Ph.D. Theses
Permanent URI for this collection
Browse
Browsing Ph.D. Theses by Author "Bilge, Ümit."
Now showing 1 - 5 of 5
Results Per Page
Sort Options
Item Capacity modeling in aggregate production planning: multi-dimensional clearing functions and iterative linear programming-simulation approaches(Thesis (Ph.D.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2012., 2012.) Albey, Erinç.; Bilge, Ümit.In this dissertation, the capacity representation in aggregate production planning models (APPM) is investigated by focusing on two main capacity modeling philosophies in the literature, namely the clearing functions (CF) and iterative linear programming-simulation approaches (IA). The underlying strength of these approaches comes from facilitating the mutual link between capacity and state of the shop floor (SF). This thesis study contributes to both CF and IA techniques. The contribution to the former field has been the introduction of product based multi-dimensional disaggregated clearing functions (MDCFs). Several forms of MDCFs are developed and incorporated into the APPMs as the capacity modeling module. As a proof of concept, postulated forms are first tested on a single machine multi-product (SMMP) system under several experimental settings. The results reveal that new MDCF forms show more accurate prediction of product-level throughput hence generate better (i.e. more profitable) plans than the existing CF approaches and the classical linear programming approach. Then, postulated forms are extended to model the capacity in multi-machine multi-product (MMMP) systems and are tested under different aggregations, manufacturing flexibility levels and execution policies. As a contribution to the field of IA, a new and more robust mechanism is proposed based on rigorous experimental analysis of the convergence behavior of an existing IA based capacity modeling mechanism. The findings in this study support the hypothesis that MDCF based APPMs lead to better production and release plans compared to the ones based on single dimensional aggregated CFs and to the models enhanced with IA.Item Evolutionary approaches to many-objective combinatorial optimization problems(Thesis (Ph.D.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2020., 2020.) Şahinkoç, Hayrullah Mert.; Bilge, Ümit.Many-objective evolutionary approaches try to characterize and overcome the challenges posed by the large number of objectives and have been shown to be very e ective for achieving good Pareto approximations. Despite the growing interest, most of the existing studies work on well-de ned continuous objective functions with designed features, and studies on combinatorial problems are still rare. The proposed many-objective evolutionary algorithm is characterized by elitist nondominated sorting and reference set based sorting where the reference points are mapped onto a xed hyperplane obtained at the beginning of the algorithm by solving single-objective problems. All evolutionary mechanisms such as reference point guided path relinking, repair and local improvement procedures are designed to complement the reference set based sorting. Moreover, the reference set co-evolves simultaneously with the solution set, using both cooperative and competitive interactions to balance diversity and convergence, and adapts to the topology of the Pareto front in a self-adaptive parametric way. The proposed algorithm works successfully both under binary and permutation encoding, as well as for correlated objectives or objectives with di erent scales. Near optimal solutions can be used to construct the hyperplane without any signi cant deterioration in the quality of the Pareto approximation. Moreover, when an optimization problem under scenario-based uncertainty is modeled as a many-objective problem, the proposed algorithm can provide good solutions simultaneously for several robust measures. Numerical experiments demonstrate the success of the proposed algorithm compared to state-of-art approaches and con rm that it can be applied sustainably to a variety of many-objective combinatorial problems.Item Network optimization models with fairness concern for disaster relief operations(Thesis (Ph.D.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2019., 2019.) Erbeyoğlu, Gökalp.; Bilge, Ümit.Humanitarian logistics activities, which aim to alleviate the suffering of the population after a sudden and calamitous event, confront us with a large number of interrelated and challenging network optimization problems. This study focuses on the preparedness and response stages of the disaster life-cycle. Humanitarian network design decisions are of critical importance since they set the frame for all further post-disaster operations. Having an adequate number of strategically located storage and distribution centers for supplies is the key that enables effectiveness, efficiency and fairness when responding to a disaster. The preparedness model proposed in this study aims to find a robust relief network design that ensures the right mix of relief items can be supplied at the right time, and satisfies the demand for all given disaster scenarios. We propose a logic-based Benders decomposition approach to solve this problem to optimality. The numerical studies demonstrate that it is possible to obtain optimal or very good solutions to instances with realistic sizes for this NP hard problem. After disaster occurrence, it is critical to have a relief distribution plan that is efficient and effective while being equitable among the beneficiaries. For this purpose, a rich vehicle routing model along with alternative linear objective functions is presented in the second stage of the study. These objectives provide a balance between timely and fair response plans. Two heuristic methods are presented for the response stage problem. Although solution quality is comparable to the solver in smaller sized problem instances, the heuristic methods provide robust solution methods since they are able to find solutions for larger cases where the solver fails. The assumptions and the parameters used in the models are justified by authorities of humanitarian organizations. The benefit of using these two complementary models consecutively to achieve a better response is also demoItem Resource portfolio problem under resource dedication policies in multi-project environments(Thesis (Ph.D.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2012., 2012.) Beşikci, Umut.; Bilge, Ümit.; Ulusoy, Gündüz.Resource related decisions are one of the important aspects of multi-project environments, since the resource based considerations define the environment as a multi-project problem by coupling projects with corresponding conceptual and physical constraints. The characterization of the way resources are used by the individual projects in a multi-project environment is named as resource management policy in this dissertation. Resource management policies can differ with respect to the environment characteristics (e.g., geographical distribution of projects, specific resource characteristics, etc.). Thus, to identify and characterize different properties and aspects of the multi-project environment, different resource management policies need to be defined. Two different resource management policies are proposed in this dissertation. The first one is the Resource Dedication (RD) policy where resources cannot be shared among projects because of the characteristics of the multi-project environment. The second policy is an extension of RD, such that renewable resource transfers among projects are allowed when one of the projects finishes before the start of the another one. This resource management policy is called the Relaxed Resource Dedication (RRD) policy. These different resource management policies are investigated in a problem environment such that general resource capacities are included into the problem as another decision level. This problem is called Resource Portfolio Problem. The main contributions of this dissertation are the definitions for RPP under different resource management policies, corresponding mathematical models and the proposed solution approaches for multi-project scheduling problems.Item Simultaneous scheduling of machines and the material handling system in a flexible manufacturing system(Thesis (Ph.D.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 1991., 1991.) Bilge, Ümit.; Ulusoy, Gündüz.The scheduling of material handling system is a crital issue in a Flexible Manufacturing System (FMS), although it has little importance in a job shop. The purpose of this dissertation is to exploit the interactions between the machine scheduling and the scheduling of the material handling system in an FMS and to integrate them by addressing them simultaneously. In the FMS under consideration, the material transfer between machines is done by a number of identical Automated Guided Vehicles (AGVs). Upon completing a loaded trip the AGV is designated to its next pick-up station. Therefore, the travel times of the empty trips depend on the ending and the starting points of the successive loaded trips assigned to a vehicle.This concept of sequence-dependent travel times increases the difficulty of the problem. As a first step, the combined machine and material handling system scheduling problem is formulated as a nonlinear mixed integer programming model which turned out to be of intractable size for real-world problems. Then, the problem is decomposed into two subproblems, one having the characteristics of the machine scheduling problem while the other is a vehicle scheduling problem and an iterative solution procedure is developed. At each iteration, a new machine schedule, generated by a heuristic procedure, is investigated for its feasibility to the vehicle scheduling subproblem. To do this, the operation completion times obtained from the, machine schedule are used to construct "time windows" for each material handling trip, and the second subproblem is handled as a "sliding time window" problem. The procedure is numerically tested on a number of example problems.