TY - GEN
T1 - A general framework for chemical process scheduling
AU - Sundaramoorthy, Arul
AU - Maravelias, Christos T.
N1 - Copyright:
Copyright 2011 Elsevier B.V., All rights reserved.
PY - 2010
Y1 - 2010
N2 - In the context of process scheduling, existing methods can be broadly classified into sequential and network-based. The former are used to address scheduling problems in processes where batches are processed in a predefined sequence of operations; mixing and splitting of batches are not allowed thereby preserving the identity of each batch. The latter are used for processes where batches are allowed to be mixed, split and recycled (network processes), which implies that the identity of a batch is not preserved and material balances (rather than batch tracking) are important. The approaches for sequential processes rely on the notion of "batch", and the decisions such as the number and size of batches are often fixed (given) (Mendez et al., 2006; Pinto and Grossmann, 1995; Hui and Gupta, 2000; Mendez et al., 2001). These sequence-based formulations typically include assignment and sequencing constraints expressed in terms of batches: each batch has to be assigned to exactly one unit at each stage/operation and the batches assigned to the same unit have to be sequenced. In addition to exploiting the sequential structure of the process, most sequence-based approaches are based on a series of restricting assumptions (e.g. unlimited utilities and storage capacity) which lead to highly specialized but computationally effective formulations. On the other hand, the approaches developed for network processes are rather general (Kondili et al., 1993; Pantelides, 1994). These network-based formulations can be used to address problems with storage as well as resource constraints. The main idea behind these approaches is the partition of the scheduling horizon into (discrete or continuous) time periods and the enforcement of i) material balance, ii) unit assignment and iii) resource constraints for each period. Importantly, these constraints are expressed similarly to flow balance constraints in networks. One of the limitations of this "flow" representation, however, is that it does not keep track of batches. Therefore, network-based approaches cannot be used to address problems in sequential processes. Therefore, there is no single approach that can be used to represent and address all scheduling problems. Furthermore, problems in facilities with different material routing restrictions (which are quite common in many sectors) are decomposed into simpler (sequential and network) subproblems, leading to suboptimal solutions.
AB - In the context of process scheduling, existing methods can be broadly classified into sequential and network-based. The former are used to address scheduling problems in processes where batches are processed in a predefined sequence of operations; mixing and splitting of batches are not allowed thereby preserving the identity of each batch. The latter are used for processes where batches are allowed to be mixed, split and recycled (network processes), which implies that the identity of a batch is not preserved and material balances (rather than batch tracking) are important. The approaches for sequential processes rely on the notion of "batch", and the decisions such as the number and size of batches are often fixed (given) (Mendez et al., 2006; Pinto and Grossmann, 1995; Hui and Gupta, 2000; Mendez et al., 2001). These sequence-based formulations typically include assignment and sequencing constraints expressed in terms of batches: each batch has to be assigned to exactly one unit at each stage/operation and the batches assigned to the same unit have to be sequenced. In addition to exploiting the sequential structure of the process, most sequence-based approaches are based on a series of restricting assumptions (e.g. unlimited utilities and storage capacity) which lead to highly specialized but computationally effective formulations. On the other hand, the approaches developed for network processes are rather general (Kondili et al., 1993; Pantelides, 1994). These network-based formulations can be used to address problems with storage as well as resource constraints. The main idea behind these approaches is the partition of the scheduling horizon into (discrete or continuous) time periods and the enforcement of i) material balance, ii) unit assignment and iii) resource constraints for each period. Importantly, these constraints are expressed similarly to flow balance constraints in networks. One of the limitations of this "flow" representation, however, is that it does not keep track of batches. Therefore, network-based approaches cannot be used to address problems in sequential processes. Therefore, there is no single approach that can be used to represent and address all scheduling problems. Furthermore, problems in facilities with different material routing restrictions (which are quite common in many sectors) are decomposed into simpler (sequential and network) subproblems, leading to suboptimal solutions.
UR - http://www.scopus.com/inward/record.url?scp=78751547528&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=78751547528&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:78751547528
SN - 9780816910656
T3 - AIChE Annual Meeting, Conference Proceedings
BT - 10AIChE - 2010 AIChE Annual Meeting, Conference Proceedings
T2 - 2010 AIChE Annual Meeting, 10AIChE
Y2 - 7 November 2010 through 12 November 2010
ER -