Abstract The aim of this paper is to discuss some important aspects concerning methods and models for shot peening simulation. An overview of papers concerning shot peening simulation from international conferences according to http://www.shotpeening.org/ICSP/confrns.htm together with other publications according to the reference list is presented and a short summary of each investigated paper/thesis is enclosed as an appendix to this report. This study is a part of an R&D project Simulation of product properties for efficient product development http://extra.ivf.se/prosim/ where the aim is to develop methods for sequential simulation of manufacturing processes. This project is financed by Vinnova and participating companies.
Since using the Finite Element Method (FEM) for simulation of shot peening is not industrial established yet the focus of the studied papers was to show that FEM is a useful tool in order to predict residual stress and plastic strain profiles. Several papers presented results that was validated with experiments and residual stress measurements with X-Ray diffraction (XRD) and concluded the usefulness of FEM. A lot of methods and models were proposed with different model descriptions, number of shots, type of analysis and type of material models and several studies proposed simplifications such as 2D models, static analysis and one shot studies etc. However one conclusion of this overview is that one should consider as many aspects as possible when modeling the process. The most advanced studies in the overview indicate a dynamic analysis where the target is a 3D model (block or cylinder) of the substrate below the surface. They also indicate that several shots should be considered where the shot is modeled in 3D with elastic or elastic perfectly plastic capabilities. Furthermore it is noted that the material model will have a large influence of the results and kinematic hardening, multilinear isotropic or kinematic multilinear hardening capabilities should be taken into consideration as well as the strain rate dependency. In several of the R&D presentations the influence of different process and material parameters on the residual stress profile was studied. The influence of shot velocity, shot diameter, impact angle, shape of shot and target hardening was discussed but since the studies were performed under different conditions it is difficult to draw generic conclusions concerning the impact of different process and material parameters on the residual stress profile.
2 Overview of previous R&D concerning shot peening simulation An overview of the studied papers/thesis is condensed in the table 1 and 2 below.
Nr Code Target conditions Model Dimension (mm) Target Yield (Mpa) Hardening Strain rate Temp. 1 Abaqus 3D block 1.5*1.5*0.85 1263Yes Yes Yes 2 Ansys 3D block 7R *4R *5R 600Bilinear no no 3 Abaqus 2D axisym - 1680Bilinear kinematic no no 4 LS-Dyna 3D block 0.8*0.8*1.6 1500Bilinear Cwpr Smnds no 5 Abaqus 2D axisym, 3D cylinder - 315, 930, 1120 Bilinear kinematic no no 6 Abaqus 3D Cylinder R=8D, H=3D 760Bilinear no no 7 Abaqus 3D block 4*4*2 1250Bilinear Cwpr Smnds no 8 Abaqus 3D block, 2D axisym 35*35*20 (35*20) 930Elastic perf plastic Yes no 9 Ansys 2D axisym Diam=100 H=50 630Multilinear isotropic no no 10 Murakami experiment 100*100*50 630 - - - Table 1: A summary of commercial codes and target conditions for shot peening simulation.
Nr Shot conditions Focus for investigation Validation Type Diam (mm) Number Vel (m/s) Friction 1 Rigid 0.5 - 2.5 1 - 7, 7 20 - 80 0.4 number of shots, velocity,angle, size XRD at Karlsruhe 2 Rigid 0.5 - 2 1, 2 50,75,100 0.25 velocity, size, shape, hrdness modul Edberg [11] 3 Elastic 0.8 1 - 7 6 Mj (75 m/s) - isotropic - kinematic hardening - 4 Elastic 0.8 4 - 25 50 - 100 0.1 number of shots, velocity, angle Torres [12] 5 Rigid 0.2-1 several 40 - 120 - Almen intensity - velocity, linear - elasto plastic XRD on Almen strip 6 Rigid 1.0 1 25 - 100 0.2 velocity, angle - 7 Rigid 0.72 several 37,73,91 0.1 FEM or Analytical expr - 8 Rigid 3.2 1 200, 300 0.1 Quasi static or dynamic XRD 9 Elastic 50.0 1 6 0.4 bilinear-multilinear, friction - no friction Murakami [10] 10 Steel 50 & 76.2 1 - 30 Drop from 2 m - Influence of number of shots and dynamic conditions on res. stress Table 2: A summary of shot conditions and focus for investigation in the studied papers
FE codes and analysis Software: The commercial code Abaqus was used in most of the cases but also Ansys and LS- Dyna was used in some cases. Dynamic or static analysis: Both dynamic and static analysis with large deformation capabilities was performed. The dynamic analysis was performed both in implicit and explicit mode and for example in [2] and [9] Ansys implicit solver was used with Newmark implicit time integration. In [3] the dynamic analysis was replaced by a static analysis with large deformation capabilities and where the velocity was replaced with an energetic criterion. One of the reasons was that dynamic simulations generate elastic vibrations in the material and the computations that are necessary to reach equilibrium can be time consuming. In [8] the effect of the velocity was matched by the size of the measured crater and a static analysis with large deformation was performed. Also in [8] a comparison between dynamic and static analysis was done and the conclusions was that for velocities below 200 m/s a static analysis is sufficient. However the experiments in [10] contradicts those conclusions showing a big experimental difference between a static and dynamic impact. The different conclusions in the presentations may be an effect of different input conditions concerning material, dimensions and impact velocity. Especially in [9] and [10] the conditions differs from real shot peening concerning impact velocity (very low velocity =6.4 m/s), shot diameter (very large shots =50 mm). The discussion above indicates that one should indeed be careful when trying to simplify the analysis. 3 Target conditions Both 2D axisymmetric models and 3D models have been used. Significant for all the shot peening simulations is the very small geometric volume (cube or cylinder) that is studied due to the fact that only a very small volume below the surface is affected by the shot peening procedure. Only in [9] a geometric large model was studied (axisymmetric model of 100 x 50 mm with shot diameter 50 mm) since the goal of this study was to validate the model against an experiment [10]. The mesh is often modeled very fine close to the impact zone in order to get a good resolution of the residual stress distribution below the surface and coarser away from the impact zone. The number of elements differs significant and in [8] a very small 2D model with 4000 8-nodes brick elements is described. However in [1] the 3D FE-mesh consists of 372000 8-nodes brick element where the smallest element size is 0.008 mm. Both [1] and [8] are validated with XRD- measurements. Mesh convergence studies has been performed in many presentations in order to secure the accuracy of the results. The characteristics of the studied target material differed from mild material (Yield =600 Mpa) to high strength steel (Yield =1680 Mpa).
Shot conditions Type of shot: Most of the shots in the discussed R&D are modeled as rigid. In [4] the shot is modeled as elastic and the results form experimental evaluation is very good. Also [3] use elastic shot and in [9] both elastic and elastic perfectly plastic shots are investigated and in [9] the validation with experimental residual stress measurements [10] is good. Number of shots: The number of shots varies from analysis with one shot [6], [8] and [9] to analysis with 19 shots where 1 to 7 shots in a sequence and then 7 simultaneous impacts are evaluated [1] and tests with 4 to 25 shots [4]. It is interesting to note that in [5] the conclusion is that the residual stress field is mainly influenced by the first impact and the difference due to the following ones are limited in a range about 15 %. Also in [4] the difference between 9 and 25 shots is small which emphasis this statement. However in [1] the difference in maximum residual stress between the 1:st and 7:th is substantial and also in the experiment [10] the difference between the 1:st and 30:th shot is very significant and the surface residual stress state at the center of the indenter goes from tension in the first shot to compression in the 30:th shot. The different statements probably depend on different conditions concerning dimensions, target material and shot speed. Thus attention should be paid concerning the number of shots that should be included in the model and this study indicates that a pattern of shots according to [1] or [4] should be used. Friction coefficient: In most of the studied papers the friction coefficient was set to 0.1. However in [9] a closer investigation of friction resulted in friction coefficient =0.4. Also in [1] the friction coefficient is set to 0.4.
Material models Stress strain relations: The description of plastic hardening differs form perfect elastic plastic [8], elastic plastic with bilinear behavior [2], [6] and [7], elastic plastic with kinematic hardening, [3] and [5] and elastic plastic with isotropic multilinear behavior [9]. Especially [3] compares isotropic and kinematic hardening and recommends kinematic hardening due to significant changes in residual stress profile, also for one-shot simulation. Also [9] investigates several material models and receives very good results with multilinear isotropic hardening compared to tests [10]. The conclusion is that one should pay attention to the choice of material model and evaluate especially kinematic hardening and multilinear isotropic hardening or combinations according to the material models described in fig 1.
4
Fig 1: Difference between bilinear, multilinear, isotropic and kinematic material models according to Ansys Inc Theory reference.
Influence of strain rate: The strain rate may have a significant effect on the stress strain curve according to fig 2:
Fig 2: Stress-strain curves for a mild steel as a function of strain rate according to [13]
Several mathematical models describes the influence of strain rate on hardening like Cowper- Symonds, J ohnson-Cook, Armstrong-Zerilli and Piecewise linear Cowper Symonds according to below.
Fig 3: Models for strain rate dependence on a mild steel according to [13] Where the solid lines indicate tests and dotted lines indicates the models.
The Cowper Symonds material model was used in [4] and [7] but the parameter settings differs in the presentations. In [8] a similar power law was used and in [1] a more complex material model was used with strain rate dependence, asymptotic characteristics of hardening 5 as well as temperature dependence included. In the other presentations the strain rate dependence was ignored and also in [8] it is indicated that the strain rate could be igno impact velocities below 200 m/s. The Cowper Symond formula is described below:
red for =() [1 +( / D) 1/P ] here () =static stress strain relation, W =plastic strain rate, D and P are adjustable
he Influence of shot peening and material parameters on the residual stress profile al
(0.5 2.5 mm) does not significant affect ss. esidual hardening: According to [2] the target hardening does not affect the maximum
iscussion and methods for shot peening simulation differ a lot in the studied presentations
arget: 3D block or cylinder of a small substrate n coeff ient: with multilinear kinematic hardening or alike and ignificant for most of the presented FE models is the fact that the studied geometric target f parameters. More information about Cowper Symonds parameter settings can be found in [14].
T Velocity: In [1] the velocity is varied between 20 80 m/s with no major change of maxim residual stress or surface residual stress. However the position of the maximum compressive stress gets deeper in the material as an effect of velocity increase. Also in [2] no major change in maximum residual stress but significant increase in the depth of the compressive layer for shot velocity 50 100 m/s. However in [4] an increase in velocity from 50 to 100 m/s may increase maximum compressive stress about 10 %. Also in [6] the shot velocity have effects on maximum compressive residual stress, especially in the low range of velocities (12 30 m/s) while at higher ranges (30 100) the influence is not so big. In [6] the maximum compressive stress is detected to 1.5*Yield stress. Shot diameter: In [1] the variation of shot diameter the maximum residual stress or surface residual stress. However the position of the maximum stress gets deeper in the material as an effect of diameter increase. The same results in [2]. Impact Angle: In [1] an increase in impact angle show a decrease in maximum residual stre Also the position of maximum residual stress moves towards the surface while the impact angle increase. Also in [6] it is stated that normal impact angle is most beneficial. Shape of shot: According to [2] the geometrical shape of the shot affects maximum r stress. Target residual stress. However increase in hardening increase the depth in compressive layer.
D The models and different approaches are suggested concerning type of model (2D or 3D), type of analysis (static or dynamic), number of shots and material model. The discussion above indeed indicates that one should be careful when trying to simplify the model and indicates a specification of a model according to below:
T Shot: Elastic or elastic perfectly plastic Frictio ic 0.1 0.4 Type of analysis: Dynamic strain rate dependence Number of shots: Several shots
S volume is very small due to the fact that only a small volume below the surface is affected o the process. A typical FE-model has the form of a block or a cylinder where the element size is very small around the impact zone.
6 It is always recommended to evaluate the model with XRD-tests or alike. Especially [1], [2], lop a he results from a shot peening simulation may give indications of the residual stress and y an be [4] and [9] has been successfully validated against experiments and residual stress measurements with XRD. In one presentation [7] an attempt has been made to deve mathematic expression for analytical prediction of residual stress.
T plastic strain profile below the surface as well as indentation depth. A simulation model ma also be a support for investigation of the influence of process parameters (impact speed, shot diameter, shape of shot, shot angle and impact pattern) and material characteristics on the residual stress and plastic strain profiles. A simulation model may also be a support to develop a better understanding of the physical effects of the shot peening process and c used for quantitative or qualitative analysis. 7 References
1. Schwarzer J ., Schulze V., Vhringer O., Finite Element Simulation of Shot Peening - A Method to Evaluate the Influence of Peening Parameters on Surface Characteristics, Proceedings from International Conference of shot peening nr 8 (ICSP-8) Munich, Germany, 2002, http://www.shotpeener.com/library/spc/2002066.htm
2. Meguid S.A., Shagal G., Stranart J .C., Daly J ., Three-dimensional dynamic Finite element analysis of shot-peening induced residual stresses, Finite Elements in Analysis and Design 31, 1999, 179-191
3 Rouhaud E., Oakka A., Ould C., Chaboche J ., Francois M., Finite Elements Model of Shot Peening, Effects of Constitutive Laws of the Material, Proceedings ICSP-9, Paris, France, 2005
4. Majzoobi G.H., Azizi R., A 3-D Numerical Method of Shot Peening Process using Multiple Shot Impacts, Proceedings ICSP-9, Paris, France, 2005
5. Guagliano M., Vergani L., Bandini M., Gili F., An approach to relate the shot peening parameters to the induced residual stresses, Proceedings ICSP-7, Warsaw, Poland, 1999, http://www.shotpeening.org/ICSP/icsp-7.htm
6. Hong T., Ooi J .Y., Favier J ., Shaw B., A numerical simulation to relate the shot peening process parameters to the induced residual stresses, Proceedings ICSP-9, Paris, France 2005
8. Boyce B.L., Chen X., Hutchinson J .W., Ritchie R.O., The residual stress state due to a spherical hard-body impact, Mechanics of Materials 33, 441- 454, 2001
9. Zion H. Lewis, A Dynamic Finite Element Simulation of the Shot-Peening Process, Ph.D. Thesis, Georgia Institute of Technology, USA, 2003
10. Kobayashi M., Matsui T., Murakami Y., Mechism of creation of compressive residual stress by shot peening, Int. J . Fatigue. Vol 20 no 5, 1998, 351-357.
11. Edberg J . Lindgren L., Mori K., shot peening simulated by two different finite element formulations, Simulation of materials Processing: Theory, Methods and Applications, Balkema, Rotterdam, 1995, pp. 425 430
12. Torres M., Voorvald H., An evaluation of shot peening residual stress and stress relaxation on the fatigue life of AISI 4340 steel, Int J . of Fatigue, Vol. 24, pp 877 - 886
13. Dietenberger M, Buyuk M, Kan C, Development of a high strain-rate dependent vehicle model, LS-Dyna Anwenderforum, Bamberg 2005
14. Cunat P, Stainless steel properties for structural automotive applications, Metal bulletin International Materials Conference, Cologne, 2000 8
Web-sites
Shot peening organization: http://www.shotpeening.org/ Conferences in shot peening: http://www.shotpeening.org/ICSP/confrns.htm Shot peening organization: http://www.shotpeener.com/ Ansys theory reference guide: www.oulu.fi/atkk/tkpalv/unix/ansys- 6.1/content/theory_toc.html 9 Appendix: Short summary of papers and thesis [1] Finite element simulation of shot peening a method to evaluate the influence of peening parameters on surface characteristics (Schulze et al)
FE-Code: ABQUS
FE-model: Target plate modeled as 3D mesh of 1.5 x 1.5 x 0.85 mm, 372000, 8-nodes linear brick with reduced integration and hourglass control, smallest element size 0.008 mm. Shots modeled as half spherical rigid surfaces connected to point mass and rotary inertia, diameter 0.56 mm. Simulation of 19 impacts. Isotropic Coulomb friction coefficient 0.4 was used to describe the contact between the shot and target. Impact order according to below.
Fig 1: FE-model and impact order according to [1].
Material: Target material is quenched and tempered steel, 42CrMo4 Yield strength 1263 Mpa, Tensile Strength 1373 Mpa. Shot material is cast steel shot, mass density 7.85 g/cm 3
Material models: A constitutive law that describes the influence on temperature and strain rate on flow stress (shear stress required to cause plastic deformation of solid metals) is implemented as a subroutine in ABAQUS (Including Boltzmann constant). A constitutive equation that describes the work hardening behavior as a function of initial yield stress and hardening rate and asymptotic characteristics of hardening is implemented. Material constants are determined by tensile tests.
Analysis: Influence of shot velocity (20, 40, 50, 65, 80 m/s), shot diameter (0.5, 1, 1.5, 2, 2.5 mm), impact angle (10, 30 and 55 degrees) and number of impacts (1 to 7 impacts in a sequence and 7 simultaneous impacts according to fig 1) on residual stress profile according to fig 2:
Validation: Experiment and X-Ray diffraction comparison according to fig 2:
10
Fig 2: Analysis of shot parameter variations and validation of model against experiments according to [1].
Summary: - Constitutive law that describe the influence of strain rate and temperature. Also work hardening included. Rigid shot, Abaqus/explicit - Number of impacts: influence the shape of the profile but not the maximum residual stress - Variations in impact velocity: No major impact of maximum residual stress, some influence of maximum residual stress position and deep of intendent. - Variations in shot diameter: No influence of surface and maximum residual stress. Some influence on position. - Good agreement with experiment. The model is an effective tool for prediction of residual stresses.
11 [2] Three dimensional dynamic Finite element analysis of shot peening induced residual stresses (Mequid et al)
FE-Code: Ansys
FE-model: Single shot and twin shot, each of radius R against 3D mesh of plate with width 7R, height=4R and breadth =5R where. Dimensions carefully determined as a result of numerous models to establish the effect of the boundary. Eight node bricks and four-node tetrahedral elements with large strain capabilities with brick elements at impact zone. A number of mesh designs in order to find the best mesh (convergence tests). Contact elements on both shot and target but only on lower half of shot surface. Elastic Coulumbs law with friction coefficient 0.25 was used. Normal and tangential stiffness of contact elements were decided to 10 8 N/m and 10 6 N/m. Symmetry boundary conditions at z=0 and x=0 planes and fully constrained at work piece bottom surface. The effect of strain rate and elastic wave propagation was not considered in the study. Dynamic analysis carried out using Newmark implicit time-integrations scheme. Tests with single shot and two shots simultaneously (dynamic cointendentation) according to below.
Fig 3. FE-models according to [2].
Material: Target in high strength steel =7800 Kg/m 3 and E=200 Gpa, Yield stress = 600Mpa and strain hardening rate=800 Mpa. Shot assumed to be rigid.
Material models: Elasto plastic with bilinear behavior.
Analysis: Single shot: Influence of shot velocity (50, 75, 100 m/s), shot size (R=0.25, 0.5 and 1.0 mm), shape of shot (a/b=0.25, 0.5, 1 and 2, see below) and hardening characteristics (50, 800 and 1600 Mpa) on residual stress profile. Co-indentation model: Influence on distance between shots (C/R =1, 1.5 and 2) on residual stress profile.
12
Fig 4: Simulation of residual stresses as a function of parameter variations according to [2]
Validation: The model was validated against an earlier work of Edberg et al [11] where the target and shot characteristics were the same as in this report.
Summary: - 3D model, medium strength material 1 2 impacts, Ansys 5.3 implicit solver, Elastic plastic with bilinear behavior. Rigid shot - Variations in impact velocity: Increase in the maximum equivalent plastic strain and increase in depth of compressed layer, no major impact on residual stress profile - Variations in shot size: No major influence on maximum residual stress and surface residual stress. Influence in position of maximum residual stress. - Variations in shot shape: Influence of the maximum residual stress - Variation in target hardening: Increase in hardening coefficient increase the depth in compressive layer, no change in maximum residual stress state and decrease in surface residual stress. - Variations in distance between 2 shots: Variations in residual stresses between the centerline of the target between the co- indenting shots Proposed finite element model is capable of capturing the main features of the residual stress field. 13
[3] Finite elements model of shot peening, effects of constitutive laws of material (Rouhaud et al)
FE-Code: ABAQUS
FE-model: 2D axisymmetric, shot with radius=0.4 mm Velocity 75 m/s (=6.03 MJ kinematic energy), Dynamic analysis generates elastic vibrations and the computations that are necessary to reach equilibrium can be time consuming. Static analysis based on an energetic criterion is therefore chosen, normal contact with no friction, strain rate sensitivity neglected. Elastic shot
Material: Yield stress 1680 Mpa
Fig 5: FE-model and models for isotropic and kinematik hardening according to [3]
Material models: Isotropic or Kinematic hardening
Analysis: Influence of kinematic hardening (none, small slope, average slope high slope) of residual stress curves
Fig 6: Comparison of isotropic hardening and kinematic hardening after one shot according to [3]
14
Fig 7: Comparison of isotropic hardening (a) and kinematic hardening (b) after several shots according to [3]
Summary: - Medium strength material, 1 7 impacts at same place, Abaqus - Isotropic or kinematic hardening, quasi-static analysis, elastic shot. - Variation of material law: Kinematic hardening can greatly enhance the shot peening models. - Kinematic hardening models are well suited for cyclic solicitations and can enhance shot peening models.
15 [4] A 3-D numerical method of shot peening process using multiple shot impact (Majzoobi et al)
FE-Code: LS-DYNA
FE-model: Target plate 0.8x0.8x1.6 mm. Shot radious 0.4 mm, friction coefficient =0.1. Number of shots: 4, 6, 8, 9, 13 and 25 shots. Impact velocities in range of 50 to 100 m/s. Elsatic shots and symmetry conditions.
Fig 8: FE-model and impact pattern according to [4]
Where () =static stress strain relation, =plastic strain rate, D and P are adjustable parameters. D was set to 2*10 5 s -1 and P to 3.3.
Analysis: Influence of number of shots and impact velocity and shot angle of residual stress profile.
Fig 9: Results according to [4]
Validation: Flat steel plate 20x300x300 mm, two single balls of 3 and 5.5 mm fired against the target using explosive actuated gun launcher (100 to 400 m/s). Examination of the 16 intendentations using optical microscopy and comparison between numerical and experimental residual stress profiles [Torres]
Summary: - 3D models, several impacts at different places, LS-Dyna - Elastic plastic with Cowper Symonds material model (influence of strain-rate), elastic shot - Number of shots: Uniform state of stress at a particular shot number - Variation in shot velocity: Increase of velocity improves the residual stress to a particular point. Further increase may reduce the maximum residual stress. - Shot peening can successfully be simulated by LS-DYNA providing a proper material model is selected. A uniform state of stress is achieved at a particular shots number. The increase of the impact velocity improves the residual stress distribution up to a particular point. Further increase in velocity may reduce the maximum residual stress. Impact angle affects the residual stress profile.
17 [5] An approach to relate the shot peening parameters to the induced residual stresses (Guagliano et al)
FE-Code: ABAQUS
FE-model: Both 2D axisymmetric for tests with one shot and 3D models for tests with one shot and several shots. First order elements, 22379 nodes, rigid shot (diam 0.2 1 mm) and dynamic impact of one or more shots against a plate (velocity 40 120 m/s).
Fig 10: FE models according to [5].
Material: Mechanical characteristics of 3 materials: Fe510 with yield stress =315 Mpa and tensile stress =510 Mpa, 35CrMo4 with yield=930 Mpa and tensile =1050 Mpa, C70 with yield strength =1120 Mpa and tensile strength =1300 Mpa.
Material model: Both linear elastic and elastic plastic with kinematic hardening rule. The linear model showed a difference of contact pressure compared to Hertz value with less than 1 %.
Analysis: Evaluation of shot velocity, diameter, material, angle and number of impacts with focus on the value of the maximum compressive stress and its depth and the depth where the residual stress change sign. Analysis of the relation between shot velocity and Almen intensity. Development of dimensionless peening index.
Fig 11: FE Analysis of shot velocity based on Almen intensity according to [5].
18 Validation: Several FE models with different numbers of elements were tested in order to evaluate the accuracy of the results. Residual stress field were measured with X-Ray diffraction method. It should be noted that the residual stress field might be affected due to material removement with an electro chemical device during X-ray measurements.
Summary: - 2D and 3D models, one shot and several shots, Abaqus/explicit - Elastic plastic with kinematic hardening rule, rigid shot - Number of impacts: The residual stress field is mainly influenced by the first impact. The following are limited in range of about 15 %. - FEM for relation between Almen intensity and shot velocity. - Angle of impact on residual stress state is limited if the shot speed is higher than 60 m/s. Surface stress is the one that is more influenced by the numerical model and parameters probably to the high surface distortion of the elements. The residual stress field is mainly influenced by the first impact and the differences due to the following one are limited in a range of about 15 %. Fatigue resistance is mainly due to residual stress field. The hardening also play a role in the increment of fatigue but is less important than the residual stresses.
19 [6] A numerical simulation to relate shot peening process parameters to the induced residual stresses (Hong et al)
FE-Code: ABAQUS
FE-model: FE-mesh acc. to below with radius R=8d shot , height H=3d shot where d shot is the shot diameter. One half of a circular plate with symmetry conditions. Restrained against all displacements and rotations on the bottom. 8-nods linear brick elements with reduced integration with element size 0.05 x d shot x 0.05d shot x 0.05d shot , initial diameter of shot =1 mm. ABAQUS analytical rigid surface with a mass position at the center to model the shot. Convergence tests with different meshes and element types to ensure that the numerical results were not affected by the choice of mesh and element types. Friction coefficient =0.2 and reference velocity =75 m/s.
Fig 12: FE-model and results according to [6].
Material: Target: E=200 Gpa, Poissons ratio=0.3, density=7800 kg/m 3 , yield stress=760 Mpa, strain hardening =500 Mpa, density of shot =7800 kg/ m 3 , mass of shot =4.085 mg.
Material model: Linear elastic strain hardening plastic material model
Analysis: The influence of the influence of impact velocity and angle.
Summary: - 3D model, Abaqus/explicit - Linear elastic strain hardening plastic material model, rigid shot - Impact velocity: Little effect on surface residual stress but significant effect on maximum subsurface residual stress and depth of compressive residual stress zone. - Impact angle: Normal angle impact results in most beneficial compressive residual stress. - The impact velocity appeared to have little effect of surface residual stress but showed significant effect on the maximum sub surface residual stress and depth of compressive residual stress zone. Maximum compressive residual stress reached 1.5 x yield stress. Impact angle close to normal impact produces most beneficial compressive residual stress as expected. 20 [7] Numerical simulation of multiple shot impact (Al Hassani et al)
FE-Code: ABAQUS
FE-model: A solid steel block of width, breadth and height 4, 4 and 2 mm. Symmetry and of the geometry is modeled. 8 node brick elements with 14 elements in each direction acc to below. Spherical rigid surfaces of diameter 0.72 mm. Each carried a mass element (1.5128 mg). In a second case the spheres were arranged above one another. Time lag between impacts was 1.27 s. Friction coefficient=0.1.
Fig 13: FE-model according to [7].
Material: High tensile steel with E=2.03 Gpa, poisons ration =0.3, mass density =7830 kg/m 3 . Bilinear material work hardening between 1250 1550 Mpa with plastic strain 0 1.
Material model: Strain rate dependence under dynamic loading according to Cowper- Symonds power law:
=() [1 +( / D) 1/P ]
Where () =static stress strain relation, =plastic strain rate, D and P are adjustable parameters. D was set to 40s -1 and P to 5.1.
Summary: - A theoretical solution was presented acc to below as a complement to the FE-simulations and compared. By using a strain hardening strain-rate sensitive material it is possible to model the effects that the repeated and progressive impacts of shots have on the residual stress profile and extent of surface hardening. Caution should be taken before the results are compared with experiments.
Fig 14: Analytical results according to [7]. 21
[8] The residual stress state due to a spherical hard-body impact (Boyce et al)
FE-Code: ABAQUS
FE-model: A Substrate was approximated both as a cylindrical block with 35 * 35 * 20 mm and as an axi symmetrical model with diameter =35 mm and depth =20 mm. An axisymmetric mesh with 4000 8-nodes elements. Friction coefficient=01. Static finite element calculations. Diameter of shot, D=3.2 mm, rigid contact surface option used to simulate the rigid intender. Impact velocities =200 m/s and 300 m/s.
Material Model: Elastic-perfectly plastic with a V. Mises surface to specify yielding. For dynamic simulation also the strain rate was included according to:
Where is the strain rate sensitive yield stress and is the yield stress at .
Analysis: Both Quasi-static simulation with Abaqus implicit and dynamic simulation - Only one shot. In the quasi-static analysis the size of the crater was adjusted by controlling the force such that the resulting crater dimensions matched with the dimensions measured on the experimental impact.
22 Validation: The results from the FE-simulations where compared against experiments with shots against a plate and afterwards with X-Ray diffraction measurements of residual strain and stresses according to below:
Fig 15: Comparison between FE-simulation and XRD-measurements according to [8].
Summary: - Experiments shows that the initial residual stress state is substantially reduced by relaxation or redistribution during subsequent fatigue cycling (30 50 % after first fatigue cycle). The model was modified to take into account the Baushinger effect and showed a reduction of residual stresses on the order of 40 % after the first cycle. - Observations of significant tensile stresses at the crater rim. In the case of multiple overlapping shots the tensile zone is offset as a result of successive shots, which indicates that incomplete peening coverage will leave uncompressed rims leading to degradation of fatigue life time. - The shape of the spatial strain gradient predicted by FEM model is consistent with experimental observations with X-Ray diffraction. - A 200 m/s impact can be modeled using a quasistatic approach, ignoring time dependent effects and simply simulate the effect of velocity by matching the relative shape of the crater. - There are substantial errors from the application of quasistatic analysis to the simulation impact of 300 m/s. These errors part attributable to time dependent effects like strain-rate sensitivity, elastic wave interactions etc).
23
[9] A dynamic Finite element simulation of shot peening process (Zion)
FE-Code: Ansys
FE-model: Axisymmetric model 100 mm diameter and 50 mm depth acc to below, 8 node 2D solid element (plane82) and contact elements (Conta172), friction coefficient =0.4 and elastic ball with diam. =50 mm. Force for static case =49 KN and velocity at impact for dynamic case =6.26 m/s. Mesh convergence checks were performed. Dynamic finite element solution with Newmark integration scheme. Implicit solution with Newmark time integration scheme was chosen due to its stability along with the fact that the stress wave propagation was not considered to play a role in the residual stress outcome:
Fig 16: Static and dynamic axisymmetric FE-models in 2D according to [9]
Material: Medium steel, S48C with Yield strength =375 Mpa, Tensile strength =630 Mpa E=207 Gpa and poissons ratio =0.28 and density =7840 g/m3. Ball with elastic-perfectly plastic behavior with both yield and tensile strength =2.534 Gpa.
Material Model: Several material models in Ansys were tested. Bilinear kinematic hardening with and without friction, multilinear isotropic hardening with and without friction. Since the aim was to capture a single load unload drop event there is no theoretical difference between kinematic and isotropic hardening in this case.
Fig 17: Difference between bilinear and multilinear material input according to [9].
Analysis: - Both static and dynamic analysis and tests with different material models and friction coefficients, Ansys implicit. - Also analysis of shot peening parameter variations on high strength steel with much smaller shots and higher velocities and analysis according to Design of Experiments (DoE)
24 Validation: The models were validated against Murakami experiments [10]. Murakami dropped steel balls (diam. =50 and 76.2 mm) against a steel plate (material S48C and dimensions 100 x 50 mm) from a height of 2 m and measured residual stresses and shapes of indentations according to below. He also made static tests where the ball was pressed against the plate with Force =49 KN and 108 KN.
Fig 19: Residual stress distribution at static tests with load 49 KN and dynanic test with dropping height 2 m according to [10], ball diam.=50 mm.
Fig 20: Comparison between FE-results and Murakami experiments according to [9].
Both the shape of indentation and the residual stress profile differed between the static and dynamic experiments. In the dynamic experiments there was a surface tension residual stress at the center of the indentation. The tension stress component disappeared when dropping several balls on the plate as showed below:
Fig 21: Results from experiments according to [10] where the surface residual stress changes as a result of several impacts.
25 Summary: - With the multilinear material model together with friction coefficient there was a good correlation between the experiments and simulation models. In the dynamic case the tension puddle according to Murakami experiment [10] was captured in the FE- simulation. - No major impact when raising the friction coefficient from 0.4 to 0.8 - DoE investigation with one shot