HOLISTIC SYSTEMS INTEGRATION

HARNESSING THE POWER OF Information

Keynote Speakers

The Structure and Dynamics of Complex Engineering Networks


Dan Braha, New England Complex Systems Institute & University of Massachusetts Dartmouth

Monday, 12 October 2009, morning


In this talk, we review recent advances in the use of Statistical Mechanics and Complex Networks Theory in understanding the structure, functionality, dynamics, robustness, and fragility of real-world large-scale engineering systems. We analyze information flow networks of large-scale product development (PD) projects in several major companies, and characterize their structure and dynamics. Strikingly, we observe that the patterns of information flows embedded in real world PD networks are similar to those discovered in other information, technological, and biological networks. We show that these properties make PD networks robust to errors and unplanned design changes. However, we identify specific information flow patterns that present "information bottlenecks," which can cause the activity to degenerate into an endless cycle of revisions. Informed engineers who identify and safeguard these bottlenecks can help protect their PD systems from escalation of effort and cost. Focusing engineering efforts on central tasks can improve process efficiency and boost performance significantly. Similar results are also obtained at the level of the design structure (e.g., open source software and electronic circuits product architectures) for which the design networks are composed of physical (or logical)


Dan Braha is a Full Professor at the University of Massachusetts, Dartmouth. He is also a Co-Faculty of the New England Complex Systems Institute (NECSI) in Cambridge, MA. Prior to that he was a Visiting Professor at the Massachusetts Institute of Technology (MIT), a researcher at Boston University, and a tenured Professor at BGU, Israel. Dr. Braha has published extensively in various prestigious journals, and edited or authored four books , including Complex Engineered Systems with Springer. He was the editor of several special journal issues, served as an Associate Editor for IEEE Transactions, as well as a member of the Editorial Board of various leading journals. He has also served on executive committees including the International Conference on Complex Systems (ICCS), and as chair at a number of international conferences. Dr. Braha is frequently invited to present his work in prestigious conferences and workshops, including the RAND Corporation and GE Global Research. Dr. Braha has contributed to a wide spectrum of research areas including complexity theory, systems engineering, design theory and methodology, artificial intelligence, data mining, and operations research. In particular, Dr. Braha has advanced the area of Complex Systems by introducing methodologies for understanding and improving the design, implementation, and dynamics of Complex Socio-Engineered Systems as well as exploring the interplay between natural and large-scale human-made systems. This research mainly focuses on understanding the functionality, dynamics, robustness, and fragility of large-scale systems. Much of his recent research is in the area of Complex Networks in general, and Dynamic Complex Networks in particular, with applications to large-scale social, economic, biological, and technological systems. More information on Dr. Braha's work can be found here.

 

 

Product Information Models to Capture and Manage Complex Product Designs

 

Sean Callahan, The Boeing Company

Monday, 12 October 2009, morning

As companies advance to greater design automation and virtual product modeling, they require specific content in their product information models. This talk defines the research and development issues involved in defining that content and managing it over the product lifecycle. The product design information under consideration ranges from the standard purview of systems engineering—requirement, functional and logical systems designs—to physical (geometric) CAD, and manufacturing planning. Extension to post-sale product support, part ordering, and other areas is also possible.
Features of this design-information landscape are discussed, including what it means to have a sufficiently complete, computer-sensible design representation. While completeness tends to dominate extant thinking on product design, there is a great opportunity to greatly increase design companies’ productivity by understanding how common design information can be shared between those more complete representations of product design configurations.
Each design of a complex product is represented by a version of a product structure. Each version is a new design and, except for the first, is created by modifying an existing design. Each subsequent version supersedes the preceding design. A variant is a particular sequence of such versions: variants provide an evolving design history. The versions within a variant are mutually exclusive in the sense that only the latest one is up-to-date, while all others have been superseded by subsequent versions.  On the other hand, all of the latest versions for all variants are independent designs that do not supersede or depend upon each other.  Each may be offered and modified independently of the others.
Finally, another aspect of design-information sharing is the ability to have multiple changes applied independently against a particular version, and then to allow such changes to be merged to develop new, complete product designs.
Current commercial products do not offer adequate solutions to these information representation problems. There are also some additional problems that stem from the way that the product lifecycle management (PLM) and design tool industries are structured.   This and Boeing’s solutions to some of these fundamental challenges are the final focus.
While this talk will focus on families of complex products, marketplace demands for lower price and greater product variety will eventually drive these information modeling requirements down to companies who develop simpler products.

 

Sean Callahan is a Technical Fellow working for The Math Group within the Boeing Research & Technology organization. He received his PhD in physics from the California Institute of Technology.  He has been at Boeing for twenty-one years and his research focuses on developing product information models to support design automation, knowledge-based engineering and computer aided engineering.  This has involved developing and patenting concepts that allow sharing of versions and variants of reusable design hierarchies (product structures). The result is that arbitrarily complex, fine-grained models of product designs can be developed and managed over a product family’s lifecycle those design can share common design data to represent common design content.  He is also focusing on integrating complex component based simulations with the system designs of those product families. He has previously focused on high performance visualization, solid modeling, and rule-based, semi-automated design.  Sean was the design data architect for a Boeing-developed system design tool being used on the 787 program, based on Siemens PLM’s product SLATE, to capture a computer sensible representation of the system design.  Sean has served as an associate editor for the Journal of Computer Science in Engineering (JCISE) in informatics and has published the results of his research in several publications.

 

 

Integrating the Product, Process, and Organization Views of Complex System Development

 

Steven Eppinger, Massachusetts Institute of Technology, Sloan School of Management

Tuesday, 13 October 2009, morning

The application of design structure matrix (DSM) methods has devolved over the past 30 years from parameter-based models, to process models, to organization- and system-architecture models.  Each of these modeling domains offers a unique and important perspective on the challenges faced in complex technical development projects, and we have learned valuable lessons from these perspectives over the years. Of course, in practice it is very difficult to model all of these perspectives for any one project.  I believe that wider industrial application of DSM will be enabled by simpler ways to model these several important domains in an integrated manner.  This presentation will review what we have learned in these domains and suggest ways they can be further integrated.

 

Steven D. Eppinger is Professor of Management Science at the Massachusetts Institute of Technology (MIT) Sloan School of Management.  He also holds the General Motors Leaders for Global Operations Chair and has a joint appointment in MIT's Engineering Systems Division.  Prof. Eppinger served as Deputy Dean of the MIT Sloan School from 2004 to 2009.  He has created an interdisciplinary product development course in which graduate students from engineering, management, and industrial design programs collaborate to develop new products. He also teaches MIT's executive programs in the areas of product development and complex project management.  Prof. Eppinger has co-authored a leading textbook entitled Product Design and Development published by McGraw-Hill.  Currently in its fourth edition, the text is used by hundreds of universities around the world.  Dr. Eppinger's research is applied to improving complex design processes in order to accelerate industrial practices.  He is a pioneer in development of the Design Structure Matrix method for managing complex system design.  He has authored over seventy articles in refereed academic journals and conferences.  He has received MIT's Graduate Student Council Teaching Award, the Sloan School's Award for Innovation and Excellence in Management Education, and twice has received the ASME Best Paper Award in Design Theory and Methodology. Prof. Eppinger received S.B., S.M., and Sc.D. degrees from MIT's Department of Mechanical Engineering before joining the MIT faculty in 1988.