Explanation in the engineering sciences : IBE, causal modelling, and (fictional) mechanisms in failure analysis and safety science

This dissertation investigates explanation in the engineering domains of failure analysis and accident causation. Explaining accidents and failures is important for learning, preventing, repairing and in some cases figuring out responsibility. The work presented here aims to clarify some widespread explanatory practices, strategies, and formats in these domains, using insights and accounts from philosophy of science. A series of case studies are used to highlight the structure and value of explanatory practices within engineering science. The crash of Air France 447 is used to highlight the way in which different explanations are evaluated when unequivocal evidence is missing. Analysing this case highlighted, among other things, that in failure analysis an important criterion driving the explanation is the ability to provide valuable redesign information. This insight is further exemplified in other cases such as malfunctioning Samsung Galaxy 7 mobile devices. Another interesting result stems from analysing how engineers explain the breaking of metals and other ductile materials. One key factor in explaining these ruptures is the presence of small voids, which when merging together to a sufficient extent provoke the disjointing of metal pieces. The explanatory strategy employed here is that of a mechanical story, despite voids or holes not being the sort of entity that can behave mechanically. An analysis of how these explanations work is presented together with a description of the cognitive benefits they provide. In addition to looking at concrete case studies, the dissertation also looks at the modelling practices used when analysing the causality of accidents. The way in which the causality of accidents is modelled has a direct impact on understanding and preventing accidents, as well as on the adjudication of legal responsibility. Different modelling strategies encode differing causal information, resulting in key differences in applicability, scope, effectiveness, and explanatory power. An account of the value of the most important modelling practices is put forward, regimented by precise tools from philosophy of science which highlight the relevant trade-offs each modelling approach offers. This analysis allows to make suggestions as to which modelling strategy 192 is best suited to achieve specific aims, such as preventing accidents or adjudicating legal responsibility. Finally, key philosophical and theoretical tools are employed in building a procedure which aims at improving the practice of Failure Analysis. This procedure structures the information contained in explanations such that it enables to find weak points, to compare competing explanations, and to provide redesign recommendations. In doing so it becomes a tool for assessing the quality of failure explanations. The procedure offers failure analysts a practical tool for critical reflection on some areas of their practice while offering a deeper understanding of the workings of failure analysis (by framing it as an explanatory practice) and redesign. Two case studies from Mechanical and Civil Engineering are used to exemplify how the procedure works/is intended to work.

Jaar van publicatie:2023

Toegankelijkheid:Embargoed