Prediction of Surface Roughness of SLM Built Parts after Finishing Processes Using an Artificial Neural Network

  1. Soler, Daniel
  2. Telleria, Martín
  3. García-Blanco, M. Belén
  4. Espinosa, Elixabete
  5. Cuesta, Mikel
  6. Arrazola, Pedro José
Journal:
Journal of Manufacturing and Materials Processing

ISSN: 2504-4494

Year of publication: 2022

Volume: 6

Issue: 4

Pages: 82

Type: Article

DOI: 10.3390/JMMP6040082 GOOGLE SCHOLAR lock_openOpen access editor

More publications in: Journal of Manufacturing and Materials Processing

Abstract

A known problem of additive manufactured parts is their poor surface quality, which influences product performance. There are different surface treatments to improve surface quality: blasting is commonly employed to improve mechanical properties and reduce surface roughness, and electropolishing to clean shot peened surfaces and improve the surface roughness. However, the final surface roughness is conditioned by multiple parameters related to these techniques. This paper presents a prediction model of surface roughness (Ra) using an Artificial Neural Network considering two parameters of the SLM manufacturing process and seven blasting and electropolishing processes. This model is proven to be in agreement with 429 experimental results. Moreover, this model is then used to find the optimal conditions to be applied during the blasting and the electropolishing in order to improve the surface roughness by roughly 60%.

Funding information

Funders

  • Basque Government
    • ELKARTEK 2019 KK-2019/00077
  • Government of Spain
    • SURF-ERA, EXP - 00137314 / CER-20191003

Bibliographic References

  • 10.1016/j.techfore.2015.07.024
  • 10.1016/j.addma.2020.101070
  • 10.1016/j.jmatprotec.2008.06.020
  • 10.1016/j.ijmecsci.2017.10.017
  • 10.1016/j.cirp.2015.05.002
  • 10.1016/j.proeng.2018.02.010
  • Campbell, (2018)
  • 10.1016/j.jmatprotec.2012.11.011
  • 10.1016/B978-0-08-100433-3.00004-X
  • 10.1016/j.matdes.2019.108235
  • 10.1016/j.addma.2019.04.011
  • 10.1016/j.addma.2018.11.001
  • 10.1007/s40032-016-0340-z
  • 10.1108/13552541311312166
  • 10.1016/j.ijmachtools.2019.01.001
  • 10.1016/j.apsusc.2019.06.042
  • 10.1016/j.cirpj.2020.05.010
  • 10.1016/S0007-8506(07)63068-6
  • 10.1504/IJAT.2014.065828
  • 10.1007/s00170-005-0197-4
  • 10.1007/s00170-008-1698-8
  • 10.1016/j.rcim.2019.01.004
  • 10.1016/j.eswa.2022.118024
  • 10.1016/j.eswa.2009.07.033
  • 10.1007/s001700300039
  • 10.1007/s00170-005-2538-8
  • 10.1109/MASSP.1987.1165576
  • 10.3390/ma12030340
  • 10.1007/s10845-009-0376-5
  • Beale, (2019)
  • Moré, (1978), pp. 105
  • Goodfellow, (2016)
  • 10.1016/S0169-2070(97)00044-7
  • 10.1137/0728030
  • Chipperfield, (1994)
  • 10.1007/PL00011391
  • 10.5897/JMER2013.0271