Manufacturing Modeling and Simulation

To develop and deploy advances in measurement science that will help optimize factory throughput, quality, and efficiency through the automation of data acquisition, manufacturing model development, and manufacturing analysis, delivering results through open-source implementations and standards bodies by 2014.

The use of commercial manufacturing simulation tools to enable optimization is currently manually intensive and unreliable. A new use of commercial manufacturing simulation tools is proposed that is largely automated, more reliable, and standards-based. An integrated data model for manufacturing activities will be defined that will automate real-time data inputs and updates to manufacturing simulation and analysis.

The task of acquiring reliable and complete manufacturing data is known to take roughly 30 % of the time required to develop a manufacturing model^[1]. Furthermore, the manufacturing data input to typical manufacturing simulation tools and real production data collected by data collection technologies are commonly incompatible. NIST will determine the measurement science necessary to automate the acquisition of reliable, complete, and compatible data, in a standards-compliant way.

Reliable manufacturing models, which can be automatically generated, will help realize optimized factory operations, since the results from what-if scenarios can be better relied upon and obtained more quickly. Measurement science techniques, including classic statistics, will be applied to transform the manufacturing data into a manufacturing model format. NIST will investigate methods for the automated generation of manufacturing models, which incorporate elements such as facilities, process planning, resource management, and part management.

To achieve automated analysis of the manufacturing operation, all aspects of the manufacturing operation must be included, such as plant, process, and personnel as well as design, production, and maintenance. Simulation offers a controlled environment to study the large scale interaction of machines and processes under different conditions. Simple parameter adjustments can be run through simulation time sequences to predict the impact of potential changes. Yet, there is a lack of decision–making strategies for optimizing manufacturing using simulation. NIST will address this problem through the identification and characterization of key manufacturing activities and then develop modeling techniques to optimize their performance.

The research plan consists of: 1) the definition of a target manufacturing system-under-test, 2) the development of publically available prototype software (measurably compliant to the system-under-test) to discover and validate obstacles to automating data analysis and factory model generation in concert with commercial manufacturing simulation tools, and 3) the development of measurement science techniques for manufacturing data acquisition, subsequent integration into the manufacturing model, and manufacturing analysis. An important measurement will be to quantify the time saved in automated manufacturing data acquisition versus current approaches to manufacturing data acquisition, and quantify the time saved in automated manufacturing model generation versus current approaches to manufacturing model generation. This will be very hard to measure, and will require close collaboration with key industry partners.

Intelligent automated manufacturing data analysis is also needed to realize optimized factory operations. Measurement science techniques, including classic statistics, will be applied to transform the manufacturing data into a format consumable by the production model. Integration of the factory data with automatically generated manufacturing models is required for intelligent analysis, and this integration must be accurate, timely and cost effective. The NIST Virtual Factory Testbed (VFT) will be used to test the effectiveness of our manufacturing models, initially using simulated manufacturing data as input to the models, before working with actual factory data from our U.S. industrial partners.

The process for automating the data acquisition, factory model generation and analysis is a multiyear effort culminating in a new technology paradigm for automated factory optimization, outlined below.

• Automated data acquisition: a) assessment of customer data and performance requirements through input from manufacturing partners; b) identification of data quality and data healing techniques; c) evaluation of data healing techniques through testing on the VFT, using industrial partner actual production data test sets and reference data sets with bad data; d) pilot evaluation of self-correcting capability of candidate data acquisition technology within continuous simulation environment of the VFT.
• Automated model generation: a) assessment of production modeling requirements and narrowing the selection of candidate factory manufacturing  models based on their importance to our U.S. manufacturing partners; b) development of an experimental measurement criteria with which to evaluate the Core Manufacturing Simulation Data (CMSD) ^[2] generated models; c) verification of the automated modeling by comparing it to known statistical performance measures generated from the VFT against performance measures generated by the CMSD automated factory model from the VFT production model and data; d) validation of the automated model generation against its intended production modeling requirements in terms of the technology used and the results obtained; and e) evaluate automated model generation pilot on U.S. industry partner’s production facility, using real data, and then verify and validate to determine accuracy and utility of the generated model.
  
• Automated analysis: a) assessment of performance metrics and optimality criteria through extensive literature research complemented with input from manufacturing partners; b) identification of key production performance measurements and optimality techniques for a focus subset; c) generation of test sets of factory models and simulation data for evaluation; d) experimental evaluation of optimization techniques using commercial modeling and analysis tools on developed test sets; and e) pilot evaluation of automated analysis on Virtual Factory testbed of production facility, that automatically generated a facility model and initiated automated collection and remediation of data; evaluation will verify and validate to determine accuracy and utility of the results of the generated analysis.

• Outcome:  MTConnect specification incorporation of ISO 13399 tooling paradigm based on the NIST translation  from ISO 13399 Express representation into XML Schema (2011)
• Output:  Conference papers on formalized approach to the integration of the various data sources of sustainable production knowledge at an automotive casting facility (2011-2012).
• Output: NIST contribution to MTConnect open source software repository  to assist U.S. industry in configuring and deploying MTConnect Agent with multiple and different devices on the network (2012)
• Output: NIST software development, testing and  participation with Boeing and Hexagon in demonstration of the Web-enabled, Real-time MTConnect Measurement Data at Hexagon Training Facility, Kent, Washington (2012)
• Output: Standards landscape report on the available technology for integration of factory floor equipment (2011)

Standards and Codes:

• MTConnect Version 1.1.0 Parts 1, 2, and 3
• The Simulation Interoperability Standards Organization (SISO), maintaining the Core Manufacturing Simulation Data (CMSD) standard
• ISO TC184 SC5 WG1 Modeling and Architecture

[1]A. Skoogh and B. Johansson, “A Methodology for Input Data Management in Discrete Event Simulation Projects,” in Proceedings of the 40th Winter Simulation Conference (WSC), Orlando, FL, 2008.

[2]CMSD is XML-based and facilitates the exchange of information between models and the production software applications, such as used in factory layout, process planning, scheduling, inventory management, production management, or supply chain management.