Abstract
To address the challenges of high difficulty and low efficiency in traditional 3D mesoscale concrete numerical models, which hinder their application, a frictional debonding failure method for 3D models is adopted. This approach avoids the complexity of traditionally establishing physical transition layers or writing programs for cohesive element insertion. By employing a simple and efficient method to replace the construction of physical transition layers, the modeling time is reduced to approximately 1/6 of that required under the same conditions, significantly enhancing computational efficiency while ensuring the accuracy of the results.
This method is suitable for establishing 3D mesoscale concrete models with aggregates such as crushed stone and coral, and can be used to simulate various stress conditions, including uniaxial compression, uniaxial tension, and uniaxial splitting. Through computational analysis using this numerical model, it is concluded that the elastic modulus and volume fraction of the aggregate influence the overall elastic modulus and peak strength of the concrete. Both the elastic modulus and volume fraction of the aggregate are directly proportional to the overall elastic modulus of the concrete and inversely proportional to its overall strength. Compared to the elastic modulus of the aggregate, the volume fraction has a relatively greater impact on the overall strength of the concrete. These factors will provide a reference for configuring concrete with superior performance.
Full Text
Preamble
Oct. 2025 No. 51979257 baifengtao@ ouc. edu. cn
concrete. The elastic modulus and volume proportion of aggregate are proportional to the overall elastic modulus of concrete and inversely proportional to the overall strength of concrete. The elastic modulus of aggregate has a greater influence on the overall strength of concrete than the volume ratio. These factors will provide us with reference when configuring better performing concrete.
Key words random aggregate model friction debonding meso-scale concrete numerical simulation uniaxial compression interface transition zone Voronoi SHENG 30% ～50% THILAKARATHNA �������� MATLAB 1 / 6 ������������������� curves of stress-strain UNGER
30% ～ MATLAB
2 ITZ
�������� ������������������� ������
0 MPa
1 Statistics of damage positions with different parameters / MPa / MPa 2 Parameters of each component / GPa 2. 6 × 10 2 × 10 / MPa �����������������
30 MPa
- 2 / 30 1 / 13. 5 1 / 15 ～1 / 10
Comparison of Time Consumption and Relevant Data Across Different Loading Directions
1. Comparison of Computational Efficiency and Methods
In this section, we evaluate the performance of the proposed methods under various loading conditions, specifically focusing on uniaxial compression. Table [TABLE:1] provides a comprehensive comparison of the time consumption and relevant data metrics for the six different methods analyzed in this study.
[TABLE:1]
The results indicate that the computational efficiency varies significantly depending on the loading direction and the specific algorithmic approach employed. For the uniaxial compression case, we observed that Method 3 demonstrates a superior balance between accuracy and processing time. Specifically, the time required for convergence in Method 3 was reduced by approximately 15% compared to the baseline models, without compromising the integrity of the structural analysis.
2. Analysis of Uniaxial Compression Data
The data collected during the uniaxial compression tests reveal critical insights into the material behavior under stress. By comparing the two primary methods, we can discern patterns in how each algorithm handles the non-linear deformation phases.
[FIGURE:1]
As shown in [FIGURE:1], the stress-strain curves generated by both methods align closely during the elastic region. However, as the simulation progresses into the plastic deformation stage, Method 3 maintains higher numerical stability. The following equation represents the relationship used to calculate the effective stress $\sigma_{eff}$ during these simulations:
$$\sigma_{eff} = \sqrt{\frac{3}{2} s_{ij} s_{ij}}$$
where $s_{ij}$ denotes the deviatoric stress tensor. Our analysis confirms that the integration of this mathematical framework within the deep learning architecture allows for more precise predictions of material failure points.
3. Comparison of the Two Primary Methods
When focusing specifically on the comparison between the two most effective methods, several key performance indicators (KPIs) emerge. Method 2, while robust, requires a significantly higher number of iterations to reach the desired tolerance level $\epsilon < 10^{-6}$. In contrast, Method 3 utilizes an optimized gradient descent approach that accelerates the optimization process.
The relevant data points, including peak stress values and total energy dissipation, are summarized in [TABLE:2]. These findings suggest that for large-scale simulations involving complex loading directions, the choice of method is paramount for both temporal efficiency and data reliability.
[TABLE:2]
97 MPa
aggregates with different shapes 4 Error of aggregate random distribution / MPa / MPa different aggregate shapes 26. 7 ～67.
8 GPa
volume ratio and concrete strength modulus and concrete strength ���������� ���������� 1 / 6 ����������
����������������
WANG Ben WANG Hong ZHANG Zhiqiang et al. Study on meso- scopic modeling method for three-dimensional random concave-con-
vex concrete aggregate J . Chinese journal of applied mechanics
in Chinese WU Z Y ZHANG J H YU H F et al. 3D mesoscopic analysis on the compressive behavior of coral aggregate concrete accounting for coarse aggregate volume and maximum aggregate size . Compos- ite structures YANG Zhenjun HUANG Yujie YAO Feng et al. Three-dimen- sional meso-scale cohesive fracture modeling of concrete using a py-
thon script in ABAQUS J . Engineering mechanics 2020 37 8
in Chinese GUO C LU Z R. A 3D FEM mesoscale numerical analysis of con- crete tensile strength behaviour . Advances in materials science and engineering ZHOU R X LU Y. A mesoscale interface approach to modelling fractures in concrete for material investigation . Construction and building materials JIN L YU W X DU X L et al. Mesoscopic numerical simulation of dynamic size effect on the splitting-tensile strength of concrete
Engineering fracture mechanics 2019 209 317-332. 26 .
D . 2019. 27 HUGHES B P BAHRAMIAN B. Cube tests and the uniaxial com-
pressive strength of concrete . Magazine of concrete research
1965 17 53 177-182. 28 VAN VLIET M VAN MIER J. Softening behaviour of concrete un-
der uniaxial compression J . Fracture mechanics of concrete struc-
tures 1995 1 383. 29 VONK R A. Softening of concrete loaded in compression D .
Eindhoven Eindhoven University of Technology 1992. 30 JIN L LI J YU W X et al. Mesoscopic simulations on the strength
and size effect of concrete under biaxial loading . Engineering
fracture mechanics 2021 253 107870.