K. Wu
Please Note
5 records found
1
Existing frequency-domain topology optimization methods typically exploit the discrete cosine transform (DCT) and discrete Fourier transform (DFT) mainly for dimensionality reduction; the capacity of spectral representations to encode periodic patterns remains untapped, and the reconstructed density fields often lack sharp, manufacturable boundaries. To overcome both limitations, this paper proposes a topology optimization framework integrating frequency-domain periodic regulation (FDPR) with the ODE-driven boundary penalization (ODE-BP) method, in which the DCT/DFT coefficients of the level-set function serve directly as design variables. A quantitative relationship is derived between the periods of the retained basis functions and the spatial repetition patterns of the reconstructed field, and a four-parameter spectral mask (target period, detuning onset, detuning bandwidth, and spectral cutoff) is introduced, which recovers strictly periodic, quasi-periodic and classical non-periodic designs. FDPR imposes (quasi-)periodicity on curved shells directly in parameter space without extra constraints, thereby avoiding mesh-matching difficulties associated with many homogenization-based approaches. Owing to the linearity of the DCT/DFT, level-set evolution in the full frequency domain is proved equivalent to spatial-domain evolution, whereas coupling the spectral parameterization with nonlinear optimizers breaks this equivalence and opens up solution spaces that are difficult to reach with conventional schemes. Two- and three-dimensional examples, covering static compliance, heat conduction, and dynamic problems; single- and bi-material designs; and strictly periodic, quasi-periodic, multi-zone, and triply periodic minimal surface (TPMS) configurations, demonstrate the effectiveness, flexibility, and clear boundary definition of the method, providing a theoretical and algorithmic foundation for (quasi-)periodic lattice design.
Residual stresses and distortions are major barriers to the broader adoption of wire arc additive manufacturing. These issues are coupled and arise due to large thermal gradients and phase transformations during the directed energy deposition process. Mitigating distortions may lead to substantial residual stresses, causing cracks in the fabricated components. In this paper, we propose a novel method to reduce both residual stresses and distortions by optimizing the fabrication sequence. This approach explores the use of non-planar layers, leveraging the increased manufacturing flexibility provided by robotic arms. Additionally, our method allows for the concurrent optimization of the structural layout and corresponding fabrication sequence. We employ the inherent strain method as a simplified process simulation model to predict residual stresses and distortions. Local residual stresses are aggregated using a p-norm function, which is integrated into distortion minimization as a constraint. Through numerical examples, we demonstrate that the optimized non-planar fabrication strategies can effectively reduce both residual stresses and distortions.
In this framework, a pseudo-density field defines the structural layout, while a pseudo-time field encode the fabrication sequence, offering detailed insights into the layer-by-layer manufacturing process. Two key advancements are developed in order to improve the manufacturability: First, a thermal regularization method is proposed to ensure a smooth and continuous pseudo-time field without local minima. Second, a layer geometry control scheme is implemented to improve the consistency of the layer dimensions.
The framework is further adopted to address challenges associated with residual stress and thermal-induced distortion in WAAM, employing the inherent strain method as a simplified process simulation model. In addition, the anisotropic nature of material properties in WAAM is considered, enabling the rational alignment of material deposition orientation to enhance performance.
Numerical results demonstrate the feasibility and effectiveness of the proposed optimization framework, inspiring the exploration of the innovative potential offered by multi-axis additive manufacturing. ...
In this framework, a pseudo-density field defines the structural layout, while a pseudo-time field encode the fabrication sequence, offering detailed insights into the layer-by-layer manufacturing process. Two key advancements are developed in order to improve the manufacturability: First, a thermal regularization method is proposed to ensure a smooth and continuous pseudo-time field without local minima. Second, a layer geometry control scheme is implemented to improve the consistency of the layer dimensions.
The framework is further adopted to address challenges associated with residual stress and thermal-induced distortion in WAAM, employing the inherent strain method as a simplified process simulation model. In addition, the anisotropic nature of material properties in WAAM is considered, enabling the rational alignment of material deposition orientation to enhance performance.
Numerical results demonstrate the feasibility and effectiveness of the proposed optimization framework, inspiring the exploration of the innovative potential offered by multi-axis additive manufacturing.
Wire and Arc Additive Manufacturing (WAAM) has great potential for efficiently producing large metallic components. However, like other additive manufacturing techniques, materials processed by WAAM exhibit anisotropic properties. Assuming isotropic material properties in design optimization thus leads to less efficient material utilization. Instead of viewing WAAM-induced material anisotropy as a limitation, we consider it an opportunity to improve structural performance. This requires the integration of process planning into structural design. In this paper, we propose a novel method to utilize material anisotropy to enhance the performance of structures both during fabrication and in their use. Our approach is based on space–time topology optimization, which simultaneously optimizes the structural layout and the fabrication sequence. To model material anisotropy in space–time topology optimization, we derive the material deposition direction from the gradient of the pseudo-time field, which encodes the fabrication sequence. Numerical results demonstrate that leveraging material anisotropy effectively improves the performance of intermediate structures during fabrication as well as the overall structure.