6307877849

Abstract

The dissertation concerns problems related to building project schedules in organizations managed by projects. In preceding years, the increasing interest in this form of production and supply of both Capital and consumer goods in various branches of economies could be noticed. Regardless of character, its Capital structure or economic goals, a project always causes expenses related to the usage of different resources (e.g. machines, means of transport, materials, or manpower), the energy consumption or buying services, both materiał and non-material. The Schedule as the one of the most important instruments in the proper management (within the scope of planning, organization and control) of realization of project, increases the chances of achieving the satisfying effect in the planned deadline and within the approved budget. Because of the importance of this tool there comes a ąuestion if it is possible to optimize the Schedule due to the economic factors based on the modern metaheuristics.

There are three fundamental premises to conduct this work. Most of the Computer based project management support systems do not give the possibility of optimization of the Schedule, enabling only its building and evaluating. On the other hand, the drawbacks of most of the optimization methods applicable to the software are: high level of the model simplification and time criteria direction, which is unsatisfactory for evaluation of economic activity. In such a case, it can be noticed that estimation of the economic result is a consequence of accepted solution after building the Schedule, instead of analysis of economic effect during decision process. We can also observe the increasing interest and remarkable progress in the area of scientific research and practical implementation of metaheuristics.

The dissertation starts with the chapter devoted to basie notions and definitions related to the project management area and scheduling. Secondly, we present the classification of project scheduling problems ffom the perspective of constraints and evaluation criteria. Next, we introduce the review of optimization methods, including: exact algorithms, priority heuristics and metaheuristics. In chapter 3 we analyze the evaluation of the project Schedule quality, both organizational and economic. We present the influence of each decision variable on time and value criteria in the scheduling process. In chapter 4 we formulate the model of project scheduling with economic factors which allows to describe the relations between decision variables and economic effects in the precise way. The constraints of the decision variables and the dynamie criteria of economic effects of Schedule estimation are formulated as well. Chapter 5 deals with the analysis of detailed procedures of the exact and heuristic methods, that can be used in the examined algorithms regarding to the previously constructed model. We present available representation schemes, Schedule generation procedures, exploration and exploitation techniques and searching strategies in the genetic algorithm, simulated annealing, tabu search, scatter search and population learning algorithm. In chapter 6, the description of assumptions and results of computational experiments in the possibly wide group of problems with the representative procedures’ parameters is presented. The dissertation is concluded by the summary of the results and the directions of further research.

The examination of genetic algorithm, simulated annealing, tabu search and hybrid algorithm (combining features of genetic algorithm and simulated annealing) in the group of different size and complexity of problems has confirmed that there is an interesting possibility to improve the project schedules effectively by the usage of modern metaheuristics. We noticed also that the optimization of time criteria does not guarantee good economic effects. We have proved that the hardness (or computational complexity) of the problem highly depends on the number of activities and restrictiveness of the resource constraints. The topology of project network is observed to have less importance. The model of project scheduling formulated during the research is found elastic and applicable to the different kinds of project management support systems, what can be direction of further research.