Mechanical Assembly and Researches Needed to Make it Smart

1. Mechanical Assembly andResearches Needed to Make it Smart S. Jack Hu and Dawn Tilbury Department of Mechanical Engineering The University of Michigan

2. Problem Statement What are the tough research problems (related to the design, engineering, validation, construction, installation, launch, and operation of assembly processes and systems) that must be solved to make “smart assembly” a reality?

3. Outline • Mechanical Assembly and Its Role in Production Realization • Performance of Assembly Systems • Smart Assembly and Research Needs • Summary

4. What is Assembly ? • Assembly: • a collection of parts • the fitting together of manufactured parts into a complete machine, structure, or unit of a machine - Merriam-Webster Online Dictionary Product Process

5. Why do we assemble things ? • Functional need. • Technical Feasibility:It is not feasible to make the product as a single piece. • Cost:It is not economical to make as a single piece.

6. Mechanical Assembly

7. Component Manufacturing Design Fabrication Assembly Assembly Assembly as process • • • as product Role of Assembly in Product Realization “Logistical issues, supply chain management, product architecture, mass customization, management of variety and product family strategies … are enabled during assembly design and are implemented on the assembly floor” - Dan Whitney, 2004

8. Role of Assembly in Product Realization • Properly defined assembly interface can allow a company to mix and match parts to create variety with minimum cost. • An appropriate assembly sequence can permit a company to customize a product when it adds the last few parts. • Properly defined subassembly permits a company to design them independently or outsource them. • A well defined assembly-centered product development process can make ramp-up faster. Whitney, 2004

9. To Assemble or Not to Assemble ?

10. 2. Performance of Assembly Systems • Quality • Perceived quality • Deviation from design intent • Cost • Productivity • Input - output • Responsiveness • Volume • Variety

11. 3. Smart Assembly • Smart tools, such as feeders, fixtures etc. • Collaborative robots working with people intelligently and smoothly. • Automatic compensation for quality • Effective response of maintenance personnel to unexpected change of system to ensure system quality and productivity. • Self repairing and reconfigurable system What is Smart Assembly ?

12. 3.1. Quality Where • x(k) : Part Deviation • u(k) : Tooling Deviation • y(k) : Measurement Points • Sources of Variation: • Part variation • Fixture variation

13. •••••• 3.1. Quality Research Challenges • Developing models incorporating all sources of variation, including operator influence • Developing models early to aid system design • In-line adjustment and compensation • Sensor placement • Controllability issues Diagnosis Prediction

14. Machine Reliability Throughput WIP Machine Rates Bottleneck Analysis Buffer Size System Configs Buffer allocation 3.2. Reliability and Maintenance

15. 3.2. Reliability and Maintenance Research Challenges • Interactions of quality and reliability in throughput modeling • Allocation of maintenance resources to ensure fast response • Self-reconfiguring of assembly system for exception handling • Task re-allocation • Self repairing

16. 3.3 Variety and Complexity • Today’s market • Higher variety at lower volume; mass customization • Mixed-model assembly lines are needed to share investment among different vehicle models and absorb demand fluctuations • Engineering challenges • When product variety is large, manual assembly processes can become very complex, increasing the likelihood of human mistakes higher warranty cost more rework & lower throughput

17. Station Level Complexity • Typical procedures/activities • Parts are stored in line-side bins • An operator reads the production tag • Picks up the right part • Uses the right tool and fixture • Follows the right assembly procedure • A series of assembly activities and corresponding choices

18. System Level Complexity • Incoming complexity • Outgoing complexity

19. Variety and Complexity Research Challenges • What is complexity ? • Understanding the relationship between complexity and performance • Balancing between product variety and manufacturing complexity

20. 3.4 Control Design & Validation • Virtual models of manufacturing systems built for layout, design, training • Useful for control systems validation and testing • Advantages of virtual validation • Controls can be programmed and tested before system is built and operating • Error handling and recovery can be exhaustively tested without damage to hardware

21. Control Design & Validation • Needs for virtual validation: • Sensor and actuator models • Control interface • Kinematic behavior • Barriers to virtual validation • Lack of standard interfaces for simulation models and control codes (proprietary interfaces) • Fidelity of simulation model not guaranteed (lack of agreement between virtual/real behavior)

22. Control Design & Validation • Research challenge: Automatic generation of control logic from virtual model • Developing libraries of control components • Appropriate granularity for the modules in the libraries, neither too big (entire system) or too small (single axis) • Incorporation of multiple modes of control (auto, manual) • Interface specifications for the modules, across proprietary platforms, to ensure seamless operation • Incorporation of the operator through the HMI panel • Safety criteria, to protect machines, parts, and operators

23. Control Design & Validation • Research challenge: Logic verification (i.e. formal mathematical proof of correctness) • Formal representations of logic specification (currently often specified in English) • Formal models of control code, including execution semantics (not adequately addressed by existing standards, e.g. 61131 or 61499) • Generating mathematical models (e.g. diff. eqns or discrete-event systems) from simulation model for formal verification

24. 3.5. Real+Virtual • Idea: Run simulation model of factory in real-time in parallel with real factory • Update simulation with run-time data at discrete points in time • “Record” a trace of system execution • “Play back” what happened yesterday (or last year) • “Rewind” to see how the current error state occurred • “Play Forward” to examine the potential future effects of different control decisions now

25. 3.6. Real+Virtual • Research Challenges: • Validate the fidelity of the simulation model by comparing simulated performance with real performance • Synchronization across multiple domains with interacting controllers • Integrating models at different levels of abstraction • Machines: force, position, accuracy, CNCs,… • Cells: material handling, maintenance, PLCs,… • System: part flow, scheduling, databases, control & communication networks, …

26. 4. Summary • Assembly plays important roles in product realization. • Performance of assembly systems include quality, cost and responsiveness. • “Smart assembly” must be designed to ensure performance. • Research needs are identified to ensure performance of assembly systems.