Chapter 3: Numerical Simulation II. Numerical simulation of SE(T) C specimens
62 All the specimens are tested for 0.125 ≤ a/W ≤ 0.625 which corresponds to crack lengths of 1mm to 5mm in a total width of 8mm. A conventional crack analysis mesh configuration is used with a focused ring of 15-node quadratic triangular prism (C3D15) elements around the crack tip, figure 8. Around this first cylinder of triangular elements five concentric rings of 20-node quadratic brick, reduced integration (C3D20R) elements are generated. These rings are subsequently used for calculation of the J-Integral wherein the values on the first ring are ignored [11]. There has to be convergence on all other element rings for results to be valid (see ABAQUS/Standard TM user’s manual [11] for a detailed analysis). The position of the center of these rings defines the crack front. A typical 3D finite element mesh contains about 10000 elements. A transverse plane surface is then chosen from the center of the rings to the crack edge which is defined as the crack plane. In the analysis ABAQUS/Standard TM duplicates the nodes on the crack plane and then assigns one set of nodes to one face and the other set to the other. This creates a crack with no opening at initial condition (zero charge) andat the crack tip a singularity in terms of stresses and strains. However since the large strain zone is very localized at the singularity the problem can be overcome satisfactorily using small-strain analysis. The crack tip strain depends on the choice of the material model used. In this analysis an incremental plasticity isotropic hardening model is used. If r is the distance from the crack tip then the strain singularity valid for small strain is:
' = E − ϑ 2
E (4)
The use of plane strain conditions in thin specimens may not represent in an accurate manner the state of stress and strain in a thin specimen. However, keeping in view the need for using a coherent and unique damage parameter for comparisons, only the plane strain condition is considered. The advantage of this parameter is that atroomtemperature, it gives almost the same values as for an elastic analysis since the crack tip plasticity is very low. However at higher temperatures this parameter shows the effects ofcrack tip plasticity while same dimensions (and units) as the K I parameter. The second advantage is important for comparison of fatigue crackpropagation curves at
The difference in the crackpropagation curves may be explained by two reasons. First the K I , have been calculated for plane strain condition with smallscale yielding. In the experiments, these conditions do not strictly prevail. In particular in the 0.6mm specimen the effect of the plane stress and large plastic zone has to be considered. The second explanation could be a different crack closure mechanism due to larger plastic deformations at the crack tip of the 0.6mm specimen compared to the 2.5mm specimen. More work has to be done on the stress state of the crack tip to have better insight into the difference incrackpropagation curves.
The difference in the crackpropagation curves may be explained by two reasons. First the KI, have been calculated for plane strain condition with smallscale yielding. In the experiments, these conditions do not strictly prevail. In particular in the 0.6mm specimen the effect of the plane stress and large plastic zone has to be considered. The second explanation could be a different crack closure mechanism due to larger plastic deformations at the crack tip of the 0.6mm specimen compared to the 2.5mm specimen. More work has to be done on the stress state of the crack tip to have better insight into the difference incrackpropagation curves.
The difference in the crackpropagation curves may be explained by two reasons. First, the K I , are calculated for a
plane-strain condition with small-scale yielding. In the experiments, these conditions do not strictly prevail. In particular, in the 0.6 mm specimen, the effect of the plane stress and large plastic zone has to be considered. The second explanation could be a different crack closure mechanism due to larger plastic deformations at the crack tip of the 0.6 mm specimen compared to the 2.5 mm specimen. More work has to be done on the stress state of the crack tip to have better insight into the difference incrackpropagation curves. Conclusion
' = E − ϑ 2
E (4)
The use of plane strain conditions in thin specimens may not represent in an accurate manner the state of stress and strain in a thin specimen. However, keeping in view the need for using a coherent and unique damage parameter for comparisons, only the plane strain condition is considered. The advantage of this parameter is that atroomtemperature, it gives almost the same values as for an elastic analysis since the crack tip plasticity is very low. However at higher temperatures this parameter shows the effects ofcrack tip plasticity while same dimensions (and units) as the K I parameter. The second advantage is important for comparison of fatigue crackpropagation curves at
Figure 3. Schematic of the SET standard specimens.
Figure 4. (a) Variation of correction factor F(a/W) vs. a/W at two H/W ratios. (b) Relative error of F(a/W) estimation between ABAQUS® and [7] and [8] The relative error is defined as: (F c -F l )/F l , where F c = calculated correction factor using ABAQUS® and F l = correction factor from literature. This error in Figure 4b does not exceed 8% over the whole range ofcrack measurement. There is also a tendency towards stabilisation of the error with increasing a/W. It was therefore considered that the procedure of evaluation of KI with ABAQUS® is relevant for the test conditions presented here.
Toulouse, INSA, UPS, EMAC, ISAE; ICA (Institut Clément Ader), 10 Av. Edouard Belin, Toulouse Cedex 4, France, 3 Mechanical Engineering
Department, University of Engineering and Technology, Taxila, Pakistan
A B S T R A C T A novel approach is used to characterise the surface damage of AISI H11 hot work toolsteel. The fatigue crack growth rate (FCGR) experiments on thin specimens, considered representative of the surface oftool steels, are carried out. Single edge notched tension specimensof 8 mm width and different thicknesses are used in the study. Initially, the ef- fect of thickness (scale) on the FCGR is investigated to establish if there exists a differ- ence between bulk and near surface properties of the toolsteel. Then, the effect of R value on the thin specimens is investigated. All the experiments are carried out atroomtemperatureand600 °C. These temperatures represent the limits of use of this steel. Paris curves are established. Effect of R ratio on the threshold value ofpropagationat el- evated temperatures is investigated in detail. A special ascending ΔK experiment for es- tablishing the threshold ofpropagationat elevated temperatures insmallspecimens is proposed, and the results are presented. The increase in R ratio increases the FCGR at low temperature, while it has no effect at600 °C. A reduction in thickness shows a reduc- tion in the FCGR. Increase intemperature increases the FCGR and dramatically in- creases the threshold ofcrackpropagation. The sharp increase in threshold value is studied in detail. Scanning electron microscopy of the specimens is performed to explain some of the characteristics observed in the specimen testing.
Figure 3. Schematic of the SET standard specimens.
Figure 4. (a) Variation of correction factor F(a/W) vs. a/W at two H/W ratios. (b) Relative error of F(a/W) estimation between ABAQUS® and [7] and [8] The relative error is defined as: (F c -F l )/F l , where F c = calculated correction factor using ABAQUS® and F l = correction factor from literature. This error in Figure 4b does not exceed 8% over the whole range ofcrack measurement. There is also a tendency towards stabilisation of the error with increasing a/W. It was therefore considered that the procedure of evaluation of KI with ABAQUS® is relevant for the test conditions presented here.
The difference in the crackpropagation curves may be explained by two reasons. First the K I , have been calculated for plane strain condition with smallscale yielding. In the experiments, these conditions do not strictly prevail. In particular in the 0.6mm specimen the effect of the plane stress and large plastic zone has to be considered. The second explanation could be a different crack closure mechanism due to larger plastic deformations at the crack tip of the 0.6mm specimen compared to the 2.5mm specimen. More work has to be done on the stress state of the crack tip to have better insight into the difference incrackpropagation curves.
(b)
e no clear crack closure could be optically
of 0.250mm specimen. (a) Slip bands (b)
may be explained by two reasons. , have been calculated for plane strain condition with smallscale yielding. In the experiments, these conditions do not strictly prevail. In particular for specimens the effect of the plane stress and large plastic zone beyond has to be considered. The second explanation could be a different crack closure mechanism due to larger plastic deformations at the crack tip of the thinner This is fairly evident in the crackpropagation curve of 0.110mm specimen (Fig. 10) the effect ofcrack closure can be . With an increase in R from 0.1 to 0.6 the , whereas the same for R = 0.1 and 0.7 respectively. The SEM micrographs of 0.250mm specimens showing presence of ents will be carried out in the future to
(b)
e no clear crack closure could be optically
of 0.250mm specimen. (a) Slip bands (b)
may be explained by two reasons. , have been calculated for plane strain condition with smallscale yielding. In the experiments, these conditions do not strictly prevail. In particular for specimens the effect of the plane stress and large plastic zone beyond has to be considered. The second explanation could be a different crack closure mechanism due to larger plastic deformations at the crack tip of the thinner This is fairly evident in the crackpropagation curve of 0.110mm specimen (Fig. 10) the effect ofcrack closure can be . With an increase in R from 0.1 to 0.6 the , whereas the same for R = 0.1 and 0.7 respectively. The SEM micrographs of 0.250mm specimens showing presence of ents will be carried out in the future to
integrate microstructural heterogeneities. The close agree- ment in thermal expansion and tensile (strength and strain-to- failure) measurements between ultrathin and bulk specimens demonstrates improvements in local characterisation capabil- ities at high temperatures. Coated superalloy exhibit a strong gradient of mechanical properties at high temperatures with a particularly poor strength of the NiCoCrAlYTa coating above the brittle-to-ductile transition temperature. A procedure to control the atmosphere was found in order to avoid surface degradation of ultrathin specimens even for alumina-forming alloys at very high temperature. Preventing alumina-forming superalloy from excessive oxidation was particularly difficult because of the very high stability of alumina, but it was also necessary to slightly oxidise the alloy in order to prevent its sublimation. The effect of such atmospheric degradations (oxidation, sublimation) is particularly critical when deal- ing with high surface/volume ratio specimensand materials subjected to phase transformations and has been qualitatively documented in the present paper. This technique can be used for the local mechanical characterisation of graded materials such as coated materials, materials presenting surface degra- dation (oxidation/corrosion, irradiation, machining, oxygen embrittlement) or surface treatments (grit blasting, carbura- tion, nitridation). This paper shows the level of control of specimen thickness and atmosphere required to investigate the mechanical behaviour ofspecimens with high surface/ volume ratio at elevated temperatures.
Université de Toulouse ; Mines d’Albi ; Mines Albi, Institut Clément Ader ; Campus Jarlard, F-81 013 Albi cedex 09, France
Abstract
The high strength of 5% chromium steels at elevated temperature is directly related to their complex microstructure generated by specific heat treatments that consist of thoroughly controlled austenitizing, quenching and tempering. Fracture toughness and fatigue crackpropagation are studied for various heat treatments. It is shown that changing the austenitizing temperature, i.e. the grain size and morphology of primary carbides, does not change the fracture characteristics. However, modifications of tempering, directly impacting the alloy hardness, result in drastic changes of fracture toughness andcrackpropagation rates. As no change in grain or martenstic lath morphology is observed, the associated modification of the Paris law coefficients is attributed to a change in the distribution and volume fraction of nanometric carbides investigated by Small Angle Neutron Scattering.
embrittlement has not to be considered for the 316 L(N) austenitic steelin the temperature range [473 e823 K]. These authors observed a slight reduction in the mechanical properties for the specimens tested by small punch test in liquid sodium but the fracture mode remains fully ductile in their conditions. They concluded that the 316 L(N) is prone to liquid metal accelerating damage (LMAD) but not Liquid Metal Embrittlement (LME). Still, this study may have missed the range or the conditions where LME can be maximum, even though a broad range of mechanical test parameters such as temperatureand strain rate was investigated [ 17 ]. Our study shows that the 304 L, the 316 L and, contrary to prior results of reference [ 13 ], the 316 L(N) austenitic steels have a lower crackpropagation resistance in oxygenated liquid sodium. This difference in LME response for 316 L(N) could be attributed to the higher triaxiality of the axisymmetric notched geometry compared with the SPT test used by Serre et al. [ 13 ]. More critically, the absence of LME could also be explained by an incomplete or absence of wetting since it is possible that an insuf ficient wetting was reached before SPT tests. Hemery et al. [ 5 ] had illustrated the in fluence of the exposure duration on the LME susceptibility and found that an increase in the exposure duration enhances the wettability of T91 martensitic steel by liquid sodium and that LME was coincident with the onset of good wetting of the specimens.
Abstract: The cyclic behaviour ofX38CrMoV5 (AISI H11) toolsteel with a nominal hardness of 47HRC has been predicted. Basically, thermo-elastoplastic and thermo-elastoviscoplastic constitutive laws are investigated. First, various uniaxial isothermal conditions (LCF) with different strain rate, strain amplitude andtemperature level are investigated. Then the constitutive laws are examined under various TMF loading conditions. The simulated results by both the approaches are compared with experimental results in terms of stress–strain behaviour and cyclic softening. Some applications of the model for simulation of thermal fatigue sample are shown. Taking into consideration the results of this work, the goal is to further characterise the limitations of these constitutive laws under complex and severe loading conditions, i.e., under variable temperature, variable strain amplitude and thermal fatigue structural specimen. Keywords: tempered martensitic steels; Low Cyclic Fatigue; LCF; Thermo-mechanical Fatigue; TMF; thermal fatigue; constitutive laws; behaviour modelling; numerical simulation; viscoplasticity.
ArcelorMittal, Global R&D Maizières Research, BP 30320, 57283 Maizières-lès-Metz Cedex, France
Abstract
The cyclic mechanical behavior, the wear and fatigue resistances and damage developments of working surface oftool steels are dependent on microstructural features. A multi-scale approach combining experimental testing, numerical treat- ments and simulations is developed to model the surface behavior ofX38CrMoV5-1 martensitic tool steels. The multi-scale modeling is coupled with finite element calculations. The elasto-viscoplastic constitutive equations used are based on crys- tal plasticity model of Méric-Cailletaud and are implemented on the finite element code ABAQUS under a small strain assumption. Trough an appropriate laboratory testing, the microstructure features comparable to the surface of industrial tools or pin/disc in tribology experiments are reproduced by considering plate specimens. Monotonic tensile testing is coupled with in-situ Digital Image Correlation tech- nique (DIC) to determine the surface strain fields. The measured local nonlinear mechanical strain fields are analyzed. The strain localization is related to stereolog- ical artifacts. The numerical treatments allow reproducing, qualitatively, the strain localization patterns at the surface observed during tensile testing. The influence of the various stereological parameters such as the morphology of martensitic laths, the crystallographic orientations, the internal hardening state of the surface profiles and their evolutions on the local strain fields are addressed. By such approach, it is possible to get a better insight of some elementary mechanisms acting on tools and/or pin/disc surfaces regarding both tensile and cyclic behavior.
Another possible impurity of concern is hydrogen which is known to induce hydrogen embrittlement phenomenon. The only possible source of hydrogen in our setup is the decomposition below saturation into dissolved oxygen and hydrogen of sodium hydroxide formed with the water vapor in flux. The contamination flux from the residual humidity supply at the sodium free level can be estimated using the collision rate from the gas kinetic theory [ 18 ]. Given that the water vapor content is constant during the test (equals to 1 ppm) then the free surface of liquid sodium wetting the notch (~8.7 mm 2 ) is collided by 3 10 5 mol of water molecules per hour. In our calculation, the geometrically constrained weight of the liquid metal droplet taken into account is estimated to lie be- tween 0.01 g and 0.1 g. If one supposes that every molecule reacts with sodium (a conservative hypothesis), this leads to an increase in the hydrogen content from 0.3 to 6 wppm (433 K e533 K in saturation). According to these assessments, considering that the average durations of the tests are 1 h, 0.5 h and 0.3 h at respectively 573 K, 623 and 673 K, the uptake of hydrogen remains therefore limited or negligible. Moreover, hydrogen embrittlement is known to occur only below roomtemperature with austenitic steels [ 19 ]. The reduction in mechanical properties observed in this study is more obvious at 623 K, i.e. more than 300 K above the temperatureat which the known weakening effects of hydrogen solutes disap- pear with austenitic steels.
damage of low-Si tempered martensitic hot work toolsteelX38CrMoV5 (AISI H13)”, Proceedings of 26th Journées de Printemps SF2M « Fatigue sous sollicitations thermiques », Paris, (2007)
[8] Z.W. Chen, D.T. Fraser, M.Z. Jahedi, “Structure of intermetallic phases formed during immersion of H13 toolsteelin an Al-11Si-3Cu die casting alloy melt”, Material Science and Engineering Forum, Vol A260, (1999), 188-196
(a) (b)
Fig. 4. Fatigue crack growth diagram for tests in air, vacuum and hydrogen environment at R=0.1 and 20Hz; (a) da/dN vs 'K; (b) da/dN vs 'K eff
after crack closure correction.
SEM observations of the fracture surface in hydrogen in the near threshold regime (Fig. 5a) appear very similar to that in air (Fig. 3a) with a transgranular cleavage-like fracture mode typical of a stage II propagation with a crack plane normal to the load axis and localized intergranular facets again in accordance with previous observations on the same alloy comparatively tested in hydrogen and ambient air [11]. On the basis of the similitude ofcrack growth data andof the surface morphology, the same water vapor adsorption assisted stage II propagation mechanism is assumed to be operative in both gaseous environments. This assumption is supported by the presence in the 4 bar