Brief Description

Schedule production of Ford Transit vehicles’ chassis at the body shop to minimize consecutive model spec changes (e.g., roof height, wheelbase length) and balance hourly labor requirements, while adhering to shop constraints (e.g., producing Jumbo type vehicles in groups of four).

Key Highlights

·      Enhanced Flexibility of body shop and all lines downstream to unforeseen events such as parts delivery delays or unplanned maintenance requirements.

·      Reduction in Scheduling Time and reduced risk of feasible problems not being solved compared to Classical Computing.

·      Ability to Scale the Optimization Problem to include other processing line requirements from both upstream and downstream productions, thanks to the high constraint limits of quantum hybrid solvers.

Detailed Description

·      Customer orders from around the world are consolidated into a pool, with each vehicle’s specifications and delivery deadlines predetermined. In body production, each spec change requires the programmable logic controller of the welding robots to load a new set of instructions. With 250 welding stations, frequent spec changes can lead to production delays.

·      Sequencing these vehicles to maximize throughput and adhere to physical and workforce constraints of the Body Shop, Paint Shop, and Assembly Line, is a highly constrained optimization problem. Additionally, due to our supply chain dynamics, the optimized schedule may need to be altered on-demand by removing vehicles requiring specific parts to prevent production halts. Similar challenges arise with unplanned maintenance events due to machine breakdowns.

·      The classical optimization model we developed, alongside our Quantum algorithm, solves this problem in approximately 10 minutes for 1,000 vehicles using a commercial solver. In contrast, Open-Source tools can take over an hour with less optimal results. Quantum annealing (via D-Wave’s hybrid solvers) solves the same problem under 5 minutes (c.16,000 constraints for 1000 vehicles). Preliminary observations (without extensive testing) suggest that as the problem constraints tighten, the commercial classical solver is more likely to fail to find a feasible solution.

·      Current scheduling problem can be extended to cover other considerations from up and down stream productions with Quantum Annealing, thanks to increasing nature of upper constraint limits observed in newer quantum chips.

Publicly Available Technical Data

·      https://www.dwavesys.com/media/5v4axazz/production-scheduling-data-sheet.pdf (Page 2)

·      https://www.dwavesys.com/media/5d0gxhqk/d-wave-optimization-guide.pdf (Page 14)

·      https://www.youtube.com/watch?v=9jwmLrdPpCU (Qubits 2024 – Boston; Customer Story)