Search

Tez (Doktora) -- İstanbul Teknik Üniversitesi, Fen Bilimleri Enstitüsü, 2014, Thesis (PhD) -- İstanbul Technical University, Institute of Science and Technology, 2014, Giderek artan ürün çeşitliliği ve gelen siparişlerin kısa sürede tedarik edilerek müşteriye iletilmesi yönünde artan eğilim şirketleri etkili ve verimli lojistik operasyonları kurma noktasında iyileştirmeler yapmak zorunda bırakmıştır. Lojistik ağının verimli ve etkili bir şekilde çalışabilmesi depolama sistemlerindeki operasyonlarla direkt ilişkilidir. Depo içi operasyonlar sırasıyla ürün kabul, ürünlerin raflara yerleştirilmesi, herhangi bir sipariş durumunda ürünlerin raflardan toplanması, toplanan ürünlerin siparişlere ayrıştırılması, ayrıştırılan siparişlerin paketlenmesi ve sevkiyatı şeklinde yol izler. Sırasıyla belirtilen depo içi operasyonların aralarındaki etkileşim çok fazladır, birbirlerinin performanslarına direkt etki ederler. Dolayısıyla sistem bir bütün olarak ele alınarak sistem performansını artıracak çalışmalara ihtiyaç duyulmaktadır. Aksi durumda operasyonlar üzerinde ayrı ayrı yapılacak performans iyileştirme çalışmaları lokal olarak bir yarar sağlar fakat global olarak sistem performansını artıracak bir faydası olmaz. Örneğin depo içerisinde sipariş toplama operasyonu performansını çok fazla artırmamız tüm sistem performansının da aynı derecede artırıldığı manasına gelmez. Sipariş ayrıştırma ve paketleme operasyon performansları da aynı şekilde artırılmazsa sipariş toplama operasyonundaki performans artışı sadece darboğaza sebebiyet verir, yani sürecin toplamında bir iyileştirme sağlamaz. Bir zincirin performansı, en zayıf halkanın peformansıyla ölçülebildiği için tüm zayıf halkaların performansını iyileştirmek gerekir. Bu çalışmada temel olarak depolama faaliyetleri içerisinde en fazla öneme sahip sipariş toplama ve sipariş ayrıştırma operasyonlarının bütünleştirilmesi üzerine çalışmalar yapılmıştır. Performans ölçütü olarak bu çalışmada üretilen iş miktarları seçilmiştir. Performans kriterlerinden maliyet kriterini minimum kılmak için sistem içerisindeki tüm operasyon ve tasarım faktörlerini lokal olarak minimum kılıp sistem maliyetlerini global olarak minimum kılmak mümkündür. Fakat diğer performans ölçütü olan üretilen iş kriteri göz önünde bulundurulduğunda, sadece sipariş toplama veya sipariş ayrıştırma işleminde üretilen iş miktarının ayrı ayrı maksimum kılınması toplam üretilen iş miktarını aynı şekilde maksimum kılmaz. Bunlar bir bütün olarak düşünülerek toplam üretilen işin maksimum kılınması gerekmektedir. Çalışmada sipariş toplama ve sipariş ayrıştırma operasyonları üzerine analitik model çalışmaları yapılmıştır. Önerilen analitik modellerin farklı depo tasarımlarında nasıl sonuçlar verdiği irdelenmiştir., In supply chain performance of companies, warehouses play an important role. Increasing trend in product variety and expectation of customers for low delivery time of orders enforces the distribution centers for making investments. As is known, warehouses do not directly make money for companies, so distribution centers operate under constant pressure from management to reduce costs and increase efficiency. This dilemma needs to be solved for increasing customer satisfaction and decreasing the cost values of the process. Increasing trend in product variety and rising tendency in delivering orders in a short time have put pressure on company managers to streamline efficient logistics operations. Efficiently and effectively workings of logistics network is directly related to operations in the warehouses. Basic warehouse operations are respectively like: Receiving of goods, storing of goods to shelves, picking of goods from shelves when ordered, sortation of randomly picked goods to related orders, packaging of the sorted goods and transporting. Integration between the operations are very high, they can directly affect each other’s performance. So, by taking whole system into consideration redevelopment plan must be done. Otherwise making separate improvements on operations may be beneficial locally but there cannot be a global improvement on whole system’s performance. For example, increasing the performance of order picking operation by a huge rate cannot directly increase the whole system’s performance by the same rate. If performance of the order sortation and packaging operation could not be increased by the same ratio, only a bottleneck will be seen between these operations. There will be no benefit of this local improvement for whole system. Most of the workings done in warehouse operations only focused and scoped out local problems. These types of research results are not sufficiently communicated to industry to make a significant impact on practice of warehouse operations. Some of the previous academic studies specified this problem which is noted the lack of integration between warehouse operations problems. Rouwenhorst, et al. (2000), Gu, et al. (2007), Baker and Marco (2009) and Gu, et al. (2010) are only some examples that cited note about this problem. In this study, we mainly work on integration of order picking and sortation operations in warehouse applications. The aim of this research is to provide assistance for a critical decision of optimal wave size in order picking and order sortation operations. Integration of these two operations is very important because order picking and sortation operations are the two warehouse functions that are very effective on the overall warehouse operational performance. And also cost related to the order picking operations account for more than % 50 of the total cost of a warehouse De Koster, vd. (2007). First operation mentioned here is order picking operation. The order picking or order preparation operation is one of basic operations of a warehouse. It consists in retrieving items from the shelves and brings to the front of a warehouse to satisfy customers orders. In distribution centers, long lists of order are put together. Each customer’s order can involve different items and different number of items. In basic order picking procedure, each picker is assigned to only one order list and it is known that the products exist in this list are stored in different locations of the warehouse. Therefore, picker will look up every part of the warehouse in order to complete the list and will scan whether the items in the list exist or not. In this procedure, there are unnecessary transportation costs and utilization of ineffective workers. In literature, there are several different order picking methods named as discrete order picking, zone picking, batch picking or wave picking. The reader is referred to De Koster, et al. (2007) for details of different order picking methods. We used a different order picking method that is a combination of zone, batch and wave picking. In this method, each picker is assigned to a zone and picks all items of orders stored in the assigned zone. The picker picks more than one item at a time. Second operation mentioned in this study is order sortation operation. Picked products of order wave are transferred to induction area by takeaway conveyor and then all of the products of wave are sequentially enter the sortation conveyor system. Before entering into the sortation conveyor system, a barcode scanner reads the items and determines the sequence of the items of different orders. In an automated order sortation system, sorters accommodate mixed orders or batches to its own shipment destinations at a time interval The sorting process involves identifying the item’s destination, tracking the item along its conveyor path and then physically transmitting the carton or packages to the appropriate destination. Determination of this sequence is very important. This sequence information is being transferred to the computer system. Later, the computer will signal the system to divert the item into the appropriate shipment destination for packing. Major parameters for sortation operation are the “length and speed of conveyors”, “the wave size”, “number of the sorting lanes” and the “sorting strategy”. There are mainly three different operation strategy models proposed in sortation systems. First one, named as Fixed Priority Rule (FPR), developed by Bozer at el (1985). In this study number of sortation lanes is equal to number of the orders on the loop conveyor. In fixed priority rules, the orders are prioritized (smallest order first and largest order first) and are assigned to the sortation lanes based on that primacy. Second one developed by Johnson, (1998), developed a dynamic sortation strategy named as Next Available Rule (NAR). In dynamic assignment model, item locations of orders in a wave are considered. Under NAR, each time an order is completely sorted, the next item to pass the bar code scanner of the associated accumulation lane defines the next order to be sorted. Third one on dynamic assignment family is developed by Eldemir and Charles (2006) named as Earliest Completion Rule (ECR). Under this rule, next order is determined by selecting the order that has the closest last box to the accumulation lane. In other words, the order that requires minimum time to sort is selected as the next order to be sorted. In most of the warehouse operation systems order picking and sortation operations are separated from each other. Order picking systems are used to retrieve the products of the orders from the storage area. Randomly picked items are sent to sortation area for being sorted to related orders. In Pick & Sort systems “wave picking” method is being used. In wave picking method, a group of orders is picked simultaneously with each picker being responsible for picking a single group of items for all the orders in a wave. After order picking operation finished, all of the products in order wave are put on the takeaway conveyor. Takeaway conveyor links the warehouse storing area (or order picking operation) to the order accumulation / sortation system. After all of the randomly selected products sorted and decomposed to the related orders, they are packed and sent to the shipping system by shipping conveyor. Efficient warehouse system requires a well-designed and successfully integrated link of these operations. While there are many implementations of integrated order picking and sortation in industry, academic researches on integrated systems are relatively scarce. Most of them are related with just order picking or order sortation operations. The present study has focused on the integrated pick-and-sort systems. We developed two different analytical models for order picking and order sortation operations that reveal the relationship between these operations. The proposed models consider not only operational parameters but also design parameters. Determining the impact of both operation and design parameters is very important for these types of expressions because, in warehouse applications design parameters affects the operational parameters and operational parameters also affects the design parameters. So, proposed solution method should give answer for the restrictions of both design and operation factors. This paper propose an analytical model to determine the throughput rate in a unit time for both order picking and order sortation operations has been proposed. Both operations’ efficiency has been determined as a function of order wave size. The reason for selecting analytical solution method is that, the problems discussed in this paper are at the operational level, which means that decisions need to be made quite frequently and the influence of these decisions is typically of a short duration and localized. Resultant analytical model give us a chance to solve the trade-off between both order picking and sortation operations by considering the restriction of design and operation factors. So proposed model in this study gives effective and reliable results and can be applied to any different warehouses. The trade-off analysis between order picking and sortation operations refers to a specific sorting system and order picking area layout. That is why it is difficult to show the performance difference of proposed model and real life results. We showed the results via an illustrative example. Results of this study can be listed as below: • Analytical model results showed that when order wave size increase the throughput rate of order picking is also increasing but throughput rate of order sortation is decreasing. • Proposed analytical expression gives a chance to calculate the optimum wave size for these two operations. Thus, managers who wish to improve the performance their system can easily determine the bottleneck operation in their system addressing the order picking or order sortation operations. • Operation and design parameters can be determined by the manager. By using the proposed model, manager can easily observe the changes in total throughput of the system by changing different design and operation paramaters., Doktora, PhD

Bu calışmada yapay sinir ağları algoritmaları kullanılarak saatlik, günlük, haftalık ve aylık doğal gaz enerjisi tahmini için uygun modellerin bulunması amaçlanmıştır. Bilindiği üzere doğal gaz tüketiminin doğru tahmin edilmesi hem gaz dağıtıcıları hem de tüketicileri açısından önemlidir. Doğal gaz dağıtıcısı açısından incelendiğinde, doğru tahmin değerleri sistemde olusabilecek hataları azaltır ve gaz dagıtım limitlerinin doğru şekilde programlanabilir hale getirir. Bu sayede gaz sistemleri cok daha gerçekçi ve karlı hale gelir. Ayrıca tüketici açısından bakıldığında doğru tahmin değerleri sistemde oluşabilecek hatalari azaltacağından ve bu sayede dogal gaz kesintisi olmayacağından dolayı iyidir.Bu calışmada öncelikle gaz tüketimine etki eden faktörleri belirlemek icin kapsamlı faktor analizi calışması gerçekleştirilmistir. Bulunan sonuçlar geri beslemeli yapay sinir ağları modellerine uygulanmıştır. Yapay sinir ağlari tahmin yöntemi uygulamalarının kullanılmasının amacı algoritmaların doğrusal olmayan verilerin modellemesinde iyi sonuçlar vermesi ve birden fazla tahminin ayni anda yapılabilmesidir. Yedi değisik yapay sinir ağları algoritması uygulanıp bunların karşılastırılması yapılmıştır. The aim of this study is to find a suitable natural gas energy forecasting model for hourly, daily, weekly and monthly values by using artificial neural networks(ANN). As it is known, accurate forecasting is important for both gas distributors and consumers. On the view point of distributors, with accurate forecasting the number of false alarms would be significantly decreased and tranship limits would be scheduled.By this way gas systems would be more reliable and profitable. Although accurate forecasting values are good for general consumers there will be no disconnect and breakdown etc. In this study wide factor analyzing study is done in order to find the factors that effects the gas consumptions. Founded results were applied to ANN feed forward back propogation algorithm models.The reasons behind choosing ANN are the ability of ANN to forecast future values of more than one variable at the same time and to model the nonlinear relation in the data structure. Seven different algorithm models were used and comperison of their performance were done. 157