A methodology for system integrity in design
Thesis DisciplineMechanical Engineering
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
The early stages of engineering design are formidable terrain for human designers as well as their methodologies and tools. The large uncertainties, fluidity of concept, and frequent lack of quantitative relationships and data make the formal design methods difficult to apply. This project sought to develop a methodology that could help manage the process. The literature on the processes and methodologies engineering design were investigated, and interpreted with respect to a proposed generic design process. Existing methodologies are generally unsuitable for early design stages as they require relatively complete information and problem definition, which may be unavailable. There is need for methodologies that can support the designer even though the design uncertainties are high. A methodology called Design for System Integrity (DSI) was developed to support the management of uncertainty at the early stages of design. Essential features of this methodology are the ability to simulate system behaviour (i.e. functional modelling) based on quantitative as well as qualitative information, the accommodation of uncertainty of analysis and process variability, and the provision of multiple viewpoints on the design, including but not limited to performance, cost, and reliability. The methodology is computationally demanding, and requires a supportive interface between the user and the algorithms, so it was necessary to embody it in software. The software development was from the ground up using the Delphi programming language rather than an expert system or database shell. The software development was a critical enabling mechanism in transforming the DSI methodology from a philosophy and design principle into an embodiment that could be explored and scrutinised. The methodology was applied to a case study of dishwasher engineering design. One of the viewpoints in this model was wash performance, and the system was able to successfully simulate this parameter in a probabilistic manner despite the uncertainties. Other views including electric shock hazard, reliability and cost were also simulated. Being able to produce not only an outcome but also measure its likelihood, is valuable information for evaluating and managing the integrity of a design. The DSI methodology may be applied in the early designs stages when uncertainty is high, as well as in the more mature stages. In addition a catalogue system was developed for functional modelling. This permits the designer to select a device from a database, and inherit the default properties of that device. Importantly, the functional modelling may occur over several viewpoints simultaneously, and placing a device into one view will cause the system to automatically create and populate additional viewpoints in the background with other properties of that device. It is also possible to substitute one device for another, corresponding to the testing of various candidates, and the properties of the device are changed to those of the new one. The results demonstrate that it is possible to develop a methodology for modelling qualitative and quantitative performance at early design stages, so as to manage the integrity of a design.