The permutation flowshop scheduling problem: analysis, solution procedures and problem extensions
- Autores: Víctor Fernández-Viagas Escudero
- Directores de la Tesis: José Manuel Framiñán Torres (dir. tes.)
- Lectura: En la Universidad de Sevilla ( España ) en 2016
- Idioma: inglés
- Número de páginas: 253
- Tribunal Calificador de la Tesis: Jaime Domínguez Abascal (presid.), Paz Pérez González (secret.), Rubén Ruiz García (voc.), Imma Ribas Vila (voc.), Rainer Leisten (voc.)
- Materias:
- Enlaces
- Tesis en acceso abierto en: Idus
- Resumen
During the past twenty five years, following the massive use of internet and the EU single Market, European manufacturing companies struggle in a more competitive market, where firms from different countries must fight for common customers. As a consequence, prices of the products have decreased and the efficiency in the production processes of the companies have become more and more important. Nowadays, this fact is also increasing due to the competition from companies in developing countries whose labour cost is substantially lower. Therefore, production management is a key element for companies to survive. Production management involves decision making over several issues such as master scheduling, material requirements planning, capacity planning, manufacturing scheduling, ... Among these decisions, manufacturing scheduling plays an essential role on resource productivity and customer service. Its role is also increasing in many service industries as transportation, computer and communications industries, which are moving towards manufacture-to-order and virtual environments. Manufacturing scheduling deals with the determination of the jobs which are processed for each resource in each instant of time, i.e. establishes the schedules of the resources along the horizon under consideration. In order to determine the best schedule for the shop floor, both the specific constraints and the goal of the shop have to be considered. In these environments, the difficulty of the scheduling problem increases and becomes NP-hard even for the most simple scheduling problems, being extremely complex for real manufacturing scenarios. Additionally, scheduling decisions should be made in short time intervals requiring a rapid response time, due to several aspects such as the lifetime of a schedule, the delay in the suppliers, arrivals of new jobs to be processed, rescheduling due to failures while processing a job, .... All these issues strongly stress the need to find fast and efficient solution procedures (i.e. heuristics and metaheuristics) for solving manufacturing scheduling problems. In practice, several processing layouts have been adopted by companies to manufacture their products. Among them, the Permutation Flowshop Scheduling Problem (PFSP in the following), which is the problem addressed in this Thesis, stands out as the most relevant, being one of the most studied problems in Operations Research. There are several reasons for this fact: On the one hand, the flow shop layout is the common configuration in many real manufacturing scenarios, as it presents several advantages over more general job shop configuration, and, in addition, many job shops are indeed a flow shop for most of the jobs. On the other hand, many models and solution procedures for different constraints and layouts have their origins in the flowshop scheduling problem, which increases the importance to find efficient algorithms for this scheduling problem. Despite the huge number of research conducted on the PFSP, we believe that there is room for improving the current state of the art in the topic by: 1. deepening the understanding of the problem with respect to their input parameters, 2. devising new approximate solution procedures for the common employed objectives, and 3. addressing problem extensions to capture more realistic situations. To carry out this goal, the following general research objectives are identified: 1. To review the PFSP literature for the most common objectives, i.e. makespan, total completion time and due-date-based objectives (total tardiness, and total earliness and tardiness). 2. To analyse the influence of the processing times and due dates of the jobs on the PFSP. 3. To provide schedulers with faster and more efficient heuristics and metaheuristics to solve the PFSP for makespan, total completion time, total tardiness, and total earliness and tardiness minimisation. 4. To demonstrate the efficiency and good performance of the solution procedures developed in Goal 3. 5. To extend the proposals in Goal 3 to some constrained PFSP based on real manufacturing environments. To achieve these objectives, the Thesis have been structured in five parts as follows: - Part I is divided into two chapters. In Chapter 1.1, we introduce this Thesis and discuss its main contributions. In Chapter 2, the problem under consideration is stated. The measures to compare approximated algorithms are discussed in Chapter 3. There, the benchmarks used to evaluated the algorithms are introduced and an alternative indicator is proposed to overcome some problems detected using the traditional ones. - In Part II, we analyse the problem in detail along three chapters. Dealing with Objective 1, the main contributions in the literature are review for the most-common objective functions in Chapter 4. Additionally, in Chapter 5, we extensively study the behaviour of the problem depending on the configuration of the shops, i.e. processing times and due dates of the jobs (see Goal 2). - In Part III, we propose new novelties efficient algorithms to solve the PFSP under several objectives. The procedures, constructive and improvement heuristics and metaheuristics, exploit the specific structure of the problem to both reduce the computational times of them and improve the quality of the solutions. Additionally, they are validated in extensive computational evaluations, comparing them with the state-of-the-art algorithms under the same conditions. More specifically, this part is divided in four chapters and addresses the general research objectives GO3 and GO4. Firstly, a new tie-breaking mechanism to minimise makespan, which can be incorporated in the two most efficient algorithms for the problem, is proposed in Chapter 6. In Chapter 7, two efficient constructive heuristics are proposed to minimise total flowtime. Several tie-breaking mechanisms are proposed and compared to minimise total tardiness in Chapter 8. Finally, four procedures to minimise total earliness and tardiness are proposed in Chapter 9. - In Part IV, focused in more real manufacturing environment, new constraints are added to the traditional problem as well as different consideration and interaction between factories are taken into account. The proposed environments are solved using efficient approximate methods taken into consideration ideas of the traditional PFSP. More specifically, an iterated non-population algorithm to minimise makespan subject to a maximum tardiness is proposed in Chapter 10. In the Chapter 11, we add the blocking constraints to the traditional PFSP. These constraints take into consideration limited buffers between the machines. This problem, of permutation nature, is solved by means of an efficient beam-search-based constructive heuristic trying to minimise the total completion time. In Chapter 12, we consider the parallel flowshop scheduling problem also denoted as distributed PFSP where several identical flowshop or even flowshop factories are available in parallel to assign the jobs. The problem is solved using a bounded-search iterated greedy algorithm - Finally, in Part V, the conclusions of this research and future research lines are discussed.
