Cutting Tool Failure and Surface Finish Analysis in Pulsating MQL-Assisted Hard Turning
- PDF / 9,250,760 Bytes
- 18 Pages / 595.276 x 790.866 pts Page_size
- 42 Downloads / 188 Views
TECHNICAL ARTICLE—PEER-REVIEWED
Cutting Tool Failure and Surface Finish Analysis in Pulsating MQL-Assisted Hard Turning Soumikh Roy . Ramanuj Kumar Amlana Panda
. Ashok Kumar Sahoo .
Submitted: 24 June 2020 / Accepted: 13 July 2020 ASM International 2020
Abstract Current work emphasized the cutting tool failure, surface roughness, surface topology and chip morphology analysis in hard turning of AISI 4340 steel under a novel pulsating MQL cooling/lubricating circumstance. Pulsating MQL is a newer cooling strategy through which lubricant is supplied into cutting region by timecontrolled pulse mode. Al2O3 top-layered coated carbide tool is utilized to furnish the turning task. Tool failure analysis is carried out considering principal flank wear and auxiliary flank wear. Principal flank wear is traced to be lower than the standard limit of 0.2 mm with the maximum surface roughness of 0.99 lm. Abrasion mechanism is the prime cause for the formation of principle flank wear in each test. Auxiliary flank wear majorly influences the dimensional error and surface quality of the turned surface. Cutting speed as well as depth of cut is the significant and impactful term toward principal and auxiliary flank wears. Feed trailed by the depth of cut and cutting speed are the significant terms toward surface roughness. Helical or ribbon pattern of chips is noticed during the test. The feed is the uppermost term which influences the chip reduction coefficient. Further, neural network modeling has been accomplished to simulate the output response data. Two different training algorithms, namely BFGS quasi-Newton (trainbfg) and Levenberg–Marquardt (trainlm), are utilized for the modeling. Three different architectures like 4-4-1, 4-5-1 and 4-8-1 have been considered to simulate the outputs. Levenberg–Marquardt (trainlm) algorithm attributed the smaller absolute percentage mean error (APME) value relative to BFGS quasi-Newton (trainbfg) algorithm. S. Roy R. Kumar (&) A. K. Sahoo A. Panda School of Mechanical Engineering, KIIT University, Bhubaneswar-24, Odisha, India e-mail: [email protected]
Neural network architecture 4-8-1 with trainlm algorithm exhibited the least values of APME and largest R-square for principal and auxiliary flank wears, while architecture 4-4-1 with trainlm algorithm exhibited the least values of APME and largest R-square for surface roughness. Keywords Hard turning Pulsating MQL Principal flank wear Auxiliary flank wear Surface roughness Neural network modeling
Introduction Nowadays, every manufacturing sector wants to achieve maximum profits and product quality with a minimum cost of production. Achieving these goals requires continuous improvement and research in the field of machining techniques. Many researchers are continuously trying to make improvements in the different machining techniques to maximize its machining efficiency. As per the current trend in manufacturing sectors, the hard turning practice is considered as the utmost preferred machining process over cylindrica
Data Loading...
