Highlights from the KBE Sessions at the Fall 2003 COE Conference
By Greg Premetz, Vought Aircraft Industries, Inc.

Our last Conference (Fall 2003) was the first time that we attempted to combine the activities of our organization with those from the International ICAD User's Group (IIUG). This is a direct result of Dassault purchasing the ICAD software company, Knowledge Technologies International, (i.e. KTI). For us, it meant that five of the ten KBE breakout sessions were related to ICAD. Although some attendees did not see the direct application to CATIA V5, others found the presentations to be educational since they showed a greater potential for KBE, other than just the creation of geometry in CAD. 

With the Conference being held in Seattle, we were able to benefit by an increase of KBE presentations from Boeing, a major practitioner of KBE. And although Boeing uses ICAD a lot, only two of their five sessions dealt with ICAD. Due to Boeing's presence, many attendees were interested in knowing Dassault's plans for incorporating ICAD into the V5 products or to what level would they be maintaining the ICAD software. Unfortunately, Dassault did not reveal any strategy for incorporating ICAD, and would only go so far as to say that they will support existing licenses through maintenance and not development. They did provide a session (KBE-10) meant to cover how "they intend to help and facilitate the customers to extract, capture, and document their ICAD-based KBE applications". But let's discuss this in detail later in the article.

Starting with KBE-1, Sheng-Chuan Wu from Franz Inc. presented a methodology for enabling engineering knowledge reuse through a "simple, universal, and ubiquitous access layer". He highlighted the issues with current KBE applications where the knowledge is statically maintained inside the application. He proposed using "web services" to create "a generic interface to discover and access information and processes". This interface would then work with his proposed methodology of Service Orientated Architecture (SOA) where the KBE applications are looked as "web services" and not as discrete functions. This allows the user to create dynamic KBE applications utilizing a "composite application" that is linked via the internet to databases, KBE Product Models, analysis routines, and other input and output applications.

He showed an architecture that incorporated Simple Object Access Protocol (SOAP) and LISP as the underlying technologies for the generic interface. SOAP is a lightweight protocol for exchange of information in a decentralized, distributed environment. It is an XML based protocol that consists of three parts: an envelope that defines a framework for describing what is in a message and how to process it, a set of encoding rules for expressing instances of application-defined datatypes, and a convention for representing remote procedure calls and responses. LISP is a very powerful language for doing KBE and Sheng-Chuan listed the numerous advantages for using this language as the core for all logic operations. He concluded the session by emphasizing the importance of having the ability for the KBE applications to evolve; and that web services and SOA provide this ability through a scalable and distributed architecture.

For KBE-2, Dennis Brauning from the Boeing Company presented some of the KBE applications that have been developed and are in use at their Wichita facility. He showed the applications used for Preliminary, Detail, and Post Design Processes, as well as, for Production. Applications included a Nacelle Design Tool, a Multi-surface Join and Offset Tool, a Generative Fuselage Tool, a Rapid Forging Design Tool, a Generative Bond Tool, a Stretch Form Block Tool, and a Tool for determining CAD geometry that is incompatible with NC Processes and then recommends replacement geometry.  

For KBE-3, Art Iodice from the Boeing Company presented their development of a KBE Web Viewer that is being currently used to interface with their Skin Layout and Stringer Clip Applications. The goals for the Web viewer are very similar to those in the KBE-1 session with similar technologies being employed (XML for the interface and ICAD LISP for the logic). Art noted the advantages such as built-in 3D visualization tools, a CAD independent platform, access to legacy ICAD applications, and the ability to support their future business model with multiple global partners. He showed multiple screenshots of the Viewer to provide an understanding of its operation.

For KBE-5, William Moon from the Boeing Company provided an interesting exposition on their application for estimating the weights of struts. Here we see a KBE Application that uses multiple toolsets (Excel, ICAD, CATIA, and in-house FEA routines) to develop its final answer. The presentation noted the importance of using Object Oriented Programming; but of equal value was the slide showing the KBE approach used for developing the application.

For KBE-6, John Borger from the Timken Company presented their work using ICAD in Knowledge Based Manufacturing. In the past, we have seen many KBE applications presented at COE, but this was one of the few that were applied to manufacturing. Beyond designing a tool, their application (or combination of applications) is able to output the Bill of Materials, the Plant Routing, the Shop Traveler, the In-Process Drawing, as well as, the Machine, Operator, and CNC Gauging Instructions.

Again, we saw multiple systems (Databases, Shop Floor Execution, and CAM) linked together to automate the manufacturing process. John explained their "reverse engineering methodology" that they use to ultimately create the NC instructions and highlighted the benefits of avoiding the creation of hard tooling and the reduction in lead-time from 10 weeks to 6 weeks.

For KBE-7, Thomas Butler and Dale Mortimer from the Boeing Company presented their work for creating parts and assemblies that are tested in the Wind Tunnel. This presentation was well received by the attendees because it showed great productivity gains utilizing readily accessible tools within the V5 environment without a lengthy development phase. By using the tools within CATIA V5 (Parametric Seed Models, Power Copy, Knowledgeware rules, and Knowledge Expert checks), they are able to quickly respond to their customer's requirements and to readily make modifications during the test plan. They gave a live demo that was very impressive by showing the ability to morph and add parts within stress analysis rules, and then to take that data and send it to NC via a Windows-based interface.

They noted that they used consultants to migrate from V4 to V5, emphasized the importance of training to quicken the return on investment, and shared their metrics in the form of time savings as they went from V4 to V5.   

For KBE-8, Dennis Brauning from the Boeing Company did an excellent job of defining KBE and the pros and cons for using it. By using the metaphor of building a house, he was able to explain the construction of a KBE application in a simplified form that covered the main building blocks like attributes, behavior, rules, technology, and flexibility. Dennis then tackled some of the key questions when starting a KBE initiative like when do you purchase off-the-shelf technology versus developing your own, who does the development (the subject matter experts or the IT group), and what language do you use. We have had presentations in the past that have covered how one would start a KBE project; but Dennis' approach was unique and well constructed.

For KBE-9, Mike Twelves from Corus Automotive presented their Geometrical Optimization KBE Application using both Linear and Non-Linear Finite Element Analysis. Here again we are able to see the power of ICAD versus a CAD dependent KBE tool. The KBE application develops all of the initial FEA parameters for the vehicle's hood to meet new legislated safety requirements.  These parameters are then used by engineering to start their design. The application can complete 64 iterations in less than 40% of the time it takes for an experienced analyst to do a single iteration. Understand that they need at least 3200 iterations to satisfy legislation.

Although the details about the development of the application were riveting, what really caught my interest was the fact that this was a collaborative effort between Corus and Ford (Jaguar).  Mike noted that 90% of the intellectual property (IP) used to develop the rules and to do the analysis was supplied by Ford. His Company added some manufacturing rules and some data that they had; but the for the most part the IP came from Ford. He said that the original agreement with Ford had both Companies sharing development costs, IP, and exploitation rights.  He said it took as long to work out the legal aspects as it did to write the application (approximately 6 months).

Since Corus is a Tier 1 Supplier, I asked if his company could use this application with other Customers like Daimler. He said yes, explaining that they solved this issue by patenting everything that was co-developed. So now if Corus has a customer who wants to use the application, they simply have to pay pre-arranged royalty rights to Ford. The savings are substantially greater than the royalties, so Corus believes they will be able to use the application to get new business from other customers. Ford, on the other hand, is able to use the application anytime they like without paying Corus. I thought it interesting that Ford and Corus could work out the issues regarding IP and still allow Corus to do business with other customers.

For KBE-10, Francois Riche and Prasanna Katragadda from Dassault Systemes presented the transition plan from ICAD to V5. This presentation was at best disappointing and at worst a waste of time. The message from Dassault is that transition means hiring Dassault consultants to glean all of the knowledge out of the existing ICAD applications and then rewrite them in the appropriate V5 tool or combination of tools, which could include Knowledgeware, Knowledge Expert, Power Copy, Product Engineering Optimizer, Product Knowledge Template, Business Knowledge Template, CAA RADE, and/or VB Script. They showed Dassault's vision for using templates and "best practices" as the core technology and process for collaborative automation in V5.  Interestingly, they mentioned using MOKA methodology for collecting, structuring, and formalizing the knowledge needed for the application. And then they gave a demo of an aerospace machined-rib being converted from ICAD to V5.

Although I see the power of all these tools, and we have seen numerous presentations that support their usage, I continue to question their scalability and ability to address complex systems. I'm still waiting for the demo where someone is able to create an entire wing using these toolsets, works great on single parts and small assemblies. But I digress from the real issue, why can't Dassault Systemes find a way to leverage the existing ICAD applications in their KBE strategy? If their current strategy promotes linking multiple toolsets to achieve the final product; then why can't one of those tools be ICAD?

And last but not least, B&L-2 was presented by Joe Konecny from MSC.Software where Joe walked through a step by step process of creating KBE application within V5 using VB script.  This was very informative for those wanting to "drive" V5 with rules outside of the model. Joe provided all of the models and source code as an added benefit for those who wanted to go back and better understand what he presented.

All in all, I would say this conference was very beneficial for those following the KBE sessions.  Certainly the ICAD perspective was a bit of a surprise for some; but I thought it added another dimension to our perception of what is attainable using KBE. I hope my explanation at the beginning of this article helps in understanding the increase of these sessions versus V5 or VB specific sessions. We will attempt to do a better job of differentiating these sessions by a clearer description in the session abstract. I just hope that everyone will take another look at the concepts that can be learned from an ICAD presentation, rather than only the mechanics of the code. In reviewing attendee comments, most of the issues centered around the need for larger rooms as many sessions were standing room only and the abundant use of acronyms that left every one wondering what they meant. And as I review this article I find myself just as guilty with using acronyms; but have a deadline to meet and no time to correct my transgressions.