- Email: [email protected]
The Enterprise-Web portal for life-cycle support
COMPUTER-AIDED DESIGN Computer-Aided Design 32 (2000) 85–96 www.elsevier.com/locate/cad
The Enterprise-Web portal for life-cycle support M. Rezayat* SDRC/University of Cincinnati, 2000 Eastman Drive, Milford, OH 45150, USA Accepted 2 September 1999
Abstract As we enter the new millennium, the approach to product development is evolving rapidly. Companies are in the process of creating a distributed design and manufacturing environment that enables integrated product, processes, and protocols development (IP3D). Certain strategies and some specific technologies are required to create such an environment. One such strategy is finding effective methods for communication and sharing of information, especially those related to design and manufacturing, throughout the entire enterprise and the supply chain. The technologies that support such a strategy must be able to deal with distributed environments and databases, must ensure reliability and security, and must be practical. Therefore, to accomplish this task efficiently, we need a roadmap that spells out the communication infrastructure (with a practical access interface) and defines the administration and management methodology. We suggest here that electronic access to design and manufacturing information within the extended enterprise must be Web-based because of its universal interface, open standards, ease of use, and ubiquity. To effectively deal with the distributed data, we recommend combining the distributed object standards (e.g. CORBA/DCOM) with the Web standards and protocols (e.g. Java, XML, IIOP) to create the Object Web. Finally, we propose that the Object Web must be combined with an enterprise’s information authoring and management systems (e.g. CAD, PDM, ERP) to create the Enterprise-Web (E-Web) portal, with the mission of providing the right information to the right person at the right time and in the right format anywhere within the extended enterprise. In this paper we identify a majority of the components needed to implement the E-Web and provide user scenarios (based on actual working prototypes) to demonstrate the effectiveness of such a system. Using the scenarios and the prototypes, we will show how E-Web can provide support for everyone associated with a product during its life cycle, thus creating a true IP3D environment. q 2000 Elsevier Science Ltd. All rights reserved. Keywords: Integrated product development and support; Process and application portal; Global information access; Supply-chain integration
1. Introduction Enterprise-Web portal will bind distributed environments and objects in the next millennium Integrated Product and Process Development (IPPD) has been in practice for some time in the major corporations. Global competition and distributed manufacturing environments will require that these corporations integrate people and computers into the total development process as well. This integration must take place through well-accepted human and electronic protocols. Therefore, we introduce the new concept of Integrated Product, Processes, and Protocols Development (IP3D), where the word “protocols” refers to both human and electronic protocols. One factor for this integration is that 50–80% of all components in products from Original Equipment Manufacturers (OEMs) are now fabricated by outside suppliers, and this trend is
* Tel.: 11-513-576-5981; fax: 11-513-576-2840. E-mail address: [email protected] (M. Rezayat).
expected to continue into the next century. Another factor is the need to access information throughout a product’s life cycle by everyone associated with its design, creation, sale, distribution, and maintenance. In this paper, we define the infrastructure for implementation of IP3D within the extended enterprise in order to support everyone that is associated with a particular product during its life cycle. In a second paper [1], we will discuss methods for bringing the enterprise knowledge to each individual as a decision-support aid for effective design and production. To remain competitive, OEMs must operate efficiently at the extended enterprise level. An extended enterprise comprises an OEM, its supply chain, subsidiaries, consultants, and partners affiliated with the life cycle of a particular family of products. By integrating the various components of an extended enterprise into the development process, OEMs can dramatically reduce the cycle time and cost, while improving quality and product variety. (In the remainder of this paper, “enterprise” represents the extended enterprise, unless indicated otherwise.) Some of
0010-4485/00/$ - see front matter q 2000 Elsevier Science Ltd. All rights reserved. PII: S0010-448 5(99)00092-5
86
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
the critical, and inter-linked, issues that must be addressed in this regard are: • Enterprise-wide access/sharing of information. • Enterprise-wide viewing/updating of information. • Enterprise-wide management of information resources.
and
A variety of commercial PIM (Product Information Management) and ERP (Enterprise Resource Planning) products are at the disposal of these companies to address the last issue [2]; but the first two issues are yet to be resolved in a satisfactory manner. Access to and viewing of information at the enterprise level is now primarily through massive sets of paper drawings. There is practically no sharing of information, updating is at best cumbersome, and searching for specific information or versions nearly impossible because of the massive number of drawings and MCAD/ECAD files. (See Appendix A for a definition of acronyms used in this paper.) The problem is magnified by the fact that the majority of the actual manufacturing is now performed by outside suppliers, who are likely to be located in different cities, countries, or even continents from the OEM. The impact of this disjointed environment on the overall product development cycle is enormous. Over the last 2 years, this author has put together a series of technical reports [3–6] and prototypes that investigate these issues at length. Based on these investigations, it is proposed here that a satisfactory solution to these problems can be achieved primarily through a client/server system that integrates people and computers into the total product and processes development. Such a system, for instance, generates requested information dynamically, displays that information in a useful manner electronically, maintains control over the created/saved files, automatically updates the information, and supports a distributed environment. Moreover, we propose that this system must have the Web as its backbone and technologies such as Java, browsers, HTML, XML, VRML/X3D, TCP/IP, and Agents as its communication and collaboration infrastructure. Java is a programming language that is highly tuned for the Web environment. It is object-oriented, robust, secure, portable, interpreted, threaded, and dynamic [7]. XML is the new standard that formalizes the semantics of the contents of Web files and facilitates electronic data interchange [8], and VRML/X3D is the standard language for delivering and displaying 3D data on the Web [9]. Finally, TCP/IP is the protocol that guarantees reliable delivery of bits from point A (a server) to point B (a client) on the network using sockets [10]. None of these technologies and protocols, however, deals effectively with distributed objects and environments; it is difficult to find the right information at the right time and interact with it across a distributed network. This is where the Object Web [11] comes into the picture.
2. Client/server computing: the Object Web An important by-product of living in this information age is that environments that do not conform to the Web (i.e. do not become Web-enabled) seem quite limited. The Web culture has taught us that computing is infinite: there is more data and functionality literally a click away. Once exposed to this concept, it is difficult to constrain users by bounded and contained environments. The Web’s ease-ofuse, scalability, flexibility, ubiquity, and diverse content gives a whole new dimension for collaboration at the extended enterprise level. However, as the number of information consumers grows within the extended enterprise, the traditional monolithic access methods begin to fail. Client/server computing, on the other hand, leverages the capabilities of the networks to balance the load and solve the scalability problem associated with information access [12]. There are a number of Web-based and non-Web-based systems that provide client–server services. They include CORBA, DCOM, HTTP/CGI, MOM, and RPC. The first two are object-based and have a sound framework, the third one is used exclusively on the Web, and the last two are more general but follow the traditional client–server architecture. All indications are that only CORBA and DCOM provide the degree of sophistication needed to implement a practical object-based, client–server system at the extended enterprise level. In Appendix B, we provide a brief description of these two standards and discuss how Web-based tools such as Java, XML, and VRML can complement these distributed object standards to provide an infrastructure for information access called the Object Web. These tools can deliver the ultimate goal of client–server computing, an environment in which components not only inter-operate but also collaborate at the semantic level to get the job done. However, to operate at the extended enterprise level, the Object Web must be complemented by authoring tools for creation of information (e.g. ECAD/MCAD packages) and management tools for controlling the flow of information (e.g. PIM/ERP systems). Fig. 1 displays the variety of individuals and environments that must be served by such a system, which we call Enterprise-Web or simply E-Web. To create and provide the right set of information and knowledge and convey it in the right format in a controlled and secure manner to everyone shown in this figure is one of the most critical aspects of a successful IP3D environment.
3. Object Web in the extended enterprise: the E-Web portal We envision that, in the near future, we can categorize the end-users throughout the extended enterprise into two types: the “information consumers” who primarily need to view the data and read/access related material (e.g. individuals in
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
87
Marketing & Management Design & Engineering
Training & Support
The Enterprise-Web User Community
Suppliers & Partners
Manufacturing & Testing Sales & Distribution
Fig. 1. The Enterprise-Web supports everyone involved with a product during its life cycle.
management, marketing, sales, support, suppliers, shopfloor personnel), and the “information creators” who not only need quick and easy access to the data, but also need the means to make substantial modifications to that data (e.g. process planners, designers, analysts, manufacturing engineers). The first set of end users (the majority) need a low-cost, low-maintenance, and easy-to-use environment to view the information and perhaps add/publish simple attributes. For these customers, the Web-based server is the only viable solution. The “information creators,” on the other hand, will use the server as a means of communication and as a decision support system on high-end graphical user interfaces for concurrent design and collaborative engineering. For these customers we need a fast and versatile search and publishing capability, accessible from CAD and PDM systems, and embedded URLs in our databases that can give them additional information as they are making modifications. Although a Web-based server may not be the only way to make data available to this group, the need for collaboration and information sharing at the extended enterprise level makes the Web-based solution very attractive. The issue common to both groups of server customers is the enterprise-wide management of data and resources. This management must be in place to ensure integrity of data and propagation of up-to-date information throughout the organization. Thus, integration with PDM/ERP products is critical to the success of a Web-based system. ECAD/ MCAD packages play the role of the authoring tool for creation of information in our proposed “dynamic” system. The system must be dynamic because if we simply take a snapshot of the database (i.e. data in libraries) and convert it to a Web-compatible “visual” document, the result is a “static” server which has lost contact with the source (which contains changes to the information since the last snapshot) and have generated a heavy-weight file which cannot easily be transmitted or searched. Moreover, any mark-up that must be transmitted back to the source and managed there will have a much tougher path. The whole concept can only work if a
light-weight client is in constant and direct contact with the source of information (i.e. the servers) which in turn can provide light-weight information quickly and easily to the user for viewing and/or mark-up through the Object Web. The source of information in our case is primarily CAD data in libraries; however, a PDM system can serve up other types of information generated elsewhere in the extended enterprise (e.g. front/back office documents). The importance of the Web in providing access to information goes far beyond pure viewing. Browsers provide a powerful publishing capability with their visual editors, drag/drop capabilities, simple one-button publishing, page wizards, and support of HTML/XML/Java tools. Moreover, the host of tools provided by companies like Microsoft and Netscape (e.g. NetMeeting and Conference) make it possible to not only generate and format information, but also to share and mark up documents in a distributed/collaborative environment. Using the Web for information management gives us the ability to post changes on the network, thus giving users in remote locations instant access to these changes. In addition, a dynamically generated Web page that reports any relevant information to the manufacturing engineers, either on request or by notification, could drastically reduce the “search” time. Furthermore, the data would be delivered without the need for the user to explicitly know where it came from or where it goes after they are done with it. Fig. 2 shows a schematic diagram illustrating the use of authoring tools (i.e. CAD/CAM/CAE) and administration/ management tools (i.e. PDM/ERP), collectively referred to as C3PE, to enhance the Object Web for enterprise use. This figure defines what we have termed as the Enterprise-Web (E-Web) portal in this paper. It is easy to see that in this EWeb environment, a client has a single access point to any information regardless of its location or storage mechanism. Moreover, the access and viewing is controlled by the C3PE system in the sense that the client is recognized (i.e. authenticated) by the system and the information it has permission to see is customized (using Java and XML) for that particular user.
88
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
Corporate System
Files Files
Files D
D
Corporate System
Files
Supplier System
Files
C3PE D
D
D
HTTP & CORBA/DCOM Servers
Intranets/Internet HTML/XML and Java/VRML
Browser Client
Files
Customer System D
Fig. 2. The E-Web system provides access to information at the extended enterprise level.
All CAD/CAM/CAE and PIM/ERP (C3PE) vendors are now embracing the Object Web and its tools to varying degrees. A majority of these vendors consider the Web technology primarily as a presentation layer, while the underlying data continues to be stored in traditional PDMmaintained databases, and managed and viewed with conventional tools. However, some of these vendors forecast that data repositories will soon be accessed and managed via the Web (encompassing both the Internet and intranets). The data access will be dynamic, and will propagate through an enterprise either by request (i.e. a “pull” data-access model) or by automatic notification using software agents (i.e. a “push” data-access model). See Ref. [13] for more information on software agents and their impact on the Web. This type of data server “push” and browser “pull” may be necessary to get around the Web’s growing bandwidth problems. New technologies like the Digital Subscriber Line (DSL), Cable Modems, and wireless data communications provide hope for cost-effective broadband access to enterprise information at all levels and from any location. In this paper we recommend designing a client/server system [4] on top of the Object Web that uses CAD and PDM/ERP systems at its core for data generation and resource management, and then takes advantage of the existing Web-based tools for propagating the information within the enterprise. This system must have the ability to be customized to users’ specifications and at the same time make this customization as simple as possible. Since different organizations are structured differently and have different types of data to communicate to different groups of individuals, such customization is critical to the success of the server. Fortunately, the biggest asset of the Web is its potential for customization through its open and easily accessed/understood standards. With Java and XML we have the ability to create platform independent programs that enable manipulation of format-independent information. With the right style sheet, this information can be
represented in many ways and on any platform [1]. Other de facto industry standards such as CORBA IIOP (Internet Inter-ORB Protocol) can provide a common way of connecting distributed objects across the Internet and intranets. Since many systems are now CORBA/DCOMcompliant, the server can very easily and effectively communicate MCAD/ECAD data to other objects on the Web. Servers must have secure and reliable access to data stored in a wide variety of persistent data engines (e.g. object databases, relational databases, file systems, dynamic websites), which may be scattered throughout the enterprise. Some products use Java DataBase Connectivity (JDBC) API to access databases ranging from dBase II to Oracle Server. Other products use JavaScript and serverside Java to access most of the existing relational databases by the ODBC (Open Database Connectivity) gateway or native methods. The important point here is that E-Web must provide secure and easy access to any existing persistent data engine in a generalized way, and this may require multiple access methods.
4. Life-cycle support by the Enterprise-Web portal The proposed E-Web portal discussed in this paper follows a multi-tier approach known as client–server– server. In such a system, the client (tier one) is sitting on the user’s desktop and is connected to the Object-Web server (tier two) which in turn is connected to the data servers (tiers three and up), with access to and control over all types of information scattered throughout the enterprise. These data servers contain all the databases, information-creator systems, and legacy applications. The second tier contains the HTTP and CORBA/DCOM servers. All transactions between these tiers is controlled and monitored by data and information management systems. These systems also may include a layer for administration and
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
management of users. The client (first tier) contains the user interface along with some Java applets that perform simple tasks. We have created several prototypes based on this architecture to show feasibility of the E-Web portal concept. These prototypes are demonstrated in this section through scenarios that cover the entire life cycle of a product. One approach for the visual user interface in E-Web is the use of VRML files. (The next generation of VRML is the XML-based Extensible 3D or X3D. The current ISO/ VRML97 standards are used in this paper.) VRML is the Internet standard for communicating “rich” 3D data. The richness of VRML is due to its interaction and navigation capabilities and also to the fact that VRML objects can be hyperlinked to multimedia (image, text, video, audio) or HTML files, as well as to other VRML objects. SDRC’s I-DEAS, for example, currently has the ability to create outstanding VRML files which are hyperlinked to other HTML pages containing relevant geometric and nongeometric information. Fig. 3 displays an image of exported VRML and HTML files which contain all types of geometry, annotations, user attributes, and display options. (You can use http://www.sdrc.com/ideas/ vrml_help/vrml_help.html to get more information and download the actual exported files.) In our prototypes, the Java-enhanced VRML files provide the bulk of the visual
89
interface between the client and the servers. We assume here that these VRML files are automatically created/ updated by the enterprise CAD systems and are maintained by the enterprise PDM systems. We also assume that the relationship between the CAD data and other associated data, stored in various databases, has already been defined and the links between them are maintained by the PDM systems. Based on these assumptions, we envision E-Web scenarios for users during a product’s life-cycle along the following lines: 4.1. Scenario 1: requirements gathering and conceptual design During early design and concept stages of product development, there is a tremendous need for searching the enterprise databases to match customer requirements (design specifications) with characteristics of existing components: 1. Designer wants to conduct a search, based on a criterion, of all parts and assemblies. 2. Designer wants to view product structure and geometry simultaneously via E-Web. 3. Designer brings up the interface client window and logs into the EPDM system.
Fig. 3. Inter-linked VRML and HTML files give access to all MCAD data in E-Web.
90
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
4. The EPDM system checks designer’s security-clearance level and provides appropriate access. 5. Designer wants to see all components in the system that have certain characteristics. 6. Designer clicks on the appropriate button and specifies criterion on the preformatted form. 7. The EPDM system locates all parts and displays them (with their status) in the bottom pane. 8. Designer selects parts and sees hierarchy and “thumbnail” geometry in the top two panes. 9. Designer can get more “precise” geometry once a particular component has been identified. 10. Designer can interact with geometry, do dimensional or volume queries, get metadata, etc. 11. Designer can forward an e-mail and ask questions, if necessary, about the original design intent. 12. Designer can annotate any of the panes and save the marked-up copy in the EPDM system. 13. The EPDM system checks the validity of the file and places it in the appropriate location. 14. The EPDM system notifies all team members which might be affected by the search findings. Fig. 4 illustrates the three-pane client for this scenario within Netscape’s Navigator. The bottom and upper-left panes are Java applets that use the Swing set (JFC) and can be connected to the PDM system through the
CORBA/DCOM servers. The upper-right pane displays the VRML file using the CosmoPlayer plug-in. This pane is connected to the PDM and CAD systems through an HTTP server. The important points here are that the flow of data is taking place in a controlled and managed environment and that the designer gets quick access to desired data without getting bogged down with placement/retrieval mechanisms or check-in/check-out procedures. 4.2. Scenario 2: detailed design and change notification In this scenario, we demonstrate the use of E-Web for engineering change notification during the detailed design stage of product development: 1. Producer (i.e. information creator) checks out parts, assemblies, or drawings from project libraries. 2. Producer makes modification per ECO and updates associated assembly instances, BOM, etc. 3. Producer places the files back into the project libraries and creates distribution files. 4. The EPDM system provides the means for electronic review and approval of design changes. 5. The design changes are reviewed and all comments are stored as part history attributes. 6. The EPDM system notifies all team members affected by the change and distributes results.
Fig. 4. The client user interface for the E-Web demonstrates search and viewing capabilities.
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
7. Each (information) consumer gets notified by e-mail and is provided with a URL for additional ECN info. 8. Each consumer can click on the URL which brings up the three-pane client interface. 9. Consumer views the thumbnail and the product structure to determine impact of change. 10. Consumer can get more “precise” information by clicking on the VRML or STEP tabs. 11. Consumer can use Microsoft’s NetMeeting to collaborate with the producer or other consumers. Fig. 5 illustrates the client user interface for the prototype supporting the engineering change process during the detailed design stage of product development. Note that everything is done electronically without the need to hold group meetings or refer to paper drawings. Also, the review and approval process can take place in a collaborative environment through E-Web and the use of conferencing tools such as NetMeeting and Conference, which are discussed in the next scenario. 4.3. Scenario 3: supplier-chain integration and production At this stage of product development, there is a need to collaborate very closely with the suppliers. The process starts with bidding and continues through manufacturing, inspection, and shop-floor assembly: 1. First Tier Supplier (FTS) is provided access to “lightweight” VRML file through the E-Web portal.
91
2. FTS wants bids from lower tier suppliers, identified by directory services, for a new sub-system. 3. FTS sends corresponding VRML files to all lower tier suppliers using e-mail. 4. All suppliers use NetMeeting to hold a live collaborative discussion session. 5. FTS brings up master VRML file and discusses various aspects with each lower tier supplier. 6. FTS brings up the white board and everyone starts the mark-up process. 7. Comments from chat sessions and mark-ups from white board are all saved in EPDM. 8. Appropriate 2D and 3D files are forwarded once the supplier chain has been identified and approved. 9. Supplier may examine the 2D and 3D information using the E-Web client and perform simple queries for data embedded in the VRML file at the assembly, part, and feature levels. 10. Supplier questions regarding a particular feature may be resolved by direct interaction with E-Web. 11. To retrieve supplier-requested data, the server goes to appropriate database(s) controlled by EPDM. 12. The information is dynamically converted and displayed in the best possible multimedia Web format. 13. Supplier can get additional information in a live collaborative session from OEM or other suppliers. 14. Once parts are shipped to OEM for assembly, inspection is done using the annotated VRML files.
WIP
6
0.0 A
0.0 6
B Thumbnail VRML
Jav 3D
X3D
Fig. 5. The E-Web prototype demonstrates engineering change notification via e-mail.
92
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
15. Conflicts between the designers, shop-floor personnel, and the suppliers are resolved through E-Web. The “light-weight” file mentioned in the above scenario contains only geometry, assembly hierarchy, feature attributes, scripting code to allow filtering and navigation, and the associated URLs for multimedia and HTML files. All other information remains stored in databases known to the EPDM system. Also, the type of data that a supplier may query includes metadata from EPDM, life-cycle and process planning information, product and manufacturing information, dimensions, constraints, tolerances, FE meshes and results, training and educational material, sales and distribution information, etc. The major point in this scenario is that the whole process takes place electronically, without the need for expensive paper drawings. Fig. 6 shows an image of a live net-conferencing session between an imaginary OEM and a supplier, who may be thousands of miles apart physically. Notice the presence of audio and video, and the ability to mark up and time-stamp the documents electronically. The discussions from each session, stored in a chronicle order in the same file, can be retrieved very easily, and the complete file can be stored in the EPDM system. 4.4. Scenario 4: product maintenance and disposal This scenario deals with post-production stages of
the product life cycle. We discuss here the case for supporting service personnel sent to the customer site for maintenance: 1. Customer calls the manufacturer to report problems with a product. 2. Representative from service department is sent to the customer site. 3. Service person performs necessary diagnostics and finds the bad component. 4. Service person looks at the product on the hand-held electronic-device for part identification. 5. Using E-Web, the product structure is navigated to locate part and its related information. 6. Once part number is found, the system checks inventory for availability. 7. Part is ordered from the nearest distribution center and scheduled for immediate delivery. 8. While waiting for the part, the service person begins the disassembly process using E-Web. 9. Service person clicks on URL associated with the part for appropriate instructions. 10. Associated URL may point to an animated VRML file which provides help in 3D. 11. New part arrives and service person replaces part according to the electronic instructions. 12. The removed component is disposed of according to instructions in appropriate URL.
Fig. 6. Using 2D markup of VRML files for communication in a live session using NetMeeting.
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
The E-Web’s product structure and the 3D visual interface help the user resolve problems in the quickest way possible. Moreover, since all actions are under the control of the EPDM system, any supporting document can be associated with each component relevant to the particular assembly. Fig. 4 illustrates that, once an instance of a part in an assembly has been selected, the user can click on this component to get information relevant to its function in the assembly. This information includes URLs for support and maintenance.
5. Discussion To cautiously predict the future trends in product development, we have proposed certain scenarios based on prototypes that use current proven and accepted technologies. These scenarios and prototypes can help define a roadmap for a Web-assisted product development environment at the extended enterprise level. The simplicity and intuitiveness of the Web is causing a revolution in the way companies are developing their next-generation products. We believe our examples illustrate the way products are going to be designed, built, marketed, and maintained in the near future. The E-Web prototypes illustrate that we will soon have access to any desired piece of data on the Web by using the keyboard, the mouse, and/or voice commands. The E-Web portal can help the enterprise user connect to other users, to the Internet, to any database or computer on the network, and essentially any electronic device with an IP address and information. It is not too difficult to imagine how the manufacturing industry will flourish in this paperless electronic world. Today, most of our detailed design and manufacturing is still performed using paper drawings and most, if not all, inspection, validation, assembly, approvals, and procurements require paper and physical prototypes. These methods have been in practice for many years with de facto industry standards and understood limitations. This paper should help us determine the feasibility of a paradigm shift to the new electronic (paperless) world and understand its consequences. Fig. 7a and b looks at the problem from the point of view of OEM and compares today’s procedures with tomorrow’s possibilities. The current process followed by a majority of OEMs includes a succession of converting the CAD model to multiple paper drawings, annotating and plotting these drawings, checking and re-checking them, sending them as a packet to suppliers, and repeating the whole procedure again when there is missing information. The result is a time-consuming and expensive procedure. However, what is even more troubling is that there is practically no collaboration involved; all parties live in a virtual information vacuum until it is their turn to look at the data. In contrast, with E-Web everyone can be involved simultaneously throughout the development cycle and the information is both shared and collaborated in a controlled
93
environment. Therefore, it is easy to see that E-Web is an integral part of implementing the IP3D strategy. This discussion leads us to an important question. Should we use the open Web-related technologies like Java/XML to view the 2D/3D data generated by the server or should we rely on more specialized proprietary viewers that may satisfy our immediate needs but have limited potential for growth at the extended enterprise level? Both strategies have benefits and limitations: these involve a trade-off between the timely delivery of specific functionality for MDA/EDA versus being highly compatible with the rest of the emerging network technologies (e.g. XML, browsers) and enterprise systems (e.g. EDA, ERP). Our long-term goals must be based on using the universal user interface provided by the Web browsers rather than introducing another layer of UI in non-Web-based viewers. Any short-term advantages that may be gained using a CAD-specific viewing tool must be weighed against that goal. The alternative is to implement IP3D using non-Web technologies and protocols. This dramatically increases the risk of evolving the customer’s information infrastructure into a proprietary world of incompatible formats. This is not acceptable. Therefore, the architecture of the E-Web portal, as part of the IP3D strategy, must be based on a seamless interface to the Object Web and must have CAD/CAM/CAE and PIM/ERP (C3PE) systems at its core, as depicted in Fig. 2. We note that not all Web technologies are ready for use at the extended enterprise level. For example, Java is still too slow for real world applications, and conferencing tools such as NetMeeting lack performance and scalability. We also mentioned the bandwidth bottleneck in a previous section. However, there are strong indications that new enhancements and technologies will soon resolve most, if not all, of these issues. Companies like IBM are working on fast and scalable conferencing tools for the enterprise and new versions of JDK (Java Development Kit) continue to improve its performance and security. Also, new standards such as XML will work with tools like Java to facilitate electronic document/data interchange on the Web. XML holds the promise of introducing structure and semantics into the enterprise documents. Thus, with XML, the E-Web portal will have standards for intelligent search, data exchange, adaptive presentation, and personalization. We will talk more about this issue in the context of Knowledge-Based Product Development (KBPD) in the second part of this paper [1].
6. Conclusions In a few years, assuming that bandwidth and security concerns are addressed, the impact of the Web in our lives will be as fundamental and deep as the impact
94
M. Rezayat / Computer-Aided Design 32 (2000) 85–96 Select a 3D model to make drawing
Transfer to drafting module and get 2D cross sections
Send back marked-up drawing for additional information
Annotate 2D crosssections and add user attributes
Send back parts that fail inspection
Resize and send to the plotter Check all drawings and validate
Send order to Procurement
Ship parts back for inspection/assembly
Send all drawings to supplier(s)
(a)
Annotate the 3D CAD model
Use E-Web server to get & put missing/additional information
Reference Design and Engineering
Check the 3D model and validate Export light-weight 3D Web file and place on the network Reference Reference Reference
Collaborate Shop Floor Inspection/Assembly (b)
Pre-set appropriate 3D views & attributes
Reference
Suppliers
Collaborate Collaborate
Collaborate
Procurement & Support
Customers
Fig. 7. (a) Today’s drawing-based manufacturing using complex and time-consuming procedures. (b) Tomorrow’s paperless product development environment using E-Web.
from electricity and the telephone. And just like these two elements, the most amazing changes that the Web will cause are the ones we do not expect. Thus, managers and planners in any industry must have a comprehensive plan to use the Web and be able to react to any unanticipated changes it might bring. To this end, we have introduced the concept of the E-Web portal to illustrate how Web-based standards and distributed-object technologies can be integrated with MCAD/ ECAD and PIM/ERP systems to provide controlled access to any type of information and resource within the extended enterprise. In this paper, we have reviewed some scenarios on anticipated changes in the manufacturing sector due to EWeb. There is no argument about the need for a dynamic information server. Currently, designers and manufacturing
engineers spend more than 50% of their time just looking for information. There is also no argument that tight integration with suppliers and customers throughout the entire development cycle is essential. However, answering questions about the underlying architecture and the user interface requires more thought. This paper is our first attempt to answer some of these questions through prototypes based on current technology. These prototypes illustrate the feasibility of an approach based on the E-Web portal strategy.
Acknowledgements Several individuals at Structural Dynamics Research Corporation contributed to the implementation of the
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
prototypes. The author acknowledges major contributions by Mark Bailey and Rick Miller. The author is grateful to Emily Ivey for providing constructive comments on the structure of the manuscript.
Appendix A. Definition of acronyms CGI Common Gateway Interface CORBA Common Object Request Broker Architecture DCOM Distributed Component Object Model DOAE Distributed Objects and Environments ECAD Electrical Computer Aided Design ECN Engineering Change Notification ECO Engineering Change Order EDA Electrical Design Automation EPDM Enterprise Product Data management ERP Enterprise Resource Planning E-Web Enterprise-Web HTML Hypertext Markup Language HTTP Hypertext Transfer Protocol IIOP Internet Inter-ORB Protocol IP3D Integrated Product/Process/Protocol Development LDAP Lightweight Directory Access Protocol MCAD Mechanical Computer-Aided Design MDA Mechanical Design Automation MIME Multipurpose Internet Mail Extensions MOM Message-Oriented Middleware OEM Original Equipment Manufacturers ORB Object Request Broker PDM Product Data Management PIM Product Information Management RPC Remote Procedure Calls SMTP Simple Mail Transfer Protocol TCP/IP Transmission Control Protocol/Internet Protocol URL Uniform Resource Locator VRML Virtual Reality Markup Language XML Extensible Markup Language
Appendix B. State-of-the-art in Distributed-Object Technology CORBA, introduced in 1991, is a specification that defines interoperability rules between distributed objects on clients and servers. The most critical part of a CORBA system is the Object Request Broker (ORB). The ORB provides the middleware services that shield clients from complexities of remote communication with data servers. It is responsible for establishing a client/server relationship between components. The ORB receives requests from the client objects and routes them to an appropriate server, which in turn delivers the service required back to the client though the ORB. In order to ensure compatibility between various ORBs, the CORBA specification requires use of Internet Inter-ORB Protocols (IIOP). In essence, IIOP
95
provides a common communication backbone between different ORBs by adding several CORBA-specific messages to TCP/IP, the common network communication protocol used by middleware. DCOM, introduced in 1996 by Microsoft, is like CORBA in that it separates interface from functionality by using an IDL (Interface Definition Language). However, currently, the IDL used by CORBA is quite different from the one used with DCOM, which causes severe interoperability problems. DCOM is based on a proprietary format and is more limiting than CORBA, e.g. DCOM has no naming or trading services, and its objects are not persistent (see Ref. [14] for more information). On the other hand, if a system mainly uses Win32 platforms, it may be simpler and more practical to implement DCOM than CORBA [15]. DCOM also has the added advantage of being a standard part of the Win32 operating systems. The fact is there are no clear winners in the standards for distributed objects. However, bridges are currently being established between these two standards. (CORBA 3.0 will support interoperability between the two standards and Microsoft, working with Iona Technologies and Visual Edge Software, is in the process of doing the same.) The bottom line is that both CORBA and DCOM are here to stay and, if necessary, these two sets of standards for distributed objects and environments can co-exist within the extended enterprise. Combined with object-based Web standards, these distributed-object standards provide powerful 3-tier client–server solutions. In a 3-tier system, generally, the user interface or the client is the first tier, distributed-object and HTTP servers are in the middle tier, and data servers and legacy applications are in the third tier. In such a system, CORBA/DCOM deals with location and access transparency, while Web-related tools such as Java and XML deal with implementation and presentation transparency. For instance, CORBA, using the ORB bus, provides two ways for objects to locate one another on the system. The first is the Naming Service in which a client object locates another object by that object’s ORB-registered name (analogous to the white pages in a phone book). The second method is through the Trader Service in which a client object asks for all objects that have a certain ORBregistered characteristic (analogous to the yellow pages in a phone directory). CORBA also contains functions for creating/deleting objects, storing them in a persistent manner, defining relationship between them, and publishing their states. Web standards (e.g. Java, XML, VRML) on the other hand simplify data exchange by providing a data dictionary, and allow platform independence and effective viewing, sharing, and editing. For instance, if the data is formatted in XML, it can be presented in different ways (depending on the need) on the client side using Java. Together, these standards comprise the so-called Object Web, in which an object can efficiently locate any other object on the network and inter-operate with it without using slow CGI scripts.
96
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
References [1] Rezayat M. Knowledge-based product development using XML and KCs, submitted for publication. [2] Berquist TP, Kahl SJ, Kumar A. Supply and demand management. Piper Jaffray Inc, March 1998. [3] Rezayat M. Web: the next generation. SDRC Technical Report, February 1997. [4] Rezayat M. A network-enabled product information server/publisher based on I-DEAS and metaphase. SDRC Technical Report, May 1997. [5] Rezayat M. Introduction to VRML and its relevance to MDA. SDRC Technical Report, September 1996. [6] Rezayat M. An introduction to computer agents and their applications. SDRC Technical Report, December 1996. [7] Lemay L, Perkins CL, Morrison M. Teach yourself Java in 21 days, professional reference edition. Sams.net, Indianapolis, Indiana, 1996. [8] St. Laurent S. XML: a primer. Foster City, CA: IDG Books Worldwide, 1998. [9] Carey R, Gavin B. The annotated VRML 2.0 reference manual. Reading, MA: Addison-Wesley Developers Press, 1997. [10] Comer DE. Internetworking with TCP/IP: principles, protocols, and architecture. New York: Prentice Hall, 1995. [11] Orfali R, Harkey D. Client/server programming with JAVA and CORBA. 2. New York: Wiley, 1998.
[12] Duchessi P, Chengalur-Smith I. Client/server benefits, problems, best practices. Commun ACM 1998:41. [13] Bigus JP, Bigus J. Constructing intelligent agents with Java. New York: Wiley, 1998. [14] Lewandowski SM. Frameworks for component-based client/server computing. ACM Comput Surv 1998:30. [15] Eddon G, Eddon H. Inside distributed COM. Microsoft Press, April 1998.
Mohsen Rezayat is a Technical Fellow at Structural Dynamics Research Corporation (SDRC) and an adjunct professor at the University of Cincinnati. He holds a PhD in engineering mechanics from the University of Kentucky, USA. As a director of research at SDRC, he has investigated extensively the simulation of manufacturing processes, integrated product development, and Web-related technologies. His current research interests include Webbased education and process modeling, portals for accessing applications and product information, and knowledge capture and reuse.
The Enterprise-Web portal for life-cycle support M. Rezayat* SDRC/University of Cincinnati, 2000 Eastman Drive, Milford, OH 45150, USA Accepted 2 September 1999
Abstract As we enter the new millennium, the approach to product development is evolving rapidly. Companies are in the process of creating a distributed design and manufacturing environment that enables integrated product, processes, and protocols development (IP3D). Certain strategies and some specific technologies are required to create such an environment. One such strategy is finding effective methods for communication and sharing of information, especially those related to design and manufacturing, throughout the entire enterprise and the supply chain. The technologies that support such a strategy must be able to deal with distributed environments and databases, must ensure reliability and security, and must be practical. Therefore, to accomplish this task efficiently, we need a roadmap that spells out the communication infrastructure (with a practical access interface) and defines the administration and management methodology. We suggest here that electronic access to design and manufacturing information within the extended enterprise must be Web-based because of its universal interface, open standards, ease of use, and ubiquity. To effectively deal with the distributed data, we recommend combining the distributed object standards (e.g. CORBA/DCOM) with the Web standards and protocols (e.g. Java, XML, IIOP) to create the Object Web. Finally, we propose that the Object Web must be combined with an enterprise’s information authoring and management systems (e.g. CAD, PDM, ERP) to create the Enterprise-Web (E-Web) portal, with the mission of providing the right information to the right person at the right time and in the right format anywhere within the extended enterprise. In this paper we identify a majority of the components needed to implement the E-Web and provide user scenarios (based on actual working prototypes) to demonstrate the effectiveness of such a system. Using the scenarios and the prototypes, we will show how E-Web can provide support for everyone associated with a product during its life cycle, thus creating a true IP3D environment. q 2000 Elsevier Science Ltd. All rights reserved. Keywords: Integrated product development and support; Process and application portal; Global information access; Supply-chain integration
1. Introduction Enterprise-Web portal will bind distributed environments and objects in the next millennium Integrated Product and Process Development (IPPD) has been in practice for some time in the major corporations. Global competition and distributed manufacturing environments will require that these corporations integrate people and computers into the total development process as well. This integration must take place through well-accepted human and electronic protocols. Therefore, we introduce the new concept of Integrated Product, Processes, and Protocols Development (IP3D), where the word “protocols” refers to both human and electronic protocols. One factor for this integration is that 50–80% of all components in products from Original Equipment Manufacturers (OEMs) are now fabricated by outside suppliers, and this trend is
* Tel.: 11-513-576-5981; fax: 11-513-576-2840. E-mail address: [email protected] (M. Rezayat).
expected to continue into the next century. Another factor is the need to access information throughout a product’s life cycle by everyone associated with its design, creation, sale, distribution, and maintenance. In this paper, we define the infrastructure for implementation of IP3D within the extended enterprise in order to support everyone that is associated with a particular product during its life cycle. In a second paper [1], we will discuss methods for bringing the enterprise knowledge to each individual as a decision-support aid for effective design and production. To remain competitive, OEMs must operate efficiently at the extended enterprise level. An extended enterprise comprises an OEM, its supply chain, subsidiaries, consultants, and partners affiliated with the life cycle of a particular family of products. By integrating the various components of an extended enterprise into the development process, OEMs can dramatically reduce the cycle time and cost, while improving quality and product variety. (In the remainder of this paper, “enterprise” represents the extended enterprise, unless indicated otherwise.) Some of
0010-4485/00/$ - see front matter q 2000 Elsevier Science Ltd. All rights reserved. PII: S0010-448 5(99)00092-5
86
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
the critical, and inter-linked, issues that must be addressed in this regard are: • Enterprise-wide access/sharing of information. • Enterprise-wide viewing/updating of information. • Enterprise-wide management of information resources.
and
A variety of commercial PIM (Product Information Management) and ERP (Enterprise Resource Planning) products are at the disposal of these companies to address the last issue [2]; but the first two issues are yet to be resolved in a satisfactory manner. Access to and viewing of information at the enterprise level is now primarily through massive sets of paper drawings. There is practically no sharing of information, updating is at best cumbersome, and searching for specific information or versions nearly impossible because of the massive number of drawings and MCAD/ECAD files. (See Appendix A for a definition of acronyms used in this paper.) The problem is magnified by the fact that the majority of the actual manufacturing is now performed by outside suppliers, who are likely to be located in different cities, countries, or even continents from the OEM. The impact of this disjointed environment on the overall product development cycle is enormous. Over the last 2 years, this author has put together a series of technical reports [3–6] and prototypes that investigate these issues at length. Based on these investigations, it is proposed here that a satisfactory solution to these problems can be achieved primarily through a client/server system that integrates people and computers into the total product and processes development. Such a system, for instance, generates requested information dynamically, displays that information in a useful manner electronically, maintains control over the created/saved files, automatically updates the information, and supports a distributed environment. Moreover, we propose that this system must have the Web as its backbone and technologies such as Java, browsers, HTML, XML, VRML/X3D, TCP/IP, and Agents as its communication and collaboration infrastructure. Java is a programming language that is highly tuned for the Web environment. It is object-oriented, robust, secure, portable, interpreted, threaded, and dynamic [7]. XML is the new standard that formalizes the semantics of the contents of Web files and facilitates electronic data interchange [8], and VRML/X3D is the standard language for delivering and displaying 3D data on the Web [9]. Finally, TCP/IP is the protocol that guarantees reliable delivery of bits from point A (a server) to point B (a client) on the network using sockets [10]. None of these technologies and protocols, however, deals effectively with distributed objects and environments; it is difficult to find the right information at the right time and interact with it across a distributed network. This is where the Object Web [11] comes into the picture.
2. Client/server computing: the Object Web An important by-product of living in this information age is that environments that do not conform to the Web (i.e. do not become Web-enabled) seem quite limited. The Web culture has taught us that computing is infinite: there is more data and functionality literally a click away. Once exposed to this concept, it is difficult to constrain users by bounded and contained environments. The Web’s ease-ofuse, scalability, flexibility, ubiquity, and diverse content gives a whole new dimension for collaboration at the extended enterprise level. However, as the number of information consumers grows within the extended enterprise, the traditional monolithic access methods begin to fail. Client/server computing, on the other hand, leverages the capabilities of the networks to balance the load and solve the scalability problem associated with information access [12]. There are a number of Web-based and non-Web-based systems that provide client–server services. They include CORBA, DCOM, HTTP/CGI, MOM, and RPC. The first two are object-based and have a sound framework, the third one is used exclusively on the Web, and the last two are more general but follow the traditional client–server architecture. All indications are that only CORBA and DCOM provide the degree of sophistication needed to implement a practical object-based, client–server system at the extended enterprise level. In Appendix B, we provide a brief description of these two standards and discuss how Web-based tools such as Java, XML, and VRML can complement these distributed object standards to provide an infrastructure for information access called the Object Web. These tools can deliver the ultimate goal of client–server computing, an environment in which components not only inter-operate but also collaborate at the semantic level to get the job done. However, to operate at the extended enterprise level, the Object Web must be complemented by authoring tools for creation of information (e.g. ECAD/MCAD packages) and management tools for controlling the flow of information (e.g. PIM/ERP systems). Fig. 1 displays the variety of individuals and environments that must be served by such a system, which we call Enterprise-Web or simply E-Web. To create and provide the right set of information and knowledge and convey it in the right format in a controlled and secure manner to everyone shown in this figure is one of the most critical aspects of a successful IP3D environment.
3. Object Web in the extended enterprise: the E-Web portal We envision that, in the near future, we can categorize the end-users throughout the extended enterprise into two types: the “information consumers” who primarily need to view the data and read/access related material (e.g. individuals in
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
87
Marketing & Management Design & Engineering
Training & Support
The Enterprise-Web User Community
Suppliers & Partners
Manufacturing & Testing Sales & Distribution
Fig. 1. The Enterprise-Web supports everyone involved with a product during its life cycle.
management, marketing, sales, support, suppliers, shopfloor personnel), and the “information creators” who not only need quick and easy access to the data, but also need the means to make substantial modifications to that data (e.g. process planners, designers, analysts, manufacturing engineers). The first set of end users (the majority) need a low-cost, low-maintenance, and easy-to-use environment to view the information and perhaps add/publish simple attributes. For these customers, the Web-based server is the only viable solution. The “information creators,” on the other hand, will use the server as a means of communication and as a decision support system on high-end graphical user interfaces for concurrent design and collaborative engineering. For these customers we need a fast and versatile search and publishing capability, accessible from CAD and PDM systems, and embedded URLs in our databases that can give them additional information as they are making modifications. Although a Web-based server may not be the only way to make data available to this group, the need for collaboration and information sharing at the extended enterprise level makes the Web-based solution very attractive. The issue common to both groups of server customers is the enterprise-wide management of data and resources. This management must be in place to ensure integrity of data and propagation of up-to-date information throughout the organization. Thus, integration with PDM/ERP products is critical to the success of a Web-based system. ECAD/ MCAD packages play the role of the authoring tool for creation of information in our proposed “dynamic” system. The system must be dynamic because if we simply take a snapshot of the database (i.e. data in libraries) and convert it to a Web-compatible “visual” document, the result is a “static” server which has lost contact with the source (which contains changes to the information since the last snapshot) and have generated a heavy-weight file which cannot easily be transmitted or searched. Moreover, any mark-up that must be transmitted back to the source and managed there will have a much tougher path. The whole concept can only work if a
light-weight client is in constant and direct contact with the source of information (i.e. the servers) which in turn can provide light-weight information quickly and easily to the user for viewing and/or mark-up through the Object Web. The source of information in our case is primarily CAD data in libraries; however, a PDM system can serve up other types of information generated elsewhere in the extended enterprise (e.g. front/back office documents). The importance of the Web in providing access to information goes far beyond pure viewing. Browsers provide a powerful publishing capability with their visual editors, drag/drop capabilities, simple one-button publishing, page wizards, and support of HTML/XML/Java tools. Moreover, the host of tools provided by companies like Microsoft and Netscape (e.g. NetMeeting and Conference) make it possible to not only generate and format information, but also to share and mark up documents in a distributed/collaborative environment. Using the Web for information management gives us the ability to post changes on the network, thus giving users in remote locations instant access to these changes. In addition, a dynamically generated Web page that reports any relevant information to the manufacturing engineers, either on request or by notification, could drastically reduce the “search” time. Furthermore, the data would be delivered without the need for the user to explicitly know where it came from or where it goes after they are done with it. Fig. 2 shows a schematic diagram illustrating the use of authoring tools (i.e. CAD/CAM/CAE) and administration/ management tools (i.e. PDM/ERP), collectively referred to as C3PE, to enhance the Object Web for enterprise use. This figure defines what we have termed as the Enterprise-Web (E-Web) portal in this paper. It is easy to see that in this EWeb environment, a client has a single access point to any information regardless of its location or storage mechanism. Moreover, the access and viewing is controlled by the C3PE system in the sense that the client is recognized (i.e. authenticated) by the system and the information it has permission to see is customized (using Java and XML) for that particular user.
88
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
Corporate System
Files Files
Files D
D
Corporate System
Files
Supplier System
Files
C3PE D
D
D
HTTP & CORBA/DCOM Servers
Intranets/Internet HTML/XML and Java/VRML
Browser Client
Files
Customer System D
Fig. 2. The E-Web system provides access to information at the extended enterprise level.
All CAD/CAM/CAE and PIM/ERP (C3PE) vendors are now embracing the Object Web and its tools to varying degrees. A majority of these vendors consider the Web technology primarily as a presentation layer, while the underlying data continues to be stored in traditional PDMmaintained databases, and managed and viewed with conventional tools. However, some of these vendors forecast that data repositories will soon be accessed and managed via the Web (encompassing both the Internet and intranets). The data access will be dynamic, and will propagate through an enterprise either by request (i.e. a “pull” data-access model) or by automatic notification using software agents (i.e. a “push” data-access model). See Ref. [13] for more information on software agents and their impact on the Web. This type of data server “push” and browser “pull” may be necessary to get around the Web’s growing bandwidth problems. New technologies like the Digital Subscriber Line (DSL), Cable Modems, and wireless data communications provide hope for cost-effective broadband access to enterprise information at all levels and from any location. In this paper we recommend designing a client/server system [4] on top of the Object Web that uses CAD and PDM/ERP systems at its core for data generation and resource management, and then takes advantage of the existing Web-based tools for propagating the information within the enterprise. This system must have the ability to be customized to users’ specifications and at the same time make this customization as simple as possible. Since different organizations are structured differently and have different types of data to communicate to different groups of individuals, such customization is critical to the success of the server. Fortunately, the biggest asset of the Web is its potential for customization through its open and easily accessed/understood standards. With Java and XML we have the ability to create platform independent programs that enable manipulation of format-independent information. With the right style sheet, this information can be
represented in many ways and on any platform [1]. Other de facto industry standards such as CORBA IIOP (Internet Inter-ORB Protocol) can provide a common way of connecting distributed objects across the Internet and intranets. Since many systems are now CORBA/DCOMcompliant, the server can very easily and effectively communicate MCAD/ECAD data to other objects on the Web. Servers must have secure and reliable access to data stored in a wide variety of persistent data engines (e.g. object databases, relational databases, file systems, dynamic websites), which may be scattered throughout the enterprise. Some products use Java DataBase Connectivity (JDBC) API to access databases ranging from dBase II to Oracle Server. Other products use JavaScript and serverside Java to access most of the existing relational databases by the ODBC (Open Database Connectivity) gateway or native methods. The important point here is that E-Web must provide secure and easy access to any existing persistent data engine in a generalized way, and this may require multiple access methods.
4. Life-cycle support by the Enterprise-Web portal The proposed E-Web portal discussed in this paper follows a multi-tier approach known as client–server– server. In such a system, the client (tier one) is sitting on the user’s desktop and is connected to the Object-Web server (tier two) which in turn is connected to the data servers (tiers three and up), with access to and control over all types of information scattered throughout the enterprise. These data servers contain all the databases, information-creator systems, and legacy applications. The second tier contains the HTTP and CORBA/DCOM servers. All transactions between these tiers is controlled and monitored by data and information management systems. These systems also may include a layer for administration and
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
management of users. The client (first tier) contains the user interface along with some Java applets that perform simple tasks. We have created several prototypes based on this architecture to show feasibility of the E-Web portal concept. These prototypes are demonstrated in this section through scenarios that cover the entire life cycle of a product. One approach for the visual user interface in E-Web is the use of VRML files. (The next generation of VRML is the XML-based Extensible 3D or X3D. The current ISO/ VRML97 standards are used in this paper.) VRML is the Internet standard for communicating “rich” 3D data. The richness of VRML is due to its interaction and navigation capabilities and also to the fact that VRML objects can be hyperlinked to multimedia (image, text, video, audio) or HTML files, as well as to other VRML objects. SDRC’s I-DEAS, for example, currently has the ability to create outstanding VRML files which are hyperlinked to other HTML pages containing relevant geometric and nongeometric information. Fig. 3 displays an image of exported VRML and HTML files which contain all types of geometry, annotations, user attributes, and display options. (You can use http://www.sdrc.com/ideas/ vrml_help/vrml_help.html to get more information and download the actual exported files.) In our prototypes, the Java-enhanced VRML files provide the bulk of the visual
89
interface between the client and the servers. We assume here that these VRML files are automatically created/ updated by the enterprise CAD systems and are maintained by the enterprise PDM systems. We also assume that the relationship between the CAD data and other associated data, stored in various databases, has already been defined and the links between them are maintained by the PDM systems. Based on these assumptions, we envision E-Web scenarios for users during a product’s life-cycle along the following lines: 4.1. Scenario 1: requirements gathering and conceptual design During early design and concept stages of product development, there is a tremendous need for searching the enterprise databases to match customer requirements (design specifications) with characteristics of existing components: 1. Designer wants to conduct a search, based on a criterion, of all parts and assemblies. 2. Designer wants to view product structure and geometry simultaneously via E-Web. 3. Designer brings up the interface client window and logs into the EPDM system.
Fig. 3. Inter-linked VRML and HTML files give access to all MCAD data in E-Web.
90
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
4. The EPDM system checks designer’s security-clearance level and provides appropriate access. 5. Designer wants to see all components in the system that have certain characteristics. 6. Designer clicks on the appropriate button and specifies criterion on the preformatted form. 7. The EPDM system locates all parts and displays them (with their status) in the bottom pane. 8. Designer selects parts and sees hierarchy and “thumbnail” geometry in the top two panes. 9. Designer can get more “precise” geometry once a particular component has been identified. 10. Designer can interact with geometry, do dimensional or volume queries, get metadata, etc. 11. Designer can forward an e-mail and ask questions, if necessary, about the original design intent. 12. Designer can annotate any of the panes and save the marked-up copy in the EPDM system. 13. The EPDM system checks the validity of the file and places it in the appropriate location. 14. The EPDM system notifies all team members which might be affected by the search findings. Fig. 4 illustrates the three-pane client for this scenario within Netscape’s Navigator. The bottom and upper-left panes are Java applets that use the Swing set (JFC) and can be connected to the PDM system through the
CORBA/DCOM servers. The upper-right pane displays the VRML file using the CosmoPlayer plug-in. This pane is connected to the PDM and CAD systems through an HTTP server. The important points here are that the flow of data is taking place in a controlled and managed environment and that the designer gets quick access to desired data without getting bogged down with placement/retrieval mechanisms or check-in/check-out procedures. 4.2. Scenario 2: detailed design and change notification In this scenario, we demonstrate the use of E-Web for engineering change notification during the detailed design stage of product development: 1. Producer (i.e. information creator) checks out parts, assemblies, or drawings from project libraries. 2. Producer makes modification per ECO and updates associated assembly instances, BOM, etc. 3. Producer places the files back into the project libraries and creates distribution files. 4. The EPDM system provides the means for electronic review and approval of design changes. 5. The design changes are reviewed and all comments are stored as part history attributes. 6. The EPDM system notifies all team members affected by the change and distributes results.
Fig. 4. The client user interface for the E-Web demonstrates search and viewing capabilities.
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
7. Each (information) consumer gets notified by e-mail and is provided with a URL for additional ECN info. 8. Each consumer can click on the URL which brings up the three-pane client interface. 9. Consumer views the thumbnail and the product structure to determine impact of change. 10. Consumer can get more “precise” information by clicking on the VRML or STEP tabs. 11. Consumer can use Microsoft’s NetMeeting to collaborate with the producer or other consumers. Fig. 5 illustrates the client user interface for the prototype supporting the engineering change process during the detailed design stage of product development. Note that everything is done electronically without the need to hold group meetings or refer to paper drawings. Also, the review and approval process can take place in a collaborative environment through E-Web and the use of conferencing tools such as NetMeeting and Conference, which are discussed in the next scenario. 4.3. Scenario 3: supplier-chain integration and production At this stage of product development, there is a need to collaborate very closely with the suppliers. The process starts with bidding and continues through manufacturing, inspection, and shop-floor assembly: 1. First Tier Supplier (FTS) is provided access to “lightweight” VRML file through the E-Web portal.
91
2. FTS wants bids from lower tier suppliers, identified by directory services, for a new sub-system. 3. FTS sends corresponding VRML files to all lower tier suppliers using e-mail. 4. All suppliers use NetMeeting to hold a live collaborative discussion session. 5. FTS brings up master VRML file and discusses various aspects with each lower tier supplier. 6. FTS brings up the white board and everyone starts the mark-up process. 7. Comments from chat sessions and mark-ups from white board are all saved in EPDM. 8. Appropriate 2D and 3D files are forwarded once the supplier chain has been identified and approved. 9. Supplier may examine the 2D and 3D information using the E-Web client and perform simple queries for data embedded in the VRML file at the assembly, part, and feature levels. 10. Supplier questions regarding a particular feature may be resolved by direct interaction with E-Web. 11. To retrieve supplier-requested data, the server goes to appropriate database(s) controlled by EPDM. 12. The information is dynamically converted and displayed in the best possible multimedia Web format. 13. Supplier can get additional information in a live collaborative session from OEM or other suppliers. 14. Once parts are shipped to OEM for assembly, inspection is done using the annotated VRML files.
WIP
6
0.0 A
0.0 6
B Thumbnail VRML
Jav 3D
X3D
Fig. 5. The E-Web prototype demonstrates engineering change notification via e-mail.
92
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
15. Conflicts between the designers, shop-floor personnel, and the suppliers are resolved through E-Web. The “light-weight” file mentioned in the above scenario contains only geometry, assembly hierarchy, feature attributes, scripting code to allow filtering and navigation, and the associated URLs for multimedia and HTML files. All other information remains stored in databases known to the EPDM system. Also, the type of data that a supplier may query includes metadata from EPDM, life-cycle and process planning information, product and manufacturing information, dimensions, constraints, tolerances, FE meshes and results, training and educational material, sales and distribution information, etc. The major point in this scenario is that the whole process takes place electronically, without the need for expensive paper drawings. Fig. 6 shows an image of a live net-conferencing session between an imaginary OEM and a supplier, who may be thousands of miles apart physically. Notice the presence of audio and video, and the ability to mark up and time-stamp the documents electronically. The discussions from each session, stored in a chronicle order in the same file, can be retrieved very easily, and the complete file can be stored in the EPDM system. 4.4. Scenario 4: product maintenance and disposal This scenario deals with post-production stages of
the product life cycle. We discuss here the case for supporting service personnel sent to the customer site for maintenance: 1. Customer calls the manufacturer to report problems with a product. 2. Representative from service department is sent to the customer site. 3. Service person performs necessary diagnostics and finds the bad component. 4. Service person looks at the product on the hand-held electronic-device for part identification. 5. Using E-Web, the product structure is navigated to locate part and its related information. 6. Once part number is found, the system checks inventory for availability. 7. Part is ordered from the nearest distribution center and scheduled for immediate delivery. 8. While waiting for the part, the service person begins the disassembly process using E-Web. 9. Service person clicks on URL associated with the part for appropriate instructions. 10. Associated URL may point to an animated VRML file which provides help in 3D. 11. New part arrives and service person replaces part according to the electronic instructions. 12. The removed component is disposed of according to instructions in appropriate URL.
Fig. 6. Using 2D markup of VRML files for communication in a live session using NetMeeting.
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
The E-Web’s product structure and the 3D visual interface help the user resolve problems in the quickest way possible. Moreover, since all actions are under the control of the EPDM system, any supporting document can be associated with each component relevant to the particular assembly. Fig. 4 illustrates that, once an instance of a part in an assembly has been selected, the user can click on this component to get information relevant to its function in the assembly. This information includes URLs for support and maintenance.
5. Discussion To cautiously predict the future trends in product development, we have proposed certain scenarios based on prototypes that use current proven and accepted technologies. These scenarios and prototypes can help define a roadmap for a Web-assisted product development environment at the extended enterprise level. The simplicity and intuitiveness of the Web is causing a revolution in the way companies are developing their next-generation products. We believe our examples illustrate the way products are going to be designed, built, marketed, and maintained in the near future. The E-Web prototypes illustrate that we will soon have access to any desired piece of data on the Web by using the keyboard, the mouse, and/or voice commands. The E-Web portal can help the enterprise user connect to other users, to the Internet, to any database or computer on the network, and essentially any electronic device with an IP address and information. It is not too difficult to imagine how the manufacturing industry will flourish in this paperless electronic world. Today, most of our detailed design and manufacturing is still performed using paper drawings and most, if not all, inspection, validation, assembly, approvals, and procurements require paper and physical prototypes. These methods have been in practice for many years with de facto industry standards and understood limitations. This paper should help us determine the feasibility of a paradigm shift to the new electronic (paperless) world and understand its consequences. Fig. 7a and b looks at the problem from the point of view of OEM and compares today’s procedures with tomorrow’s possibilities. The current process followed by a majority of OEMs includes a succession of converting the CAD model to multiple paper drawings, annotating and plotting these drawings, checking and re-checking them, sending them as a packet to suppliers, and repeating the whole procedure again when there is missing information. The result is a time-consuming and expensive procedure. However, what is even more troubling is that there is practically no collaboration involved; all parties live in a virtual information vacuum until it is their turn to look at the data. In contrast, with E-Web everyone can be involved simultaneously throughout the development cycle and the information is both shared and collaborated in a controlled
93
environment. Therefore, it is easy to see that E-Web is an integral part of implementing the IP3D strategy. This discussion leads us to an important question. Should we use the open Web-related technologies like Java/XML to view the 2D/3D data generated by the server or should we rely on more specialized proprietary viewers that may satisfy our immediate needs but have limited potential for growth at the extended enterprise level? Both strategies have benefits and limitations: these involve a trade-off between the timely delivery of specific functionality for MDA/EDA versus being highly compatible with the rest of the emerging network technologies (e.g. XML, browsers) and enterprise systems (e.g. EDA, ERP). Our long-term goals must be based on using the universal user interface provided by the Web browsers rather than introducing another layer of UI in non-Web-based viewers. Any short-term advantages that may be gained using a CAD-specific viewing tool must be weighed against that goal. The alternative is to implement IP3D using non-Web technologies and protocols. This dramatically increases the risk of evolving the customer’s information infrastructure into a proprietary world of incompatible formats. This is not acceptable. Therefore, the architecture of the E-Web portal, as part of the IP3D strategy, must be based on a seamless interface to the Object Web and must have CAD/CAM/CAE and PIM/ERP (C3PE) systems at its core, as depicted in Fig. 2. We note that not all Web technologies are ready for use at the extended enterprise level. For example, Java is still too slow for real world applications, and conferencing tools such as NetMeeting lack performance and scalability. We also mentioned the bandwidth bottleneck in a previous section. However, there are strong indications that new enhancements and technologies will soon resolve most, if not all, of these issues. Companies like IBM are working on fast and scalable conferencing tools for the enterprise and new versions of JDK (Java Development Kit) continue to improve its performance and security. Also, new standards such as XML will work with tools like Java to facilitate electronic document/data interchange on the Web. XML holds the promise of introducing structure and semantics into the enterprise documents. Thus, with XML, the E-Web portal will have standards for intelligent search, data exchange, adaptive presentation, and personalization. We will talk more about this issue in the context of Knowledge-Based Product Development (KBPD) in the second part of this paper [1].
6. Conclusions In a few years, assuming that bandwidth and security concerns are addressed, the impact of the Web in our lives will be as fundamental and deep as the impact
94
M. Rezayat / Computer-Aided Design 32 (2000) 85–96 Select a 3D model to make drawing
Transfer to drafting module and get 2D cross sections
Send back marked-up drawing for additional information
Annotate 2D crosssections and add user attributes
Send back parts that fail inspection
Resize and send to the plotter Check all drawings and validate
Send order to Procurement
Ship parts back for inspection/assembly
Send all drawings to supplier(s)
(a)
Annotate the 3D CAD model
Use E-Web server to get & put missing/additional information
Reference Design and Engineering
Check the 3D model and validate Export light-weight 3D Web file and place on the network Reference Reference Reference
Collaborate Shop Floor Inspection/Assembly (b)
Pre-set appropriate 3D views & attributes
Reference
Suppliers
Collaborate Collaborate
Collaborate
Procurement & Support
Customers
Fig. 7. (a) Today’s drawing-based manufacturing using complex and time-consuming procedures. (b) Tomorrow’s paperless product development environment using E-Web.
from electricity and the telephone. And just like these two elements, the most amazing changes that the Web will cause are the ones we do not expect. Thus, managers and planners in any industry must have a comprehensive plan to use the Web and be able to react to any unanticipated changes it might bring. To this end, we have introduced the concept of the E-Web portal to illustrate how Web-based standards and distributed-object technologies can be integrated with MCAD/ ECAD and PIM/ERP systems to provide controlled access to any type of information and resource within the extended enterprise. In this paper, we have reviewed some scenarios on anticipated changes in the manufacturing sector due to EWeb. There is no argument about the need for a dynamic information server. Currently, designers and manufacturing
engineers spend more than 50% of their time just looking for information. There is also no argument that tight integration with suppliers and customers throughout the entire development cycle is essential. However, answering questions about the underlying architecture and the user interface requires more thought. This paper is our first attempt to answer some of these questions through prototypes based on current technology. These prototypes illustrate the feasibility of an approach based on the E-Web portal strategy.
Acknowledgements Several individuals at Structural Dynamics Research Corporation contributed to the implementation of the
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
prototypes. The author acknowledges major contributions by Mark Bailey and Rick Miller. The author is grateful to Emily Ivey for providing constructive comments on the structure of the manuscript.
Appendix A. Definition of acronyms CGI Common Gateway Interface CORBA Common Object Request Broker Architecture DCOM Distributed Component Object Model DOAE Distributed Objects and Environments ECAD Electrical Computer Aided Design ECN Engineering Change Notification ECO Engineering Change Order EDA Electrical Design Automation EPDM Enterprise Product Data management ERP Enterprise Resource Planning E-Web Enterprise-Web HTML Hypertext Markup Language HTTP Hypertext Transfer Protocol IIOP Internet Inter-ORB Protocol IP3D Integrated Product/Process/Protocol Development LDAP Lightweight Directory Access Protocol MCAD Mechanical Computer-Aided Design MDA Mechanical Design Automation MIME Multipurpose Internet Mail Extensions MOM Message-Oriented Middleware OEM Original Equipment Manufacturers ORB Object Request Broker PDM Product Data Management PIM Product Information Management RPC Remote Procedure Calls SMTP Simple Mail Transfer Protocol TCP/IP Transmission Control Protocol/Internet Protocol URL Uniform Resource Locator VRML Virtual Reality Markup Language XML Extensible Markup Language
Appendix B. State-of-the-art in Distributed-Object Technology CORBA, introduced in 1991, is a specification that defines interoperability rules between distributed objects on clients and servers. The most critical part of a CORBA system is the Object Request Broker (ORB). The ORB provides the middleware services that shield clients from complexities of remote communication with data servers. It is responsible for establishing a client/server relationship between components. The ORB receives requests from the client objects and routes them to an appropriate server, which in turn delivers the service required back to the client though the ORB. In order to ensure compatibility between various ORBs, the CORBA specification requires use of Internet Inter-ORB Protocols (IIOP). In essence, IIOP
95
provides a common communication backbone between different ORBs by adding several CORBA-specific messages to TCP/IP, the common network communication protocol used by middleware. DCOM, introduced in 1996 by Microsoft, is like CORBA in that it separates interface from functionality by using an IDL (Interface Definition Language). However, currently, the IDL used by CORBA is quite different from the one used with DCOM, which causes severe interoperability problems. DCOM is based on a proprietary format and is more limiting than CORBA, e.g. DCOM has no naming or trading services, and its objects are not persistent (see Ref. [14] for more information). On the other hand, if a system mainly uses Win32 platforms, it may be simpler and more practical to implement DCOM than CORBA [15]. DCOM also has the added advantage of being a standard part of the Win32 operating systems. The fact is there are no clear winners in the standards for distributed objects. However, bridges are currently being established between these two standards. (CORBA 3.0 will support interoperability between the two standards and Microsoft, working with Iona Technologies and Visual Edge Software, is in the process of doing the same.) The bottom line is that both CORBA and DCOM are here to stay and, if necessary, these two sets of standards for distributed objects and environments can co-exist within the extended enterprise. Combined with object-based Web standards, these distributed-object standards provide powerful 3-tier client–server solutions. In a 3-tier system, generally, the user interface or the client is the first tier, distributed-object and HTTP servers are in the middle tier, and data servers and legacy applications are in the third tier. In such a system, CORBA/DCOM deals with location and access transparency, while Web-related tools such as Java and XML deal with implementation and presentation transparency. For instance, CORBA, using the ORB bus, provides two ways for objects to locate one another on the system. The first is the Naming Service in which a client object locates another object by that object’s ORB-registered name (analogous to the white pages in a phone book). The second method is through the Trader Service in which a client object asks for all objects that have a certain ORBregistered characteristic (analogous to the yellow pages in a phone directory). CORBA also contains functions for creating/deleting objects, storing them in a persistent manner, defining relationship between them, and publishing their states. Web standards (e.g. Java, XML, VRML) on the other hand simplify data exchange by providing a data dictionary, and allow platform independence and effective viewing, sharing, and editing. For instance, if the data is formatted in XML, it can be presented in different ways (depending on the need) on the client side using Java. Together, these standards comprise the so-called Object Web, in which an object can efficiently locate any other object on the network and inter-operate with it without using slow CGI scripts.
96
M. Rezayat / Computer-Aided Design 32 (2000) 85–96
References [1] Rezayat M. Knowledge-based product development using XML and KCs, submitted for publication. [2] Berquist TP, Kahl SJ, Kumar A. Supply and demand management. Piper Jaffray Inc, March 1998. [3] Rezayat M. Web: the next generation. SDRC Technical Report, February 1997. [4] Rezayat M. A network-enabled product information server/publisher based on I-DEAS and metaphase. SDRC Technical Report, May 1997. [5] Rezayat M. Introduction to VRML and its relevance to MDA. SDRC Technical Report, September 1996. [6] Rezayat M. An introduction to computer agents and their applications. SDRC Technical Report, December 1996. [7] Lemay L, Perkins CL, Morrison M. Teach yourself Java in 21 days, professional reference edition. Sams.net, Indianapolis, Indiana, 1996. [8] St. Laurent S. XML: a primer. Foster City, CA: IDG Books Worldwide, 1998. [9] Carey R, Gavin B. The annotated VRML 2.0 reference manual. Reading, MA: Addison-Wesley Developers Press, 1997. [10] Comer DE. Internetworking with TCP/IP: principles, protocols, and architecture. New York: Prentice Hall, 1995. [11] Orfali R, Harkey D. Client/server programming with JAVA and CORBA. 2. New York: Wiley, 1998.
[12] Duchessi P, Chengalur-Smith I. Client/server benefits, problems, best practices. Commun ACM 1998:41. [13] Bigus JP, Bigus J. Constructing intelligent agents with Java. New York: Wiley, 1998. [14] Lewandowski SM. Frameworks for component-based client/server computing. ACM Comput Surv 1998:30. [15] Eddon G, Eddon H. Inside distributed COM. Microsoft Press, April 1998.
Mohsen Rezayat is a Technical Fellow at Structural Dynamics Research Corporation (SDRC) and an adjunct professor at the University of Cincinnati. He holds a PhD in engineering mechanics from the University of Kentucky, USA. As a director of research at SDRC, he has investigated extensively the simulation of manufacturing processes, integrated product development, and Web-related technologies. His current research interests include Webbased education and process modeling, portals for accessing applications and product information, and knowledge capture and reuse.