A Fractal Representation for Systems
Integration of Process Knowledge into Design Support Systems
To facilitate the design of evolving systems, tools are needed to capture all performance issues and evaluate design ideas and proposals quickly during the conceptual stage, so that better designs can be generated. By using such tools, engineers can quickly identify and understand how their design decisions impact and are impacted by choices made concerning other components in the system. Thus, rational design decisions will be made during conceptual design, minimizing if not eliminating the
... eliminating the need to address design problems during implementation. This paper presents a methodology, based on axiomatic design theory, for constructing a system architecture for complex systems that standardizes the classification of functions and modules used to represent a system. This is important for several reasons, including capturing the performance requirements and components of the system in a logical, coherent, and comprehensive manner, facilitating communication between engineers and managers on a large design project, and providing good technical documentation of the design decisions made and the reasoning behind them. A system architecture is applicable to systems of any size, including systems that are subsystems of a larger system. Thus, the decomposition of a system follows the same general pattern and layout at each level of the design hierarchy where the parent design parameter is a "system" in its own right. Hence, the overall system is represented in a recursive manner throughout the design hierarchy. This representation has several advantages, the most significant of which is that the design of individual subsystems can be generated by different teams of engineers on a large design project, while the functionality of the overall system as well as the interrelationships between the different subsystems are thoroughly and consistently represented. manageable tasks to accomplish. These teams are supervised by one or more layers of management, which have the difficult task of coordinating the activities of the different teams and providing a bridge of communication as and when necessary. Typically, each design team will try to optimize its design based on its assigned tasks and constraints. It is very easy, however, for a designer to be unaware of other designer's decisions which will have a negative impact on his/her design. Thus, when each of the individual designs are put together, it is quite common to discover that the overall system does not function as intended. This results from the fact that each individual subsystem design has been locally optimized, without accounting for the complex interactions between the subsystems. Thus, a change to one portion of the design can negatively impact other portions unintentionally, because no framework exists to trace the impact of the design choices between the task divisions. The main source of difficulty in designing large systems is that, in most cases, good representations of the design either do not exist or are not used to their full potential. [Söderman, 1998] . Hence, it is extremely important to have a methodology to trace the impact of design decisions on both a local level as well as a system level, since the real goal of the design effort is to optimize the performance of the system, which does not necessarily mean optimizing the performance of each component. This issue can be addressed by applying axiomatic design theory to the design of complex systems. This approach generates a system architecture, which captures the hierarchical structure of the functional requirements (FRs), design parameters (DPs), and constraints (Cs) of a system. Using the design axioms, the quality of the design can be evaluated by means of design matrices, so that potential problems of a particular design can be detected and addressed during the design process. Axiomatic design is reviewed briefly in Section 2.