NSF Workshop on Decision-Based Design: A Position Paper on the Modeling of the Design Process

Abir Z. Qamhiyah

Abstract

Mechanical engineering design is a creative activity supported by knowledge, skills, analysis, and testing. This paper presents an outline of my view of the nature of the design process and the activities that designers perform during the process. Solely on the basis of that perception, I propose that there are at least two distinct approaches to the modeling of the design process. In the first approach, the combined effects of the Adiscrete design activities (such as creation, generation of alternatives, selection, etc.) are modeled to represent the status of the design process at a given instance. In the second approach, the design process is modeled through monitoring the history of the Acontinuous design activities (such as information processing, collaboration). Decision making is one of the continuous activities in the design process.

The Nature of the Design Process

Mechanical engineering design is a creative activity supported by knowledge, skills, analysis, and testing. When attempting to formalize the design process, I believe that it is necessary to distinguish between the following two types: (1) incremental design and (2) revolutionary design. I use the term Aincremental design to refer to the design process in which the design concept and the design configuration of the system are provided as a starting point for the designer. The requirement in this case is modifying the design of the system=s components to improve its performance, facilitate its assembly, reduce its maintenance needs, etc. However, I use the term Arevolutionary design to describe the design process that must begin with conceiving the design concept and then imagining alternative configurations. Revolutionary design places more emphasis on the creativity of the designer or design team. This emphasis diminishes in incremental design, which leads to defining what the designer experiences as an analysis-based optimization process. The parties directly involved in a design process include the design team, the organizations cooperating in the product development cycle (marketing, design, manufacturing, service centers, and recycling), and the customer. Indirectly, all life forms and the environment are involved in the design process. However, these are represented by governments and policies. Depending on the product, the impact of the outcome of the design process could exceed the design team, the product development organization, and the customer, influencing our environment with all of its life forms. I refer to this scenario as the Aaction/reaction chain of the mechanical engineering design process. Such a diverse chain implies that the interests of the parties involved in the process are as diverse and at times conflicting. Despite the complexity associated with the design process, the simple fact remains that it must result in the definition and realization of the end product. This end product cannot be defined without streamlining activities, resolving conflicts, and consolidating distributed efforts. Furthermore, the product cannot be realized if it is too complex to be fabricated or if it violates the laws of nature governing both the state of its existence and its internal and external mechanisms of interaction.

The Activities Designers Perform During the Design Process

Some of the activities designers perform at different intervals in the design process include (1) identification of the need for a system to perform a given function, (2) conversion (mapping) of the identified facts in relation to the performance of the desired system into system specifications, (3) creation of the design concept, (4) generation of alternative design configurations, (5) verification of the conformance of the alternative design configurations to the design concept, (6) elimination of some of the alternative design configurations, (7) modification of the remaining alternative design configurations, (8) definition of the form of the components and subcomponents (detailed design) of the remaining design configurations, (8) evaluation of the alternative design configurations using analysis, testing, simulation, prototyping, etc., (9) ranking of the alternative design configurations, and (10) selection of a single design configuration for refinement. To ensure the progression of the design process, designers will continuously need to (a) manipulate information (storing, retrieving, acquiring, extracting, transmitting, processing, etc.), (b) collaborate with the parties involved in the design process and the product=s overall development cycle (streamlining of efforts, resolution of conflicts, and consolidation of distributed efforts), and make decisions pertinent to every activity in the design process.

Modeling of the Design Process

On the basis of my perception of the nature of the activities that occur in the design process, I believe that at least two distinct approaches could be used for modeling the process. The first approach models the combined effect of the discrete activities that occur in the design process, representing the status of the process at a given instance. The second approach models the process by monitoring the variation in the continuous design activities during the entire process. Discrete design activities are performed at intervals in the design process, while continuous design activities are common to all intervals in the design process. The activities labeled (1) to (10) in the previous section are examples of what I call discrete activities in the design process, while the activities labeled (a) to (b) are examples of what I call continuous design activities. If the design process is modeled using the first approach, then the model will be in the form of a multidimensional function. The axes of that multidimensional space (the function variables) must be an abstraction of the discrete activities that designers perform during the design process. In that case, the multidimensional function is the weighted sum of the abstractions of the design activities, with each weight multiplied by a corresponding function of a time variable. The weights are dynamic (vary in value) during a given process. The time functions associated with the weights are also dynamic (variable formulation). These time functions will regulate the duration and pattern of influence of a given weighted design activity at a specific interval in the design process. Although the tasks of assigning the variable values to the dynamic weights and the variable formulation of the dynamic time function are complex, developing the metrics that quantitatively represent the discrete design activities, which are qualitative in nature, is the most difficult obstacle to be overcome in the first modeling approach. If the design process is modeled using the second approach, then the model will be in the form of a path that the process traces as a function of time. For example, the history of information manipulation activities during the design process can be traced in the form of a list of information content, followed by the information manipulation activity that is performed on the information content, followed by the resulting information content, etc. Similarly, the history of the collaborative efforts that occur during the design process can be traced in the form of a list of the situation, collaborative effort, the new history of the situation as a result of the effect of the collaborative effort, etc. Application of the same modeling principle to the decision-making activity results in tracing its history in the form of a list of the alternatives encountered at any instance, the ranking methodology, the options selected, etc. In my opinion, of the previous three examples outlining the method of application of the second modeling approach, the example that traces the history of information manipulation seems to be the most feasible. I can justify this viewpoint by pointing out that the measurement and description of information content intuitively appears to be less complex than the development of metrics for the description of collaborative or decision-making situations and activities.

Closure

One of the methods for modeling the design process is through the activities performed by the designers during the process. In this brief paper, I have presented my position in relation to this method of modeling the design process. Two approaches that fall under the method of modeling the design process through its activities have been marked out in this paper. In the first approach, the combined effects of the Adiscrete design activities are modeled to represent the status of the design process at a given instance. In the second approach, the design process is modeled through monitoring the history of the Acontinuous design activities. Decision making is one of the continuous activities in the design process. Despite the complexity associated with the first modeling approach, I believe that in terms of generality, this approach is more promising. I also believe that application of the second modeling approach to the information manipulation activity is the most feasible.