Influence and Potential of Additive Manufactured Reference Geometries for Ultrasonic Testing

Abstract

This study researches and discusses the impact of different manufacturing-induced effects of additive manufacturing (AM), such as anisotropy on sound propagation and attenuation, on the production of test specimens for ultrasonic testing (UT). It was shown that a linear, alternating hatching pattern led to strong anisotropy in sound velocity and attenuation, with a deviation in sound velocity and gain of over 840 m/s and 9 dB, depending on the measuring direction. Furthermore, it was demonstrated that the build direction exhibits distinct acoustic properties. The influence of surface roughness on both the reflector and coupling surfaces was analyzed. It was demonstrated that post-processing of the reflector surface is not necessary, as varying roughness levels did not significantly change the signal amplitude. However, for high frequencies, pre-treatment of the coupling surface can improve sound transmission up to 6 dB at 20 MHz. Finally, the reflection properties of flat bottom holes (FBH) in reference blocks produced by AM and electrical discharge machining (EDM) were compared. The equivalent reflector size (ERS) of the FBH, which refers to the size of an idealized defect with the same ultrasonic reflection behavior as the measured defect, was determined using the distance gain size (DGS) method—a method that uses the relationship between reflector size, scanning depth, and echo amplitude to evaluate defects. The findings suggest that printed FBHs achieve an improved match between the ERS and the actual manufactured reflector size with a deviation of less than 13%, thereby demonstrating the potential for producing standardized test blocks through additive manufacturing.

1. Introduction

Additive manufacturing has developed rapidly in recent years as it offers unique and unparalleled opportunities for the production of complex geometries. Laser-Powder Bed Fusion (LPBF) produces parts by precisely remelting powdered raw material and manufacturing a part layer by layer [1]. The flexibility in geometry design provided by this technology can also be used in non-destructive testing (NDT), where the use of reference test blocks with known flaws is necessary for the calibration and validation of results. Ultrasonic testing (UT) is a semiquantitative, non-destructive testing method used to detect material defects within the test subject volume based on the refraction and reflection of elastic waves. To qualitatively determine the size of the detected flaw, the amplitude of the recorded signals is compared to the signal amplitude of a known reflector. In practice, the use of theoretically ideal flat bottom holes (FBH) as reference reflectors for UT is common. Currently, reference test blocks with FBH as circular reflectors are produced by using subtractive methods such as drilling or electrical discharge machining (EDM). These methods require a drilling channel, which, in general, is easier to detect than the actual disc reflector, thereby making this method susceptible to human bias. Moreover, the accuracy of the flat bottom hole heavily depends on the chosen manufacturing process. If EDM is used, the initially flat electrode rounds off during the process, resulting in a rounded reflector surface. Consequently, the reflector produced this way shows reduced reflection properties compared to an ideal flat-bottom hole [2]. The LBPF manufacturing process enables the production of complex geometries with internal structures, which is not possible in a single production step, as previously discussed. Regarding UT reference blocks, LBPF would enable the production of internal test defects in various shapes with high accuracy. It has been shown that AM applications can outperform conventional manufacturing methods in terms of mechanical properties [3,4]. However, the manufacturing process affects the microstructure, surface roughness, and micro porosity of the test specimens, which may affect the UT results. Since the LBPF process is characterized by a laser moving along a pre-defined path over the powder bed and locally melting it, this results in highly anisotropic solidification due to the high cooling rates and the associated high thermal gradients [5]. The 316L (1.4404) material typically used for additive manufacturing has a face-centered cubic crystal structure, which is responsible for grains aligning their growth direction with the heat flow [6]. This anisotropy is influenced by process parameters such as laser power, scanning speed, hatching distances, and scanning strategies [5]. The large grains produced during the manufacturing process influence ultrasonic propagation and lead to increased attenuation [7]. Therefore, the goal is to quantify the effects of anisotropy and grain growth on ultrasonic propagation. In addition, additively manufactured workpieces might possess rough surfaces. This may lead to issues during ultrasonic testing, as an increase in surface roughness results in a decrease in the signal amplitude of an echo [8,9,10,11]. Therefore, the influence of different surface roughnesses on sound attenuation was investigated. The evaluation of discontinuities in ultrasonic testing is based on equivalent defect sizes. Using the reference block method to evaluate discontinuities offers the advantage that all test-relevant influences of the material are already included in the DAC curve. At the same time, new ultrasonic techniques can be verified using an existing reference block. However, the production of defined reflectors presents a difficulty. The insertion of circular disk reflectors by means of EDM leads not only to an unwanted drilling channel but also to a rounded surface, which has reduced reflectivity compared to an ideal circular disk. The LBPF process offers the possibility of producing complex geometries with internal structures. This allows internal test defects of various shapes to be produced. However, the influence of the manufacturing process of the reference reflectors on reflectivity needs to be investigated.

A research group led by Mihaljević [12] investigated the production of inclined, internal flat-bottom holes. Their findings revealed that the determined equivalent reflector size (ERS) using the distance gain size (DGS) method could decrease to 30% of the nominal value, even when the angle deviated by only half a degree. The reasons for these deviations were not identified within the paper. Nemitz et al. [13] explored the possibility of calibrating the TFM method using an additively manufactured calibration reflector, while Weidig and Straube [14] additively produced a test body for training purposes in NDT. Wang et al. [15] performed phased array ultrasonic testing of micro-defects in additively manufactured titanium blocks. Although there are studies focused on determining the elastic constants of additively manufactured samples using ultrasound such as the shear modulus or the Poisson’s ratio [16] and the relationship between porosity and sound velocity [17], no sources were found that address the relationship between the orientation of laser paths in the manufacturing process and acoustic properties.

Therefore, three overlapping and synergistically complementary questions arise:

• Can additive manufacturing be used to manufacture test bodies overcoming current limitations posed by conventional processes while still meeting required specifications?
• Can artificial defects be found in additively manufactured test bodies? And how does the quality of additively manufactured reflectors compare with reflectors manufactured using metal cutting processes?
• How does the layered structure and the rough surface resulting from the manufacturing process influence ultrasonic testing?

This study includes systematic investigations of the influences on UT caused by the manufacturing process. It covers anisotropy effects, surface roughness, and flaw size accuracies.

2. Materials and Methods

2.1. Investigation of the Influence of Laser Path Orientation on Acoustic Properties

2.1.1. Sample Preparation

Cubes with side lengths of approximately 10 mm, 20 mm, 30 mm, and 50 mm were manufactured from 316L (1.4404) powder by LPBF using an Aconity Mini LPBF machine. For the 30 mm and 50 mm cubes, only one cube of each size was produced due to the possibility of performing multiple measurements per surface. To investigate the influence of anisotropy on the ultrasound, a Hatch pattern with large anisotropy was selected. This alternating linear hatching pattern with a 180° rotation between consecutive layers is shown in Figure 1. In this pattern, the laser is always active while moving in the X-direction (scanning direction). Hence, the Y-direction is always perpendicular to the laser direction (vertical direction). The Z-direction describes the build direction. The build plate was preheated to reduce warping. Preliminary tests were carried out to select the parameters, ensuring a high density and a stable process.

The parameters used are listed in

Table 1. Parameters of the samples used to investigate the influence of laser path orientation on acoustic properties.

Scanning Speed (mm/s)	Power (W)	Hatch Distance (µm)	Laser Focus Diameter (µm)	Layer Thickness (µm)	Building Panel Temperature (°C)
700	250	100	120	30	150

.

2.1.2. EBSD Analysis

The microstructure produced by the parallel orientation of individual laser paths was visualized using electron backscatter diffraction (EBSD). EBSD analyses were conducted on the three 10 mm cubes in two different planes.

2.1.3. Sound Velocity and Attenuation

As mentioned above, the microstructural anisotropy resulting from manufacturing can lead to an anisotropy for sound propagation. To investigate this, sound velocities in the different directions of the cubes were determined. For this, ultrasonic waves were sent into the cube and reflected at the back wall of the cube, as shown in Figure 2. The time that it took the waves to travel to the back wall and back to the probe was recorded in the UT device. The sound velocity was calculated by v = s/t.

The sound attenuation was assessed by comparing the amplitudes of the recorded back wall echoes. The first back wall echo is the echo from direct reflection. As ultrasound travels back and forth, a second echo is recorded after double the time. To measure the sound attenuation, the first and second back-wall echoes were recorded, and each echo was set to 80% of the Full-Screen Height (FSH) using the gain control of the UT device. Here, FSH refers to the full vertical display range of an ultrasonic testing device, where 100% represents the total screen height and is used as a standardized, device-independent reference for comparing signal amplitudes. The amount of necessary gain was measured for the ultrasound attenuation for each testing direction.

An Olympus A543S-SM transducer with a frequency of 5 MHz and a 6.35 mm diameter was used for the tests. Measurements were conducted in the X and Y directions from both sides of each cube. In the build direction, measurements were only possible from the top, as the cube had to be detached from the build plate by erosion, resulting in a significantly different surface texture that was not comparable to the other measurements. This resulted in a total of 6 measurements per orientation for the X and Y direction and 3 measurements in build direction for each of the 3 cubes. Additionally, for the 30 mm and 50 mm cubes, the measuring points were randomly distributed over each side.

2.2. Influence of Surface Roughness on Sound Amplitude

2.2.1. Sample Preparation

To investigate the influence of the surface roughness of additively manufactured samples on attenuation, a total of 18 cubes with a side length of 10.0 ± 0.1 mm were produced from 316L (1.4404) powder using a Truprint 3000 machine from Trumpf. In contrast to the investigation discussed above, a linear hatching pattern with a 67° rotation between consecutive layers was used in this case, and the build plate was not preheated. The parameters used were based on specified and proven parameter sets for this machine and this material and result in the most isotropic material behavior possible with the corresponding settings listed in

Table 2. Parameters of the samples used to investigate the influence of surface texture on sound attenuation.

Scanning Speed [mm/s]	Power [W]	Hatch Distance [µm]	Laser Focus Diameter [µm]	Layer Thickness [µm]	Building Panel Temperature [°C]
800	150	100	80	30	0

.

2.2.2. Surface Treatment and Roughness Measurement

To achieve varying surface roughnesses, the samples were ground with sandpaper of different grits (K40, K80, K120, K240, and K400), with one test series left untreated (see Figure 3). For each grit, 3 cubes were prepared by completely removing the original surface. Initially, only one surface was prepared in this manner, and a measurement series was conducted from each side. Later on, the opposite side was also prepared in the same way, and an additional measurement series was performed.

The surface roughness was measured using the tactile method using an ATORN easyROUGHNESS roughness tester equipped with a 5 µm probe tip. The untreated samples were mounted so that the roughness profile was recorded in the build direction of the LPBF process. Since surface roughness measurements took place after the ultrasonic inspections, both ground surfaces could be measured. For the ground samples, the roughness profile was recorded perpendicular to the grooves. On each of the two sample sides, 3 randomly distributed measurements were performed, resulting in a total of six surface roughness values per sample. The measurement class Sc3 was used, which specifies a total measurement length of 4.8 mm and a cutoff wavelength λc of 0.8 mm. The recommended measurement class Sc4 for the expected average roughness Ra of 2–10 µm, according to DIN EN ISO 21920-3 [18], could not be implemented due to the required evaluation length of 12.5 mm.

2.2.3. Sound Attenuation

Sound attenuation was conducted by comparing the gain values for setting the echo to 80% FSH, as described in Section 2.1.3 on Sound Velocity and Attenuation. In addition to the A543S-SM transducer used, analysis using Olympus V208-RM with a frequency of 20 MHz and a diameter of 3.175 mm, along with a delay line, was implemented. However, only two opposing pairs of surfaces were utilized as coupling or reflector surfaces, leading to the following measurement setups:

Variant 1: Coupling of the transducer to the ground surface. The back wall, which serves as the reflector, remains as-printed.

Variant 2: Coupling of the transducer to the as-printed surface. The back wall is ground in this case.

Variant 3: Coupling of the transducer to the ground sample surface. The back wall is also ground in this case.

2.3. Test Bodies with Flat-Bottomed Hole Reflectors (FBH)

2.3.1. Sample Preparation

To investigate the impact of the manufacturing process of flat-bottomed hole reflectors (FBH) on their reflection behavior, four test blocks were produced (Figure 4) using a Truprint 3000 from Trumpf. Each test body measured 160 mm in length and 31 mm in width and height and was made from 316L (1.4404) powder.

Linear hatching with a 67° rotation between consecutive layers was employed, and the build plate was not preheated. The remaining parameters are the same as listed in Table 2.

Half of the test blocks were additively manufactured with flat-bottom holes; in the other half, no additively manufactured flat-bottom holes were printed but were subsequently added by EDM. For this, the surface of these specimens had to be milled after the AM process in order to accurately position the flat-bottom holes. For comparison, the surface of the test blocks with AM flat-bottomed holes was also prepared in the same way. This preparation resulted in an improved surface quality compared to the surface as printed. Additionally, the material removal led to a reduction in width and height of about 1 mm.

The experiments aimed to investigate the influence of the manufacturing process on the reflectivity of FBH with different sizes and for different depths. The two variants of test blocks that were used in this research are listed in

Table 3. Depth and diameter of the FBH for variants 1 and 2.

Variant 1		Variant 2
Diameter FBH (mm)	Depth (mm)	Diameter FBH (mm)	Depth (mm)
0.6	27	1.5	20
1	27	1.5	25
1.5	27	1.5	27
2	27	1.5	28
3	27	1.5	29

. For each variant, one test block with 5 FBH manufactured by AM and EDM, respectively, was produced. In variant 1, the FBHs have different diameters and are located at a constant depth of 27 mm. In variant 2, the FBHs are positioned at varying depths, and each has a diameter of 1.5 mm.

2.3.2. Distance/Gain/Size (DGS) Method

The aim of this investigation was to quantify the reflectivity for the FBH depending on the manufacturing technique. This was achieved by determining an equivalent FBH size. UT testing was performed using a normal probe with 5 MHz. The recorded amplitude depends on the reflectivity and size of the reflector and the distance between flaw and transducer. For this purpose, so-called DGS (Distance/Gain/Size) diagrams are given for each transducer. In a DGS diagram, the depth is plotted logarithmically against a gain. With increasing depth, a circular disk with a known diameter requires a higher gain to achieve the same screen height. This relationship is shown in the diagram using a set of curves. A circular disk with an infinite diameter, which represents the back wall, serves as a reference. The measured amplitude for a flaw at a certain depth may then be compared to the values in the curves. For clarity and due to measurement accuracy, values in the DGS diagram were rounded to millimeters and decibels, respectively.

The first step of the experiment was to set the back wall echo to 80% screen height and use the required gain as calibration gain. After that, the ultrasound was aimed at each FBH subsequently, and the gain needed to set the echo to 80% screen height and the distance to the reflector were measured. The difference between the calibration gain and the reference gain, along with the FBH distance, was plotted in the DGS diagram to determine the equivalent reflector size.

3. Results

3.1. Investigation of the Influence of Laser Path Orientation on Acoustic Properties

The results of the EBSD analysis can be seen in Figure 5. In Figure 5a, the image shows the laser path direction oriented in the image plane, with several paths positioned next to each other along the Y-axis. The second axis represents the build-up direction in which the heat flow is oriented from top to bottom, as indicated by an arrow, and the grain growth is directed upwards against the heat flow, as indicated by a dashed arrow. A strong ⟨001⟩ crystallographic texture with individual laser paths next to each other is observable. Figure 5b shows an EBSD image in which the laser paths are horizontal in the X-direction, and the arrow represents the build-up direction. The heat flow and the resulting grain growth (dashed arrow) can also be identified in this orientation. A strong ⟨101⟩ crystallographic texture can be observed.

The measurements of sound velocities showed a significant dependency on the measurement direction. The highest sound velocity of 5897 ± 58 m/s was measured in the Y-direction, while in the X-direction, the sound velocity was only 5150 ± 63 m/s. In both X- and Y-directions, the sound velocities did not depend on the size of the sample. In the build direction, the measured velocity values ranged from 5539 ± 35 m/s to 5865 ± 45 m/s, with the highest velocity recorded in the 50 mm sample (see Figure 6 and

Table 4. Measured values of sound velocity and gain as function of testing direction with linear, alternating hatching, with rounded values.

	Side Length of the Cubes	10	20	30	50
Sound Velocity (m/s)	X-Direction	5129	5190	5156	5125
	Y-Direction	5898	5888	5832	5971
	Build Direction	5792	5652	5539	5865
Gain (dB)	X-Direction	44	51	57	63
	Y-Direction	38	43	49	52
	Build Direction	38	45	53	55

). However, no dependency between cube size and velocity could be concluded from the data. However, it is suspected that the scattering of sound velocities is bigger in the build direction due to manufacturing inconsistencies, different cooling conditions, and potential defects.

The gain factor to reach the above-mentioned 80% FSH is also dependent on the measurement direction (see Figure 7 and Table 4). In the X-direction, the samples required the highest gain and, therefore, exhibited the greatest sound attenuation. For the 10 mm samples, the needed gain is 44.5 ± 3.1 dB in the X-direction, while in the Y-direction, it was 38.3 ± 0.8 dB, and in the build direction, it was only 37.9 ± 1.4 dB. This difference was dependent on the length of the sound path and increased with longer sound paths, i.e. bigger samples. For the 50 mm samples, the gain in X-direction was 63.4 ± 3.3 dB and 51.8 ± 2.1 dB in the Y-direction. Except for the 10 mm samples, the gain in the build direction was slightly higher than in the Y-direction but significantly lower than in the X-direction.

3.2. Influence of Surface Roughness on Sound Amplitude

Pre-treatment of the surfaces with 40 grit sandpaper resulted in a decrease in the average roughness height (R_A) from 8.24 ± 0.70 µm to 2.16 ± 0.31 µm (see Figure 8). With 80 grit sandpaper, an R_A of 0.91 ± 0.11 was achieved. Finer grits only led to a slight further reduction in the average roughness height, down to 0.30 ± 0.03 µm for 400 grit sandpaper. Due to the small differences in the average roughness heights, the following diagrams are presented in relation to the pre-treatment.

The results for variant 1, with the ground coupling surface, are shown in Figure 9 and Figure 10. Firstly, it could be observed that the needed gain to obtain a signal of 80% FSH was significantly higher for the 20 MHz probe than for the 5 MHz probe. This can be explained by the additional sound path and attenuation in the delay line. In addition, higher frequencies are more susceptible to attenuation. Higher attenuation is expressed by the fact that a higher gain is required to adjust the echo to 80% FSH. The first echo from the measurement with a surface ground with 40 grit sandpaper needed 28.7 ± 0.6 dB gain for the 5 MHz probe and 68.2 ± 0.5 dB for the 20 MHz probe. For the 5 MHz probe, the gain for the first echo slightly decreased to 27.1 ± 0.7 dB at the surface ground with 80 grit sandpaper but did not improve for the finer preparation. For the 20 MHz probe, the maximum gain of 70.2 ± 1.2 dB is needed for the surface ground with 80 grit sandpaper. The echo then slightly decreased, reaching a minimum of 67.5 ± 0.9 dB with 240 grit sandpaper. This resulted in a difference of 2.8 dB between the highest and lowest values for the 20 MHz probe. The second echo also exhibited a qualitatively similar pattern. As the second echo showed the same behavior as the first echo, this indicates that the variations in gain were due to coupling losses. While performing ultrasonic testing, the coupling of transducers was the biggest source of amplitude variation. Pre-treatment of the coupling surface with a finer sandpaper did not lead to a significant reduction in attenuation, and therefore, no significantly lower gain values.

The results for variant 2, with ground reflection surface, are shown in Figure 11 and Figure 12. The 5 MHz probe again showed slight fluctuations in gain. The gain needed for the first echo decreased from 31.9 ± 0.2 dB for the reflection surface ground with 40 grit sandpaper to 30.0 ± 0.5 dB for the 80 grit sandpaper, before returning close to the initial value at 31.5 ± 0.5 dB for the 400 grit sandpaper. The second echo exhibited a qualitatively similar pattern. For the 20 MHz probe, the needed gains range from 72.7 ± 0.4 dB for 40 grit sandpaper to 71.6 ± 0.5 dB for 120 grit sandpaper. Taking into account the scattering, this does not show a clearly recognizable trend, but a rather constant high gain. The second echo also followed a qualitatively similar pattern.

For samples in variant 3, with both surfaces ground, measurements on untreated samples were included for comparison, as this represents a similar treatment of both sides. The results are shown in Figure 13 and Figure 14. These measurements, particularly with the 5 MHz probe, showed high measurement uncertainty. It was evident for both frequencies that surface treatment resulted in a significant reduction in needed amplification. The gain for the 5 MHz probe decreased from 30.9 ± 1.5 dB on an untreated surface to 29.5 dB when ground with 40 grit sandpaper, and for the 20 MHz probe, from 72.3 ± 1.0 dB to 68.6 ± 0.3 dB. It was observed that the effect of surface treatment was considerably greater at 20 MHz compared to 5 MHz. Specifically, treating both surfaces led to a decrease in gain by 2.9 dB at 5 MHz and by 6 dB at 20 MHz.

Figure 15 shows the required gains for the 20 MHz transducer for surfaces prepared with 40 grit sandpaper for all three variants. It can be seen that the values for variants 1 and 3 with the treated coupling surface are quite similar, at 68.6 ± 0.3 dB and 68.2 ± 0.5 dB, respectively, while the gain measured for the untreated coupling surface is higher, at 72.7 ± 0.4 dB. This means attenuation was significantly higher for the untreated coupling surface. Similar behavior could be seen for the other surface preparations. This suggests that the surface quality of the coupling area has a significantly greater impact on the required gain than the reflector surface itself. Consequently, a reflector surface with high surface roughness does not necessarily result in a higher attenuation. This is particularly relevant for practical applications, as real reflector surfaces for internal features cannot be machined.

3.3. Test Bodies with Flat-Bottomed Hole Reflectors (FBH)

Table 5. Measured values for the test block with AM FBHs of variable diameters (variant 1, AM).

Gain (dB) Back Wall		31.9
FBH (mm)	Produced Depth (mm)	Measured Depth (mm)	Gain FBH (dB)	∆dB	Determined FBH (mm)	Deviation (%)
0.6	27	Not resolvable	-	-	-
1	27	26.8	58.3	26	0.9	10
1.5	27	26.8	51.8	20	1.3	13.33
2	27	26.8	46	14	1.9	5
3	27	26.8	38.8	7	2.9	3.33

and Figure 16 show the manufactured sizes and depths in comparison to the measured sizes and depths with ultrasonic testing for the FBH of different sizes manufactured by AM (variant 1). To apply the DGS method, it must be ensured that the tested reflectors are in the far field and not in the near field of the probe used. According to the manufacturer’s specifications, this is well below the measuring range of the test specimen used at 5 mm [19], meaning that the DGS method can be carried out with the test setup used. It can be seen that the determined sizes of the FBH derived from the experiments measured with UT were smaller than the actual manufactured sizes. The deviation for the smaller diameters was bigger than for the bigger sizes. The smallest FBH, with a diameter of 0.6 mm, could not be detected using ultrasound. The measured depths for all FBH match the manufacturing depths.

Table 6. Measured values for the test block with EDM FBHs of variable diameter (variant 1, EDM).

Gain (dB) Back Wall		32.4
FBH (mm)	Produced Depth (mm)	Measured Depth	Gain (dB) FBH	∆dB	Determined FBH (mm)	Deviation (%)
0.6	27	Not resolvable	-	-	-
1	27	26.9	56.5	25	1.0	0
1.5	27	26.9	52.5	21	1.3	13.33
2	27	27.0	47.3	15	1.8	10
3	27	26.9	40.8	9	2.6	13.33

and Figure 17 show the manufactured sizes and depths in comparison to the measured sizes and depths with ultrasonic testing for the FBH of different sizes manufactured by EDM (variant 1). The measured sizes of the FBH were smaller than the manufacturing sizes. The deviations were bigger than for the AM FBH, with the 1 and 1.5 mm hole as an exception. Again, the 0.6 mm FBH could not be detected. The measured depth also matches the manufacturing depth.

In the variant 2 test blocks, the reflector depths were varied, and the FBH diameter was set to 1.5 mm. The results for the AM FBH are shown in

Table 7. Measured values for the test block with AM FBHs of variable depths (variant 2, AM).

Gain (dB) Back Wall		31
FBH (mm)	Produced Depth (mm)	Measured Depth	Gain (dB) FBH	∆dB	Determined FBH (mm)	Deviation (%)
1.5	20	19.8	45.4	14	1.5	0
1.5	25	24.8	50.4	19	1.4	6.67
1.5	27	26.8	52.4	21	1.3	13.33
1.5	28	27.8	Not assessable	-	-
1.5	29	Not resolvable	-	-	-

and Figure 18; the results for the EDM FBH are shown in

Table 8. Measured values for the test block with EDM FBHs of variable depths (variant 2, EDM).

Gain (dB) Back Wall		34
FBH (mm)	Produced Depth (mm)	Measured Depth	Gain (dB) FBH	∆dB	Determined FBH (mm)	Deviation (%)
1.5	20	21.7	63.4	29	0.7	53.33
1.5	25	24.5	50.4	16	1.6	6.67
1.5	27	26.4	51.4	17	1.6	6.67
1.5	28	27.7	Not assessable	-	-
1.5	29	Not resolvable	-	-	-

and Figure 19 The reflectors with a 1 mm depth could not be resolved with UT for both manufacturing techniques, as the back wall echo could not be separated from the FBH echoes. The reflector at a distance of 2 mm from the back wall could be detected, but the amplitude of the signal could not be assessed. In order to be able to assess the echo, it must no longer be influenced by the back wall echo; otherwise, measurement deviations will occur. This is not the case due to the proximity of the reflector to the back wall. In the test block with EDM FBH, the size for the distance of 10 mm from the back wall was accurately determined. The FBHs with distances of 5 mm and 3 mm from the back wall were slightly underestimated in diameter with measurements of 1.4 mm and 1.3 mm, respectively. The AM FBH holes with distances of 5 mm and 3 mm from the back wall were slightly overestimated, with measurements of 1.6 mm each. The signals for the reflector with a distance of 10 mm from the back wall were disturbed, supposedly by manufacturing flaws in the test block. This resulted in significant deviations in both distance and echo height measurements. The FBH with a diameter of 0.7 mm was found at a distance of 22 mm instead of 20 mm.

4. Discussion

4.1. Investigation of the Influence of Laser Path Orientation on Acoustic Properties

Carter et al. [20] found that the grain structure of LPBF components strongly depends on the laser path. Both Carter et al. and Kim et al. [20,21] demonstrated that the choice of an island pattern as a hatch pattern manifests in the grain structure, as was observed in EBSD analyses. In the build direction, elongated, columnar grains were visible. According to Carter et al. [20], these are attributed to the epitaxy between layers and to the tendency of heat flowing away from the melt pool in the build direction. In the literature, the sound velocity for 316L samples produced with rotating hatching with 67° rotation between layers is given as 5786 ± 48 m/s for longitudinal waves as in-plane values [22]. For the samples investigated, a sound velocity of 5897 ± 58 m/s was measured in the Y-direction, which is slightly higher than the literature values. Conversely, the sound velocity in the X-direction was measured at 5150 ± 63 m/s, which is significantly below the reference values. These differences are attributed to the fact that linear hatching results in a highly directional microstructure, which significantly affects ultrasound propagation. This may be further investigated by EBSD analysis but lies outside the scope of this research. The above-cited velocities from literature were obtained from specimens created with a rotating hatching with a 67° rotation between layers. K.D. Koube et al. describe an “anisotropic grain structure with a characteristic "fish-scale" pattern” in their EBSD analyses [22], which is caused by the manufacturing process. It can be expected that anisotropic properties are present within one of these "fish-scales". However, as the ultrasonic wave passes through several of these fish scales with different crystal orientations during a measurement, this results in isotropic acoustic properties for the overall component. This is similar to the microstructure of structural steel, where the individual grains also have anisotropic properties, but the overall component has isotropic properties due to the high variation of grain orientation within the microstructure. In samples with linear alternating hatching, the grains are predominantly oriented in one direction, resulting in highly direction-dependent properties. This indicates that there is a correlation between the influence on the sound velocity and the attenuation within the anisotropic structure of additively manufactured samples due to the large grains. This effect is already known for welds made of austenitic steel [7]. While the X-direction showed the lowest sound velocity and highest attenuation, the Y-direction had the highest sound velocity and, except for the 10 mm cubes, the lowest attenuation. For the 10 mm samples, the attenuation in the build direction was in the same range as in the Y-direction, although the sound velocity was lower. For all other sample sizes, a decrease in sound velocity corresponded to an increase in attenuation.

4.2. Influence of Surface Roughness on Sound Amplitude

The surface roughness influences the coupling properties during ultrasonic testing of metallic components. As the surface roughness increases, the transmitted sound amplitude in the component decreases. This effect is related to the frequency of the selected probe and increases for higher frequencies [8,9]. Due to the leading wedge of the probe used, it is not possible to assess whether the increase in gain is due to the higher frequency or the leading wedge. İşleyici [8] states that an increase in surface roughness leads to a decrease in signal amplitude. However, he did not report a proportional relationship between an increase in surface roughness and a decrease in signal amplitude. Therefore, it is not possible to directly correlate the signal amplitude with the surface roughness. Nagy et al. [10] found that scattering at rough surfaces is a major contributor to signal loss. The observed phenomenon that the transmitted wave experiences less attenuation than the reflected wave contradicts the findings presented in this study. In contrast, Wang et al. [9] demonstrated in their work that the influence of surface roughness on ultrasonic attenuation during testing of back-surface micro-cracks is less significant at the back wall compared to the front surface. This aligns with the observations that surface preparation of the coupling interface reduces the required gain, whereas preparation of the reflection surface shows no measurable effect on gain. Hanks et al. [11] investigated the effect of surface roughness on ultrasonic inspection of electron beam melted components. In their study, multiple defects were embedded in a test part, and it was evaluated whether their detectability depended on surface roughness and testing frequency. Their results indicate that for 2.25 MHz and 5 MHz, detectability improves as surface roughness decreases. No embedded defects could be detected at roughness levels above 40.9 µm, while all defects were detectable below 7.9 µm. However, their measurements using a 10 MHz probe showed no significant influence of surface roughness within the tested range on defect detectability. This contradicts prior findings [8,23] as well as the presented results, which indicate that the influence of surface roughness increases with higher frequencies. The authors did not provide an explanation for this discrepancy. It should be noted, however, that the roughness values used in their study exceeded the range applied in our investigations.

The results suggest that for the ultrasonic inspection of additively manufactured test components, surface quality must be considered and, if necessary, improved in order to ensure sufficient inspection reliability. However, preparation of the reflector surface is not required, as it has no measurable influence on signal amplitude at the low surface roughness levels achieved. This is particularly advantageous, since internal reflectors cannot be reworked after manufacturing.

4.3. Test Bodies with Flat-Bottomed Hole Reflectors (FBH)

Combining all results for different flaw sizes, it was revealed that the deviation for AM FBH is generally lower, at 0.1 mm, than for the EDM FBH. The reflector with a diameter of 3 mm showed a larger deviation and was measured at 2.6 mm using UT. This finding seems to contradict previous studies by Jüngert et al. [2], which suggest that for bigger electrode diameters, the influence of electrode rounding is less significant. According to their findings, manufacturing-related deviations from the ideal FBH decrease proportionally with increasing diameter. However, the manufacturing accuracy and surface quality of the FBH can influence the result. In addition, a slightly inclined FBH can lead to a reflection of the sound beam away from the test head and therefore reduce the sound amplitude. This would lead to a smaller reflector using the DGS method. In order to be able to classify the resulting deviation, a CT scan of the sample is recommended, which could not be carried out in the scope of this work. In addition, only one test specimen could be examined for each reflector type in this study. This means that only a limited sample size is available, and therefore, no overall conclusions can be drawn for the dependence of size on the deviation.

For tests with varying reflector depths, it became evident that the UT probe lacked a sufficient depth resolution to differentiate between the back wall and the FBH at a 1 mm distance from the back wall. Additionally, the probe had a wide sound beam, which causes the echo at a 2 mm distance from the back wall to still be influenced by the signals from the back wall echo. Using a probe with a higher frequency, such as 10 MHz, might improve the results.

Comparing the results of the AM FBH to those of Mihaljević et al. [12] shows a significant improvement in determining the diameter of the FBHs. Mihaljević found that for three additive-manufactured flat-bottom holes with a 2 mm diameter, the measured equivalent reflector sizes ranged from 0.51 mm to 1.60 mm using the DGS method. This results in deviations between 0.4 mm and 1.5 mm. However, Mihaljević used FBHs that were inclined at angles between 45° and 70° and filled with powder. The work did not evaluate the impact of the inclination and the residual powder filling on the results. To measure inclined reflectors, UT using an angle beam probe is required. As shown in Section 3.1, the sound velocity and attenuation may vary for the different directions in the build part and complicate the interpretation. Furthermore, the effect of powder filling inside the reflector is unclear; it might cause some sound energy to be transmitted through the powder. The reflectors shown in this paper do not have powder filling, resulting in a strong reflection due to the strong boundary between metal and air.

5. Conclusions

For many purposes in UT, the use of test blocks with known reflectors is necessary. The production of test blocks may be difficult for conventional manufacturing and limited by geometry. Additive manufacturing of parts offers the opportunity to induce flaws of different sizes and geometries. However, the manufacturing process influences the properties of the material.

It can be shown that the use of a linear, alternating hatch pattern results in a strong anisotropy for sound velocity and attenuation. The anisotropy was strongest in the direction of the laser path. A significant reduction in sound velocities and a significant increase in attenuation were found. This affects the propagation of ultrasonic waves and has to be taken into account when interpreting data from UT at AM parts. To reduce the effects of anisotropy for the following investigations, a rotating hatching with 67 ° between layers was used.

Surface roughness is an issue when AM parts have to be tested as printed. The manufacturing process produces a rough surface. This could affect the coupling of the sensors as well as the reflectivity. This paper shows that the influence of the roughness at the coupling surface was significant for both 5 MHz and 20 MHz transducers. The influence of the roughness of the reflector surface did not show a significant change for different roughnesses. Higher transducer frequencies, as expected, were more affected by surface roughness, but even for 20 MHz, there was no strong effect caused by the roughness of the reflector surface. This shows that it is possible to produce defects without machining surfaces and still obtain good UT results.

It was shown that it is possible to produce test specimens with flat-bottom holes and to quantify their size and depth using ultrasonic testing. The accuracy of additive-manufactured flaws was investigated in comparison to conventionally EDM-manufactured FBHs. It can be shown that the FBH diameters were underestimated for both manufacturing techniques. The deviation was higher for the EDM-manufactured FBHs. This means the use of AM for manufacturing test blocks with known flaws is beneficial for flaw accuracy. The flaw location in depth could be determined with high accuracy for all measurements, as long as the echoes of the reflectors could be separated from the back wall echoes.

The investigations shown here are intended as a foundation for further studies to evaluate the specific effects of additive manufacturing on ultrasonic testing results. Building on this, the potential of manufacturing test specimens using additive techniques will be explored further. The influence of the powder that remains in internal reflectors and the inclination of reflectors is part of future investigations. In addition, the quality of reflectors manufactured in different ways is to be compared using computer tomography images. Investigations into other reflector types, such as grooves or cross-holes, are also to be expanded.

Based on these findings, it should be possible to produce a test body for ultrasonic testing with properties and reflector types adapted to the testing requirements. The ability to reproduce realistic component geometries is a major advantage of this method.