Multi-skill resource-constrained project scheduling problems : models and algorithms

Detalhes bibliográficos
Autor(a) principal: Almeida, Bernardo F.
Data de Publicação: 2017
Idioma: eng
Título da fonte: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
Texto Completo: http://hdl.handle.net/10451/34858
Resumo: Tese de doutoramento, Estatística e Investigação Operacional (Otimização), Universidade de Lisboa, Faculdade de Ciências, 2018
id RCAP_67dcbf900fa650c070aadf205ecb7c2d
oai_identifier_str oai:repositorio.ulisboa.pt:10451/34858
network_acronym_str RCAP
network_name_str Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
repository_id_str https://opendoar.ac.uk/repository/7160
spelling Multi-skill resource-constrained project scheduling problems : models and algorithmsTeses de doutoramento - 2018Domínio/Área Científica::Ciências Naturais::MatemáticasTese de doutoramento, Estatística e Investigação Operacional (Otimização), Universidade de Lisboa, Faculdade de Ciências, 2018In this dissertation, project scheduling problems with multi-skill resources are investigated. These problems are commonly found in companies making use of human resources or multi-purpose machinery equipment. The general problem consists of a single project comprising a set of activities. There are precedence relations between the activities. Each activity requires one or several skills for being processed and for each of these skills, more than one resource may be needed. The resources have a unitary capacity per time unit and may master more than one skill. The resources can contribute with at most one skill to at most one activity that requires it, in each time unit. It is usually assumed that the resources are homogeneous, i.e., the proficiency at which each skill is performed is the same across all resources that master that skill. Preemption is not allowed, which implies that once an activity starts being processed it cannot be interrupted. When a resource is assigned to perform a skill for an activity, it remains in that status for the whole processing time of the activity. The objective of the problem is to schedule all the activities, satisfying all constraints such that the makespan of the project is minimized. After introducing a framework to the realm of project scheduling problems with multi-skill resources and highlighting the main objectives and contributes of this thesis, a state-of the-art review on the topic is presented. The particular problem investigated in this document is then described in detail and its specific features are discussed. To that end, a continuous-time mathematical formulation from the literature is revisited, an example of the problem is presented and some aspects related to the computation of feasible solutions are discussed. This last topic is of major relevance when dealing with problems that combine personnel staffing with project scheduling. In order to properly assess the quality of solutions obtained by the methodological developments proposed in this thesis, it became necessary to develop an instance generator to build a set of instances larger than those existing in the literature. After formally proposing such generator, we detail the characteristics of the two sets of instances considered for the computational experiments to be performed. In the next sections of the document, the solution methodologies developed within the scope of this thesis are presented and thoroughly discussed. A wide range of mathematical formulations is studied, two of which are first proposed in this document. From the assessment of their ability both to compute feasible and possibly optimal solutions and to derive good lower bounds (stemming from their linear programming relaxations) to the problem, it will become clear that the so-called discrete-time formulations yield the strongest lower bounds whereas a continuous-time formulation from the literature proved to be the most suitable for solving instances of the problem to optimality. This trend is observed for both sets of instances considered. Two constructive lower bound mechanisms proposed for the resource-constrained project scheduling problem are extended to account for the existence of multi-skill resources and multi skill requirements of the activities. The results reveal that such methods improve the lower bounds achieved by the studied mathematical formulations for some instances. Real-world project scheduling problems usually involve a large number of activities, resources and skills. Hence, the use of exact methods such as the standard techniques for tackling the aforementioned mathematical models, is often impractical. When faced with this kind of situations, a project manager may consider preferable to have a good feasible solution, not necessarily an optimal one, within an admissible time, by means of an approximate method. A close look into the problem being investigated in this thesis reveals that it has some features that are not present in some well-studied particular cases of it, namely the notion of skill—multi skill resources and skill requirements of the activities. Hence, with the objective of developing approximate solution methodologies that better exploit the specific characteristics of the problem at hand, two new concepts are introduced: activity grouping and resource weight. The well-known parallel and serial scheduling schemes, proposed originally for the class of resource-constrained project scheduling problems, are extended to our problem setting and the two above-mentioned concepts are incorporated into these two new frameworks. Such frameworks use well-known activity priority rules for defining the order by which the activities are selected to be scheduled and resource weight rules to determine a set of resources that meets the requirements of all the activities to be scheduled at each time with the least total cost (weight). Thereafter, two heuristic procedures making use of those schedule generation schemes are proposed, namely a multi-pass heuristic built upon the parallel scheduling scheme and a biased random-key genetic algorithm. The idea of computing a feasible solution using the so-called backward planning is also considered in both methods. The multi-pass heuristic retrieves the solution with the minimum makespan after performing a specific number of passes, each associated with a unique combination of the considered activity priority rules, the developed resource weight rules and the two precedence networks: forward and backward. The biased random-key genetic algorithm is a metaheuristic whose developed chromosome structure encodes information related to: (i) the priority values of the activities; (ii) the weights of the resources; (iii) how a chromosome is decoded, i.e., the scheduling scheme and precedence network scheme to be used for computing the associated makespan. By embedding all this information into the chromosomes, it becomes possible to take advantage of the evolutionary framework of the biased random-key genetic algorithm, which tends to allow the evolution of such data (change in their values) over time, towards better makespan valued solutions. Three variants of the biased random-key genetic algorithm are considered with regard to the type of scheduling generation scheme to be used for decoding its chromosomes: (i) all chromosomes are decoded with the parallel scheduling scheme; (ii) all chromosomes are decoded with the serial scheduling scheme; (iii) the scheduling scheme to be used for decoding each chromosome depends on the value of the associated parameter which is embedded in the chromosome. The computational results revealed that the proposed multi-pass heuristic is an efficient algorithm for computing feasible solutions of acceptable quality within a small computational time whereas the biased random-key genetic algorithm is a robust algorithm and a more competitive approximate approach for computing feasible solutions of higher quality, especially for harder instances such as those of medium and large dimensions. We conclude this thesis with an overview of the work done and with some directions for further research in terms of methodological developments and of some potentially interesting extensions of the addressed problem.Conceição, Francisco Alexandre Saldanha da Gama Nunes da, 1968Correia, Isabel Cristina Silva, 1967-Repositório da Universidade de LisboaAlmeida, Bernardo F.2018-09-25T15:58:32Z201820172018-01-01T00:00:00Zdoctoral thesisinfo:eu-repo/semantics/publishedVersionapplication/pdfhttp://hdl.handle.net/10451/34858TID:101439814enginfo:eu-repo/semantics/openAccessreponame:Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)instname:FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologiainstacron:RCAAP2025-03-17T13:57:44Zoai:repositorio.ulisboa.pt:10451/34858Portal AgregadorONGhttps://www.rcaap.pt/oai/openaireinfo@rcaap.ptopendoar:https://opendoar.ac.uk/repository/71602025-05-29T02:59:07.159537Repositórios Científicos de Acesso Aberto de Portugal (RCAAP) - FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologiafalse
dc.title.none.fl_str_mv Multi-skill resource-constrained project scheduling problems : models and algorithms
title Multi-skill resource-constrained project scheduling problems : models and algorithms
spellingShingle Multi-skill resource-constrained project scheduling problems : models and algorithms
Almeida, Bernardo F.
Teses de doutoramento - 2018
Domínio/Área Científica::Ciências Naturais::Matemáticas
title_short Multi-skill resource-constrained project scheduling problems : models and algorithms
title_full Multi-skill resource-constrained project scheduling problems : models and algorithms
title_fullStr Multi-skill resource-constrained project scheduling problems : models and algorithms
title_full_unstemmed Multi-skill resource-constrained project scheduling problems : models and algorithms
title_sort Multi-skill resource-constrained project scheduling problems : models and algorithms
author Almeida, Bernardo F.
author_facet Almeida, Bernardo F.
author_role author
dc.contributor.none.fl_str_mv Conceição, Francisco Alexandre Saldanha da Gama Nunes da, 1968
Correia, Isabel Cristina Silva, 1967-
Repositório da Universidade de Lisboa
dc.contributor.author.fl_str_mv Almeida, Bernardo F.
dc.subject.por.fl_str_mv Teses de doutoramento - 2018
Domínio/Área Científica::Ciências Naturais::Matemáticas
topic Teses de doutoramento - 2018
Domínio/Área Científica::Ciências Naturais::Matemáticas
description Tese de doutoramento, Estatística e Investigação Operacional (Otimização), Universidade de Lisboa, Faculdade de Ciências, 2018
publishDate 2017
dc.date.none.fl_str_mv 2017
2018-09-25T15:58:32Z
2018
2018-01-01T00:00:00Z
dc.type.driver.fl_str_mv doctoral thesis
dc.type.status.fl_str_mv info:eu-repo/semantics/publishedVersion
status_str publishedVersion
dc.identifier.uri.fl_str_mv http://hdl.handle.net/10451/34858
TID:101439814
url http://hdl.handle.net/10451/34858
identifier_str_mv TID:101439814
dc.language.iso.fl_str_mv eng
language eng
dc.rights.driver.fl_str_mv info:eu-repo/semantics/openAccess
eu_rights_str_mv openAccess
dc.format.none.fl_str_mv application/pdf
dc.source.none.fl_str_mv reponame:Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
instname:FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologia
instacron:RCAAP
instname_str FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologia
instacron_str RCAAP
institution RCAAP
reponame_str Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
collection Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
repository.name.fl_str_mv Repositórios Científicos de Acesso Aberto de Portugal (RCAAP) - FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologia
repository.mail.fl_str_mv info@rcaap.pt
_version_ 1833601552287268864