The heart of any aerospace engineering design system are the databases that store the geometric details of the products being considered. We begin with this consideration rather than the tools used for concept design since we believe that an understanding of the geometry environment to be used is the most important first step in tool selection. At the point where manufacture begins, this is now almost universally one of the mainstream CAD packages such as Dassault-Systeme’s Catia™ or Siemens’ NX™. Here we use Rhinoceros™ for initial concept definition, which is a very low cost tool and ideally suited for parametric programming, and Solidworks™ for detail design, since this is a modest-cost detailing option but one that still offers very significant functionality. While there are some public-domain CAD codes capable of production-level drawing, the CAD description is so central to detail design that the use of a commercial code always seems to us to be justified. Commercial codes also interface well with the range of modern rapid manufacturing technologies being used to source components for the Southampton air vehicles.
Solidworks™ is, of course, capable of parametric modeling of the components being designed but, in common with all today’s generation of CAD tools, is not sufficiently intelligent to allow a complete parameterization of every dimension and choice in the design of a UAV: there are simply too many interactions between parts, surfaces, and functions for this to be possible. As soon as any significant change is made to several parameters at once, the models typically fail to rebuild. This causes a fundamental difficulty in realizing a decision support vision where the aim is for rapid and radical changes to be made possible with strictly limited human intervention. Accordingly, we have restricted the use of a fully parametric
Figure 9.1 Outline design workflow.
Figure 9.2 Analysis tool logic.
approach to those places where it is convenient/most important; there we use Rhinoceros™ and have had to accept that this implies a subsequent and more manual detailing phase for almost all aspects of the design after concept decision making and prior to manufacture, and where only limited parametric approaches can be taken. This means that analysis codes have to be able to work with both a fully detailed and a nondetailed set of geometries - the former to allow final design checks on stressing and aerodynamic performance, and the latter when carrying out design searches to achieve the best design balance. Although research is currently under way in several universities and companies to address such shortfalls, it seems likely that some manual, CAD-based, detailing, based on engineering experience and judgment, will still be required before production, for many years to come. The assumption made here is that such work does not impact significantly on the overall design balance being struck by the main decision support system which uses a simpler fully parametric geometry.