Residual stress and micro cracks that occur during additive manufacturing (AM) process can result in irreversible damage and structural failure of the object after its manufactured. The complex nature of some AM methods means that not all NDT techniques are effective in detecting residual stresses. EM-ReSt functions as add on for the existing AM processes, comprising two sets of NDT techniques: Electromagnetic Acoustic Transducers (EMAT) and Eddy Current Transducers (ECT). A crucial (and novel) extension of the proposed EM-ReSt system is the incorporation of machine learning algorithms applied to the inspection sensors data for for estimation of likelihood of the AM technique to introduce anomalies into the printed structures.
The key objectives of this project are the theoretical and experimental investigation of potential and limitations of EMAT and ECT for monitoring of AM manufacturing process; the development of EMAT and ECT systems specifically designed for online and continuous AM structures; integration of the two systems together in AM machines; and Machine Learning and Big Data analysis of monitoring data for the optimisation of AM parameters and potentially the development of preliminary standards in AM.
The proposed EM-ReSt system offers several performance advantages over current AM NDT inspection methods like thermography, X-ray computed tomography (CT scan) or digital radiography. EM-ReSt is fast (ms/measurement and overall scanning time does not exceed a minute), reliable (99% Probability of Detection), non-destructive online monitoring of AM techniques, can achieve 15% reduction of faulty outputs with the use of 10 times more cost-effective monitoring system. In addition, the technology allows for a low profile sensing hardware with potential for EMAT and EC miniaturization.
For EM-ReSt project, Brunel Innovation Centre (BIC) will develop a Machine Learning (ML) algorithm for real-time residual stress estimation from signals obtained from ultrasonic EMAT (and (EC data. For EM-ReSt, BIC is leading the FEA modelling work for EMAT wave propagation and sensitivity to defects and stress concentration. BIC is mainly leading the development and the training of an ML-driven Automated Defect Recognition System that will assess the integrity of the AM part based and provide a decision whether the built component is acceptable or would it require remanufacturing.
- Ether NDE
- Hybrid Manufacturing Technologies
- Brunel University London