Design of hybrid cellular manufacturing systems in erratic demand conditions using axiomatic design

Düzensiz talep koşullarında melez üretim sistemlerinin aksiyomlarla tasarımı

Abstract

Today, product demands have become erratic due to increase in product variety, and competitiveness among companies. Manufacturing systems must be redesigned in order to adapt to these changes, and respond quickly to erratic demand. Lead time of the manufacturing system should be decreased in order to achieve these goals. Therefore, one-piece flow is employed during design, at this study. Moreover, as the companies produce wide range of products, these products should be in different stages of their life cycle. For instance, products that are at their birth phase must be produced in functional layout. Sample parts or prototypes are examples to these part types. However, parts with relatively stable design and high demand must be produced in cells. This results in coexistence of both cells and functional layout at the same factory. These systems are called hybrid cellular manufacturing systems (HMS). These systems are met frequently in real world. However, solution approaches that focus on erratic demand conditions and HMS design are very limited. For these reasons, a road map for the design of HMS in erratic demand conditions using Axiomatic Design (AD) was proposed at this study. Exceptional operations causing inter-cell traffic should be eliminated, so as to implement one-piece flow. For this purpose, a one-piece flow algorithm where alternative machines are utilized is developed at this study. As technology improved, machines have become able to carry out operations of other machines, and this caused them become alternative for each other. At our one-piece flow algorithm, alternative machines are evaluated, and exceptional operations are tried to be eliminated by reassigning them to alternatives systematically. This algorithm is one of the unique components of our study, because it utilizes alternative machines both in cells and in functional layout in order to eliminate exceptional operations. After the elimination of exceptions, unidirectional flow must be established in each cell. This enhances one-piece flow in cells, and decreases production lead time of the parts. Simulation technique is used so as to check sufficiency of system capacity, and bottlenecks are identified on the basis of simulation results. Bottleneck identification is also a rule-based approach using AD principles. Afterwards, bottlenecks must be eliminated. For this purpose, a systematic procedure is developed and pursued. In order to eliminate bottlenecks, some sequential solutions are proposed using AD. Solutions must be implemented starting from the simplest to the hardest. The hardest solution is purchasing machines. This approach was applied on a small hypothetical example, to reflect most of the possible situations. First, part-machines matrix of the problem was clustered using Average Linkage Method, and some exceptional operations were found. The improved machine- part matrix was established, including alternative machines for each operation. Exceptional operations were tried to be eliminated by reassigning them to alternatives based on the rules. By pursuing these rules, all exceptional operations were eliminated in the problem. Afterwards, unidirectional flow in each cell was achieved by using the heuristic of Aneke and Carrie (1986). The resulting system consists of cells with one-piece flow and functional layout with batch flow, without exceptions. The sufficiency of the resources was checked by using the simulation technique. Bottlenecks were identified by evaluating the lead times and queue lengths. Addition of new machines were also tried to remove the bottlenecks. Moreover, the systems before and after the implementation of the proposed algorithm were modeled, and were compared. It is clear that the proposed algorithm decreased the production lead time of the parts produced. Finally, the model explained above was applied at the factory of a Turkish automotive-parts manufacturer. The system was designed as a HMS where partial one-piece flow is achieved through the proposed model and the algorithm. The lead time performance of the proposed design where one-piece flow is applied within the cells was compared with that of the former design with batch flow, using simulation models. Based on simulation results, the proposed design had superior performance. Moreover, existence of bottleneck machines was analyzed using the simulation model. This analysis showed that machines were not bottleneck resources.

Özet

Çalışmamızda kısmî tek parça akışının uygulandığı melez üretim sistemlerinin tasarımı için Aksiyomlarla Tasarım (AD) yöntemini kullanan bir yol haritası önerilmiştir. Tek parça akışının uygulanabilmesi için parçaların hücreler arası hareketlerine neden olan istisnai operasyonlar yok edilmelidir. Çalışmamızda alternatif makineler incelenerek, geliştirilen kurallar çerçevesinde istisnai operasyonlar sistematik biçimde alternatiflere atanmaya çalışılır. Tek parça akışına yönelik tasarım algoritması çalışmanın özgün yönlerinden biridir. Çünkü istisnai operasyonları yok etmek için alternatif makineleri kullanan, fonksiyonel alandan da yararlanmaya olanak tanıyan sistematik bir yaklaşım önerilmiştir. İstisnai operasyonların yok edilmesinin ardından hücre içerisinde tek yönlü akışın sağlanması hücrelerde tek parça akışını mümkün hale getirir. Bu aşamadan sonra benzetim tekniğinden yararlanarak darboğaz kaynakları belirleme ve yok etme prosedürü uygulanır. Bu prosedür de AD’yi temel alan sistematik bir yaklaşımdır. Darboğazlar belirlenirken hücredeki hangi iş merkezinin veya iş merkezlerinin darboğaz olduğu kurallar yardımıyla belirlenir. Daha sonra iş merkezinde hangi kaynağın buna neden olduğu sistematik biçimde araştırılır. Belirlenen darboğazların yok edilmesi için de AD kullanılarak bir dizi çözüm önerilmiştir. Son olarak sistemin yerleşim planı, melez yerleşime uygun biçimde hazırlanır. Önerilen yol haritası düşünsel bir örnek üzerinde uygulanmıştır. Ayrıca istisnai operasyonları içeren sistem ile TPA algoritması uygulandıktan sonra elde edilen sistem Arena 10.0 benzetim yazılımı ile ayrı ayrı modellenmiştir. Geliştirilen algoritmanın, düşünsel örnekteki sistemin parça imalat temin süresini çarpıcı biçimde azalttığı gösterilmiştir.