Operations-driven Performance Measurement for Smart Manufacturing Systems

Summary

Collecting and analyzing data to inform changes to improve system performance is a hallmark of Smart Manufacturing. making full use of that data is impeded by a lack of well-defined methods and standards. Smart Manufacturing Systems rely on the ability to quickly measure and collect a wide variety and volume of data from manufacturing operations. Currently performance improvements are gained through targeted applications. A system of measurement science can make these improvements more obtainable for a wider array of applications and a wider set of manufacturers. NIST will contribute to the measurement science needed to evaluate performance in the design and analysis of smart manufacturing systems.

This project develops methods, tools, and standards for identifying and analyzing performance issues through systems characterization and data analysis. System characterization, whether it be the individual manufacturing process or the broader manufacturing system, defines the frame of reference needed to evaluate and improve the performance of a given system against a norm. In collaboration with ASTM, we are developing methods for describing, measuring, and evaluating the sustainable performance of manufacturing systems. These methods will enable the representation of those systems during system design and analysis allowing for trade-off analyses and optimization. In addition, we are developing methods for using operational data to uncover performance problems to further improve system design.

Description

Objective:  Develop and deploy standards, guidelines, and reference data for measuring the performance of manufacturing systems to inform the design and aid in analysis of factory improvements.

Technical Idea:  Manufacturers are adopting smart systems to drive gains in agility, productivity, quality, and sustainability. Smart manufacturing systems apply integrated information systems along with intelligent software applications to optimize system behavior. Performance measurement is key to designing and analyzing manufacturing systems. The derivation of effective performance metrics is difficult because of the complexity of these systems.

Operational-level performance measurements are ubiquitous in smart manufacturing systems. A key technical barrier is to develop models that use the data produced by the operations along with the desired performance metrics to addressing performance challenges.
Sustainable manufacturing is a primary performance problem in the context of smart manufacturing because of the inherently a complex set of trade-offs which must be optimized together with other competing performance objectives. Understanding the changes to any system in terms of the multiple objectives is already challenging, but is made much more difficult when criteria against which sustainability assessments can be made are not measured nor available in such a way as to be shared at a system level.

This project focuses on techniques for operations-driven performance measurement. In this context investigations explore the use of formalized performance metrics, standardized modeling methods, and standard guidelines for applying performance measures.

The project will continue development of standardize methods for evaluating the performance of manufacturing processes and standardized methods for characterizing the performance of manufacturing processes as building blocks for system analysis with specific focus on sustainability evaluation.

The project will extend this work by developing a repository of models of specific manufacturing processes, called Unit Manufacturing processes or UMPs, and tools for self-assessment of smart manufacturing systems. These outputs will provide frames of reference to guide decision making in preparing for future systems.

Research Plan:  A process model repository poses several research and standardization challenges. The concept of the Unit Manufacturing Process (UMP) repository is premised on a standardized format for representing those models. A foundation for that format was an earlier output of this project and accepted as an ASTM standard (ASTM E60.13) in FY16. This year’s research work will build on that foundation expanding it to support a rich set of manufacturing processes and capabilities to compose those models for system level evaluations. To collect a set of UMP models we will host a challenge competition, inviting the public to use the ASTM standards to characterize their own manufacturing processes. From these submissions, along with our own research, we will study the feasibility of creating a large collection of reusable abstractions of manufacturing processes. Specifically, we will develop requirements for

• the additional information will need to be standardized for UMP models to be composed for system level assessment;
• the types of abstractions that will be needed to facilitate use of the repository, such as process taxonomies, visualizations, and manufacturing ontologies;
• approaches to validation of the UMP models.

In addition, we will continue working with ASTM to provide standardized guidance on the selection and use of Key Performance Indicators (KPI) for sustainable manufacturing. KPI definition is necessary for completeness of UMP models and every UMP may have a unique set of KPIs. Guidelines on how those are selected will serve the purpose of strengthening performance characterizations of individual process as well as supporting manufacturing in improving their systems.

The second objective of self-assessment tools for SM is also based on the collection of manufacturing knowledge. The production of performance data and the automation of the information flows to consume that data is fundamental to achieving data-driven manufacturing. We will explore methods for aid manufacturers to make better use of their data. Using the Factory Design and Improvement reference model that was developed earlier in the project we will contribute to guidelines to help manufacturers improve their use of digital technology.
Additionally, we will apply analysis techniques to discover performance problems from operational data and structure the results into a diagnostic framework. We plan to work with industrial partners to obtain operational data for exploration and to define tools suitable to a variety of manufacturers.

In future years, we will continue to pursue the concept of reusable abstractions of manufacturing processes based on the lessons learned from our competition results and our industrial pilots. We anticipate further research on reusability and standards definition will be needed to make the vision a reality. In addition, we will identify the use of the UMP repository in the role of developing treatment
plans. The integration of the self-assessment tools with the UMP repository will foster the development of virtual treatment plans that are more than static recommendations but rather models that can be activated. If successful, this integration will be explored in FY18 and FY19.