por: michel Tavares, Rolf Schroeter, Maria Vitoria Sikora

Acessar artigo completo

Ano: 2019

Instituições de ensino: Universidade Federal de Santa Catarina, Universidade Federal de Santa Catarina, Universidade Federal de Santa Catarina

Idioma: Portugues



The reaming process is usually performed after other processes such as drilling and boaring, hence the cost of these previous processes are added to any failure cost in reaming. As a consequence, there is a demand to study reaming physical phenomena such as the relation between the reaming torque and thrust and the process parameters. In the present work, reaming operations in GG25 cast iron were conducted ranging two process parameters: the cutting speed (32; 65; e 97 m/min) and the feed rate (0.12; 0.16; 0.32; e 0.48 mm). The applied tool was an eight straight flute right hand cut carbide reamer with a 18.98 mm diameter. The pre-holes were 18.50 mm wide in diameter and 9 mm in deep. During the experiments, the torque and thrust values were acquired with a piezoelectric platform. Three different models were developed with the results: a Kienzle with constants values from reference tables; a Kienzle model with calculated constant values; and a feed-forward neural network model. The mean absolute percentage error (MAPE) was used as the comparison criteria, as well as the simulation of reaming realized with a cutting speed of 85 mm/min and a feed of 0.24mm. The neural network model presented lower MAPE and better simulation results. The use of table constants for a Kienzle model generated errors greater than 500% for thrust, but with a bias, that allowed the use of a correction factor.

Palavras chave

Acessar artigo completo


Veja outros trabalhos acadêmicos

Outros conteúdos


Tire suas dúvidas e ajude outras pessoas no CIMM:

Faça uma pergunta