Prediction and Analysis of Multi-Response Characteristics on Plasma Arc Cutting of Monel 400™ Alloy Using Mamdani-Fuzzy Logic System and Sensitivity Analysis

• Main Authors: Rajamani Devaraj, Emad Abouel Nasr, Balasubramanian Esakki, Ananthakumar Kasi, Hussein Mohamed
• Format: Article
• Language: English
• Published: MDPI AG 2020-08-01
• Series: Materials
• Subjects: ANOVA; kerf taper; heat affected zone; material removal rate; response surface methodology; intelligent modeling
• Online Access: https://www.mdpi.com/1996-1944/13/16/3558

Nickel-based alloys, especially Monel 400™, is gaining its significance in diverse applications owing to its superior mechanical properties and high corrosion resistance. Machining of these materials is extremely difficult through the traditional manufacturing process because of their affinity to rapid work hardening and deprived thermal conductivity. Owing to these difficulties a well-established disruptive metal cutting process namely plasma arc cutting (PAC) can be widely used to cut the sheet metals with intricate profiles. The present work focuses on an intelligent modeling of the PAC process and investigation on the multi-quality characteristics of PAC parameters using the fuzzy logic approach. The Box-Behnken response surface methodology is incorporated to design and conduct the experiments, and to establish the relationship between PAC parameters such as cutting speed, gas pressure, arc current, and stand-off distance and responses which include the material removal rate (MRR), kerf taper (KT), and heat affected zone (HAZ). The quadratic regression models are developed and their performances are assessed using the analysis of variance (ANOVA). Fuzzy set theory-based models are formulated to predict various responses using the Mamdani approach. Fuzzy logic and regression results are compared with the experimental data. A comparative evaluation predicted an average error of 0.04% for MRR, 0.48% for KT, and 0.46% for HAZ, respectively. The effect of variations in PAC process parameters on selected responses are estimated through performing the sensitivity analysis.