FBE- Endüstri Mühendisliği Lisansüstü Programı - Doktora

Bu koleksiyon için kalıcı URI

http://hdl.handle.net/11527/32

Gözat

Yazar "Baş, Esra" ile FBE- Endüstri Mühendisliği Lisansüstü Programı - Doktora'a göz atma
  • Öge
    An optimization model to control the flow of relief commodities in humanitarian supply chain under uncertainty
    (Lisansüstü Eğitim Enstitüsü, 2021) İsmail, İsraa ; Baş, Esra ; 692574 ; Endüstri Mühendisliği
    In emergency situations, disaster relief organizations are faced with the difficult decision of how to allocate scarce resources in an efficient manner in order to provide the best possible relief action. It is quite common to have uncertain, incomplete, or ambiguous information about the demand, supply, or performance measure estimates as a result of inaccurate predictions of disasters' consequences and, in many cases, lack of access to information regarding the available supply and relief interventions in the affected areas. In addition, roads destructions, traffic incidents, weather conditions, and strict security procedures following the occurrence of disasters or conflicts may impede relief distribution and delay delivery times. This thesis aims to provide an analytical model that helps relief organizations in reducing human suffering following a disaster while maintaining an acceptable level of cost efficiency. A mathematical model is introduced to optimize the relief distribution problem which considers the social cost —the total sum of logistics and deprivation costs. The model is a multi-objective, multi-modal, multi-commodity, and multi-period optimization model for distributing and allocating relief commodities from supply sources to demand nodes in affected areas while considering the vehicle dispatching decisions. A new way to present the unmet demand and capture the corresponding deprivation time is introduced and the problem formulation is adapted accordingly. The model is solved using the rolling horizon method in a sequence of iterations. In each iteration, part of the planning horizon is modeled in detail and the rest of the time horizon is represented in an aggregated manner. This improves the computational efficiency and helps to reach a satisfactory solution for large scale problems in a reasonable time. In addition, the rolling horizon approach allows to update the problem parameters with actual data once revealed during the planning horizon which improves the decision making in such a dynamic environment. The thesis proceeds with extending the basic deterministic model by incorporating uncertainty originated from the inherited natural variation and the linguistic subjectivity. First, the fuzzy nature of the deprivation cost function is addressed through possibilistic mixed integer programming with fuzzy objectives to reflect variation in deprivation costs perceptions. The proposed model attempts to partially account for the influence of socio-economic characteristics on the degree of vulnerability to deprivation. It aims to minimize the risk of higher deprivation cost (perceived by vulnerable people), and minimize the most possible value (perceived by average typical individuals) while maximizing the possibility of lower deprivation cost. Next, the inherited uncertainty in model parameters such as supply, demand, and travel time are accounted for in a robust optimization formulation. The reasonable worst case robust optimization approach is adapted from literature and utilized to model uncertainties in supply and demand which are assumed to be uniformly distributed and bounded in a predefined continuous interval. The level of conservatism of the robust approach is adjusted in such a manner that keeps the probability of constraints violation at minimum under any uncertainty realization. Since the model seeks to minimize the deprivation cost, expressed as a function of deprivation time and updates the deprivation status of demand nodes at the beginning of each time period in the multi-period planning horizon, the uncertainty realizations in travel time are discretized and delays are expressed as number of time periods behind the expected delivery time. Therefore, the thesis also introduces a novel quasi robust optimization approach to model the uncertainty in travel time with discrete settings; where delays in travel time at each arc are assumed to be proportional to the arc load assigned in the optimization model solution. The model is coded on Java NetBeans IDE 8.2 platform and solved using Gurobi 8.1 optimization package. To validate and empirically test the performance of the model, problem random instances are generated according to predefined criteria to cover wide range of scenarios regarding the supply resources availability / scarcity with respect to expected demands. In addition to the empirical analysis, the thesis presents a real case study of internal displacement in northwest Syria to practically test the basic and possibilistic formulations. The information used in this study depends on published reports issued by UN acting parties as well as interviews with NGO's and humanitarian agencies. Computational results show interesting features of the proposed model which are highlighted in the following points: (1) Denoting the unmet demand for each deprivation time as a continuous variable improves the solution efficiency compared with models which use binary variables to capture the deprivation time since the latest delivery. (2) The rolling horizon methodology is found to be efficient in solving large scale instances and have a great benefit in capturing the dynamic changes in demand and supply parameters. (3) Considering the demographic structure in affected areas and reflecting it to the deprivation cost function in a fuzzy formulation helps to reach better prioritization of relief distribution and hence to attain a higher level of equity. (4) Formulating the problem as robust and quasi-robust optimization model to tackle uncertain parameters helps decision makers to reach a trade-off between feasibility and optimality under a wide range of possible scenarios. Finally, some limitations of the current research can be reported and pointed out for future research. For example, this study borrows the deprivation cost parameters from literature models and applies them to the current case analysis. It is recommended to pay more effort in measuring and surveying the effect of deprivation to different groups of individuals, in a case such as internally displaced persons in Syrian camps, using econometric models. Another limitation is that the newly introduced arc-load based quasi-robust optimization model still lacks a rigid mathematical foundation to measure the probability of constraints violation under each uncertainty setting. Additional work on the theoretical foundation still needs further investigation.
  • Öge
    Çeşitli Belirsizliklerin Olduğu Sermaye Paylaştırımı Ve Sermaye Bütçeleme Problemlerine Bulanık Ve Dirençli Optimizasyon Yaklaşımları
    (Fen Bilimleri Enstitüsü, 2009-01-09) Baş, Esra ; Kahraman, Cengiz ; Endüstri Mühendisliği ; Industrial Engineering
    Sermaye bütçeleme problemleri bütçe ve borç alma gibi kısıtların dikkate alındığı, belirli sayıdaki yatırım seçeneklerine sermaye paylaştırım oranlarının incelendiği doğrusal ya da doğrusal olmayan problemlerdir. Bu tezde, çeşitli sermaye paylaştırımı ve sermaye bütçeleme problemlerindeki belirsizlik bulanık ve dirençli optimizasyon yaklaşımları dikkate alınarak incelenmiştir. Bulanık matematiksel programlama, bulanık bir modelin belirli bir bulanık bağıntı kullanılarak netleştirilmesini içermektedir. Dirençli optimizasyon yaklaşımı ise, bir problemdeki belirsiz parametrelerin olabilecek en kötü gerçekleşme durumunda bile problemin eniyi çözümü verecek şekilde çözülmesini içerir. Bu tezde, başlangıçta Lorie-Savage sermaye paylaştırımı problemine literatürdeki bulanık matematiksel programlama yaklaşımları incelenmekte ve nümerik analizlerle inceleme ayrıntılandırılmaktadır. Tezin bulanık optimizasyona ayrılan diğer bölümlerinde ise, Weingartner’ın saf sermaye paylaştırımı doğrusal programlama modeli t-norm bulanık bağıntısı ve Bernhard’ın ikinci dereceden programlama modeli t-norm ve t-conorm bulanık bağıntıları dikkate alınarak modellenmekte ve modeller ayrıntılı olarak incelenmektedir. Tezin dirençli optimizasyona ayrılan bölümünde ise, Weingartner’ın saf sermaye paylaştırımı ve planlama ufku modelleri için literatürde önerilen dirençli optimizasyon yaklaşımları, diğer bazı parametrelerin de belirsiz varsayılmasıyla genişletilmiştir.