Bir buzdolabı işletmesinin montaj hattı dengelemesinde süreç madenciliği yaklaşımının kullanılması
Using process mining approach in the assembly line balancing of a refrigerator plant
- Tez No: 635028
- Danışmanlar: DR. ÖĞR. ÜYESİ CEMİL CEYLAN
- Tez Türü: Yüksek Lisans
- Konular: Endüstri ve Endüstri Mühendisliği, Mühendislik Bilimleri, Industrial and Industrial Engineering, Engineering Sciences
- Anahtar Kelimeler: Belirtilmemiş.
- Yıl: 2020
- Dil: Türkçe
- Üniversite: İstanbul Teknik Üniversitesi
- Enstitü: Fen Bilimleri Enstitüsü
- Ana Bilim Dalı: Endüstri Mühendisliği Ana Bilim Dalı
- Bilim Dalı: Mühendislik Yönetimi Bilim Dalı
- Sayfa Sayısı: 107
Özet
Artan rekabet ortamında endüstriyel işletmeler kaliteden ödün vermeden, maliyet azaltarak, verimlilik artışları sağlayarak, pazar paylarını arttırarak karlılıklarını arttırmayı hedeflemektedirler. Bunun tabiki de bir çok yöntemi bulunmaktadır. En önemli yöntemlerinden biri de özellikle üretim tesislerinin montaj hatlarında dengeleme çalışmalarının yapılmasıdır. Üretim en temel anlamda, toplumun ihtiyaçlarını karşılamak amacıyla ekonomik bir değeri olan herhangi bir ürün veya hizmetin oluşturulması ve topluma kazandırılması, bundan da katma değer sağlanmasının amaçlanmasıdır. Üretimin olduğu bir çok yerde montaj, montaj işleminin yapıldığı bir çok yerde de montaj hatlarının varlığından bahsedebiliriz. Montaj işlemini bir ürünün komponentlerini çeşitli prosesler sonucu bir araya getirerek o ürünün oluşturulması, montaj hattını ise bu ürünün komponentlerinin çeşitli proseslerden geçerek bir araya getirildiği yer şeklinde tanımlayabiliriz. Montaj hattının olduğu her yerde montaj hatlarının dengelenmesi ihtiyacı da endüstriyel şirketlerinin amaçları, hedefleri doğrultusunda zamanla ortaya çıkmaktadır. Montaj hatlarının dengelenmesi ile iş istasyonu sayısının minimize edilmesi ve/veya çevrim süresinin azaltılması amaçlanır. Bu amaç doğrultusunda işletmeler ve ürünler bazında çeşitli iyileştirme faaliyetleri hayata geçirilmektedir. Endüstriyel tesislerin karlılıklarının arttırılmasında uygulayabilecekleri bir diğer önemli yaklaşım ise süreç madenciliğidir. Süreç madenciliği bir çeşit veri analizi yöntemi olup ilgili ham veriyi kullanıp, bir bakıma görselleştirerek sürecin modelini oluşturmayı esas alır. Oluşturulan bu süreç modelini analiz ederek iyileştirmeye açık noktaları, bu iyileştirmelerin nasıl yapılabileceğini ve etki boyutunun ne olduğunu anlayabilmemize olanak sağlar. Bu sayede endüstriyel işletmeler amaçladıkları karlılık seviyelerine ulaşabilirler. Literatürde montaj hatlarının dengelenmesi ile süreç madenciliği ayrı ayrı bir çok kez geçmelerine rağmen bir arada neredeyse ele alınmamaktadırlar. Bu tez çalışmasında bu iki kavram beraber ele alınarak sonuçları değerlendirildikten sonra birbirleri üzerindeki etkileri incelenmiştir. Bu tez çalışmasında öncelikli olarak montaj hatlarında dengeleme ve süreç madenciliği kavramları üzerinde duruldu ve literatürdeki kullanım alanları incelendi ve detaylandırıldı. Arçelik Eskişehir buzdolabı işletmesi üretim hattı 4'te iki ayrı uygulama gerçekleştirildi. İlk olarak süreç madenciliği uygulaması gerçekleştirildi. Süreç madenciliği uygulamasında ProM 6.9 süreç madenciliği aracı kullanılmıştır. ProM aracı ile süreç madenciliğinin üç türünün uygulaması gerçekleştirildi. Süreç keşfi adımında dört farklı madencilik algoritması uygulanarak bu madencilik algoritmalarının süreç modelleri oluşturuldu. Daha sonra süreç uyum kontrolü adımında dört farklı madencilik algoritması için oluşturulan süreç modellerinin kalite kriterleri mukayese edildi, tümevarımsal madencilik algoritması ile oluşturulan süreç modelinin gerçek veriyi en iyi temsil eden model olduğu sonucuna varıldı. Süreç madenciliği uygulamasının son adımı olan süreç iyileştirme adımında ise dar boğaz olan istasyonlar belirlenerek sosyal ağ analizi gerçekleştirildi. Süreç madenciliği uygulamasının ardından montaj hattı dengeleme çalışması yapıldı. Hat dengelemede en uygun yöntem olarak sezgisel hat dengeleme methodlarından en yaygın kullanılanların başında gelen sıralı pozisyon ağırlığı methodu uygulandı. Sıralı pozisyon ağırlığı methodu ile montaj hattı dengelendiğinde mevcut duruma göre aynı çevrim süresi ile toplam iş istasyonu sayısı ve operator sayısı 45'ten 41'e düşerek %9'luk bir iyileşme gerçekleştirildi. Son olarak süreç madenciliği ve montaj hattı dengeleme çalışmalarının sonuçları bir arada değerlendirilerek elde edilen verimliliğin nasıl maksimize edilebileceği ile ilgili yol haritası oluşturuldu.
Özet (Çeviri)
In an increasingly competitive environment, industrial enterprises aim to increase their profitability while reducing costs, increasing productivity, increasing their market shares without compromising on quality. Absolutely there are many methods to achieve this. One of the most important methods is balancing especially on assembly lines of production facilities. Production is basically the aim of providing any product or service with an economic value in order to meet the needs of the society and bringing it to the society and providing added value from it. We can mention that the existence of assembly in many places where production is available and existence of assembly lines in many places where assembly process is performed too. We can define assembly process as one product components were combined as a result of various processes. In addition, we can define assembly line as the place where the components of this product are put together by passing through various processes too. Wherever there is an assembly line, the need to balance the assembly lines emerges over time in line with the goals and objectives of the industrial companies. By balancing assembly lines, it is aimed to minimize the number of workstations and / or reduce cycle time. For this purpose, various improvement activities are carried out on the basis of businesses and products. Another important approach that industrial plants can use to increase their profitability is process mining. Process mining is a kind of data analysis method and it is based on creating the process model of the process by using the relevant raw data and visualizing it in a way. By analyzing this created process model, it allows us to understand the points open to improvement, how these improvements can be made and what the impact size is. In this way, industrial enterprises can reach their desired profitability levels. In the literature, although the assembly line balancing and process mining passes many times separately, they are almost not handled together. In this thesis, after these two concepts are handled together and their results are evaluated, their effects on each other are examined. In this thesis, primarily the concepts of process mining and assembly line balancing (ALB) are discussed and their usage areas in the literature are examined and detailed. In the literature section related to the concept of process mining, we primarily focused on how the concept of process mining is handled in the literature. We mention that process mining methodology and explained one by one the steps that make up the process mining methodology. These steps are respectively planning, extraction, data processing, mining & analysis, evaluation and process improvement & support. Then we mention that three main types of process mining. This three main types of process mining are also referred to as process mining techniques in the literature. These techniques are respectively process discovery, process conformance checking and process enhancement. In process discovery step, we give detail about four process mining algorithms which are used to create process model, these are alpha miner, heuristic mine, inductive miner and fuzzy miner and we gave information about how the process discovery goes in the literature. In process conformance checking step, we give detail about four quality criterion of process model and we mention that how the process conformance checking goes in the literature. And in last process mining techniques, process enhancement step, we also mention that how the process enhancement goes through the literature. Lastly in process mining literature section, we mention that studies on process mining and we shared some examples of the results of the improvement activities carried out as a result of these studies. In the literature section related to the concept of assembly line balancing, first of all in the literature, we focused on how the concept of assembly line balancing is handled on and what their effects might be. Then, we focused on some basic assembly line balancing concepts such as work element, work station, cycle time, total work time, balance delay, line efficiency, smoothness index, precedence diagram and bottleneck. Then we classified the assembly line balancing problems by the purpose, the line configuration, product type and the duration of the processing times. Then, we discussed the methods used in solving assembly line balancing problems one by one, these are heuristics methods, analytical methods and meta-heuristics methods. Lastly in assembly line balancing literature section, we mention that studies on assembly line balancing and we shared some examples of the results of the improvement activities carried out as a result of these studies. After the literature review of the concepts of assembly line balancing and process mining, two separate case studies were carried out on the assembly line 4 of Arcelik Eskisehir refrigerator plant. Firstly, process mining case study was carried out. ProM 6.9 process mining tool is used in process mining study. Three types of process mining were implemented with the ProM tool. In the process discovery step, primarily we created the event log of our raw data, which consists of 450 cases and 14,665 events, then we interpreted the outputs of this event log. Later, four different mining algorithms were applied and process models of these mining algorithms were created and their soundness (correctness) and outputs were interpreted. Since the process models created with the inductive mining algorithm guarantee to create a sound model every time, our process model that we have created with this mining algorithm is a sound model. However, it has been observed that the models we created with alpha and heuristic mining algorithms are not correct models. In fuzzy algorithm, the situation is somewhat different. Since the process models created with this mining algorithm are not in the Petri net, but in the form of a process graphic, the soundness of the process models created with this mining algorithm can not be mentioned. Then in the process conformance checking step, the quality criteria of the process models created for four different mining algorithms were compared, conformance checking analysis were carried out in the ProM tool and it was concluded that the process model created with the inductive mining algorithm was the model that best represented the actual raw data. In the process enhancement step which is the last step of the process mining case study, the work stations with bottlenecks were identified and social network analysis was carried out. After the process mining study, assembly line balancing study was done. The production capacity of the assembly line which we carried out line balancing (LB) study was 600 units / shift and the number of work stations and operators was 45. Ranked positional weight (RPW) method, one of the most suitable and common method in heuristic line balancing methods, was applied. Balance delay, line efficiency and smoothness index values were calculated both for the current situation and for the situation after the assembly line was balanced with the ranked positional weight method. Also the total number of work stations and operators were compared between the current situation and for the situation after the assembly line was balanced with the ranked positional weight method. In the results section, process mining results, line balancing results and process mining and line balancing results are evaluated together respectively. In the evaluation of process mining results, conformance analyzes of the process models created with alpha miner, heuristic miner, inductive miner and fuzzy miner algorithms were performed and trace fitness values were compared. Trace fitness value could not be calculated with fuzzy algorithm. This is because the process model created using this mining algorithm is not in the form of a Petri network, but in the form of a process chart. It is seen that the process model created with the alpha algorithm has a very low trace fitness value of 0,13. The reason for this may be the large discrepancies between the process model created with this mining algorithm and the event log used in the study. This means that the process model created with the alpha algorithm is insufficient to represent our data. It can be seen that trace fitness value of the process model created with the heuristic algorithm has a good value such as 0,84. Considering this value alone, it can be concluded that the process model created by the heuristic algorithm can actually represent our data, there are no large inconsistencies between our event log and process model and our model is appropriate in summary. However, it can be concluded that the process model created with heuristic mining is also missing when the process model created with the inductive algorithm, has higher trace fitness value and when the process model is evaluated in terms of control criteria. The process model created with the inductive algorithm was found to have a very high trace fitness value of 0,90. It can be said that the process model created with the inductive algorithm can represent our data at a very high level, there are no significant inconsistencies between our event log and process model and our model is a suitable model in summary. Therefore, it can be concluded that the work stations numbered 1650, 1590, 1890, 1960, 1680 and 830 are bottleneck when we perform the performance / conformance analysis of the process model we created using the inductive mining algorithm that best represents our data. Average sojourn times of these work stations are 15,90 minutes, 15,34 minutes, 14,76 minutes, 13,75 minutes, 13,44 minutes and 12,57 minutes, respectively. In the evaluation of assembly line balancing results, when the assembly line was balanced with the ranked positional weight method, %9 improvement was achieved by decreasing the total number of work stations and operators from 45 to 41 with the same cycle time according to the current situation. In addition, when the assembly line was balanced with the ranked positional weight heuristic method, the decrease in balance delay from %37,22 to %31,10 with %16 improvement, increase from %62,77 to %68,89 with %10 improvement in line efficiency, the smoothness index decreased from 113.57 to 92.06 with improvement of %19 according to the current situation. Finally, the results of the process mining and assembly line balancing studies were evaluated together and a road map was created on how to maximize the efficiency. In the process mining step, we identified six workstations (830, 1590, 1650, 1680, 1890, 1960) as bottlenecks. In assembly line balancing step, eight different workstations 1360/1390, 1870/1890, 1960/1980, 2240/2250) were combined and reduced to four stations. It is noteworthy that workstations numbered 1890 and 1960 are common for both tools. In addition, these workstations are successors of each other, that is, after the workstation numbered 1890 in the U-type assembly line where our study is made, the refrigerator palette is processed in the workstation numbered 1960. This information appears as a common output of both tools. If priority improvement activities are carried out at workstations numbered 1890 and 1960, there is the potential to increase capacity on the assembly line. In the literature, there are very few resources utilizing the process mining tool in assembly line balancing in production facilities. In the following studies, by evaluating these two studies together, it is possible to benefit from potential improvement opportunities on concepts such as capacity, overall man efficiency (OME) and overall equipment efficiency (OEE) by considering process mining outputs as the input of line balancing results.
Benzer Tezler
- Bir imalat işletmesinde bilgisayar destekli kapasite planlama sisteminin tasarımı
Başlık çevirisi yok
SERVET HASGÜL
- Bilgi toplumuna geçiş sürecinde işletmelerde beyin göçü sorununu gidermede üst yönetimin rolü
The Role of top management solve the brain draiw problem in process of knowledge society in business
DERYA ŞERBETÇİ
- An investigation into the implementation of multi objective optimization in sustainable manufacturing: A case study from household goods industry
Sürdürülebilir üretimde çoklu amaç optimizasyon uygulamasının araştırılması: Ev eşyaları endüstrisi ile ilgili vaka çalışması
GÖZDE GÖK
Yüksek Lisans
İngilizce
2015
Endüstri ve Endüstri MühendisliğiOrta Doğu Teknik ÜniversitesiMakine Mühendisliği Ana Bilim Dalı
PROF. DR. METİN AKKÖK
YRD. DOÇ. DR. HAKKI ÖZGÜR ÜNVER
- İşletmelerde kariyer yönetimi ve Eskişehir Arçelik buzdolabı işletmesinde bir uygulama
Başlık çevirisi yok
ZEYCAN ÇELEBİ
- İşletmelerde zaman yönetimi Eskişehir Arçelik buzdolabı işletmesinde bir uygulama
Başlık çevirisi yok
HÜSNÜ USLU
