Decision Based Design Open Workshop - 9th Face-to-Face Workshop

Design is a process that proceeds along multi-axes. Design problems require decomposition across at both the system level and the abstraction level. The degree to which a design activity is decomposed depends on many factors including the nature of the product and the team of people working on the activity.

Design involves activities that are both intuitive and logical. Logical methods are historically more popular, well-defined procedures, and based on first principles. However the intuitive aspects are also important and are used as the basis for formalising the alternative generation step of design.

While design is a highly complex process, it is a surprising how little time is spent on conceptual design. Practising engineers have little experience using the formal methods based on intuitive principles to generate alternatives. We can attribute this to the fact that generating new ideas accounts for only 20% of the design activities that engineers are called upon to perform. Creativity is the bedrock of conceptual design and is not enhanced through formal methods based on logic. Conceptual design is a stage of the design process that is over-looked in the efforts to create formal methods. Generating alternatives, one of the key activities of conceptual design, is critical to design success. It is not clear how DBD aids in conceptual design.

Decision theory is a normative process concerned with making trade-offs under uncertainty. Some of these issues are intractable. That is why we attempt to quantify all design decisions. What we discover is that we are unable to optimize all attributes of a design at the same time.

Design engineers are trained in normative techniques, like optimization. But many aspects of design decision-making are not solvable by optimization. This leads design engineers to be uncomfortable with the qualitative and subjective aspects of decision-making.

Decision theory canÕt replace design expertise. However, it is a way to enable designers to deal with tradeoffs usually made further down the line. Decision theory provides a framework for structuring design processes so that some of the predictable difficulties are overcome. The goal is to aid us in focusing design efforts on the area where we will accrue the greatest payoff. Our goal should be to use decision theory techniques appropriately.

Table 1. Benefits and Limitations of Decision-Based Design Throughout the Design Process

Design Stage	Limitations	Benefits
Customer or Designer Need Expressed		Separates true objectives from superflous. Defines true tradeoff range Avoid biases, inconsistencies, paradoxes
Creative Synthesis or Alternatives	Cannot replace creativity Cannot replace engineering expertise	Frees designer to think in terms of function rather than form Initial filter for materials, configuration, manufacturing options based on attribute and range definition
Analysis	Cannot define analytic constraints (strength of materials, kinematics, structural analysis, etc.)	Identifies relevant analysis, based on attributes and range Indicates where experimentation or other effort is worthwhile to improve analytic model
Tradeoff Evaluation	Cannot determine which tradeoffs are technically feasible Cannot define Pareto Optimal frontier Cannot provide solution procedures, optimization algorithms	Rank orders alternatives Identifies alternatives worth further analysis Determines desirable tradeoffs Focuses effort where payoff is greatest Defines objective function for optimal solution
Decision Making under Uncertainty	Cannot remove uncertainty	Models uncertainty Includes effect of uncertainty on rank order of alternatives Avoids irrationality under uncertainty Determines when it is worthwhile to gather more information vs. act, via EVSI
Team Design	Does not resolve Arrow's impossibility Theorem	Provides framework for obtaining preference information Communicates team preference information Breaks decision problems into components on which consensus can be reached

Many design methods are proposed that are not fully validated. Some of these methods give absurd results. Worse, we often use methods improperly because we believe all methods can be used all ways.

We must develop a more rigorous habit of validation for our design theory research. A set of guidelines was suggested.

Design methods need to be understood in historical perspective. That is, the philosophy of the period drives the perspectives proposed by researchers of the day. We can learn much about design theories and methods by understanding the changes of the philosophy of design through the last few centuries.