Optimising the factory schedule as production gets harder to plan

The problem

Every factory solves the same problem shift after shift: which job runs on which machine, and in what order, to meet due dates and protect throughput within routing, setup, batching and capacity constraints. In its flexible form the job-shop scheduling problem is strongly NP-hard. As product mixes widen, lines share more equipment, and orders change inside the day, the schedule a planner fixed in the morning is stale by mid-shift.

Where we think we can add value

Semiconductor fabs run several hundred operations across roughly a hundred tool groups with re-entrant flow; because re-solving in time is not feasible they fall back on dispatching rules and accept idle tools. Pharmaceutical and fine-chemical plants campaign many low-volume products on shared multi-purpose equipment with cleaning and validation windows. Primary metals couple casting, hot and cold rolling and coating with sequencing rules on width, gauge, grade and temperature. Solving these schedules quickly, and re-solving as conditions change, is where we believe we can help.

Status

What is measured and what is not. The market figures on this page come from public research and are established context. Our measured results to date are in mine planning. The performance we describe for manufacturing is a hypothesis we want to test on a real problem with a real operator, not a delivered result. See /mining.

We have solved a problem of this class at scale: a scheduling problem of more than 200,000 variables returned in a few minutes on a single consumer-grade GPU, against legacy tooling that takes hours or days.