Gemi üretiminde kapasite optimizasyonu

Abstract

Gemi üretimi sırasında farklı disiplinden birçok faaliyet bir arada yürütülmektedir. Bu durum karmaşık bir üretim ortamı meydana getirdiğinden, gemi inşaatı planlaması ve idaresi zor bir sanayii dalıdır. Dolayısıyla ancak gerçekçi bir üretim planlaması ve üretim sürecinin etkin bir şekilde takibi yapılarak üretimin kontrol altında ilerlemesi sağlanabilir.Bloklar profil, plaka ve diğer bileşenlerin bir araya getirilmesi ile oluşturulurlar. Bunlar gemi üretiminde en fazla zaman alan ve üretimi sırasında birçok güçlükle karşılaşılan parçalardan biridir. Blok üretimindeki faaliyetler muhtelif iş merkezlerinde icra edilirler. Gemi yapım endüstrisinde otomasyona dayalı işlerin sınırlı kalması gemi imalatının diğer aşamalarında olduğu gibi blok üretiminde de karşılaşılan en büyük zorluklardandır. Bu sebeple iş gücü büyük oranda insan kaynağına bağlıdır. Dolayısıyla iş gücünün planlanması gemi yapımındaki en ehemmiyetli hususlardan birisidir. Öte yandan üretim planlaması ve kontrolünde yeterince başarılı olamayan özellikle küçük ve orta ölçekli tersaneler üretim süreci boyunca büyük sıkıntılarla karşı karşıya kalmaktadırlar. Bununla birlikte imalat faaliyetlerinin emek yoğun karakteristiği sebebiyle birim performans değerleri sürekli değişiklik göstermektedir.Tezde dört aşamalı bir hiyerarşik kapasite optimizasyonu metodolojisi önerilmiştir. İkinci ve üçüncü aşamada, stokastik araştırmaya dayanan metasezgisel algoritmalar kullanılmıştır. Bu doğrultuda öncelikle blok üretimdeki ana faaliyetler belirlenmiş ve kritik iş merkezleri tespit edilmiştir. Daha sonra bahsedilen iş merkezlerindeki faaliyetler belirli bir planlama devrinde tamamlanması gereken görevler olarak gruplandırılıp iş paketleri tesis edilmiştir. Bunlar temel alınarak metasezgisel optimizasyon algoritmaları ile kapasite ve işlem süreleri hesaplanmıştır. Bunun ardından uzman görüşleri doğrultusunda inşa edilen bir algoritma ile optimizasyondan elde edilen plan son bir işlemden geçirilmiştir. Böylece kısıtlı bilginin mevcut olduğu durumlarda olabildiğince dengelenmiş kapasite ihtiyacı miktarlarının hesaplanması ile karar vericiye yardımcı olacak bilginin üretilmesi sağlanmıştır. Bu çalışmada geçmiş bir tersane projesi analiz edilerek üretilen veriler doğrultusunda sayısal bir örnek verilmiştir. Toplamda 92 adet blok üretileceği düşünülmüştür. Teslim süresi 21 ay ve her ayda 22 iş günü olduğu kabul edilmiştir. Normal mesaide 8, fazla mesaide 3 çalışma saati olduğu varsayılmıştır. Ayrıca normal mesai ile karşılaştırıldığında, fazla mesaide % 35 oranında performans düşüşü olduğu kabul edilmiştir. Blokların çelik konstrüksiyon imalatının gerçekleştirildiği iş merkezlerinde gerekli olan kapasitelerin belirlenebilmesi için bir kapasite tahsis modeli geliştirilmiştir. Bunun yanı sıra muhtelif kapasite tahsis seçeneklerinin bir birleri ile kıyaslanabilmesi için kapasite tahsis performansı olarak isimlendirilen bir indeks tanımlanmıştır. Burada ele alınan altı durumdan dördünde optimizasyondan elde edilen kapasitelerin işlenmesi ile, ikisinde ise optimizasyon yapılmadan hesaplanan ortalama kapasitelerin işlenmesi ile kapasite tahsis performansları hesaplanmıştır. Ayrıca proje planının başında ve sonundaki basamaklı yapıdan dolayı projenin iş merkezlerine tam yüklenmesini daha iyi yansıtacağı düşüncesi ile birinci ve sonuncu iş paketleri kapasite tahsis performansı hesabı dışında tutulup yine aynı altı durum incelenmiştir. Her iki halde de optimizasyondan elde edilen kapasite değerlerinin kullanıldığı durumlar bariz bir şekilde ortalama hesaptan elde edilen durumlardan daha iyi kapasite tahsis performansı değerine sahiptir. Tüm iş paketleri hesaplamaya dâhil edildiğinde durum 3, ilk ve son iş paketleri hariçte tutulduğunda ise durum 2 en iyi kapasite tahsis performansına sahiptir. Birinci ve ikinci halde elde edilen indeks değerleri birbirleri ile kıyaslandığında ise en fazla iyileşmenin durum 2'de gerçekleştiği görülmektedir. Bu bakımdan çözüm modelinde dördüncü aşamada durum 2 esas alınmıştır. Çalışma özellikle gelişmekte olan, üretim planlama ve kontrolünde yeterli düzeye gelemeyen tersanelerdeki kısıtlı bilgi ortamına uygun bir plan hazırlanması problemi üzerine bina edilmiştir. Neticede, böylesi bir üretim ortamındaki gemi inşası için blok üretim planını, çözüm modeli içinde tanımlanan matematiksel modeller doğrultusunda ideale en yakın bir şekilde elde etmek için planlamacıya yardımcı olacak bir sistem ortaya çıkmıştır.

The construction process of ships built in shipyards is known to be long and laborious. A shipbuilding project consists of a variety of activities that require multidisciplinary cooperation. Therefore, during ship production process many activities from different disciplines are carried out together. As this creates a complex production environment, shipbuilding is an industrial branch that is difficult to plan and manage. Therefore, only realistic production planning and effective monitoring of the production process can ensure that production proceeds under control. One of the main stages of production activities is the production of blocks. The blocks are time consuming and difficult components to produce in the shipbuilding process. They are the structures formed by joining the cut metal sheets, profiles and other components. Additionally, during this main stage, ship's hull blocks are built gradually by combining the small parts. Block built activities are carried out at the different stations of shipyards. These work centres can mainly be explained as follows: In the preparation section, the necessary pre-activities for production such as cutting and marking are carried out. The output of this station is called a single part. In the pre-production section, the individual parts are combined to form sub-assemblies and matrix structures. In the panel production section, panels with large dimensions are produced by combining the metal sheets and profiles. The difference between pre-production and panel production is the dimensions of the parts produced in both sections. In the block production section, the parts produced in the previous sections are combined and the blocks are constructed. In line with expert opinions and observations made in the production area preparation, pre-production and block production route is defined as critical route.The fact that automation-based works are limited in the shipbuilding industry is one of the biggest challenges encountered in block production as in other stages of shipbuilding. Since the workforce is highly dependent on human resource, the planning of the workforce is one of the most important issues in shipbuilding. On the other hand, especially small and medium-sized shipyards, which are not successful in production planning and control, face great difficulties during the production process. At the same time, due to the labour intensive characteristic of manufacturing activities, unit performance values are constantly changing. Factors such as the initiation of production with incomplete design and revisions following the amendment requests from costumers lead to the fact that the work to be performed is not fully known in advance. Because of the work amount, revisions and worker performance uncertainty factors to be experienced in the production process, resource requirement cannot be determined clearly.Although working with subcontractors provide workforce flexibility, unforeseen capacity requirement change in the production stations negatively influence the overall workforce planning and productivity. The capacity requirement, which has previously been determined incorrectly, should somehow be compensated throughout the production process so that it does not delay the delivery date of the ship. In this case, those working in other fields are supplied to the related project. Meanwhile, the equipment in the normal work areas must be moved to the new work area. Along with this move, adaptation to the new job will be required. In addition to all these, some problems will arise where they stop working and come from. Therefore, those who are supplied in this way will be more costly than the normal workers. Therefore moving the workforce allocated initially is not desirable.In the most general sense, duration (d) can be expressed by d=L/(P*R). Here L denotes amount of work, P denotes average performance and R denotes amount of resource. In the preceding equation the denominator (P*R) can be called as capacity (Q). Performance is defined as the work done by one resource in a unit time and capacity is defined as the work done per unit time in a station. As seen, capacity is a parameter which includes performance and resource amount. After determining how much capacity a job needs to be allocated to complete it in a given time, the required amount of resource can be determined according to the current performance value determined from the field. A certain amount of work can be completed in a shorter time by increasing the capacity allocated to it. In order to increase the capacity, at least either the performance should be improved or the amount of resource should be increased.It can be said that three basic knowledge is required to prepare a good plan. The first is the determination of the amount of work to fulfil. This can be achieved by completing the design and analysing the product. The second is the determination of production performance in work centres and can be achieved with effective production control. The third is the determination of the capacity to be allocated to the work centres in order to complete the product in the desired lead time. This is achieved by the calculations made by the planner. The problem addressed in this thesis is the determination of the third component. In this study, a sub-work based capacity allocation and planning research with lack of information is conducted. This dissertation is focused on providing a predetermined completion time by minimizing capacity requirement change in contrast to minimizing the makespan of the project. For this purpose, a four-stage hierarchical capacity optimization methodology is proposed. In the second and third stages, metaheuristic algorithms based on stochastic exploration are used. Initially, work packages are determined within the two stages. Depending on the available information, multiple work package configurations can be produced. Therefore, it is much more time consuming to directly search for a good work package configuration. In the first stage a preliminary procedure is applied to simplify the final work package configuration search. At the second stage, the final work package configuration is searched in accordance with the assumptions specified in the problem definition by using the simulated annealing algorithm. At the third stage, the plan is constituted by calculating capacities of critical route stations by metaheuristic optimization. In this context the capacity requirement for each sub-work, and corresponding durations are searched with the particle swarm optimization algorithm. At the last stage, the results obtained from optimization are reconsidered in line with expert preferences in an attempt to bring closer the results to reality. Through this stage, final adjustments are made on the optimization results. An evaluation algorithm was constructed in the direction of expert opinions to put the optimization results into the final form.A numerical example was provided using data generated by analysing a previous project of a shipyard. There are totally 92 blocks used in calculations. The lead time is 21 months. Each month has 22 working days. It is assumed that the normal shift has 8 and over shift has 3 working hours. Compared to the normal shift, the over shift is considered to have a performance reduction of 35%. A capacity allocation model has been developed to determine the required capacities in work centres. In addition, an index called capacity allocation performance is defined so that the various capacity allocation options can be compared with each other. In four of the six cases discussed here, capacity allocation performances are calculated by processing the capacities obtained from optimization. In the other two cases, capacity allocation performances are calculated based on the average capacities calculated without optimization. In addition, the first and the last work packages are excluded from the capacity allocation performance calculation and the same six cases are examined in order to better reflect the full load of the project to the work centres due to the stepped structure at the beginning and end of the project plan. In both states, the case where the capacity values obtained from the optimization are used, obviously have better capacity allocation performance than those obtained from the average calculation. Case 3 has the best capacity allocation performance when all work packages are included in the calculation, and case 2 when the first and last work packages are excluded. When the index values obtained in the first and second state are compared to each other, it is seen that the maximum improvement occurs in case 2. In this respect, the fourth stage of the solution model is based on case 2. The study is built on the problem of preparing a plan suitable for the limited information environment in the shipyards. As a result, a system has emerged to assist the planner achieving the block production plan in such environment to the near optimum solution in line with the mathematical models described in the model.