UC Berkeley

Advanced manufacturing systems, cyber-enabled tools outfitted with intelligent sensors and communication abilities, have become increasingly capable, ubiquitous, and standardized in an effort to increase the productivity and quality of manufactured goods. As demands on the complexities of both the products and the manufacturing processes increase, new workflows and systems are required in order to develop successful programs for automated equipment, interact and collaborate with manufacturing equipment in a productive manner which leverages the respective strengths of humans and machines alike, and plan and schedule the complex workflows in a production system. Additionally, any influx of cyberinfrastructure brings with it security risks and vulnerabilities, and unique considerations must be made for manufacturing systems. This dissertation provides contributions in all of the aforementioned areas.

Part I of this dissertation discusses advanced toolpath planning and optimization approaches for additive manufacturing systems, which are adaptive to complex geometries, problematic in-process physical phenomena, and demands on throughput or resolution. The first chapter details a method for computing additive manufacturing toolpaths on complex design surfaces, moving the field beyond the paradigm of flat slicing and discretization of intricate part geometries, while guaranteeing collision-free movement throughout the printing process. While the former is an entirely geometric-based toolpath planning approach, additive manufacturing processes incur many failures due to the evolution of complicated thermomechanical fields, which are strongly dependent on the toolpath selected for the process. The next chapter contributes work integrating physical knowledge into the toolpath planning process by employing efficient physical process simulation techniques in concert with evolutionary optimization methods to yield toolpaths tailored for a specific part's geometry, orientation, and material composition, capable of reducing or eliminating process-induced defects during additive manufacturing. This method of coupling physical simulation with artificial intelligence (AI) techniques can serve as a model for other fields and applications. Another issue facing additive manufacturing is relatively long processing times when compared to conventional manufacturing techniques. The following chapter presents a prototyped and tested design concept for a variable-aperture nozzle, which can dilate to increase throughput, while adjusting to print with finer resolution in regions of the part where needed. Such a mechatronic system must be programmed to control the aperture to the optimal diameter at each moment during the process, and computational approaches are provided for determining the sequence of optimal diameters.

Part II broadens the scope from the manufacturing process to the entire production system, introducing approaches for collaborative workflows between humans and machines on the factory floor, analyzing relationships between key performance indicators to achieve more stable and sustainable production, and identifying the cybersecurity risks present in manufacturing systems. The first chapter presents a shop-floor system to work adjacent to, and cooperatively with, human machine operators to anticipate and prevent critical CNC machining errors. Factory-level planning is often performed with incomplete knowledge, based instead on key performance indicators (KPIs), but the KPIs themselves are not mutually exclusive, despite how they are treated in international standards. The following chapter contributes an analysis of the interactions between different KPIs, in particular those which relate to the concept of machine health, such as preventive maintenance time, corrective maintenance time, and machine failure metrics, with the goal of enabling more stable and sustainable production. The final chapter provides an analysis of the various data flows present in a modern manufacturing enterprise, considering data on the factory floor, between facilities, and along tiers of the supply chain. This framework allows for the identification and management of manufacturing-specific cybersecurity risks.