Proceedings of the First International Conference on Combinatorial and Optimization, ICCAP 2021, December 7-8 2021, Chennai, India

Research Article

Investigation on Joining Divergent Geometric Profiles using 20KHz Ultrasonic Sound Waves

Download58 downloads
  • @INPROCEEDINGS{10.4108/eai.7-12-2021.2314484,
        author={Pradeep Kumar  J and Sambath Kumar  M and Nishandh  D and Sandeep Kumar  K.B and Sathish Kumar  P},
        title={Investigation on Joining Divergent Geometric Profiles using 20KHz Ultrasonic Sound Waves},
        proceedings={Proceedings of the First International Conference on Combinatorial and Optimization, ICCAP 2021, December 7-8 2021, Chennai, India},
        publisher={EAI},
        proceedings_a={ICCAP},
        year={2021},
        month={12},
        keywords={ultrasonic metal welding response surface methodology artificial neural network},
        doi={10.4108/eai.7-12-2021.2314484}
    }
    
  • Pradeep Kumar J
    Sambath Kumar M
    Nishandh D
    Sandeep Kumar K.B
    Sathish Kumar P
    Year: 2021
    Investigation on Joining Divergent Geometric Profiles using 20KHz Ultrasonic Sound Waves
    ICCAP
    EAI
    DOI: 10.4108/eai.7-12-2021.2314484
Pradeep Kumar J1,*, Sambath Kumar M1, Nishandh D1, Sandeep Kumar K.B1, Sathish Kumar P1
  • 1: PSG College of Technology
*Contact email: jpk.prod@psgtech.ac.in

Abstract

In the modern scenario of solid-state welding, Ultrasonic metal welding (USMW) has emerged as one of the successful and efficient method of joining metal specimens with dissimilar profiles (cylindrical – flat). As the methods and procedures involved in repairing flaws are not cost effective, many industries require a systematic approach to forecast weld strength before manufacturing the weld joints. This study is carried out to develop a mathematical model for predicting the weld strength using response surface method. Experiments are conducted based on response surface design matrix comprising of five factors such as the weld time, amplitude, weld pressure, sheet thickness and wire diameter and the weld strength of each experimental trials evaluated in terms of T-peel load are measured. Also, a feed forward back propagation artificial neural network with supervised training has been developed to predict the T-peel load and it tends to be consistent throughout the entire range values.