Contextualising manufacturing data for lifecycle decision-making

Feasibility Study for an Automated Engineering Change Process

Engineering change is a significant cost sink in many projects. While avoiding and mitigating the risk of change is the ideal approach, mistakes and improvements are recognized inevitably as more is learned over time about the quality of the decisions made in a product's design. This paper presents a feasibility and performance analysis of automating engineering change requests to demonstrate the promise for increasing speed, efficiency, and effectiveness of product-lifecycle-wide engineering-change-request processes. To explore this idea, a comparatively simple case study is examined both to mimic the reduced set of alterable aspects of a typical change request and to highlight the need of appropriate search algorithms as brute force methods quickly prohibitively resource intensive. Although such cases may seem trivial for human agents, with the volume of expected change requests in a typical facility, the potential opportunity gain by eliminating or reducing the amount of human effort in low level change requests accumulate into significant returns for industry on time and money. Within this work, the genetic algorithm is selected to demonstrate feasibility due to its broad scope of applicability and low barriers to deployment. Future refinement of this or other sophisticated algorithms leveraging the nature of the standard representations and qualities of alterable design features could produce tools with strong implications for process efficiency and industry competitiveness in the execution of its projects.

Foundations of information governance for smart manufacturing

The manufacturing systems of the future will be even more dependent on data than they are today. More and more data and information are being collected and communicated throughout product development lifecycles and across manufacturing value chains. To enable smarter manufacturing operations, new equipment often includes built-in data collection capabilities. Older equipment can be retrofitted inexpensively with sensors to collect a wide variety of data. Many manufacturers are in a quandary as to what to do with increasing quantities of data. Much hype currently surrounds the use of AI to process large data sets, but manufacturers struggle to understand how AI can be applied to improve manufacturing system performance. The gap lies in the lack of good information governance practices for manufacturing. This paper defines information governance in the manufacturing context as the set of principles that allow for consistent, repeatable, and trustworthy processing and use of data. The paper identifies three foundations for good information governance that are needed in the manufacturing environment-data quality, semantic context, and system context-and reviews the surrounding and evolving body of work. The work includes a broad base of standard methods that combines to create reusable information from raw data formats. An example from an additive manufacturing case study is used to show how those detailed specifications create the governance needed to build trust in the systems.