Automated conceptual design of manufacturing workcells in radioactive environments
Williams, Joshua Murry
MetadataShow full item record
The design of manufacturing systems in hazardous environments is complex, requiring interdisciplinary knowledge to determine which components and operators (human or robotic) are feasible. When conceptualizing designs, some options may be overlooked or unknowingly infeasible due to the design engineers' lack of knowledge in a particular field or ineffective communication of requirements between disciplines. To alleviate many of these design issues, we develop a computational design tool to automate the synthesis of conceptual manufacturing system designs and optimization of preliminary layouts. To generate workcell concepts for manufacturing processes, we create a knowledge-based system (KBS) that performs functional modeling using a common language, a generic component database, and a rule set. The KBS produces high-level task plans for specific manufacturing processes and allocates needed material handling tasks between compatible human and/or robotic labor. We develop an extended pattern search (EPS) algorithm to optimize system layouts based on worker dose and cycle time minimization using the functions and sequencing of generated task plans. The KBS and EPS algorithm were applied to the design of glovebox processing systems at Los Alamos National Laboratory (LANL). Our computational design tool successfully generates design concepts with varied task allocation and processing sub-tasks and layouts with favorable manipulation workspaces. This work establishes a framework for automated conceptual design while providing designers with a beneficial tool for designing manufacturing systems in an interdisciplinary and highly constrained domain.