arXiv subjects

Andrew Cahaly

Publications and source records attributed to Andrew Cahaly.

3 recordsLinked to original sources

arXiv2025

PCIC: Cylindrical Volume Moment Calculation and Interface Reconstruction for Sub-Grid Modeling in Volume of Fluid Methods

The accurate modeling of topology changes remains a significant challenge in geometric Volume of Fluid (VOF) simulations. When using traditional single-plane reconstruction (PLIC), fluid structures smaller than the mesh size cannot be resolved and spurious numerical breakup is triggered; this impacts important flow statistics such as drop size distributions. Recent advances have introduced paraboloid and two-plane reconstructions, which have improved high-curvature performance and enabled sub-grid film reconstructions, respectively. However, sub-grid ligament reconstructions have remained elusive. In this work, a novel cylindrical interface reconstruction strategy called PCIC is introduced for sub-grid ligament modeling. PCIC is facilitated by deriving the analytical volume moments of quadratic cylinders clipping polyhedra; this allows for exact mass conservation during volume moment transport. From the transported moments, a straight circular cylinder can be defined in the center cell of a 5x5x5 stencil. First, a quadratic principal curve is fitted to the normalized first-order moments in the stencil (the liquid barycenters), from which the cylinder's orientation and origin are approximated. The cylinder radius is then chosen to conserve volume. On-the-fly ligament detection is achieved using connected-component labeling and moments of inertia criteria, which allows for simulations to automatically choose between PLIC and PCIC in each interface cell at runtime. PCIC is demonstrated in multiphase flow test cases, where it exhibits robust reconstruction of sub-grid ligaments. This allows for relatively low-resolution PCIC simulations to provide comparable results to traditional high-resolution simulations.

math.MG↗

arXiv2024

PLIC-Net: A Machine Learning Approach for 3D Interface Reconstruction in Volume of Fluid Methods

The accurate reconstruction of immiscible fluid-fluid interfaces from the volume fraction field is a critical component of geometric Volume of Fluid (VOF) methods. A common strategy is the Piecewise Linear Interface Calculation (PLIC), which fits a plane in each mixed-phase computational cell. However, recent work goes beyond PLIC by using two planes or even a paraboloid. To select such planes or paraboloids, complex optimization algorithms as well as carefully crafted heuristics are necessary. Yet, the potential exists for a well-trained machine learning model to efficiently provide broadly applicable solutions to the interface reconstruction problem at lower costs. In this work, the viability of a machine learning approach is demonstrated in the context of a single plane reconstruction. A feed-forward deep neural network is used to predict the normal vector of a PLIC plane given volume fraction and phasic barycenter data in a $3\times3\times3$ stencil. The PLIC plane is then translated in its cell to ensure exact volume conservation. Our proposed neural network PLIC reconstruction (PLIC-Net) is equivariant to reflections about the Cartesian planes. Training data is analytically generated with $\mathcal{O}(10^6)$ randomized paraboloid surfaces, which allows for sampling a wide range of interface shapes. PLIC-Net is tested in multiphase flow simulations where it is compared to standard (E)LVIRA reconstruction algorithms, and the impact of training data statistics on PLIC-Net's performance is also explored. It is found that PLIC-Net greatly limits the formation of spurious planes and generates cleaner numerical break-up of the interface. Additionally, the computational cost of PLIC-Net is lower than that of (E)LVIRA. These results establish that machine learning is a viable approach to VOF interface reconstruction and is superior to current reconstruction algorithms for some cases.

physics.comp-ph↗

arXiv2022

The Periodic Signals of Nova V1674 Herculis (2021)

We present time-series photometry during eruption of the extremely fast nova V1674 Herculis (Nova Her 2021). The 2021 light curve showed periodic signals at 0.152921(3) d and 501.486(5) s, which we interpret as respectively the orbital and white dwarf spin-periods in the underlying binary. We also detected a sideband signal at the /difference/ frequency between these two clocks. During the first 15 days of outburst, the spin-period appears to have increased by 0.014(1)%. This increase probably arose from the sudden loss of high-angular-momentum gas ("the nova explosion") from the rotating, magnetic white dwarf. Both periodic signals appeared remarkably early in the outburst, which we attribute to the extreme speed with which the nova evolved (and became transparent to radiation from the inner binary). After that very fast initial increase of ~71 ms, the spin-period commenced a steady decrease of ~160 ms/year -- about 100x faster than usually seen in intermediate polars. This is probably due to high accretion torques from very high mass-transfer rates, which might be common when low-mass donor stars are strongly irradiated by a nova outburst.

astro-ph.SR↗