Most of the efforts in optimisation so far have been focused on the development of novelor the improvement of existing numerical methods for an effective computation ofoptimal solutions. Particular attention has been put on balancing multiple conflictingobjectives, handling the interaction between different disciplines, reducing computationalcost and managing uncertainty. Nonetheless, specific issues of this design methodologystill remain to be properly addressed. In this research, attention is concentrated onadvancing engineering optimisation as a tool for design exploration. The work is put inthe context of conceptual aircraft design.The overall aim of the present research is to develop a methodology that allows thedesigner to effectively conduct an exploration and analysis of alternative design solutionsvia a set of methods that can be used separately or conjointly.The initial part of the thesis introduces two novel methods for assisting the formulation ofan optimisation problem, which generally is assumed to be given a priori. Nonetheless,the correctness of the optimisation statement, which is not addressed by establishedoptimisation methods, turns out to be decisive for the feasible design set determination.The designer is thus provided with an adaptive formulation of functional and designvariableconstraints, which allows the exploration of further promising solutions initiallynot contained in the feasible design set. Meaningless results or the loss of importantsolutions can therefore be partially avoided.In a second instance, attention is focused on the visualisation needs for designexploration. A suitable visualisation methodology has been developed to make the largemultidimensional results of complex design optimisation procedures fully readable andexplanatory. This is achieved by integrating advanced visualisation techniques which provide the designer with diverse perspectives of the data under study and allow him/herto conduct a number of analysis tasks on it, without the need to be an expert in numericaloptimisation methods.Last, but not least, a methodology to address conceptual design change problems isproposed. The decision-maker is enabled to formally state the new design requirementsand priorities introduced by the conceptual change via an adequate problemreformulation. All the data previously collected can thus be re-used and exploited to drivean effective exploration of alternative design solutions through design space regions ofinterest.The evaluation of the proposed methodologies has been carried out with a number of testcases. Analytical examples have been used for the assessment of effectiveness, whereascodes representative of aircraft sizing procedures have been adopted to evaluate themethodologies functionality. A visualisation user interface prototype has also beendeveloped for demonstration and evaluation purposes.
展开▼
