Adjility

Automating mixed batch
job scheduling

Case Study

Key problem

A division of Dow has a number of plants which use a mixed batch-job structure, which makes it hard to schedule. In addition, there are frequent changes in requirements, and there are unanticipated disruptions that require rescheduling.

Solution

We built a custom two-stage heuristic algorithm which generated schedules in less than a minute -- it handles a wide variety of constraints, and does changes to the schedule with minimal disruption.

Result

The process of doing a schedule or rescheduling has evolved from a week-long manual process every month to one that takes a few minutes for computation.
In addition, the application has been used for longer term capacity planning.

Summary

Much of supply chain software stops at the level of optimizing master production plans - at this level the problems are still tractable and "modelable" using traditional mechanisms. Getting into plants, however, can be messy - and scheduling is often done using manual processes. Formal models of these, when built, often need to be trashed because of small structural or policy changes in the plant.Our approach to this for Adjility's client - a Fortune 50 chemical company - was different:
Hard problem to solve. Awesome approach!
Al CrowtherCEO, Adjility.

The approach


  1. A planner that measures and directs the solution algorithm, and a detailed scheduler that faithfully represents the ground realities.
  2. Build a perturbation algorithm that intelligently traverses the solution space without repetition.
  3. Allow incorporation of additional constraints and policies by customizing the detailed scheduler to each scenario.
  4. This approach not only allowed the quick and efficient solution of these problems, but was helpful in providing a mechanism to simulate structural changes and load increases.