Indirect porosity detection and root-cause identification in WAAM

Joselito Yam II Alcaraz (), Wout Foqué (), Abhay Sharma () and Tegoeh Tjahjowidodo ()
Additional contact information
Joselito Yam II Alcaraz: Department of Mechanical Engineering
Wout Foqué: Department of Mechanical Engineering
Abhay Sharma: Department of Materials Engineering
Tegoeh Tjahjowidodo: Department of Mechanical Engineering

Journal of Intelligent Manufacturing, 2024, vol. 35, issue 4, No 10, 1607-1628

Abstract: Abstract Due to the complexity of the Wire-arc Additive Manufacturing (WAAM) process, it is prone to the occurrence of defects in the product. One of the most common defects is porosity, which is detrimental to the mechanical and fatigue properties of the product. To guarantee the product quality, it is important to detect its occurrence. In this research, a strategy was developed to detect the occurrence of porosity and to identify its root cause. To develop a porosity detection strategy, a monitoring set-up was designed in which the WAAM process is monitored by a current sensor, condenser microphone, structural Acoustic Emission sensor, spectrometer and flow meter. Seventeen experiment samples were produced and the sensor data were stored using a data acquisition system. A sensor data analysis showed that current and acoustic signals could be correlated with the occurrence of porosity. From these sensor signals, 74 features were extracted. A validation of the samples, based on milling, grinding and image processing, was developed to obtain the pore density values for the corresponding feature set. A long short-term memory neural network was trained through supervised learning to classify feature data into acceptable or non-acceptable pore densities. A test accuracy of 82.52% was achieved. Apart from the porosity detection, a strategy for porosity root cause detection was developed in order to distinguish between process related porosity and contaminant related porosity. An accuracy of 90.67% was achieved, whereby the remaining inaccuracy is interpreted as a physical phenomenon and the results from the model are interpreted actually as more accurate than the prior assumption of the training labels.

Keywords: Quality monitoring; Smart manufacturing; Artificial intelligence; Wire arc additive manufacturing; Long short-term memory network (search for similar items in EconPapers)
Date: 2024
References: View complete reference list from CitEc
Citations: View citations in EconPapers (1)

Downloads: (external link)
http://link.springer.com/10.1007/s10845-023-02128-x Abstract (text/html)
Access to the full text of the articles in this series is restricted.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:joinma:v:35:y:2024:i:4:d:10.1007_s10845-023-02128-x

Ordering information: This journal article can be ordered from
http://www.springer.com/journal/10845

DOI: 10.1007/s10845-023-02128-x

Access Statistics for this article

Journal of Intelligent Manufacturing is currently edited by Andrew Kusiak

More articles in Journal of Intelligent Manufacturing from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().