New Continuous-Time Scheduling Formulation for Continuous Plants under Variable Electricity Cost

Pedro M. Castro, Iiro Harjunkoski, Ignacio E. GrossmannIndustrial 2009 · DOI

TL;DR

This paper introduces a novel continuous-time scheduling formulation for continuous plants that effectively handles time-varying electricity costs and availability. While the continuous-time model is a breakthrough for incorporating dynamic utility profiles, the discrete-time approach proved more practical for industrial-scale problems due to its natural handling of discrete events and better computational performance. Both approaches demonstrated significant potential cost savings (around 20%) by optimizing production schedules around electricity price fluctuations.

What Problem Does This Paper Address?

Industrial plants, especially those with electricity-intensive processes, face increasing pressure to minimize production costs amidst fluctuating electricity prices and availability. Traditional scheduling methods often struggle to efficiently incorporate these dynamic energy constraints, leading to suboptimal decisions and higher operational expenses. This paper tackles the challenge of developing scheduling models that can accurately represent and optimize production for continuous plants operating under time-dependent electricity costs, multiple intermediate due dates, and variable utility availability. The authors aim to bridge the gap between theoretical modeling and practical industrial needs by proposing both continuous-time and discrete-time formulations based on the Resource-Task Network (RTN) framework.

Methods

The core of this work lies in developing scheduling formulations using the Resource-Task Network (RTN) representation. The authors propose two main approaches:

  • Continuous-Time Formulation: This is presented as a major breakthrough, being the first model of its kind capable of effectively incorporating time-variable utility profiles. It builds upon existing RTN-based formulations but extends them to handle dynamic electricity costs and availability. A key innovation is the disaggregation of processing tasks to account for different energy levels, leading to a more detailed model.
  • Discrete-Time Formulation: This approach uses a fixed-length time grid. While it might require a sufficiently fine grid to accurately capture events, it handles discrete events (like electricity price changes or demand points) more naturally. The authors compare its performance against the continuous-time model.

Both formulations aim to minimize total electricity cost while respecting constraints on resource availability (processing units, storage, utilities) and meeting product demands by specified due dates.

Datasets

The paper uses data generated randomly from a real industrial problem for its computational studies. Specific examples (EX1-EX10) are presented, varying in the number of products, machines, and storage units. The illustrative examples (EX1, EX2, EX3) use specific parameters for product demands, machine power requirements, processing rates, and storage capacities.

Key Insights and Ideas

  • Continuous-time modeling of dynamic electricity costs is a significant challenge: While a breakthrough, the continuous-time model's computational burden increases dramatically with the number of event points, making it intractable for larger industrial problems.
  • Discrete-time formulation offers a practical compromise: For industrial-scale problems, a discrete-time approach with a sufficiently fine time grid provides a more natural and computationally efficient way to handle discrete events like electricity price changes.
  • RTN is a powerful framework: The Resource-Task Network provides a unified and flexible way to represent complex process scheduling problems, accommodating both continuous and discrete tasks.
  • Electricity cost variability drives significant savings: Optimizing production schedules to leverage lower electricity cost periods can lead to substantial cost reductions (around 20%), highlighting the importance of accurate energy cost modeling.
  • Storage capacity and machine availability are critical constraints: These factors significantly influence the scheduling decisions and can lead to increased costs if not managed effectively.

Key Results

  • Cost Savings: The proposed scheduling approaches can lead to potential electricity savings of around 20% by shifting production from high-cost to low-cost periods.
  • Computational Performance: The discrete-time formulation demonstrated superior performance for industrial-sized problems, finding solutions with optimality gaps below 1% in practical computational times (around 5 minutes). The continuous-time formulation was limited to smaller problems and required significantly more computation time.
  • Model Tightness: The discrete-time formulation was found to be "tighter" than the continuous-time one, exhibiting lower integrality gaps.
  • Impact of Constraints: For EX2, setting maximum power consumption to zero for a specific period resulted in a 15.8% cost increase compared to EX1, illustrating the significant impact of power availability constraints.
  • Storage Flexibility: EX3, with an additional storage unit, achieved a 2.5% cost reduction compared to EX1, demonstrating the benefit of increased storage flexibility.

Limitations and Future Work

  • Continuous-time model scalability: The authors acknowledge that the continuous-time formulation is limited to small problems due to its computational intensity, which strongly depends on the number of event points.
  • Discrete-time solution degeneracy: The discrete-time formulation can lead to solutions with a high number of task instances and changeovers, which might be undesirable in practice. A post-processing procedure is suggested to address this.
  • Future Work: The paper suggests that further research could focus on developing more efficient continuous-time models or hybrid approaches that combine the strengths of both continuous and discrete formulations. Investigating the impact of other factors not explicitly modeled, such as machine breakdowns or urgent orders, would also be valuable.

Summary updated

Abstract

This work addresses the scheduling of continuous plants subject to energy constraints related to time-dependent electricity pricing and availability. Discrete- and continuous-time formulations are presented that can address these issues together with multiple intermediate due dates. Both formulations rely on the resource−task network process representation. Their computational performance is compared for the objective of total electricity minimization with the results favoring the discrete-time model due to the more natural way of handling such a wide variety of discrete events. In particular, it can successfully handle problems of industrial size. Nevertheless, the new continuous-time model is a major breakthrough since it is the first model of its type that is able to effectively incorporate time-variable utility profiles. When compared to a simple manual scheduling procedure, the proposed scheduling approaches can lead to potential electricity savings around 20% by switching production from periods of high to low electricity cost.
BibTeX
@inproceedings{castro2009new,
  author = {Pedro M. Castro and Iiro Harjunkoski and Ignacio E. Grossmann},
  title = {{New Continuous-Time Scheduling Formulation for Continuous Plants under Variable Electricity Cost}},
  booktitle = {Industrial},
  year = {2009},
  doi = {10.1021/ie900073k},
}
Figure