PSI - Issue 76

Robin Motte et al. / Procedia Structural Integrity 76 (2026) 74–81

76

Table 1. Welding parameters. Parameter

CONV

CMT

Average wire feed speed [mm / s]

125 243 21.8

125 226 17.5

Average current I [A] Average voltage U [V] Welding speed W s [mm / s] Stick-out length [mm] Arc length correction [%]

10 15

10 15 15

-

Shielding gas

Ar + 18%CO 2

Ar + 18%CO 2

Shielding gas flow [l / s] Heat input [kJ / mm]

0.3

0.3

0.42

0.32

2.2. Fatigue testing

The specimens were dynamically loaded in four-point bending by an Instron 8872 servohydraulic fatigue testing system at a stress ratio R = σ min /σ max = 0 . 1. A load cell with a 10 kN capacity was used. The setup is shown in Figures 2(a) and (b). 6 CONV and 5 CMT specimens were tested under constant amplitude conditions. All tests were conducted at 10 Hz, except at the stress range of 260.23 MPa where 3 Hz (for the CONV specimen) and 4 Hz (for the CMT specimen) were used, as the servohydraulic test rig could not maintain a stable sinusoidal load at higher frequency. An InfraTec ImageIR ® 8300 infrared camera with spatial resolution of 640 × 512 pixels, a measurement accuracy of ± 1Kor ± 1% of the reading and a temperature resolution of better than 20 mK at 30 °C (InfraTec GmbH (2018)) was used to monitor the as-built surface of the specimen (as shown in Figure 2(a)). For this purpose, a Thorlabs mid infrared enhanced gold coated mirror with a surface area of 50.8 × 50.8 mm² (Thorlabs, Inc.) was placed underneath the specimen (see Figure 2(b)). The temperature of the front of the four-point bending fixture was subtracted from that of the specimen to account for any fluctuations in the environmental temperature. For this reference temperature, the average in an area of 36 × 284 (pixels)² (approximately 7.278 × 57.412 mm²) was used. Additionally, the entire setup was covered in black cloths to avoid stray reflections. The as-built surface of the specimen and the four-point bending fixtures were coated in matte black spray paint to increase their emissivity to allow for accurate temperature measurements. Temperature data was acquired at a frequency of 200 Hz during 500 cycles for every 2000 cycles using the IRBIS ® 3 software. A thermogram obtained during one of the experiments is presented in Figure 2(c), showing the region of interest on the underside of the specimen and the reference region. According to Krapez et al. (2000), the temperature variation at the surface of a specimen during a fatigue test, without external heat sources, can be expressed by Equation 1. Here, T exp is the experimentally measured temperature, T 0 is the temperature at the start of the test, ∆ T is the mean rise of the temperature per cycle. T 1 and ϕ 1 are respectively the amplitude and phase of the first harmonic, while T 2 and ϕ 2 represent the same parameters for the second harmonic. The parameter ω corresponds to the angular frequency of the fatigue loading (with frequency f ) and t is time. Hence, the phases ϕ 1 and ϕ 2 are the phase shifts of the first and second harmonic to the sinusoidal force. It is assumed that higher order harmonics can be neglected (Chhith et al. (2017)).

T exp ( t ) = T 0 +∆ T · f · t + T 1 sin( ω t + ϕ 1 ) + T 2 sin(2 ω t + ϕ 2 )

(1)

For an adiabatic system, the amplitude T 1 is given by equation 2, where α is the coe ffi cient of thermal expansion, ρ is the mass density, C p is the specific heat capacity at constant pressure, and σ kk is the sum of the amplitudes of the principal stresses within a single cycle (Krapez et al. (2000); Talemi et al. (2016)). Bercelli et al. (2023) monitored fatigue crack initiation of as-built WAAM specimens in bending by detecting a decrease in T 1 . As the stresses at the