Jinsheng (Frank) Lai 「赖近昇」

[P1] Mechanical Properties and Feasibility of GFRP from Decommissioned Large-Scale Wind-Turbine Blades for Wave Energy Converter: A Case Study.

Yan-Wen Li*, Jin-Sheng Lai, Bin-Zhen Zhou, Li Cheng. Polymers (JCR Q1), Published, 18(7), 892, 2026.

[Paper] [DOI]

[P2] Applications of Deep Reinforcement Learning in Optimization of Offshore Wind Power Operation and Maintenance.

Jin-Sheng Lai*. ICEPET, Accepted, 2026.

[Paper]

[P3] Structural Reuse of Decommissioned Wind-Turbine-Blade GFRP in Marine and Civil Infrastructure: Performance and Design of Hybrid Bonded-Bolted GFRP-to-Duplex Stainless Steel Joints.

Jin-Sheng Lai, Tian-Hui Fan, Li Cheng, Yan-Wen Li*.

Construction and Building Materials (JCR Q1), Required Reviews Completed, 2026.

High-Value Repurposing of Decommissioned Wind Blade GFRP: Materials, Joints, and Marine Applications

Master's Thesis, 2024-Present. Supervisor: Yan-Wen Li.

• Executed mechanical characterization of 100+ repurposed-blade GFRP specimens and derived engineering design values via ANOVA and eigenvalue analysis; an LCA showed 71.5% lower carbon footprint and 37.4% lower structural mass than an all-steel wave energy converter.
• Proposed position-specific repurposing strategies based on tip/mid/root material property variations along the blade.
• Independently designed and tested hybrid bonded-bolted GFRP-duplex stainless steel lap joints; identified the adhesive layer as the governing failure mode with bolts providing redundancy; validated numerically using Abaqus FEA; manuscript in preparation.
• Established hydrostatic equilibria, robustly screened the Type-II minimum retained tip length, and projected BEM-derived hydrodynamic coefficients onto a 7-DOF hinged raft-type two-body model. Built a one-way coupled time-domain-to-FEA workflow incorporating nonlinear hydrostatics, quadratic drag, PTO and mooring loads, and transferred synchronized critical load states for structural assessment.
• Flume experiment (in progress) to validate the hinged raft-type two-body WEC model.

Structural Modeling and Optimization of a 15MW CS-UHPC Composite Semi-Submersible Floating Wind Turbine Platform

Research Project, Jun. 20, 2026-Present. Supervisor: Yan-Wen Li.

Responsible for the numerical modeling and computational analysis of a 15MW CS-UHPC composite semi-submersible platform. Establishing a time-domain fully coupled workflow in which OpenFAST provides aero-servo-elastic tower-base loads and ANSYS AQWA resolves platform hydrodynamics, motions, and mooring response, with platform kinematics and tower-base loads exchanged at each time step. The resulting tower-base forces and moments, fairlead tensions, platform motions, and distributed wet-surface hydrodynamic pressures are transferred to ANSYS Mechanical for finite-element stress, buckling, and structural optimization studies of the composite floating system.

OpenFAST, ANSYS AQWA, ANSYS Mechanical, Aero-Hydro-Servo-Elastic Coupling, Finite Element Analysis, Structural Optimization.

Modeling and Intelligent Operation and Maintenance of Offshore Wind Systems

Self-initiated independent project, Jun. 2025-Mar. 2026.

Independently studied and implemented deep reinforcement learning from scratch, formulating offshore wind O&M as a Markov Decision Process with a hybrid state space (turbine health, environmental conditions, resource availability, and market signals) and a hybrid discrete-continuous action space. Trained and evaluated a DDPG agent on public and simulated data from a 20-turbine 5 MW offshore wind farm over a 3-year horizon: it outperformed DQN and PPO baselines in revenue, failure frequency, MTBF, and MTTR, and stayed robust under stressed conditions (wind-speed variance +20%, failure rate +15%). This work led to the ICEPET 2026 conference paper.

Python, PyTorch, DDPG, Markov decision process.

Environmental Sensitivity Analysis of a Floating Wind Turbine and Semi-submersible Foundation

Research Assistant, Oct.-Nov. 2024. A collaborative project between Prof. Yan-Wen Li and Prof. Chaohe Chen.

Contributed to numerical modeling and environmental sensitivity analysis in ANSYS AQWA for a floating wind turbine and semi-submersible foundation. Assessed tower-base and tower-top motion responses across variations in wave height, wind speed, wave period, and wave-current incidence angle. The analysis identified wave height as the dominant driver of horizontal motion amplitudes, whereas wave period exerted a stronger influence on heave response.

ANSYS AQWA, Hydrodynamic Simulation, Environmental Sensitivity Analysis.

Compressive Response of 3D-Printed 2D Metamaterial Reinforced Composites with Diversified Geometries

Research Assistant at Versatile Structure Lab, advised by Prof. Nan Hu, May. 2025-Aug. 2025.

Tested and evaluated the compressive behavior of 24 diverse 2D composite architectures, and proposed a comprehensive performance evaluation coefficient to screen for designs with superior specific energy absorption and specific strength.

3D Printing, Mechanical Testing.

Crack Recognition Based on Backpropagation Neural Networks

Coursework Project, advised by Liangwan Rong, Oct. 2024-Feb. 2025.

Engineered a Multi-Layer Perceptron neural network with optimized activation functions and cross-entropy loss with backpropagation for automated Structural Health Monitoring.

Python, Neural Networks.

Vibration Damping Performance Analysis of Graded Yield Metal Dampers in a Steel Frame Structure

Undergraduate Thesis, Nov. 2023-Mar. 2024. Supervisor: Prof. Yun Chen.

Investigated the energy dissipation mechanism of graded yield metal dampers in steel frame structures through modal decomposition response spectrum analysis and elastic time-history analysis across minor, moderate, and major earthquake scenarios.

SAP2000, Seismic Analysis.

Flexural Property and Mechanism of Coir-FRP Seawater and Sea-Sand Concrete for Artificial Fish Reefs

Research Assistant, Nov. 2022-May. 2023, advised by Prof. Fei Zhao.

Optimized coir-fiber-matrix compatibility through combined chemical modification, tested flexural strength, integrated a non-linear MATLAB prediction model, and used SEM analysis to study the fiber bridging toughening mechanism.

Lab Testing, MATLAB, SEM.