Abstract

This thesis presents a new methodology for optimizing earthmoving operations using computer simulation and genetic algorithms. It provides an optimization tool geared towards selection of near-optimum fleet configurations. The optimization used in the selection of fleets accounts for availability of equipment and aims at minimizing the total project cost or duration. The simulation process, in the proposed methodology, utilizes discrete event simulation (DEVS) and object oriented modeling. Different features of object orientation are employed including classes, dynamic data structure and polymorphism. The three-phase simulation approach, rather than process interaction, was employed to control the dynamics of the simulation process and track involved activities. This simulation approach is considered most appropriate for object oriented simulation (OOS). The optimization process uses a developed genetic algorithm to search for a near-optimum fleet configuration that reduces project total cost. The algorithm considers a set of qualitative and quantitative variables that influence the production of earthmoving operations. Qualitative variables represent the models of equipment used in each fleet scenario, whereas, quantitative variables represent the number of equipment involved in each scenario. The proposed methodology accounts for: (1) uncertainties associated with earthmoving operations; (2) optimization of project duration or its total cost, considering equipment availability; and (3) realistic estimates of haulers' travel time. It also makes full use of object oriented features and is implemented in a prototype software system named SimEarth . The system consists of five main components: (1) EarthMoving Simulation Program ( EMSP ); (2) Equipment Cost Application ( ECA ); (3) Equipment Database Application ( EDA ); (4) EarthMoving Genetic Algorithm ( EM_GA ), and (5) Output Reporting Module ( ORM ). Beside these main components, SimEarth is supported by: (a) Hauler's Travel Time Application ( HTTA ), and (b) EarthMoving Markup Application ( EMMA ). All system components are implemented in Microsoft ( MS ) environment except the dynamic sub-module of ORM component, which is implemented utilizing " Proof Animation " software. Five numerical examples were analyzed in order to validate and demonstrate the essential features of the system's components. A comprehensive case study of an actual project was analyzed in order to test the performance of the developed system (including the dynamic interaction among its components) and to illustrate the practical features of the developed methodology. The project involves the construction of a large rockfill dam, located in the northern part of the province of Quebec.