Additively manufactured (AM) 316L stainless steel exhibits complex microstructures, residual stresses, and scattering sources, such as grains and pores, that evolve during production and after heat treatments. In this study, we investigate the effect of heat treatments at 725°C, 900°C, and 1100°C on AM 316L using ultrasound measurements. 316L samples were printed using the same laser powder bed fusion (LPBF) parameters and then subjected to different heat treatments. Frequency-dependent attenuation and sound velocity are calculated using ultrasound backscattered signals measured before and after treatment using a 5-MHz transducer. The attenuation spectra and principal component analysis (PCA) are used as an explorative step to differentiate material states before and after heat treatments. The spectra are then modeled using a sum of power laws with cross-validation across 25 measurement points. The model shows a fit with R2 greater than 0.90. The estimated model parameters reveal contributions from multiple scattering regimes, including Rayleigh, transition-region, and absorption-related contributions, which are linked to the microstructure variations present in AM steels. This is validated by light optical microscopy (LOM), X-ray diffraction (XRD), electron backscatter diffraction (EBSD) maps, and grain size distribution measurements, which confirm microstructural evolution and stress relief. This integrated framework demonstrates the capability of using ultrasound for physically interpretable, nondestructive evaluation of heat treatment-induced microstructural evolution in AM steel components.

Additive manufacturing is known for producing complex metal components, particularly with materials like 316L stainless steel. However, ensuring the quality and microstructural consistency of such components remains a challenge, as traditional testing methods are often destructive and time-intensive. Data driven models that are used for non-destructive evaluation are often difficult to interpret. This study explores the use ultrasound measurements combined with a multivariate statistical technique (partial least squares), to estimate the material properties of steel samples and examining the relationships between ultrasound signals at various frequencies and material properties such as porosity, grain size, and hardness. This aims to enhance the interpretability of ultrasound testing for additive manufacturing. Our findings indicate that ultrasound backscatter can be effectively linked to key material properties.

Additive manufacturing (AM) enables the manufacturing of complex and tailored products for an unlimited number of applications such as aerospace, healthcare, etc. The technology has received a lot of attention in lightweight applications where it is associated with new design possibilities but also reduced material costs, material waste, and energy consumption. The use of ultrasound has the potential to become the material characterization method used for AM since it is quick, safe, and scales well with component size. Ultrasound data, coupled with supervised learning techniques, serves as a powerful tool for the non-destructive evaluation of different materials, such as metals.

This research focuses on understanding the additive manufacturing process, the resulting material properties, and the variation captured using ultrasound due to the manufacturing parameters. The case study included in this thesis is the examination of 316L steel cubes manufactured using laser powder bed fusion. This study includes the estimation and prediction of manufacturing parameters using supervised learning, the assessment of the influence of the manufacturing parameters on the variability within samples, and the quantitative quality assessment of the samples based on the material properties that are a result of the changes in manufacturing parameters.

The research is vital for analyzing the homogeneity of microstructures, advancement in online process control, and ensuring the quality of additively manufactured products. This study contributes to valuable insights into the relationship between manufacturing parameters, material properties, and ultrasound signatures. There is a significant variation captured using ultrasound within the samples and between samples that shows the backscattered signal is sensitive to the microstructure that is a result of the manufacturing parameters. Since the material properties change with the change in manufacturing parameters, the quality of a sample can be described by the relation between the material properties and backscattered ultrasound signals.

The thesis is divided into two parts. The first part focuses on the introduction of the study, a summary of the contributions, and future work. The second part contains a collection of papers describing the research in detail.

Additive Manufacturing is used for printing parts with high precision and complex geometries, but achieving consistent material properties and avoiding defects is a challenge. This paper presents the use of ultrasound technology as a non-destructive method to optimize the additive manufacturing process. A factorial design is used to print 18 samples using the key process parameters such as Power, Speed, and Hatch Distance. The ultrasound measurements are carried out using a 7.5 MHz focused transducer to capture within-sample variation. The manufacturing parameters and ultrasound variation metric is converted to a response surface model which is then used to identify optimal manufacturing conditions that can help minimize process induced variation and get a consistent microstructure and achieve consistent mechanical properties.

Metal based additive manufacturing techniques such as laser powder bed fusion can produce parts with complex designs as compared to traditional manufacturing. The quality is affected by defects such as porosity or lack of fusion that can be reduced by online control of manufacturing parameters. The conventional way of testing is time consuming and does not allow the process parameters to be linked to the mechanical properties. In this paper, ultrasound data along with supervised learning is used to estimate the manufacturing parameters of 316L steel samples. The steel samples are manufactured with varying process parameters (speed, hatch distance and power) in two batches that are placed at different locations on the build plate. These samples are examined with ultrasound using a focused transducer. The ultrasound scans are performed in a dense grid in the build and transverse direction, respectively. Part of the ultrasound data are used to train a partial least squares regression algorithm by labelling the data with the corresponding manufacturing parameters (speed, hatch distance and power, and build plate location). The remaining data are used for testing of the resulting model. To assess the uncertainty of the method, a Monte-Carlo simulation approach is adopted, providing a confidence interval for the predicted manufacturing parameters. The analysis is performed in both the build and transverse direction. Since the material is anisotropic, results show that there are differences, but that the manufacturing parameters has an effect of the material microstructure in both directions.

Metal based additive manufacturing techniques such as laser powder bed fusion (LPBF) can produce parts with complex designs as compared to traditional manufacturing. The quality is affected by defects such as porosity or lack of fusion that can be reduced by online control of manufacturing parameters. The conventional way of testing is time consuming and does not allow the process parameters to be linked to the mechanical properties. In this paper, ultrasound data along with supervised learning is used to estimate the manufacturing parameters of 316L steel cubes. Nine cubes with varying manufacturing parameters (speed, hatch distance and power) are examined with ultrasound using focused transducers. The volumetric energy density (VED) is calculated from the process parameters for each cube. The ultrasound scans are performed in a dense grid in the built and transverse direction. The ultrasound data is used in partial least square regression algorithm by labelling the data with speed, hatch distance and power and then by labelling the same data with the VED. These models are computed for both measurement directions and as the samples are anisotropic, we see different behaviours of estimation in each direction. The model is then validated with an unknown set from the same 9 cubes. The manufacturing parameters are estimated and validated with a good accuracy making way for online process control.

Additive manufacturing is used to produce complex and tailored products that cannot be achieved using conventional manufacturing approaches. The products can be made from different materials including polymers, metals, etc. The material is added layer by layer to create a final product. The mechanical properties of the final part depend on the process parameters. To improve the quality of the product these manufacturing parameters need to be optimised and for this purpose machine learning along with ultrasound measurements can be used. In this paper, the manufacturing parameters of 50 mm thick polymer cubes are linked to the ultrasound data using partial least squares regression. Three cubes with varying layer heights are made from PLA and ABS each, and backscattered responses of ultrasound are recorded from these six cubes. The ultrasound data is used in the partial least squares algorithm to estimate the layer height and the filament type. The clusters that are formed using the first few components obtained from the algorithm show that the data points of the six cubes can be distinguished and themanufacturing parameters are estimated with good accuracy.

Estimation of frequency-dependent attenuation and speed of sound using ultrasound is of great importance. The acoustic properties can be used for material characterization and to study the local variations in a solid. As ultrasound is a mechanical wave, it is directly sensitive to changes in the material properties. The layered nature of additively manufactured prod-ucts pose a challenge for the estimation of acoustic properties. The non-parametric approaches using frequency transforms are sensitive to noise. In this paper, a parametric model is used to estimate the phase velocity and attenuation of 3D-printed cubes. The received signal from the cubes is a superposition of the backscattered responses from multiple layers of the printed part. A reference echo from aluminium is used as an input to the linear model and to estimate the received ultrasound response. The estimate of the ultrasound signal using the linear model is similar to the measured data and it suggests that it can be used to estimate wave propagation in additively manufactured products. The estimated acoustic properties show an increasing trend with the frequency and dispersion can be seen due to the layered nature of the material.