Maximum acceleration on a standard street

10.6 Modeling for Performance Evaluation

295 In this discussion on analytical modeling, a number of issues were raised: ■ What level of accuracy is needed? Analytical models can be used instead of physical models only when there is a high degree of confidence in their fidelity. ■ Are analytical models available of sufficient fidelity to give the needed accuracy? If not, then physical models are required. Often it is valuable to do both to confirm one’s understanding of the product. ■ Are deterministic solutions sufficient? They probably are in the early evalu- ation efforts. However, as the product is finalized, they are not sufficient, as knowledge of the effect of noises on the dependent parameters is essential in developing a quality product. ■ If no analytical techniques are available, can new techniques be developed? In developing a new technology, part of the effort is often devoted to generating analytical techniques to model performance. During a design effort, there is usually no time to develop very sophisticated analytical capabilities. ■ Can the analysis be performed within the resource limitations of time, money, knowledge, and equipment? As discussed in Chap. 1, time and money are two measures of the design process. They are usually in limited supply and greatly influence the choice of the modeling technique used. Limitations in time and money can often overwhelm the availability of knowledge and equipment. 10.6.5 Step 5: Understand the Physical Modeling Capabilities Physical models, or prototypes, are hardware representations of all or part of the final product. Most design engineers would like to see and touch physical real- izations of their concepts all the way through the design process. However, time, money, equipment, and knowledge—the same resource limitations that affect analytical modeling—control the ability to develop physical models. Generally, the fact that physical models are expensive and take time to produce, controls their use. However, the ability to develop physical prototypes of complex components has improved greatly since the mid-1980s. During this period, rapid prototyp- ing methods were developed. These systems use solid models of components to deposit materials or laser-harden polymers to rapidly make a physical model. The components made by some of the methods are actually usable in tests; others are only visual and usable to test fit and interference. You can’t BS hardware.