35 th International Conference on Computers and Industrial Engineering 1929 CONCURRENT OPTIMIZATION OF MULTI RESPONCE QUALITY CHARACTERISTICS BASED ON TAGUCHI METHOD Ümit Terzi*, Kasım Baynal *Department of Industrial Engineering, University of Kocaeli, Vinsan Campus, Kocaeli , Turkey +90 262 3351148 / 1296 , umitterzi@kou.edu.tr Abstract Taguchi method is one of the most frequently used tools for improving quality. However, most published applications have been concerned with the optimization of a single performance characteristic. In this study, three different methods are used to improve a production prosess with multiple performance characteristics and results are evaluated. Keywords: Taguchi Method; Goal Programming; Fuzzy Logic; Multi-Response Characteristics 1. Introduction Quality is a key factor for sustainable success in global economy. Taguchi method (Taguchi, 1986 and Ross, 1988) is one of the most frequently used tools for improving quality. This experimental design technique acquires data in a plan of experiments, using orthogonal arrays. The orthogonal array produces smaller, less costly experiments that have high rates of reproducibilty (Peace, 1991), which facilitates experimental design to be used in industry. Taguchi focused on information of both the mean and the variability of a quality characteristic using the signal to noise (SN) ratio. Hence, optimal factor/level combination obtained from the Taguchi method can be determined to simultaneously reduce the quality variation and bring the mean value close to the target value (Tong, Wang, Chen, & Chen, 2004) The method has been succesfully used in many areas. However most published applications have been concerned with the optimization of a single performance characteristic. Generally quality of products or services are measured with more than one characteristic. Although importance and types of characteristics vary according to the problem, all of the performance goals must be achieved as much as possible. There are numerous methods for solving multiple objective problems. Selection of the method for handling multi-response characteristics is crucial, because of the variance of characteristics must be minimized. Besides, easiness of the method is also important that it would designate if it will be used commonly. In this study three methods; fuzzy logic, additive methods and goal programming that we propose are used to solve a multi-response problem and the methods are compared. A cable production process with 7 control factors and 3 characteristics was chosen as the implementation area of these methods. Additive method and fuzzy logic are used to obtain different Multiple Indexes (MI) to be able to analyse multiple quality characteristics together. Goal programming method which uses linear regression to formulate objectives is presented as an alternative. Experimental study is used as a tool to present steps of the tecniques and the results are evaluated to compare the effectiveness of the approaches. 2. Experimental Study The aluminium wire production process steps are as seen in Fig. 1. Aluminium billets are heated through four stations (T 1 , T 2 , T 3 , T 4 ) and extruded with a velocity of V. After extrusion the wires are exposed to aging at T o C for H hours. There are 3 important responses of quality. These are Strenght, Percentage of Extension and Conductivity.