The Stages of Design
Despite all the advances that have been made in computational analysis, it must be remembered that the fundamental hallmark of design is not analysis but synthesis - the choice of appropriate mechanism types, power sources, setting of dimensions, choice of features and subcomponents, selection of materials, and manufacturing processes - all these are acts of synthesis, and it is the skillful making of decisions in these areas that is the hallmark of the good designer. Of course, designers use analysis all the time, but design is about decision making, and analysis is, by contrast, an act of gaining understating but not of making decisions. Moreover, to be a good designer, the most often cited personal prerequisite is experience - this view is backed up by many observational studies in engineering design offices.
So, even though design decision making is commonly preceded by a great deal of information gathering and analysis, and although the gathering of such information, often using computational models, may be a very skilled and time-consuming activity, it should be made clear that whatever the cost, this remains just a precursor to the decisions that lie at the heart of design. Our researches have primarily addressed the tools that help support this decision-making process. The process may be thought of as being built from four fundamental components:
- 1. Taxonomy. The identification of the fundamental elements that may be used, be they gears, servos, airfoils, and so on;
- 2. Morphology. The identification of the steps and their order in the design process;
- 3. Creativity or synthesis. The creation of new taxonomies, morphologies, or (more rarely) fundamental elements (such as the linear induction motor);
- 4. Decision making. The selection of the best taxonomy, morphology, and design configurations, often based on the results of much analysis.
Perhaps the simplest and most traditional way of representing the design process is as a spiral, see Figure 1.2. The idea behind this view is that design is also iterative in nature with every aspect being reconsidered in turn and in ever-more detail as the design progresses. It begins with an initial concept and constraint review that attempts to meet a (perceived) customer need
Figure 1.2 The design spiral.
specified in relatively few major requirements. For an aircraft, this might cover payload, range, speed, and anticipated cost; for an engine, it might be thrust, weight, fuel efficiency, and cost. This phase of the design process is often called concept design. It is then traditionally followed by preliminary design, detailed design, and the generation and verification of manufacturing specifications and tooling designs before production commences. For major aerospace systems, once the product is operational, a continuing “in-service” design team takes over to correct any emerging problems and deal with any desired through-life enhancements. Finally, decommissioning and waste disposal/reuse must be considered. In Rolls-Royce, for example, this is called the Derwent process, and is characterized by significant business decision gates at each stage.
In our UAV designs, mainly because the products being designed are nothing so complex as an airliner or a jet engine, the preliminary design phase is not separated from the concept stage - all design variables are considered to lie in the concept phase unless they live only within the CAD-based detail definition process, in which case they are considered detailed design variables. Similarly, considerations of manufacturing methods are not separated from the detailed design work - rather they are closely integrated so that the designs take maximum advantage from the manufacturing systems employed - here the focus has been to use modern rapid prototyping and numerically controlled machining so as to reduce the amount of manual input during manufacture to the absolute minimum. As we shall see, these approaches reduce, but do not eliminate, the dislocations that arise when a design moves from one such stage to the next. To support a truly seamless design process, the designer must be able to move between concept thinking and manufacturing details with ease whenever needed, without worrying that any changes made will make masses of existing design work redundant. Parametric, CAD-based detail modeling can help in this respect, but it is difficult to entirely remove them. The approaches described here represent our best attempt to mitigate such problems.
Nowadays, even the most traditional manual approach to design will probably make use of some computing facilities, but it will generally not draw on modern search and optimization strategies, knowledge-based systems, or grid-/cloud-based computing. It will, in all probability, make heavy use of extremely experienced design staff and detailed experimental activities. At the other end of the spectrum, as much of the design process is automated as possible, high-level value metrics are used to balance competing concepts, formal optimization is used to generate new combinations of parameters, and data/knowledge stores are maintained electronically and centrally shared. Table 1.1 illustrates the kind of variation that can be encountered in organizations with different levels of design system maturity. Our aim is to operate at the highest maturity level in this spectrum. Before proceeding, however, it is useful to briefly step through the more commonly referred to stages of design in a little more depth. As just noted, in our UAV work the concept and preliminary stages are combined, as are the detailed and manufacturing stages. Formal decommissioning, over and above simply using the University’s normal recycling systems, is not considered at all.
-  Widely distributed computing systems networked together to form a single, often remote, resource.