Online Monitoring Of Laser-Based Powder Bed Fusion By Acoustic Emission

Abstract

Metal additive manufacturing (AM) has seen great advances in capabilities and the technology has matured to the point where industries, such as aerospace, are readily implementing it for production. The main concern remains qualification and quality control. Laser powder bed fusion (LPBF) is one of the more popular AM technologies which, as the name suggests, uses a laser to melt and solidify a powder in such a way as to create a three-dimensional part. The part is built in a layer-wise fashion, stacking each layer on top of the previous layer. The quality of the part being built is dependent on the quality of the previous layer as it forms the foundation. The advantage of the layer-wise process is that it also makes online monitoring of the building process a viable option. Monitoring the process can allow for very tight control and thus improve the quality or notify the operator that the component has defects and, therefore, is not fit for service. Current commercial online monitoring systems are mostly in the form of some sort of imaging or temperature monitoring system. These have the ability to monitor any defects in the powder delivery and laser scanning (melt pool). The size and shape of single tracks ultimately determine the quality of the parts, as it is the building block of the LPBF process. All the different process parameters interplay with each other and operate within a process window. The powder layer should be carefully controlled because the input energy from the laser is set and any change in material volume/powder thickness will change the resulting track’s shape. This study investigates whether gas-borne acoustic emission (AE) signal can be used for online monitoring during LPBF. The amazing amount of information that can be interpreted through listening has been proved for manufacturing processes such as laser welding and monitoring of components in service, such as electrical generators. The experiments were carried out on a commercial machine, EOS M 280, with Ti6Al4V ELI alloy. The influence of the machine noise, microphone and scanning position is investigated, and the signal filtered accordingly. Defects due to changes in process parameters are shown, and more specifically, laser power, scanning speed and powder layer thickness. The AE is correlated to the resulting single track shape. Each combination of sets of process parameters produces a specific sound. The sound pressure level and frequency of AE signal are clearly correlated to defects in single tracks that are supported by physical cross-sectioning. The information about the characteristics of LPBF and its AE is used to develop two possible methods which can detect a defective layer thickness. The algorithm compares the test signal to signals from optimal parameters and parameters which produce defects. The signals are run through a series of processing steps and the results are then correlated to each other. It is shown that the proposed algorithm can detect a defective layer with high accuracy.