University of the Aegean
Νοήμονες μέθοδοι επίλυσης προβλημάτων εξομάλυνσης πόρων σε έργα
Abstract
dc:descriptionThe present thesis deals with Resource Leveling optimization problems in projects. Resourceleveling is a problem that has not been accurately solved so far, for large scale projects. Forsmall scale projects accurate solutions can be found in literature, achieving global optimum. Formedium and large scale project as well as for projects with particular characteristics (largeduration of activities, complex associations among activities) approximate solutions ofacceptable quality can be found in literature, but further field for improvement exists.The present dissertation achieves improved approximate solutions in resource leveling formedium and large scale projects, implementing and comparing four (4) new methodologies, allrelated to computational intelligence. In addition, the dissertation proposes a number of suitablebenchmark problems for resource leveling in projects, for further experimentation andcomparison on new related methods in the future.Concerning methodological innovation, the dissertation proposes a new, more effective geneticalgorithm from the ones existing in literature and two more hubrid intelligent techniques whichprove to handle effectively the problem of resource leveling optimization. The proposedapproach is based on the development of an innovative methodology for the production offeasible alternative starts of the project’s activities for the formation of improved resourceprofiles. All known resource profile evaluation functions are tested (seven in total, according torelated literature) which correspond to different approaches to the optimization problem, relatedto the preciousness of the resource, the need for uniform resource distribution, etc .Due to the existing variety of resource profile evaluation functions, a sequential application ofall known functions with all possible combinations is proposed, thus aiming at the best possibleaverage resource profile tracking, in such a way that all the involved resource profile evaluationfunctions can be partially satisfied. The result of this method in most cases agrees with theresults obtained from the main proposed GA approach.Within the dissertation, several studies of other researchers for the problem of resource levelingare reported and analyzed in detail. The proposed resource leveling methodologies provesuperior to other competitive approaches existing in literature for medium and large scaleprojects, or compete to them worthily in small scale problems where the exact optimum isknown. Comparisons are possible only with those research reports that provide related projectdata. Nevertheless, most projects presented in related literature so far, concern small scaleproblems.From the present dissertation a number of published papers in referred conferences and journalshas resulted (while a few more have been submitted for publication), which covers all partial research questions posed and investigated from the PhD candidate during his study period
Degree
thesis:*- Grantor dc:publisher
- University of the Aegean
- Year dc:date
- 2015
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Kyriklidis, Christos
- Κυρικλίδης, Χρήστος
Subjects
dc:subject × 9Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/42879