Tedarik süreçlerinde döngüsel sefer sistemlerinin yol ve kapasite ölçütlerine dayalı modellenmesi ve uygulanması

Abstract

Lojistik maliyetlerinin, işletmenin sahip olduğu toplam maliyetler içerisindeki payı oldukça büyüktür. Küreselleşme ile de artık dünyanın her noktasına teslimat büyük önem kazanmıştır, bu da lojistik faaliyetlerinin optimizasyonunun önemini oldukça arttırmıştır. Bu nedenle son 50 yıldır araç rotalama problemlerine çeşitli yaklaşımlar geliştirilmiştir. Bir veya daha fazla tesisten, belirli müşterilere ürün dağıtılması ya da toplanması işlemlerinin yer aldığı problemler, araç rotalama problemleri (ARP) olarak adlandırılmaktadır. ARP'nde temel amaç bütün maliyetleri en aza indirmek, gidilen yolun en küçüklenmesi ile verimli sonuçlar elde etmektir. Yalın düşünce felsefesinin önemli ayaklarından biri yalın lojistiktir. Doğru ürünü, doğru zamanda, doğru yere ve doğru kalitede teslim edebilmek için iyi tasarlanmış bir sisteme sahip olmak gerkmektedir. Araç rotalama problemlerinde de yalın lojistik sistemelerinin araçlarından olan döngüsel sefer yöntemi bu ihtiyaçlar için uygundur. Bu çalışmada araç rotalama problemlerinin çeşitlerinden ve literatürdeki gelişiminden bahsedilmiştir. Ardından geliştirilen bir matematiksel model ile desteklenen çalışma, mobilya üreticisi probleminde uygulanmıştır. Uygulanan model, optimal sonucu verirken, daha kısa sürede optimal sonuca yakın sonuçlar üreten sezgisel bir algoritma da kullanılmıştır. Gerçek hayatta birçok türde ARP ile karşılaşılmaktadır. Bu çalışmada da araçlar için gidilen yolu en küçükleyen ve diğer kısıtları optimize eden matematiksel model önerilmiştir. Bir mobilya üreticisinin tedarikçileri ile birlikte ele alındığı modelde, kullanılan araç sayıları, kapasiteleri ve harcanan zaman problemin sınırlayıcı kısıtları olarak yer almaktadır. Sonraki aşamada ise mevcut problemi optimale yakın bir değerle çözüme kavuşturan bir sezgisel algoritma kurgulanmıştır. Önerilen bu iki yöntem ile kullanılmayan araç kapasitesinin ve gidilen yolun ortak optimalliği amaçlanılmaktadır. Uygulama 5 tedarikçiye sahip bir tesiste yapılmıştır. Daha sonra 13 ve 31 tedarikçili senaryolar ile desteklenmiştir. 5 tedarikçili veriler problemin modellenmesine yardımcı olurken, problemin genel modellenmesi açısıdndan yetersiz kalmıştır. Bu nedenle problemin çözümünde farklı senaryolar uygulanmıştır. Her bir problem için 2 aşamalı tamsayılı modeller çözüldükten sonra farklı ağırlıklara sahip sezgisel metodlar ile test yapılmıştır. Sezgisel olarak ele alınan 15 ayrı senaryo sonucu değerlendirilmiş ve ardından sezgisel metotlar arasında en uygun olanı seçilmiştir.

Logistic is the management of the flows between initial point and customer. Especially, the impact of globalization, people trade without borders. Therefore, trade areas become larger and optimization of logistic operations become more important. Logistic cost is one of the major cost of production. For these reasons, various approaches have been developed about vehicle routing problem in the last 50 years. These approaches contain providing high quality products with low price and quick response to customer demands. In order to achieve these aims, logistic operations in vehicle routing problems are become more important. Vehicle routing problems have been widely studied owing to the economic importance of developing optimization in logistics systems. In real world there are many applications in routing problems occurs. The classical solution of vehicle routing problem is a set of routes, which starts with initial node and end in the depot. However, majority of real world problems are more complex. Therefore, these problems are NP hard problems that can be exactly solved for small examples of the problems. The letters NP stand for Nondeterministic Polynomial. Even though the heuristic approaches do not guarantee the exact optimality, they yield best results in practice. Lean logistics is the logistics part of lean thinking. The main objective is to deliver and serve right materials to right locations with right time and right quantity. While achieve these targets, working with high efficiency become more important. In order to meet these targets in vehicle routing problems, milk-run systems can be used as a lean logistics technique. Milk-run is a round trip and all about logistics support for the supply chain. This technique aims to reduce transportation cost, traveling distance and time consumption. Vehicle routing problems (VRP) has attracted researches who have wanted to derive an efficient solution method for the logistic problems. These problems are to form a tour and n nodes beginning and ending at the origin (first node), that gives the minimum distance. Dantzig and Ramser consider the first algorithmic approach vehicle routing problems in 1959. The objective of this study is to minimize the total route cost. Then Clarke and Wright improve this metod and consider travel time, distance, amount of vehicle and other costs. The size of VRP, which can be solved optimally, limited. VRP have numerous application in real life industry. These problems can be extended by considering more depots, more vehicles, different capacities and additional route and distance restrictions. Vehicle routing problems are based on distribution, transportation and logistics operations. These problems focus on distributing the products from one or multi plant(s) to determined customers. The main issue of these problems is to minimize the total distance with cost reduction operations to obtain efficient solutions. Therefore, these problems may contain milk-run systems. There are many types of vehicle routing problem. For these problems, there are several mathematical models proposed to minimize the total distance with optimizing the other constraints. In this study, furniture manufacturer and suppliers were taken together in milk run logistic system and a new mathematical approach proposed. The main objectives of this study are to minimize the unused capacity of vehicles and total distance with time constraints. Finding the optimal solution takes too much time in VRP. Therefore, the new heuristic approach was introduced to obtain the closer results to optimal. For both models were aimed to minimize the total distance and total unused capacity. There are several variations and specializations of the VRP. For instance in order to find the optimal routes for fleet of vehicles to visit the pickup and deliver points is about VRP with pickup and delivery or VRP with time windows are considering determined time. These examples may be multiplexed. Occasionally some types of VRP considered together. In this study time constraints have been discussed. Generally, lean implementation starts with lean manufacturing. The manufacturer, which was mentioned in this thesis, make production according to the lean manufacturing principles. This firm completed the lean manufacturing transformation and ready to implement lean logistic systems. Therefore, in this study, lean logistic philosophy was applied truly. Lean logistic principles describe the logistic systems with avoiding the waste resources. The goal is to optimize logistic processes, with optimize the total traveling distance. In this study, first vehicle routing and lean logistics concepts are mentioned. Specifically milk-run systems, vehicle routing problems and cycle shipments are also introduced. The definition of these concepts are provided clearly with their corresponding literature reviews. Then, the proposed models are explained with not only their literature applications, but also implemented to furniture manufacturer. Proposed models are implemented to plant, which has 5 suppliers and these models were supported with 13 and 31 supplier scenarios. The necessary suppliers data are gathered from manufacturer. However, 5 supplier dataset is not enough to explain the model clearly. Therefore, models were supported and implemented with 13 and 31 suppliers' dataset. After solved the 2-stage integer models for each problem, tests were made with heuristic methods with different alternatives. Then the closest heuristic method was chosen to use in VRP. In this study, 15 different scenarios are solved and compared. In real life, nearly all problems have multi objectives. However, sometimes goals may conflict each other and controling of constraints become more difficult. Therefore multi-criteria decision-making techniques are proposed in literature. To solve these type of problems goal-programming models are used instead of linear programming models. Goal-programming models set of goals that are prioritized in some sense. The integer goal-programming model proposed in this study is solved by using Gams software with Cplex solver. Preemptive goal programming model is applied to find the optimal solution. This optimization model has two stage. First stage is about minimizing the total distance. After solving this model by Gams, second stage starts. The outputs of the first model are the inputs of the model of the next stage. Second stage aims to minimize the unused capacity of vehicles. Obtaining the optimum solutions from these models are taking too much time in large problems. Therefore, heuristic approach were proposed to obtain rapid solutions. In this approach, different scenarios were applied. Microsoft Excel program helps to solve these scenarios. Exact solution methods are certain to find an optimum solution, but a worst require an excessive number of calculations. Hence heuristic methods are applied, because these techniques are not always certain to find an optimum solution, but require a small number of calculations with less computational effort. Briefly in this study, theoretical information and applicable contents of this study consisted of the analysis and guideline of lean logistics, milk-run systems and supplier logistics. In the last part of this study, the results of the application are compared and discussed. I hope that, this study will be helpful and useful to the companies about improving their existing distribution systems.