PSI - Issue 57

Felix-Christian Reissner et al. / Procedia Structural Integrity 57 (2024) 411–419

415

F.-C. Reissner et al. / Structural Integrity Procedia 00 (2023) 000–000

5

3.1. Simulation of the shot peening process The shot peening process is simulated by a combination of DEM and FEM. The DEM method is used to generate the shot for the shot peening. The shot is modeled as single spheres which are generated as DEM spheres by a DEM generator Dassault Systèmes (2016) in Abaqus 2018. To reduce the computation time, the contact between the generated spheres is neglected. Furthermore, the spheres were generated from a plate shaped generator with a uniform distribution regarding the generation position, a diameter d = 0 . 2 mm , a density ρ = 7800 kg m 3 , a velocity of v = 70 m s and a massflow of ˙ m = 0 . 02 kg s . Under these conditions, it was found that a cover of 100 % is reached after 0 . 75 · 10 − 3 s and a cover of 200 % after 1 . 5 · 10 − 3 s simulation time. Because of the symmetrical load case, only a quarter of the cross bore and the chamfer are modeled in FEM, Figure 3. To consider kinematic and isotropic hardening effects as a result of the shot peening, the Chaboche-Lemaitre model Chaboche (1986) is used. It has been shown by Ould et al. (2006) and Rouhaud et al. (2005) that if only isotropic hardening is considered, the residual stresses are overestimated. The parameters of the Chaboche-Lemaitre model are fitted on the basis cyclic of tension and compression tests. Since the FEM elements on the surface are distorted after the shot peening simulation, the residual stresses are calculated by taking the average of the absolute maximum principal stresses (Max Principal (ABS)) in a 25 mm 2 square area.

Fig. 3. Global model and submodel.

To validate the numerical models, XRD measurements are performed. In the case of the shot peening simulation of the cast iron specimens, the simulation only captures a general qualitative trend, compared to the measurement, Figure 4 (a). In the case of the steel specimens, the simulation reproduces the measure ments qualitatively well, Figure 4 (b). Quantitatively, however, there is deviation of about 400 MPa at the surface. A possible explanation for the deviation lies in the fitting of the material model. Experimentally, it is not possible to reproduce the high plastic deformation speed that occur during shot peening. As a result, the material model is not adapted to the real kinematics. Another explanation lies in the method of XDR measurement, where the measurement area is determined manually and the average of four points is used as the result. Since the residual stresses are very heterogeneously distributed and the area over which the average is taken in the simulation is larger than the area for the XRD measurement, the measurement could have been taken from an area with higher residual stresses than the true average residual stresses. However, it is assumed that the XRD measurement has a standard deviation of 50 MPa due to the heterogeneously distributed residual stresses. 3.2. Load simulation While the shot peening simulation is conducted with an explicit solver, the simulation of the entire shaft is conducted with an implicit solver to ensure a higher accuracy of the simulation. To account for residual stresses from the shot peening simulation, the quarter model of the cross bore is used as a submodel within the model of the entire shaft, Figure 3. The residual stresses from the shot peening simulation are transferred to the submodel as initial conditions.