TOWARD THE IDEAL OF AUTOMATING PRODUCTION OPTIMIZATION

1. TOWARD THE IDEAL OF AUTOMATING PRODUCTION OPTIMIZATION Frederick Proctor, NIST John Michaloski, NIST Jorge Arinez, GM Jonatan Berglund, Chalmers University of Technology Proceedings of ASME 2013 International Mechanical Engineering Congress & Exposition San Diego Califormia, U.S.A. Nov 15–21, 2013

2. Overview • Optimization goal • Automate data collection, model generation and factory analysis • First step - need factory model to optimize • Core Manufacturing Simulation Data (CMSD) • Factory Information Model (for discrete event simulation) • Modeling factory jobs, parts, process plans, resources • Work determined missing elements in CMSD • Case study • Precision casting line at GM • CMSD model of cast line with conveyors, robots, performance, MTBF, MTTR, cycle time

3. Optimization • Discrete Event System (DES) Factors • Non trivial hard problem – exhaustive search • Improved, but not perfect, factory data collection • Time-intensive procedures required to develop DES models • Can we improve? • More automated optimization • Goal: Automated data acquisition, automated generation of production models, and the automated integration of data • More expressive optimization goals • CMSD extensions to narrow search • Required, but not described, CMSD extensions to handle optimization reports, comparisons, etc.

4. Core Manufacturing Simulation Data • CMSD - an information model to facilitate the definition/exchange of manufacturing simulation related information for/between simulations and other manufacturing applications. • Major categories of information • Part, part type, BOM, and inventory data • Resource, resource class, setup, and employee skill data • Process plan, maintenance plan, and shift related data • Order, job, and schedule related data

5. CMSD (of interest) Overview • The goal of CMSD is to provide a neutral framework that facilitates the creation of collections of related manufacturing information suitable for use in the creation or enhancement of manufacturing simulations and other manufacturing applications. \ • We found that CMSD would be better served if supplemented by: • more incremental approach to file development, • more recording of manufacturing operation simulations for comparison • intrinsic language to describe optimality in the system.

6. Example CMSD Document

7. NIST ISD Adaption of CMSD • First, although designed in UML, CMSD has a C#/.Net Framework mapping in which to read CMSD files. • Using the C# EXE, the xsd.exe software tool from Microsoft generated an XSD. • Next, the commercial tool XMLSpy was used as it provides facilities to load XSD documents, validate the XSD files, and then generate C++ archival (read and writing from files) code based on a XML parser. • XMLSpy autogenerated C++ code for XML reading and validation using Microsoft MSXML technology.

8. Case Study General Motors: Saginaw Metal Casting Operation

9. Production Facility w/ Real Data

10. SMCO Precision Sand Casting • 250,000 sq ft. and over a thousand pieces of equipment • Makes V8 engine blocks • Uses Polyurethane Cold Box Sand Cores • Cores Produced & Assembled Into 700 lb. Mold • Studied the Cast/Molten Line 123225.76 m²

11. Concentrated on Casting Line CHILL (goes to basement for cooling) CORE PACKAGE & Base Plate with CASTING (goes to basement for cooling and shakeout) MOLTEN ALUMINUM BASE PLATE COVER CORE CORE PACKAGE CHILL ALLOY ELEMENT

12. NIST CMSD Builder

13. CMSD Gaps • Incremental CMSD files • Bulky but not impossible to use 1 large file • Allowed easier comparison • Optimization criteria • Unbounded search otherwise • Replication comparison • No facility to compare day-to-day optimizations

14. CMSD Incremental Files

15. CMSD Resource

16. CMSD Part Information (in UML)

17. CMSD Process Plan

18. Optimization Criteria • Express optimize criteria • Maximize, Minimize, Range • Comparisons • Compare MTBF, MTTR differences in resource • Compare buffer sizes • Minimize blocked and starved states • Benchmarking – compare improvements • Not covered – • Strategies • Forecast, trends, capacity • Asset costs, ROI

19. CMSD Job w Optimization

20. Job with Resource Parameter Range

21. Discussion • CMSD improvements feasible and useful • Incremental CMSD information models helpful • Optimization criteria is worthwhile addition to CMSD • More CMSD and general optimization formalism required • One step toward long term goal • Automate the generation of DES model from CMSD and then run simulations based on the factory described in CMSD with automated data acquisition. • Like to push factory model building onto machines with some “introspective” capabilities – location, # axes, etc. • Lot of hard coding now, but easily automated into CMSD • Given data from a brownfield factory and then validating model is difficult.

22. Comments • It would be unreasonable to expect a large car company to change its steady-state production based upon the preliminary findings of a DES system. • Unreasonable not to have undergone some preliminary optimization of buffers, equipment layout, etc. before assembling the production line. • Day-to-day matters will become routine and change itself can be difficult [Goldratt]

23. Questions?

24. View on Factory Floor

25. CMSD Information Modeling • Resource - describes equipment that performs manufacturing activities. Resources in the CMSD are used to represent stations, machines, cranes, employees, tools, and fixtures. (for this iteration we assumed no trained personnel were required.) • Part - provides a means to specify the characteristics of the materials and subcomponents that are used to make end products. • Process plan - specifies the set of production activities needed to transform materials and subcomponents into finished products. Each process plan is built of process steps (with associated resource(s)) that must be executed for the part to be finished. • Process - defines a manufacturing activity or group or manufacturing activities that encompass a detailed strategy for creating a part. The process will most likely contain information that describes the resources that will be used, the parts that will be consumed and produced, the sequence in which resources will be used, and the sequence of activities within a group of activities. • Job - defines normal, maintenance or repair operation, but in our case the job represents normal manufacturing and is the central construct of the system. Each job (assuming it came from customer order as described earlier), would generate an appropriate number of parts into ``spawned'' jobs (type of job) and under each spawned job contains the part knowledge exhibited within the job e.g., process plan, the resources, etc. The spawned job would contain a copy of the initial job that described all the parts and quantities.