1 July 2025 in Physics of Fluids
In this Letter, a physics-based data-driven strategy is developed to predict vortex-induced drag on a circular cylinder under non-uniform inflow conditions—a prevalent issue for engineering applications at moderate Reynolds numbers. Traditional pressure-signal-based models exhibit limitations due to complex vortex dynamics coupled with non-uniform inflow. To address this issue, a modified fully connected neural network architecture is established that integrates upstream velocity measurements (serving as an inflow calibration) with pressure-signal-based inputs to enhance predictive capability (R2∼0→0.75). Direct numerical simulations at Reynolds number Re=4000 are implemented for model training and validation. Iterative optimizations are conducted to derive optimized input configurations of pressure sensor placements and velocity components at upstream locations. The optimized model achieves an R2 score of 0.75 in forecasting high-amplitude drag coefficient fluctuations (Cd=0.2–1.2) within a future time window of one time unit. An exponential scaling between model performance and optimized pressure signal inputs is observed, and the predictive capability of sparsely distributed but optimized sensors is interpreted by the scaling. The optimized sensor placements correspond to the physical mechanism wherein flow separation dynamics play a governing role in vortex-induced drag generation. This work advances machine learning applications in fluid–structure interaction systems, offering a scalable strategy for forecasting statistics in turbulent flows under real-world engineering conditions.
7 April 2026 in Engineering Computations
Purpose: Generating watertight and orientable surfaces from 3D inputs such as triangle soups presents several challenges. In many cases, triangle- and vertex-normal information is required to construct a consistent surface, yet reliable orientation cues are often unavailable. For models containing internal structures, preserving these interior regions is particularly difficult. Design/methodology/approach: To address these issues, we present an automatic method that generates watertight, orientable manifolds for complex inputs. We first construct the model's outer surface and retain the tetrahedral mesh to enable subsequent inner wrapping. To build the inner wrapping surface, we introduce a layer-concealment process that mitigates interference from the outer structure. We then restart the wrapping procedure with additional Steiner points inserted around the inner region. Finally, a global carving step extracts both the outer and inner surfaces. Findings: We evaluate our method on a variety of complex models, and the results demonstrate robust performance. Originality/value: Given a triangle soup with possible inner structure, we present an automatic and robust algorithm to generate proper watertight and manifold surface mesh. We tend to deal with the complex defective input and provide surface meshes of different regions for subsequent computations. We generate the menbrane-like layer around input geometry and thereby extract these wrapping surfaces from both inside and outside.