Online Scheduling

In the past researchers have focused on using optimization-based techniques to compute a schedule. However, in a dynamic environment, disturbances or new information may render the computed schedule suboptimal or infeasible, hence there is a need to reschedule. The revision of an existing schedule in response to such disturbances or new information is called online scheduling. In online scheduling, optimization problems are solved on a regular basis to generate an open-loop schedule. The current decisions from the open-loop schedule are implemented and the horizon is rolled forward. Based on the solution to a series of open-loop problems, we determine the actual implemented schedule (the closed-loop schedule).

Closed-loop schedule generation schematic

An example of closed-loop schedule generation. Batches of I1 and I2 produce products P1 and P2, respectively and are executed in the same unit. In iteration 0, to meet the demand, batches of I1 and I2 are slotted to occur consecutively. In iteration 1, a single period delay in batch of I1 has been detected, hence the I2 batch starts later. Note that only current decisions from each open-loop iteration are implemented (shown as red arrows).

The group studies how the open-loop schedules affect the closed-loop performance and found that there is considerable potential to improve the performance, but this task is non-trivial. We have proposed systematic methods to design online scheduling algorithms, focused on determining the horizon length and re-optimization time step, as well as modifications of the optimization model (e.g., addition of new constraints). We have developed a framework for studying online scheduling in the presence of endogenous uncertainties such as the uncertainties associated with batch processing times, yields, and unit breakdowns. Furthermore, we have made initial efforts towards incorporating feedback (i.e., real-time data, automation-derived information) into the scheduling problem. Through such efforts in our group, we continue to understand and address the challenges in online scheduling.   


Gupta D, Maravelias CT. Framework for studying online production scheduling under endogenous uncertainty. Computers & Chemical Engineering, 129, 106517, 2020.

Gupta D, Maravelias CT. On the design of online production scheduling algorithms. Computers & Chemical Engineering, 135, 106670, 2019.

Gupta D, Maravelias CT. On deterministic online scheduling: Major considerations, paradoxes and remedies. Computers & Chemical Engineering, 94, 312-330, 2016.

Rawlings BC, Avadiappan V, Lafortune S, Maravelias CT, Wassick J.M. Incorporating automation logic in online chemical production scheduling. Computers & Chemical Engineering, 128, 201-215, 2019.