated. 4 Computer-Aided Process P l a n n i n g (CAPP) Standard Sequence File Part Family Matrix Part Family Search Header Input Data Standard Sequence Retrieve/Edit Part *\ Classification V Code J Process Plan Process Plan Formatter ( Application Programs Operation Ran File Work Element Processor Operation Plan Retrieve/Edit Process Plan Store F i g u r e 2: O p e r a t i o n s Overview Code Char. Γ 1 Set Classification Code Length I 2 3 4 5 7 6 0 1 8 9 11 10 X 2 X X 3 X 4 X X X X X X X X X X X X X X 7 X X 8 X 9 X X A X X Β X 1 2 , 3 1 4 5 . 6 „7 I 8 16 X X X 15 X X 6 5 14 13 X X X 12 X X X 1 I 9 I 10 , 11 , I Part Attributes Figure 3: Part Family Matrix File 5 Computer-Aided Process Planning (CAPP) Header Data Elements Identification Reference Part No. Part Name Part Family No. Plan Type Issue No. Etc. Part Type Source Standard Cost Lead Times Project Etc. Usage Material ι I Next Assy. Effectivity Quantity Units/Prod. Etc. Material Code Material Size Material Spec Weight Supplier Etc. I 1 Planning Change Miscellaneous Change Letter Change Planner Change Date Etc. Work Standards Standard Times Q.A. Class Order No. Mfg. Source Figure 4: Process Plan Each process plan is built by invoking a set of operation codes (op codes), each of which identifies a unique operation. T h e o p codes are set up during initialization in the op code file and operation descriptions can be in the standard facility format. Each process is numbered and cross referenced to the part family number to which it refers. The standard plan structure is shown in Figure 5. Operation Plan (OP Plan/OP Codes) • M / C Number Dept Work Center IM/C Reference 1 Location M/C Code Set Up Data Time Stds Data Work Elements or Work Instructions Work Element Parameters Figure 5: Standard Plan Structure 6 Computer-Aided Process Planning (CAPP) T h e basic C A P P system requires the user to insert his own time standards, but several companies have found it a fairly straight forward operation to link a computerized standard time program to the system. T h e C A P P system also allows for unique part process plans to be stored separately for parts which do not reasonably fit into any commonly used part family. T h e operation of the system requires the user to feed in the classification code for the part concerned. T h e system searches the part family matrix file and attempts to fit the part into a part family for all the digits in t h e code. If unsuccessful in matching all digits, the system will find the part family which matches the earlier digits in the code and will tell the user how many digits can be matched. Once a part family has been found, the process plan for that family can be called up on the screen. T h e process plan can be used as is or can b e modified where necessary for a particular part, and header data can b e altered. If a standard plan is used as is, the part number concerned is added to the list of the part numbers which form a part family. If a standard plan has to b e modified, it can be stored as a separate part plan, printed out and then deleted, or stored as a revised process plan for the appropriate part family. T h e basic system menu and o n e of the sub-program menus are shown in Figure 6, as examples of the system in operation. CAPP is comparatively simple to use, once it is set up, and requires no special computer skills from the process planner. T h e system has been operated successfully by many companies in its basic form. However, it was first used in 1976 and some of the major companies who started with the system have made modifications to enhance its usefulness and to make it m o r e compatible with their own operations. 1.1.5 CAPP Developments CAM-I C A P P is known as a variant process planning system; i.e., it operates by taking an existing process plan and changing it to suit the new part. T h e r e are, however, obvious advantages in being able to create a brand-new process plan for each part as required. Systems which o p e r a t e in this way are known as generative process planning systems and development work on such systems has been the subject of much research over the last decade or so. A generative process planning system must be capable of applying a set of rules and massive amounts of data to the construction of a process plan in much the same way as a human planner reasons when he constructs such a plan. Hence, the development of a generative process planning computer system requires the use of computer systems termed artificial intelligence (AI) systems. M o r e specifically, current research is based a r o u n d a subset of AI technology called knowledge-based expert systems. Expert systems consist of a factual database and a set of inferential or heuristic (rule of thumb) rules which use the factual database to solve problems. Investigations into heuristic methods of process planning define process as a problem which is only defined by its target (finished part), starting point (raw material), and a set of constraints (available processes and equipment and technological relationships). In addition, it is probably necessary to introduce time/cost criteria to choose an optimal solution. T h e conclusions from such studies tend to agree that the use of expert systems or AI are probably the most promising way to tackle the complex relationships involved in the planning process. 1.1.6 CAPP a n d CAD/CAM C A D / C A M systems have been developed to the point where a design engineer and a manufacturing engineer seated at graphics terminals connected to a common computer database can produce parts from computer designs without benefit of paperwork at any point. Thus, engineering drawings, process planning documents, and other manufacturing paperwork are all stored in the computer. If the computer is connected directly to the appropriate C o m p u t e r Numerical Control (CNC) machines as in a flexible 7 Computer-Aided Process Planning (CAPP) CAPP SYSTEM MAIN MENU FS FP DP HD OS Part Family Search Format Plan Delete Plan Create New Plan Retrieve Header Retrieve OP Code Sequence Retrieve OP Plan Process Plan Review Logoff OP PP LO FS FP DP HD HD/ (Part No, Plan Type, Status) OS/ (Part No, Plan Type, Status) OP/ (Part No, Plan Type, Status) PP/ (Part No, Plan Type, Status) LO Space for inserting code for menu item required, e.g. FS Part Family Search MS SP DS Matrix Search Retrieve Standard Plan Display Plan No's Search Attribute Continue Search Return to Main Menu MM Logoff SA NS/ SP/ DS/ SA/ CS MM LO CS MM LO (Classification Code) (Part Family No.) (Part Family No.) (Attribute No., Value) Space for inserting code for menu item required, e.g. DS/26 Figure 6: Basic System Menu and Sub-program Menu manufacturing system, the process plan can be followed without the need for human intervention. Current C A P P systems make this possible, the newer C A P P systems will obviously enable it to be done even more efficiently. L 2 C A P P a n d GT Current industrial applications indicate that the combining approach of variant and generative CAPP methods is suitable for practical usage. A variant model is found most applicable for those products which the company produces regularly, and a generative model is useful for the parts for which engineering designs and manufacturing requirements vary. Since the current state of the art of the generative method requires further development, it is m o r e practical to develop a semi-generative method which combines the specific features of variant and generative approaches. 8 Computer-Aided Process Planning (CAPP) 1.2.1 Group Technology As noted above, G T is a manufacturing philosophy which identifies and exploits the underlying sameness of parts and manufacturing processes. In conventional piece-part manufacturing, each part is treated uniquely from design to manufacture. However, by grouping similar parts into part families based on design or process information, it is possible to reduce manufacturing costs. G T classification and coding and part family concepts have been used as the key elements for most variant C A P P systems, which obtain the process plan by retrieving and modifying the part family-related standard process plan. T h e application of G T classification and coding and part family concepts are used, not only for classifying the parts, but also as the foundation for constructing logic and rules for generating process plans. T h e system provides a highly automated C A P P process which generates the applicable process plans. T h e system is relatively easier to develop and program, and needs less computation time and storage space than most of currently available generative C A P P systems. The implementation of an in-house G T oriented C A P P system is a multi-phased procedure and must be introduced in steps. T h e proposed method makes the maximum usages of the G T concept. T h e entire process includes developing an in-house G T classification and coding system, forming part families, building an effective manufacturing information retrieval system, and developing the in-house semigenerative C A P P system. 1.2.2 GT Classification The first and the most important step for implementation of G T is developing a suitable classification and coding system. Many generic classification and coding systems and schemes have been developed in various forms. They are readily available but not directly applicable in their original form, without being properly modified and tailor-fitted to meet the specific needs of a particular application. A tailor-fitted process contains three steps. First, a random sample of parts for a fixed production period is collected. T h e sample is used as representative of the whole population of parts that is produced. Second, a suitable publicly available system is used for sample coding of selected sample parts. Third, the statistical data of the sample part population is evaluated using the coding results. T h e trial system is modified, based on the sample coded results and also to meet the specific needs of the application. Following these exercises, an in-house classification and coding system with 13 digits is developed. T h e system contains such information as part type, general shape, material, operation and surface finishing requirements, and heat-treatment. Figure 7 shows the outline of the classification and coding system. A n effective, user-friendly, menu-driven computer program can be developed on a personal computer. 1.2.3 P a r t Families Since most of the benefits obtained from applying G T are based on the underlying sameness of parts and their manufacturing processes, the method used to form the proper part family is a very critical element. T h e methods of forming G T part families have been improved using appropriate mathematical programming techniques and suitable computer programs. T h e simple ocular observation method and the manual production flow analysis method have been replaced by computerized methods using suitable algorithms. Recently, many advanced methods including cluster analysis and pattern recognition techniques have been proposed to form the G T part-families. Most of the existing cluster analysis algorithms for forming G T part families use production flow data as the input. T h e part families formed by these methods are operation-sequence dependent and do not reflect the design aspect. However, from the computer integrated manufacturing implementation point of 9 Computer-Aided Process Planning (CAPP) DIGITS DESCRIPTIONS 1 & 2 Product Type 3 & 4 Material 5 Size Parameter 1 6 Size Parameter 2 7 Size Parameter 3 8 Shipping Weight 9 General Shape 10 Operation and Accuracy (Outside) 11 Operation and Accuracy (Inside) 12 Operation and Accuracy (Ends) 13 Heat Treatment Figure 7: Outline of an In-house Classification and Coding System view for standardization of design, process planning, and other implementations of the G T concepts, part families should reflect both design and manufacturing characteristics. Since a well-designed G T classification and coding system should reflect correctly both manufacturing and design characteristics, part families reflecting both design and production processes can be obtained. The actual industrial applications using G T codes and multi-objective cluster analysis provides a method that can form the G T part families and retrieve similar parts and related part families properly. 1.2.4 Information Retrieval System A n important task in large manufacturing concerns is identifying and analyzing the numbers, types, and characteristics of parts being produced. All manufacturing, planning, and control functions are dependent on the understanding of the part population and its manufacturing requirements. With an interactive computer program, this information retrieval system easily finds the required production features, the material types, and operation requirements based on the specific priorities required. It is always the desire of management that, at any given time, the part population of the factory is precisely known in terms of part specifications, production requirements, and part features, so that an information base exists for use in design rationalization, manufacturing planning, production control, marketing, and purchasing decisions. U n d e r a G T environment, this includes the ability to retrieve and group part families for use in various manufacturing functions. Without a well-designed classification and coding system, retrieval of required manufacturing information is difficult. 10 Computer-Aided Process P l a n n i n g (CAPP) 1.2.5 Variant CAPP Module T h e first step in building a relatively proficient C A P P system is to use the variant approach. As in most variant C A P P systems, a standard process plan can be automatically retrieved for a new order, using its G T classification code and the part family number. T h e built-in standard process plans are entered for each part family, and the final process plan for a particular part can b e obtained by editing the standard process plan. T h e unique feature of this particular variant C A P P module is that it uses a multi-objective clustering algorithm as t h e part family searching mechanism instead of using the part family matrix matching or G T classification codes matching method like most GT-based variant C A P P systems. T h e advantage of using this method is that if the new part cannot be grouped into an existing part family, a number of similar parts and their part families can be reached according to different searching requirements. In this way, the user can develop the new standard process plan based on some other similar parts and their part family related standard process plan. The structure of the variant module is shown in Figure 8. In order to obtain the process plan for a new part, the part must first be coded. Using the interactive coding program, a G T code can be assigned to the new part. After the part is coded, the user needs to input the priority for the part family searching. For instance, if a new part has the particular requirement on operations, the user can put the digits of operation requirements as higher priority digits. If, however, a part is made of special material and needs special heat treatment, then the user can specify the digits which describe the material and heat treatment as m o r e important features. A similar part and part family can be retrieved and the standard process plans of these retrieved part families can be obtained from the standard process plan data file. T h e final process plan for the new part is obtained by modifying the retrieved standard process plan. A practical industrial application shows that the multi-objective clustering algorithm-based variant C A P P system can classify a part into a part family or find the proper part families m o r e accurately for the new part than other variant-type C A P P systems. Using the relational database management technique, the computation time for searching similar parts and part families can be reduced tremendously. 1.2.6 Semi-Generative CAPP System Using the variant process plan, the final process plan for a particular part can be obtained by editing the standard plan. Nevertheless, it requires the process planner to make decisions at the editing stage. In order to improve the performance of the C A P P system further, an application-oriented, rule-based semigenerative C A P P system can be developed to enhance the proposed variant system. Variant and Generative C A P P Module Although a variant process planning system can help the process planner to make the process plan efficiently, the person using the variant C A P P system must be an experienced process planner, so as to be competent in making decisions during editing the standard plans. Therefore, the variant process plan system is usually suitable for preparing process plans for those parts which have relatively stable process routes. When there are many differences between the process plans of the similar parts, it a generative process plan system. Instead of using existing standard process plans, plan system utilizes built-in logic to select and sequence necessary operations. All process planning procedure are m a d e by the system, and the o p e r a t o r need not be planner. 11 is m o r e suitable to use a generative process of the decisions in the an experienced process Computer-Aided Process Planning (CAPP) Coding Part Family Searching Semi-Generative Main Logic Tree Editing Process Plan Figure 8: Structure of Variant Module Developing a generative process planning system, however, is really an enormous task. A truly generative CAPP system is hard to construct. E n o r m o u s activities are involved in developing such things as general process planning logic sets and constructing a suitable database structure. Since most so-called generative CAPP systems require that the user key-in part features, generated process plans are therefore dependent on how the part features are interpreted. In addition, a completely generative C A P P system makes every decision in the process planning procedure from scratch. This requires tremendous computer memory and computation time. Semi-generative CAPP Module T h e activity of process planning is a company-oriented task. This means that even a manufacturing process which has been successfully used in o n e company might not be suitable for producing the same part in another company, because of different manufacturing conditions, machine tools, skill of the workers, corporate culture, and unions. Therefore, it is hard to build a generic generative CAPP system which can be adapted for different manufacturing systems. However, when applying specific manufacturing applications, the task of developing the generative logic for process planning is greatly simplified. Based on the above considerations, a semi-generative CAPP system holds promise for many applications. T h e reason that the system is termed "semi-generative" is that both a variant retrieving method and a generative decision logic approach are used. Some decisions are generated by the built-in, process 12 Computer-Aided Process Planning (CAPP) planning generating logic, while some other process plan decisions are retrieved in the form of tables and decision sets. T h e G T part family concept is applied as the key element for constructing the rule-based C A P P expert system. Instead of developing the process planning logic based on part features like most other generative C A P P systems did, the process planning logic or decision rules are divided into two types; part family related decisions and operation related decisions. T h e part family related logic is the set of logic using the part family as the driven machine. Since different parts require a different process plan which is generated by a different logic set, similar parts can share the same process plan logic set. Therefore, the logic used for generating the process plan can be simplified by using the part family concept. In the part family oriented logic set, the decisions are made step by step according to the part family number. In each node of the logic tree, the operation or operations, division number and machine number will be defined. T h e detail operation instructions can be further retrieved by the operation number. Detailed operation instructions, like operation sequences, machine selections, machining parameters, and time standards can be generated with very few computations. T h e other type of decisions are operation related. They are the rule sets which are related to operations rather than part families. F o r instance, some parts have auxiliary holes which should be drilled or drilled and tapped. However, there exist various combinations on radial or axial holes, the location of the holes, and the diameters of the holes. This is a set of decisions which is independent from the part families; e.g., a rule set called drilling. Depending on the size, location, and type of hole, parameters such as operation, operation sequences, and machine station can be obtained. The method of developing the generative process planning logic using G T part family concepts simplifies the procedure of constructing the process planning logic. Because of this simple and clear logic tree, the computation time is reduced. System Structure T h e system structure of the semi-generative C A P P module is shown in Figure 9. T h e part to be planned is coded first, using the in-house classification and coding system. After coding, the part will be automatically classified into o n e of the existing part families, if o n e is found. T h e main process plan logic calls the part family oriented rule sets and operation oriented rule sets, the machining and operation data bank, and time standard data bank and generates the process plan. Figure 10 shows the flow diagram of the entire system. 13 Computer-Aided Process Planning (CAPP) Figure 9: Structure of the Semi-generative Module 14 Computer-Aided Process Planning (CAPP) ^ Start MOCA PF Searching CED Figure 10: System Structure of the Proposed G T Based CAPP System 15 Computer-Aided Process Planning (CAPP) 2. System Implementation 2.1 Developing t h e Process P l a n T h e basic information which links quote/order processing data with process planning data includes the part number, part description, and stock number. The part master identifies all of the resources which the company builds or uses in serving its customers. With data from quote/order processing, delivery requirements are established by date and quantity. With this information in place, automating the process planning function can proceed. 2.1.1 Basic Considerations T h e first step in automating the process planning function is selecting a few jobs and defining procedures for developing formal process plans and defined a series of operations. A less formal plan consists of simply a "build to print" instruction. Most major manufacturers require a format which spells out all of the manufacturing requirements, including material, operations, tools, fixtures, gauges, and inspection data. It also provides details about setup and fixturing, speeds and feeds, workpiece orientation, dimensions and tolerances, and step-by-step process descriptions, including Q C . Process engineers develop these plans on sheets summarizing the manufacturing process, which usually accompanies the workpiece as it travels through the shop. These summary sheets are called travelers or routers and are a necessary part of process planning. 2.1.2 D a t a b a s e Contents Every workpiece, tool and fixture must have a manufacturing plan or set of plans. These plans must be stored in an easily accessible manner. T h e r e can be more than o n e plan for manufacturing a particular item, because different machines may be used or a special rework plan may be needed. Each computer record should identify tools, fixtures, N/C programs, machines, setup times, piece times (labor and machine), and detailed operation instructions. Vendor operations and inspections must also be identified. Materials, tools, fixtures, and other elements required for an operation should b e specified and tracked to b e sure that they are in inventory, scheduled to b e made, or ordered in time to meet production requirements. Routers, travelers, and other work control documents must accompany single workpieces or sets of workpieces in production. T h e paperwork must list the operations in summary form in a manner consistent with the plan. Workpieces for a particular job must be grouped in lots. Lot size can vary from o n e to the total order quantity. Information defined in the process planning module serves as a basis for estimates and quotes. Labor vouchers are validated against the plan. Based on what the labor voucher shows, the production control department can estimate how the j o b is progressing and where trouble spots or bottlenecks might be. T h e plan helps production schedulers estimate the load on each workstation and foresee contention. Standard times help job costing by establishing a baseline against which actual costs can be measured. Work performance can be measured on an equivalent unit basis, a frequent requirement for large contracts with progress payments. 17 Computer-Aided Process Planning (CAPP) 2.1.3 Start-Up Methods T h e r e are several ways to begin orienting process planning toward computer augmentation: • Produce a highly detailed process plan for every workpiece. These plans are prepared manually, then copied and updated as needed. To start, select several jobs and use a process planning method to prepare the paperwork for the master plan and the travelers. This approach has little impact on production and allows the shop to experiment with the software. It also facilitates training. • Use the existing process planning system. Develop data conversion programs to move the data from the existing system to the new system. Data is moved all at once and everyone uses the new system immediately. Although this immediate conversion can put pressure on users, they have the advantage of moving from o n e automated system to another. This prior experience eases the transition from the manual system to the automated one. • Develop a prototype process planning module, designed to gradually introduce the idea of automated process planning to the engineering department and to establish a standard method of defining plans and releasing work to the shop floor. This prototype is simpler than the finished system, but all of the basic information can be captured and transferred to the new system. A n internal data processing department provides resources to support this kind of independent effort. • Introduce automated process planning to the shop gradually. Keep a skeleton plan on the system, but develop detailed process sheets with word processing software on PCs. Thus, the most labor intensive aspects are handled off the system. Copies of the master plan, with all of the operation sheets, are kept in a master file and backup copies of the P C files are kept for security. 2.1.4 Educating t h e User Process plans are the heart of any good system, but creating good plans is very difficult. Sharing information is the major challenge. For example, engineering departments do not always have the best information. Engineers must stay in touch with production control and shop floor foremen. Separate departments have a tendency to isolate themselves and to make decisions based on incomplete information. Access to a central database helps solve this problem. Process planning requires identifying the best technical as well as the most economically feasible means of producing a part. Sequencing operations, selecting the right machine, and estimating the duration of each operation are among the most difficult pieces of information to collect or define. Rush or non-standard jobs are particularly difficult to handle. Establishing standard times for operations sparks the greatest controversy. Engineers feel that, without actual timings or shop floor data, they are guessing. Operators feel threatened by the standards, which can create the impression of poor performance if set unrealistically high. Managers worry that some employees will slack off if standards are too low. Most of these problems can be solved by education. As planning is implemented and work release controlled, shops begin to collect shop floor data against the plans. T h e original estimates are replaced with standards based on actual data collected. This relieves the burden on the engineering staff for collecting and analyzing data. Publishing standard times and explaining their use eases the concerns of operators as well as managers. Operators are able to evaluate goals and judge whether or not they are 18 Computer-Aided Process Planning (CAPP) realistic. In some cases, managers are surprised to find that operators are eager to recommend improvements and identify problems once they understand what is expected of them. Shops must accept the basic premise that the definition of the plan becomes the standard against which performance is measured. Therefore, it is extremely important to create plans carefully and follow them closely. A major area of difficulty is the process of work release and determining manageable lot sizes. Introducing formal work release procedures requires each production run to have its own original traveler. Those in charge of work release have to learn to assign jobs to specific plans and create individual travelers for each lot of workpieces. Several problems typically occur: • Some production control personnel create single lots for the entire order quantity, even when a large number of parts are to be delivered over a long period of time. These lots are obviously too large and impossible to schedule or track effectively. • Some individuals create lots based on container size; e.g., if a tub holds eight pieces, a lot of eight pieces is created. • Lots are created using the wrong process plans. Often, multiple plans for the same part exist, depending on what has to be done to the part. This situation creates the danger that the wrong traveler might be used and released to the shop floor. Education provides the solution to these problems; e.g., a lot is defined as a quantity of parts in production that can b e tracked and scheduled in some near term capacity. T h e concept involves determining lot sizes based on delivery requirements. O n e or m o r e sets for shipping can be combined to make a single lot. Likewise, production control personnel can determine lot sizes based on the most economical quantity to produce according to delivery dates. Customer orders with "as required" delivery dates can be a problem. Scheduling these orders is difficult because the quantities and delivery dates are not known. These orders are handled "on demand" and worked into the schedule as time permits. 2.1.5 Procedures Establishing standard procedures and creating automated tools to support them is a lot of hard work but ultimately streamlines the process. In addition, keeping plans in a central place helps to meet the demand for documentation, planning, and traceability. Online displays allow users to view plans and travelers as they are being created. T h e bill of materials lists resources for every operation. These resources can be other workpieces, various materials, tools, fixtures, or anything that is required to perform the operation. If the resource has to be made, a plan is developed and a resource-traveler created. In this way, a tree of plans can be constructed to correspond to the production requirements of a particular assembly. Split lots are always a problem. T h e most c o m m o n reason for splitting is to meet schedule demands of customers. T h e procedure to solve this problem provides traceability of lots as they split. When a lot is split, the last successful operation of the original lot is identified. Then o n e or m o r e parts can be separated to form a second lot for additional processing. This method provides a history of the job through production. 19 Computer-Aided Process Planning (CAPP) 2.2 Selecting a n d Implementing a C A P P System 2.2.1 Approaches Two of the most difficult questions confronting the selection of a C A P P system are: • What system best suits my company's needs • W h a t steps are required to successfully install the system or systems chosen T h e probability of o n e system's fulfilling all of a company's needs is relatively small. As o n e might expect, n o single system performs every function required by a company. F o r example, some systems interface to design, through the ability to translate I G E S (Initial Graphics Exchange Specification) files. These systems are able to communicate with C A D systems more readily than systems without this capability. However, companies manufacturing products to customer's designs have little control of design. They may not be as interested in the ability to perform design retrieval for productibility analysis and standardization. O n the other hand, those same companies might be very concerned with the ability to transmit C A D data, CAPP instructions, and machine language instructions to the customer's plant or to distribute customer information throughout their factory or among plants in geographically diverse locations. Moreover, the product manufactured might lend itself to either variant or generative CAPP systems. In generative environments, companies may wish to consider a rule base for design producibility as well as a rule base for manufacturing instructions. In variant environments the ability to direct process planners or design engineers to preferred practices is desirable. If left uncontrolled, variant systems may direct users to a number of similar designs or processes, but unless steps are taken to prevent it, the user may select a poor design or process on which to base the variant. T h e mix of products manufactured may be an issue. In metal working environments in which G T and process planning originated, traditional G T codes perform satisfactorily. However, in companies producing printed circuit boards (P/CBs) or complex electromechanical devices, classification by shape may be unimportant, or perhaps useless, while the classification of other features, such as the components contained on a P/CB, will require the classification power of a relational database to perform effectively. Further, in these instances, ANSI S Q L or other relational database application languages are often required to make use of the data classified. For example, it is not always the case that a design or manufacturing rule may be based solely on the presence or absence of data in a single classification field. In most instances, manufacturing rules are contingent on a variety of characteristics. So, for example, when planning the manufacture of a P/CB, o n e might wish to know whether the components mounted on the board may be inserted automatically by machines. A component that is normally machine insertable, however, may not be machine insertable, if it is located in close proximity to a larger component that interferes with the insertion head. Because many large components are not adaptable for insertion by machines, they must be inserted by hand. In this case, the application language rules would call for the insertion, if the larger c o m p o n e n t were absent o r located at a distance sufficient to safely insert the other component. In the absence of these conditions, the process instructions would direct the individual along a different course. Work measurement is an issue at s o m e companies because of union agreements, compliance with military specifications, or incentive payment systems based on piecework or other forms of measured output. In these environments, the ability to standardize production methods and easily revise labor standards is essential. T h e situations exist in competitive markets typified by repetitive production, such as that found in consumer products. In other environments, one-of-a-kind or custom designs produced infrequently make labor standards a 20 Computer-Aided Process Planning (CAPP) relatively low-priority item required only for project control. In these very different environments, the automated capability to measure human or machine-paced work will be weighted differently. A n o t h e r factor compounding the selection decision is the difficulty involved with assessing the strengths and weaknesses of competing systems whenever they approach the problem from different directions. Many potential C A P P users, for instance, have situations in which both variant, generative, or semigenerative approaches are appropriate. In these instances, evaluating a purely generative system against a purely variant system can be quite difficult. Likewise, the differences between systems' capabilities to interface with other business systems, such as Manufacturing Resource Planning ( M R P ) , also vary considerably, further complicating attempts to make a solid evaluation. A n alternative approach, then, is to identify the business needs to be addressed by the systems under consideration and assess the degree to which competing systems meet business needs. This approach avoids fruitless attempts at comparisons between systems that approach the problem differently and also identifies areas in which perhaps more than o n e system will be required in order to meet all important business needs. 2.2.2 Identifying Business Needs An approach to identifying business needs utilized by major consulting, systems, or engineering firms involves flow charting the business systems to which the CAPP system must interface. T h e procedure involves a system "walk-through" from point of origin to completion, while, at each stop, interviewing the system users, collecting the documents and other information requirements used by the system, and finally, flow-charting the information processing and distribution of documents and information throughout the system. System Walkthrough O n e manufacturing engineering manager trained in systems design was amazed at the system interface required to replace paper, pencil, and file cabinet process planning systems with a computerized one. Even though his objectives were limited to automating the manual work, the information interface to other business systems proved much larger than o n e might anticipate for so modest a goal. As pointed out earlier, this is because the process plan contains information that permeates the business operation. Accountants use it to value inventory and as the basis for standard costs. Standard costs, in turn, affect pricing in the market. T h e difference between the market price and standard cost establishes profit from the sales. Usually, estimating or quoting activities for new orders and delivery lead times are based on manufacturing methods. As these systems are automated, the cooperation with data-processing elements of the business must grow. Communication of the information contained on the process plan must switch from paper to electronic media. Now, communication protocols and the timing of information transmissions become important. Because other data-processing users may now access the system, activities that previously lacked a time dimension may now be constrained by one. For example, electronic process plans may all be required to be u p to date and complete during the evaluation of a year-end inventory, or routing files may be absent or out of date when an M R P run is under way. In paper and pencil systems, the most recent copy on file always served this purpose. In electronic systems where only o n e image exists, it cannot be "locked out for update" when required by other elements of the business system. Many departments within the company rely on data contained in the process plan. Production scheduling and control systems rely heavily on information within the process plan for lead time calculations and capacity requirements for personnel and productive equipment. Make-buy decisions affect the purchasing department, engineering bills of material, and process instructions. O n the manufacturing floor, routings 21 Computer-Aided Process Planning (CAPP) may be used to report direct labor and material applied, trigger vendor operations, procure outside services, and record quality transactions such as scrap, rework, or acceptance into finished inventory. In some companies the process plan is used to report personnel-related information such as attendance or tardiness. In incentive or piecework environments, it plays a direct role in the calculation of employee earnings. T h e raw materials from which products are made, the perishable tooling of materials used in its production, and the capital equipment upon which it is produced constitute the largest source of investment for most manufacturing companies. When this information is compiled and distilled by management, strategic decisions affecting the company's course and position in the market are made. T h e accuracy of the process planning database is crucial for sound business decisions, and the interface to other business systems is important even when CAPP's immediate goal is limited to automating the functions of paper, pencil, and file cabinet. User Interviews The most important step at each stage of the process consists of interviewing the users of the system. These interviews are a source of invaluable information on the many small ways that improvements to the system can be made. O n e should not be surprised to find that much of the data collected can be eliminated because it is no longer utilized by any downstream process, many things currently being done manually might be better d o n e electronically, and the throughput time required to sequentially process through a paper system compared to that required for processing with an electronic system can be dramatically reduced. A n objective of the system designer should be to provide each user of the system with only the information required. N o information should be collected that is not utilized elsewhere in the processing system. Likewise, no useful piece of information should be omitted from the formal system design and accomplished through informal methods. At this state of the system design, interviews with users are important to gather information and understand how the system works so that the CAPP system can be successfully interfaced. Later, when the design of the new system nears completion, the system installer will once again return to these users to review the proposed design and verify that the planned system meets their needs. System Flow Chart It is good practice to start at the origin of the information system. For most process planning systems this will be at the point of quotation or order entry. From that point, o n e should follow the processing steps through to the final system transaction and output. At each stop, it is important to record the information input to the process step, the method by which the information is processed, and the o u t p u t of that process. Flowcharts may be used to document the channels through which the processed o u t p u t is forwarded. This is accomplished by interviewing the users at each step in the process and collecting completed forms used to capture, process, or transmit information. W h e r e electronic systems are involved, the same steps take place; however, the information is captured in the form of software documentation for the electronic systems employed. Often, however, electronic systems are not documented or are incompletely documented. In these instances, flowcharts which identify each piece of information and the way in which it is dispositioned must be constructed by the systems analyst. This step, while time consuming, is very important. When the new C A P P system is installed, users of 22 Computer-Aided Process Planning (CAPP) every business system to which it interfaces must receive the information required to perform their job or function. These flowcharts, forms, and software documentation assure that their information needs are met by the new C A P P system. O n e of the ways in which this assurance is provided comes from the forms collected. As the manufacturing engineering manager, mentioned earlier, progressed through sales order entry and the various downstream departments, he noticed that many of the forms in use n o longer fit the information processing system. In some cases, personnel simply used old forms because a supply of several hundred still existed. These older forms often lacked a formal method for capturing important information and, conversely, often provided blocks for information that was no longer useful. T h e reason for taking examples of completed forms at each step in the business system is to identify instances in which useless information is captured and not subsequently processed. Conversely, whenever annotations are made on the reverse of forms, the margins of forms, or through informal channels not provided in the formal business system, o n e can take steps to ensure that these deficiencies are corrected and formally addressed in the revised system. New business forms are very cheap when compared to the risk of processing incomplete information, missing a customer delivery, or incorrectly pricing a major job. D o not succumb to the temptation to use old, inappropriate forms simply because several hundred or several thousand exist in the office supply's storeroom. Throw them out and design forms appropriate to the business needs of the system. Take advantage of opportunities to install better controls through the use of serially prenumbered forms with controlled distributions, whenever the importance of the activity being controlled, such as purchase orders or manufacturing orders, so warrants. After the information is collected, it must be distilled and digested so that the business system is understood. This is accomplished through the use of flowcharts. Flowcharts record the activity that takes place at each step, documenting the information input to the activity, the processing that takes place, and the information content of the o u t p u t distributed from the activity. Flowcharts explain both the processing of paper and the electronic-based business systems. As our manufacturing engineering manager proceeded through the various departments, he documented the information that was received by each user of the system, what he or she did with that information, collected examples of the completed forms used in processing the information, and noted where the information was sent. System Design With the systems "walkthrough" completed, the system installer is in a position to construct a preliminary system design based on the information requirements that must be met by the C A P P system. Understanding of the information requirements that the C A P P system must provide to other business systems and the timing and controls necessary for each process step at which that information must be exchanged is a necessary function of this process. This understanding helps the installer to specify the minimum capabilities of the C A P P system as well as other timing and execution issues that must be addressed prior to installation of the system. To complete the preliminary system design, the system installer will require the assistance and participation of a number of other professionals. For example, expertise in the data processing, accounting, and materials area is often required in order to assure that the redesigned system performs adequately. With input from these personnel, a detailed system specification can be written and implementation activities coordinated among several departments. T h e preparation of this specification may be expected to touch most, if not all, of the company's business systems. In addition to capturing and providing a processing system and destination for all the information, system 23 Computer-Aided Process Planning (CAPP) design tradeoffs must also be evaluated. System architecture is often an area in which many alternatives are available. In single-product systems, access may be controlled by the system manager through the assignment of various file, field, o r device privileges appropriate to a variety of users. In multi-product operations, however, coordination is much m o r e difficult. Decisions must be made to determine what data is to be shared among various products, what is to be product specific, and how all the data within the system is to be secured, maintained, and recovered if lost. T h e transition from paper to paperless systems operation adds another dimension to the system design effort. Sensitive or classified data must be protected. Since such data is most vulnerable during transmission, o n e may expect to be required to secure the entire area in which transmission takes place in order to satisfy the requirements of government security officers. Transmission on unsecured media is not likely to be permitted regardless of password security or encryption techniques employed. Specification Review T h e last and most important step is to review the proposed system specification with all users. O n c e again, begin at the point of origin and trace the system through to the end of processing. At each step, identify for each user where the information required for the performance of the task will come, how processing will take place, and how the information will be distributed subsequent to that processing step. While time consuming, this second iteration through the system is essential. It is rare that even the most talented, experienced system designers complete a workable system specification in o n e pass. By reviewing the detailed system design, insight can be obtained into the design, the alternative selected, and the advantages and disadvantages of the tradeoffs made e n r o u t e to the final design. Most importantly, the users provide a valuable check on the design assumptions around which the successful design of the system will depend. 2.2.3 System Selection Survey Market Several means exist by which a survey of commercially available C A P P systems can be made. A variety of seminars sponsored by professional societies dealing with CAPP, G T , and related subjects are scheduled annually and often provide an afternoon or evening in which attendees can view and discuss systems from various suppliers. Also, tool shows and expositions often provide the opportunity to see and evaluate various C A P P systems. Identify Candidates When the vendors who seem most likely to meet the needs of the business have been identified, a second screening may be made on the basis of literature provided by each vendor. O n the basis of this review the three or four suppliers most likely to meet the needs of the business can be identified and examined more closely. In order to m a k e the final determination, the field is narrowed to the two C A P P systems that most closely meet the business needs identified in the preliminary design effort. Then the finalists are put through a number of steps, including site surveys, sales presentations, systems analyses, technical demonstrations, and formal proposals. 24 Computer-Aided Process Planning (CAPP) 2 3 Cost Justification Although current C A P P systems are not intelligent enough to capture all the seemingly infinite number of variables that are associated with process planning, they are maturing and becoming more of an economically useful tool. As these systems grow in power and sophistication, they are becoming recognized as a "necessity" for those companies who wish to stay competitive in the future. Many large companies have delved into C A P P and found it to be economical in spite of some limitations and costs. As an increasing n u m b e r of larger companies use C A P P systems from a growing number of vendors, a trend takes place that drives the cost and availability of such systems down within the grasp of medium and small size companies. Regardless of size, there comes a time when a company in the market for a C A P P system needs to justify the cost of such a system. Before management commits significant dollars or resources in obtaining and implementing a C A P P system, they want to know if the benefits will outweigh costs. For this reason two questions need to be answered. H o w much is it going to cost, and what savings can be expected? Questions pertaining to such things as acquisition costs of hardware, software, maintenance, training are straight forward. These can be readily obtained from a vendor. If the C A P P system is to be developed by the user, then software development cost, dedicated h u m a n resources, and implementation lead time are less concrete. Capturing cost of an in-house developed system is even m o r e elusive. Some form of a project management system is usually employed. T h e scope of work is drawn up and the resources needed to support the work are estimated. F r o m these estimated resources a ball park figure can be arrived at for the total system cost. However, the question of savings becomes even more elusive than defining the cost of an in-house developed system. Some fortunate companies have a cost tracking system that is well developed and supported. This makes the job of estimating and tracking savings much easier. A good cost tracking system is used as a solid standard on which savings can be based and projected. A number of companies however, are not as advanced in their accounting methods and do not break out as many cost categories. Thus, they struggle to find an exact cost in their existing system. A real problem with projecting savings is to determine how much work a piece of software as complex as a C A P P system will save in an area as all encompassing as process planning. After installation, system performance can be measured and compared with historical data, if that data exists. In this case foresight is needed first to justify the system. 2.3.1 System Analysis A gas turbine engine manufacturer produces a wide variety of parts most generally characterized by complex shapes and tight tolerances. D u e to the inherent quality and reliability requirements for aircraft propulsion engines and the complexity of design, process plans are very lengthy, complex and meticulous. A typical process plan ranges between 30 to 50 operations. Each operation contains machine tool used, perishable tooling, jig and fixture tool numbers, operation description, machining dimensions and tolerances, caution notes, specifications in a condensed form, speeds, feeds, depth of cut, and graphics. T h e engine builder uses G T techniques. All parts are produced in o n e of four shops, namely, wheels, gears, cases, and sheet metal. Though this is not a complete breakdown of parts into "families of parts" as pertaining to G T it does significantly reduce the magnitude of searching needed to develop the complete process plan as well as benefit the shop floor in facilitating production. T h e company examined the detailed requirements for a C A P P system and defined the modules needed to satisfy its design. Because of the interest aroused by this study, it became necessary to provide to management as soon as possible a complete project analysis with associated costs and benefits. 25 Computer-Aided Process Planning (CAPP) A single engineer handled the analysis. T h e only other human resources dedicated were a few hours each week of an experienced process planner. This in-house C A P P system development was labeled C A P E (computer-aided production engineering). Parts of the system were purchased from outside vendors, but for the most part, it is an in-house design. C A P E is an enhanced variant process planning system, which interfaces with the existing company database and integrates with all other departments and functions concerned with process planning. 2.3.2 Process Plan Analysis T h e strategy used for cost analysis was simple: select a representative family of parts and develop a standard routing for that family. Next, compare and contrast the standard routing with the existing routings and carefully n o t e all differences. This would be followed by an analysis of those differences to see what, if any, the cost impact would be. Lastly, the findings of the analysis would be applied to actual cost data for a projected cost savings. D u e to time constraints, only o n e part family was chosen for analysis. T h e family chosen was that of spur gears with a trepan configuration. This choice was made for two reasons. First, there was an existing classification and coding scheme in the gear shop with a significant number of the parts coded. Second, it was felt that spur gears would be as representative a part family as the shop could offer, considering the wide variety of part types. Seven parts fell within this spur gear family and were used for the analysis. After the process plans were gathered, a production flow analysis was performed on the seven parts of this family. A matrix was constructed consisting of a vertical listing of the routing; i.e., a list in descending order of each machine tool used in the routing of the part. T h e horizontal axis of the matrix consisted of the respective part numbers of the spur gear family. T h e matrix had to be built o n e column at a time. Each operation for a part had to be as closely fitted and aligned to the previous part routing as possible. This was made fairly simple, due to the great number of common operations that are a natural outcome of G T families. T h e operation n u m b e r was placed at the intersection of each part number and machine tool. Operation numbers are assigned sequentially, regardless of machine type or process. The load center number which appears beside each operation takes the place of what many companies call an operation number. Following the matrix construction another vertical column was added. This column was a machine class or process type column. An example of this would be calling out an operation as using a medium horizontal N/C lathe as opposed to a specific company name and model type N/C lathe. This column was added for use in generating a graphics flow chart of the production flow on a higher level to avoid a chart with too much detail and clutter. At this point, the experienced process planner was called upon to go through each routing in detail to create a standard routing. This was a time consuming task taking o n e of the three weeks allotted to this project. A process planner makes hundreds of calculations, judgments, and decisions to create a standard routing. Although the seven parts of the spur gear family were nearly identical in geometry and material, their routings differed considerably. T h e differences were because of a number of reasons, among which were obsolete machinery, changes in methods, new equipment, different approaches and preferences of the planner who created the routing, changes in tooling, and changes in company policy. From this careful examination of each routing, a standard process plan was created that would cover all seven spur gears. T h e standard routing was then compared with the matrix created earlier, and the matrix was adjusted to reflect the standard routing. Operations were deleted, combined, and, in some cases, added to the matrix. T h e o u t c o m e of all the changes was a decrease in the number of operations. Before the analysis, there had been a total of 221 operations among the seven parts studied. After the standard routing was 26 Computer-Aided Process Planning (CAPP) produced and compared, there were 201 operations among the seven parts. N/C operations were reduced from 14 operations to nine. For demonstration purposes and as a visual aid a flow chart was made of the spur gear family routing flow through the shop. This simple line and circle flow diagram was used to visualize the magnitude of changes that had taken place on the matrix. It is easier to verify a standard routing when all possible paths are laid out for easy visual access. In addition, this chart was also used in presenting the C A P P system to management. A network, if not too complex visually, is m o r e easily understood in a meeting atmosphere than volumes of tabular data such as the matrix. After the standard routing was created, the changes were transferred to the matrix. This resulted in moving, deleting, and adding operation numbers at the intersections of the part number and the machine or operation description. W h e n this process was finished, all vacant lines were removed. T h e analysis discovered that a number of operations were performed in the same sequence from part to part with the only difference being the brand name of the machine to which it was routed, which led to further condensation of the matrix; e.g., four lines of medium horizontal N/C lathes with different manufacture names being combined into o n e line. 2.3.3 Cost Analysis After the estimated percentage savings were calculated, costs were gathered. Baseline cost data in this case was from the previous calendar year for each task area. Cost data was gathered from a number of individuals and reports. In most cases, the information was present and accurate. Only in a few instances was a "best guess" made by an experienced person who had a good handle on the situation of that department. T h e collected cost information was formatted on a P C using a spreadsheet software package. T h e format used was to group each major section; e.g., run-time, N/C, scrap. The percent of impact was calculated by the a m o u n t that each shop spent for a particular area compared with the total for four shops. A phasing factor was added, which took into consideration the time each shop would require to move from the manual system to the C A P P system. T h e phasing was linear and reached 97 percent at its highest point in anticipation that a full implementation would not be realized due to various unforeseen conditions and circumstances. Savings for each shop were totaled as well as saving by each column or quarter for the different areas. This pattern was repeated for each of the areas. Implementation costs included a combination of purchased software, hardware requirements, and personnel needed to design, implement, and maintain the total system over a period of five years. Wages and salaries were adjusted using a six percent per year inflation factor. Net savings per quarter were derived from the total savings less total costs. These net savings were then summed to a total net savings over five years. 2.3.4 Savings Substantial savings were projected from this study. A rate of return was calculated at greater than 50 percent, with the pay back occurring in the eleventh quarter. With these figures it was deemed a worthwhile project to undertake. A presentation was made to management using these figures as well as the graphical flow chart of the spur gear routing as mentioned earlier. T h e outcome was a general approval of the C A P E system with further deliberation on the technical requirements of the software. 27 Computer-Aided Process Planning (CAPP) 3. Manufacturing Interfaces 3.1 A Distributed, Hierarchical Architecture 3.1.1 Process P l a n n i n g Manufacturing personnel are generally familiar with the organizational distinction between manufacturing engineering and production control. Manufacturing engineering has responsibility to generate the process plan. Production control has responsibility to manage the production activities, hereafter referred to as process control. Process control is the domain where product is fabricated, assembled, tested, and transported. It is also where the process plan is interpreted, optimized, executed, and monitored according to specification. It is the domain of process planning to specify the manufacturing process requirements in terms of sequence and content, and to arrange this data in a format acceptable to multiple process control subsystems. Process planning is positioned between product design engineering and process control of the factory floor systems. Because the two domains share product and producibility data requirements in common, it is compelling to merge the product design with process design. Traditionally, this has not been the case. There is, however, a need to recognize the necessity for strong two-way integration between the two domains to support new product startup and producibility goals. While it ultimately is a goal to completely a u t o m a t e the generation of the process plan, this architecture does not require it. In fact, with the wide variety and increasing introduction pace of new manufacturing process technology, the practicality of o n e package meeting the data needs of all process control systems could likely become impossible. T h e architecture lays the foundation upon which process planning applications can plug in or out or be implemented or replaced, as required in an integrated environment. Currently there is little integration of shared data across factory systems. R o u t e s and bills of material are p a r a m o u n t to the integrity and proper revision control of the manufacturing process. When a revision is made to a process, this change should be automatically communicated to all affected systems. Currently, manually maintained, standalone systems requiring people to run around to keep them operating is the glue. F o r the future, manufacturing strategies must insist on integrating this reference data so that data changed at the p r o p e r source is appropriately communicated to all affected systems automatically. 3.1.2 Architectural F u n d a m e n t a l s O n e of the most fundamental elements of complex system design is the architecture upon which it is built. Architecture can b e defined as many things by many people, but at the very least it describes the system components, their interfaces, and their relationship to o n e another. Just as an architect designs the structure before it is physically built, so does the software architect design the system architecture prior to application development. Interface Protocol Standards In designing a building, the architect must adhere to certain standards, the purpose of which is to guarantee a level of compatibility between independently designed functions that have to interface with o n e another. A n example of two independently designed components that ultimately interface is a wall socket and a lamp plug. Although the socket and plug perform different functions, they both have been designed to meet the same interface standard. Because a standard exists, a radio may be plugged in instead of a lamp without impact on the system. 29 Computer-Aided Process Planning (CAPP) Because of standards, applications may be switched in and out, and as long as they do not violate their interface standard, the other system components are not impacted. This approach allows for flexibility in the implementation of a system where competing modules can be selectively implemented based upon requirements. A given application module may be replaced by another, but if the interface remains constant, the rest of the system should not have to change. Data Server/Database T h e architecture is data server rather than database oriented. In a database oriented architecture, the knowledge of the data structures have to be coded into the application. In a data server approach, applications obtain this reference data via data management interface calls. By doing so, o n e database product can be modified or replaced by another database product. As long as the interface call does not change, the application will not have to change. This also can open the way to a database machine whose sole role is to provide data storage/retrieval to any number of requesting client applications. Function/Form A house design is based upon the function that each room accommodates. "Kitchen" conjures a mental picture of a cooking facility. T h e basic function of cooking will remain the same regardless of the size or number of kitchens in the system. W h e t h e r a large stove or a small stove is requested, users know that it goes in the kitchen and not somewhere else. This analogous approach provides great insight into how a complex system is designed. Within process planning, there are four basic categories of data: workflow (route), bill of material, quality, and equipment. These categories should be as distinct and well defined as a kitchen is from the rest of the house and understood as to their role in the process plan. C o m m o n Support Tools Within most systems, a common set of tools can be identified that are used throughout the system. These tools include networking, human interface, and data management. O n e such tool used extensively within the architecture is called the message bus. T h e use of this tool sets the stage for distributed processing and removes the knowledge of where on the local area network the cooperating application resides. This transparency is critical to the ability to move application modules to other computer hardware without having to make changes to the applications making the calls. For an application to request service, all it has to do is call the message bus and pass the list of arguments using a predefined interface call. This would be analogous to our writing a letter (the list of arguments) and addressing an envelope. W e do not care what the postal system does in delivering the letter, only that it reaches its destination in a timely manner. T h e message bus takes care of physically routing the packet of data to the appropriate destination. Hierarchical Approach T h e architecture assumes a hierarchical approach in positioning the process planning functions to minimize redundancy. Plant levels such as workstation, cell, and shop can be arranged in a manner similar to that of a traditional top-down organization chart. That is, a shop has any number of cells reporting to it; each cell has any number of workstations reporting to it, and so on. 30 Computer-Aided Process Planning (CAPP) T h e above considerations are fundamental to the architecture discussed below. T h e architecture has made extensive use of a protocol standards, modularity of functions, and data servers, all based on the use of a distributed network message bus system. 3.1.3 Architectural Overview There are three major subsystems within the process planning architecture. They are 1) the bill of description ( B O D ) , 2) the process planning development hierarchy that is made u p of functions used to generate the process plan, and 3) the bill of process (BOP). Data management interfaces are shown as horizontal lines between these subsystems. Figure 11 shows this relationship. Not shown are the command level interfaces connecting the subsystems within themselves as well as with external entities, such as process control and product design engineering. Business Factory Level Generator Bill of Managerial > Shop/Cednter Functional ^Cell/Line Data Management Operation^ System -Workstation Recipe -^Automation Module Shop/Center Level Generator Description Cell/Line Level Generator Data Management System Factory Bill of Process Workstation Generator Instruction Set Generator PROCESS PLANNING DEVELOPMENT HIERARCHY Data Input |^ Off Line Process Development • On Line PD _Process "Controllers" Figure 11: Architectural Overview Bill of Description T h e B O D is a data management system focused on management of data that is independent of any specific process plan but may be referenced by any number of process plans. It manages relatively static data that describes product/parts, plant resources, and general process knowledge requirements. G T coding and classification approaches would fall within this domain. Notice that this data remains constant and does not change unless there is a change in product design, plant resource, or manufacturing strategy. Part Product data must be provided by design engineering in a state acceptable to manufacturing. This requires an interface to the C A D systems, commonly referred to as the C A D / C A M interface. This includes part master and manufacturing bill of material (BOM). 31 Computer-Aided Process Planning (CAPP) Plant Plant data is owned primarily by manufacturing. It describes the resource capabilities, capacities, configuration, and control hierarchy of each plant entity. Data owned by finance systems, but considered to be plant data, are master data tables such as operation codes and employee identification numbers. Plant data is used by a number of systems including process modeling/simulation, short interval scheduling, capacity resource planning in M R P II, as well as other process control systems. Process Knowledge Process knowledge is a grouping that represents the rules, knowledge, and strategies that are used to generate/develop a process plan. T h e process rules are general in nature in that they may apply to any number of process plans and are therefore specific to a process in a global sense as opposed to a specific process plan. Only as the manufacturing process strategy changes do the process rules have to be modified. Data Management Interfaces Referring to the B O D in Figure 11, are two sets of data management interfaces. The interface diagrammed to the left is used to maintain the B O D data. This interface set supports a full range of database manipulation capabilities including creation, modification, and deletion of data. The functions responsible for maintaining this data can be either front-end terminal screen-based programs that retrieve and modify the data, or some other on-line data management system that uses this interface set to load the data into the B O D . However, the most desirable option when dealing with another on-line system is to place data servers on that data source thereby effectively integrating that system into the larger B O D data management system environment. Regardless of the source of the data, any application can make data management queries conform to the protocol interface specification. This is the set of interfaces extending out of the B O D and into the process planning functions. As denoted by the o n e way flow of data, these support read-only data access. This guarantees that the data can only be changed by the originating source of that data. Therefore, the modify and delete interface calls are not allowed for use by the process planning function. T h e Development Hierarchy T h e primary responsibility of this subsystem is to generate a cost effective process plan that meets the product and process specifications as defined in the B O D . In a traditional sense, the process engineer has to interpret the product specifications, understand the process resources available, and, drawing from previous experience and process knowledge, generate the process plan. Before the plan can be released, it must be verified and approved. T o verify a plan may involve process simulations against a specific o r average product load using varying resource allocations. Verification may also take place to assure that all recipe programs are syntactically correct and that the data is of the correct format and meets the data requirements of the various process control systems. Hierarchy Different types of data requirements can be seen in Figure 11, depicting five levels of a production facility. Business data refers to the level of process plan data that plant wide systems such as M R P II would require. Managerial data refers to the level of data that is required by shop level subsystems to optimize, 32 Computer-Aided Process Planning (CAPP) execute, and monitor production tasks across a process line. Functional data refers to the process plan requirements of the cell/line controller to specify the activities to be executed by its workstations. Operational data describes the coordination activities within the workstation. T h e recipe data provides the predefined sequence of activities required by an automation module. T h e difference between on-line and off-line underscores the distinction between creation of the process plan and the access of this reference data by process control. T h e process planning components that create the process plan data are commonly referred to as generators; e.g. instruction set, workstation, cell/line, shop/center, or factory generators. T h e automation module level is the level in process control that the instruction set generation addresses. T h e instruction set generator is probably the best understood component in the hierarchy. T h e output of instruction set generators are known as N/C programs, patterns, scripts, pick and place programs, automated test programs, vision programs, and instruction sets for human operators. T h e major responsibility of the instruction set generator is to arrange the automation module program instructions into an optimal order, given the input limitations provided from its workstation generator. For an instruction set generator to generate a program, it requires a data management interface to the B O D for product and standard set up information; and a data management interface to the B O P for contextual process plan development information such as setup requirements/restrictions and the specific component parts involved in the step. It also requires a command via the command level interface from its higher level workstation generator to initiate the data management interface calls and begin the instruction set development activity. T h e responsibility of the workstation generator is to manage the instruction sets generated by its instruction set generators as well as provide operating data such as state tables, which the workstation controller requires. T h e workstation generator understands the relationships that exist between the workstations and their automation modules. Each of the cell/line, shop/center, and factory generators perform the same type of responsibilities that the workstation generator performs. However, each generator addresses the data requirements appropriate at its own level. Each level has the capability of receiving and responding to high level commands, making decisions within the command limits allowed, and assigning and monitoring process planning tasks at its lower level. T h e exchange of data between any two generators is termed as the command interface protocol. The generator that originates the command is the source and is at a higher level in the hierarchy than the commanded generator. Commands travel downwards and command responses are returned upwards. Therefore, commands travel vertically within the hierarchy whereas data management interface protocols travel horizontally. T h e types of data addressed by the command interface are: • the part number of the part being built • the material/components involved in the activity • the equipment configuration (set-up) requirements • a unique tag identifying the specific output process plan data so that subsequent communication between any two generators is guaranteed Efforts continue on standardizing process control command level interfaces. T h e manufacturing message specification (MMS) is o n e such standard that, although targeted for relatively unintelligent devices, provides the command interface protocol concepts required for communication between process planning generators. 33 Computer-Aided Process Planning (CAPP) Categories of Data A n analysis of the data needs of the process control system highlights four basic categories of d a t a workflow (route), material, quality, and equipment. These categories of data are found replicated within the hierarchy as often as the data is required by process control, based to greater or lesser degree upon the need of the corresponding process controller. T h e process planning functions required to generate this data will exist at the same level that the data is required. In process control, the workflow controller optimizes the production build schedule through use of short interval scheduling (simulation) techniques. This optimized list becomes input into an execution controller that controls the workflow based on run time status. A monitor controller compares the theoretical optimized schedule against the actual load status and makes a determination of when to take corrective actions including regenerating a n o t h e r optimum build schedule. T h e workflow generator in process planning must take the data needs of each of these workflow subsystems into account when it is generating the process plan. Typically, the data needs of the optimization controller are different from the needs of the execution controller; and the needs of the monitor are vastly different from the needs of the other two subfunctions, even though all three have some of the same data needs in common. O n e c o m m o n data thread across most process control workflow subsystems is the need of routing information. The workflow generator must be able to generate o n e logically consistent route definition across multiple subsystems, while meeting the independent data needs of each of these workflow subsystems. T h e route definition should also be able to support the branching logic for rework and other alternate next step scenarios, and arrange this in a hierarchical manner consistent with the way process control expects to assign the process plan to its process control hierarchy of controllers. Once the workflow is defined, the three other functional categories of data can be developed and associated to a specific point within the workflow definition. T h e material generator working off of the route definition determines the consumption p o i n t - s o m e t i m e s called an operation or s t e p - o f every component part listed on the manufacturing B O M . T h e quality generator must provide the data necessary for process control to monitor the process and to determine what, if any, action should be taken to resolve an unacceptable quality problem. T h e equipment generator among other things specifies the number and type of tools required to perform the task, the tool wear characteristics-used by process control to track preventive maintenance criteria, and the recipe data required by the technology dependent automation modules. B O P Data Management Systems T h e B O P data management system maintains the process plan data generated by the process planning generator functions. It is a logical data management system in that the data may very well be physically stored closest to its ultimate point of use. T h e approach taken recognizes that workflow (route) definition provides the framework upon which all other process plan data is associated. When a point in the route is established, o n e can begin to associate detail data that helps describe the activity at that process point. As such, the B O P must be capable of modeling a hierarchical route where a route step at o n e level can explode into a subroute composed of a set of steps describing the lower level route requirements. As shown in Figure 11, each level within the B O P hierarchy addresses its own level of process plan detail data. 34 Computer-Aided Process Planning (CAPP) Once a step in a route has been defined, attribute data can be associated to that step by the use of "views." A view defines some a m o u n t of data that describes what is to be done by a subsystem at a step. If it were a transport system, then the data associated to the step will be of the content and format useful to the transport controller to execute its requirement. A n o t h e r subsystem may be involved in the same step but for a different purpose; e.g., the M R P system that is tracking work in process and labor would require a different set of data that would describe the step requirements in the content and format acceptable to MRP. View data can be of any content and format required by a workflow optimizer, workflow execution module, workflow monitor, material controller, quality controller, or equipment controllers, including route step data, a recipe, a list of tools required, fixtures, graphical/textual work instructions, and data collection requirements. Giving a group of data a view n a m e provides a way to store and retrieve process plan data. It can be stored and accessed by referencing the view n a m e of the step in question. Data Management Interface Because of the need to appropriately model the various subsystem views of data and represent it hierarchically, a general interface to read and write process plan data to and from t h e B O P data management system has been devised. The B O P has two sets of clients. Internal clients, such as process planning generators, have read/write access because of their role in the generation of the process plan. External clients, such as process control, have read-only access since their role is to execute the process plan. There are four basic read-only call interfaces used typically by process control subsystems to obtain route and other workflow related data. G E T _ R O U T E is generally used when the route data is batch loaded as a private route copy into a n o t h e r subsystem. This might be d o n e to download a workcell route to a vendor specific F M S subsystem. If this were the case, then "FMS" could be the view n a m e and the group of data required by the F M S could be retrieved by G E T _ R O U T E . A run time interactive scenario is also supported when process control subsystems request route data on a step by step basis, as each step is completed. R u n time requests use G E T _ F I R S T _ S T E P , G E T _ A L L _ N E X T _ S T E P S , and G E T _ N E X T _ S T E P . This approach frees up the requirement of the subsystem to maintain a local private representation of its routing requirements and puts the data caching back into the hands of the B O P data management system. T h e same approach can be taken to define the interface calls that support the process planning generators' ability to create, modify, and delete B O P data. 3.1.4 Integration to External Architectures As important as it is to have a well structured process planning architecture, strong integration must also exist with the product design architecture. Figure 12 highlights how this integration can occur by use of the message bus. At the top of Figure 12 is a series of interfaces to external architectures such as product design engineering, order transaction processing (for job shop/customer order specific requirements), producibility, process control, and other systems/functions external to the process planning architecture. Each of these architectures must be tied into the message bus using an agreed upon set of interfaces, thus achieving a closed loop system. T h e formats of these interfaces need to be general in nature because they are expected to support a wide variety of data transmission requirements. Use of the message bus allows for a distributed environment where generators can b e scattered throughout 35 Computer-Aided Process Planning (CAPP) Product Design Engineering τ Annotated Product Data Product Data Producability Order Transaction Prooessing Process Control τ Process Planning Interface Message Bus Product ECO/ECR Change Notification Process ECO/ECR Change Notification Process Plan Detail Data Customer Order Rqmts 1 Bill of Description (BOD) 1 Process Planning Generators Bill of Process (BOP) Figure 12: Process Planning External Architecture the network. And because this generally means distributed databases, a basic problem involves keeping the system updated to the most current information. That problem is solved in the process planning architecture, as data is changed by o n e subsystem, notification of that change is "broadcast" via the message bus to all affected systems. The message bus is a utility which guarantees delivery of the broadcasted message to any subsystem which has previously registered itself as an interested party for a particular broadcast message type. The broadcast message can contain all of the data associated with the change, or it can contain pointers that map back to the change data. In the first scenario, the message is looked upon as a "push" of data to the subsystem. Generally this approach is used when the content of data is minimal. On the other hand, if the content of the data is great, then a "pull" approach may be preferred which gives the individual subsystems the ability to request the changed data at some convenient future moment. T h e pull approach is used in the process planning architecture to obtain product description data from design engineering architectures as well as customer order information from order transaction processing systems. A notification of E C O / E C R (engineering order/change) product data is received as a "pull". This means that the detail product data is not passed as part of the notification message but is subsequently pulled into the B O D which manages among other data, the raw product detail data. Once product data is in the B O D , it is checked for completeness. If the data is found lacking, then a subsequent unacceptable broadcast can be made back to the C A D / C A E source requesting the modifications to be made. If it is extremely critical due to time or cost constraints, then it may be annotated as long as the annotated product data is broadcast back to the true source/owner of that data, the product design engineering community. 36 Computer-Aided Process Planning (CAPP) T h e process planning generators then perform their value added functions by merging the B O D data into a consistent process plan. T h e process plan data is physically stored and managed by the B O P data management system. O n c e the process plan is complete, notification that the process plan has been released is broadcast via the message bus. Any interested party can then retrieve the released process plan data via the message bus using the standard data server interface calls supported by the BOP. 3J2 An Integrated Scheduling System 3.2.1 System Overview A n automatic C A P P system based on G T concepts has been developed and combined with a group scheduling algorithm called key machine loading (KML) to form an integrated computer-aided process planning and scheduling system (ICAPPS). New concepts, like part machining surface set (PMSS) and cell machining surface set (CMSS) proved effective in combining the automatic process plan generation with GT. T h e K M L algorithm guarantees no-idle time for key machine tool and minimum lead time for the job. T h e integration of C A P P and scheduling is accomplished through the use of the cell machine loading and alternatives algorithm (CMLA) and scheduling source file (SSFL). This allows the system to modify the process plan automatically and thus, to avoid possible conflicts when changes in j o b orders and machine tool allocations occur. 3.2.2 CAPP and the Process Plan Generally, C A P P and scheduling algorithm are performed separately. And yet such an integration is necessary, if modification of the process plan due to scheduling conflicts is to be carried by the system. This, in turn, would lead to the increased efficiency of those two important functions linking C A D and CAM. T h e task of integration is a difficult one, judging even from the fact that process planning is concerned with technical requirements, while scheduling is concerned with the time factors. In practice, modification of the process plan sheet is routinely necessary, because of the need to utilize alternative machine tools. This happens either because of overloading or breaking down of some machines or because of changes occurring in the job orders. An automatic search for the most appropriate alternative machine tool and subsequent modification of the process plan constitutes the rationale for the development of an integrated system. T h e automatic process planning generation is combined in this system with G T concepts as are some other new approaches. 3.2.3 ICAPPS Subsystems As can be seen from the block diagram in Figure 13, the ICAPPS system consists of three main subsystems: • • • the process planning generator the key machine loading algorithm the operation planner T h e C A D database is a common database used by C A D and CAPP. A small database is established to store part geometry data and technical requirements. T h e boundary representation of the solid geometry model is chosen to describe the parts. In the boundary m o d e l a face is defined by a list of edges, each edge being represented by two vertices. This is a general rule; in a real system, some special faces can be defined by conventional rules. 37 Computer-Aided Process Planning (CAPP) Design CAD I Geometric Model Technical Drawing Technical 1 Files 1 f PPG PMSS Generator ψ CMTT File —>• CDB Machining Sequence and SSFL Generator > < CMSS Files PAEK Rules 1 KML Algorithm Fixtures Tool Files Operation Planner CNC Tape Files CDB Process Plan Sheet and Job Order PPG PMSS CMSS SSFL PAEK KML CMTT CMLA CDB - Process Planning Generator Part Machining Surface Set Cell Machining Surface Set Scheduling Source File Production Axioms and Experience Knowledge Key Machine Loading Algorithm Cell Machine Tool Table Cell Machine Loading & Alternatives Cutting Data Base Figure 13: ICAPPS System Technical requirements can b e divided into the index data and technical data. Index data contains indices of brief information about all faces to be generated, such as shapes and surface roughness. It is easier to extract all faces by using t h e index data than by comparing the part geometry data with its blank data. Therefore, it would be advantageous to use t h e index data when a part and its blank a r e complicated. 38 Computer-Aided Process Planning (CAPP) T h e Process Planning G e n e r a t o r ( P P G ) T h e functions of the P P G are: • to extract faces from the boundary model and to integrate them into the PMSS • to compare the PMSS with the CMSS to find the best group of machine tools • to generate an approximately optimum machining sequence based on AI techniques and optimization criteria • to estimate the machining time for parts on every machine tool and to produce a scheduling source file T h e P P G obtains information from the data files, and functions as described below. T h e Part Machining Surface Set (PMSS) The PMSS is a set of surfaces to be generated through machining in the order which is arranged according to certain rules. Every machining surface or element of the PMSS is expressed by the shape symbol, size, and necessary technical data. The PMSS provides all essential information of a part for process planning. T h e technical requirements are represented by the accuracy and surface roughness. T h e r e can be the accuracy of size, shape, or position. T h e level of the accuracy-high or e c o n o m i c a l - d e p e n d s on the actual situation in the job shop. There are two rules for arranging the order of PMSSs for rotary parts: from left to right for external surfaces and from right to left for internal ones. For prismatic parts, the rule is from the datum plane to the top plane, with holes and slots included between the planes. Through the extension and combination of these rules more complicated PMSS can be arranged. T h e Cell Machining Surface Set (CMSS) T h e CMSS is a set of preselected surfaces which can be machined by a group of machine tools. In a job shop, only a part of the functions of a machine tool is used. T h e machining operation is related to tools, fixtures, and attachments. It is necessary to consider them when analyzing the shape generation on machine tools in the process of selecting a set of feasible surfaces to be generated by a cell. This set of surfaces forms a CMSS. T h e main parameters used to represent the CMSS are similar to those used in PMSS. Extracting Generated Surfaces T h e process of extracting surfaces to be generated is shown in Figure 14. At first, all faces to be generated are extracted from the part geometry model, then they are integrated into the geometry description of machining surfaces ( G D M S ) according to integration rules. O n c e G D M S is formed, it will be combined with technical data to form a machining surface. Finally, by repeating this process, the PMSS is generated. 39 Computer-Aided Process Planning (CAPP) Extract all faces to be generated according to the index data Data file for all faces to be generated Integr. Rules technical data to form a machining surface ι End Figure 14: Extracting Generated Surfaces Finding the Best Cell Generally, cells can be divided into two kinds: that for machining rotary parts and that for prismatic parts. When a part can be machined in o n e cell, this cell is the best choice. In order to find the best cell, the PMSS is compared with the CMSS. If the PMSS is included in the CMSSX (X is the cell number), then the CMSSX is the best cell. If n o n e of the CMSS can include the PMSS, the CMSSY, which can include most of the elements of the PMSS and requires the minimum number of additional machine tools to complete this part, is the best cell. T h e next cell is selected on the basis of a neighbor principle. Production Axioms and Experience Knowledge (PAEK) T h e production axioms are logical and general concepts and can be used to make logical inference. They are the result of analysis of shape generation process on machine tools and methods implemented to guarantee part requirements. These production axioms are adapted to any job shop manufacturing system except for some special cases. T h e production axioms concern: • the shapes of surfaces and corresponding machining methods and machine tools • some fixed machining sequences for generating surfaces 40 Computer-Aided Process Planning (CAPP) T h e experience knowledge usually deals with a particular manufacturing system and with the performance of an individual machine tool. It can be applied to: • shapes of surfaces and corresponding individual machine tool • typical machining sequence for a surface in a particular manufacturing system • other constraints (tools, fixtures...) Cell Machine Tool Table (CMTT) and Cutting Database (CDB) T h e C M T T is used for selecting the best machine tool from the point of view of the best manufacturing performance. In order to create this file certain rules and criteria have to be established. T h e data for C D B comes from industrial practice. Machining Sequence Generation The information obtained from the data files is used to generate the machining sequence. For this purpose AI techniques and optimization criteria are applied. AI techniques involve backtrack search strategy and production rules which are based on the production axioms and experience (expert knowledge), which are used to select the machining sequence of some surfaces. After this step, there are generally several surfaces left in the parts, which need to be machined. Optimization criteria is used to minimize machining time of the remaining surfaces. This criteria depends on a number of factors, including total length of a part, average diameter, width, cutting length, cutting speed, feed rate, and material. Some of these factors can b e treated as the main factors. In this system, the cutting depth, d, and the cutting length, 1, are considered as the two main factors. According to the variations in d and 1, several machining sequences can be produced, from which the best o n e can then be selected. T h e procedure of process planning can be divided into two steps. T h e first step deals with the surfaces which require high accuracy and low surface roughness. T h e second deals with those whose accuracy and roughness are economical to achieve. Every step has many substeps which vary from rotary parts to prismatic parts. T h e process planning generator will estimate the machining time on every machine tool on the basis of the similar principle and will produce a data file called scheduling source file (SSFL) for scheduling. T h e estimation depends mainly on the experimental data in a job shop. A n example is shown in Figure 15. Key Machine Loading Algorithm (KML) Generally, a load unbalance exists among a group of machine tools. A machine tool loaded more than others is called the key machine tool (KMT). T h e K M T with no idle time might ensure increased productivity in some cases. T h e K M L algorithm is based on this idea and is constructed to guarantee that key machines have n o idle time and the job can be completed on due date under a minimum lead time when batch jobs are scheduled among a group of machines. T h e K M L algorithm uses the C M L A file. This is an important file for integration of C A P P and scheduling. This file gives information regarding job starting time and completion time, machine loading, and alternatives. T h e modification of the process plan sheet depends on this file. 41 Computer-Aided Process Planning (CAPP) i PMSS Files I Compare PMSS with CMSS Determine The Best Cell CMSS Files I Select Finishing Methods and corresponding MT Select Allowance I Determine Optimum MQ Select Economic Methods and Corresponding MT CDB I Estimate Machining Time on Every MT CMTT I Primary Process Plan Files Scheduling Source File Blank Data MQ - Machinging Sequence MT - Machine Tool Figure 15: Flow Chart of Process Planning Jobs can be ordered based on the priorities fixed according to the due date. A batch of jobs is set on the same due date. In the batch job case with K M T as the first machine in a group, the K M L algorithm minimizes the total flow time T f , then calculates the minimum lead time T m which is given by T m = T f - T p, where Τ is a certain time period. 42 Computer-Aided Process Planning (CAPP) If T m is negative, no lead time is necessary. In cases when K M T is not the first machine, after the algorithm minimizes flow time T f , the minimum operation time T k m which guarantees that K M T has no idle time, is calculated along with the new completion time for all jobs. T h e new minimum lead time T m is then calculated: T m = T k m + ( T c - T p) , where T c is the last job completion time on the last machine. When scheduling random jobs among cells, K M L uses a dispatch rule called FCFS-A, where A means alternative. In this case, the algorithm deals with two problems: • scheduling jobs among all cells if some machine have enough idle time • cancelling some jobs in order to insert some urgent jobs T h e Operation Planner T h e Operation Planner has the following functions: • selecting cutting speed and feed rate and calculating machining time • modifying "machine loading table • selecting tools and fixtures • printing the process plan sheet and job orders • producing C N C tape file 43 Computer-Aided Process Planning (CAPP) 4. Applications 4.1 Machined Cylindrical Metal Parts C o m p u t e r managed process planning (CMPP) is an advanced process planning system for machined cylindrical parts. While C M P P is a partial implementation of C A P P technology, the concept can be extended to other applications. C A P P typically addresses up to four process planning functions: sequence of operations, tolerance control, reference surface selection, and process plan output. C M P P is a generative system that addresses all four functions. It is unique due to the fact that the software and complete documentation are available to U.S. companies and government organizations at no cost. 4.1.1 Characteristics The characteristics of C M P P are: • G e n e r a t i v e - I t develops process plans rather than simply recording o p e r a t o r decisions. • Interactive-Interactive changes to the process plan can be made by a process planner, if desired. • Manufacturer I n d e p e n d e n t - I t is not dependent upon any o n e manufacturing process philosophy. It can be used by different manufacturers on different type parts. • Process Decision Models U s e d - U s e r dependent process decision models, written in an English-like language that is understood by the system, provide manufacturing logic for part families. • C o m p u t e r Part Modeling U s e d - A process plan is developed, operation by operation, to transition the raw material into a finished part based on a detailed description of the part and raw material. 4.1.2 Geometric Coverage C M P P concentrates o n the capability to deal with cylindrical surfaces and features. However, some noncylindrical features are also handled, but less completely than the basic cylindrical geometry. • Cylindrical S u r f a c e s - T h e system handles diameters, faces, tapers, and circular arcs • Cylindrical F e a t u r e s - F e a t u r e s such as grooves, recesses, reliefs, and notches are supported • Non-cylindrical F e a t u r e s - O t h e r features such as holes, gear teeth, windows, flat slots, and splines are also supported along with functions such as heat treating and surface finishing 45 Computer-Aided Process Planning (CAPP) 4.1.3 C M P P Major Components Functionally, C M P P uses four input sources and provides three outputs (Figure 16). The inputs are: part data, process decision models, machine tool data, and user interactive instructions. T h e system outputs a routing sheet, dimensioned operations sketches, and a tolerance chart. Machine jData Process Decision Models! Part Data" Computer Managed Process Planning (CMPP) Routing Sheets — • • Tolerance Chart Dimensioned Operation Sketches User Figure 16: Functional View of C M P P C M P P Software is composed of three major components: database system, part input system, and processing planning system (Figure 17). It performs four functions: generates a summary of operation; selects surfaces for dimensioning references, clamping and locating; calculates machining dimensions and tolerances; and outputs process documentation. The database and part input systems are not strictly process planning components but are necessary in order to plan a process. T h e components of and functions accomplished by the software are described in the following paragraphs. Figure 17: Major CMPP Components Database System The database system provides the manufacturer independence for CMPP. It builds local files of user dependent manufacturing logic and parameters. As shown in Figure 18, it consists of four files. • T h e Manufacturing Vocabulary File lists the local manufacturing vocabulary terms that may be used to process decision models. It is composed of the English-like description of each term. 46 Computer-Aided Process Planning (CAPP) • T h e Machine Tool File lists the machine classes and individual machine tools within each machine class of the user's workshop. • T h e Cut Parameter File contains data on the machine tools' metal removal and tolerance holding capabilities. • T h e Process Decision Model File contains local manufacturing logic. A process decision model states manufacturing rationale for determining the sequence of operations of a part. A n English-like, problem oriented Computer Process Planning Language ( C O P P L ) is used to define manufacturing practice for families of parts. However, the system does not require a specific G T coding system for part families. C O P P L process decision models are compiled into CMPP-executable form, stored in the database, and executed for individual parts during process planning. T h e C O P P L language uses an open-ended vocabulary of manufacturing terms. This vocabulary can be extended to include terms required by a particular user. Each vocabulary term has a "definition" which is understood by the process planning system. Define Vocabulary Compile Process Decision Models ^ PROCESS PLANNING Build Machine Tool File SYSTEM ι • Build Cut Parameter File Figure 18: Data Base System Part Input System T h e part input system (Figure 19) receives part data and constructs computer models for use in process planning. Part data input is a three-step process: initial part data collection, part data editing (data can be completed or changed), and generation of an internal C M P P part model from edited part data. Each model is a detailed description of a part and the raw material from which it is fabricated. T h e model describes each surface, feature, dimension, tolerance, and notes. T h e system is designed for interactive part data input, so that it can be used as a standalone system. However, it can also be interfaced with a C A D engineering database for greater effectiveness. 47 Computer-Aided Process Planning (CAPP) Collect Part Data Edit Part Data Build Part Model PROCESS -PLANNING SYSTEM Figure 19: Part Input System Process Planning System The process planning system (Figure 20) applies the manufacturing practices and resources defined in C M P P to a part to produce a process plan. The four technical functions are a summary of operations, reference surfaces, stock removal analysis, and process data. Generate Summary of Operations 1 Select Reference Surfaces Calculate Dimensions & Tolerances Output Process Data Figure 20: Process Planning System 48 Computer-Aided Process P l a n n i n g (CAPP) • A Summary of Operations is generated by executing a process decision model. T h e model determines the operations in sequence. Operation description, machine type, setup orientation, and list of cut surfaces are generated for each operation. A summary of operations can also be produced without a process decision model by interactively specifying each operation. T h e summary of operations is presented in a matrix format so that each horizontal line describes the operation, machine, and surfaces/features affected. Each vertical line designates the manufacturing steps which produce an individual surface or feature of the part. • Reference Surfaces are determined after the summary of operations has been generated. Surface selection for dimension reference, clamping, and locating for each machining operation is provided by execution of decision models or interactively. • Dimension, Tolerance, and Stock Removal Analysis is performed as the third step. An initial starting tolerance is assigned to each cut and tolerance accumulations are calculated. When necessary, these tolerances are selectively tightened to meet tolerance requirements. This adjustment is made by a sophisticated algorithm with the objective of optimizing tolerance capabilities of shop equipment. Toleranced stock removals are then calculated, based on raw material tolerances, machining tolerances, and stock removal capabilities of machine tools. Potential problems of insufficient stock removal are diagnosed and flagged for interactive resolution during this step. Finally, nominal stock removals are used to c o m p u t e nominal machining dimensions for each cut. • Process Data, the final output, includes a printed routing sheet or sequence of operations and operation sketches, and tabular tolerance charts. T h e sketch is a not-to-scale drawing of the workpiece with cut surfaces highlighted by heavy lines. Surfaces not yet formed are shown as dashed lines inside the workpiece outline. Toleranced machining dimensions for cuts in the operation are included on the sketch. 4.1.4 CAD/CMPP/CAM Integration Although designed to o p e r a t e as a standalone system, C M P P achieves maximum advantage when integrated into a C A D / C A M system. Use of part data from an engineering database; o u t p u t of process documents to a graphics system; and use of o u t p u t data for tool design, N/C programming, and other manufacturing services maximizes its usefulness. Thus, C M P P offers benefits in process planning, other manufacturing services, and t h e shop. Case studies and estimates indicate cost savings of 25-45 percent on process planning labor depending on the extent of use. Savings in other manufacturing services depend on the extent of use of the o u t p u t data. Savings on the shop floor are the result of improved and standardized process plans. Although the percentage of savings is greater for process planning, total dollar savings may be greater in the other areas, due to their greater cost. Savings are especially significant when complex parts fabricated from expensive materials are produced. 4.1.5 Field Testing a n d Implementation C M P P was field tested by three aerospace companies during the development process. Benefits verified in process planning include labor savings, lead time reduction, and surge capability. Savings of 25-45 percent are indicated, depending on whether it is implemented during low rate production or at full-scale production. This is consistent with surveys and studies of expected savings. N o experience data is yet available on savings from manufacturing services on the shop floor. T h e survey and studies indicate that full integration would result in substantial dollar savings, although the percentage would be smaller than for the process planning function. 49 Computer-Aided Process Planning (CAPP) Implementation of C M P P requires the commitment of top management, since it involves a radical change in the way of doing business and a significant investment in funds to develop the local database. C M P P users have found useful applications for portions of the technology, such as the tolerance module. Partial implementation in a variety of industries has been reported, including jet engine, control system components, gears, bearings, electric motors, petroleum, cutting tools, and electro-optical components. Initially, C M P P was demonstrated o n IBM and U N I V A C computers with Tectronix terminal and CalComp plotter. U s e of C M P P on any o t h e r equipment required modification of the software. T h e capability to interface with C A D was demonstrated, but was not part of the delivered software; however, ease of implementation has been and is being addressed. A contract was awarded to modify C M P P for application on Digital E q u i p m e n t Corporation's ( D E C ) V A X computers. Additionally, neutral input and output interface will be implemented by the use of DI3000 graphics software. These enhancements increase the transportability and reduce the implementation cost of C M P P . 4.2 Printed Circuit Board Assembly CAPP systems today are either highly domain specific; i.e., deal with a specific technology or processes, or are shells that become highly domain specific when implemented. This reflects the underlying theory is inadequate to unify, in a practical manner, the diverse environments manufacturing. Until this theory is better developed, C A P P systems will remain very domain set of the fact that that exist in specific. Interest in C A P P is also shown by the increasing number of commercially marketed software packages that are designed to assist in manufacturing planning. In addition to helping construct a process plan, some of these systems help in the production of shop aids, postprocessor instruction sets, and test routines. Moreover, several firms in military electronics have constructed rather elaborate C A P P systems to meet, among o t h e r things, a need for exhaustive, traceable documentation of the process used on each and every copy of a product produced. Indeed, any firm producing electronics which must be supported in the field over a period of years, needs to know the production process that was used to manufacture individual units so that decisions dealing with recall and modification can be made in the most economical manner. 4.2.1 The P/CB Assembly Task T h e specific planning task that occurs in the manufacture of electronic equipment concerns the attachment of components to a P/CB. It is also referred to as P/CB assembly or "populating the board." In comparison to metalworking and many other manufacturing processes, circuit board assembly is a relatively simple process. O n e begins with a bare P/CB and attaches to it various types of mechanical and electrical components and subassemblies. While some components are mechanically attached, most are first temporarily affixed to the board by inserting the leads of the components into holes drilled in the board or by the use of adhesives. T h e attachment is made permanent by a soldering operation that can take various forms depending on the type of component that is to be attached. T h e soldering operation must also provide the connections required to make the circuits o n the board electrically functional. The bare P/CB can be viewed as a component itself. T h e process of attachment can b e accomplished manually, by robot, or by a variety of auto-insertion machines that are dedicated to the insertion of particular types of components. T h e first step in the planning task is the selection of an appropriate attachment process. This selection is determined by the geometry of the part to be attached, the proximity of the part to other parts on the board, and a variety of thermal, chemical, electrical, and material handling considerations. While machine attachment generally 50 Computer-Aided Process Planning (CAPP) produces higher yields than manual attachment, not all components can be economically attached by machine. Moreover, the economics of production require that some components be attached manually, even though they feasibly could be auto-inserted or inserted by robot. In short, the process starts with a bare P/CB and a collection of electrical and mechanical components. T h e problem is to specify the manner in which each component will be attached to the circuit board. Order of Operations T h e general flow of work is shown in Figure 21 and proceeds as follows: TEMPORARY Clinching Wax Manual Robotic Fixed Automata Manual Robotic Fixed Automata Pre-attachment Operations Attach Certain Mechanical Components ATTACHMENT Flux Adhesive / VCD Dip Machine Machine Axial Component Insertion Dip Insertion Manual/Robotic Insertions of other Components Perform] Leads Coat TYPICAL COMPONENTS Tin Solder Latches and other mechanical assemblies Labels Single in-line packages (SIPs) Dual in-line packages (DIPs) Radial components Axial components Cans "J" packs Leaderiess chip carriers (LCCs) Connectors Chips Rat packs Heat sinks Sockets LED and optical displays Post-solder Attachments Have Vapor Phase Laser LR. Hand Solder Test SUPPLIED Bulk or loose Reel Tube Magazine Carrier Figure 2 1 : Typical Steps in the Assembly of a P/CB (1) Preassembly operations on components for purposes of trimming and forming leads and other modifications to vendor supplied components. (2) Attachment of certain mechanical components by hand, robot, or fixed automation device. (3) Insertion of dual in-line packages (DIPs) by DIP auto-insertion machines. 51 Computer-Aided Process Planning (CAPP) (4) Insertion of axial components by variable center distance ( V C D ) machines. An axial component is o n e whose leads pass axially through the normally cylindrically shaped components; e.g., a typical resistor. (5) Manual or machine insertion of radial and other components as well as axials and DIPs, which for o n e reason or another could not be machine-inserted during earlier steps. A radial component is o n e whose leads extend radially out from a circular shaped body such as a disk capacitor. (6) Soldering. (7) A final hand or robotic assembly work center for attaching components that cannot be soldered or components that, because of size, shape, or location must be attached and soldered after all others have been attached and soldered. (8) Test. In the case of boards that involve surface mount devices (SMDs), the processing steps are slightly modified. Machines exist to automatically place special types of devices and there are other variations that may be employed. However, the process is basically the same as above. An interesting feature of process planning for electronic assembly is that the assembly operations for any particular board are an ordered subset of operations taken from a technologically, strictly ordered set of operations determined by the machines and processes available in the factory in which the board is being made. T h e fact that these operations are strictly ordered prevents the type of combinatorial explosion of possible operation sequences that can occur in other technologies. As noted above, the assembly planning process is o n e of deciding which components will be inserted at which step in the process or, equivalently, how they will be attached to the P/CB. U p to several hundred components may be attached to a board and, in a typical factory, hundreds of different types of board are produced each year. Depending on t h e philosophy of inventory management as well as the degree of automation, production is managed as a variation on o n e of the two basic approaches: • • Batch Production Flexible Line Production Batch Production Each work station or machine center on the production line is stocked specifically to meet the needs of a particular board and a relatively large lot size of a particular board is run. Steps are normally taken to balance t h e line to minimize idle time. When the lot is complete, the line is restocked and perhaps reconfigured to produce a new board. U n d e r batch production, a process plan is written for the particular board to be produced and the line is stocked and configured according to this production plan. In some factories, the factory floor is not configured as a production line. Rather, machines and workers are functionally grouped and batches of boards move through the factory floor from work center to work center in typical job-shop fashion. Materials are commonly issued in kits. 52 Computer-Aided Process Planning (CAPP) Flexible Line Production Taking into account the production schedule for some period of days or weeks into the future, all components needed for assembly of the scheduled boards are assigned to particular work stations. As the boards move through the line, they are automatically identified at each work station and the appropriate component attachments are completed. Each work center can be viewed as a flexible machine center (FMC) and nearly all the design and scheduling concepts applicable to F M C s can be applied. Lot sizes of o n e may b e accommodated and just-in-time concepts applied. Balancing the line becomes a very difficult task, but substantial a m o u n t s of idle time can be tolerated and justified on the basis of reduced setup time, inventory reduction, improved yields, and the other benefits of this type of manufacturing philosophy. This type of system is relatively intolerant of large a m o u n t s of component insertion by humans, because the work task can change from m o m e n t to moment. Communicating o p e r a t o r instructions and monitoring the operator's actions can be difficult. C R T monitors which present the o p e r a t o r instructions are used to reduce the confusion, as are a variety of other machines which mechanically identify, for the h u m a n operator, the c o m p o n e n t to b e inserted and its location o n the board. Under the flexible line production philosophy, a process plan must also exist for each board to be produced. However, t h e plan moves with the board, being accessed, say, by barcode, at each work station. Moreover, the configuration and stocking of the line must attempt to take into account the requirements of the collective process plans of all boards to be produced during the planning period. Ideally, o n e would have all of these process plans available in advance in order to attempt to optimize the location of components in relation to machine center capacity. However, such is rarely the case, and day-to-day modifications are made. 4.2.2 The P l a n n i n g T a s k In order to fully appreciate t h e planning task, it is helpful to understand how circuit boards are typically designed. Circuit Board Design Computer-aided circuit board design was o n e of the first and most successful applications of C A D . N o r t h e r n Telecom uses the Circuit Board Design System (CBDS) which was developed and is maintained by Bell N o r t h e r n Research and is marketed by IBM. T h e r e are a number of competing systems, but all the principal C A D systems for circuit board designs provide the designer with a highly automated environment for conceptualizing and analyzing the functional properties of the circuit, choosing actual components for implementing the circuit, and laying out the board. In addition, these systems produce a B O M and other files that are used by purchasing and manufacturing for planning, machine programming, and other activities. Figure 22 illustrates the relationship of process planning to C A D , C A E , and other elements of a CIM architecture for the production of electronic systems. F r o m a manufacturing standpoint, there are two data sets operating the C A D background which are of particular interest. O n e is the c o m p o n e n t database from which either the designer or the C A D system, itself, automatically selects the actual component to be used o n the board. This database may identify preferred components from a purchasing, electronic, and manufacturing standpoint. However, the designer may choose any c o m p o n e n t in t h e database, as required, to meet the design objectives and, while corporate approval may b e required, may add new components to this database. Because of the rapid changes in electronic technology and the need to remain competitive, the c o m p o n e n t database is constantly being changed. The other data set is in the form of sets of rules that either control or influence the selection of 53 Computer-Aided Process Planning (CAPP) NEW PRODUCT REQUIREMENTS DESIGN ENGINEERING (CAE) NEW PRODUCT REQUIREMENTS Electrical Silicon Mechanical Software Documentation Simulation Standard Cells Finite Element Solids Modeling DATA STORAGE & ADMINISTRATION MANUFACTURING ENGINEERING (CANENG) Product Line Up Project Support Networks Test Programs Costing Post Processing Process Dev. & Planning PRODUCTION Part Assembly System Assembly Functional Test Packaging Change Management Design Transfer Archive/Retrieve »DUCTION CONTROL PROC PRODUCTION PLANNING Schedules Capacity Planning Resources Material Detail Schedules Customer Configuration Τ Τ SHIPMENTS FORECASTS/ORDERS STS/i Figure 22: A CIM Architecture for the Production of Electronic Systems components and the manner in which they are placed on the board. These rules have two modes of application. In the first mode, they are automatically applied by the C A D system to place components and route the electrical paths on the board. In the second mode, these rules act as an advisor to the designer. If a rule is violated, the designer is issued a warning. Rules concerning manufacturing feasibility or desirability are normally included to o n e extent or the other in the rule base for board design. For example, most systems will include rules for the placement of manufacturing fixture locating holes on the board. By counting the numbers of and weighting the types of violations, designs can be scored from the standpoint of ease of manufacturing. While the usefulness of these scoring techniques is debatable, it illustrates the manner in which manufacturing concerns can be handled at the design level. Accordingly, the design of circuit boards is performed in an environment that, in theory, can b e very supportive of the application of criteria to simplify and reduce the cost of manufacturing. Planning Constraints T h e planning task would be relatively straightforward if there were a unique and stable relationship between a component to be attached and a method of attachment, and if the design of the board was in consonance with preferred manufacturing principles. Indeed, o n e could simply assign an operation code to all the c o m p o n e n t s in one's component database. Then, given a B O M , assign an operation to each component, thereby specifying which components should be available at each work station. However, the realities of circuit board assembly include the following: 54 Computer-Aided Process P l a n n i n g (CAPP) (1) Insertion feasibility is dependent on the specific configuration of the board; e.g., a D I P component which can be auto-inserted by a D I P machine on o n e board cannot b e inserted on another board, because there is insufficient clearance for the insertion tool. (2) Insertion feasibility and attractiveness is dependent on the specific machine and tooling. This is a problem when o n e or m o r e design groups are producing designs for boards that may be produced at several different facilities, each with different process capabilities. (3) T h e c o m p o n e n t is unavailable at the preferred insertion work station. Auto-insertion machine, manual, and robotic work stations can only accommodate a limited variety of inventory. While this problem can be overcome with sophisticated, material delivery systems, it will be m o r e economical for many factories to simply pre-stock each work station. Accordingly, a component that could be auto-inserted may end up being inserted by hand due to the physical limitation on the inventory that can be accommodated at the auto-insertion work station. (4) D u e to machine breakdowns, the preferred or normal insertion method is unavailable. If a machine is to be down for any significant period of time, the parts assigned to that machine must b e affixed elsewhere and this implies a change in operator and machine instruction sets. If feasible, an alternative to this is to reschedule production, although dynamic production scheduling can b e complex and m o r e trouble than it is worth. In any event, o n e needs to minimize t h e time needed to respond to unforeseen problems on the line, and this response will most typically be a change in process plan. Shop Aids T h e overall planning process involves the development of the shop aids necessary to support these operations. Shop aids include the instruction sets that drive automatic machines, test programs for operating automatic test equipment, and operator instructions that must be issued to the floor in order to execute the assembly of the board. In general, these may be grouped into two broad categories: (1) assembly instructions for h u m a n operators and (2) the instruction sets, in the form of punched tape or computer files, to o p e r a t e automatic equipment and to guide the process controllers that initiate and terminate automatic machine operation, identify products, and control material transfer devices. Most of these shop aids require a precise knowledge of c o m p o n e n t location and board geometry. This information is provided by the C A D system. However, a first-cut at assigning components to operations can be accomplished without regard to part or board geometry. U n d e r ideal design conditions, the insertion methods normally associated with the components would all b e feasible and preferable. This preliminary assignment of components to operations can produce a tentative list of the operations that will produce the board and a list of parts by each operation. These latter lists, coupled with component location and board geometry are what is needed by the persons and computer programs which produce the shop aids. Computer postprocessors which produce the instruction set for the operation of auto-insertion equipment are of particular interest. T h e reason for this is that, while the feasibility of part insertion due to specific board geometries can be addressed at a preliminary planning level, it must be addressed at the postprocessor level. It is simpler to let the C A P P initially deliver a candidate list of parts for insertion to the postprocessor, and let the postprocessor determine what it can insert while it is determining how to sequence t h e insertions. If a part cannot be auto-inserted because of the specific joint configuration of the component, board or tooling, the part is returned to C A P P for reassignment to another operation. T h e order of choices is normally, first, auto-insertion, second, robotic, and lastly, manual. This process is shown in Figure 23 and implies that the postprocessor must be able to identify insertion feasibility as well as develop the sequence of c o m p o n e n t insertions and the path of the insertion tool. This does not significantly complicate the algorithms used by the postprocessor. 55 Computer-Aided Process Planning (CAPP) Other Shop Aids Component by Operation Machine •Instruction File Figure 23: The CAPP-Postprocessor Relationship Accordingly, CAPP is responsible for producing an initial process plan and for dealing with those initial proposals which were identified as infeasible or unattractive actions by the postprocessors. 4.2.3 P a r t s Numbering and Coding T h e introduction of a new parts numbering or coding system is normally an expensive and time-consuming task, whose cost is commonly very much underestimated. Accordingly, it should be approached cautiously. C A P P systems seem to imply the need for a part numbering or classification system which provides a oneto-one link between a part and its assembly method. In turn, this means that each new component that is added to the component database must be coded or otherwise linked to an assembly method. Maintenance of such a system is expensive if the component database is constantly changing, and potentially confusing if different manufacturing facilities view the assembly of the c o m p o n e n t in different ways. At Northern Telecom, the C A D file produces a parts list for the board. It contains a descriptive engineering code, a nondescriptive specific part number, and a consistent, short English language description of each part. There is enough information contained in the engineering code and the English phrase to infer a great deal about the way in which each component can be attached. This happens even though neither the engineering code nor the English description of the part were originally designed to designate an assembly method. That is, it is possible to infer an assembly method from existing part descriptions, and it is not necessary to create and maintain a new coding system. This is accomplished by parsing the engineering code and the English phrase for key words and groups of characters in the context of the board that is under consideration. Alternatively, the parsing routing could b e used to automatically create an assembly code file by component. This has not been necessary and would produce n o benefits from the standpoint of the CAPP system. 56 Computer-Aided Process Planning (CAPP) 4.2.4 CAPP Aids CAD T h e process plan that is produced at Northern Telecom lists the operations that are involved in the production of a board and shows an estimate of the direct labor and/or process time associated with each of these operations. W e are currently interested in assessing the degree to which making this cost information available to the designer will influence the degree to which manufacturing cost and yield are considered by t h e designer. This design aid could only work with a generative C A P P . 4.2.5 P/CB Production • A C A P P system reduces the labor cost of preparing a process plan. By itself, this is probably not a strong argument for C A P P in electronic assembly planning. T h e a m o u n t of time spent o n process planning, particularly initial process planning, is relatively low in comparison to other technologies. • A C A P P system reduces the training costs of new process planners and mitigates the impact of losing key personnel. It is not clear that a generative C A P P system for electronic assembly can serve as an effective training aid for new process planners. A variant system is better at this. C A P P may be useful when it comes to acquainting new designers of electronic boards with manufacturing issues. By pointing out costs, the C A P P system might well be an effective device for increasing designer sensitivity to these issues. • A C A P P system improves the consistency with which plans are produced. While this is true, the issue of consistency is not of as much concern as in other technologies. In part, the reason for this is that there are not the processing options available in electronic assembly that exist in metalworking. However, the process of structuring C A P P uncovers misconceptions regarding assembly feasibility. These misconceptions are uncovered as o n e assembles the knowledge-base and rules for its application. T h e elimination of these misconceptions will improve productivity. T o some extent, this is the result of pursuing a generative rather than variant strategy. • A C A P P system provides the response time necessary to run a highly a u t o m a t e d CIM system. As electronic assembly moves toward fewer manual insertions, the ability to rapidly change the process plan for a board under a flexible machine center philosophy becomes increasingly important. U n d e r a CIM architecture, there are compelling arguments for a generative CAPP system, which can o p e r a t e quickly and with as little human intervention as possible. • A C A P P system provides a useful tool to improve the manufacturability of circuit board designs. T h e extent to which this is true depends on the organizational and product development philosophies of the firm. In firms which already assign a high priority to ease of manufacturing, attaching the C A P P to the C A D system may not be useful. F o r others, it may b e useful to provide the designer with a C A P P system which gives the designer immediate access to an estimated conversion cost and explains the reason for the conversion cost by showing the cost implications of the design on an operation-by-operation basis. C A P P should b e designed to allow the process planner maximum control over its operation, and alternatives should b e thoroughly explored before introducing new parts coding or classification schemes. They may not be necessary. 57 Computer-Aided Process Planning (CAPP) 4 3 The Route Generator In 1986, D E C was faced with how to integrate M R P II shop floor control, automated material handling, factory data collection at the shop/center hierarchical level, and several cells at the cell/line hierarchical level. With s o m e of these systems having already been implemented, and several others on the drawing board, each system was designed independently of the other. Consequently, each system would have addressed its routing and process data needs independently. If this had been allowed to continue, each system would have had to be implemented alone, in the sense of manually maintaining its own private copy of the r o u t e and process data. T h e problem with standalone implementations that require much of the same data in c o m m o n is version control. R o u t e maintenance and version control arises when route-sensitive systems are implemented in t h e same environment but are not designed for integration. For example, when there is a need to update t h e r o u t e because of an engineering change order, to change the standard route because of a sudden downward shift in quality levels, or to modify the process to take advantage of new technology/ideas, making a change to the manufacturing process can easily spill over into requiring modifications to process definitions on other systems. Thus, redundant, manually maintained route data is undesirable. 4.3.1 T h e G e n e r a t o r T o resolve this problem, instead of requiring multiple, redundant data maintenance activities to be made on each system, the R o u t e G e n e r a t o r ( R G ) was developed as a single point of definition and control for the generation and revision management of routes and related process planning data for all route-sensitive systems. O n e temptation to overcome was to try to build upon the existing process definition data structures within the M R P II system. However, it was soon realized that the M R P architecture was incompatible, that it did not support hierarchical routings, and that, as a business system, it did not support process engineering activities; e.g., it assumes a straight line route flow. In most process flow charts, there are several branch points that a process can take dependent upon any number of constraints. T h e R G is an interactive, graphical, video terminal-based work flow generator application compliant with the architecture previously described. It is used to converge the routing needs of each of these systems into lone logical master route. O n c e defined, routing data is derived from this master route for any system. Because there is o n e point of control for routing data, the integrity of routes between systems should b e insured. T h e benefit of the R G is its ability to capture and communicate the various route requirements for each of the factory control systems. Each process control system accesses its route data by making B O P data management interface calls. Because this data comes from the same source, routing data is consistent across all systems and is at the proper revision level. Multiple revisions of a r o u t e (process) can co-exist for the same part number at the same time in the manufacturing site. Additionally, the preferred process to be used to build and test a part can change from time to time. Therefore, the R G allows for revision control of both part and process. This allows factory systems to phase-in new processes and parts without losing visibility into the requirements for a specific instance of a part and process. Systems that use the R G are truly integrated and have a common reference system ( B O P ) that can be used to manage any number of system views from same route definition. 4.3.2 Controller Views T h r e e process control systems, each of which require a route, are referred to as controllers because of their ability to control s o m e aspect of the manufacturing system. Each of these systems potentially requires 58 Computer-Aided Process Planning (CAPP) more detail about the r o u t e flow, and each has different views and data needs of the same route. They are: • Manufacturing Resource Planning ( M R P II) • W o r k Flow Control ( W F C ) and Material Handling • Factory Data Collection ( F D C ) Manufacturing Resource Planning M R P II requires the most abstract view of the route. M R P IPs view is not for real time control purposes and, therefore, does not require extensive route control data. Instead, M R P II is interested in financial and capacity planning data related to the route. T h e data required is used by M R P II for capacity determination and costing. W o r k Flow Controller The W F C takes a material handling point of view, based on a "material drop-off perspective. A W F C needs route data so that it can match route setup requirements to current resource capabilities and machine setup states to achieve optimal work flow. Factory Data Collection F D C potentially requires the most detailed route flow view. T h e F D C route flow view can include standard operations, alternate operations, and various alternate rework operations irrespective of the way an associated material handling system may be configured. Alternate rework operations may be determined by the degree of seriousness of a failure o r some o t h e r significant set of parameters that are used to determine what operation is to be performed. 4.3.3 Combined R o u t e Flow T h e R G permits the creation of a single, user-defined logical route, providing multiple views for factory systems in need of r o u t e flow data, yet maintains the continuity of o n e logical route. Below is an example of how multiple r o u t e views can be derived from o n e logical r o u t e flow. T h e example begins by listing a product r o u t e flow, without giving consideration as to how the process steps in the route are eventually executed by the various process control system controllers. T h e example route flow shown in Figure 24 is based strictly on the product requirements and the logical activities required to build and test the product. T h e sample r o u t e list is then translated into a logical process flow, as shown in Figure 25. It is at this time that the engineer would define the route flow in a way as depicted in Figure 25. T h e master r o u t e flow is then analyzed with respect to the needs of each route-sensitive system. In effect, t h e generator for each controller selects certain process steps and ignores others, creating different views of the same master route. 59 Computer-Aided Process Planning (CAPP) M R P II View of the Master R o u t e T h e Process Steps required for M R P are depicted in Figure 26. The normal cycle time for the initial assembly process is short enough to not warrant detailed tracking by M R P II. H e n c e only steps 1, 5, 6, 9 and 12 are "seen" by M R P . T h e M R P II view is: • • • • • Process Process Process Process Process Step Step Step Step Step 1: Assembly Step 1 5: Assembly Step 5 6: Test Step 1 9: Test Step 2 12: Final Touch-up Process Step 1 2 3 4 5 6 7 8 9 10 11 12 Description Assembly Step 1 Assembly Step 2 Assembly Step 3 Assembly Step 4 Assembly Step 5 Test Step 1 Rework Test Step 1 Retest Step 1 Test Step 2 Rework Test Step 2 Retest Step 2 Final Touchup Figure 24: Sample Product R o u t e List pass pass Figure 25: Master R o u t e Flow 60 Computer-Aided Process Planning (CAPP) 1 >-5->6 pass >9 >M2 Figure 26: M R P View of the Master R o u t e T h e activities that M R P II does not see are Process Steps 2, 3, 4, 7, 8, 10 and 11. W F C View of the Master R o u t e T h e W F C view is a different subset of the master route. The W F C view is affected by the physical layout of the factory floor and t h e material drop-off points at each of the work stations. Assuming that Assembly Step 4 and Assembly Step 5 share the same physical material drop-off point, the W F C view requires Assembly Steps 1 through 4 but not Assembly Step 5 (see Figure 27). T o the W F C , the next standard operation after Assembly Step 4 is Test Step 1 (Process Step 6). 1 >2 >3->>4— pass ^ 6 >M2 Figure 27: W F C View of the Master R o u t e T h e W F C view is: • • • • • • • Process Process Process Process Process Process Process Step Step Step Step Step Step Step 1: Assembly Step 1 2: Assembly Step 2 3: Assembly Step 3 4: Assembly Step 4 6: Test Step 1 9: Test Step 2 12: Final Touch-up N o t e in this W F C example that Rework and Retest are performed internally to each test step so that the W F C never needs to know about Rework and Retest operations. Of course, if the W F C did need to know about rework loops, then Rework and Retest Steps could be identified as part of its view. F D C View of the Master R o u t e FDC's view is yet another, m o r e detailed view of the route master flow. F D C needs to view any physical drop-off steps required by a W F C , as well as Rework and Retest Steps. N o t e , however, that the F D C view does not include Assembly Steps 2 and 3 (see Figure 28). The assumption is that F D C is not involved in these two assembly steps, but a W F C must still deliver the product to these steps. 61 Computer-Aided Process Planning (CAPP) F D C s view requires all the Process Steps except Steps 2 and 3: • • • • • • • • • • Process Process Process Process Process Process Process Process Process Process Step Step Step Step Step Step Step Step Step Step 1: Assembly Step 1 4: Assembly Step 4 5: Assembly Step 5 6: Test Step 1 7: Rework Test Step 1 8: Retest Step 1 9: Test Step 2 10: Rework Test Step 2 11: Retest Step 2 12: Final Touch-up N o t e that F D C needs to know that the Rework Step is not the standard Process Step but, rather, is an allowed alternative only if a failure path is required at the completion of either test step. T h e R G is table driven such that an engineer wishing to expand the number of system views simply registers the new view name in a master table along with a pointer that maps to the view-specific subroutine that addresses the data requirements of that system view. Later, when that view is indicated to the R G , the customized view-specific subroutine is executed, prompting the user for the view-specific data. Still later, once the route definition has been approved and released, the process controllers make the appropriate B O P data management interface calls referencing the same view name as originally input by the engineer. pass pass Figure 28: F D C View of the R o u t e Master R o w T h e B O P data management interface calls are very powerful. Generally, controllers have two modes in which they operate. O n e mode is to pass the status outcome of a previous step and allow the BOP evaluation routines to evaluate the proper next step based upon the view being requested. Referring to Figure 28, in a simple case where Step 6 results in a failure, then the logical next Step is a rework step, Step 7. W h e n F D C requests the next step following Step 6 and supplies a "fail" status, the BOP evaluates this request and returns the rework step, Step 7. If, however, the next step request was made by WFC, the rework Step 7 is not seen and the correct response returned is Step 9. This is because it was determined that the W F C view would not have to be concerned with whether or not a failure was detected at Step 6. T h e o t h e r m o d e is for the controller to request all of the next steps for a given step that meet the view specified. In this mode, the controller must be able to evaluate the alternate next steps and make the decision on its own as to which step will be selected. 62 Computer-Aided Process Planning (CAPP) 4.3.4 Process M a s t e r While the R G supports the traditional file structure of separate routes for each product, it additionally offers a G T capability by combining the otherwise separate routes into o n e superset route. This master r o u t e superset, called the Process Master (PM), is defined as a generic route flow template containing any number of Process Steps (PSs) which can be applied to define the r o u t e of a family of similar parts or products. T h e P M is well suited to support manufacturing operations of closely related product lines. In fact, the closer t h e alignment of process similarities within the PM, the m o r e efficient the P M becomes. For example, a D E C video terminal model number VT240 is a m o n o c h r o m e terminal. A VT241 is identical to the VT240, but has color capability added. T h e VT240 and VT241 terminals would be considered excellent candidates for inclusion in the same family to share a single P M because of their similar route flows. A P M can have an unlimited number of PSs to support the members in the family of parts, but the most efficient use of the P M occurs with the highest percentage of like PSs. As a point of comparison, the VT100 terminal deviates significantly from the VT240 series and would be considered less likely to effectively share the same P M . Since the VT100 has fewer PSs in common with the VT240-241 process master, combining these two dissimilar processes into o n e process master would not be desirable. T h e r e are tremendous advantages that can b e derived from combining similar processes into o n e process master, such as a reduction in data maintenance and redundancy, and the ability to apply an integrated G T approach to manufacturing process families. R o u t e View of the P M Using the previous example of the P M family, the VT240 and VT241 would each require a R o u t e View ( R V ) of the PM. Both of the R V s would be mapped to the same PM. Almost all of the PSs would be mapped for each R V due to the similarities between the build processes. Consequently, two part process views have roughly 80 percent of the process steps in common. However, if the VT100 were added as a member to the same PM family, not only would fewer of the existing PSs b e usable by the VT100 route, but additional non-sharable steps would need to be added. It would simply b e m o r e efficient to create a separate P M for a family of VTlOOs, rather than minimize the utility of the existing family. Conceptual A p p r o a c h T h e R G uses t h e P M to define a superset of valid sequential activities for use as a template for each of the individual parts within the P M flow. This P M template consists of the following components: • • • PM Header Process Steps Controller View The components of the P M must be created in the order listed above. T h e P M H e a d e r data is created first and identifies the historical information about its creation. T h e PSs then define the superset route flow for a specific PM. T h e controller view points to the specific process steps that are viewed by each controller. 63 Computer-Aided Process Planning (CAPP) T h e R V points or "maps" to a subset of the PM by selecting the appropriate PSs from the PM template. T h e R V consists of: • • • Part Master R o u t e View Header Controller Specific T h e R V Header is attached to the PSs of the same PM. T h e R V inherits all data associated with each PS, as well as any controller views associated to those PSs. Controller specific data is attached to the R V Header. T h e P M and the R V The relationship of the two data structures, the PM and the R V , is depicted in Figure 29. PROCESS MASTER PART PROCESS VIEW Process Master Header Part Master y y Route View Header Process Steps «<• y y Controller Specific Views Controller Views Figure 29: Data Relationship Between the Process Master and the R o u t e View 64 Computer-Aided Process Planning (CAPP) 5. Planning Facilities 5.1 The Process Industry Solution Center Time is the essence of manufacturing competitiveness. For this reason, the focus of t h e Process Industry Solution Center, a joint partnership between Hewlett-Packard and Fisher Controls, is enterprise-wide information integration; from product development through production, distribution, marketing, and sales. T h e Process Industry Solution Center is a meeting and demonstration facility developed around the operations of a simulated chemicals manufacturer. T h e Process Industry Solution Center debuted in 1988 as the C I M Technology Center. T h e Goal of the Process Industry Solution Center is to illustrate to customers as well as help them: • Balance costs, quality, features and availability • Shorten development, manufacturing and delivery cycles The Process Industry Solution Center believes C o m p u t e r Integrated Manufacturing (CIM) must build on the manufacturing management foundation already established within an organization. T h e basic strategy of the Process Industry Solution Center is to, as painlessly as possible, migrate to an open, standards-based integrated environment by building on and adding to existing applications. Because no two process manufacturers are alike, the Center doesn't p r o m o t e "one-size-fits-all" solutions. Its approach is to analyze operations and information flows, identify strategic opportunities for improvement, and quantify the benefits these opportunities offer. 5.1.1 The M a r k e t Each of t h e industries addressed by the Solution Center rely on a combination of batch-oriented processes and discrete manufacturing techniques. Market conditions are forcing these sectors to examine CIM as a business solution. F o r instance, the rate of new drug introductions in pharmaceuticals requires producers to become m o r e flexible, while tightening F D A and E P A regulations, and consumer expectations, are pressing them to improve quality. These demands are boosting the market for CIM. According to Frost & Sullivan, the market for integrated batch process control systems serving these sectors will exceed $1 billion by 1994, a 100 percent increase over five years. Over half of the growth will come from foods, fine chemicals, and pharmaceuticals. 5.1.2 T h e Solution C e n t e r Demo T h e first part of the presentation focuses in the importance of time-based competitiveness: the ability to get product to market sooner. T h e next part homes in on modeling both the business and manufacturing process, with the goal of simplification. Various scenarios and control methodologies are explored, informed by the collective experience of the Solution Center's staff in process applications and CIM. T h e final part of the program focuses on implementation. H e r e , a 30-minute comprehensive 65 Computer-Aided Process P l a n n i n g (CAPP) demonstration shows how a batch-oriented process m a n u f a c t u r e r - t h e Mighty Fine Chemical Co., a producer of specialty chemicals-would use integrated O p e n Systems to cut product lead times. As the d e m o begins, the Mighty Fine plant is already running with a predetermined master schedule and a short-term production schedule, extending four or five days. T h e process itself mixes and reacts up to four ingredients into four different end-products, simulating the type of flexibility characteristic of today's specialty chemical producers. In t h e d e m o , the M R P II system ( C I M P R O , from Datalogix International) has already downloaded the current work order backlog for the week, which the finite scheduling system (Schedulex from Numetrix, Ltd.) converts into round-the-clock daily schedules. Those schedules are then downloaded to the supervisory control system. A choice of the following supervisory systems are featured: • Monitrol from Hilco Technologies • PMIS from Bradley-Ward • CIM 21 from Industrial Systems Inc. Within the plant, the supervisory system sends orders to the Fisher P R O V O X p l u s plant floor distributed control system (DCS) to make various batches, o n e at a time. T h e D C S in turn is linked to a Fisher Controls U N I V O X line operator station. O n c e the D C S reports that the batch has been made, the supervisory system schedules a quality control test from the plant production laboratory by notifying the laboratory information management system (LIMS), based on HP's L a b / U X software. T h e plant is also equipped with a maintenance management system (MCS-II from Diagonal Data) that schedules and tracks routine and emergency repairs. Provision is made for the operation of a PLC-controlled line for packaging the finished chemicals into drums. T o demonstrate the flexibility of the system, a breakdown is staged on the production line. T h e main reactor, a 2,000-gallon tank, suddenly malfunctions, requiring a rerouting of production to a larger, less efficient 3,000-gallon tank; the issuance of a repair work order; the scheduling of extra production lab tests to determine the quality of the batches produced both before and after the breakdown; and the adjustment of plant schedules to account for the less efficient production capability. T h e adjustments start with notification by the D C S to the supervisory system. This in turn prompts the issuance of repair work orders from the MCS-II maintenance management system, extra quality tests from the H P L a b / U X LIMS system, and a rerunning of production schedules on the finite scheduler. 5.1.3 New Tactical Demo Added T h e Tenafly center has long provided an executive-level system demonstration aimed at corporate management and M I S staff. It has now added a hands-on tactical version for technical users, such as plant engineers, to actually operate all systems except for the M R P master scheduling. F o r instance, in the H P L a b / U X LIMS system, the o p e r a t o r can override the system and perform historical trending o r sample tracking, adjust automatic specification testing procedures, or change automatic test schedules for out-of-spec samples. In the DCS, he or she can change both the formulations, process trains (routings), and the operating parameters of the reactor tanks. T h e computer systems are all H P 9000 workstations running H P / U X . They are networked via TCP/IP Ethernet. Applications are integrated using the Network File System (NFS) and H P Sockets, an applications and communications interface. T h e supervisory and plant management functions are managed 66 Computer-Aided Process Planning (CAPP) by the H P computers. They link to the Fisher control environment via Fisher's C H I P / U X (Computer Highway Interface Program/Unix). T h e changes to the Solution Center's p r o g r a m s - f r o m awareness to service and i m p l e m e n t a t i o n - a r e the result of customer input. For instance, though both H P and Control Associates will provide implementation and strategic consulting services, the Center will also work with third-party service partners designated by the customer. T h e demos at the Process Industry Solution Center will continue to evolve. According to process industry consultant Henry F a s t e n of H P , who organized much of the computer hardware and software portion of the demo, additional applications are being added, including M A N B A S E from M A I Manufacturing Systems Inc. T h e d e m o will in many cases offer choices of multiple packages for any given application, such as the three already offered-Monitrol, PMIS and CIM 2 1 - a t the supervisory level. Currently three days a month are reserved for visits by area companies to the Solution Center. A short list of previous visitors include American Cyanamid, Ciba Geigy, Hoescht Celanese, Hoffman LaRoche, Lederle Labs, Lever Brothers, and U n i o n Camp. An estimated $7 million in sales were attributed to the CIM Technology Center between 1988 and 1991, a level of success that will continue at its successor, the Process Industry Solution Center. 5.1.4 Conclusion Endeavors such as the P.I. Solution Center provide users, and vendors an opportunity to view an operational system-even if only hypothetical. This opportunity fosters further development and activity in CIM. This will, o n e hopes, provide for m o r e efficient manufacturing processes. 5.2 Honeywell's Integrated Manufacturing Facility Many manufacturers have already studied the feasibility of integrating building control with manufacturing control in a single, integrated system, in an effort to increase manufacturing competitiveness. Building-control functions usually include monitoring and controlling environmental applications in a facility, such as heating/ventilating/air conditioning ( H V A C ) , humidity, air pressure, lighting, fire alarm systems, utilities systems, plant security systems, safety interlocks, and energy management and conservation. Manufacturing control functions typically include monitoring and controlling such manufacturing applications as data acquisition, regulatory control functions, discrete/logic functions, batch functions, sequencing, event-initiated processing, analysis and report generation, historization, and interfacing to plant-wide communications systems. 5.2.1 Sample Configuration, Typical Manufacturing Facility T h e physical operations of typical manufacturing facilities usually include several areas, such as receiving, manufacturing, packaging, storage, and shipping. O t h e r physical operations in an industrial facility may include utilities (such as boilers, refrigeration units, and cogeneration units) and environmental process units (such as wastewater system, thermal oxidizers, and scrubbers). O t h e r manufacturing occurs in extremely sensitive, controlled environments, such as clean rooms and laboratories. Figure 30 shows all physical operations within a typical manufacturing facility (manufacturing process areas, utilities, and environmental process units). 67 Computer-Aided Process Planning (CAPP) Manufacturing Facility Laboratory a Receiving Area —• Mfg. Area —• —• Packaging Area Storage and Shipping V Clean Room Utilities: • Boilers • Refrigeration Units • Cogeneration Units Environmental Process Units: • Wastewater • Thermal Oxidizers • Scrubbers Laboratories/ Cleanrooms Figure 30: Physical Operations, Typical Manufacturing Facility Historically, building-control and manufacturing-control networks were designed separately, with overlapping capabilities. Since the two types of networks were seldom integrated, they almost always had: • • • • Different system architectures Incompatible hardware and software platforms Inconsistent data bases Lack of communication between manufacturing control and building control R e d u n d a n t , costly, cumbersome communication networks still exist in many industrial facilities. Figure 31 illustrates the traditional building control focus. Leading manufacturers are now gaining a competitive advantage by analyzing the interrelationships of their building control network, their manufacturing control network, and their business goals. The Honeywell approach to creating an Integrated Manufacturing Facility provides manufacturers with two types of plant-wide data communications: • Bidirectional horizontal communication-from the receiving dock to the shipping dock: this type of communication links work areas and utilities with the entire manufacturing process 68 Computer-Aided Process Planning (CAPP) Building Control Focus Manufacturing Facility Laboratory Receiving Area —• Mfg. Area ·. —» Packaging Area —» t Storage and Shipping Clean Room Utilities: Environmental Process Units: • Boilers • Refrigeration Units • Cogeneration Units • Wastewater • Thermal Oxidizers • Scrubbers Security Laboratories/ Cleanrooms Figure 3 1 : Traditional Building Control Focus • Bidirectional vertical communication-from sensors to the boardroom: plant floor information flows upward to the administrative/management network, and strategic/tactical information and operating strategies flow down to the shop floor Building and manufacturing information is made available to those who need it: operators, engineers, maintenance and support personnel, and plant management. These bidirectional horizontal and vertical communications capabilities are designed to help manufacturers make m o r e informed, timely decisions concerning their operations. 5.2.2 Challenges Facing Today's Manufacturing Facilities "Under-rooP industrial facilities that house all their manufacturing equipment in an enclosed environment face many control challenges. These challenges result from the lack of integration with a building control system and include: • • • Inconsistent product quality Inflexible control and automation systems Excessive downtime (often caused by a safety system fault) 69 Computer-Aided Process Planning (CAPP) • • Inefficient analysis and reporting capabilities from using separate systems Inconsistent measurement, control, and management of facilities costs-particularly utilities These shortcomings can be addressed individually, as follows: • Inconsistent Product Q u a l i t y - I n d o o r humidity, temperature, air pressure, and air quality must be measured and controlled. T h e internal building environment often has a direct impact on product quality and plant profitability in many manufacturing processes. • Inflexible Control and Automation S y s t e m s - M a n y manufacturers are unable to quickly and flexibly change the environmental conditions of their facility as quickly as they change recipes for product manufacturing. This inflexibility can adversely affect the quality of fine chemical products and o t h e r products that require close tolerances for temperature, humidity, and air pressure parameters during their production. • Excessive D o w n t i m e - T h e inability to integrate a safety shutdown alarm with an intelligent, staged shutdown of the process can cause excessive downtime and lost production time. • Separate Systems Performing Analysis and R e p o r t i n g - C o n v e n t i o n a l manufacturing facilities usually have analysis and reporting capabilities spread throughout the plant in independent systems. Multiple operating stations generate reports in different formats among various manufacturing areas. This approach makes it difficult to collect, analyze, and historize important events. In addition, operators must be trained to use a variety of systems, formats, and protocols. • Inconsistent M e a s u r e m e n t , Control, and M a n a g e m e n t of Facilities C o s t s - C o s t s of utilities and environmental units are difficult to measure and control. Many facilities previously had n o way to track these costs or allocate them back to product lines, plant areas, or departments. In addition, the production cycle and the physical facility may have peak energy needs that occur simultaneously, resulting in "high peak" energy billing and/or "brown out" conditions. Integrating the manufacturing process with utilities and environmental process units can help manufacturers: predict when peak energy usage will occur; schedule and allocate energy usage costs efficiently; and distribute utility resources (chilled water, steam, compressed air). 5.2.3 The Integrated Manufacturing Facility Solution Figure 32 illustrates an Integrated Manufacturing Facility Solution, as envisioned by Honeywell. T h e integration of a building control network with a manufacturing control network can offer users these competitive advantages: • Better Product Q u a l i t y - I n t e g r a t e d building and manufacturing control strategies help keep the internal facility environment consistent with the product being manufactured, resulting in products of consistent quality and m o r e productive workers • Flexible Manufacturing Capabilities-Integrating process control and building control allows manufacturers to respond to changing market demands with quicker setups and faster transitions between product runs 70 Computer-Aided Process Planning (CAPP) Plant Management Manufacturing Facility Manufacturing Control Focus Laboratory Receiving Area JZJL Mfg. Area * Packaging Area Storage and Shipping Process Control Clean Room Utilities: Security Environmental Process Units: • Boilers • Refrigeration Units • Cogeneration Units Γ Building Control Network Material Handling • Wastewater • Thermal Oxidizers • Scrubbers Laboratories/ Cleanrooms Packaging Control Manufacturing Control Network Figure 32: Integrated Manufacturing Facility • Increased Uptime, Reduced Product Costs, and Lower Scrap Rates--The Integrated Manufacturing Facility can warn production operators when environmental conditions are out of specification, or when a potential safety hazard exists before out-of-specification conditions or hazards occur; early alarming gives operators time to react to this information and correct it, before a manufacturing control problem occurs; if a shutdown situation should occur, early n n alarming can help stage a soft shutdown and eliminate costly startups • Coordinated Supervisory and Reporting Capabilities-The Integrated Manufacturing Facility integrates analysis of building control and manufacturing control, providing: • • • • • • Improved ability to comply with F D A and E P A regulations A convenient "single-window" for reporting and tracking History and trending analysis for both systems A consistent operator interface Reduced operator training Improved Management and Tracking of Facilities C o s t s - T h e Integrated Manufacturing Facility can measure, control, and manage facilities costs, providing-plant operations personnel with tighter tracking and billing of operational costs to appropriate manufacturing units, and optimized energy conservation 5.2.4 Manufacturing Applications Internal cultural and organizational issues can sometimes impede even the best integration efforts. Fortunately, many industries are moving quickly toward focusing on both building control and 71 Computer-Aided Process Planning (CAPP) manufacturing control. These manufacturers are looking beyond departmental barriers to create integration solutions for a variety of applications-from simple, single-loop applications to those that are extremely complex. Fast, fundamental benefits occur in industries that recognize the positive impact of control integration on the quality of products being manufactured. For example: • P h a r m a c e u t i c a l s - h a v e stringent ambient temperature, humidity, and pressure requirements in manufacturing, laboratory, and animal room environments. In addition, many F D A requirements now extend beyond the process system to include building control. Integration between building control and manufacturing control can help pharmaceuticals manufacturers control these processes m o r e tightly, thereby improving product quality and meeting increasingly strict federal regulatory requirements. • Food, Beverage a n d C o n s u m e r Goods-face increasingly complex formulations and processes. A need exists in these industries for information to b e shared between facility services and the process, so that a wider variety of products can be tightly controlled during the production process. Integration provides better data-sharing and communications throughout the facility, as well as consistently high-quality products. • Textile Fiber Industries and P a p e r Products C o n v e r s i o n - h a v e strict humidity and temperature control requirements. Integration of building control with manufacturing control is crucial to the consistent, cost-effective production of quality products. • Silicon Chip M a n u f a c t u r e r s - s i l i c o n chip manufacturers face sophisticated control needs within clean room and assembly areas. In addition, the building control system often: • Annunciates alarms for toxic material detection and provides safety warning systems for notification of occupants • Provides focused control solutions for many sensitive applications • Controls process waste Integration of building control with manufacturing control in this type of application helps manufacturers maintain quality and reduce scrap. 5.2.5 Approaches To Achieving An Integrated Manufacturing Facility T h e r e are two basic approaches to achieving an Integrated Manufacturing Facility. In the first approach, manufacturing process requirements drive the integration, and commercial building controllers are integrated into a higher-level industrial control system. In the second approach, building control system requirements drive the integration, and individual industrial control devices are integrated into a building control system. Manufacturing process requirements drive, e.g., photographic film manufacturing (see example below) and parenteral pharmaceuticals manufacturing; building control requirements drive, e.g., laboratories and cleanrooms. By way of example, the following paragraphs describe the use and benefits of the Integrated Manufacturing Facility in a photographic film production facility. In many manufacturing processes, such as photographic films, the building environment is a process parameter that directly impacts product quality and plant profitability. Examples of building environment 72 Computer-Aided Process Planning (CAPP) process parameters include the measurement and control of humidity, temperature, pressure, and air quality . As manufacturers a u t o m a t e their process control systems, they also study how to tightly integrate their building control systems with their process. These building control systems typically control heating, ventilating and air conditioning ( H V A C ) and are candidates for integration into a single system-again, known as an Integrated Manufacturing Facility. Because the building environment and process are highly interactive during photographic film production, a film plant needs to be completely enclosed, or "under roof." T h e product, be it in the form of raw material, in-process inventory, or finished goods, is sensitive to environmental conditions during both storage and processing. Because of the length of time film spends in various phases of storage, the storage environment is often critical to the quality of the final product. T h e processing phase is typically much shorter in duration and therefore is less dependent on environmental conditions. However, this phase is still important, because conversion takes place during processing. 5.2.6 Process Description Time of exposure to conditions of temperature, humidity, and light during storage is a critical quality parameter. Product-dependent variables during manufacturing include drying pressure, temperature, humidity, and air quality. H o w these variables are controlled can make the difference between an acceptable product and a rejected product. The process flow in a film plant begins with raw material storage of chemical compounds, such as paper and film backings, water solvents, resins, and active ingredients. T h e initial processing stage is called solution manufacturing. Next comes a coating process. After the coating phase, t h e film dries in a multi-zone continuous dryer. Pressure, temperature, and flow rate are critical to maintaining product quality during this drying process. T h e final phase of photographic film production is the converting and packaging process. Most of this process is performed in a "lights out" environment because of film's inherent sensitivity to light. Most film plants today have separate process control, building control, fire protection, and safety systems. This control approach can result in duplication of monitoring and operating costs and/or poor or nonexistent communications between systems. Several problems in a photographic film plant present opportunities for using an Integrated Manufacturing Facility. F o r example, pressure must be tightly controlled in the various storage and processing rooms, as well as within some of the process units, such as continuous dryers. Product quality often depends on very stable and consistent airflow patterns throughout the manufacturing process. In addition, separate safety shutdown systems are difficult to monitor and diagnose. Unnecessary shutdowns caused by faulty sensors or system errors usually take hours to investigate and recover from, often causing hours of lost production and product waste. Connecting these critical personnel safety systems into an integrated manufacturing facility system not only reduces waste and downtime but also speeds diagnostic maintenance and system recovery efforts. 5.2.7 Solutions and Benefits Plant personnel have repeatedly expressed a need for a system that provides alarms, history, trends, and reports to: prevent upsets leading to quality defects; satisfy regulatory, customer, and management requirements; and establish correlations between quality of products and environmental conditions. A system of this type would provide operators with quality data that displays storage parameters as well as 73 Computer-Aided Process Planning (CAPP) process parameters for any batch, lot, or roll. Potential benefits of building and manufacturing integration in a photographic film plant include improved quality-i.e., m o r e consistent production results from minimizing process variability-and waste reduction, increased uptime, and reduced training, spares, and support costs for various separate systems-i.e., an integrated system quickly identifies problems, establishes correlations, saves time, and eliminates finger-pointing between working groups or departments. This leads to m o r e productive use of personnel in solving plant problems, lowering costs, and improving overall manufacturing competitiveness. 5.2.8 Conclusion A n Integrated Industrial Facility can provide world class manufacturing benefits to manufacturers in a variety of industries. Honeywell claims to be the first control supplier to bring this integration technology to control customers on a worldwide basis. Integration of manufacturing control and building control can help manufacturers achieve and maintain: • • • • • • • • • • Consistent production of quality products Flexible manufacturing capabilities Increased u p t i m e Reduced product costs Lower scrap rates Tight, trackable compliance with F D A and E P A regulations Improved management and tracking of facilities costs Integrated plant-wide communications High levels of employee safety Energy reduction programs Honeywell believes it can help achieve and sustain the Manufacturing Competitiveness of its customers through the Integrated Manufacturing Facility. 74 Computer-Aided Process Planning (CAPP) Vendors T h e following list of products was taken from a 1989 research report by the staff of United Technologies Research Center, E. Hartford, CT. At that time there were 134 C A P P systems; however, the majority of these systems were either research systems, in-house systems, or systems developed in academe. Often the researchers failed to give full information regarding the systems developer or the systems, themselves. T h e systems listed below are those that are complete (not partial, quasi-, or n o n - C A P P systems), commercially available, and included sufficient information to be useful for manufacturing applications. Information on all systems studied by the researchers may be obtained by contacting CAM-I Inc., 1250 E. Copeland Rd., Suite 500, Arlington, T X 76011; 817/860-1654; F A X 817/275-6450 System: Vendor: AC/PLAN American Channels, Inc Lexington, M A • System: Vendor: Alphagraphics Birsch, Birn & Partners Ft. Lauderdale, F L • System: Vendor: A variant, G T oriented system; primary focus on classification and coding. Autoplan Metcut Research Associates, Inc. Cincinnati, O H • System: Vendor: A hybrid (variant/generative combination) system using graphics for interactive selection of resources. C A P P (CAM-I's A u t o m a t e d Process Planning) Computer-Aided Manufacturing International Arlington, T X • System: Vendor: O n e of t h e oldest and most commonly used variant process planning system; uses G T methods. CutPlan Metcut Research Associates, Inc. Cincinnati, O H • System: Vendor: A n enhanced variant planning system; uses keywords to pick o u t the most similar plan; uses search function to choose operations from other plans. GT-based, variant system. CutTech Metcut Research Associates, Inc. Cincinnati, O H 75 Computer-Aided Process Planning (CAPP) • System: Vendor: Expert machining operation detailing program; selects tools, cutting parameters (selection, speeds, feeds), tolerances, cost estimating. GECAPP-PLUS General Electric Albany, N Y • System: Vendor: Hybrid modular system; modules may be purchased separately; limited generative capability. IntelliCap CimTelligence Lexington, M A • System: Vendor: A variant system combining AI, G T , and relational database technologies. IntelliGen CimTelligence Lexington, M A • System: Vendor: A n AI, GT-based shell system designed to capture user's expertise. LOCAM Logan Associates Natick, M A • System: Vendor: A popular hybrid system; user logic stored using decision tables. MultiCAPP Organization for Industrial Research Bedford, M A • Variant system similar to CAM-I CAPP; besides G T code, retrieves plans by matrix, variance code matching (similar part families), and information within a plan. 76