Table 1 A list of features used to train BiLSTM model for predicting Δa23

From: An interleaved physics-based deep-learning framework as a new cycle jumping approach for microstructurally small fatigue crack growth simulations

Feature

Description

mmax

Maximum of the Schmid factors evaluated across 12 slip systems

\(\mathop{\sum }\nolimits_{1}^{5}m\)

Sum of the five largest Schmid factors out of those evaluated across 12 slip systems

\({d}_{fs}^{x},{d}_{fs}^{y}\)

Nearest distance to the free surface in X and Y directions

a

Half-crack length

\(\left\vert \gamma \right\vert\)

Absolute angular position of CFP

ωavg

Disorientation angle

E[001]

Elastic modulus along the loading direction [001]

Γ

A binary variable indicating whether a CFP is at the intersection of a grain boundary

Δak−1

Crack extension at crack growth increment k − 1

\({D}_{5}^{{\rm{avg}}}\)

Fatigue indicator parameter extracted using the micromechanical fields at k = 0

\({\epsilon }_{{\rm{eq}}}^{{\rm{avg}}}\)

Equivalent plastic strain extracted using the micromechanical fields at k = 0

\({\sigma }_{{\rm{triax}}}^{{\rm{avg}}}\)

Stress triaxiality extracted using the micromechanical fields at k = 0

\({\epsilon }_{33}^{{\rm{avg}}}\)

Strain along Z-direction extracted using the micromechanical fields at k = 0

\(\nabla {\epsilon }_{33}^{{\rm{avg}}}\)

Strain gradient along Z-direction extracted using the micromechanical fields at k = 0