본문 바로가기

Investigation of viscous effects on a rectangular barge undergoing roll and heave motions in regular waves: A physics-informed neural network approach

문서 유형
연구출판물 · 학술지 논문
등재
SCIEScopus
저널 순위
JCR Q1Scopus Q1
저널
PHYSICS OF FLUIDS37(12) · #127143 · 2025
피인용
Web of Science 1Scopus 1
FWCI (Scopus)
0.22
JCR IF
JCR 2024 4.3
ISSN
1070-6631

저자

  1. 즈엉 띠엔 쭝제1저자Pusan Natl Univ, Dept Naval Architecture & Ocean Engn, Busan 46241, South Korea
  2. Park, Sung BooSamsung Heavy Ind, Ship & Offshore Performance Res Ctr, Daejeon 305380, South Korea
  3. 정광효교신저자Pusan Natl Univ, Dept Naval Architecture & Ocean Engn, Busan 46241, South Korea

초록

This study examines viscous effects on a rectangular barge undergoing coupled roll and heave motions in regular waves by reconstructing flow and pressure fields using a physics-informed neural network (PINN) framework based on particle image velocimetry (PIV) data. Three neural network models were developed to infer the water volume fraction, velocity, and pressure. The velocity model employed a stream-function formulation to inherently satisfy the continuity equation, while the pressure model incorporated the Navier-Stokes equations with a zero-gauge pressure boundary condition in the air phase. The models were fine-tuned using the limited-memory Broyden-Fletcher-Goldfarb-Shanno (L-BFGS) optimizer following initial training with Adam. The PINN-reconstructed velocity and vorticity fields closely matched experimental measurements, confirming the reliability of the approach. The results show that vortices generated near the barge coincide with local reductions in dynamic pressure, and that increasing wave height intensifies vortex strength, resulting in larger viscous effects without altering the overall vortex evolution. Under waves at the natural roll period, higher wave heights primarily increased roll damping, explaining the nonlinear roll response observed experimentally. These findings provide insight into vortex-driven mechanisms of viscous roll damping and enhancement and demonstrate the capability of PINNs for high-fidelity flow reconstruction in complex wave-structure interactions.

키워드

FLOW

식별자

Scopus EID
2-s2.0-105025713629
WoS UID
001643448200017