Capacity analysis with simulation for continious production lines and buffer allocation

Developing technology enables many of today's companies competites with each other on many subjects such as raw material supply, delivery of final goods to customer because of increasing human population and consuming rate. Higher competitions between companies and devoloping technology standarts on supply chain lead companies to either protecting or increasing their profit margins difficultly. These conditions forces companies to minimize their raw material, final goods and process inventories. Although it is good for companies not to carry inventories, in real world's production area it is required to remove malfunctions such as; different production rates of machines and unexpected breakdowns. For these kind of reasons, inventory level determination is a common study area for both academicians and sector proffessionals. While keeping inventories determination of buffer stock area is a critical and important topic to study. Because the buffer areas which have been specified wrongly increases lead time and causes non-value added area in production floor. As a result of increased lead time more unexpected situations are faced during production process and non-value added area in production floor causes economical losses. The main purpose of this thesis study is allocation of buffer stock for production facilities operating on process type production systems . Another reason to study on this topic is realize the planned production plan in most profitable way. With this study solution alternatives have been offered for the problem of buffer stock allocation in process type production systems. As it is mentioned before the benefit of this study will be decreased lead time and elemination of non-value added production areas which will return to producer companies as economically. Two general methods have been studied to solve these kind of problems. These methods are mathematical modeling and simulation. However some reasons such as high number of product types and breakdown of machines in production systems and the variety of distribution function of repair duration make simulation modeling more useful than mathematical modeling for this study. According to calculations breakdown times and repair durations fit to Weibull, Exponential, Beta, Gamma and Erlang distributions. The benefit of simulation modelling is that analyzing system scenarios faster, more reliable and nearly free. Because of the machines of the production system are not reliable, the machines can be broken down randomly. Hence, another benefit of the simulation modelling is that failures can be foreseen and required precautions can be taken proactively. Despite of other similar studies in literature performance criteria has not been formed of only one criteria and it is decided to create a simulation model with all criterias which affects system performance. After creating this simulation model it is aimed to find the relation between performance criterias and best scenario. ANP method has been used as multi criteria decisin making method on performance criteria to find the best scneario. The proposed method has been tested in a production facility which produces shampoo, to find the accuracy of the simulation model. The factory is located in Gebze/Turkey and export their products to all over the world especially East Europe and Middle Asia. The factory produces 120 different type of shampoo and they are packaged into 180 ml, 200 ml, 350 ml, 400 ml, 650 ml and 700 ml sizes. While applying production plan in facility, buffer stock allocation has been performed by the experiences of expert responsible from production without making any calculation. Because of this stiation many problems have been faced such as inefficent usage of machines and buffer stock availability and production durations which are longer than usual, all of these problems affects production plans in negative way. In order to minimize these kind of problems simulation model of production system has been created. While creating the simulation model of this systems ARENA 10.0 software has been used. While creating simulation model, weekly demand of each products, breakdown of machines, repair durations, the capacities of buffer stock tanks and machines have been provided by data taken from factory. Furthermore, different set-up time which occurs while transforming from producing a product to another product have been added to simulation model. One of the main difference of this study which distinguishes this study from other studies in literature is that the constraint of liquid product can not be collected in buffer stock tanks. In this study buffer stock tanks have been allocated to packaging lines with different numbers under four different scnearios of studied facility constraints and the value of all performance criterias which affect the system have been collected. After that the interaction of performance criterias with each other have been determined by ANP method via surveys performed on engineers who work in this production factory. The cohorency control for the gathered survey have been conducted and after it is prooved that yhe survey can be used with ANP method, this method have been applied to survey. The coefficents of performance criteris which have been determined by simulation modelling is calculated by ANP method. After that these coefficents have been harmonized with simulation results and it is aimed to find how many buffer stocks should be allocated to which packaging lines for weekly production planning. Because of this answer includes more than one performance criterias and the interaction of these performance criterias are also examined, it is optimum solition under current conditions. Finally, some suggestions have been given for the further studies such as experimental design can be extended in a production system which does not have a physcial constraint. As an another suggestion the survey used in ANP method can be applied to more people so the reliablity of pairwise comparision matrix can be increased. However it must be considered that multicriteria decision making with ANP methods is based on expert opinions; for this reason the same results can not be taken from different production facilities.