A robust design in hardfacing using a plasma transfer arc

This paper presents the use of the Taguchi-regression method in developing the optimal plasma transferred arc welding (PTAW) process for obtaining high hardfacing quality characteristics. An “optimal” process means that the best performance characteristic would be produced while the least number of process parameters are involved. In the experimental tests, the surface hardening process is conducted using Cobalt-based and Nickel-based powdery metal materials together with L18 orthogonal arrays. The dependent variable, wear, obeys the-smallerthe-better quality characteristic, and the performance statistics, the signal-to-noise ratios (SNRs), are obtainable. The experimental results show that the most efficient process parameters based on analysis of variance are set as follows: hardfacing material, the accelerating voltage, the powder feed rate , and the pre-heat treatment temperature. They account for almost 90% of the total variance of wear. Under the optimal setting, the average error rates for using Taguchi and Taguchi-regression methods are 7.05 and 5.50%, respectively. The outcome of the experiment indicates that predicted values of the optimal setting fit well with the actual data. The reproducibility of the optimal plasma transfer arc hardfacing is obtained from the experimental data of the confirmation run. A reliable analysis based on S.-Y. Tu Department of Mathematics, University of Michigan at Flint, Flint, MI 48502, USA M.-D. Jean ( ) Department of Electrical Engineering, Yung-Ta Institute of Technology & Commerce, Pingtung, Taiwan, 316 Chunshan Rd, Lin-Lo, Ping-Tung, Taiwan 909, R.O.C E-mail: [email protected] Tel.: 886-8-7233733 Fax: 886-8-7236807 J.-T. Wang Department of Mathematics, Computer Science, and Statistics, State University of New York College at Oneonta, USA C.-S. Wu Welding Technology Section, Metal Industries Research & Development Centre, Kaohsiung, Taiwan the results from plasma transfer arc hardfacing conditions can be achieved through the Taguchi-regression method.