PSI - Issue 80

Francisco J.G. de Oliveira et al. / Procedia Structural Integrity 80 (2026) 1–10 Author name / Structural Integrity Procedia 00 (2023) 000–000

3

3

a)

b)

c)

x

0

Fixed Root

Distributed fibre-optic sensing

U ∞

y

Kármán vortex street

7.28 D 1.6D 1.6D

C2

C1

U ∞

A

x

z

Downwash

B

C1

C2

Turbulence generating grids

~3 D

~7 D

Tip Vortices

Fig. 1. a) : Schematic of the experimental setup of the cylindrical body’s experiment. 3 PIV fields of view (FOV) were used, 2 (FOV A and B ) along the transverse plane, and FOV C along the spanwise extent of the cylinder. b) : fibre optic sensor array location along the surface of the cylinder. c) : Characterisation of the free-stream turbulence (FST) conditions used in the experiment, each case with di ff erent incoming turbulence intensity ( TI ) and integral length scale L / R . Adapted from de Oliveira et al. (2024, 2025).

a scaled wind turbine model, enabling extension of the methodology to a rotating system exposed to realistic inflow turbulence.

2.1. Cantilevered cylinder in turbulent cross-flow

Experiments were performed in the hydrodynamics flume of the Department of Aeronautics at Imperial College London. The facility has a constant cross-section of 58 × 60 , cm 2 , providing a uniform incoming stream. The test specimen was an acrylic cylinder of diameter D = 50 , mm and aspect ratio L / D ≈ 10, mounted as a cantilever with 95% of its span submerged. To suppress buoyancy e ff ects, the hollow cylinder was filled with water during operation. The cylinder root was rigidly clamped to the top frame of the flume, and its free end was located approximately 80 , mm above the floor, allowing the formation of tip vortices and their advection downstream. The cylinder surface was instrumented with single-mode fibre optics bonded using cyanoacrylate glue. Four sens ing lines were applied at polar positions θ f = − 45 ◦ , − 135 ◦ , 45 ◦ , 135 ◦ , covering both the windward and leeward faces of the structure. Each line comprised 140 sensing points with a spatial resolution of ∆ y ≈ 2 . 6 , mm, equivalent to ∆ y / D ≈ 0 . 05, yielding distributed strain fields along an e ff ective span of L i ≈ 7 . 28 D . Strain measurements were ac quired with a LUNA ODiSI-B system based on optical frequency-domain reflectometry (OFDR), exploiting Rayleigh backscattering to achieve high-density spatial coverage (Xu and Sharif-Khodaei, 2020; Li and Sharif-Khodaei, 2025). The acquisition frequency was 50 , Hz, with a low-pass filter applied at 15 , Hz, above the dominant vortex shedding frequency of St ≈ 0 . 2, where St is a normalised frequency as St = fD / U ∞ . To characterise the surrounding flow, two-dimensional particle image velocimetry (2D-PIV) was employed. A high speed Nd:YLF laser operating at 1000 Hz illuminated seeded tracer particles, while two Phantom V640L cameras recorded image pairs at f acq = 50 , Hz, capturing approximately 200 vortex-shedding cycles per run. The PIV fields of view (FOVs) were positioned at y / D = 3 . 4 (midspan, regular vortex shedding region) and y / D = 7 . 4 (free end, tip-vortex region), with interrogation windows down to 16 × 16 px at 50% overlap, giving a spatial resolution of ∆ x / D ≈ 0 . 024. The alignment of fibre sensing lines and PIV fields allowed direct correlation between local flow structures and distributed strain. Free-stream turbulence (FST) conditions were generated using a family of passive grids placed upstream of the cylinder. By varying mesh size and distance from the test section, a parameter space of turbulence intensity ( TI = 1 . 3 − 14%) and transverse integral length scale ( L / D = 0 . 08 − 0 . 8) was explored. These conditions enabled isolation of the direct e ff ect of fluctuating inflow velocity from the indirect e ff ect of altered vortex shedding dynamics on the structural response. For more details on the setup, the reader is directed to de Oliveira et al. (2025).