Issue 78

V. I. Kolesikov et alii, Fracture and Structural Integrity, 78 (2026) 220-232; DOI: 10.3221/IGF-ESIS.78.13

In RVE reconstructions, two separate workflows in Dream 3D for columnar and equiaxial microstructure sub-RVEs, respectively, were developed. Then, these sub-RVEs were used in Dream 3D to combine them into multilayer RVEs studied in this work. Custom scripting to merge all grain data in such multilayer RVEs was necessary.

R ESULTS AND DISCUSSION CP-simulation and RVE reconstruction

F

ig. 1a shows a schematic representation of the simulated perfectly bonded layered composite under application of a vertical load, represented by the Piola-Kirchhoff stress tensor ( ത ). The former is subjected to a horizontal deformation gradient rate represented by the corresponding tensor ( ത̇ ).

Axes and crystal ODFs for phase with columnar microstructure

a)

c)

X-axis

Load

Y-axis

Z-axis

Z

RVE

Deformation gradient rate

X

b)

Axes and crystal ODFs for phase with equiaxial microstructure

X RVE with 32 × 32 × 64 pixels, random crystal orientation and periodic boundaries

X-axis

Z-axis

Y-axis

Phase with columnar microstructure

Phase with equiaxial microstructure

Phase with columnar microstructure

Figure 1: a) schematic representation of geometrical model under simulation: the perfectly bonded layered composite under vertical load subjected to horizontal deformation gradient rate (the DAMASK based simulation). The green line indicates the interface between mated materials, while red line marks the position of tribological contact (not modeled here) b) The RVE general structure composed from two sets of different microstructures (phases). c) The ODFs for two RVE microstructure (phases), illustrating their spatial and crystal orientations. The RVE geometry in the DAMASK simulation is defined in SI units (meters), but the absolute physical size of the RVE is arbitrary because the spectral solver with homogeneous boundary conditions is scale ‑ invariant. The solution is affected by the relative dimensions of microstructural features (grain shapes, phase fractions, topology) only. Therefore, the RVE length is treated as a normalized (dimensionless) computational size rather than a physical length. The simulations were run for t =144 s using the DAMASK Spectral Solver under the prescribed mixed boundary conditions of the homogeneous type represented by Eqn. (1):

   

   

−

   

    

3

0 0 10 0 0 0 0 0 *

8 * * * * * * * * 10

˙

= 

F

= 

P

s-1 and

Pa.

(1)

− 

222