TY - JOUR
T1 - Combining the advantages of discrete- and continuous-time scheduling models
T2 - Part 2. systematic methods for determining model parameters
AU - Lee, Hojae
AU - Maravelias, Christos T.
N1 - Funding Information:
H. Lee would like to acknowledge support from University of Wisconsin - Wisconsin Distinguished Graduate Fellowship, as well as the Kwanjeong Educational Foundation, South Korea.
Publisher Copyright:
© 2018 Elsevier Ltd
PY - 2019/9/2
Y1 - 2019/9/2
N2 - The Discrete-Continuous Algorithm (DCA) is a novel framework that harnesses the strengths of discrete- and continuous-time scheduling formulations (Lee and Maraveloas,2018). Its flexibility in the selection of two user-defined parameters, namely discretization step length (δ) and horizon relaxation (η), can lead to significantly improved computational performance and solution quality. In this paper, we propose systematic methods to determine these parameters. Specifically, we evaluate the parameters based on error evaluation functions and cumulative error functions that consider various aspects of the scheduling instances. Through an extensive computational study, we show that the proposed methods bring up to × 104 speedups, while leading to identical or better solutions in the majority of the instances compared to traditional methods.
AB - The Discrete-Continuous Algorithm (DCA) is a novel framework that harnesses the strengths of discrete- and continuous-time scheduling formulations (Lee and Maraveloas,2018). Its flexibility in the selection of two user-defined parameters, namely discretization step length (δ) and horizon relaxation (η), can lead to significantly improved computational performance and solution quality. In this paper, we propose systematic methods to determine these parameters. Specifically, we evaluate the parameters based on error evaluation functions and cumulative error functions that consider various aspects of the scheduling instances. Through an extensive computational study, we show that the proposed methods bring up to × 104 speedups, while leading to identical or better solutions in the majority of the instances compared to traditional methods.
KW - Chemical production scheduling
KW - Continuous-time representation
KW - Discrete-time representation
UR - http://www.scopus.com/inward/record.url?scp=85056319646&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85056319646&partnerID=8YFLogxK
U2 - 10.1016/j.compchemeng.2018.10.020
DO - 10.1016/j.compchemeng.2018.10.020
M3 - Article
AN - SCOPUS:85056319646
SN - 0098-1354
VL - 128
SP - 557
EP - 573
JO - Computers and Chemical Engineering
JF - Computers and Chemical Engineering
ER -