Industrial Scheduling Problems Research Articles

An optimal control approach for scheduling mixed batch/continuous process plants with variable cycle time

Effective scheduling of operations in the process industry has the potential to achieve high economic returns. Process plants containing both batch and continuous units present a difficult scheduling problem. When these processes are modelled with batch cycle times as decision variables, the complexity of the problem is increased significantly. These problems when modelled in the conventional MILP formulation are extremely difficult to solve as they are NP-hard. Many solution methods require unacceptable amounts of time/memory to solve even a simple problem. This paper considers an alternate way to represent the problem in an attempt to improve solution performance. This method is based on optimal control and the hierarchical splitting of the optimisation problem. Results are compared to the traditional MILP formulation. The motivation for this work is an existing scheduling problem in the sugar milling industry. A smaller problem, which has characteristics of the sugar milling problem is considered for performance comparison. A substantial reduction in the complexity required to solve several test-cases of this problem is achieved using the optimal control formulation, while maintaining good quality solutions.

Sets formulation to schedule mixed batch/continuous process plants with variable cycle time

Effective scheduling of operations in the process industry has the potential to achieve high economic returns. Process plants containing both batch and continuous units present a difficult scheduling problem. When these processes are modelled with batch cycle times as decision variables, the complexity of the problem is increased significantly. These problems when modelled in the conventional MILP formulation, are extremely difficult to solve as they are NP-hard. Many solution methods require unacceptable amounts of time/memory to solve even a simple problem. A formulation based on the set-covering principle that constructs feasible sub-schedules is considered. Modelling considerations, such as identifying and generating sub-schedules are discussed. The motivation for this work is an existing scheduling problem in the sugar milling industry. A smaller problem, which contains the important characteristics of the sugar milling problem, is initially considered for performance comparison. A substantial reduction in the complexity required to solve the problem is achieved using the sets formulation. The sets implementation is then applied to a sugar mill scheduling case study. It is found that the method is able to return objective function values 45% better than current industry practice. Conventional MILP approaches are unable to solve this larger case study.

A short-term scheduling problem in the paper-converting industry

Due to increasing competition in international markets, the efficient usage of available production resources has become more and more important to industry. Production planning facilities play a key role in achieving this goal. In this paper we will discuss a short-term production scheduling problem in the paper-converting industry and formulate it as two alternative variants of a mixed integer linear programming (MILP) model on which a sequential updating procedure is applied. The objective is to obtain short-time production schedules for large-scale real-world systems. The features of interest are those of practical economical significance as well as computational efficiency.

An implementation of man-software cooperative scheduling: The IO software

Scheduling problems in industry are more and more specific, and a generic software can hardly provide the user with an acceptable solution in a fully automated way. As a consequence, cooperation between man and software becomes a major issue for industrial schedulers. We analyse in this paper the specific requirements induced by a cooperative approach and we suggest some solutions oriented towards visualization tools, constraint management and interactive planning. The concept of generalized constraints is suggested in order to increase the modelling capacities of the software. The implementation of these facilities in an industrial software called IO is described.

Heuristic Control of a Constraint-Based Algorithm for the Preemptive Job-Shop Scheduling Problem

In the recent years, constraint programming has been applied to a wide variety of academic and industrial non-preemptive scheduling problems, i.e., problems in which activities cannot be interrupte...

A fuzzy linear programming model for production planning in coal mines

This article shows how fuzzy linear programming may be used to solve production scheduling problem in coal industry. First, a fuzzy linear programming model is presented. The proposed model is then tested on a hypothetical problem developed by using production cost estimates from independent coal mines in the states of Virginia, Illinois and Pennsylvania. The results of the model indicate that the model has potential for solving production scheduling problems in the coal industry.

Interactive scheduling in the chemical process industry

Due to the typical nature of the industry, the scheduling problem in the Chemical Process Industry (CPI) is a particularly difficult one. Consequently, fully automated scheduling in CPI is not in reach for the moment. Only humans are sufficiently dynamic to deal with the inevitable chaos, caused by a multitude of possible conflicting choices, and by the fuzziness and uncertainty of the parameters involved. We believe that a computer program (for the moment) can at best be an intelligent assistant. That is why we implemented SKYE, an interactive decision-aid tool for human schedulers, which is capable of handling a very large set of scheduling problems in the CPI. SKYE combines stochastic techniques with deductive techniques such as, heuristics, constraint propagation and temporal reasoning. For demonstration purposes we apply SKYE as a stand alone automatic module to a set of small example problems.

Primary production scheduling at steelmaking industries

This paper describes primary production scheduling in the steel industry—the problem and the approaches to the solution. The scheduling problem in steel plants is known to be among the most difficult of several industrial scheduling problems. We first describe the main steelmaking processes and show how scheduling affects the effectiveness of plant operations. We characterize the problems associated with scheduling steelmaking activities to achieve business objectives of delivering quality steel on time to customers, while minimizing operating costs. We then describe several different approaches for computerized scheduling solutions. They include application of techniques in operations research, artificial intelligence, and a hybrid of these two. We conclude by describing advanced techniques for integrated scheduling of steel plants.

Production scheduling problems in a textile industry

In this paper scheduling problems arising in a textile industry are analysed. The maximum tardiness is taken as an objective function. Bounds on the optimal solutions are derived in the general case; the worst-case behaviour of approximate algorithms is investigated in case of uniform looms and a polynomial algorithm is presented for the case of equal looms with identical deadlines. Finally an LP-based heuristic is presented for the general case.

Scheduling manufacturing systems

In this paper, three industrial scheduling problems arising in manufacturing systems are discussed. The first problem is the single machine scheduling problem with sequence-dependent setup times and precedence constraints. The mixed-integer formulation proposed is simple and easy to implement. The problem is solved with an efficient heuristic algorithm. The second problem is a machine cell scheduling problem. A new dispatching rule is developed to minimize the total tool setup time. The performance of the proposed rule was tested against three priority rules: first-in—first-out, shortest processing time, and longest processing time. The results of the simulation experiment showed that the dispatching rule performs better than the rules tested. The third problem is concerned with scheduling laser cutting operations. The overall objective is to minimize the number of overproduced parts and randomly nested parts. An integer programming formulation is proposed.

An approximate method for the production scheduling of industrial batch processes with parallel units

In the first part of this paper, we present the results of a survey of production scheduling problems occurring in a broad range of industries, including the manufacture of polymers and aromas, and the processing of minerals. The surprising result of interviews with plant managers and production schedulers is the remarkable similarity in the process structures, product sequencing constraints, and scheduling objectives important to these diverse industries. Scheduling to reduce tardiness of customer orders was found to be the most important objective. A key requirement was the ability to handle realistic constraints on product sequencing and restrictions on the number of units which can make a given product. Thus, a class of industrial scheduling problems important to a broad range of process industries is defined. This is significant, since a great deal of research effort has, in the past, been devoted to problems with other scheduling objectives, without sequencing constraints and with identical units. In the second part, a solution method is presented which is shown to obtain optimal or near-optimal solutions for processes involvin stage of nonidentical parallel units. Sequence-dependent cleanouts, unit- and product-dependent processing times and product-to-unit assignment constraints are modeled. Test problems with up to 12 process units and up to 100 customer orders were solved, which is the typical size of industrial-scale problems. An evolutionary strategy was developed which generates an initial solution using heuristics, then systematically improves the solution until certain necessary conditions for optimality are satisfied. The method found optimal or near-optimal schedules for highly-constrained and difficult test problems with modest computational requirements. These results are significant since an effective solution method for single-stage industrial problems of this complexity has not heretofore been demonstrated.

Scheduling in a two-stage manufacturing process

This paper describes an industrial scheduling problem taken from a cable manufacturing process. An heuristic solution procedure known as the CMD rule is presented for this multi-product, multistage, continuous flow problem. The solution procedure is designed to minimize the sum of idle time of machines and in-process waiting at the second stage. A comparison of this rule with some standard scheduling rules is provided.

New alternatives for computer-scheduling of multi-product plants running batchwise under constraints

The paper discusses a new approach used for operations scheduling of multi-product plants. The algorithm allows schedules to be built and improved even when complex sets of constraints are to be dealt with. The ‘post-scheduling’ technique that was developed uses global functions or ‘modules’ called in an interactive fashion according to the decisions made by an operator and based on the results obtained at a previous step. The paper also studies the human decisional impact in combinatorial problem-solving and presents a generalization of constraint handling in scheduling systems. Global results deriving from an industrial scheduling problem that was solved using this method are also presented

Scheduling and the problem of computational complexity

Why is it that the problem of scheduling is so computationally difficult to solve? At last recent developments in modern mathematical complexity theory are providing some insights. The paper describes in essentially non-mathematical terms the computational technique known as the ‘Branch and Bound Method’. This, the best general optimising technique available for scheduling, is also shown to have its limitations. It now appears that efficient computational and optimising algorithms are unlikely ever to be found for all except special cases of the general industrial scheduling problem. It seems that heuristic (rule-of-thumb) methods leading to approximate solutions are likely to offer the only real promise for the future.

An Industrial application of the traveling salesman's subtour problem

Abstract Practical applications of the classical traveling salesman problem are rare because in most real world situations there are constraints upon the tour. To increase the range of useful applications, the classical structure must be modified to realistically deal with these constraints. In this application an industrial scheduling problem, called the traveling salesman's subtour problem, is formulated and defined. A time constraint restricts the salesman to a small subset of his total customers per planning period. Hence, the solution requires the simultaneous selection of the proper subset and the identification of the calling order. The problem is formulated to maximize expected sales subject to the time constraint. Lagrangean relaxation is used and by taking advantage of the underlying transportation structure of the problem the “gaps” are taken into account, thus providing a tight upper bound for the optimizing branch and bound routine in the algorithm.

The theory-practice gap in job-shop scheduling

This job-shop scheduling problem is perhaps the most common form of industrial scheduling problem, it is also probably the most complex. This paper examines the nature of the problem and reviews the state of the art of scheduling theory and its relevance to the solution of practical problems.