An artificial intelligence evaluation system using neural network was developed to upgrade the creep and creep-fatigue damage assessment methodology for heat resistant steels of fossil power plants through image analyses of EBSD(Electron BackScateer Diffraction pattern) maps. KAM(Kernel Average Misorientation) maps were obtained for creep and creep-fatigue damaged austenitic stainless steel SUS304HTB and the stratified data were manipulated to evaluate damage degree. The system consisted of an input layer, intermediate layers and an output layer. As the activation function, ReLU(Rectified Linear Unit) function was used for the intermediate layers and Softmax function was used for the output layer. The evaluation results of the proposed system were compared with the results of the conventional quantitative damage evaluation method(the master curve method). As a result, the estimated damage accuracy of the artificial intelligence evaluation system developed in this research was proved to make some improvement compared with the estimated damage accuracy using the conventional evaluation method. The introduction of the neural network is considered to be effective for evaluating even for insufficient number of experimental data. Thus machine learning methodology utilizing neural network is proved to have the potential of versatile data analysis method applicable to various sorts of metallographic investigation.

抄録全体を表示

The EBSD(Electron BackScatter Diffraction pattern) observations were conducted on unused and creep damaged materials of Mod.9Cr steel. GROD(Grain Reference Orientation Deviation) was used as the parameter to evaluate the grain inside distribution of deformation. The grain inside GROD distribution function was well approximated by the form of continuously distributed dislocation function. The coefficient A multiplied by grain size l was correlated with macroscopic creep strain and applied to evaluate strain distribution near creep void. Using A × l parameter, strains near creep voids were proved to be almost similar level of the macroscopic creep strains.

抄録全体を表示

EBSD (Electron Backscatter Diffraction pattern) observations were conducted and GROD (Grain Reference Orientation Deviation) distribution patterns were obtained by drawing intersection lines settled on GROD maps for creep damaged samples of austenitic stainless steel SUS304HTB for boiler tube use. GROD distribution patterns were well fitted by continuum dislocation distribution functions and the representing parameter A of the distribution functions showed liner relationship against macroscopic creep strain. Using this relationship, creep strains near creep voids were evaluated as the similar levels of macroscopic creep strain.

抄録全体を表示

Indentation technique is used as one of the evaluation methods of material on microscopic region. But this method shows the shallower the indenter presses, the harder the result of a measurement. This is called as size effect of hardness. The strain gradient theory is studied on this problem. In this study, observation by EBSD(Electron Back Scatter Diffraction) were done around several sized indentations and crystal orientations were measured. The direction of expanse of the plastic region was different and it was found that the slip direction affects it. And it was found that the relative plastic region size to the maximum indentation depth decreased when the maximum indentation depth became extremely small and when the tip of indenter was dull. Provided that the strain near the indentation is same, the strain gradient increase if the plastic deformation region decrease. So, many GN dislocations, which prevent atoms from moving, occur and material becomes harder.

抄録全体を表示

Creep damage and creep -fatigue damage were imposed to the austenite stainless steel SUS304HTB for boiler tube use. EBSD (Electron BackScatter Diffraction pattern) observations were conducted on the creep and the creep-fatigue damaged samples using GROD (Grain Reference Orientation Deviation) maps and Phase maps. Area averaged GROD: GRODave and area fraction of alpha phase f_a(%) were used as the evaluation parameter to creep strain and accumulated creep strainrange. GRODave showed linear relationship with those strains except at final rupture stage. fa showed similar linear relationship with those strains except at the final rupture stage and the initial stage. Proportional relationship was also observed between the equivalent diameter d_v obtained from the creep void area and the estimated creep strain value ε_<cg> obtained from GRODave as well as f_a. These results indicated the EBSD observation could be effective for quantitative evaluation of permanent strain and phase transformation during creep and creep-fatigue damage process.

抄録全体を表示

An artificial intelligence evaluation system using neural network was developed for upgrading the creep damage assessment methodology through image analysis of EBSD(Electron BackScateer Diffraction pattern) maps. KAM(Kernel Average Misorientation) maps were obtained for creep damaged austenitic stainless steel SUS 304HTB and the stratified data were manipulated as the representatives of damage degrees. The system consists of an input layer, intermediate layers and an output layer. As the activation function, ReLU(Rectified Linear Unit) function is used for the intermediate layers and Softmax function is used for the output layer. The evaluation results of the proposed system were compared with the results of the conventional quantitative damage evaluation method. As a result, the estimated damage accuracy of the artificial intelligence evaluation system developed in this research was proved to be improved by about 3.3% compared with the estimated damage accuracy using the conventional evaluation method. Furthermore, the accuracy was improved by about 6.7% after the optimization of the neural network compared with the conventional evaluation method. Moreover, it was proved this system had sufficient robustness through the check tests for the case of extremely missing EBSD image. Thus machine learning utilizing neural network was expected to be a potential method for versatile data analysis applicable to various sort of metallographic study.

抄録全体を表示

Creep-fatigue crack propagation tests of lead-free solder were conducted under various loading waveforms. The waveforms adapted in displacement-controlled tests were four triangular waveforms: pp waveform having fast loading-unloading, cc waveform having slow loading-unloading, cp waveform with slow loading followed by fast unloading, and pc waveform with fast loading followed by slow unloading, and three waveforms with hold time: cc-h waveform having a hold time under tension and compression, cp-h waveform having a hold time under tension, and pc-h waveform having a hold time under compression. In load-controlled conditions, three waveforms are adapted: pp, cc-h and cp-h. Microscopic observation using SEM and EBSD was conducted near the crack tip region and on fracture surfaces. Plenty of microcracks were observed near the main crack tip on the surfaces of specimens, while almost all of them disappeared after removing the surface layer of about 0.4mm. On the removed surfaces, microcracks were observed near the main crack tip for unsymmetrical waveforms, cp, pc, cp-h, while no microcracks for symmetrical waveforms, pp, cc, cc-h. The existence of microcracks is responsible for crack acceleration under unsymmetrical waveforms. EBSD observation showed the formation of subgrains within original grains of Sn and eutectic phases near the crack tip, and the grain size decreased with increasing crack propagation rates. The grain average of GROD also decreased with increasing crack propagation rate, while that of KAM was nearly constant without respect to the crack propagation rate. Striations, fragmentation, and intergranular fracture facets are three main features of creep-fatigue fracture surfaces. The features of striations were clear for the cases of pp, cc, pc, while vague for the other cases. The spacing of striations was nearly equal to the crack propagation rate, supporting the linear summation rule of creep and fatigue crack propagation rates. Plenty of intergranular facets were observed for cp, cp-h. Fragmentation was abundant for waveforms including creep contribution.

抄録全体を表示