History of the DOME Project
DOME Project
The DOME (Distributed Object-based Modeling Environment) project, initiated at the MIT CADlab and first published in 2000, is focused on creating usable design tools for exploring large-scale, systems-oriented design problems while accommodating the physics, heterogeneity, exceptions, uncertainties, context-dependencies, and rapid change associated with real-world problems.
DOME provides a common Internet-based computational infrastructure that makes the exchange and coordination of related parameters between different simulations a transparent process. DOME enables a new distributed, declarative, emergent system definition process that does not require an explicit master integration model. DOME allows design teams to rapidly create integrated design models in a spontaneous, ad-hoc manner, allowing them to focus on developing and analyzing design scenarios rather than being constrained by a rigidly defined integrated design environment.
DOME's implications for changing how products and services are designed may rival how the World-Wide Web (WWW) transformed the process of disseminating information. In contrast to the web, which allows users to both navigate and contribute to a network of static documents, DOME allows users to utilize, and contribute to, networks of collaborating dynamic computational processes or simulations. In simple terms, one can imagine DOME as a simulation-web that allows users to join models and create distributed spreadsheet-like simulations that can span the globe. For example, an industrial designer might publish into the DOME environment key parameters of a geometric CAD model, much like one can publish a web page in the WWW. Then an engineer, working in another location, might subscribe to parameters offered by the CAD model and link them to parameters in their own cost analysis spreadsheet. Consequently, if either the industrial designer makes changes in the CAD model or the engineer makes parameter changes in the cost model, appropriate changes are automatically triggered in the respective software applications. DOME automatically solves relationships between distributed simulations and synchronizes the execution of parametric models, regardless of what software application they are implemented in. System designers need only focus on defining integrative relationships between simulation parameters without worrying about the order in which relationships must be executed to correctly coordinate the underlying simulations. Further, the tacit system knowledge embodied in relationships defined by integrators is not revealed to the disciplinary applications to which they subscribe.
In practice, large-scale integrated simulation is often deemed implausible and many researchers have noted that difficulties arise because of the large scale, complexity, rate-of-change, heterogeneity, and proprietary barriers associated with comprehensive product design simulation. Since integrated analytical models are difficult to change, some researchers believe that integrated systems can actually have negative effects by prematurely freezing the system level architecture of a product.
While there are many issues to address, we believe that a root-cause of these difficulties is that the method for defining and building integrated systems does not suit the product design and product evolution process. Integrated simulation environments apply explicit procedural system definition techniques, while design synthesis can be viewed as a distributed, emergent, system building process. The DOME project has developed a new World-Wide-Simulation-Web concept for synthesizing emergent computational simulations. DOME is a technological innovation that enables an emergent model integration process in a WWW-like simulation environment. This new integration process overcomes the difficulties long associated with traditional efforts to create large integrated product simulations. A federated solving architecture allows individuals to declare local relationships between sub-system parameters without knowledge of the overall system structure.
A publication describing the DOME project received the 1998 ASME International Design Engineering Division, Design for Manufacturing Committee IBM Best Paper Award and there is an industry driven working group that meets quarterly to share their DOME experiences. The information provided in the DOME Project section draws upon ideas from a number of published academic papers (available through CADlab publications). Some key references are cited below.
Pahng, Francis; Nicola Senin, and David R. Wallace, "Distributed object-based modeling and evaluation of design problems", Computer-aided Design 30, pp. 411-423, 1998.
Abrahamson, Shaun; David R. Wallace, Nicola Senin, and Peter Sferro, "Integrated Design in a Service Marketplace", Computer-aided Design 32, pp. 97-107, 2000.
Borland, Nicholas; and David R. Wallace, "Environmentally-conscious Product Design: A Collaborative Internet-based Modeling Approach", Journal of Industrial Ecology 3, 2000.
Senin, Nicola; David R. Wallace, and Nicolas Borland, "Distributed Object-based Modeling in Design Simulation Marketplace", accepted for publication, ASME Journal of Mechanical Design, 2000.
Wallace, David R; Elaine Yang, and Nicola Senin, "Integrated Simulation and Design Synthesis", accepted for publication, Journal of Concurrent Engineering, 2002.
Xia, Hao-Xiang, "Toward Emergent Problem Solving on the World-Wide-Simulation Web", MIT CADlab Report, 2002.
What is DOME?
There are three key aspects of the DOME concept and DOME3 implementation.
A new philosophy of how to build integrated networks: Just as the hypertext concept underlies the WWW's model for building information networks, the concept of distributed, declarative, emergent system synthesis anchors DOME as our vision for a World-Wide-Simulation-Web (WWSW).
An architectural representation that supports the integration philosophy: In order to realize the hypertext concept in the WWW, a number of enabling protocols and standards were needed, such as http, html, xml, etc. Likewise, DOME3 exploits many of these same protocols and standards, but also includes an object representation necessary to support the WWSW and system simulation.
A software implementation to allow the use of the integration philosophy: In order to use the hypertext concept/WWW several applications are required: authoring tools to generate content (e.g., DreamWeaver, FrontPage); tools to move html files to web accessible computers (e.g., FTP applications), tools that serve html to remote users in the WWW (e.g., Apache, iis), and tools to view/browse the content of the WWW (e.g., Safari, Netscape, Internet Explorer). DOME3 provides a precisely analogous set of software applications: build; deploy; server; and run.
Traditional System Design
There have been many efforts to develop integrated modeling environments in both academia and industry. However, it does not appear that the process for defining integrated systems has been questioned. Available integration tools employ a wide variety of architectures, data models and communication technologies, but all appear to rely on some form of a consolidated explicit description for the complete integrated system model. However, we believe that the significant difficulties associated with scale, complexity, rate-of-change, heterogeneity, and proprietary barriers arise because explicit system definitions are required.
We propose that large integrated simulation efforts have had limited success because the methods used for structuring and building integrated simulations do not support the nature of design activities and the product evolution process. There is a mismatch between the nature of large-scale design synthesis activities and integrated simulation model building techniques.
The 'V' system engineering approach and other similar methods patterned after IDEF (Integration DEFinition for Function Modeling) are widely used in industry for structuring integrated design simulation environments. Starting from the top left of the V, the system is first fully defined or scoped through a top-down, cascading decomposition of subsystem requirements. Then, individual subsystems are constructed, individually tested, and integrated by working back up the V. When one reaches the top right of the V, the system is complete and ready for deployment or use.
This is an explicit, procedural process for defining and building systems. The method assumes that the definition of the overall system and control of its execution is consolidated. The system is usable when it is deployed in a complete form and as a result the architecture of the system is fixed. Designers can set the values of parameters but not change the 'plumbing' of the integrated system simulation. It is often very difficult and expensive to change the underlying system architecture.
Although explicit procedural system definition methods can be readily applied to design or simulation synthesis problems that are addressed by a single individual or small number of individuals, when addressing larger products (even ones that are quite stable and mature) organizations struggle to explicitly define and order their functional system model.
Distributed, Declarative, Emergent System Design
In contrast to traditional system definition techniques, design practice involves rapid synthesize/test/evaluate cycles. This may occur in a bottom up, top down, or middle out fashion. During these cycles many discoveries are made and the design or system undergoes many fluid transformations. Many participants interact and make local decisions during the process, and the collective result defines a complex whole.
These characteristics of design synthesis are all consistent with the properties of emergent systems. In fact, most complex systems occurring in nature or involving human behavior are emergent in nature. The economy is often cited as an example of an emergent system.
Managing the development of emergent systems using the almost universally-adopted, strict, explicit, consolidated approaches is very inefficient. In contrast, effective organizations can arise through an emergent process within appropriate environments where individuals can understand the implications of their actions on overarching goals. Therefore, the goal of DOME is to provide a new integrated simulation building technology that allows system simulations to emerge throughout the design activity. The traditional difficulties associated with scale, complexity, rate of change, heterogeneity, and proprietary information can be resolved through a distributed, declarative, emergent simulation synthesis process that, without consolidated control, can maintain parametric consistency between the distributed simulation models.
Distributed: control and execution of models and resources are in many locations.
Declarative: one can define local interactions with a larger system without requiring detailed knowledge of the entire system or the system-level causal implications of the local connections that are being made
Emergent: a system of models maintain parametric consistency without an overall system-level controller model or meta model.
Protecting Proprietary Models
Much of the integrated modeling community has focused on data model integration. The goal is to achieve a data model that allows everyone to interpret and reconstruct a complete representation of the problem in different environments. In other words, integration is achieved by exchanging models or having shared access to a common data model. While standardization is necessary at some level, it is very difficult to achieve this at a modeling level. As such, the most frequent strategy is to standardize on specific proprietary tools so individuals can exchange files, often leading to problems with unnecessary rework, the inability to choose suppliers with flexibility, and the need to share detailed proprietary models.
DOME focuses on parameter or data-type level integration by creating an infrastructure so it is easy for people to make connections and maintain consistent states between different simulations, regardless of the type of software tool, simulation or data model that resides behind the services. Individuals can provide Internet accessible service interfaces to simulations while maintaining full ownership of the underlying models. When necessary, entire files or data models can also be bundled as DOME parameters.
A Product Development Service Marketplace
DOME is a modeling infrastructure intended to create a global community, or marketplace, of individuals offering access to simulation services related to their own specialties, much as the WWW has enabled world-wide access to information. Within the envisioned simulation marketplace, it will be possible to quickly create and holistically assess technology systems from many viewpoints, allowing individuals to design and understand complex technology systems.
Participants and product development organizations are empowered to publish their geometric design, CAE, manufacturing, or marketing capabilities as live services that are operable over the Internet. Product developers, small or large, can subscribe to and flexibly inter-relate these services to embody a distributed product development organization, while simultaneously creating system models that allow the prediction and analysis of integrated product performance.
We believe that product development services will become commodities, much like many component-level products are today. It will be possible to rapidly interchange equivalent design service providers so that the development of the product and the definition of the product development organization become part of the same process. Computer-aided design tools will evolve to facilitate the publishing of live design services.
Funding
The DOME project is grateful for the research funding provided by its sponsors. The initial seeds funds for the project were provided in 1996 by the Alliance for Global Sustainability (AGS) and the Leaders for Manufacturing Program. Ford Motor Company and the Center for Innovation in Product Development have been very strong supporters since 1998. Additional support has been received from LG Electronics, and Bose.