On deterministic online scheduling: Major considerations, paradoxes and remedies
Introduction
Process industries employ a complex chain of operations (processes) which compete for limited resources. Hence, a resource allocation problem arises wherein decisions with regards to the start time of operations, how much to process, the equipment to utilize and several associated decisions need to be taken on a repetitive basis.
Traditionally, schedules have been constructed by experienced scheduling personnel (schedulers) using spreadsheets or heuristics-based software. In the last two decades, however, there has been an increasing thrust towards using advanced computational tools to achieve more profitable schedules. Hence, working with scheduling models to optimally plan operations has become an important problem in process systems engineering (Shobrys and White, 2002, Kelly and Mann, 2003a, Kelly and Mann, 2003b, Harjunkoski et al., 2014). Although the process systems engineering (PSE) community has worked on building accurate models and better solution methods, the aspect of rescheduling has received limited attention. Rescheduling has been emphasized in some works (Cott and Macchietto, 1989, Kanakamedala et al., 1994, Huercio et al., 1995, Rodrigues et al., 1996, Vin and Ierapetritou, 2000), but in most cases scheduling is still thought to be a static open loop problem wherein if rescheduling is carried out, the emphasis is only on restoring feasibility or optimality to the current static schedule. Quantifying the quality of the implemented schedule obtained by rescheduling has not been addressed. Accordingly, the goal of this paper is to study how the open-loop optimization problem impacts the quality of the implemented (closed-loop) schedule.
In a dynamic environment multiple disturbances such as task delays, yield losses, unit breakdowns or rush order arrivals can render a previously computed schedule suboptimal or even infeasible. In addition, Subramanian et al. (2012) emphasized that rescheduling also needs to be performed due to advancement of the scheduling horizon over which the schedule is computed even when there are no disturbances. Rescheduling has been demonstrated to lead to lower inventory accumulation in supply chains (Subramanian et al., 2014).
In this work, we first and foremost show that open-loop and closed-loop scheduling are two different problems, even in the deterministic case, when no uncertainty is present. Thereafter, we investigate how the design of the open-loop optimization problem affects the quality of the resulting implemented closed-loop schedule. The design attributes we study are scheduling horizon length, rescheduling frequency and optimality gap of each open-loop optimization. We choose illustrative process networks and extensively study combinations of the aforementioned attributes over a reasonably exhaustive set of short term demand patterns, production load, and scheduling objectives. From these test cases, we identify trends, and problem instance characteristics which can facilitate in carefully choosing the three attributes.
The paper is structured as follows: In Section 2, we present background on chemical production scheduling and discuss past work on rescheduling. In Section 3, we discuss various design considerations in closed-loop scheduling through motivating examples. In Section 4, we describe the problem library we use. We then present results in Section 5, followed by, a discussion in Section 6. Thereafter, we present several case studies in Section 7, and then provide conclusions in Section 8. We use lower case Latin characters for indices, uppercase bold Latin characters for sets, lowercase Greek characters for parameters, and uppercase Latin characters for variables.
Section snippets
Previous research
A schedule can become suboptimal or even infeasible due to, for example, a disruptive event or a change in resource availability. The revision of an existing schedule in response to disruptions or changes is termed as reactive scheduling or rescheduling. When the focus of rescheduling is only on a schedule for a fixed horizon, it results in a shrinking-horizon problem. On the other hand, when the current schedule is possibly revised, and also additional decisions are made for the time ahead, it
Considerations in online scheduling
In a dynamic environment with new incoming information, disruptive events or changes in availability of resources, new schedules are computed and implemented. Although this practice of rescheduling is adopted routinely in production facilities, it is not known how it affects the overall performance of the production systems (Vieira et al., 2003). While formulating a lot-sizing formulation for reduced nervousness in production schedules Kazan et al. (2000) noted two factors that motivate
Problems, instance types, instances and runs
Once problem features are fixed (variable batch-sizes, utilities, setups), a problem is defined by the objective function (OBJ). An instance type, is a problem expressed for a process network (PN) and a demand pattern (D). A demand pattern, which is further discussed in Section 4.1.2, is a distribution of order due-times and sizes of each order. For each instance type (OBJ.PN.D), we define an instance designated by a sample (S) from the demand pattern under consideration. Finding the
Results
In Fig. 11, Fig. 12, Fig. 13, Fig. 14, Fig. 15, Fig. 16, each subplot corresponds to a single instance type, and carries a tag of the form FpLm which denotes that orders are due, on an average every p hours, and each order has an average size of m tons for each product. For example, F6L8 means orders are due, on an average, every 6 h with mean order size of 8 tons of each product. Table 1 shows the values of FpLm used and the corresponding approximate production capacity utilization for PN-3 (
Rescheduling frequency
In general, higher rescheduling frequency leads to better closed-loop quality, however, there exists an upper threshold, beyond which closed-loop performance does not improve. Importantly, lower than a certain frequency leads to very bad closed-loop schedules. One of the factors on which this lower threshold depends, is the frequency of orders. If orders are due frequently, then rescheduling infrequently leads to poor closed-loop schedules. Even when orders are due infrequently, it is still
Case study-I: Design attributes
We compute a closed-loop schedule for PN-1 (Fig. 1) for 1-week (cost minimization) to meet a demand pattern created from tuple (p = 6, Δp = 1, m = 6, Δm = 0). First, we consider MH = 24, RF = 3, and OPTCR = 0%, and obtain a closed-loop cost of $190,744. Next, we consider MH = 36, RF = 6, and OPTCR = 5%. We use a larger OPTCR and RF = 6 to emulate the practical situation where solving a larger problem to optimality will require more time than the time available between iterations. Surprisingly, in the second
Conclusions
We presented a framework for the analysis of closed-loop schedules. First, we showed that open-loop and closed-loop scheduling are two different problems, even in the deterministic case when no uncertainty is present. Applying methods to improve solution to the open-loop problem, does not necessarily translate to good solutions for the closed-loop problem. Second, we found that it is important to reschedule periodically, even when there are no “trigger” events, something that is in contrast
Acknowledgements
The authors would like to acknowledge support from the National Science Foundation under grants CMMI-1334933 and CBET-1264096, as well as the Petroleum Research Fund under grant 53313-ND9. CTM would like to thank Dr. James Rawlings for fruitful discussions on model predictive control and, specifically, on the role of feedback.
References (62)
- et al.
Scheduling optimization under uncertainty – an alternative approach
Comput. Chem. Eng.
(2003) - et al.
Minimizing the effects of batch process variability using online schedule modification
Comput. Chem. Eng.
(1989) - et al.
Medium-term planning of a multiproduct batch plant under evolving multi-period multi-uncertainty by means of a moving horizon strategy
Comput. Chem. Eng.
(2010) - et al.
A time scale-bridging approach for integrating production scheduling and process control
Comput. Chem. Eng.
(2015) - et al.
Scope for industrial applications of production scheduling models and solution methods
Comput. Chem. Eng.
(2014) - et al.
A framework for schedule evaluation with processing uncertainty
Comput. Chem. Eng.
(1999) - et al.
Incorporating on-line scheduling strategies in integrated batch production control
Comput. Chem. Eng.
(1995) - et al.
A new robust optimization approach for scheduling under uncertainty: II. Uncertainty with known probability distribution
Comput. Chem. Eng.
(2007) - et al.
New lot-sizing formulations for less nervous production schedules
Comput. Oper. Res.
(2000) - et al.
Hierarchical decomposition heuristic for scheduling: coordinated reasoning for decentralized and distributed decision-making problems
Comput. Chem. Eng.
(2008)
A general algorithm for short-term scheduling of batch operations-I. MILP formulation
Comput. Chem. Eng.
Process scheduling under uncertainty: review and challenges
Comput. Chem. Eng.
A new robust optimization approach for scheduling under uncertainty: I. Bounded uncertainty
Comput. Chem. Eng.
Dynamic scheduling in multiproduct batch plants
Comput. Chem. Eng.
State-of-the-art review of optimization methods for short-term scheduling of batch processes
Comput. Chem. Eng.
Preprocessing and tightening methods for time-indexed MIP chemical production scheduling models
Comput. Chem. Eng.
Reactive scheduling framework based on domain knowledge and constraint programming
Comput. Chem. Eng.
A comprehensive constraint programming approach for the rolling horizon-based scheduling of automated wet-etch stations
Comput. Chem. Eng.
A fuzzy logic based production scheduling/rescheduling in the presence of uncertain disruptions
Fuzzy Sets Syst.
Reactive scheduling approach for multipurpose chemical batch plants
Comput. Chem. Eng.
Modeling and solving real-time scheduling problems by stochastic integer programming
Comput. Chem. Eng.
A general algorithm for short-term scheduling of batch operations – II. Computational issues
Comput. Chem. Eng.
Planning, scheduling and control systems: why cannot they work together
Comput. Chem. Eng.
A state-space model for chemical production scheduling
Comput. Chem. Eng.
Economic model predictive control for inventory management in supply chains
Comput. Chem. Eng.
A branch-and-bound algorithm for the solution of chemical production scheduling MIP models using parallel computing
Comput. Chem. Eng.
Multiple and nonuniform time grids in discrete–time MIP models for chemical production scheduling
Comput. Chem. Eng.
Theoretical framework for formulating MIP scheduling models with multiple and non-uniform discrete–time grids
Comput. Chem. Eng.
A fuzzy robust scheduling approach for product development projects
Eur. J. Oper. Res.
Improving the efficiency of discrete time scheduling formulation
Comput. Chem. Eng.
A robust optimization model for stochastic logistic problems
Int. J. Prod. Econ.
Cited by (59)
Integration of chemical process operation with energy, global market, and plant systems infrastructure
2024, Computers and Chemical EngineeringA decomposition-based two-stage online scheduling approach and its integrated system in the hybrid flow shop of steel industry
2023, Expert Systems with ApplicationsDiscrete-time and continuous-time scheduling approaches for crane-assisted pipeless process plants
2023, Computers and Chemical EngineeringProduction scheduling under demand uncertainty in the presence of feedback: Model comparisons, insights, and paradoxes
2022, Computers and Chemical EngineeringA self-learning state-aware rescheduling approach for robot-aided aero-engines fleet inspection
2022, Journal of Manufacturing Systems