Resumen en Inglés:

Abstract This paper proposes an adaptive enhanced proportional topology optimization algorithm (AEPTO) to address the problems of poor convergence, multiple grayscale units, and unclear topology structure in handling stress constrained minimum volume problems using the proportional topology optimization algorithm (PTO). This algorithm systematically improves the PTO algorithm in terms of interpolation method, filtering processing, density update, and convergence control by introducing a new material interpolation model based on logistic function, Cauchy type density filter, arctangent density update strategy, and dynamic adaptive historical balance coefficient. Through numerical experiments on topology optimization of three typical structures, namely cantilever beams, L-shaped beams, and MBB beams, under stress constraints, it is shown that the AEPTO algorithm significantly reduces the volume fraction of the structure, improves convergence speed, reduces the number of gray units, and the optimized topology structure is clearer and the material distribution is more reasonable. It has good engineering applicability and optimization efficiency, providing an effective gradient free optimization method for lightweight structural design.

Resumen en Inglés:

Abstract This study presents a standards-oriented shape-optimization framework for modifying the front-end geometry of an existing tram. A finite element model was developed for the CEN/TR 17420 Type A scenario at 20 km/h using Hybrid III and VIVA+ AM50 pedestrian models. Although the baseline geometry satisfied the HIC15 requirement, lateral deflection at the 50% offset was only 582.62 mm for Hybrid III and 321.58 mm for VIVA+. Six front-end shape variables were parameterized by mesh morphing and optimized using the Global Response Surface Method (GRSM), with lateral deflection maximized subject to HIC15 and design-variable constraints. One selected design from each of the six optimization cases satisfied both Type A collision-response criteria. Across these designs, lateral deflection ranged from 895.25 to 1,284.46 mm, while HIC15 ranged from 7.36 to 711.49. The results show that mesh morphing and GRSM can support retrofit-oriented tram-front design by quantifying the trade-off among lateral deflection, HIC15, and the allowable extent of front-end modification.

Resumen en Inglés:

Abstract To further analyze the mechanical properties of random porous materials under combined loading, this paper constructs a series of random honeycomb models based on Voronoi diagrams and 3D printing. A compression-shear combined loading fixture was designed. Experiments on random honeycomb materials were conducted through the universal testing machine and the proposed fixture. The results show that reducing the cell size decreases both normal and shear stresses when relative density is consistent, while increasing the loading angle increases shear stress and decreases normal stress. Furthermore, to explore the dynamic mechanical properties, the finite element model is established. The results show that random honeycomb materials exhibit three deformation modes: random mode, transitional mode, and impact mode. Then, the sensitivity analysis of the loading angles and velocities on the yield stress is carried out. The normal stress is mainly affected by the loading velocities, and both the loading angles and velocities have significant effect on the shear stress. Finally, a phenomenological yielding criterion is established.

Resumen en Inglés:

Abstract This study evaluates and improves models for predicting the high-velocity penetration response of ultra-high-performance concrete (UHPC) used in protective structures. The research scope involves systematically comparing three common concrete constitutive models: Holmquist–Johnson–Cook (HJC), Riedel–Hiermaier–Thoma (RHT), and Karagozian & Case (K&C). First, the models' theoretical differences regarding strain-rate effects and tension-compression asymmetry were analyzed. Next, their material parameters were calibrated using fundamental mechanical tests and literature data. An LS-DYNA finite element model was then established and validated against ballistic experiments (633-959 m/s) by evaluating penetration depth, surface crater morphology, and internal damage evolution. The results indicate that the HJC model is the optimal choice for high-velocity penetration simulations. It yielded the lowest average depth error (-11.89%), and its localized damage pattern best matched UHPC's high-strength and high-toughness traits. In contrast, the RHT model caused premature matrix softening under high hydrostatic pressure, leading to a -21.76% error. The K&C model overpredicted surface funnel-shaped crater damage, with its penetration depth error increasing to -26.34% at high velocities. Furthermore, extended simulations at 800–1500 m/s revealed that the traditional Forrestal analytical model deviates significantly in the ultra-high-velocity regime. To solve this, a modified analytical model tailored for UHPC was developed. By accounting for dynamic yield and high-pressure shear rheological softening mechanisms, the new model introduces a dynamic strengthening coefficient and a rheological softening factor. Ballistic validation demonstrates that, within the validated range, the theoretical predictions of this modified model match the experimental data well, with a maximum relative error of 14.54%. Overall, this research provides reliable numerical model selection and theoretical support for designing and evaluating UHPC protective structures against kinetic energy projectile impacts.