betaFeedback

WorksPhysics and Astronomy

Cable construction effects on the lateral forces and wakes of helically stranded cables under oblique inflow

Abstract

Helically stranded cables in aerial towing systems experience oblique inflow, but are often modeled as smooth cylinders, neglecting three-dimensional cross-stream coupling and surface texture. We perform large-eddy simulations and spectral proper orthogonal decomposition for a smooth cylinder and four stranded cables at four inclinations from 22.5∘22.5^\circ to 90∘90^\circ, all stationary, with Re≈1.0×104Re \approx 1.0 \times 10^4. Aerodynamic differences are governed mainly by cross-sectional symmetry and strand-scale groove depth. Force-source decomposition shows that surface pressure contributes 94.9%–98.3% of the mean normal force and dominates the mean lateral force. High-symmetry, quasi-circular envelopes maintain balanced circumferential pressure, whereas deep-grooved, low-symmetry tri-lobed geometries produce asymmetric separation and a persistent lateral pressure imbalance. In the wake, high-symmetry cables sustain spanwise-synchronized Karman shedding at moderate-to-high inclinations, concentrating fluctuation energy in a dominant low-rank mode. Deep grooves disrupt shedding coherence, deflect the recirculation zone laterally, and spread energy across broadband scales. Mean lateral force and lateral-force fluctuations are therefore decoupled: the quasi-circular (1×19)(1 \times 19) cable has minimal mean lateral force but strong periodic fluctuations, while the deep-grooved (3×7)(3 \times 7) cable has weak fluctuations but appreciable mean lateral force. These results support construction-dependent cable selection and aerodynamic load modeling.

Introduction

An aerial towing system uses a cable as the primary mechanical link to connect an aircraft to a towed payload Figure 1, with certain designs also transmitting power and signals.[1, 2] Cables housing optical fibers or electrical conductors are typically protected by elastomeric or rubber jackets; however, mechanically loaded cables are commonly left unjacketed to minimize overall mass, envelope diameter, and bending stiffness. Consequently, their wire strands and helical inter-strand grooves are exposed directly to the airflow, introducing multiscale variations in surface inclination and curvature along both the axial and circumferential directions. Under oblique inflow, this helical surface produces spatially varying fluid–surface interactions that locally alter surface pressure, wall shear, boundary-layer separation and reattachment, and the downstream wake.[3, 4, 5]

Figure 1. Schematic of an aerial towed system.

Because aerial towing cables frequently extend over several kilometers, minor variations in local aerodynamic loading can accumulate along their length, significantly affecting total aerodynamic drag, tension distribution, and the equilibrium cable configuration.[6, 7] When predicting towing-system trajectories or local cable tension, treating the cable simply as a smooth cylinder characterized by a single drag coefficient is therefore insufficient. Instead, linking the local flow around the stranded surface to the three-component distributed aerodynamic loads is essential for high-fidelity models of towed decoys and trailing antennas.[8, 9, 10] Dynamic analyses of long flexible catenary cables and digital-twin frameworks for towed cable–body systems likewise depend on sectional fluid-load models.[74, 75] Here, surface-pressure redistribution and flow separation determine the pressure force contribution, wall

shear provides the viscous contribution, and axial superposition or cancellation determines which local changes survive spanwise integration. Comparable integrated force coefficients can therefore conceal dissimilar surface-loading distributions and wake states, limiting their direct applicability across different cable constructions, inflow angles, and cable lengths.

Although flow past a smooth circular cylinder provides the canonical baseline for bluff-body aerodynamics, even subtle departures from circularity can alter local surface curvature, pressure recovery, and recirculation regions. For instance, at Re=104Re = 10^{4}, Wang et al. observed both monotonic and nonmonotonic variations in mean surface pressure and wake velocity with increasing roundness imperfections, demonstrating that aerodynamic responses can reverse depending on disturbance amplitude.[11] Discrete surface features exhibit a similar sensitivity: spanwise wires and longitudinal fins markedly restructure the near-wake flow and the unsteady loading, and even a single surface groove shifts the separation angle and the balance between pressure and viscous drag according to its location.[12, 13, 14] Moreover, experiments with isolated large-scale surface perturbations demonstrate that boundary-layer separation, local pressure distributions, drag, and vortex shedding frequency depend strongly on perturbation location.[15]

Surface roughness, including that produced by marine fouling on cables and mooring lines, further perturbs the boundary layer and separated shear layers at geometry-dependent scales.[16] Across subcritical to transcritical regimes, large-eddy simulations of grooved and rough cylinders have shown that surface topography alters shear-layer transition, separation lines, near-wake topology, Reynolds stresses, and total drag in a regime-dependent manner, and helical grooves can suppress the formation of the Kármán vortex street.[17, 18, 19] At lower Reynolds numbers, Kumar and Roy found that the height and spacing of roughness elements govern the formation of local recirculation pockets, induce opposing shifts in pressure and friction drag, and modify both the vortex-formation length and shedding frequency.[20] Overall, these investigations indicate that the aerodynamic impact of surface disturbances depends on their specific geometry, spatial scale, Reynolds number, and inflow conditions.

On a stranded cable, these disturbances appear as strand crests and inter-strand grooves repeating along helical paths, giving the cable a three-dimensional geometry that is absent from both uniformly roughened surfaces and bluff bodies with invariant cross sections.

This helical geometry periodically shifts cross-sectional profiles and strand orientations along the cable axis, so that boundary-layer separation, shear-layer development, and wake topology vary with axial phase under identical inflow. For structurally related bluff bodies, such axial variations induce a strongly three-dimensional flow organization; even an abrupt spanwise change in diameter splits the wake into distinct vortex cells connected by vortex dislocations.[21] The spanwise correlation of cylinder wakes also depends on the aspect ratio or computational span considered.[65, 73] Numerical simulations of twisted cylinders show that spanwise geometric phase induces three-dimensional separation, delays shear-layer rollup, and produces wavelengh locking, oblique vortex shedding, and phase-dependent vortex-formation lengths.[22, 23, 24] Similarly, investigations of helical-straked cylinders found that the helical orientation of the strakes controls Reynolds stress distributions and wake coherence, to the point of breaking up the three-dimensional vortex structures.[25, 26, 27] For flexible cylinders, the strake configuration and yaw angle further govern the suppression of vortex-induced vibration and the resulting fatigue damage.[76, 77, 78] Studies of variable-pitch, wavy, and helically grooved cylinders likewise show that the axial wavelength and groove arrangement modify the near-wake vortex development and aerodynamic loading.[28, 29, 30] Although these configurations confirm the aerodynamic significance of axial phase modulation, they differ from stranded cables in cross-sectional topology, boundary conditions, and flow regimes, which prevents direct transfer of their quantitative results. Nevertheless, they identify axial position, helical phase, and repeat wavelength as the physically meaningful parameters for interpreting local separation, wake evolution, and three-component aerodynamic loading on stranded aerial cables.

Beyond the cable’s helical geometry, oblique inflow introduces both normal and axial velocity components relative to the cable axis, which alters the velocity composition and spatial development of the local three-dimensional flow. Studies of smooth yawed cylinders provide an essential baseline for isolating this inflow effect. Wang, Cheng, and Ting demonstrated that cylinder yaw staggers sectional crossflow forces along the span, causing large local forces to largely cancel in spanwise averages, even though the normal-velocity-normalized Strouhal number remains close to the crossflow value.[31] Earlier simulations and wake measurements similarly linked yaw to axial transport, oblique vortex shedding, and reorganized wake vorticity.[32, 33, 34] Subsequent delayed detached-eddy simulations and proper orthogonal decomposition (POD) at Re=104Re = 10^{4} further connected this axial flow to sectional pressure patterns, intermittent lift enhancement, and the convective propagation of force-bearing structures along the span.[35] These findings clarify the limits of the independence principle: normal velocity scaling collapses certain frequency characteristics and first-order force statistics, but it does not ensure similarity in local pressure, sectional loading, force fluctuations, or wake phase.[36] At a fixed freestream speed, changing the cable inclination angle simultaneously alters the normal and tangential velocity components and the normal Reynolds number, so angle-dependent analyses are required to distinguish among these coupled quantities.[37, 38]

When oblique inflow encounters a helical cable, the geometry-induced axial phase variation interacts with the normal and axial velocity components, generating a nonzero mean lateral force perpendicular to the plane formed by the cable axis and the freestream velocity. This coupled effect has been documented experimentally for various helical cables in yawed flow, and pressure measurements on three-strand twisted ropes attribute the transverse force to the circulation induced by the strand geometry and its orientation to the inflow.[39] In particular, Nebres et al. demonstrated that this lateral load arises primarily from asymmetric surface pressure and separation: the side where strands align more closely with the flow experiences delayed boundary-layer separation and lower mean pressure, producing a net transverse force toward that side.[4] Similarly, in tests motivated by aerial trailing-wire antennas, Stuart et al. observed that reversing helix handedness reversed the sign of the groove-induced side-force increment at small inclination angles, with surface tuft alignments confirming a groove-guided, circulation-like mechanism.[3] Although these investigations, together with side-force measurements on helically grooved slender bodies,[40] show that helical surface orientation influences mean lateral loading through asymmetric near-wall flow, they remain limited to isolated configurations. A unified, mechanism-resolved framework is still needed to elucidate how local loading, wake development, and three-component force redistribution systematically evolve with cable construction and inflow inclination.

To bridge this gap, recent investigations have begun resolving the aerodynamic action of stranded and helical surfaces through local surface flows and downstream wake dynamics. In large-eddy simulations and complementary particle image velocimetry (PIV) measurements of 6×16 \times1 and 3×13 \times1 cables in normal flow, Abdelhady and Wood identified near-stagnant fluid within the inter-strand gaps near the strand crevices, strong axial variations in separation angle, and a deeper base suction than on a circular cylinder. Although the dominant shedding frequency matched that of a circular cylinder, the velocity spectra, Reynolds stresses, and POD modal energy distributions were clearly different.[5, 41]

Wind-tunnel tests on stay cables with helical fillets and wire-mesh coverings similarly showed that localized surface features alter separation lines, transition states, and wake development, with local fluctuating loads partially canceling under spanwise integration.[42, 43, 44]

Comparative wind-tunnel and numerical studies across cable surface treatments, cable–wind angles, and cross-sectional shapes further show that the cable geometry significantly modifies the drag, the lift, and their dependence on Reynolds number and inflow angle.[45, 46, 47]

These findings demonstrate that a comparable shedding frequency is insufficient to infer similar local loading or wake organization. Although existing studies document isolated variations in local pressure, separation, and wake topology, they have not linked these near-wall flow modifications to the three aerodynamic force components under matched cable geometries and inflow inclinations. For aerial towing cables, the central unresolved question is how these local flow reorganizations are reflected in the normal, lateral, and tangential force components across varying cable constructions and flow angles.

In summary, prior studies show that surface disturbances alter pressure recovery and separation, helical geometry imposes phase-dependent axial flow variations, oblique inflow reshapes wake transport, and the combination of helical geometry and oblique inflow generates mean lateral forces and nonuniform local loading. However, these insights remain fragmented across disparate investigations, and a unified framework linking geometry, inflow, local flow physics, and three-component aerodynamic forces is still lacking for aerial cables. In particular, a construction- and angle-matched comparative analysis is required to determine how normal, lateral, and tangential forces are partitioned into pressure and viscous tractions, and which surface-pressure, separation, and wake signatures accompany this redistribution. Such detailed mechanical insight is essential because integrated aerodynamic coefficients alone cannot explain cable performance: identical global forces can mask markedly different pressure and shear distributions, and local force variations can be diminished or reversed by axial phase cancellation. Local flow reorganization and lengthwise spatial superposition therefore need to be treated together to resolve how near-wall mechanisms determine the integrated aerodynamic loading and to provide the flow-physics basis for assessing the directional aerodynamic forces on aerial towing cables.

Here, large-eddy simulations are employed to systematically investigate the flow physics around a smooth circular cylinder and four representative stranded cables across four inflow inclinations. We evaluate the normal, lateral, and tangential aerodynamic forces across these 20 cases, linking them to pressure/viscous force-source decompositions, surface- and base-pressure recovery, phase-resolved wake fields, and spectral as well as modal diagnostics, the latter having proven effective in resolving coherent structures in turbulent bluff-body wakes and separated flows.[48, 49] This work aims to determine the coupled dependencies among stranded geometry, oblique inflow, local surface flow, and wake organization, thereby

clarifying the physical origins of the observed aerodynamic trends. This perspective provides a basis for screening candidate towing-cable constructions according to their directional mean loading and lateral unsteadiness. The remainder of this paper is organized as follows: Sec. II outlines the computational models, flow parameters, and numerical verification; Sec. III presents the aerodynamic load balances, surface separation patterns, wake transport dynamics, unsteady spectral coherence, and modal organization; and Sec. IV summarizes the main conclusions.

Geometry and Numerical Method

Geometry and flow-condition definitions

The computational matrix comprises a smooth circular cylinder and four stranded cable configurations, denoted 1×71 \times7, 1×191 \times19, 3×73 \times7, and 7×77 \times7. Here, m×nm \times n represents a cable formed by mm strands, each containing nn wires. Each configuration is evaluated at four inflow inclinations, α=22.5∘\alpha= 22.5^\circ, 45∘45^\circ, 67.5∘67.5^\circ, and 90∘90^\circ, yielding 20 cases in total. All models share an identical envelope diameter, D=4D = 4 mm, providing a consistent reference scale for comparing the local flow physics and aerodynamic responses across configurations. In subsequent discussions, “single-strand” specifically denotes the 1×71 \times7 and 1×191 \times19 configurations, while “multi-strand” refers to the 3×73 \times7 and 7×77 \times7 cables, thereby distinguishing strand-level architecture from wire-level surface texture.

The axial model length LzL_z spans one lay pitch. The single-strand and outer-strand lay pitches are set to 6D=246D = 24 mm, falling within the limits specified by applicable engineering standards.[50, 51] For multi-strand cables, the internal-wire pitch is Pinner=3D=12P_{\mathrm{inner}} = 3D = 12 mm, which equals half the outer-strand pitch and complies with construction-dependent standards.[51] Single-strand cables adopt a right-handed (Z) lay, and multi-strand cables combine a right-handed (Z) outer-strand lay with a left-handed (S) internal-wire lay. To facilitate mesh generation and geometrically approximate the contact-induced compaction observed in practical cables, adjacent wires are slightly overlapped and merged. Figure 2 illustrates the nominal cross sections and representative cable geometries.

Figure 2. Nominal cable cross sections and representative stranded geometries, with cable-fixed coordinates and inflow velocity components: (a) cross-sectional contours of the five configurations; (b) single-strand 1×71 \times7 cable (Z lay); and (c) multi-strand 3×73 \times7 cable (Z/S lay).

A cable-fixed right-handed Cartesian coordinate system is defined by the orthonormal basis (en,eℓ,et)=(eX,eY,eZ)(\boldsymbol{e}_n,\boldsymbol{e}_\ell,\boldsymbol{e}_t) = (\boldsymbol{e}_X,\boldsymbol{e}_Y,\boldsymbol{e}_Z), where XX, YY, and ZZ designate the solver Cartesian directions, and the subscripts nn, ℓ\ell, and tt denote the normal, lateral, and tangential directions, respectively; the coordinates (x,y,z)(x,y,z) are measured from the cable axis along en\boldsymbol{e}_n, eℓ\boldsymbol{e}_\ell, and et\boldsymbol{e}_t. The inflow inclination α\alpha (equivalent to a yaw angle of 90∘−α90^\circ-\alpha as commonly defined for yawed cylinders) is defined as the angle between the freestream velocity vector U∞U_\infty and the cable axial direction et\boldsymbol{e}_t, such that α=90∘\alpha= 90^\circ corresponds to pure normal crossflow. The total freestream speed is fixed at U∞=37 m s−1U_\infty= 37\ {\rm m\,s}^{-1}, which, evaluated with standard sea-level air properties and the envelope diameter DD, corresponds to a nominal Reynolds number of Re≈1.0×104Re \approx1.0 \times10^4. With zero lateral velocity prescribed at the inflow boundaries, the incoming freestream velocity is decomposed as

U∞=Unen+Utet,Un=U∞sin⁡α,Ut=U∞cos⁡α.(1)U_\infty= U_n\boldsymbol{e}_n + U_t\boldsymbol{e}_t,\qquad U_n = U_\infty\sin\alpha,\qquad U_t = U_\infty\cos\alpha. \tag*{(1)}

Consequently, varying α\alpha changes the normal and tangential velocity components at a fixed total speed, with all subsequent flow quantities and aerodynamic loads referenced to this cable-fixed framework.

Two reference dynamic pressures are distinguished:

qtot=12ρU∞2,qn=12ρUn2,q_{\mathrm{tot}} = \frac{1}{2}\rho U_{\infty}^{2}, \qquad q_{n} = \frac{1}{2}\rho U_{n}^{2},
Ci(qχ)=FiqχAref,Aref=DLz,i∈{n,ℓ,t}.(2)C_{i}(q_{\chi}) = \frac{F_{i}}{q_{\chi} A_{\mathrm{ref}}}, \qquad A_{\mathrm{ref}} = D L_{z}, \qquad i \in\{n,\ell,t\}. \tag*{(2)}

Here, qχq_{\chi} with χ∈{tot,n}\chi\in\{\mathrm{tot},n\} denotes normalization by the total- or normal-speed dynamic pressure, with qχq_{\chi} applied to both force and surface-pressure coefficients. Surface-pressure coefficients are defined as Cp=(p−p∞)/qχC_{p}=(p-p_{\infty})/q_{\chi}, where pp is the time-averaged wall pressure and p∞p_{\infty} is the time-averaged static pressure at the inlet. Total-speed coefficients describe load variations at fixed flight speed; normal-speed coefficients allow comparison with canonical crossflow scaling. The corresponding Reynolds and Strouhal numbers are

Re=ρU∞Dμ,Ren=ρUnDμ,Stn=fDUn,(3)Re = \frac{\rho U_{\infty}D}{\mu}, \qquad Re_{n} = \frac{\rho U_{n}D}{\mu}, \qquad St_{n} = \frac{fD}{U_{n}}, \tag*{(3)}

where ff is the frequency; at the dominant vortex shedding frequency, StnSt_{n} is the shedding Strouhal number, denoted StpSt_{p} in § 4 D, where StnSt_{n} also serves as the spectral frequency variable. Thus, varying α\alpha at fixed U∞U_{\infty} simultaneously alters both RenRe_{n} and UtU_{t}.

The mean pressure and viscous force components are reconstructed by integrating the time-averaged wall fields over the solid surface SS:

Fi,p=−∫Sp n^s⋅ei dS,Fi,τ=∫Sτw⋅ei dS(4)F_{i,p}=-\int_{S}p\,\hat{\mathbf{n}}_{s}\cdot\mathbf{e}_{i}\,\mathrm{d}S, \qquad F_{i,\tau}=\int_{S}\boldsymbol{\tau}_{w}\cdot\mathbf{e}_{i}\,\mathrm{d}S \tag*{(4)}

where n^s\hat{\mathbf{n}}_{s} is the outward unit normal vector on the solid surface, pp is the time-averaged wall pressure defined above, and τw\boldsymbol{\tau}_{w} represents the signed viscous wall traction vector; the total force component in Eq. (2) is Fi=Fi,p+Fi,τF_{i}=F_{i,p}+F_{i,\tau}. Normalizing these two contributions by qtotArefq_{\mathrm{tot}}A_{\mathrm{ref}} yields the pressure and viscous force coefficients analyzed subsequently:

Ci,p=Fi,pqtotAref,Ci,τ=Fi,τqtotAref.(5)C_{i,p}=\frac{F_{i,p}}{q_{\mathrm{tot}}A_{\mathrm{ref}}}, \qquad C_{i,\tau}=\frac{F_{i,\tau}}{q_{\mathrm{tot}}A_{\mathrm{ref}}}. \tag*{(5)}

Unless otherwise stated, all force coefficients, velocities, and surface pressures reported herein represent time averages, and fluctuation intensities are denoted by the subscript rms (root mean square).

To characterize the helical periodicity of the cables, an axial phase coordinate is defined along the cable axis et\mathbf{e}_{t}. For each configuration, the phase-resolved cross sections are indexed by the geometric repeat period PrepP_{\mathrm{rep}}, which represents the axial distance over which the nominal cross-sectional pattern recurs. The local axial phase ϕ\phi is expressed as

ϕ=2πzPrep,(6)\phi=\frac{2\pi z}{P_{\mathrm{rep}}}, \tag*{(6)}

where zz is the axial coordinate along et\mathbf{e}_{t}. For the single-strand configurations, the corresponding repeat periods are Prep=4P_{\mathrm{rep}}=4 mm for 1×71 \times7 and 2 mm for 1×191 \times19, reflecting their discrete rotational symmetries over one lay pitch. For multi-strand cables, different values of PrepP_{\mathrm{rep}} and phase coordinates are evaluated separately for the outer-strand and internal-wire structural scales as needed.

Computational domain, mesh, and numerical settings

Three-dimensional large-eddy simulations (LES) are employed to resolve the unsteady flow past the stationary cable geometries. The large, energy-containing turbulent structures are directly resolved on the computational grid, while the subgrid-scale (SGS) stresses are represented by the constant-coefficient Smagorinsky–Lilly model with Cs=0.1C_{s}=0.1. This numerical approach resolves the unsteady boundary-layer separation, wake evolution, and aerodynamic load fluctuations required for the present mechanism analysis while avoiding the high computational cost of direct numerical simulation (DNS). Air is modeled as an incompressible Newtonian fluid with density ρ=1.225 kg m−3\rho=1.225\ \mathrm{kg\,m^{-3}} and dynamic viscosity μ=1.7894×10−5 Pa s\mu=1.7894 \times10^{-5}\ \mathrm{Pa\,s}.

The computational domain is a rectangular box extending 46D46D, 21D21D, and 6D6D along the normal (en\mathbf{e}_{n}), lateral (eℓ\mathbf{e}_{\ell}), and axial (et\mathbf{e}_{t}) directions, respectively. Here, the inlet and outlet boundaries are situated 15.5D15.5D upstream and 30.5D30.5D downstream of the cable axis, while the lateral boundaries are placed at 10.5D10.5D on either side. At the inlet, the velocity vector is prescribed using the normal and tangential components defined in Eq. (1). A zero-gauge-pressure condition is imposed at the downstream outlet, symmetry boundary conditions are applied to the lateral boundaries, and the stationary cable surface is treated as a no-slip wall. Translational periodic boundary conditions are applied between the two axial end planes. The computational domain, coordinate orientations, and boundary conditions are illustrated in Figure 3.

Figure 3. Computational domain, coordinate system, and boundary conditions.

Spatial discretization is performed using an unstructured polyhedral mesh with boundary-layer prism layers clustered near the cable wall. Local mesh refinement is applied around the outer wire ridges, inter-wire grooves, and throughout the near-wake region, gradually coarsening toward the far-field boundaries. Four grid levels—designated coarse, medium, fine, and finest—are generated for the grid-independence study. All four grids share the same base settings: a near-wall surface size of 0.025D0.025D, a far-field maximum cell size of 0.5D0.5D, and 15 prism layers with the first-cell height tailored to ensure y+≈0.5y^+ \approx0.5. The grid refinement is controlled solely by the cell size within the local refinement zone, set to 0.125D0.125D, 0.100D0.100D, 0.080D0.080D, and 0.070D0.070D across the four levels, respectively. Figure 4 displays the surface mesh of the 1×71 \times7 cable alongside a cross-sectional view comparing the coarse and finest volume meshes.

The SIMPLE algorithm is used for pressure–velocity coupling. Pressure spatial discretization is handled by a standard second-order scheme, momentum convection is discretized using bounded central differencing, and temporal advancement is achieved via a bounded second-order implicit formulation. The nondimensional time step is defined as

Δt∗=ΔtU∞D.(7)\Delta t^* = \frac{\Delta t U_\infty}{D}. \tag*{(7)}

In normal-velocity units, the time step is ΔtUn/D=Δt∗sin⁡α\Delta t U_n/D = \Delta t^* \sin\alpha. For the time-step sensitivity evaluation at α=90∘\alpha= 90^\circ, three nondimensional time steps are compared: Δt∗=0.0023\Delta t^* = 0.0023, 0.00460.0046, and 0.00920.0092. After initial transient effects diminish, the production statistics of the mean-flow fields and of the force and probe histories are accumulated over durations of 0.0120.012–0.0360.036 s, depending on the case, which corresponds to at least 94 convective time units D/UnD/U_n, or about 19 vortex shedding cycles, in every case.

Grid and time-step verification

Grid and time-step verification was conducted for a smooth circular cylinder and a 1×71 \times7 stranded cable at normal inflow (α=90∘\alpha= 90^\circ); the results are summarized in Table I and Figure 5. Four mesh levels were evaluated first, followed by three nondimensional time steps on the fine mesh. The assessment considers the mean normal-force coefficient CnC_n, the root-mean-square lateral-force fluctuation Cℓ,rmsC_{\ell,\mathrm{rms}}, the normal Strouhal number StnSt_n, the normalized minimum centerline velocity Umin⁡/UnU_{\min}/U_n, the recirculation length Lr/DL_r/D, and the base pressure coefficient CpbC_{pb}. Here, Umin⁡U_{\min} denotes the minimum normal velocity along the downstream centerline; Lr/D=x0/D−0.5L_r/D = x_0/D - 0.5 denotes the wake recirculation length, where x0x_0 is the streamwise position, measured from the cable axis, of the first downstream zero crossing of the time-averaged normal velocity along the centerline, so that the subtracted 0.5D0.5D refers to the length to the nominal rear envelope; and CpbC_{pb} is the base pressure coefficient, defined as the time-averaged pressure coefficient at the leeward base point. These metrics together characterize the integrated mean loading, the dominant shedding unsteadiness, the wake recirculation, and the leeward base pressure response.

Figure 5. Time-averaged surface-pressure distributions at α=90∘\alpha= 90^\circ: (a) smooth cylinder and (b) 1 × 7 cable on the four meshes; (c) smooth cylinder and (d) 1 × 7 cable for the three time steps on the fine mesh. The angle ϕs\phi_s is measured from the windward stagnation line. The black dashed line is digitized from Norberg[54] at Re=8.0×103Re = 8.0 \times10^3, and the star marks the measured base pressure at Re=1.01×104Re = 1.01 \times10^4.

For the smooth cylinder, refining the mesh from coarse to fine reduces CnC_n from 1.215 to 1.090 and shifts StnSt_n from 0.201 to 0.210, while the further discrepancy between the fine and finest meshes is limited to 1.9%, 2.6%, 1.3%, 0.2%, 1.6%, and 3.0% across CnC_n, Cℓ,rmsC_{\ell,\mathrm{rms}}, StnSt_n, Umin⁡/UnU_{\min}/U_n, Lr/DL_r/D, and CpbC_{pb}, respectively. With the fine-grid solution taken as the reference, the coarse and medium cylinder meshes exhibit relative departures of 11.5% and 4.1% in CnC_n, and 4.2% and 3.2% in StnSt_n, respectively. For the 1×71 \times7 cable, the corresponding departures in CnC_n are within 0.3% and 0.1%, though Cℓ,rmsC_{\ell,\mathrm{rms}} remains 9.5% and 6.3% higher on the coarse and medium grids. The fine-to-finest variations are 3.6%, 7.0%, 0.4%, 4.9%, 5.1%, and 6.4% for CnC_n, Cℓ,rmsC_{\ell,\mathrm{rms}}, StnSt_n, Umin⁡/UnU_{\min}/U_n, Lr/DL_r/D, and CpbC_{pb}, respectively, with the complete grid-comparison data compiled in Table 1.

CaseSettingCnC_nCℓ,rmsC_{\ell,\mathrm{rms}}StnSt_n−Umin⁡/Un-U_{\min}/U_nLr/DL_r/D−Cpb-C_{p_b}
Grid sensitivity
Cylindercoarse1.2150.4780.2010.2770.921.238
medium1.1350.3240.2030.3221.031.117
fine1.0900.2750.2100.3061.221.054
finest1.0690.2680.2070.3061.201.022
1 × 7 cablecoarse1.1220.2190.1980.2941.141.053
medium1.1170.2120.2060.2991.121.032
fine1.1190.2000.2050.2941.181.047
finest1.0780.1860.2060.2791.240.980
Reference results
CylinderDNS[52]1.143–0.2030.2490.821.129
Exp[52, 53, 54]1.104–1.186–0.193–0.2020.228–0.3800.78–1.021.07
1 × 7 cableExp[55, 56]1.120–1.360–0.180–0.220–––
Exp (PIV)[41]––0.21–––
Time-step sensitivity on the fine mesh
CylinderΔt∗=0.0023\Delta t^{*} = 0.00231.0870.2650.2080.3001.241.040
Δt∗=0.0046\Delta t^{*} = 0.00461.0900.2750.2100.3061.221.054
Δt∗=0.0092\Delta t^{*} = 0.00921.1200.3090.2050.3281.091.095
1 × 7 cableΔt∗=0.0023\Delta t^{*} = 0.00231.1010.1930.2070.2821.191.012
Δt∗=0.0046\Delta t^{*} = 0.00461.1190.2000.2050.2941.181.047
Δt∗=0.0092\Delta t^{*} = 0.00921.0830.1720.2070.3231.300.992

Table 1. Grid and time-step sensitivity and reference results at 90° inflow.

Against references at Re≈104Re \approx10^4, the fine-grid cylinder CnC_n lies 1.3% below the lower experimental bound and 4.6% below the DNS of Dong et al.^{52}, StnSt_n exceeds the measured range by 4.0%, and −Umin⁡/Un-U_{\min}/U_n falls within the experimental interval. The recirculation length is the least well reproduced quantity: the fine-grid value Lr/D=1.22L_r/D = 1.22 exceeds the measured range of 0.78–1.02, consistent with the known sensitivity of the wake closure point to shear-layer transition and spanwise resolution.^{64,65} For the stranded cable, the fine-grid CnC_n lies within 0.1% of the lower experimental bound, and StnSt_n lies within the experimental range and 2.4% below the PIV value for a six-outer-strand cable at Re=3100Re = 3100. Consequently, the fine mesh is adopted for all subsequent production simulations.

On the fine mesh, the time-step sensitivity is evaluated using Δt∗=0.0023\Delta t^* = 0.0023, 0.00460.0046, and 0.00920.0092. For the smooth cylinder, Δt∗=0.0023\Delta t^* = 0.0023 and 0.00460.0046 yield essentially identical results for the mean normal force, shedding frequency, minimum centerline velocity, recirculation length, and base pressure coefficient. Increasing the time step to Δt∗=0.0092\Delta t^* = 0.0092 shifts CnC_n from 1.090 to 1.120, elevates Cℓ,rmsC_{\ell,\mathrm{rms}} from 0.275 to 0.309, and contracts Lr/DL_r/D from 1.22 to 1.09. For the 1×71 \times7 stranded cable, while StnSt_n remains between 0.205 and 0.207 for all three settings, the mean force, lateral-force fluctuation, minimum centerline velocity, and reCylinder references are at Re≈104Re \approx10^4: experimental CnC_n and StnSt_n as compiled by Dong et al.[52], Umin⁡/UnU_{\min}/U_n and Lr/DL_r/D from PIV[52] and laser Doppler velocimetry (LDV),[53] and −Cpb-C_{p_b} from Norberg[54]; benchmarks at Re=1000Re = 1000–39003900 are given in Refs. [57, 58, 59, 60, 61, 62, 63]. For the 1 × 7 cable, CnC_n is from overhead conductors,[55] StnSt_n from wire ropes of other constructions[56] at Re=5.8×103Re = 5.8 \times10^3–1.38×1041.38 \times10^4, and the PIV value[41] from a six-outer-strand cable at Re=3100Re = 3100. Because the present verification runs and the subsequent production runs are separate simulations with targeted adjustments to the statistical averaging duration, some corresponding quantities exhibit minor differences of up to 3%.

Mesh of the 1 × 7 cable

Figure 4. Mesh of the 1 × 7 cable: (a) surface mesh on the cable wall; and (b) volume-mesh cross section in the z=0z = 0 plane, with the coarse (upper half) and finest (lower half) meshes.

circulation length exhibit their largest departures at Δt∗=0.0092\Delta t^{*} = 0.0092. Therefore, the fine mesh paired with Δt∗=0.0046\Delta t^{*} = 0.0046 is selected as the baseline numerical setting for all subsequent production simulations. Figure 5 compares the corresponding mean surface-pressure distributions. For the smooth cylinder, the fine and finest meshes differ by less than 0.02 on the windward half and by at most 0.03 near the base, and Δt∗=0.0023\Delta t^{*} = 0.0023 stays within 0.02 of the baseline. Against the measurements of Norberg[54], the fine-grid leeward plateau agrees to within 0.02 at Re=8.0×103Re = 8.0 \times10^3, and −Cpb-C_{p_b} lies within 1.5% of the

the value at Re=1.01×104Re = 1.01 \times10^4, whereas the computed suction peak is about 0.05 deeper and lies about 4° farther upstream. For the 1 × 7 cable, the coarse, medium, and fine distributions differ by less than 0.04; the finest mesh and Δt∗=0.0092\Delta t^{*} = 0.0092 make the leeward plateau about 0.05 less negative.

Results and discussion

Global three-component aerodynamic loads and pressure–viscous contributions

We first examine the complete parametric matrix of five geometries and four inflow inclinations to determine how stranded cable surfaces modify the global three-component aerodynamic response. Aerodynamic force coefficients are evaluated under both total-speed (qtotq_{\mathrm{tot}}) and normal-speed (qnq_n) normalizations [Eq. (2)], separating flight-speed load variations from crossflow-scaling departures across geometries. Figure 6 summarizes the mean normal, lateral, and tangential-force coefficients, the lateral-force fluctuation, and the absolute pressure share of the mean normal force across all 20 cases. Here, the absolute pressure share of component ii is defined as ∣Fi,p∣/(∣Fi,p∣+∣Fi,τ∣)\lvert F_{i,p}\rvert/(\lvert F_{i,p}\rvert+ \lvert F_{i,\tau}\rvert), which compares the relative magnitudes of pressure and viscous contributions irrespective of algebraic sign.

Figure 6. Detailed aerodynamic parameters for all 20 cases: (a) mean normal-force coefficient under total-speed normalization Cn(qtot)C_n(q_{\mathrm{tot}}); (b) mean normal-force coefficient under normal-speed normalization Cn(qn)C_n(q_n); (c) mean lateral-force coefficient Cℓ(qtot)C_\ell(q_{\mathrm{tot}}); (d) root-mean-square lateral-force fluctuation Cℓ,rms(qtot)C_{\ell,\mathrm{rms}}(q_{\mathrm{tot}}); (e) mean tangential-force coefficient Ct(qtot)C_t(q_{\mathrm{tot}}); and (f) absolute pressure share of the mean normal force.

Under total-speed normalization (qtotq_{\mathrm{tot}}), the mean normal-force coefficient CnC_n increases with inflow inclination α\alpha across all five geometries [Figure 6(a)], though both absolute magnitudes and relative configuration rankings evolve. At α=22.5∘\alpha= 22.5^\circ, CnC_n spans from 0.140 for the 7 × 7 cable to 0.166 for the smooth cylinder; at α=90∘\alpha= 90^\circ, this spread widens to 0.900–1.105. This monotonic increase is associated with the growth of the normal velocity component UnU_n as α\alpha increases at fixed total speed. Furthermore, under normal-speed normalization (qnq_n), the coefficients contract into a much narrower band of approximately 0.90–1.18 [Figure 6(b)], yet they fail to collapse onto a single universal curve: at α=67.5∘\alpha= 67.5^\circ, the 1 × 19 cable reaches Cn(qn)=1.182C_n(q_n) = 1.182, compared with 1.018 for the 3 × 7 cable. Normal-speed scaling therefore accounts for the kinematic crossflow contribution and leaves a residual response that depends on the cable surface topology.

Unlike the monotonic behavior of normal loads, the mean lateral-force magnitude ∣Cℓ∣\lvert C_\ell\rvert displays a nonmonotonic, configuration-sensitive response to inclination [Figure 6(c)]; at α<90∘\alpha< 90^\circ, the mean lateral force on the stranded cables is directed toward −eℓ-\boldsymbol{e}_\ell (Cℓ≤0C_\ell\le0), and magnitudes are reported below. In particular, ∣Cℓ∣\lvert C_\ell\rvert remains small in absolute terms at α=22.5∘\alpha= 22.5^\circ, becomes large for several stranded cables at intermediate inclinations (α=45∘\alpha= 45^\circ and 67.5∘67.5^\circ), and returns to near zero under pure crossflow at α=90∘\alpha= 90^\circ. The largest mean lateral force arises in the 3 × 7 cable at α=67.5∘\alpha= 67.5^\circ, reaching ∣Cℓ∣(qtot)=0.354\lvert C_\ell\rvert(q_{\mathrm{tot}}) = 0.354, compared with 0.231 for 7 × 7, 0.117 for 1 × 7, and 0.045 for 1 × 19. Already at α=45∘\alpha= 45^\circ, the 1 × 7 and 3 × 7 configurations produce comparable lateral-force magnitudes of ∣Cℓ∣(qtot)=0.218\lvert C_\ell\rvert(q_{\mathrm{tot}}) = 0.218 and 0.213, respectively. Throughout the entire inclination range, the smooth cylinder maintains a near-zero mean lateral force (∣Cℓ∣(qtot)<0.01\lvert C_\ell\rvert(q_{\mathrm{tot}}) < 0.01), confirming that the nonzero lateral load is an aerodynamic signature of surface stranding.

The inclination corresponding to peak lateral loading differs systematically by cable family: the single-strand cables (1 × 7 and 1 × 19) peak in magnitude at α=45∘\alpha= 45^\circ, and the multi-strand cables (3 × 7 and 7 × 7) continue to increase and peak at α=67.5∘\alpha= 67.5^\circ. Among all stranded geometries, 1 × 19 exhibits the smallest lateral-force magnitude across intermediate inclinations (∣Cℓ∣(qtot)≤0.103\lvert C_\ell\rvert(q_{\mathrm{tot}}) \le0.103). This shift in peak inclination largely reflects the normalization: under qnq_n, all stranded cables attain their largest mean lateral-force coefficient near α=45∘\alpha= 45^\circ. Because ∣Cℓ∣(qtot)=∣Cℓ∣(qn)sin⁡2α\lvert C_\ell\rvert(q_{\mathrm{tot}}) = \lvert C_\ell\rvert(q_n)\sin^2\alpha and $\sin^2\alpha increases by a factor of 1.71 from 45∘45^\circ to 67.5∘67.5^\circ, the qtotq_{\mathrm{tot}} peak moves to 67.5∘67.5^\circ when ∣Cℓ∣(qn)\lvert C_\ell\rvert(q_n) at 67.5∘67.5^\circ retains more than about 59% of its 45∘45^\circ value. This holds for the multi-strand cables, whose ∣Cℓ∣(qn)\lvert C_\ell\rvert(q_n) decreases by only 3% (3×7) and 20% (7×7) between the two inclinations, but not for the single-strand cables, whose ∣Cℓ∣(qn)\lvert C_\ell\rvert(q_n) decreases by about 70%. The surface-pressure origin of these lateral-force variations is analyzed in § 4 B.

The lateral-force fluctuation, measured by Cℓ,rmsC_{\ell,\mathrm{rms}} [Fig. 6(d)], displays an independent, configuration-dependent trend across the parameter matrix. At α=67.5∘\alpha= 67.5^\circ, the 1×19 cable develops the strongest unsteady fluctuation (Cℓ,rms(qtot)=0.300C_{\ell,\mathrm{rms}}(q_{\mathrm{tot}}) = 0.300); however, the 3×7 cable yields only 0.053, despite generating the largest mean lateral-force magnitude (∣Cℓ∣(qtot)=0.354\lvert C_\ell\rvert(q_{\mathrm{tot}}) = 0.354). Under pure crossflow (α=90∘\alpha= 90^\circ), the mean lateral force returns to near zero across all geometries, yet the fluctuation levels remain clearly configuration-dependent: among the stranded cables, Cℓ,rms(qtot)C_{\ell,\mathrm{rms}}(q_{\mathrm{tot}}) ranges from 0.019 for 3 × 7 to 0.204 for 1 × 19, all below the smooth-cylinder level. The mean lateral-force magnitude ∣Cℓ∣\lvert C_\ell\rvert quantifies the stationary asymmetry of the time-averaged flow, and Cℓ,rmsC_{\ell,\mathrm{rms}} reflects the intensity of unsteady vortex shedding fluctuations about this mean state. These two metrics convey complementary dynamic information, and their different combinations separate the aerodynamic responses of the 1 × 19 and 3 × 7 cables.

The mean tangential-force coefficient CtC_t reflects the three-dimensional flow response under oblique inflow [Fig. 6(e)]. Although the axial freestream component produces a baseline viscous tangential load on the smooth cylinder, the departures observed across the stranded cables indicate how the helical surface modifies axial momentum exchange. Across all inclined cases (α≤67.5∘\alpha\le67.5^\circ),

CtC_t remains nonzero and spans the range of 0.012–0.041, attaining its highest levels at small to intermediate inclinations before returning to near zero at α=90∘\alpha= 90^\circ under pure crossflow. This variation does not scale monotonically with the nominal number of strands or outer wires: the 3 × 7 cable attains Ct(qtot)=0.040C_t(q_{\mathrm{tot}}) = 0.040 at α=22.5∘\alpha= 22.5^\circ; both the 1 × 7 and 7 × 7 configurations reach peak values of approximately 0.041 at α=45∘\alpha= 45^\circ.

Pressure–viscous force decomposition shows that the mean normal force is dominated by pressure in all 20 cases, with surface pressure accounting for 94.9%–98.3% of the total load [Fig. 6(f)]. For the mean lateral force, the load is likewise dominated by asymmetric surface pressure. At α=67.5∘\alpha= 67.5^\circ, for example, the lateral force of the 3 × 7 cable consists of Cℓ,p=−0.347C_{\ell,p} = -0.347 and Cℓ,τ=−0.006C_{\ell,\tau} = -0.006; that of the 1 × 19 cable comprises Cℓ,p=−0.042C_{\ell,p} = -0.042 and Cℓ,τ=−0.003C_{\ell,\tau} = -0.003, confirming that viscous shear represents only a minor fraction of the lateral balance in both configurations. The mean tangential load, however, arises from a delicate balance between viscous friction and surface pressure: at α=67.5∘\alpha= 67.5^\circ, for 3 × 7, a negative pressure contribution (Ct,p=−0.010C_{t,p} = -0.010) partially offsets the positive viscous shear (Ct,τ=0.021C_{t,\tau} = 0.021), resulting in a small net force. For 1 × 19, both contributions act in the same direction (Ct,p=0.007C_{t,p} = 0.007 and Ct,τ=0.016C_{t,\tau} = 0.016). Figure 7 summarizes these signed contributions for each force component.

Figure 7. Signed pressure and viscous force contributions across normal, lateral, and tangential directions for the 1 × 19 and 3 × 7 cables at α=67.5∘\alpha= 67.5^\circ.

Overall, the integrated load responses across the 20 cases lead to two main observations regarding stranded cable aerodynamics. First, a lower mean normal load does not imply an overall reduction in aerodynamic burden: at α=67.5∘\alpha= 67.5^\circ, the 3 × 7 cable exhibits a relatively modest normal force but generates the largest mean lateral force across all cases, reflecting a three-dimensional redistribution of the load. The ratio ∣Cℓ∣/Cn|C_\ell|/C_n, which does not depend on the normalization, reaches 0.354/0.869=0.410.354/0.869 = 0.41 for the 3×73 \times7 cable, so the resultant of the normal and lateral forces in the cross-sectional plane is deflected by about 22∘22^\circ from the plane containing the inflow and the cable axis, compared with 2.6∘2.6^\circ for the 1×191 \times19 cable (0.045/1.0090.045/1.009). At α=45∘\alpha= 45^\circ, this deflection exceeds 20∘20^\circ for both the 1×71 \times7 and 3×73 \times7 cables. A load model that neglects the lateral component would therefore misplace the force direction by an angle of this order.

Second, the mean lateral force is decoupled from the lateral-force fluctuation: the 3×73 \times7 cable maintains a low lateral-force fluctuation (Cℓ,rms(qtot)=0.053C_{\ell,\mathrm{rms}}(q_{\mathrm{tot}}) = 0.053) together with its pronounced mean lateral force (∣Cℓ∣(qtot)=0.354|C_\ell|(q_{\mathrm{tot}}) = 0.354). The 1×191 \times19 cable, however, develops a strong lateral-force fluctuation (Cℓ,rms(qtot)=0.300C_{\ell,\mathrm{rms}}(q_{\mathrm{tot}}) = 0.300) despite a minimal mean lateral force. These global differences cannot be fully understood from integrated forces alone; they are linked with how helical surface grooves channel the crossflow and reorganize local surface-pressure fields, boundary-layer separation lines, and wake vortex structures. § 4 B therefore investigates these local flow fields for representative cable geometries and inflow inclinations.

Surface pressure, wall shear, and separation organization

To clarify the surface origin of the integrated load responses obtained in § 4 A, the time-averaged surface pressure and wall shear stress are evaluated over one lay pitch (Lz=6DL_z = 6D) and mapped onto a unified circumferential coordinate ϕs\phi_s around the nominal outer contour, measured from the windward stagnation line and increasing toward +eℓ+\mathbf{e}_\ell (distinguished from the axial phase coordinate ϕ\phi defined in § 2). Several diagnostic metrics are used to characterize the key surface features: the windward front-point pressure coefficient CpfC_{pf}, the circumferential minimum pressure coefficient CpmC_{pm}, taken as the minimum of the area-weighted, axially averaged distribution Cp(ϕs)C_p(\phi_s), and the root-mean-square mirror-point pressure difference Cp,asym,rmsC_{p,\mathrm{asym,rms}}, which characterizes the circumferential pressure asymmetry,

Cp,asym,rms={1Ns∑k=1Ns[Cp(ϕs,k)−Cp(360∘−ϕs,k)]2}1/2(8)C_{p,\mathrm{asym,rms}} = \left\{ \frac{1}{N_s} \sum_{k=1}^{N_s} \left[C_p(\phi_{s,k}) - C_p(360^\circ- \phi_{s,k})\right]^2 \right\}^{1/2} \tag*{(8)}

evaluated over the Ns=180N_s = 180 bins of 1∘1^\circ width in 0∘<ϕs<180∘0^\circ< \phi_s < 180^\circ. Following the kinematic separation criterion of Haller, applied here to the circumferential component of the time- and axially averaged skin-friction coefficient, boundary-layer separation and reattachment locations are identified from its zero crossings. The circumferential angles of the primary separation points on either side, measured from the windward stagnation line, are denoted by θfl\theta_{\mathrm{fl}} on the +eℓ+\mathbf{e}_\ell side (0∘<ϕs<180∘0^\circ< \phi_s < 180^\circ) and θfr\theta_{\mathrm{fr}} on the −eℓ-\mathbf{e}_\ell side. Together with CpbC_{pb}, these indicators summarize the surface-pressure, wall-shear, and separation characteristics across all 20 cases in Figure 8.

Figure 8. Surface-pressure, wall-shear, and separation-organization indicators across all 20 cases: (a) windward front-point pressure coefficient Cpf(qtot)C_{pf}(q_{\mathrm{tot}}); (b) base pressure coefficient Cpb(qtot)C_{pb}(q_{\mathrm{tot}}); (c) circumferential suction peak (minimum pressure) coefficient Cpm(qtot)C_{pm}(q_{\mathrm{tot}}); (d) circumferential pressure asymmetry metric Cp,asym,rms(qtot)C_{p,\mathrm{asym,rms}}(q_{\mathrm{tot}}); (e) windward-left primary separation angle θfl\theta_{\mathrm{fl}}; and (f) reflected windward-right primary separation angle 360∘−θfr360^{\circ}-\theta_{\mathrm{fr}}.

Across the parameter space, Figure 8(a)–Figure 8(c) summarize the variations in the three primary pressure metrics: CpfC_{pf} spans from 0.104 to 1.001, CpbC_{pb} from −1.007-1.007 to −0.116-0.116, and CpmC_{pm} from −1.606-1.606 to −0.197-0.197. The contrast between the 1×191 \times19 and 3×73 \times7 cables at α=67.5∘\alpha= 67.5^\circ illustrates this variability. The 3×73 \times7 cable exhibits a lower front pressure coefficient (Cpf=0.798C_{pf} = 0.798) and a less negative base pressure coefficient (Cpb=−0.667C_{pb} = -0.667) than the 1×191 \times19 cable (Cpf=0.853C_{pf} = 0.853 and Cpb=−0.962C_{pb} = -0.962). However, its circumferential minimum pressure is much more negative, reaching Cpm=−1.606C_{pm} = -1.606 compared with −1.057-1.057 for 1×191 \times19. The deeper suction peak of the 3×73 \times7 cable is consistent with local flow acceleration over its prominent outer-strand crests, whereas its less negative base pressure reflects a more modulated leeward recovery (Figure 9) than the nearly uniform leeward suction of the 1×191 \times19 cable.

Figure 9. Axial phase organization of the mean surface-pressure coefficient Cp(qtot)C_p(q_{\mathrm{tot}}) across the smooth cylinder and stranded cables at α=45∘\alpha=45^{\circ} and 67.5∘67.5^{\circ}. Unwrapped horizontal coordinates ϕs∈[0∘,360∘]\phi_s \in[0^{\circ},360^{\circ}] trace the circumferential envelope, and the vertical span z/D∈[−3,3]z/D \in[-3,3] covers one lay pitch (Lz=6DL_z=6D). Solid and dashed lines mark the inter-strand and inter-wire grooves, respectively.

Across all 20 cases, the suction minimum and the integrated normal load correlate strongly and negatively (r=−0.87r = -0.87), and the front stagnation pressure correlates almost linearly with the normal load (r=0.99r = 0.99). Both correlations, however, largely reflect the common dependence of these quantities on α\alpha. At a fixed inclination, the deepest suction does not imply the largest normal load: at both α=45∘\alpha= 45^\circ and 67.5∘67.5^\circ, the 3×73 \times7 cable combines the deepest suction peak with the lowest normal load among the five configurations. The normal load is therefore determined by the opposing contributions of windward compression and localized suction peaks, so no single scalar pressure indicator is sufficient to represent the integrated aerodynamic force.

A similar contrast between local suction and leeward recovery appears at α=45∘\alpha= 45^\circ, where the 3×73 \times7 and 1×191 \times19 cables again form the opposite bounds among the stranded geometries. The 3×73 \times7 cable exhibits the least negative base pressure (Cpb=−0.310C_{pb} = -0.310) together with the deepest suction peak (Cpm=−0.961C_{pm} = -0.961), and the 1×191 \times19 cable develops the most negative base pressure (Cpb=−0.507C_{pb} = -0.507) with the shallowest suction (Cpm=−0.757C_{pm} = -0.757). The 1×71 \times7 and 7×77 \times7 cables lie strictly between these two bounds, with intermediate base pressure levels (Cpb=−0.420C_{pb} = -0.420 and −0.445-0.445) and suction minima (Cpm=−0.914C_{pm} = -0.914 and −0.891-0.891). Figure 8(b) further shows a systematic dependence on inflow inclination: the cross-configuration spread in CpbC_{pb} (defined as the difference between the maximum and minimum values among all configurations at a given inclination) widens monotonically with increasing inclination, expanding from 0.046 at α=22.5∘\alpha= 22.5^\circ to 0.197 at 45∘45^\circ, 0.295 at 67.5∘67.5^\circ, and 0.305 under pure crossflow (α=90∘\alpha= 90^\circ). The leeward base pressure therefore differentiates the cable topologies increasingly as the crossflow component grows, and the geometric sensitivity is largest under pure

crossflow.

The circumferential pressure asymmetry shows a systematic sensitivity to both cable configuration and inflow inclination, as evaluated by Cp,asym,rmsC_{p,\mathrm{asym,rms}} in Figure 8(d). For the smooth cylinder, Cp,asym,rmsC_{p,\mathrm{asym,rms}} remains bounded between 5.4×10−45.4 \times10^{-4} and 2.6×10−32.6 \times10^{-3}, reflecting time-averaged crossflow reflectional symmetry across all inclinations. The stranded cables, however, develop much larger pressure asymmetry under oblique inflow: the multi-strand 3×73 \times7 cable reaches 0.4230.423 at α=67.5∘\alpha= 67.5^\circ and 0.2860.286 at 45∘45^\circ; the 7×77 \times7 cable attains 0.2540.254 at 67.5∘67.5^\circ. For the single-strand 1×191 \times19 cable, the asymmetry peaks instead at a moderate inclination (0.1300.130 at 45∘45^\circ) before decreasing with inclination to 0.0520.052 at 67.5∘67.5^\circ and 0.0130.013 at 90∘90^\circ. Across the entire 20-case parameter space, Cp,asym,rmsC_{p,\mathrm{asym,rms}} correlates closely with the mean lateral-force magnitude ∣Cℓ∣\lvert C_\ell\rvert (r=0.99r = 0.99), with the linear correlation tightening to 0.9970.997 at α=45∘\alpha= 45^\circ and 0.9980.998 at 67.5∘67.5^\circ. Combined with their mutual approach to zero under pure crossflow at α=90∘\alpha= 90^\circ, this high correlation indicates that the mean lateral force is predominantly driven by oblique-flow-induced surface-pressure redistribution, while asymmetric wall shear stresses exert a weaker influence, consistent with the force decompositions in § 4 A.

The primary separation angles show how stranded surface topography restructures the time-averaged near-wall flow [Figs. 8(e) and 8(f)]. Here, θfl\theta_{\mathrm{fl}} denotes the windward-left separation angle, while 360∘−θfr360^\circ- \theta_{\mathrm{fr}} represents the reflected windward-right separation angle plotted on the same angular scale from the front stagnation line. For the smooth cylinder, the two separation angles coincide to within 0.1∘0.1^\circ across all inclinations, remaining bounded between 86.3∘86.3^\circ and 91.0∘91.0^\circ. On the stranded cables, by contrast, the helical geometry introduces side-to-side differences in separation. For the 3×73 \times7 cable, the windward-right separation angle 360∘−θfr360^\circ- \theta_{\mathrm{fr}} remains pinned near the first inter-strand valley between 67.7∘67.7^\circ and 68.9∘68.9^\circ; meanwhile, the windward-left angle θfl\theta_{\mathrm{fl}} shifts under oblique inflow, registering 69.9∘69.9^\circ, 98.9∘98.9^\circ, 99.2∘99.2^\circ, and 69.0∘69.0^\circ at α=22.5∘\alpha= 22.5^\circ, 45∘45^\circ, 67.5∘67.5^\circ, and 90∘90^\circ, respectively. For the 1×191 \times19 cable, both separation lines remain near 90∘90^\circ, with θfl\theta_{\mathrm{fl}} spanning 86.0∘86.0^\circ–92.4∘92.4^\circ and 360∘−θfr360^\circ- \theta_{\mathrm{fr}} spanning 92.2∘92.2^\circ–96.7∘96.7^\circ. With increasing inclination, the 1×71 \times7 cable yields paired angles (θfl,360∘−θfr)(\theta_{\mathrm{fl}}, 360^\circ-\theta_{\mathrm{fr}}) of 96.4∘/93.9∘96.4^\circ/93.9^\circ, 102.5∘/106.4∘102.5^\circ/106.4^\circ, 108.3∘/115.0∘108.3^\circ/115.0^\circ, and 105.5∘/105.5∘105.5^\circ/105.5^\circ; the corresponding 7×77 \times7 pairs evolve through 98.2∘/97.9∘98.2^\circ/97.9^\circ, 97.1∘/102.1∘97.1^\circ/102.1^\circ, 108.0∘/104.3∘108.0^\circ/104.3^\circ, and 105.5∘/105.5∘105.5^\circ/105.5^\circ. Consequently, both the 1×71 \times7 and 7×77 \times7 cables converge to the same symmetric separation angle of 105.5∘105.5^\circ under pure crossflow at α=90∘\alpha= 90^\circ; the 3×73 \times7 cable, however, maintains a pronounced separation asymmetry at α=45∘\alpha= 45^\circ and 67.5∘67.5^\circ. This circumferential offset between the primary separation lines accompanies the asymmetric surface-pressure distribution that generates the mean lateral force.

Figure 9 presents the axial phase organization of the mean surface pressure in unwrapped two-dimensional representations for all five configurations at α=45∘\alpha= 45^\circ and 67.5∘67.5^\circ. In these maps, the horizontal coordinate ϕs\phi_s traces the circumferential surface position, and the vertical axis spans one lay pitch (Lz=6DL_z = 6D). Across the configurations, the surface pressure forms configuration-specific helical patterns whose widths, inclinations, and local patch structures reflect the underlying cable geometry. For the smooth cylinder, the pressure contours remain strictly parallel to the axial direction, displaying complete axial homogeneity. For the single-strand cables (1×71 \times7 and 1×191 \times19), the field is characterized by continuous, inclined helical bands aligned with the outer wire paths. In contrast, the multi-strand cables (3×73 \times7 and 7×77 \times7) exhibit composite two-scale topologies that superimpose fine internal-wire-scale pressure variations onto the broader strand-scale variations, with the tri-lobed envelope of the 3×73 \times7 cable producing the strongest spatial modulation. Near the windward stagnation line (ϕs∼0∘\phi_s \sim0^\circ), the compression band remains nearly uniform along the axis on the smooth cylinder and the 1×191 \times19 cable, whereas the deep inter-strand valleys of the 3×73 \times7 cable fragment it into localized pressure dips that shift circumferentially along the axis. When integrated over the body, the alignment between these local pressure patches and the projected surface normal vectors contributes directly to the aerodynamic force balance, consistent with the large mean lateral force and moderate normal load of the 3×73 \times7 geometry.

These local surface characteristics link back to the global force balances established in § 4 A. Because the mean normal force is dominated by pressure, variations in windward stagnation compression, suction peak depth, and base recovery determine the net normal resistance. The lateral force, on the other hand, arises from how asymmetric pressure fields are projected through the three-dimensional surface orientation. For the multi-strand cables (3×73 \times7 and 7×77 \times7), the opposite-lay combination of the 24 mm outer-strand pitch and 12 mm internal-wire pitch alters the local surface geometry at two different axial length scales. Under oblique inflow, this dual-scale geometric modulation introduces an axial phase dependence into the surface pressure and separation fields. For example, the local separation pinning and suction extrema of the 3×73 \times7 cable generate a large mean lateral force together with a moderate normal load. The more symmetric, single-scale surface of the 1×191 \times19 cable, however, maintains a minimal mean lateral force but a higher normal resistance. These surface-pressure patterns connect the near-wall topology to the downstream wake dynamics examined in § 4 C, where the axial-phase-dependent surface structures are linked with the wake reorganization and lateral momentum transport.

Mean-wake organization, recirculation, and axial-phase-dependent transport

The global force balances in § 4 A and surface flow fields in § 4 B reveal a marked difference in the dynamic response of the cable families at intermediate inclinations. At α=67.5∘\alpha= 67.5^\circ, the 3×73 \times7 cable develops a large mean lateral force alongside a weak lateral-force fluctuation (∣Cℓ(qtot)∣=0.354\lvert C_\ell(q_{\mathrm{tot}})\rvert= 0.354 and Cℓ,rms(qtot)=0.053C_{\ell,\mathrm{rms}}(q_{\mathrm{tot}}) = 0.053); the 1×191 \times19 cable, however, generates a small mean lateral force along with a strong lateral-force fluctuation (∣Cℓ(qtot)∣=0.045\lvert C_\ell(q_{\mathrm{tot}})\rvert= 0.045 and Cℓ,rms(qtot)=0.300C_{\ell,\mathrm{rms}}(q_{\mathrm{tot}}) = 0.300). To identify the off-body flow features associated with this contrast, we examine how the helical strand-groove topography propagates into the wake, altering the transverse mean-wake displacement, recirculation-zone structure, and axial-phase-dependent momentum transport.

To characterize the near-body flow structure, the centerline metrics Umin⁡/UnU_{\min}/U_n, x0/Dx_0/D, and Lr/DL_r/D retain the definitions given in § 2 C. Because oblique inflow and helical surface grooves induce asymmetric wake deflection away from the geometric centerline (y=0y = 0), relying solely on centerline profiles is insufficient to capture the displaced deficit core. We therefore evaluate the cross-sectional flow field at x/D=2x/D = 2, introducing a nonnegative velocity deficit weight

g(y)=max⁡(1−un(y)/Un,0),(9)g(y) = \max(1-u_n(y)/U_n,0), \tag*{(9)}

where yy is the lateral coordinate and un(y)u_n(y) is the time-averaged normal velocity along x/D=2x/D = 2 in the reference section z=0z = 0. The integrated sectional deficit (the subscript 2 denoting x/D=2x/D = 2) and deficit-center displacement are evaluated as

D2=∫g(y) d(y/D),(10)D_2 = \int g(y)\,\mathrm{d}(y/D), \tag*{(10)}
ycD=∫(y/D)g(y) d(y/D)∫g(y) d(y/D).(11)\frac{y_c}{D} = \frac{\int(y/D)g(y)\,\mathrm{d}(y/D)}{\int g(y)\,\mathrm{d}(y/D)}. \tag*{(11)}

The effective lateral width and profile skewness are determined from the corresponding higher-order central moments:

(σyD)2=∫[(y−yc)/D]2g(y) d(y/D)∫g(y) d(y/D),(12)\left(\frac{\sigma_y}{D}\right)^2 = \frac{\int[(y-y_c)/D]^2g(y)\,\mathrm{d}(y/D)}{\int g(y)\,\mathrm{d}(y/D)}, \tag*{(12)}
W2σD=2σyD,(13)\frac{W_{2\sigma}}{D} = 2\frac{\sigma_y}{D}, \tag*{(13)}
γ1=∫[(y−yc)/D]3g(y) d(y/D)[∫g(y) d(y/D)](σy/D)3.(14)\gamma_1 = \frac{\int[(y-y_c)/D]^3g(y)\,\mathrm{d}(y/D)}{\left[\int g(y)\,\mathrm{d}(y/D)\right](\sigma_y/D)^3}. \tag*{(14)}

Here, the integrals are evaluated across the full transverse wake domain using the dimensionless measure d(y/D)\mathrm{d}(y/D). The four quantities D2D_2, yc/Dy_c/D, W2σ/DW_{2\sigma}/D, and γ1\gamma_1 thus separate variations in deficit magnitude, transverse displacement, lateral spread, and profile asymmetry.

Similar deficit-weighted statistical moments have been employed in wake-deflection and rotor-interaction analyses;[67] in the present formulation, they decouple lateral wake deflection from wake spreading and shape distortion.

The 3 × 7 angle sequence in Figure 10 indicates that the growth of mean lateral force and surface-pressure asymmetry is reflected in the wake as a transverse displacement of the deficit center. From 45° to 67.5°, yc/Dy_c/D increases from 0.103 to 0.173 [Figure 10(d)]. Meanwhile, the integrated sectional velocity deficit remains nearly unchanged [D2=0.862D_2 = 0.862 and 0.849, Figure 10(c)], and the centerline minimum velocity stays close to Umin⁡/Un≈−0.12U_{\min}/U_n \approx-0.12 [Figure 10(a)], while the recirculation length contracts from Lr/D=1.222L_r/D = 1.222 to 0.410 [Figure 10(b)]. Stronger lateral loading is therefore accompanied primarily by transverse deflection of the deficit core, with little change in the overall sectional velocity deficit. In addition, at the same inclination of 67.5°, the 1 × 19 cable retains a wake that remains close to the centerline (yc/D=0.010y_c/D = 0.010) due to its balanced crossflow, consistent with its weaker mean lateral force. This correspondence holds along the 3 × 7 inclination sequence; the 7 × 7 cable at 67.5°, by contrast, carries a mean lateral force of ∣Cℓ∣/Cqtot=0.231|C_\ell|/C_{\mathrm{qtot}} = 0.231 with almost no displacement of the deficit center [Figure 10(d)].

Figure 10. Mean-wake indicators across all 20 cases: (a) centerline minimum normal velocity Umin⁡/UnU_{\min}/U_n; (b) recirculation length Lr/DL_r/D; (c) integrated sectional velocity deficit D2D_2 at x/D=2x/D = 2; (d) transverse wake deficit-center displacement yc/Dy_c/D at x/D=2x/D = 2; (e) wake width W2σ/DW_{2\sigma}/D based on twice the deficit standard deviation at x/D=2x/D = 2; and (f) deficit profile skewness γ1\gamma_1 at x/D=2x/D = 2.

Under pure crossflow, the restored crossflow symmetry eliminates both the mean lateral force and the transverse wake displacement [yc/D=−0.001y_c/D = -0.001, Figure 10(d)] for the 3 × 7 cable. Concurrently, the wake develops an elongated recirculation region: the recirculation length extends to Lr/D=2.455L_r/D = 2.455 [Figure 10(b)], the centerline reverse flow deepens to Umin⁡/Un=−0.352U_{\min}/U_n = -0.352 [Figure 10(a)], and the integrated deficit and width increase to D2=1.541D_2 = 1.541 and W2σ/D=0.764W_{2\sigma}/D = 0.764 [Figure 10(c) and Figure 10(e)]. These variations reflect two wake states: oblique inflow steers the deficit transversely and shortens the recirculation length, whereas pure crossflow restores symmetry within a deeper, broader, and slower-recovering wake.

To evaluate the cumulative velocity deficit across the wake and distinguish localized profile reorganization from changes in total velocity loss, we define a planar integrated velocity deficit

Idef=∫Adefmax⁡(1−un/Un,0) dAD2,(15)I_{\mathrm{def}}=\int_{\mathcal{A}_{\mathrm{def}}}\max(1-u_n/U_n,0)\,\frac{\mathrm{d}A}{D^2}, \tag*{(15)}

where Adef\mathcal{A}_{\mathrm{def}} denotes the sampling domain spanning 0.5≤x/D≤120.5 \le x/D \le12 and ∣y∣/D≤4|y|/D \le4. Resolved directional turbulent fluctuations are quantified by the Reynolds stress tensor

Rij=⟨uiuj⟩−⟨ui⟩⟨uj⟩,(16)R_{ij}=\langle u_i u_j\rangle-\langle u_i\rangle\langle u_j\rangle, \tag*{(16)}

where ⟨⋅⟩\langle\cdot\rangle denotes time averaging, uiu_i is the instantaneous resolved velocity (elsewhere, unu_n and uℓu_\ell denote time-averaged components), and i,j∈{n,ℓ,t}i,j \in\{n,\ell,t\} represent the normal, lateral, and cable-tangential directions in the cable-fixed frame, respectively. The resolved turbulent kinetic energy per unit mass is then evaluated as

kres=Rnn+Rℓℓ+Rtt2.(17)k_{\mathrm{res}}=\frac{R_{nn}+R_{\ell\ell}+R_{tt}}{2}. \tag*{(17)}

For any flow-field quantity QQ, its area-weighted spatial average over a designated planar domain A\mathcal{A} is computed as

⟨Q⟩A=1AA∫AQ dA,(18)\langle Q\rangle_{\mathcal{A}}=\frac{1}{A_{\mathcal{A}}}\int_{\mathcal{A}}Q\,\mathrm{d}A, \tag*{(18)}

where AA≡∫AdAA_{\mathcal{A}}\equiv\int_{\mathcal{A}}\mathrm{d}A is the area of the domain. For the retained stress and energy summaries, A\mathcal{A} is the entire fluid cross section of the computational domain (46D×21D46D \times21D); these plane-averaged values are reported without additional notation and are identified as plane-averaged in the text and figure captions.

plane-averaged in the text and figure captions. The near-wake mean-flow indicator is the near-wake-averaged lateral velocity magnitude ∣uℓ∣/Un|u_\ell|/U_n, evaluated over 0.5≤x/D≤40.5 \le x/D \le4 and ∣y∣/D≤4|y|/D \le4, with the absolute value taken before area averaging. Both RijR_{ij} and kresk_{\mathrm{res}} have units of velocity squared and are normalized by Un2U_n^2. The velocity covariance and its half-trace follow the standard Reynolds decomposition,[68] applied here to the resolved velocity field.

To quantify the axial flow modulation introduced by helical stranding, Figure 11 shows the relative range across axial sections for each scalar wake indicator QpQ_p (Lr/DL_r/D, IdefI_{\mathrm{def}}, kres/Un2k_{\mathrm{res}}/U_n^2, or ∣uℓ∣/Un|u_\ell|/U_n), defined as

Figure 11. Relative range δQ\delta Q [Eq. (19)] across axial sections of four wake indicators against inflow inclination: (a) recirculation length Lr/DL_r/D; (b) velocity deficit integral IdefI_{\mathrm{def}}; (c) plane-averaged resolved turbulent kinetic energy kres/Un2k_{\mathrm{res}}/U_n^2; and (d) near-wake-averaged lateral velocity magnitude ∣uℓ∣/Un|u_\ell|/U_n. Each range is the maximum-minus-minimum difference divided by the absolute value of the arithmetic mean across the axial sections, reported as a percentage.

δQ=max⁡pQp−min⁡pQp∣Np−1∑p=1NpQp∣×100%,(19)\delta Q=\frac{\max_p Q_p-\min_p Q_p}{\left|N_p^{-1}\sum_{p=1}^{N_p}Q_p\right|}\times100\%, \tag*{(19)}

where pp indexes the axial sampling planes (Np=6N_p=6 for the 3×73\times7 cable and Np=4N_p=4 for all other stranded configurations). By normalizing the maximum section-to-section difference by the absolute value of the section-averaged mean, this metric measures the sensitivity of each wake feature to geometric phase.

At α=67.5∘\alpha=67.5^\circ, the 3×73\times7 cable exhibits strong axial variations in local recirculation and transverse flow: the relative range reaches 90.8% for the recirculation length Lr/DL_r/D and 17.0% for the near-wake-averaged lateral velocity ∣uℓ∣/Un|u_\ell|/U_n, but the planar deficit integral IdefI_{\mathrm{def}} and the plane-averaged resolved turbulent kinetic energy kres/Un2k_{\mathrm{res}}/U_n^2 vary by only 1.72% and 3.14%, respectively. The recirculation closure point shifts between x0/D=0.920x_0/D=0.920 and 1.574, spanning a streamwise distance of 0.654D0.654D across the axial sections. At the same inclination, the 1×191\times19 cable maintains uniform wake profiles along the span, with relative ranges of 6.56% for Lr/DL_r/D and 1.51% for the lateral velocity. This contrast indicates that the large lateral force on the 3 × 7 cable is accompanied by a local distortion of the recirculation zone along the span, with little change in the overall wake velocity loss. The 7 × 7 cable under pure crossflow (α=90∘\alpha= 90^\circ) shows the opposite pattern: the recirculation length remains relatively stable (5.96%), while the integrated deficit varies by 17.5%. In that case, the helical stranding manifests mainly as a section-to-section variation of the integrated deficit, with the streamwise location of the recirculation closure point remaining comparatively stable.

Figure 12(a)–Figure 12(f) show the crossflow distribution of the turbulent fluctuations associated with these global loads. At α=67.5∘\alpha= 67.5^\circ, the lateral Reynolds stress profiles (Rℓℓ/Un2R_{\ell\ell}/U_n^2) for the 3 × 7 cable form broadened, asymmetric lobes whose peak values and positions change across axial sections: peak values remain between 0.094 and 0.130, with their transverse locations shifting between y/D=−0.52y/D = -0.52 and 0.51. The 1 × 19 cable, however, maintains a sharp, centrally concentrated peak near y/D=0y/D = 0, reaching Rℓℓ/Un2=0.406R_{\ell\ell}/U_n^2 = 0.406–0.4540.454. This comparison relates the mean lateral force and the lateral-force fluctuation observed in § 4 A to the wake turbulence. For the 1 × 19 cable, turbulent fluctuations are focused along the wake centerline, reflecting organized vortex shedding that drives a strong lateral-force fluctuation (Cℓ,rms(qtot)=0.300C_{\ell,\mathrm{rms}}(q_{\mathrm{tot}}) = 0.300) without a steady sideways

Figure 12. Directional velocity fluctuations and area-weighted turbulent stresses across representative stranded cables and inclinations: (a)–(c) local lateral velocity variance Rℓℓ/Un2R_{\ell\ell}/U_n^2 at x/D=2x/D = 2; (d)–(f) local tangential velocity variance Rtt/Un2R_{tt}/U_n^2 at x/D=2x/D = 2; and (g)–(i) plane-averaged diagonal components 103Rii/Un210^3R_{ii}/U_n^2 (i=n,ℓ,ti = n,\ell,t; no summation) across axial sections.

bias. For the 3 × 7 cable, the helical grooves steer the wake sideways and disperse fluctuation energy across a broader transverse region, and the lateral-force fluctuation is lower (Cℓ,rms(qtot)=0.053C_{\ell,\mathrm{rms}}(q_{\mathrm{tot}}) = 0.053). At α=45∘\alpha= 45^\circ, the lateral fluctuation peaks of the 3 × 7 cable remain on the negative-yy side (y/D=−0.36y/D = -0.36 to −0.23-0.23). The tangential normal-stress profiles (Rtt/Un2R_{tt}/U_n^2, Figure 12(d)–Figure 12(f)) likewise show section-dependent shifts in peak position and magnitude, indicating that the helical surface redistributes the directional turbulence energy in space.

To reveal the spatial origin of this axial variability, the 3 × 7 cable at α=67.5∘\alpha= 67.5^\circ is visualized in Figure 13 as the representative case combining the largest mean lateral force, maximum transverse wake deflection, and widest variation in recirculation length. Across its six axial phase planes, the low-velocity deficit core continuously alters its shape and cross-stream position in response to the rotating strand–groove exterior contour. Concurrently, localized regions of positive and negative lateral velocity shift their positions and extent within the near wake (x/D≤3x/D \leq3). When viewed alongside the Reynolds stress profiles in Figure 12, these two-dimensional velocity fields show that the asymmetric mean lateral force is reflected in the wake as a three-dimensional restructuring of the deficit core and transverse flow, which is closely linked to the helical boundary-layer separation lines.

Cross-sectional distributions of the mean velocity field across six axial phase sections of the 3 × 7 stranded cable

Figure 13. Cross-sectional distributions of the mean velocity field across six axial phase sections of the 3 × 7 stranded cable at α=67.5∘\alpha= 67.5^\circ: (a)–(f) normalized normal velocity un/Unu_n/U_n and (g)–(l) normalized lateral velocity uℓ/Unu_\ell/U_n. Subplot titles indicate the normalized axial position z/Dz/D.

Complementing the scalar ranges in Figure 11, the spatial phase spread locates section-to-section differences in the crossflow plane and shows how far they persist downstream. It is defined as

ΔQ(x,y)=max⁡pQp(x,y)−min⁡pQp(x,y),(20)\Delta Q(x,y) = \max_p Q_p(x,y) - \min_p Q_p(x,y), \tag*{(20)}

where pp indexes the axial sections and Qp(x,y)Q_p(x,y) represents the time-averaged velocity or resolved turbulent kinetic energy at each crossflow position. As shown in Figure 14 for the four representative stranded cases, the lateral velocity spread (Δuℓ/Un\Delta u_\ell/U_n) for the 3 × 7 cable at α=67.5∘\alpha= 67.5^\circ is large throughout the near wake and exceeds that of the single-strand 1 × 7 cable. Under pure crossflow, the 3 × 7 cable develops a broader normal velocity spread (Δun/Un\Delta u_n/U_n), and the 7 × 7 cable at α=45∘\alpha= 45^\circ shows the smallest overall variation among the cases examined. Across all configurations, the velocity and resolved turbulent kinetic energy spreads decay rapidly with downstream distance: the geometric-phase imprint is concentrated in the near-wake recirculation and shear-layer zones (x/D≲3x/D \lesssim3), although for the 3 × 7 cable a normal velocity spread of order 0.1Un0.1U_n persists to x/D≈5x/D \approx5;

Axial phase-to-phase spread distributions across four representative stranded cable cases

Figure 14. Axial phase-to-phase spread distributions, ΔQ(x,y)=max⁡pQp(x,y)−min⁡pQp(x,y)\Delta Q(x,y) = \max_p Q_p(x,y) - \min_p Q_p(x,y), across four representative stranded cable cases: (a)–(d) normal velocity spread Δun/Un\Delta u_n/U_n; (e)–(h) lateral velocity spread Δuℓ/Un\Delta u_\ell/U_n; and (i)–(l) resolved turbulent kinetic energy spread Δkres/Un2\Delta k_{\mathrm{res}}/U_n^2.

farther downstream, the imprint is largely smoothed out by turbulent mixing.

In summary, the wake characteristics identified in § 4 C link the near-wall surface patterns to the differences in global loads. The prominent mean lateral force on the 3 × 7 cable is accompanied by transverse wake deflection and localized distortions of the recirculation zone associated with asymmetric boundary-layer separation, while the transverse dispersal of turbulent fluctuations coincides with a weak lateral-force fluctuation. The 1 × 19 cable, by contrast, shows no mean-wake deflection; its symmetric mean wake concentrates unsteady energy along the centerline, and its lateral-force fluctuation is strong. This comparison shows that the decoupling between the mean lateral force and the lateral-force fluctuation is reflected in the three-dimensional organization of the near wake. However, time-averaged wake profiles alone cannot reveal whether local shedding frequencies align with the global force response. Section III D therefore investigates the time-dependent aerodynamic coupling, utilizing lateral-force spectra and downstream force–probe coherence to determine how wake vortex dynamics relate to global unsteady loading.

Unsteady force signatures and wake–force coupling

The nominal twelve-wire outer layer of the single-strand 1×191 \times19 cable provides a wire-scale surface topography; the tri-lobed, three-strand 3×73 \times7 cable instead combines the helical lay angles of both the outer strands and the internal wires. The contrasting surface pressure, separation patterns, and wake topologies identified at 67.5∘67.5^\circ in § 4 B and III C motivate a direct comparison of lateral-force fluctuation amplitudes, spectral characteristics, and local force–velocity coherence. Subsequent diagnostics focus primarily on the lateral force, where the decoupling between the mean lateral force and the lateral-force fluctuation is most evident; normal and tangential loads remain part of the three-component balance. The unsteady lateral-force coefficient and its centered root-mean-square fluctuation are defined as

Cℓ(t)=Fℓ(t)qnAref,Cℓ=⟨Cℓ(t)⟩,C_\ell(t) = \frac{F_\ell(t)}{q_n A_{\mathrm{ref}}}, \qquad C_\ell= \langle C_\ell(t) \rangle,
Cℓ,rms=⟨[Cℓ(t)−Cℓ]2⟩.(21)C_{\ell,\mathrm{rms}} = \sqrt{\left\langle[C_\ell(t)-C_\ell]^2 \right\rangle}. \tag*{(21)}

Here Fℓ(t)F_\ell(t) is the instantaneous lateral force, angle brackets denote the time-averaging operator over the sampling window, and Cℓ,rmsC_{\ell,\mathrm{rms}} quantifies the overall intensity of unsteady force fluctuations around the mean. Because these coefficients are referenced to qnq_n, their values are labeled Cℓ(qn)C_\ell(q_n) and Cℓ,rms(qn)C_{\ell,\mathrm{rms}}(q_n) in Secs. III D and III E, as distinguished from the qtotq_{\mathrm{tot}}-normalized values in Secs. III A–III C.

The dominant vortex shedding frequency and common-band normalized spectral distribution are defined as

Stp=fpDUn,St_p = \frac{f_p D}{U_n},
S^ℓ(Stn)=Sℓ(Stn)∫BSℓ(s) ds.(22)\widehat{S}_\ell(St_n) = \frac{S_\ell(St_n)}{\int_{\mathcal{B}} S_\ell(s)\,\mathrm{d}s}. \tag*{(22)}

Here StnSt_n is the normal Strouhal number of Eq. (3) evaluated at frequency ff, B=[0.05,1.0]\mathcal{B} = [0.05,1.0] is the evaluation band, fpf_p denotes the frequency of the force power spectral density (PSD) maximum for 0.1≤Stn≤1.00.1 \le St_n \le1.0, which excludes the subharmonic modulation of the recirculation region, and Sℓ(Stn)S_\ell(St_n) is the one-sided PSD per unit StnSt_n, with ss serving as the integration dummy variable. Power spectral densities are evaluated using Welch’s method[69] applied to synchronized force time histories, with Hann-windowed segments of 16 384 samples, 50% overlap, and linear segment detrending (8192-sample segments for the coherence estimates below); the resulting frequency resolution is ΔStn≈0.013\Delta St_n \approx0.013–0.0350.035, depending on inclination. Converting the spectra from physical frequency (Hz) referenced to total dynamic pressure qtotq_{\mathrm{tot}} yields

Sℓ(Stn)=UnD(qtotqn)2×Sℓ,qtotf(UnDStn).(23)S_\ell(St_n) = \frac{U_n}{D}\left(\frac{q_{\mathrm{tot}}}{q_n}\right)^2 \times S^{f}_{\ell,q_{\mathrm{tot}}}\left(\frac{U_n}{D}St_n\right). \tag*{(23)}

where qtotq_{\mathrm{tot}} denotes the total dynamic pressure and superscript ff designates the one-sided PSD per unit Hz. This conversion accounts for dynamic pressure rescaling and the frequency Jacobian (df/dStn=Un/D)(\mathrm{d}f/\mathrm{d}St_n = U_n/D). Normalizing by the integrated area over the common frequency band decouples the spectral distribution from the absolute fluctuation amplitude, with band integrals computed using piecewise-linear interpolation between discrete frequency bins.

Figure 15 summarizes the unsteady lateral forces and spectral distributions across configurations and inclinations. For the smooth cylinder, a clear shedding peak is maintained near Stn≈0.205St_n \approx0.205–0.2140.214 across all inclination angles [Fig. 15(b)], reflecting stable quasi-periodic vortex shedding, consistent with the independence principle for yawed cylinders.[32, 36] A similarly concentrated narrow-band response is observed for the 1×71 \times7 and 1×191 \times19 cables, where the primary frequencies remain within Stn≈0.197St_n \approx0.197–0.2180.218. In contrast, the spectrum of the 3×73 \times7 cable is more sensitive to inclination, whereas the 7×77 \times7 cable departs from narrow-band behavior only at 22.5∘22.5^\circ (ηp=0.81\eta_p = 0.81) [Fig. 16(a)]. Although 3×73 \times7 maintains a prominent primary peak at 22.5∘22.5^\circ (Stn≈0.222St_n \approx0.222), its spectral response broadens considerably at higher inclinations (45∘45^\circ–90∘90^\circ) [Fig. 15(d)]. In particular, at 90∘90^\circ, where the lateral-force fluctuation is suppressed to Cℓ,rms(qn)=0.019C_{\ell,\mathrm{rms}}(q_n) = 0.019 [Fig. 15(a)], the shedding peak at Stn≈0.185St_n \approx0.185 becomes weaker than the broadband low-frequency content at Stn≈0.03St_n \approx0.03–0.050.05, which reflects slow modulation of the elongated recirculation region. Centerline probes beyond the recirculation zone (x/D=4x/D = 4 and 88) confirm shedding at Stn≈0.20St_n \approx0.20. This broadband redistribution coincides with a substantial reduction in force fluctuations; for instance, Cℓ,rms(qn)C_{\ell,\mathrm{rms}}(q_n) for 3×73 \times7 decreases from 0.1900.190 at 22.5∘22.5^\circ to 0.0360.036 at 45∘45^\circ [Fig. 15(a)].

To measure spectral concentration and relative harmonic content, the peak-neighborhood energy fraction ηp\eta_p and second-harmonic ratio H2H_2 are defined as

ηp=∫BpSℓ(s) ds∫BSℓ(s) ds,H2=Sℓ(2Stp)Sℓ(Stp),(24)\eta_p = \frac{\int_{\mathcal{B}_p} S_\ell(s)\,\mathrm{d}s}{\int_{\mathcal{B}} S_\ell(s)\,\mathrm{d}s}, \qquad H_2 = \frac{S_\ell(2St_p)}{S_\ell(St_p)}, \tag*{(24)}

where Bp=B∩[Stp−0.04,Stp+0.04]\mathcal{B}_p = \mathcal{B} \cap[St_p - 0.04, St_p + 0.04] with the fixed half-width 0.040.04 defining peak-neighborhood power within B\mathcal{B}, and H2H_2 compares PSD point values evaluated via linear interpolation. For the representative cases in Figure 16(a) at 67.5∘67.5^\circ, the peak-neighborhood energy fraction ηp\eta_p reaches 0.9710.971 for 1×191 \times19, compared with only 0.6280.628 for 3×73 \times7. This spectral variation is closely associated with the wake organization: the regular, spanwise-coherent shedding of 1×191 \times19 concentrates energy around the fundamental frequency, whereas the three-dimensional shear-layer interactions around 3×73 \times7 distribute fluctuations across adjacent bands and exhibit greater relative energy at the second harmonic (H2=6.11×10−3H_2 = 6.11 \times10^{-3} versus 2.11×10−42.11 \times10^{-4}).

Figure 16. Spectral energy concentration and downstream force–wake association: (a) peak-neighborhood energy fraction ηp\eta_p across inflow inclinations; (b) downstream magnitude-squared coherence γCℓ,uℓ,j2\gamma^2_{C_\ell,u_\ell,j} across all five geometries at α=67.5∘\alpha= 67.5^\circ; (c) downstream coherence evolution of the 3×73 \times7 cable across all four inclinations; and (d) unwrapped downstream probe phase lag θCℓ,uℓ,j\theta_{C_\ell,u_\ell,j} along the wake centerline for the smooth cylinder, 1×191 \times19, and 3×73 \times7 cables at α=67.5∘\alpha= 67.5^\circ.

The frequency-specific linear association of global lateral force with the lateral velocity uℓ,ju_{\ell,j} at downstream probe jj is quantified by magnitude-squared coherence,[70]

γCℓ,uℓ,j2(f)=∣SCℓuℓ,j(coh)(f)∣2SCℓCℓ(coh)(f)Suℓ,juℓ,j(coh)(f),(25)\gamma^2_{C_\ell,u_{\ell,j}}(f) = \frac{\left|S^{(\mathrm{coh})}_{C_\ell u_{\ell,j}}(f)\right|^2}{S^{(\mathrm{coh})}_{C_\ell C_\ell}(f)S^{(\mathrm{coh})}_{u_{\ell,j}u_{\ell,j}}(f)}, \tag*{(25)}

where SCℓuℓ,j(coh)(f)S^{(\mathrm{coh})}_{C_\ell u_{\ell,j}}(f) is the cross-spectral density, and SCℓCℓ(coh)(f)S^{(\mathrm{coh})}_{C_\ell C_\ell}(f) and Suℓ,juℓ,j(coh)(f)S^{(\mathrm{coh})}_{u_{\ell,j}u_{\ell,j}}(f) are the autospectral densities evaluated from synchronized time segments. On the discrete coherence frequency grid fk(coh)f_k^{(\mathrm{coh})}, the representative shedding frequency fcf_c nearest to the force spectral peak fpf_p and the corresponding cross-spectral phase lag θCℓ,uℓ,j\theta_{C_\ell,u_{\ell,j}} are identified via

kc=arg min⁡k∣fk(coh)−fp∣,fc=fkc(coh),k_c = \operatorname*{arg\,min}_{k}\left|f_k^{(\mathrm{coh})} - f_p\right|,\qquad f_c = f_{k_c}^{(\mathrm{coh})},
θCℓ,uℓ,j(fc)=arg⁡SCℓuℓ,j(coh)(fc).(26)\theta_{C_\ell,u_{\ell,j}}(f_c) = \arg S^{(\mathrm{coh})}_{C_\ell u_{\ell,j}}(f_c). \tag*{(26)}

Here, θCℓ,uℓ,j\theta_{C_\ell,u_{\ell,j}} designates the phase lag of local wake velocity fluctuations relative to the global lateral load (negative when uℓ,ju_{\ell,j} lags CℓC_\ell), as distinguished from the axial geometric phase ϕ\phi defined in (6). To continuously monitor the spatial advection of shedding structures, the phase is unwrapped along the downstream probe array (x/D=1x/D = 1–88), with an operational threshold of γ2(fc)≥0.5\gamma^2(f_c) \ge0.5 enforced for physical phase interpretation.

The resulting force–wake coherence diagnostics in Figure 16 indicate that the spatial organization of the wake depends on the exterior surface morphology and its orientation to the inflow. As shown in Figure 16(a), the smooth cylinder and single-strand cables (1×71 \times7, 1×191 \times19) maintain lateral fluctuation energy tightly concentrated within the primary shedding band (ηp≈0.85\eta_p \approx0.85–0.980.98) across all inclinations. At α=67.5∘\alpha= 67.5^\circ, this high spectral concentration is accompanied by persistent downstream linear coupling [Fig. 16(b)], with 1×191 \times19 exhibiting high coherence (γ2=0.914\gamma^2 = 0.914–0.9600.960) throughout x/D=1x/D = 1–88, corresponding to regular, spanwise-coherent Kármán vortices advecting downstream with minimal structural deformation. In contrast, the multi-strand 3×73 \times7 cable exhibits marked spatial decorrelation: coherence falls to 0.5110.511 at x/D=2x/D = 2 and remains degraded into the intermediate wake (0.5090.509 at x/D=8x/D = 8), consistent with the spectral broadening identified in Figure 16(a), where ηp\eta_p drops to 0.6280.628. This loss of coherence depends strongly on the inclination [Fig. 16(c)]: the 22.5∘22.5^\circ case retains strong wake association (γ2=0.900\gamma^2 = 0.900–0.9430.943), while the 45∘45^\circ and 90∘90^\circ cases drop below the interpretive threshold across the near wake, indicating that the integrated lateral force is only weakly coupled to the local wake velocity at these inclinations. At 90∘90^\circ, however, the centerline probes at x/D=4x/D = 4 and 88 still detect shedding at Stn≈0.20St_n \approx0.20.

Along the wake centerline [Fig. 16(d)], the unwrapped phase trajectories for the smooth cylinder, 1×191 \times19, and 3×73 \times7 cables decrease at an approximately uniform rate beyond x/D=2x/D = 2 (dθ/d(x/D)\mathrm{d}\theta/\mathrm{d}(x/D) between −91∘-91^\circ and −96∘-96^\circ over 2≤x/D≤82 \le x/D \le8), characteristic of stable vortex convection at Uc≈0.75U_c \approx0.75–0.79 Un0.79\,U_n. The phase curves of the 1×191 \times19 and 3×73 \times7 cables intersect near −500∘-500^\circ at x/D=4x/D = 4 and differ by Δθ≈52∘\Delta\theta\approx52^\circ at x/D=8x/D = 8, which may reflect a slight convective delay and delayed vortex formation associated with the helical strand grooves.

In summary, the unsteady force and coherence diagnostics in Sec. III D show two distinct aerodynamic responses associated with cable exterior morphology under moderate-to-high oblique inflow. The single-strand 1×191 \times19 construction retains a quasi-circular contour dominated by circumferentially regular, wire-scale textures. This preserves spanwise-coherent Kármán vortex shedding, which concentrates unsteady lateral energy within a narrow primary band (ηp≈0.97\eta_p \approx0.97) and yields a large lateral-force fluctuation (Cℓ,rms(qn)≈0.35C_{\ell,\mathrm{rms}}(q_n) \approx0.35) and persistent downstream force–wake coherence, while the mean lateral force remains small. The single-strand 1×71 \times7 cable likewise retains coherent shedding, but carries an appreciable mean lateral force at intermediate inclinations. In contrast, the multi-strand 3×73 \times7 geometry introduces deep helical strand-scale recesses that induce asymmetric separation and lateral wake deflection at interme-

Figure 15. Unsteady lateral aerodynamic forces and spectral characteristics across cable configurations and inclinations: (a) root-mean-square lateral-force coefficient Cℓ,rms(qn)C_{\ell,\mathrm{rms}}(q_n); (b) dominant vortex shedding frequency (peak Strouhal number) StpSt_p; (c) area-normalized lateral-force spectra S^ℓ(Stn)\widehat{S}_{\ell}(St_n) comparing 3×73 \times7 and 1×191 \times19 cables at α=67.5∘\alpha= 67.5^\circ; and (d) spectral evolution of the 3×73 \times7 cable across all four inclinations α=22.5∘\alpha= 22.5^\circ–90∘90^\circ.

diate inclinations (α=45∘\alpha= 45^\circ–67.5∘67.5^\circ). The resulting three-dimensional shear-layer interactions disrupt spanwise vortex synchronization, re-distribute fluctuation energy across adjacent frequencies, and reduce the lateral-force fluctuation (Cℓ,rms(qn)≈0.06C_{\ell,\mathrm{rms}}(q_n) \approx0.06), while a mean lateral force is maintained. These results show that the vortex-induced lateral-force fluctuation and the mean lateral force are decoupled in stranded cable aeromechanics. To resolve the spatial coherent structures governing these distinct spectral signatures at the primary shedding frequency, Sec. III E subsequently examines the modal organization of the wake using SPOD.

Modal organization at the force-nearest wake frequency

The preceding analyses show that the nominal exterior geometry and inflow inclination jointly control surface-pressure distributions, boundary-layer separation lines, and the spectral concentration of unsteady lateral forces. A key unresolved question is how these contrasting force signatures are manifested in the spatial organization of the downstream wake at the primary vortex shedding frequency. As in the preceding sections, the analysis focuses on the representative contrast between the 1 × 19 and 3 × 7 cables at α=67.5∘\alpha= 67.5^\circ. Spectral proper orthogonal decomposition (SPOD)[71, 72] is employed as an energy-optimal, frequency-resolved diagnostic. By targeting the discrete SPOD frequency closest to the force spectral peak fpf_p, the analysis extracts the dominant spatial coherent structures within the wake and evaluates their streamwise and axial coherence.

For each cross-sectional snapshot, the velocity fluctuation vector in the cable-fixed frame is denoted by v′=[u′,uz′,ut′]T\mathbf{v}' = [\mathbf{u}',u'_z,u'_t]^{\mathrm{T}}. To account for the local mesh resolution, the area-weighted inner product is defined through the diagonal matrix W=diag⁡(a1,…,aN,a1,…,aN,a1,…,aN)\mathbf{W} = \operatorname{diag}(a_1,\ldots,a_N,a_1,\ldots,a_N,a_1,\ldots,a_N), where aja_j represents the area of fluid cell jj. Following Welch’s averaged-periodogram approach[69] as adapted for SPOD,[71] the velocity fluctuations, weighted by W1/2\mathbf{W}^{1/2}, are segmented into Hann-windowed blocks with 50% overlap to construct the cross-spectral density (CSD) matrix Gk\mathbf{G}_k at discrete frequency fkf_k. The energy-optimal coherent structures are then extracted via the Fredholm eigenvalue problem:

GkΦk=ΦkΛk,r1(k)=λ1(Gk)tr⁡(Gk)(27)\mathbf{G}_k\mathbf{\Phi}_k = \mathbf{\Phi}_k\mathbf{\Lambda}_k,\qquad r_1(k) = \frac{\lambda_1(\mathbf{G}_k)}{\operatorname{tr}(\mathbf{G}_k)} \tag*{(27)}

where Λk=diag⁡(λ1,λ2,…)\mathbf{\Lambda}_k = \operatorname{diag}(\lambda_1,\lambda_2,\ldots) contains the modal energies in descending order (λ1≥λ2≥⋯\lambda_1 \ge\lambda_2 \ge\cdots), and r1r_1 represents the leading modal energy fraction quantifying the degree of low-rank spectral organization. The analysis isolates the discrete frequency bin fSPODf_{\mathrm{SPOD}} closest to the force spectral peak fpf_p. The reconstructed physical mode ψ\boldsymbol{\psi}, with the leading column of W−1/2Φk\mathbf{W}^{-1/2}\mathbf{\Phi}_k, satisfies the spatial normalization ∑jaj∣ψj∣2=1\sum_j a_j|\boldsymbol{\psi}_j|^2 = 1, so that ψ\boldsymbol{\psi} has units of m−1\mathrm{m}^{-1}; consequently, the modal fields depict normalized wavepacket topology and phase distributions, independent of the absolute fluctuation amplitude.

To assess the statistical convergence and spatial uniformity of the coherent structures, two metrics are evaluated. The temporal convergence of the leading SPOD mode across the sampling record is quantified by the first-half/second-half modal overlap:

OFL=∣ψFHWψL∣[(ψFHWψF)(ψLHWψL)]1/2(28)O_{\mathrm{FL}} = \frac{|\boldsymbol{\psi}_{F}^{H}\mathbf{W}\boldsymbol{\psi}_{L}|}{\left[(\boldsymbol{\psi}_{F}^{H}\mathbf{W}\boldsymbol{\psi}_{F})(\boldsymbol{\psi}_{L}^{H}\mathbf{W}\boldsymbol{\psi}_{L})\right]^{1/2}} \tag*{(28)}

where ψF\boldsymbol{\psi}_{F} and ψL\boldsymbol{\psi}_{L} denote the leading modes computed from the first and second halves of the Welch block ensemble, respectively. Values approaching unity indicate that the extracted mode represents a statistically stationary, converged coherent wake mode over the sampling record. Furthermore, to quantify the structural invariance along the cable axis, the modal similarity between different axial cross sections pp and qq is evaluated via the modal assurance criterion (MAC):

MACpq=∣ψpHWcψq∣2(ψpHWcψp)(ψqHWcψq)(29)\mathrm{MAC}_{pq} = \frac{|\boldsymbol{\psi}_{p}^{H}\mathbf{W}_{c}\boldsymbol{\psi}_{q}|^2}{(\boldsymbol{\psi}_{p}^{H}\mathbf{W}_{c}\boldsymbol{\psi}_{p})(\boldsymbol{\psi}_{q}^{H}\mathbf{W}_{c}\boldsymbol{\psi}_{q})} \tag*{(29)}

evaluated over the common fluid domain with area weighting Wc\mathbf{W}_c. The resulting MAC range ([0, 1]) provides a metric of spanwise coherence: values near unity denote an axially invariant shedding topology, whereas lower values indicate spatial distortion and phase decorrelation along the span.

The primary construction contrast is evaluated at α=67.5∘\alpha= 67.5^\circ. For the 1 × 19 cable, the force-nearest SPOD frequency yields a leading energy fraction of r1=0.876r_1 = 0.876 [Fig. 17(b)], a temporal overlap of OFL=0.946O_{\mathrm{FL}} = 0.946 [Fig. 17(c)], and an axial similarity range of MAC∈[0.971,0.992]\mathrm{MAC} \in[0.971, 0.992] [Fig. 17(d)]. A comparable degree of modal coherence is obtained for the 7 × 7 cable at the same inclination, yielding r1=0.905r_1 = 0.905, OFL=0.952O_{\mathrm{FL}} = 0.952, and MAC∈[0.888,0.964]\mathrm{MAC} \in[0.888, 0.964]. Compared with these two cables, the 3 × 7 cable exhibits marked spectral dispersion and spatial decorrelation, registering r1=0.437r_1 = 0.437, OFL=0.579O_{\mathrm{FL}} = 0.579, and MAC∈[0.526,0.865]\mathrm{MAC} \in[0.526, 0.865]. The spatial modal fields in Figure 18 reflect this quantitative difference: the 1 × 19 and 7 × 7 cables display compact, antisymmetrically alternating vortex cores [Figs. 18(a) and 18(b)] with focused near-wake contours [Figs. 18(e) and 18(f)]. In contrast, the 3 × 7 cable develops a broadened, asymmetric mode whose envelope lies slightly on the positive lateral side near x/D≈2x/D \approx2 and shifts toward the negative lateral side farther downstream [Figs. 18(c) and 18(g)], accompanied by its large mean lateral force (Cℓ(qn)=−0.415C_{\ell}(q_n) = -0.415) and suppressed fluctuations [Cℓ,rms(qn)=0.062[C_{\ell,\mathrm{rms}}(q_n) = 0.062, Figure 17(a)]. When the inclination is reduced to α=22.5∘\alpha= 22.5^\circ, the 3 × 7 configuration recovers dominant low-rank organization (r1=0.856r_1 = 0.856, OFL=0.944O_{\mathrm{FL}} = 0.944, MAC∈[0.944,0.977]\mathrm{MAC} \in[0.944, 0.977]), restoring symmetric alternating shedding and a concentrated focal contour [Figs. 18(d) and 18(h)], and the mean lateral-force magnitude decreases to ∣Cℓ(qn)∣=0.051|C_{\ell}(q_n)| = 0.051. This recovery indicates that modal decoherence and steady lateral deflection are associated with the deep strand-scale grooves at intermediate inclinations.

Figure 17. Modal diagnostics and aerodynamic force characteristics across representative cases at the primary wake frequency: (a) mean lateral-force coefficient Cℓ(qn)C_\ell(q_n) and centered root-mean-square Cℓ,rms(qn)C_{\ell,\mathrm{rms}}(q_n); (b) leading SPOD modal energy fraction r1r_1; (c) first-half/second-half modal overlap OFLO_{FL}; and (d) spanwise phase-MAC variation range across axial sections.

Leading SPOD lateral modal fields across representative cases

Figure 18. Leading SPOD lateral modal fields at the primary wake frequency across representative cases: (a)–(d) real part Re⁡(ψℓ)\operatorname{Re}(\psi_\ell), showing the phase-aligned coherent structures; (e)–(h) modal envelope magnitude ∣ψℓ∣|\psi_\ell|, showing the distribution of normalized modal amplitude. The black outline traces the actual cable surface envelope.

These modal diagnostics complete the chain linking the exterior surface geometry to the aerodynamic loads. For the single-strand 1 × 19 cable, the quasi-circular contour with high circumferential symmetry confines boundary-layer separation to the lateral sides, maintaining spanwise-synchronized Kármán vortex shedding; this concentrates fluctuation energy into a dominant low-rank coherent wake mode, yielding a strong lateral-force fluctuation with a minimal mean lateral force. On the other hand, the tri-lobed contour of the 3 × 7 cable features a threefold symmetry with strand-scale helical grooves that channel the crossflow and induce asymmetric local separation and lateral crossflow deflection at intermediate inclinations (α=45∘\alpha= 45^\circ–67.5∘67.5^\circ). The resulting shear-layer distortion disrupts spanwise vortex synchronization and spreads fluctuation energy across higher modes, and a mean lateral force is sustained. This interpretation is consistent with the behavior of the 7×77 \times7 cable, whose regular hexagonal envelope provides sufficient symmetry to maintain coherent vortex shedding while the cable carries an appreciable mean lateral force. It is also consistent with the recovery of shedding coherence in the 3×73 \times7 cable at the small inclination α=22.5∘\alpha= 22.5^\circ, where crossflow deflection is diminished. The aeromechanics of stranded cables under oblique inflow is thus associated mainly with cross-sectional symmetry and groove depth. By resolving how coherent wake modes either preserve spanwise synchronization or undergo lateral deflection, the force-nearest SPOD modes provide a spatial link between local surface flow deflection and the directional aerodynamic response of inclined towing cables.

Conclusions

Large-eddy simulations of a smooth cylinder and four stranded cable constructions across four inflow inclinations at Re≈1.0×104Re \approx1.0 \times10^4 indicate that exterior surface morphology and flow orientation jointly determine the three-dimensional aerodynamic response through coupled surface separation, wake reorganization, and direction-specific loading. The differences between the cable geometries are associated mainly with the cross-sectional circumferential symmetry order and the strand-scale groove depth. Cables featuring highly symmetric, quasi-circular outer envelopes and fine wire-scale surface textures (represented by the 1 × 19 construction) promote balanced crossflow and regular vortex shedding. Configurations characterized by low discrete symmetry and deep strand-scale helical grooves (represented by the 3 × 7 construction), on the other hand, present a tri-lobed cross section that induces clear lateral flow deflection and wake restructuring under oblique inflow.

The cable exterior morphology leads to distinct flow evolutions from the near-wall region into the wake. Under moderate-to-high oblique inclinations, quasi-circular envelopes confine boundary-layer separation symmetrically to the lateral flanks, sustaining spanwise-synchronized Kármán vortex shedding that concentrates fluctuation energy into a dominant low-rank coherent mode. In contrast, deep strand-scale grooves channel the local crossflow and are accompanied by asymmetric separation-line pinning and pronounced axial phase modulations; these three-dimensional shear interactions disrupt spanwise vortex coherence, deflect the recirculation zone laterally, and disperse fluctuation energy across broadband scales. The retention or breakdown of shedding coherence is linked with the intensity of groove-induced crossflow channeling: the hexagonal 7 × 7 cable preserves coherent shedding owing to its regular circumferential symmetry, while the 3 × 7 cable recovers coherence at small inclinations as crossflow steering diminishes.

These near-wall separation and wake characteristics are reflected in the integrated aerodynamic loads through the surface tractions. Force-source decompositions show that surface pressure accounts for 94.9%–98.3% of the mean normal force across all cases and dominates the mean lateral force (98% for the 3 × 7 cable at α=67.5∘\alpha= 67.5^\circ). Tangential loads remain secondary and arise from a construction-dependent balance between viscous friction and surface pressure. The mean lateral force and the lateral-force fluctuation are decoupled across configurations: quasi-circular geometries maintain balanced separation on opposing flanks, preserving small mean lateral forces while allowing regular vortex shedding to drive periodic lateral-force fluctuations. Deep-grooved configurations, however, lock in asymmetric separation that generates a persistent net transverse pressure difference and an appreciable mean lateral force, while the disruption of shedding coherence attenuates lateral-force fluctuations. The 7 × 7 cable at α=67.5∘\alpha= 67.5^\circ combines an appreciable mean lateral force with strong fluctuations, so the two quantities vary independently with construction and inclination.

For aerial towed systems employing high-aspect-ratio or long-span stranded cables with stringent towing stability requirements, aerodynamic assessments should not be confined to canonical equivalent-cylinder simplifications, and adopting refined criteria based on paired construction–inclination evaluations is recommended. For fixed rigid cable segments, the mean lateral force and the lateral-force fluctuation are decoupled. A quasi-circular construction such as 1 × 19 carries little mean lateral force but strong periodic fluctuations, whereas a deep-grooved construction such as 3 × 7 carries weak fluctuations but a relatively large mean lateral force. From an aerodynamic standpoint, constructions of the 1 × 19 type therefore suit towing applications that require a small mean lateral force for lateral stability, and constructions of the 3 × 7 type suit applications in which the lateral-force fluctuations transmitted to the towed system at moderate-to-high inclinations must be limited. The 1 × 7 and 7 × 7 constructions combine an appreciable mean lateral force with moderate-to-strong fluctuations at intermediate inclinations. These direction-specific mean and fluctuating loads provide a physical basis for the aerodynamic load models used in multibody dynamics and trajectory simulations of aerial towed systems that account for cable motion and fluid–structure coupling.

Because large-eddy simulations of long cables over many shedding cycles and many configurations remain computationally demanding, the present conclusions are drawn for fixed rigid cable segments at a single freestream Reynolds number, Re≈1.0×104Re \approx1.0 \times10^4 (Ren≈3.8×103Re_n \approx3.8 \times10^3–1.0×1041.0 \times10^4), with a spanwise periodic length of one lay pitch (6D6D), which may not capture the spanwise correlation of much longer cables. Building on the present study, multibody dynamics and trajectory simulations that incorporate the spatial coherence of high-aspect-ratio flexible cables can evaluate the effect of the identified lateral loads on operating towed systems. Wind-tunnel aeroelastic experiments on vortex-induced vibration and fluid–structure interaction at representative inclinations such as α=67.5∘\alpha= 67.5^\circ, together with fatigue tests addressing inter-wire fretting under nonstationary aerodynamic excitation, can further extend these findings.

Author Declarations

Conflict of Interest

The authors have no conflicts to disclose.

Author Contributions

Tielin Ma: Conceptualization (lead); Supervision (equal); Project administration (equal); Resources (equal); Writing – review & editing (supporting). Junlin Zhang: Methodology (lead); Software (lead); Validation (lead); Formal analysis (lead); Investigation (lead); Data curation (lead); Visualization (lead); Writing – original draft (lead); Writing – review & editing (supporting). Jingcheng Fu: Conceptualization (supporting); Funding acquisition (lead); Supervision (equal); Project administration (equal); Writing – review & editing (lead). Chongwen Jiang: Resources (equal); Supervision (supporting). Chuan Ye: Funding acquisition (supporting); Resources (supporting).

Data Availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

References

References

  1. [1]M. Kakavand and A. Nikoobin, “Modeling and stability analysis of aerially towed systems using geometric computational dynamics,” J. Guid. Control Dyn. 43(1), 73–84 (2020).DOI
  2. [2]A. R. Dehadrai, I. Sharma, and S. S. Gupta, “Three-dimensional dynamics of towed underslung systems using geometrically exact beam theory,” AIAA J. 59(4), 1469–1482 (2021).DOI
  3. [3]T. D. Stuart, J. M. Clifton, and L. V. Schmidt, “Wind-tunnel tests of an inclined cylinder having helical grooves,” AIAA J. 33(4), 665–670 (1995).
  4. [4]J. Nebres, S. Batill, and R. Nelson, “Flow about yawed, stranded cables,” Exp. Fluids 14(1–2), 49–58 (1993).DOI
  5. [5]M. Abdelhady and D. H. Wood, “Studying the flow dynamics and heat transfer of stranded conductor cables using large eddy simulations,” Phys. Fluids 33(1), 015120 (2021).DOI
  6. [6]Y. Liu, W. Yan, Y. Shen, Y. Ding, Q. Liu, and C. Wang, “Analysis of variable-length towing cable system with air–water medium,” Phys. Fluids 38(5), 053324 (2026).DOI
  7. [7]J. Wu, X. Yang, S. Xu, and X. Han, “Numerical investigation on underwater towed system dynamics using a novel hydrodynamic model,” Ocean Eng. 247, 110632 (2022).DOI
  8. [8]D. Ma, S. Wang, M. Yang, and Y. Dong, “Dynamic simulation of aerial towed decoy system based on tension recurrence algorithm,” Chin. J. Aeronaut. 29(6), 1484–1495 (2016).DOI
  9. [9]J. Cheng, X. Liu, and F. Deng, “Numerical study on parameters of the airborne VLF antenna by quasi-stationary model,” Aerospace 10(1), 29 (2023).DOI
  10. [10]L. Feng, X. Liu, and Z. F. Nio, “Optimization study of steady-state aerial-towed cable circling strategy based on BP neural network prediction,” Aerospace 11(7), 594 (2024).DOI
  11. [11]R. Wang, S. Cheng, and D. S.-K. Ting, “Numerical study of roundness effect on flow around a circular cylinder,” Phys. Fluids 32(4), 044106 (2020).DOI
  12. [12]A. Ekmekci and D. Rockwell, “Effects of a geometrical surface disturbance on flow past a circular cylinder: A large-scale spanwise wire,” J. Fluid Mech. 665, 120–157 (2010).DOI
  13. [13]M. R. Islam and A. Mohany, “Vortex shedding characteristics in the wake of circular finned cylinders,” Phys. Fluids 32(4), 045113 (2020).DOI
  14. [14]A. Pravidansam and M. S. Afzal, “Numerical investigation of flow past a circular cylinder modified with a single groove at low Reynolds number,” Phys. Fluids 35(2), 027125 (2023).DOI
  15. [15]J. Nebres and S. M. Batill, “Flow about a circular cylinder with a single large-scale surface perturbation,” Exp. Fluids 15(6), 369–379 (1993).DOI
  16. [16]M. Maduka, F. Schoefs, K. Thiagarajan, and A. Bates, “Hydrodynamic effects of biofouling-induced surface roughness – Review and research gaps for shallow water offshore wind energy structures,” Ocean Eng. 272, 113798 (2023).DOI
  17. [17]W. Cheng, D. I. Pullin, and R. Samtaney, “Large-eddy simulation of flow over a grooved cylinder up to transcritical Reynolds numbers,” J. Fluid Mech. 835, 327–362 (2018).DOI
  18. [18]I. Rodriguez, O. Lehmkuhl, U. Piomelli, J. Chiva, R. Borrell, and A. Oliva, “LES-based study of the roughness effects on the wake of a circular cylinder from subcritical to critical Reynolds numbers,” Flow Turbul. Combust. 99(3–4), 729–763 (2017).DOI
  19. [19]M. Xu, W. Hu, and Z. Jiang, “Impact of helical grooves on drag force and flow-induced noise of a cylinder under subcritical Reynolds numbers,” Phys. Fluids 36(7), 074111 (2024).DOI
  20. [20]D. Kumar and S. Roy, “Simulation of flow over a fixed and oscillating circular cylinder with different surface roughness patterns using a multi-block-multi mesh framework immersed boundary method,” Phys. Fluids 35(11), 113606 (2023).DOI
  21. [21]C. Tian, F. Jiang, B. Pettersen, and H. I. Andersson, “Vortex dislocation mechanisms in the near wake of a step cylinder,” J. Fluid Mech. 891, A24 (2020).DOI
  22. [22]J. H. Jung and H. S. Yoon, “Large eddy simulation of flow over a twisted cylinder at a subcritical Reynolds number,” J. Fluid Mech. 759, 579–611 (2014).DOI
  23. [23]H. S. Yoon, M. I. Kim, and H. J. Kim, “Reynolds number effects on the flow over a twisted cylinder,” Phys. Fluids 31(2), 025119 (2019).DOI
  24. [24]Y. Yang, H. Li, and X. Jin, “Three-dimensional wake dynamics of a twisted cylinder,” Phys. Fluids 36(7), 073629 (2024).DOI
  25. [25]A. Thakurta and R. Balachandar, “On the wake dynamics of wall-mounted helical straked cylinders,” Phys. Fluids 37(1), 015235 (2025).DOI
  26. [26]R. I. A. Simanto, S. Jin, and B.-K. Ahn, “Investigation of vortex flow fields around a circular cylinder using particle image velocimetry. i. Suppression of vortex shedding by helical strakes,” Phys. Fluids 38(6), 064111 (2026).DOI
  27. [27]H. Zhao and Y.-W. Lee, “Visualization and parametric study on vortex shedding suppression of cylindrical structures in offshore engineering using large eddy simulation,” J. Mar. Sci. Eng. 11(5), 1090 (2023).DOI
  28. [28]J. Moon and H. S. Yoon, “Effect of variable pitch on flow around a helically twisted elliptic cylinder,” AIP Adv. 10(9), 095215 (2020).DOI
  29. [29]C. Guo, H. Guo, J. Hu, K. Song, W. Zhang, and W. Wang, “Large eddy simulation of flow over wavy cylinders with different twisted angles at a subcritical Reynolds number,” J. Mar. Sci. Eng. 7(7), 227 (2021).
  30. [30]M. Xu, W. Hu, and Y. Yang, “The effect of the helical groove start angle on the flow drag and noise of a cylinder,” Phys. Fluids 36(12), 124128 (2024).DOI
  31. [31]R. Wang, S. Cheng, and D. S.-K. Ting, “Effect of yaw angle on flow structure and cross-flow force around a circular cylinder,” Phys. Fluids 31(1), 014107 (2019).DOI
  32. [32]M. Zhao, L. Cheng, and T. Zhou, “Direct numerical simulation of three-dimensional flow past a yawed circular cylinder of infinite length,” J. Fluids Struct. 25(5), 831–847 (2009).DOI
  33. [33]T. Zhou, H. Wang, S. F. M. Razali, Y. Zhou, and L. Cheng, “Three-dimensional vorticity measurements in the wake of a yawed circular cylinder,” Phys. Fluids 22(1), 015108 (2010).DOI
  34. [34]S. E. Ramberg, “The effects of yaw and finite length upon the vortex wakes of stationary and vibrating circular cylinders,” J. Fluid Mech. 128, 81–107 (1983).DOI
  35. [35]R. Wang, S. Cheng, and D. S.-K. Ting, “Numerical study of flow characteristics around a 30 degree yawed circular cylinder at Re = 10⁴,” Phys. Fluids 35(10), 105134 (2023).DOI
  36. [36]X. Lou, C. Sun, H. Jiang, H. Zhu, H. An, and T. Zhou, “Three-dimensional direct numerical simulations of a yawed square cylinder in steady flow,” J. Mar. Sci. Eng. 10(8), 1128 (2022).DOI
  37. [37]W. Han, X. W. Kang, Y. L. Wang, and X. H. Zheng, “Computation of aerial towed cable aerodynamic performances with large eddy simulation,” Appl. Mech. Mater. 105–107, 606–610 (2012).DOI
  38. [38]G. Li, M. J. Janocha, G. Yin, M. C. Ong, and W. Li, “On the importance of angle-dependent hydrodynamic coefficients in the equilibrium configuration analysis of synthetic fiber towing ropes,” Ocean Eng. 300, 117509 (2024).DOI
  39. [39]S. Schacht and S. Kosleck, “Analysis of fluid induced transverse forces on multi-strand twisted ropes,” in Proceedings of the ASME 2022 41st International Conference on Ocean, Offshore and Arctic Engineering (OMAE2022), Vol. 5B (ASME, Hamburg, Germany, 2022) p. V05BT06A020.DOI
  40. [40]K. B. Lua, T. T. Lim, S. C. Luo, and E. K. R. Goh, “Helical-groove and circular-trip effects on side force,” J. Aircr. 37(5), 906–915 (2000).DOI
  41. [41]M. Abdelhady and D. H. Wood, “An investigation of the wakes of stranded cables using particle image velocimetry,” Phys. Fluids 33(3), 035132 (2021).DOI
  42. [42]H. Christiansen, J. B. Jakobsen, J. H. G. Macdonald, G. L. Larose, and H. Bosch, “Aerodynamics of a stay cable with helical fillets - Part I: Stability and load characteristics,” J. Wind Eng. Ind. Aerodyn. 177, 376–391 (2018).DOI
  43. [43]H. Christiansen, J. B. Jakobsen, J. H. G. Macdonald, G. L. Larose, and H. Bosch, “Aerodynamics of a stay cable with helical fillets - Part II: Fluctuating load and wake characteristics,” J. Wind Eng. Ind. Aerodyn. 177, 392–404 (2018).DOI
  44. [44]L. Matejicka, H. H. Koss, and C. T. Georgakis, “Cross-flow aerodynamics of bridge cables with wire meshes,” J. Wind Eng. Ind. Aerodyn. 223, 104941 (2022).DOI
  45. [45]K. Kleissl and C. T. Georgakis, “Comparison of the aerodynamics of bridge cables with helical fillets and a pattern-indented surface,” J. Wind Eng. Ind. Aerodyn. 104–106, 166–175 (2012).DOI
  46. [46]S. McTavish, A. D’Au Teuil, and A. Raeesi, “Effect of cable surface characteristics and flow turbulence on the aerodynamic behaviour of stay cables in dry conditions,” J. Wind Eng. Ind. Aerodyn. 207, 104414 (2020).DOI
  47. [47]H. Sang, X. Wang, X. Sun, S. Ma, and F. Huang, “Numerical study on the hydrodynamic performance of wave glider umbilical cables with different cross-sectional shapes,” Ocean Eng. 325, 120800 (2025).DOI
  48. [48]A. Maleki, R. R. Mankbadi, and V. Golubev, “Spectral proper orthogonal decomposition analysis of coherent vortical structures in the wake of a rectangular cylinder,” Phys. Fluids 37(4), 045149 (2025).DOI
  49. [49]X. Shi and P. E. Sullivan, “Plasma actuator separation control investigated with spectral proper orthogonal decomposition,” AIAA J. 62(5), 1755–1774 (2024).DOI
  50. [50]YB/T 5197-2005: Steel Wire Ropes for Aeronautical Use, National Development and Reform Commission of the People’s Republic of China (2005), clauses 7.2.2 and 7.2.3 (in Chinese).
  51. [51]MIL-DTL-83420F: Wire Rope, Flexible, for Aircraft Control, U.S. Department of Defense (1997), 15 August 1997, Secs. 3.4.3.1–3.4.3.3.
  52. [52]S. Dong, G. E. Karniadakis, A. Ekmekci, and D. Rockwell, “A combined direct numerical simulation–particle image velocimetry study of the turbulent near wake,” J. Fluid Mech. 569, 185–207 (2007).DOI
  53. [53]C. Norberg, “LDV-measurements in the near wake of a circular cylinder,” in Advances in the Understanding of Bluff Body Wakes and Vortex-Induced Vibrations (BBVIV-1) (ASME, Washington, DC, 1998).
  54. [54]C. Norberg, “Effects of Reynolds number and a low-intensity freestream turbulence on the flow around a circular cylinder,” Tech. Rep. 87/2 (Chalmers University of Technology, 1987).
  55. [55]D. J. Richards, “Aerodynamic properties of the Severn crossing conductor,” Tech. Rep. RD/L/R 1111 (Central Electricity Research Laboratory, Leatherhead, England, 1962).
  56. [56]S. M. Batill, R. C. Nelson, and J. V. Nebres, “An experimental investigation of the flow field around swayed stranded cables,” Tech. Rep. CR 2210-89-1 (Naval Coastal Systems Center, Panama City, FL, 1989).
  57. [57]A. Žukauskas, “Convective heat transfer in cross flow,” in Handbook of Single-Phase Convective Heat Transfer, edited by S. Kakaç, R. K. Shah, and W. Aung (Wiley, New York, 1987) Chap. 6.
  58. [58]M. M. Zdravkovich, “Conceptual overview of laminar and turbulent flows past smooth and rough circular cylinders,” J. Wind Eng. Ind. Aerodyn. 33(1–2), 53–62 (1990).
  59. [59]M. Abalkhady and D. Wood, “Effect of thermal boundary condition on forced convection from circular cylinders,” Numer. Heat Transfer, Part A 76(6), 420–437 (2019).DOI
  60. [60]S. Cao, S. Ozono, Y. Tamura, Y. Ge, and H. Kikugawa, “Numerical simulation of Reynolds number effects on velocity shear flow around a circular cylinder,” J. Fluids Struct. 26(5), 685–702 (2010).
  61. [61]W. Sarwar and F. Mellibovsky, “Characterization of three-dimensional vortical structures in the wake past a circular cylinder in the transitional regime,” Phys. Fluids 32(7), 074104 (2020).DOI
  62. [62]D. A. Lysenko, I. S. Ertesvåg, and K. E. Rian, “Large-eddy simulation of the flow over a circular cylinder at Reynolds number 3900 using the OpenFOAM toolbox,” Flow Turbul. Combust. 89(4), 491–518 (2012).DOI
  63. [63]R. D. Henderson, “Nonlinear dynamics and pattern formation in turbulent wake transition,” J. Fluid Mech. 352, 65–112 (1997).DOI
  64. [65]H. Ali, N. B. Khan, M. Jameel, A. Khan, M. Sajid, A. Munir, A. E.-S. Ahmed, K. A. M. Alharbi, and A. M. Galal, “Numerical investigation of the effect of spanwise length and mesh density on flow around cylinder at Re = 3900 using LES model,” PLoS One 17(4), e0266065 (2022).DOI
  65. [67]G. A. Speakman, M. Abkar, L. A. Martínez-Tossas, and M. Bastankhah, “Wake steering of multirotor wind turbines,” Wind Energy 24(11), 1294–1314 (2021).DOI
  66. [68]S. B. Pope, Turbulent Flows (Cambridge University Press, Cambridge, 2000).
  67. [69]P. Welch, “The use of fast Fourier transform for the estimation of power spectra: A method based on time averaging over short, modified periodograms,” IEEE Trans. Audio Electroacoust. 15(2), 70–73 (1967).
  68. [70]G. C. Carter, “Coherence and time delay estimation,” Proc. IEEE 75(2), 236–255 (1987).
  69. [71]A. Towne, O. T. Schmidt, and T. Colonius, “Spectral proper orthogonal decomposition and its relationship to dynamic mode decomposition and resolvent analysis,” J. Fluid Mech. 847, 821–867 (2018).DOI
  70. [72]O. T. Schmidt and T. Colonius, “Guide to spectral proper orthogonal decomposition,” AIAA J. 58(3), 1023–1033 (2020).DOI
  71. [73]C. Norberg, “An experimental investigation of the flow around a circular cylinder: Influence of aspect ratio,” J. Fluid Mech. 258, 287–316 (1994).DOI
  72. [74]L. Ruan, H. Zhu, and C. Hu, “A study on vortex-induced vibration of a long flexible catenary cable in perpendicular flow,” Ocean Eng. 305, 117937 (2024).DOI
  73. [75]C. Westin and R. A. Irani, “Framework for a physics-based digital twin of a towed cable-body system,” Ocean Eng. 328, 121025 (2025).DOI
  74. [76]W. Xu, Y. Luan, Q. Han, C. J. Ji, and A. Cheng, “The effect of yaw angle on VIV suppression for an inclined flexible cylinder fitted with helical strakes,” Appl. Ocean Res. 67, 263–276 (2017).DOI
  75. [77]L. K. Quen, A. Abu, N. Kato, P. Muhammad, L. K. Tan, and H. S. Kang, “Performance of two- and three-start helical strakes in suppressing the vortex-induced vibration of a low mass ratio flexible cylinder,” Ocean Eng. 166, 253–261 (2018).DOI
  76. [78]Y. Lu, X. Yu, Y. Liao, and W. Xu, “The influence of the yaw angle on the VIV fatigue damage of a yawed cylinder with or without helical strakes,” Appl. Ocean Res. 102, 102295 (2020).DOI

Paper details

vlnc:W_a9xrne6d
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC0-1.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents