Search

Thus far in the available literature, capability-based distributed layout (CBDL) design approaches were only developed under certain environments. Indeed, uncertainties embedded in the machine unavailability (or random machine breakdowns), product demands, and process flow data were not considered by the previous studies to achieve a robust CBDL design. However, many real-life facility layout design applications may involve different types of uncertainties simultaneously, like fuzziness and stochasticity. Based on this motivation, for the first time in the literature, this paper introduces a novel robust capability-based distributed layout (R-CBDL) design problem under a mixed fuzzy-stochastic environment. First, a new fuzzy-stochastic optimization model of the R-CBDL design problem is developed by considering the random machine breakdowns and fuzzy demand/process flow data. Then, a hybrid solution approach based on a chance-constrained stochastic programming technique with a well-known interactive fuzzy resolution method is proposed. Thus, the random machine breakdowns and fuzzy part flow rates among different machining capabilities could be easily handled via the proposed approach. Fortunately, the proposed approach can also generate various risky and risk-free robust layout design alternatives under different probabilistic scenarios and uncertainty levels (α-cuts) according to the facility designer’s risk attitude. To show the validity and applicability of the proposed R-CBDL problem and hybrid solution approach, an extensive computational study with comparative analysis is first presented based on an illustrative numerical example under different machine capability overlap cases and probability distributions. Then, the performance of the proposed approach is also tested on a real-life cellular manufacturing system of a company. The computational experiments have shown that the proposed approach can accomplish more efficient robust layout design alternatives with on average 24.5% better total expected layout score when compared to the existing cellular layout of the manufacturing company.

The educational content of the undergraduate departments in all of the universities all overthe world is mainly based on the knowledge, skills, and capabilities that the graduates willrequire in their business life. Many graduates will use these skills or capabilities, which areusually gained from the compulsory and elective courses in the universities to meet therequirements and specifications in their business life. Therefore, the university coursetimetabling which is NP-hard and a class of scheduling problems is of great importance totraining well-equipped individuals in both business and daily life. However, none of theavailable studies in the literature have considered the optimal distribution of thecapabilities/skills over the curriculum that are essentially gained from compulsory andopened elective courses. Based on this motivation, this research first introduces a novelcapability-based course timetabling approach, which provides a wide variety of courses witha maximal capability set to the students during their university education. To do this, severallearning outcomes and course contents are first clustered to obtain basic capabilities that thestudents should acquire. Then, a mixed-integer non-linear programming model with bothhard and soft constraints is developed to provide appropriate distribution of these capabilitiesover the whole curriculum and also to gain the maximal capability for the students in differentclasses. By making use of the proposed capability-based course timetabling approach, inaddition to gaining the maximum number and variety of capabilities, it is also intended toassign the opened courses to the proper time slots by minimizing the distance between timedifferences of the given courses by each lecturer. Moreover, another objective is alsoformulated as an aggregated penalty function to minimize the weighted deviations in all thesoft constraints. To handle these conflicting objectives simultaneously and to producecompromise course timetables for the university department managers, a fuzzy goalprogramming approach with different importance and priorities is applied. Finally, thevalidity and applicability of the proposed capability-based course timetabling approach arealso demonstrated by a real-life case study at DEU Industrial Engineering department.&nbsp

This paper presents a web-based decision support system (DSS) which uses a fuzzy-stochastic mathematical programming based model for strategic and tactical intermodal fleet management. Indeed, several sub-problems such as load planning, fleet sizing and composition, fleet allocation, vehicle inventory control, fleet expansion/reduction and empty vehicle repositioning decisions are incorporated into the proposed DSS. Therefore, it has a modular structure to support these interactive decisions in an integrated manner. In the model component of the proposed DSS, in addition to optimize overall transportation costs, users are able to provide environmentally conscious and customer-oriented freight and fleet plans by minimizing total transit times and CO2emissions. In the data component, an object-relational database management system namely Oracle was utilized. The LINGO 15.0 optimization code of the proposed model is run over C# and object-oriented matching was utilized for connection among Oracle database and C# programs. The web-based user interface is designed by using .Net and C# programs on Microsoft Visual Studio. The proposed system is tested on a real-life application in an international logistics company of Turkey. By making use of such a DSS, effective and efficient fleet and freight plans can be generated under different types of uncertainties and risk-levels.

Bu tez kapsamında, tam belirsizlik koşulları altında tedarik zinciri ağ optimizasyonu ve intermodal ulaştırma sistemlerinde filo planlama problemleri çalışılmıştır. Öncelikle, tüm model parametreleri ve karar değişkenleri tamamen bulanık ulaştırma problemlerinin çözümü için sınırlandırılmış bulanık aritmetik temelli bir dönüştürme yöntemi ve direkt çözüm yaklaşımı geliştirilmiştir. Geliştirilen doğrusal programlama tabanlı dönüştürme yönteminden farklı olarak, söz konusu problemlerin doğrudan çözümü için metasezgisel temelli bulanık çözümleme algoritması da önerilmiştir. Bu çalışmalara ek olarak, tamamen bulanık tersine lojistik ağ tasarım problemlerinin çözümüne yönelik sınırlandırılmış bulanık aritmetik kullanılmadan parametrik bir çözüm yöntemi de geliştirilmiştir. Önerilen bütün çözüm yöntemlerinde, farklı belirsizlik düzeylerinde bulanık etkin çözümler üretilirken, karar vericinin riske karşı duyarlılığı da dikkate alınabilmektedir. Daha sonra, finansal ve toplama risklerinin birlikte yönetildiği, çok amaçlı bir akü kapalı çevrim tedarik zinciri ağ tasarımı için senaryo tabanlı rassal-bulanık optimizasyon modeli geliştirilmiştir. Son olarak, Türkiye'de faaliyet gösteren uluslararası bir lojistik işletmesinin çok modlu yük ve filo planlama problemlerinin çözümü için rassal-bulanık matematiksel programlama tabanlı bir karar destek sistemi geliştirilmiştir. Geliştirilen karar destek sistemi ile birçok karar vericinin farklı seviyelerdeki karar problemlerine entegre bir şekilde çözümler üretilebilmektedir. Bir başka ifadeyle, çok modlu yük planlama, filo büyüklüğü-komposizyonu belirleme, filo atama, filo stok kontrolü, filo genişletme-daraltma ve boş araç yeniden konumlandırma problemleri geliştirilen modelde eş zamanlı olarak ele alınabilmektedir. Gerçekleştirilen detaylı hesaplamalı analizler göstermektedir ki önerilen karar destek sistemi yardımıyla toplam taşımacılık maliyeti, toplam teslimat süresi ve çevresel kriterler gibi çelişkili sürdürülebilirlik hedefleri doğrultusunda etkin ve etkili yük ve filo planları oluşturulabilmektedir. This thesis deals with network optimization in supply chains under fully uncertain environments and fleet planning problems in intermodal transportation networks. Firstly, a novel transformation method and a direct solution approach based on constrained fuzzy arithmetic concept are developed for solving fully fuzzy transportation and transshipment problems where all of the model parameters and decision variables are stated as fuzzy numbers. In contrast to the linear programming based transformation method, a fuzzy decoding procedure is also applied within a metaheuristic algorithm in order to solve the problems directly. Then, a parametric solution method is also proposed to solve fully fuzzy reverse logistics network design problems without using constrained fuzzy arithmetic concept. All of these developed solution methodologies are also able to consider decision maker's attitude toward risk while generating fuzzy efficient solutions under various uncertainty levels. Afterwards, a new scenario based stochastic and possibilistic programming model is formulated for a multi-objective lead-acid battery closed-loop supply chain network design problem by managing both of the financial and collection coverage risks. Finally, a decision support system based on a fuzzy-stochastic mathematical model is developed for a real-life intermodal load and fleet planning problems of an international logistics company in Turkey. The proposed decision support system integrates various levels of decisions associated with several decision makers. In fact, multi-mode load planning, fleet sizing-composition, fleet allocation, fleet inventory control, fleet expansion-reduction and empty vehicle repositioning problems are incorporated within the proposed model. The extensive computational results have shown that effective and efficient load and fleet plans can be provided by making use of the proposed model in terms of the conflicting sustainability objectives such as total transport cost, transit time and environmental issues. 446