The attractive mechanical properties of nickel-based superalloys primarily arise from an assembly of γ′ precipitates with desirable size, volume fraction, morphology and spatial distribution. In addition, the solutioning cooling rate after super solvus heat treatment is critical for controlling the features of γ′ precipitates. However, the correlation between these multi-dimensional parameters and mechanical strength has not been established to date. To efficiently construct the correlation among the heat treatment process, multi-dimensional microstructure parameters and mechanical properties, utilizing scanning electron microscope (SEM) and microhardness tester with automated capabilities for characterization, the gradient cooling sample in a high-throughput manner can be analysed, the heat treatment cooling rate, microstructure, and microhardness data can be efficiently obtained.

In this study, hundreds of SEM images with different γ′ precipitates were investigated. The SEM images of both nickel-based superalloys before-aging and aged nickel-based superalloys were studied simultaneously. First, mean-shift algorithm was used to segment SEM images and C fuzzy algorithm classified the different grayscales to obtain the geometric features of γ′ precipitates. Secondly, linear regression fitting algorithm was applied to reduce the dimensionality of the input features, and sort out the importance of each parameter on hardness. Thirdly, an ANN model based on back propagation algorithm was trained with the inputs of a set of normalized parameters, and one output variable of microhardness values. The input data generating from SEM image processing was classified into both the training set and test set. The training set was used to obtain the predictive model, the test set is used to validate the model accuracy.

In order to improve the fidelity of ANN model, the statistical characteristics of γ′ precipitates were added into the training set as the second training data set. In the third data set, cooling rate was further added into the training set, which may contain information of secondary γ′ precipitates. As more types of parameters were fed into the model, the accuracy of prediction of both aged nickel-based superalloys and nickel-based superalloys before aging got increased gradually. With the correlation model of the microstructural features, cooling rate and hardness of nickel-based superalloys, it can be seen that when cooling rate was increasing, the average area of the γ′ precipitates was increasing significantly. Meanwhile, the hardness of the aged superalloys was trending inversely with the increase of the precipitate size. The controlling variables in this ANN model agrees well with the internal variables of classic dislocation models of nickel-based superalloys. Overall, this methodology is able to build a nonlinear relationship between microstructural features and mechanical properties under the inspiration of material genome spirit with proper supervised guidance.


Bridging Precipitate Microstructure and Mechanical Hardness of Nickel-Based Superalloy Through Artificial Neural Network Model

Division / Sponsors: 
Aerospace, Bioengineering, Applied Mechanics, Design Engineering Division, Advanced Energy Systems, Manufacturing Engineering Division, Noise Control and Acoustics, Nondestructive Evaluation, Diagnosis, and Prognosis (NDPD), Nuclear Engineering Division, Process Industries Division, Fluids Engineering, Heat Transfer, Materials, Microelectromechanical Systems, Safety Engineering and Risk Analysis, Computers & Information in Engineering


Paper Proceedings: https://asmedigitalcollection.asme.org/IMECE/IMECE2020/volume/84478

IMECE2020 - Virtual Conference

Supporting the need among researchers worldwide for additional online and multimedia content, ASME now offers the ASME Conference Video Collection. This platform will collect and provide long-term access to conference presentation videos, encompassing all technology areas within ASME's technical conference and events program. About 30,000-40,000 engineers across all mechanical engineering and related disciplines participate in ASME conferences each year. This includes authors of approximately 6,000 to 8,000 technical papers published by ASME, the majority of whom also submit video presentations for our virtual conferences. The video collection will be available as an add-on to existing product packages or as a separate subscription.