Sensitivity and optimisation of damage in forming processes
Loading...
Date
2024
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Die Herstellung technischer Komponenten in der heutigen Industrie erfordert ein großes Maß an Wissen, um Teile mit den gewünschten Eigenschaften zu produzieren. Dieses benötige Wissen umfasst das Materialverhalten, den Herstellungsprozess, sowie die Belastung, welche das Bau-teil im Einsatz erfährt. Eine Kombination dieser Faktoren ermöglicht es dem Ingenieur, ein Pro-dukt zu entwerfen, welches den Anforderungen der jeweiligen Anwendung gerecht wird. Elasti-sche und plastische Materialeigenschaften sind heutzutage reichlich erforscht. Schädigungs-mechanik stellt jedoch weiterhin eine anspruchsvolle Herausforderung dar und nimmt eine wich-tige Rolle in der aktuellen Forschung ein. Schädigung wird als eine Ansammlung von Defekten auf der Mikroskala verstanden, welche zu einer Verschlechterung des makroskopischen Mate-rialverhaltens führt. Die Berücksichtigung von Schädigungsmodellierung im Konstruktionspro-zess ermöglicht die Vorhersage des Schädigungszustandes, welcher durch den Fertigungspro-zess entsteht. Die Schädigung kann durch Anpassung von Prozessparametern reduziert wer-den, um Teile zu erzeugen, die eine geringere Schädigungsanreicherung aufweisen und in der Anwendung schädigungstoleranter sind. Hier ermöglicht der Einsatz numerischer Optimierung die automatische Generierung schädigungstoleranter Prozesse und Bauteile.
Diese Arbeit befasst sich mit der numerischen Schädigungsoptimierung und lässt sich in zwei Bereiche einteilen. Der erste Bereich befasst sich mit den akademischen Herausforderungen eines solchen Optimierungsproblems. Ein nicht-lokales duktiles Schädigungsmodell wird mit Sensitivitätsinformationen angereichert, um Formoptimierung zu ermöglichen. Die analytische Herleitung der Gradienten erlaubt die Anwendung gradientenbasierter Optimierungsstrategien, wodurch die Rechenzeit erheblich reduziert wird. Dieser Ansatz ist zwar rechnerisch effizient und elegant für die Herleitung der Sensitivitätsinformationen, der Anwendungsbereich ist jedoch auf akademische Probleme beschränkt. Daher befasst sich der zweite Bereich dieser Arbeit mit den industriellen Herausforderungen. Dafür wird eine zusätzliche Optimierungsumgebung um eine kommerzielle Finite Elemente Software herum entwickelt. Diese ermöglicht eine Schädi-gungsoptimierung von Umformprozessen, sodass Parameterkombinationen gefunden werden können, die Bauteile mit geringerer Schädigungsanreicherung erzeugen.
Manufacturing of engineering components in today's industry requires a great amount of knowledge in order to produce components with the desired capabilities. This knowledge en-compasses the material behaviour, the manufacturing process which creates the part, as well as the loading which the product is subjected to during its intended use. Combination of those factors allows an engineer to design a product which satisfies the requirements for the task at hand. While elastic and elasto-plastic effects are well understood nowadays, damage mechan-ics take a more leading role in research. Damage can be understood as an accumulation of de-fects on the microscale, which leads to a degradation of the macroscopic behaviour of the ma-terial. Including damage modelling into the engineering process, e.g. by using damage material models for the simulations, it is possible to predict damage accumulation due to the manufactur-ing process. This damage can subsequently be reduced by adjusting process parameters, in order to generate parts which not only show overall lesser damage accumulation, but also fur-ther resistance against damage when in use. While typically such design processes are done manually, the utilisation of numerical optimisation allows automatic generation of damage re-sistant designs. Here, the use of numerical optimisation enables the automatic generation of damage-tolerant processes and components. %It is therefore possible to automatically generate parts which not only show less damage evolution under prescribed loads but at the same time reduce the weight and therefore material cost of the part. This thesis deals with numerical damage optimisation and consists of two overarching parts. The first part covers the academic challenges of an optimisation problem. A nonlocal ductile damage model is enhanced with sensitivity information to allow for shape optimisation. Analyti-cal derivation of the gradients enables the application of gradient-based optimisation strategies which greatly reduce the computation time. While this approach is computational efficient and is elegant in deriving the sensitivity information, its scope of application is limited to academic problems. Consequently, the second part of this thesis deals with the industrial challenges. For this, an additional optimisation framework approach around a commercial finite element soft-ware is derived. This framework allows damage optimisation of forming processes, such that parameter combinations can be found which produce parts with less damage accumulation.
Manufacturing of engineering components in today's industry requires a great amount of knowledge in order to produce components with the desired capabilities. This knowledge en-compasses the material behaviour, the manufacturing process which creates the part, as well as the loading which the product is subjected to during its intended use. Combination of those factors allows an engineer to design a product which satisfies the requirements for the task at hand. While elastic and elasto-plastic effects are well understood nowadays, damage mechan-ics take a more leading role in research. Damage can be understood as an accumulation of de-fects on the microscale, which leads to a degradation of the macroscopic behaviour of the ma-terial. Including damage modelling into the engineering process, e.g. by using damage material models for the simulations, it is possible to predict damage accumulation due to the manufactur-ing process. This damage can subsequently be reduced by adjusting process parameters, in order to generate parts which not only show overall lesser damage accumulation, but also fur-ther resistance against damage when in use. While typically such design processes are done manually, the utilisation of numerical optimisation allows automatic generation of damage re-sistant designs. Here, the use of numerical optimisation enables the automatic generation of damage-tolerant processes and components. %It is therefore possible to automatically generate parts which not only show less damage evolution under prescribed loads but at the same time reduce the weight and therefore material cost of the part. This thesis deals with numerical damage optimisation and consists of two overarching parts. The first part covers the academic challenges of an optimisation problem. A nonlocal ductile damage model is enhanced with sensitivity information to allow for shape optimisation. Analyti-cal derivation of the gradients enables the application of gradient-based optimisation strategies which greatly reduce the computation time. While this approach is computational efficient and is elegant in deriving the sensitivity information, its scope of application is limited to academic problems. Consequently, the second part of this thesis deals with the industrial challenges. For this, an additional optimisation framework approach around a commercial finite element soft-ware is derived. This framework allows damage optimisation of forming processes, such that parameter combinations can be found which produce parts with less damage accumulation.
Description
Table of contents
Keywords
Nonlocal damage, Finite element method, Cariational sensivity analysis, Structural optimisation, Shape optimisation, Process optimisation, Abaqus