Effects of Seabed Scour on the Structural Performance and Safety of Fixed-Bottom Offshore Wind Turbine Foundations: A Review

Abstract

Offshore wind power has expanded rapidly in recent decades and is expected to continue growing through the deployment of larger turbines in deeper waters. The safe and stable operation of offshore wind turbines (OWTs) depends critically on foundation performance, which is severely threatened by seabed scour. To better understand these threats, this paper reviews recent studies on the effects of seabed scour on the structural performance and safety of fixed-bottom offshore wind turbine foundations (OWTFs), focusing on four aspects: (1) the changes induced by scour in both the seabed morphology and the soil mechanical state, which form the basis for the subsequent analyses; (2) the effects of scour on the bearing capacity and natural frequency of the foundation, covering both its static and dynamic performance; (3) the behavior of scoured foundations under long-term cyclic loading during normal operation and under transient seismic action during extreme events; and (4) the role of scour protection in enhancing structural safety. Several future research directions are also highlighted. This review offers helpful insights for assessing scour-induced risk and for designing scour protection from a structural safety perspective, thereby supporting the safe operation of OWTs.

1. Introduction

Driven by the rapid growth in energy demand and increasing environmental concerns associated with fossil fuel consumption, renewable energy has become an important component of the global energy transition due to its sustainability, cleanliness, and vast resource potential [1]. Wind energy is widely recognized as one of the most promising renewable energy sources and is commonly classified into onshore and offshore wind power. The advantages of offshore wind energy, including abundant wind resources and proximity to coastal demand centers, have promoted the rapid deployment of offshore wind turbines (OWTs) worldwide [2]. According to the report issued by the Global Wind Energy Council, the cumulative installed capacity of grid-connected offshore wind power worldwide has reached 92.3 GW over the past decade, and an additional 129 GW of offshore wind capacity will be installed globally by 2030 [3]. Figure 1 illustrates the projected expansion of global offshore wind installations, with a compound average annual growth rate of 24% until 2030 and 13% up to 2035, highlighting the strong growth potential of the offshore wind industry.

The growing demand for clean energy is driving offshore wind energy development toward deeper waters, farther offshore regions, and larger-capacity turbines [4,5]. This trend exposes OWTs to more complex environmental loads, posing greater challenges to their structural safety and reliability. The foundations are critical to the safe operation of OWTs and typically account for 20–30% of the total cost [6]. Figure 2 illustrates the main foundation types used for OWTs. These foundations are generally classified into fixed-bottom and floating systems. Among them, monopiles remain the dominant solution, accounting for approximately 70% of all commissioned offshore wind turbine foundations (OWTFs). Jackets, tripods, suction buckets, and other multi-leg or wide-shallow forms are adopted at greater water depths or under specific seabed conditions [7,8,9]. Although floating foundation systems have attracted increasing research interest [10,11], their practical deployment remains relatively limited [12]. Therefore, this review focuses exclusively on fixed-bottom OWTFs.

Scour-induced sediment loss reduces the effective embedment depth and lateral soil support, leading to decreases in both bearing capacity [14,15] and structural natural frequency [16,17], thereby compromising the safety and serviceability of OWTs. The detrimental effects of scour are exemplified by the Robin Rigg wind farm, where two monopile foundations were decommissioned after only six years of operation due to scour-induced problems [18]. Ensuring the safe operation of OWTs throughout their service life therefore requires both accurate assessment of scour-induced foundation risks and effective scour-protection design.

Some existing reviews have focused on scour mechanisms and prediction around marine foundations [19,20], while others have assessed the post-scour performance of OWTFs [21] or summarized scour-protection techniques [22,23]. These reviews, however, generally treat scour prediction, foundation performance, and protection as separate topics, and examine bearing capacity, natural frequency, cyclic response, and seismic behavior in isolation. This review instead synthesizes these strands from the single, unifying perspective of structural safety, with novelty in three respects. First, it traces how scour, long-term cyclic loading, and extreme seismic events combine—rather than simply add up—to erode the foundation safety margin, following one thread from scour-induced soil-state changes through bearing capacity, natural frequency, cyclic response, and seismic performance to protection. Second, it shows that the natural-frequency shift commonly used for monitoring may substantially underestimate the true degradation in strength, deformation, and fatigue life, and therefore explicitly decouples structural safety from frequency change: a stable frequency does not guarantee a safe foundation. Third, instead of cataloging protection techniques, it critically assesses whether and how scour protection restores the safety margin, including whether measures optimized for one performance metric may adversely affect another.

This is a narrative review. The relevant literature was identified through searches of the Web of Science and Elsevier databases using combinations of the keywords “offshore wind turbine”, “scour”, “bearing capacity”, “natural frequency”, “cyclic loading”, “seismic response”, and “scour protection”, supplemented by backward and forward citation tracking of key references. Priority was given to peer-reviewed studies published in approximately the past decade, while seminal earlier works on scour mechanisms and foundation designs were retained for context. The emphasis was on studies that address foundation structural performance and safety rather than sediment-transport hydrodynamics alone. Reflecting both engineering practice, where monopiles account for roughly 70% of installed fixed-bottom foundations, and the distribution of the available literature, the evidence reviewed here is weighted toward monopiles. Where sufficient data exist, multi-leg (jacket, tripod), suction-bucket, and composite foundations are discussed comparatively, and the comparatively thinner and less validated evidence base for these types is itself identified as a recurring knowledge gap.

The remainder of this review is organized as follows, as outlined in Figure 3. Section 2 introduces the scour process and classification, and outlines the associated changes in scour-hole geometry and soil mechanical state that form the basis for the subsequent analyses. Section 3 and Section 4 investigate the effects of scour on foundations’ bearing capacity and dynamic characteristics. Section 5 and Section 6 address scour-related safety issues under long-term cyclic and seismic loading. Section 7 assesses the effectiveness of scour-protection measures in restoring foundation safety. Finally, the key findings and future research directions are highlighted.

2. Scour Characteristics and Soil-State Modifications

2.1. Physical Process and Classification of Scour

In the marine environment, seabed scour is defined as the erosion and removal of sediment surrounding a structure caused by the action of flowing water [24]. Sumer and Fredsøe distinguished two categories of scour in offshore environments [25]. Local scour is characterized by an inverted conical scour hole around the monopile, formed by sediment erosion driven by flow disturbances generated by the structure. In contrast, global scour involves a broader lowering of the seabed caused by large-scale sediment removal. Since local and global scour modify the pile–soil system in fundamentally different ways, they should not be treated as equivalent in foundation-performance assessments.

The development of a local scour hole around a pile is driven by near-pile hydrodynamic features. As depicted in Figure 4, when ocean currents encounter piles, the interaction of downward flow and horseshoe vortices at the upstream induces bed shear stresses that surpass the critical threshold for the sediment particle incipient, followed by downstream transport driven by flow contraction and vortex shedding, which gradually leads to the development of local scour hole around the piles [26].

Furthermore, scour can be classified into clear-water and live-bed regimes according to the upstream sediment supply [27]. In clear-water scour, no sediment is supplied from upstream, and the scour depth gradually reaches an equilibrium value. In contrast, live-bed scour is sustained by continuous sediment transport, resulting in a dynamic equilibrium with fluctuating scour depth. This distinction is important because, under reversing tidal and wave–current conditions, scour evolves through alternating erosion and backfilling rather than remaining constant. From a structural-engineering perspective, however, foundation response is governed primarily by the resulting scour-hole morphology and soil mechanical state rather than by the detailed near-pile hydrodynamics.

2.2. Scour-Hole Characteristics

2.2.1. Scour-Hole Geometry

The scour hole around a pile typically exhibits an approximately inverted conical geometry, and is commonly characterized by three key parameters of scour depth (S_d), scour-hole angle (φ), and bottom width (S_wb), as illustrated in Figure 5. These parameters evolve progressively during scour development rather than remaining constant [28,29].

Scour depth is the primary parameter controlling foundation embedment and is therefore a central concern in design. Extensive studies have examined pile scour under steady currents [30], waves [31], and combined wave–current conditions [32], yielding a range of empirical prediction formulas. Whitehouse et al. [33] recommended a design scour depth of 1.38 D for monopiles (D denotes the monopile diameter), while the API [34] proposed a more conservative 1.5 D and the DNV [35] recommended 1.3 D in the absence of model test data. Taken together, these studies [32,36] suggest a maximum scour depth in the range of roughly 1.3 D–2.5 D rather than a single fixed value, reflecting the sensitivity of scour to site conditions.

The standards also differ in how scour enters the design. IEC 61400-3-1 [37] treats it in principle only, requiring that scour be accounted for or mitigated. API RP 2GEO [34] frames it as an effective-stress problem, distinguishing general (Δz_GS) from local scour (Δz_LS) and reducing CPT resistance in capacity evaluation. DNV-ST-0126 [35] carries this into the soil-reaction model, lowering the seabed for p–y and t–z curves under general scour and accounting for the hole depth, extent, and slope under local scour. The standards thus share a conservative treatment based on idealized conditions.

These empirical multiples are best regarded as conservative first-estimate envelopes rather than accurate predictors. Derived largely from idealized current or wave conditions, single-diameter rules do not fully capture the combined wave–current climates, complex or multi-leg foundations, or cohesive and layered seabeds, and reliable prediction is further limited by the many uncertain parameters governing scour development [16,38]. Recent field and numerical studies report equilibrium scour depths that deviate appreciably from these fixed multiples, and current research increasingly favors site-specific, time-dependent assessment [39,40].

From a structural perspective, scour depth represents the reduction in effective pile embedment. As scour progresses, the loss of passive soil support causes the pile to behave increasingly as a cantilever under loading. Scour width and scour-hole angle characterize the loss of surrounding soil support around the pile. Liang et al. [41] found that the foundation diameter primarily governs scour-hole width, which is commonly assumed to be about twice the scour depth at the seabed surface [42]. The scour-hole angle varies significantly with the soil type. In sandy soils, the side-slope angle is largely controlled by the submerged angle of repose and typically ranges from 30 to 35° [43]. In contrast, interparticle cohesion enables scour holes in cohesive soils to sustain much steeper side slopes [44].

2.2.2. Scour Morphology Across Different Foundation Types

Different foundation types produce distinct scour morphologies and sediment-loss patterns owing to differences in the flow fields they generate. Figure 6 shows the typical scour topographies for various foundation types.

Monopile foundations typically develop local scour holes driven by horseshoe vortices formed upstream of the pile, with side slopes approaching the sediment angle of repose. Zhao et al. [45] observed that scour depth increases with the height-to-diameter ratio up to two, beyond which it reaches a plateau. The scour-hole planform is strongly influenced by pile diameter. Small-diameter piles typically develop elliptical scour holes, whereas large-diameter piles tend to form asymmetric heart-shaped holes owing to wake-induced reverse scour [46].

For tripod foundations, scour develops around each of the three legs. The location of the maximum scour depth depends on the hydrodynamic conditions. Under clear-water scour, Yuan et al. [47] reported a maximum depth of approximately 3.5 leg diameters at the upstream leg, substantially greater than that typically observed for monopiles. Under combined wave–current conditions, however, the maximum scour shifted to the central column [48]. They further showed that scour depths around tripods are substantially greater than those around monopiles, requiring a correction factor of about 1.3 when monopile-based formulas are used. Unlike monopiles, sediment loss is distributed among three scour holes, which may progressively expand and coalesce into a larger scour feature.

A distinctive feature of jacket and other multi-column foundations is the coexistence of local and global scour. While local scour develops around individual piles, flow contraction and vortices generated by the structural members induce a broader scour depression around the entire foundation [49]. Welzel et al. [49] found that the upstream piles scour more deeply than the downstream piles. They also reported that the global scour intensifies as the relative wave–current velocity ratio U_cw increases. According to Chen et al. [54], jacket scour transitions from an inverted-cone to a concentric-ring morphology with sand ripples as U_cw decreases below 0.49.

In bucket foundations, scour is dominated by wake-vortex shedding, with the maximum scour depth occurring away from the foundation [50]. The resulting scour morphology is strongly influenced by the flow conditions. A steady flow produces two spoon-shaped scour holes behind the bucket, whereas a bidirectional flow generates four saddle-shaped holes with a maximum depth that is about 16% lower [50]. Under combined wave–current action, scour becomes more extensive, and the deepest scour region shifts from behind to in front of the foundation [51]. Consequently, bucket foundations exhibit widely distributed sediment loss and a flow-dependent scour pattern.

The umbrella suction anchor foundation (USAF) exhibits a scour morphology distinct from that of pile foundations because of its circumferential arrangement of umbrella-shaped anchor branches. Yu et al. [52] observed an axisymmetric scour pattern around the USAF, with the deepest scour occurring downstream, smaller scour holes forming between adjacent anchor branches, and a maximum sediment deposition at the downstream 45° positions. The umbrella-shaped branches also reduced the scour intensity in front of the foundation. Yang et al. [53] further showed that the scour morphology depends strongly on the loading condition. Currents produce a symmetric spoon-shaped pattern, waves generate an inverted-frustum-shaped pattern, and the combined wave–current action causes the most severe scour, with the maximum depth occurring outside the anchor branches. Consequently, scour-induced sediment loss is distributed relatively uniformly around the foundation over a broad area.

2.3. Scour-Induced Changes in Soil Mechanical Behavior

Scour alters not only the seabed geometry but also the mechanical state of the surrounding soil. The removal of overburden stress and the presence of weaker backfilled material modify pile–soil interaction in ways that cannot be captured by embedment loss alone.

First, scour-induced soil removal reduces the vertically effective stress, potentially transforming the surrounding soil from a normally consolidated (NC) to an overconsolidated (OC) state. This reduction is governed mainly by scour depth, width, and side-slope angle [55], and can extend to the soil between piles in pile-group foundations [56]. As a result, the soil shear strength and small-strain stiffness decrease, weakening pile–soil interaction. Although the response varies between sand and clay, scour fundamentally alters the mechanical state of the residual soil rather than simply lowering the seabed level. The implications for bearing capacity are discussed in Section 3.

The second effect concerns backfilled material. In reversing tidal or wave–current environments, repeated scour–backfill cycles may occur. The backfill typically comprises loose sand or silty sand with a lower strength and stiffness than the undisturbed seabed. As a result, restoring the seabed elevation does not necessarily restore pile support. Whitehouse et al. [33] suggested, based on field observations from the North Sea wind farms, that such loose backfill layers may contribute to the lower-than-predicted natural frequencies of OWTs. Because backfilled soil is more easily eroded than the in situ seabed, extreme events can rapidly remove it, causing scour depth to return to or exceed its equilibrium value. This questions the use of long-term average scour depths as a design basis. How backfill evolution and repeated scour–backfill cycles should be incorporated into these designs remains an open question.

3. Effects of Scour on the Bearing Capacity of OWTFs

Throughout this review, the effect sizes reported by a single study are presented as indicative or extreme-case values rather than design-representative ones, whereas effects supported by consistent trends across multiple studies are identified as such. Therefore. throughout the review, individual percentages should be read as indicative of the direction and potential scale of an effect, whereas the general trends corroborated across studies carry the main message. Because these magnitudes depend strongly on the soil type, foundation type, and scour configuration, the reported percentages should be read as indicators of potential severity rather than transferable design factors.

The bearing capacity of OWTFs is the fundamental prerequisite for structural safety. By removing the surrounding soil, scour reduces the effective embedment depth and foundation capacity while increasing load eccentricity, thereby directly eroding the foundation safety margin. Because bearing capacity governs not only the static safety but also the foundation response under cyclic, dynamic, and seismic loading, it provides the basis for the performance assessments presented in the following sections. This section therefore begins with a review of the effects of scour on foundations’ bearing capacities.

3.1. Bearing Capacity Under Individual Loading Conditions

Because lateral loading governs OWTF performance, most studies have focused on unidirectional lateral loading. Under this condition, scour-hole geometry and scour-induced changes in soil state are the primary controls on bearing-capacity degradation.

Scour depth is the dominant geometric parameter governing bearing capacity across a wide range of foundation types and soil conditions, whereas the effects of scour width and side-slope angle are generally limited [57,58,59,60]. Using a simplified method verified against FLAC3D analysis for a laterally loaded pile in soft clay, increasing the scour depth from zero to 8 D increased the pile-head displacement and maximum bending moment by about 300% and 200%, respectively, whereas increasing the scour width from zero to infinity increased them by only about 18% and 6% [58]. The effect of scour-hole slope angle is also limited. Steeper slopes retain more soil around the pile, slightly increasing lateral resistance and reducing pile displacement and bending moment [57,58]. The influence of scour depth is particularly pronounced for small-diameter flexible piles and is strongly direction-dependent. In an analytical p–y solution, a fourfold increase in upstream scour depth increased the maximum pile-head displacement in sand by about 5.2 times, whereas the corresponding increase for downstream scour depth was less than 67% [61].

The distinction between local and global scour is critical because, at the same scour depth, local scour generally yields a higher bearing capacity due to the additional overburden pressure provided by the remaining side-slope soil. In centrifuge model tests at a fixed scour depth of 2 D, the maximum bending moment increased by 12.5%, 21.7%, and 34.4% under narrow local scour, wide local scour, and global scour, respectively, highlighting the substantial differences among scour configurations [62]. Consequently, treating local scour as global scour or neglecting the three-dimensional scour-hole geometry can lead to systematic errors in bearing-capacity predictions, the magnitude and direction of which depend on the soil type and modeling simplification adopted (

Table 1. Assessment biases from simplified treatment of local and global scour.

Simplified Treatment	Bias Direction	Magnitude	Soil Type	Ref.
Treat local scour as global scour	Underestimates capacity	Actual loss ≈ 1/2–1/3 of predicted	Sand	[65]
Treat local scour as global scour (pile group)	Underestimates capacity	~29%	Sand	[66]
Treat local scour as global scour (pile group)	Underestimates capacity/moment	17%/8–15%	Dense sand	[67]
Neglect 3-D geometry	Overestimates pile-head displacement	49–68%	Sand	[57]
Neglect 3-D geometry	Overestimates displacement/moment	10–19%/6–8%	Soft clay	[58]

). To account for three-dimensional scour effects in engineering analyses, several equivalent parameters have been proposed. Qi et al. [63] showed through centrifuge tests that local scour provides greater lateral soil stiffness than global scour at the same relative depth, leading to the concept of an effective soil depth. Li et al. [64] subsequently introduced the equivalent scour depth and developed an empirical expression for the moment-capacity reduction factor.

Recent studies have shown that realistic scour-hole asymmetry can noticeably affect capacity predictions. Compared with a symmetric circular scour hole, a heart-shaped scour hole produces a greater reduction in lateral capacity, with the difference increasing from 0.2% at 1 D scour depth to 7.0% at 5 D [68]. Depth asymmetry dominates this effect, while width and side-slope-angle asymmetry have negligible influence [69].

Scour also affects bearing capacity through stress-history effects, as the unloading of the surrounding soil increases the overconsolidation ratio (OCR) of the residual soil—by about 200% at 1.5 m below the post-scour mudline in one numerical and theoretical analysis [70]. The consequences differ markedly between sand and soft clay. In sand, the increased OCR generally enhances bearing capacity (Figure 7a), raising the predicted lateral resistance by about 100% at a pile-head displacement of 25 mm [70], so neglecting this effect is conservative. In soft clay, by contrast, scour-induced unloading reduces bearing capacity (Figure 7b): p-y analysis based on a modified Matlock formulation predicted a 21.8–32.7% reduction in ultimate soil resistance at scour depths of 5–10 D [71], so neglecting the stress-history effect here overestimates resistance and may compromise design safety. Subsequent studies refined these scour-modified p–y models by adding correction factors for scour-hole geometry, vertical loading, and pile diameter [59,72,73]; however, He et al. [74] used three-dimensional finite-element analyses to show that the modified p–y approach of Lin et al. [71] may overpredict the stress-history effect on ultimate resistance.

3.2. Bearing Capacity Under Combined Loading Conditions

OWTFs are typically subjected to combined vertical (V), horizontal (H), and moment (M) loading from environmental actions. Bearing-capacity assessment must therefore consider combined loading effects rather than individual loading conditions alone.

For monopiles with a length-to-diameter ratio (L/D) of about five, vertical loading increases lateral capacity by only 1–3.5% under no-scour conditions and about 6% under scour conditions [62,75,76]. The increase is attributed to vertical-load-induced soil densification, which is largely confined to the upper 2 D of the soil profile and becomes more concentrated under scour owing to the reduced embedment depth [76]. A sharply different conclusion is reached by Liang et al. [72], whose analytical model, based on critical-state soil mechanics, predicted that vertical loading significantly amplifies the adverse impact of scour, growing nonlinearly with scour depth. This contradiction was instructive rather than a mere scatter as it traced to a specific methodological difference, where the studies reporting a negligible or beneficial effect [62,75,76] were centrifuge, laboratory, or numerical studies that captured the shallow soil densification under vertical load but did not represent the stress-history (unloading) changes caused by scour, while Liang et al. [72] explicitly modeled this effect and reported that whether it was included was what governed the sign and magnitude of the vertical-load response. The disagreement thus reflects a modeling choice rather than experimental uncertainty, and because no physical-model or field study has yet isolated the stress-history contribution, the net effect of vertical loading on scoured piles remains unresolved.

Failure envelopes, which define the capacity boundary in load space, have become a standard tool for assessing foundation performance under combined V–H–M loading. Increasing scour depth alters both the size and shape of the envelope: the normalized V–M and V–H envelopes contract toward the origin, while the H–M envelope expands and becomes increasingly asymmetric, its peak shifting toward negative H [60,77]. Scour therefore degrades horizontal and moment capacity far more severely than vertical capacity; in a three-dimensional finite-element analysis of a suction caisson in clay, for example, the horizontal and moment capacities each fell by roughly half while the vertical capacity dropped by only about a fifth [78]. Vertical loading itself plays a dual role, as below the vertical bearing-capacity limit, moderate vertical loads expand the envelope and increase the combined capacity, while beyond a certain threshold the envelope contracts [79,80]. Its effect on combined capacity is thus non-monotonic with an optimal range. The contrast with the limited influence of vertical loading on monopile lateral capacity arises from the fundamentally different load-transfer mechanisms of shallow foundations and slender monopiles.

3.3. Foundation-Type Sensitivity in Bearing Capacity

Scour sensitivity depends strongly on foundation geometry and structural redundancy. Monopiles and pile groups are generally more vulnerable because scour directly removes the soil providing their resistance. In contrast, jackets and other multi-member foundations are more robust owing to their load redistribution among members. Regardless of the foundation type, capacity degradation is governed primarily by scour depth rather than the lateral extent of scour (

Table 2. Sensitivity of different foundation types to scour-induced capacity reduction.

Foundation Type	Scour Depth	Capacity/Response Reduction	Key Controlling Factors and Features	Ref.
3 × 3 pile group (sand)	6 D	Lateral capacity ↓ 58–67%	More sensitive than monopile; front-row piles most loaded	[67]
Pile group (sand)	>1.5 Deff	Lateral capacity ↓ >40%	Depth > slope angle, spacing; longer piles mitigate	[66]
Pile group (clay)	0.7 → 2.8 Deff	Lateral capacity ↓ 26–46%	Local-as-global simplification overestimates by up to 44%	[81]
Tripod	piles 0 → 6 D	Uplift capacity ↓ ~26%	Scour at piles > at central column; low L/D more sensitive	[82]
Composite bucket	0.25 H (≈3 D)	Horizontal ↓ 18%, moment ↓ 22%	Depth-dominated; saturates beyond 0.67 D extent	[77]
Suction caisson (clay)	0.54 L	Horizontal ↓ 48.4%, moment ↓ 46.7%, vertical ↓ 20.1%	Depth-dominated; width >5 D ≈ full-layer scour	[78]
Monopile–friction-wheel	Sd/tw > 1	Accelerated degradation	Depth > slope > extent; accelerates beyond wheel thickness	[83,84]

Note: The downward arrow (↓) indicates a reduction.

).

Each foundation type also displays certain noteworthy mechanisms. For pile groups, scour truncates the passive soil-failure mechanism and redistributes the internal load, so that the front-row piles carry the greatest load and are the most prone to failure, while increasing the pile length effectively mitigates the loss [66,67,81]. For tripods, scour around the piles has a greater effect than scour beneath the central column, but once the two scour holes merge into a large, deep depression the capacity loss accelerates [82,85]. Notably, this tripod analysis is anchored to in situ scour measurements from the Alpha Ventus test site, one of the few field datasets available for multi-leg foundations [82]. For bucket and suction-caisson foundations, scour reduces the area of the failure envelope nonlinearly without changing its shape, and their capacity is likewise affected by soil stress history [77,78,86]. For monopile–friction-wheel composite foundations, the capacity degradation accelerates markedly once the dimensionless scour depth exceeds the thickness of the friction wheel, reflecting the rapid loss of the wheel’s enhancing effect once it is undermined [83,84].

Despite the substantial body of work reviewed above, several limitations are shared across the current literature and define the main knowledge gaps. First, the great majority of post-scour capacity assessments are numerical or analytical; independent experimental validation, particularly at the full scale or in the field, is largely absent. Second, the scour hole is usually idealized as symmetric, of known dimensions, and in equilibrium, whereas real scour is three-dimensional, asymmetric, and time-evolving; the influence of asymmetric and progressively developing scour has received comparatively little attention. Finally, much of the evidence derives from small-diameter or reduced-scale models, so its extrapolation to the very-large-diameter monopiles now entering service remains an open question.

In summary, although the influence of scour on bearing capacity varies with the loading condition and foundation type, it is consistently governed by scour depth and depends strongly on scour morphology and soil type. The distinction between local and general scour, and between sand and soft clay, cannot be neglected in assessment, as overlooking it leads to either overly conservative or unconservative designs. Scour thus progressively erodes the static bearing-capacity margin of the foundation. Beyond this quasi-static reduction, scour also alters the dynamic behavior of the structure, which is examined in the following section through its effect on the natural frequency.

4. Effects of Scour on the Natural Frequency of OWTs

The natural frequency of the support structure is a key parameter governing the dynamic response and safety of OWTs. Accurate prediction of the natural frequency is therefore essential, as resonance must be avoided throughout the service life. Current design codes require the natural frequency to remain between the rotor frequency (1P) and the blade-passing frequency (3P) [87]. As shown in Figure 8, this range corresponds to the soft–stiff design region, which is generally considered optimal for OWTs [88]. Notably, DNV-ST-0126 [35] (§7.6.5) already requires that the risk of scour be conservatively accounted for when the natural frequency is calculated, recognizing scour as a design-relevant driver of frequency change rather than a purely operational concern.

Seabed scour reduces pile embedment and lateral soil-restraint stiffness, thereby altering the dynamic characteristics of the system and lowering its natural frequency. As a result, the natural frequency moves closer to the 1P range, reducing the available resonance margin. The extent of this reduction depends mainly on scour depth, soil properties, and foundation type.

4.1. Influence of Scour Depth

Among all influencing factors, scour depth is the most direct and significant factor governing the change in natural frequency. Prendergast et al. [90] were the first to quantify this relationship through scaled and field tests, and subsequent studies have repeatedly confirmed the same trend (

Table 3. Effect of scour depth on the natural frequency of OWTs.

Turbine/Pile	Scour Depth	Frequency Reduction	Decay Form	Ref.
8.76 m steel pipe pile (field)	S = 6 m	33 Hz → ~2 Hz	—	[90]
3.6 MW monopile	1.3 D/1.66 D	6.5–8.5%/12.5%	Nonlinear with S/D	[16]
5 MW monopile	1.5 D	~8%	Monte Carlo	[91]
Monopile	2 D	7.6%	—	[92]
Monopile (wave–current flume)	S/D = 0 → 1.275	f1/f0 ↓ ~0.126	Logarithmic	[93]
Tripod suction bucket	S/D = 0.6	5.3%	Power law Δf/f0 ≈ −0.12(S/D)1.5	[94]

Note: The arrow "→" indicates a change to another value, and the downward arrow (↓) denotes a reduction.

).

Reporting the range of reduction alone, however, is insufficient for prediction; the functional form of the decay is equally important. Wave–current flume tests show that the natural frequency decreases logarithmically rather than linearly with scour depth [93], and a power-law sensitivity model has been proposed for tripod suction-bucket foundations [94]. More notably, the decay rate is not governed by S/D alone but also depends on the turbine size; at the same scour depth, smaller turbines exhibit larger frequency reductions [95].

These nonlinear, size-dependent decay patterns carry an important engineering implication: estimating the frequency reduction by linear extrapolation or a single scaling factor may substantially underestimate the resonance risk of a particular turbine at a particular scour depth. The assessment of frequency reduction must therefore be based on models that reflect the true decay form rather than on simple proportional scaling.

4.2. Influence of Soil Properties

At the same scour depth, the soil properties strongly affect the magnitude of the natural-frequency reduction. The relative density of sand provides the most typical example. The full-scale analysis of Prendergast et al. [16] showed that the frequency reduction in loose sand is markedly larger than that in dense sand. Together with the field observations at Robin Rigg [18], these scaled and full-scale studies make the scour–natural-frequency relationship the most extensively field-corroborated aspect of the topics reviewed here. At a scour depth of 1.3 D, the reduction was 8.5% in loose sand but only 6.5% in dense sand. At 1.66 D, the reduction in loose sand reached 12.5%. This difference has a direct safety implication. Abhinav and Saha [91] found that, in loose sand, a scour depth of 1.5 D drove the first natural frequency into the 1P band and created a resonance risk, whereas in dense sand, the same scour left the system within the safe band (Figure 9a). Turbines founded in loose sand therefore possess a smaller resonance safety margin and are more vulnerable to scour-induced resonance.

The two vibration modes respond differently to scour. Integrated wind–wave–soil–structure interaction (SSI) simulations for a flexible monopile show that the first natural frequency is more sensitive to scour than the second in terms of frequency reduction (Figure 9c,d) [91]. In contrast, the second mode exhibits a larger change in mode shape [91], indicating that frequency- and mode-shape-based indicators capture different aspects of scour effects.

Seabed stratigraphy and composition exert a comparable influence. In model tests on a typical layered seabed, Niu et al. [96] found that a high clay content yielded the lowest and most scour-sensitive first frequency, whereas a high silt content gave the highest and most stable one (Figure 9b), reflecting the differing balance between stiffness and damping in seabed of different compositions. Seabeds rich in clay therefore undergo larger frequency shifts and lose their safety margin more rapidly under scour.

In addition, most of the above studies treat the soil parameters as deterministic values. Prendergast et al. [98] introduced random-field theory and showed that the frequency uncertainty caused by the spatial variability of the soil is of the same order of magnitude as the frequency change caused by scour itself. This finding has important implications for natural-frequency monitoring.

4.3. Foundation-Type Sensitivity in Natural Frequency

Different foundation types differ markedly in the sensitivity of their natural frequency to scour, owing to differences in overall stiffness and configuration, forming a broad spectrum (

Table 4. Sensitivity of the natural frequency to scour for different foundation types.

Foundation Type	Scour Depth	Frequency Reduction	Sensitivity Features	Key Notes	Ref.
Monopile	1.5 D	1st: 5–12.5% (loose > dense); 2nd: ~4%	Most sensitive; may enter 1P band in loose sand; 1st > 2nd	Smallest margin in loose sand	[91,97]
Tripod	Sp = 2.6rp	1st: 0.9–1.4%/2nd: 1.5–2.1%	Far below monopile	Stable freq. ≠ overall safety: deflection ↑20%, fatigue life ↓~44%	[85]
Tripod suction bucket	S/D = 0.6	1st: 5.3%/2nd: 3.8–5.3%	Power law; ~50% of monopile	Multi-leg stress redistribution mitigates degradation	[94]
Composite caisson–pile	0.25–0.5d	Resonant freq. ↓ nonlinearly	Depth-dominated; width/slope minor	Stress history dominates low-freq. response	[99]
Jacket	Fore/aft 1.4 D/0.8 D	1st: 1.7%	Low; governed by global truss stiffness	Protection may raise freq. toward 3P	[17]

Note: The upward arrow (↑) denotes an increase, and the downward arrow (↓) denotes a reduction.

). Monopiles are the most sensitive. In loose sand, scour can even drive the first natural frequency into the 1P band. Tripod suction-bucket foundations are intermediate, as their multi-leg configuration mitigates stiffness degradation through stress redistribution, giving a frequency reduction roughly half that of a monopile. Multi-leg and composite foundations, such as tripods, composite caisson–pile systems, and jackets, exhibit the smallest reductions, typically a fraction of those for monopiles because their natural frequency is governed by the global truss stiffness rather than by the shallow soil restraint of a single pile.

A small frequency reduction, however, does not equate to overall safety. A single tripod foundation study is a telling example, as although its first natural-frequency reduction is only 0.9–1.4%, far smaller than that of a monopile, the same scour increases the below-mudline pile deflection by about 20%, raises pile stresses, and substantially shortens the fatigue life [85]. This shows that using natural frequency as the sole indicator of safety under scour can be misleading—for multi-leg foundations with high stiffness redundancy, a “stable” frequency may mask substantial degradation in local stress and fatigue. Frequency assessment must therefore complement, rather than replace, the assessment of bearing capacity and fatigue.

In summary, scour reduces the natural frequency of OWTs by weakening pile–soil stiffness, thereby narrowing the resonance margin between 1P and 3P. The reduction is governed primarily by scour depth, soil conditions, and foundation type, and is most severe for monopiles in loose sand. Natural-frequency changes therefore serve not only as an indicator of structural safety but also as a useful tool for scour monitoring.

Realizing this potential relies on structural health monitoring (SHM). In practice, the natural frequency of an operating turbine is tracked from ambient vibration measurements recorded by tower-mounted accelerometers, using operational modal analysis to extract modal frequencies without interrupting operation [15]. Two strategies are then used to infer scour depth: relating the measured frequency to scour depth through a pre-established frequency–scour-depth relationship [100,101], or treating scour depth as an unknown identified by iteratively updating a finite-element model to minimize natural-frequency residuals [102]. The appeal of this approach is that it is global and non-intrusive, inferring foundation condition without underwater inspection.

Field deployment nonetheless faces real challenges. The scour-induced frequency shift is small and must be separated from larger variations caused by operational state, damping, tides, temperature, and marine growth [103,104], while the harsh offshore environment complicates sensor survivability and data transmission [105]. Distinguishing scour from other sources of stiffness change, such as grout degradation, is a further difficulty. Frequency tracking is therefore best used as a screening indicator that triggers targeted inspection rather than a stand-alone measurement of scour depth.

Although the scour–natural-frequency relationship is the most extensively validated topic in this review, several gaps remain. The soil is often simplified through uniform stiffness-degradation factors or elastic beam idealizations, leaving soil nonlinearity and spatial variability incompletely captured. Standardized procedures for comparing foundation types are also lacking: because monopiles, tripods, and suction buckets scale differently, a consistent basis such as an effective diameter definition is needed before frequency-based criteria can be transferred across types. Finally, most methods are calibrated for current turbine sizes, and their extrapolation to the 15 MW-and-larger turbines and very soft soils now entering service requires validation.

It should be noted that the results reviewed in this section consider scour alone; the combined effects of scour with long-term cyclic loading and seismic action are discussed in the following sections.

5. Long-Term Service Performance Under Scour Conditions

5.1. Accumulated Deformation and Frequency Evolution

Throughout their service life, OWTs experience continuous cyclic loading from wind and waves while scour progressively develops around the foundation. Although both act on the same pile–soil system, they influence it in opposite ways: scour weakens pile–soil stiffness by reducing effective embedment, whereas cyclic loading can densify sand and enhance soil stiffness [106]. Their effects therefore cannot be treated independently or simply superimposed, making their interaction essential for assessing long-term foundation safety.

For OWTs, the serviceability limit state (SLS) limits the accumulated mudline rotation to 0.5° [35], whereas the ultimate limit state (ULS) corresponds to 2° [107,108]. Under cyclic loading alone, these limits are generally satisfied because progressive sand densification restores soil stiffness and stabilizes foundation rotation. Flume tests by Zhang et al. [109] showed that low-amplitude cyclic loading (ζ_b = 0.15) remained below the SLS after 36,000 cycles. When scour and cyclic loading act together, however, the safety margin is markedly reduced (Figure 10a). By decreasing effective embedment and lateral soil restraint, scour prevents rotation from stabilizing, increasing it to about 1.6–2.0 times that under cyclic loading alone. Under low-amplitude loading, the rotation that remained within the SLS under cyclic loading alone exceeded it by about 95% when scour was included. Under high-amplitude loading (ζ_b = 0.38), the coupled effect drove rotation beyond the ULS after about 30,000 cycles (Figure 10b). These results show that foundations considered safe under single-factor assessment may become unsafe under coupled loading, with failure progressing from SLS exceedance to ULS failure as the loading amplitude increases.

Further studies show that this risk is not unique to monopiles but is common across foundation types, although its magnitude differs. For jackets on large-diameter piles, the final cumulative deformation under the coupled action was about 40.8% greater than under cyclic loading alone (at ζ_b = 0.26) [110]. For a tetrapod jacket, the coupled cumulative deformation reached about 1.35–1.43 times that under cyclic loading alone; more critically, the coupling was enough to push both a low-amplitude case past the SLS and a high-amplitude case past the ULS, ultimately causing structural failure [111]. Nevertheless, jacket foundations are markedly less sensitive than monopiles: their coupled deformation amplification is far smaller (roughly 35–44% versus 58–104%), owing to their higher overall stiffness and the load redistribution among piles [111]. The long-term safety margin under coupled action therefore varies considerably between foundation types.

From an engineering perspective, identifying the onset of coupled scour–cyclic-loading risk is essential. Experiments indicate that monopiles in sand enter a critical state when the normalized scour depth exceeds one pile diameter and the cyclic load ratio is greater than 0.2, leading to rapid deformation accumulation [112]. Numerical studies have further shown that this response depends nonlinearly on scour depth: shallow scour has little effect, whereas deeper scour substantially reduces pile–soil stiffness and increases cumulative deformation, residual settlement, pile bending moment, and fatigue risk [113]. These results demonstrate that long-term safety is governed primarily by the scour depth and not simply its presence. Since scour depth itself evolves under cyclic loading [109,114], deformation under coupled conditions cannot be predicted by simply superimposing the two effects.

Besides cumulative deformation, the coupling of scour and long-term cyclic loading can also threaten resonance safety by altering the structural natural frequency. To avoid resonance, the first natural frequency must remain within the 1P–3P range throughout the service life [115]. Because scour and cyclic loading have opposing effects on pile–soil stiffness, the natural frequency evolves non-monotonically. Zhang et al. [109] showed that cyclic loading initially increased the natural frequency through soil densification, but progressive scour eventually reversed this trend as soil loss became dominant (Figure 11). Consequently, single-factor analyses may misjudge the onset of resonance.

Table 5. Reported departure of the coupled response from single-factor or superposed estimates under the combined action of scour and long-term cyclic loading, illustrating that the two effects cannot be obtained by simple superposition.

Response Quantity	Coupled Response vs. Single-Factor/Superposition	Departure Type	Ref.
Accumulated rotation (monopile)	1.5–2.0× the cyclic-only value; an SLS-compliant case exceeds SLS by ~95%	Coupled >> single factor	[109]
Cumulative deformation (LDP jacket)	~40.8% larger than cyclic-only (ζb = 0.26)	Coupled > single factor	[110]
Cumulative deformation (tetrapod jacket)	1.35–1.43× cyclic-only; drives an SLS-/ULS-compliant case past the limit	Coupled >> single factor	[111]
Coupled amplification, monopile vs. jacket	~58–104% (monopile) vs. ~35–44% (jacket)	System-dependent	[111]
Natural-frequency evolution	First rises then falls; matches neither scour-only nor cyclic-only trend	Not predictable from either factor	[109]

summarizes the reported magnitude of this non-superposition effect for scour combined with cyclic loading across foundation types, showing that the coupled response can substantially exceed—or, through competing mechanisms, fall below—the single-factor estimate.

These coupled effects also expose a limitation in current design guidance. Cyclic degradation is itself recognized in the standards: API RP 2GEO [34] notes that cyclic loading generally reduces lateral resistance below its static value, while DNV-ST-0126 [35] (§7.7.3) requires cyclic strength degradation to be considered for suction-bucket foundations. Neither standard, however, provides a quantitative procedure for the coupling of scour with long-term cyclic loading.

5.2. Fatigue Damage

Besides causing accumulated deformation, scour can significantly accelerate fatigue damage and shorten fatigue life. By reducing pile–soil stiffness and altering stress distribution, it modifies the structural dynamic response and stress concentration. Sensitivity analyses showed that scour is one of the dominant geotechnical factors governing fatigue performance; a fully developed 8 m scour hole increased the fatigue damage equivalent bending moment at the mudline by about 40%, exceeding the influence of soil stiffness and damping [116].

Existing studies consistently show that scour can significantly reduce structural fatigue life. The extent of this reduction is highly sensitive to the scour depth and is also strongly influenced by the turbine operating conditions. Under parked or low-wind-speed conditions, aerodynamic damping is limited, making the structure more vulnerable to the adverse effects of scour. A single study reported that a 2 D depth scour reduced the fatigue life at the tower base by 84.2% at 3 m/s, compared with only 14.5% at 25 m/s under the same scour condition [117]. This large reduction reflects the parked, low-wind state in that study, in which aerodynamic damping is minimal instead of a generally transferable value. The effect of scour on fatigue life also depends on the foundation type. For monopile foundations, a scour depth of 1.3 D was found to reduce fatigue life by approximately 24.1% [118]. For tripod foundations, scour significantly shortened the fatigue life of critical fatigue hotspots, particularly at the main column and brace tubular joint [85].

It is noteworthy that scour affects fatigue life through multiple mechanisms. For operating monopiles, fatigue-life reduction is governed primarily by the increase in quasi-static bending moment rather than by changes in natural frequency. In one study at a scour depth of 1.5 D, the first natural frequency changed only marginally, yet the maximum bending moment rose and the location of maximum stress shifted downward, producing a fatigue-life reduction of about 45% [119]. Fatigue assessments under scour conditions should therefore consider changes in quasi-static bending moment in addition to modal parameters.

Unlike static analyses, fatigue assessments must also account for the time-dependent evolution of scour. Conventional designs typically assume a single extreme scour depth, often leading to conservative predictions. Incorporating time-varying scour can reduce conservatism while maintaining safety; for example, an adaptive design approach based on scour evolution reduced the stress-to-fatigue-capacity ratio and enabled steel savings compared with the conventional assumption of a constant scour depth of 1.3 D [120]. Moreover, backfilling can partially restore fatigue life, as shown in Figure 12, with recovery showing an approximately linear relationship with the effective backfill depth [119].

Despite considerable progress, current fatigue assessments generally treat scour as a static or slowly varying boundary condition. How the temporal evolution of scour interacts with cyclic loading induced stiffness degradation and accumulated deformation to influence fatigue-damage accumulation remains largely unexplored. In other words, the accumulated deformation discussed in Section 5.1 and the fatigue damage in Section 5.2 are usually assessed independently, whereas their coupled evolution over the service life is still not well-understood and represents an important research need.

6. Seismic Performance Under Scour Conditions

6.1. Scour-Induced Amplification of Seismic Response

Unlike the progressive performance degradation discussed in Section 5, earthquakes are extreme, transient events whose effects are manifested mainly through the rapid amplification of structural response. By weakening the pile–soil interaction, reducing the lateral stiffness of the foundation, and lowering the natural frequency of the structure, scour amplifies the displacement, internal-force, and acceleration responses during seismic excitation. A scoured OWT therefore generally experiences larger deformation and internal forces under the same earthquake, and its seismic safety margin is correspondingly reduced. Experimental and numerical studies consistently confirm that this adverse effect occurs across a wide range of foundation types and site conditions.

The amplification has been quantified across several foundation types and is generally substantial, affecting displacement, internal force, and acceleration alike, while under unfavorable conditions, it can be very large—exceeding 100% of the tower-top displacement for a jacket in soft soil in the most extreme case reported (

Table 6. Scour-induced amplification of the seismic response of OWTs.

Foundation Type	Scour Depth	Seismic Response Amplification	Ref.
Monopile	2 D	Bending moment +~90%; pile-head displacement + ~37%(frequency −~15%)	[123]
Bucket jacket (wide shallow)	0.3 L	Horizontal displacement + ~30%; lateral acceleration + ~97%; bending moment + ~58%	[124]
Tetrapod pile jacket	2 D	Peak displacement, moment, acceleration rise to ~1.8–1.9× no-scour values	[125]
Jacket (soft soil)	3 D	Tower-top displacement +164.7%; pile shear +142%; bending moment +156%	[121]
10 MW monopile	3 D	Seismic bending moment +34.1% (frequency changes only slightly)	[122]

) [121]. More important than any single figure is a consistent qualitative pattern: scour amplifies dynamic responses more strongly than changing the natural frequency. For a 10 MW monopile, for instance, increasing the scour depth from zero to 3 D shifted the first natural frequency only slightly yet raised the maximum seismic bending moment appreciably [122], which is consistent with the conclusion of Section 4 that structural safety cannot be judged from the change in natural frequency alone.

A further reason why the first frequency alone is an incomplete indicator is that the seismic response is shaped by the full modal spectrum, and scour does not affect all modes equally. Which mode is most sensitive depends on the foundation: for the conventional flexible monopile examined in Section 4, the first frequency is more sensitive to scour than the second, whereas for a large-diameter semi-rigid monopile the opposite holds—a local scour of one diameter lowered the first frequency by only about 1% but the second by nearly 19% [126]. In both cases, the higher modes, which carry much of the seismic acceleration and pile-head force, can be shifted appreciably by scour, even when the change in the controlling first frequency is modest. Frequency-based screening based on the fundamental mode alone may therefore overlook part of the scour-induced seismic risk.

The underlying mechanism is spectral rather than purely stiffness-related. By reducing the natural frequency, scour shifts the structural vibration characteristics relative to the earthquake response spectrum. Because seismic energy is distributed over a broad frequency range, the resulting dynamic amplification depends on the proximity between the structural frequency and the dominant spectral content of the ground motion [127,128]. Thus, a frequency reduction is not inherently detrimental; its effect depends on whether it moves the structure toward or away from the dominant spectral band. Accordingly, seismic safety assessment should consider ground-motion spectral characteristics rather than frequency changes alone.

This coupling also lies beyond current design provisions. API RP 2GEO [34] (§6.3.2) requires that foundation instability and liquefaction be investigated in seismically active regions, referring detailed evaluation to API 2EQ. These provisions, however, address seismic ground failure on an unscoured seabed; they do not account for how scour modifies the seismic response or shifts the timing of liquefaction. The absence of a quantitative scour–liquefaction interaction criterion, as discussed further in Section 8, thus represents a specific gap between the coupled behavior reported here and existing design guidance.

6.2. Key Factors Governing the Seismic Response Under Scour

The seismic response of scour-affected OWTFs is governed by a combination of geometric, structural, loading, and geotechnical factors.

Scour geometry is the primary factor controlling the severity of the seismic response. Different scour patterns modify the foundation boundary conditions in different ways. Compared with local scour, global scour generally produces greater degradation of soil restraint and therefore causes a much stronger increase in seismic demand, in centrifuge shaking-table tests complemented by finite-element modeling, the measured maximum pile bending moment under global scour exceeded that under local scour by approximately 80% at larger scour depths [129]. Among the geometric parameters, scour depth has the greatest influence, whereas scour width and scour-hole angle become influential only within certain frequency ranges [117,130].

The relative importance of structural inertia and soil kinematic interaction is another governing mechanism. As the stiffness of the foundation system decreases, the dominant source of seismic response gradually shifts. Slender and relatively flexible foundations are primarily inertia-controlled, with the first vibration mode governing the structural response. In contrast, foundations with higher lateral stiffness or larger characteristic dimensions become increasingly affected by soil kinematic interaction and higher-mode participation [126,129]. Consequently, evaluating scour-induced seismic behavior solely through the fundamental vibration mode may underestimate the dynamic response of modern OWTFs with complex dynamic characteristics.

The effect of scour is further influenced by load combinations. By reducing the lateral stiffness of the soil–foundation system, scour modifies the natural frequency of the entire structure and changes the relative contributions of wind, wave, and earthquake loading. As scour deepens, the influence of environmental loading becomes increasingly significant, while seismic loading continues to dominate acceleration-related responses [131]. Consequently, different response quantities exhibit different controlling mechanisms: displacement and bending moments are generally governed by the combined wind–wave actions, whereas acceleration remains more sensitive to earthquake excitation [132]. These observations highlight the necessity of evaluating scour under fully coupled wind–wave–earthquake loading rather than considering individual load components independently.

The stress history of the surrounding soil also affects the dynamic response after scour. The removal of the overburden induces unloading of the remaining soil, thus increasing the overconsolidation ratio and modifying both stiffness and strength. Ignoring this stress-history effect may lead to overestimation of the natural frequency by approximately 4–8% and pile-head displacement by 7–16%, particularly in soft clay [133,134]. These findings indicate that representing scour simply by removing all soil above the scoured seabed may produce overly conservative predictions because the residual stress state continues to influence soil stiffness.

The interaction between scour and liquefaction is highly nonlinear, and the available evidence indicates that its direction is complex rather than uniformly adverse. For monopiles in sand, the amplifying effect of scour on the seismic response has been reported to be less pronounced in liquefied than in non-liquefied soils, because the already low stiffness and strength of liquefied soil leave little additional margin for scour to degrade [135]. Consistent with a non-monotonic interaction, centrifuge tests show that the soil remaining above a scour hole can suppress excess pore-pressure build-up and enhance liquefaction resistance [123], while shaking-table tests find that although deeper scour advances the onset of liquefaction, it also accelerates pore-pressure dissipation and reconsolidation [136]. The consequences further vary considerably among foundation systems [124,136], indicating that foundation configuration plays an important role. Taken together, these studies have shown that the scour–liquefaction interaction is genuinely important, but its net severity has not yet been established.

Overall, existing studies have indicated that the seismic response of scour-affected OWTFs is controlled by multiple interacting factors. Therefore, reliable seismic assessment requires these factors to be considered simultaneously rather than in isolation.

6.3. Seismic Response of Scoured Foundations of Different Types

The influence of scour on the seismic response of OWTs differs markedly with foundation type and can be categorized into two contrasting patterns. For slender pile-supported foundations, including monopiles, pile groups, tripods, and jackets, scour generally amplifies the seismic response, with differences mainly in the degree of amplification and the most affected structural components. In contrast, scour may reduce the seismic response of wide-shallow bucket foundations despite increasing their wind–wave response. This contrast is commonly attributed to the position of the fundamental frequency relative to the dominant frequency band of the ground motion: scour-induced frequency reduction tends to move the already low-frequency slender foundations toward resonance, whereas the higher-frequency bucket foundations remain well-separated from the dominant spectral band, which is consistent with the spectral mechanism discussed in Section 6.1.

For monopiles, scour generally increases displacement, bending moment, and acceleration, with these changes exceeding the accompanying reduction in natural frequency. In a numerical model of a 10 MW monopile, the maximum bending moment under combined wind–wave loading increased by up to 38.8% at the maximum scour depth considered [122]. Higher-order modes also become increasingly important for large-diameter monopiles. Yuan et al. [126] showed that scour reduced the second natural frequency by nearly 19%, compared to only about 1% for the first mode, whereas pile-head acceleration increased markedly but superstructure acceleration changed little. These findings suggest that seismic assessment based solely on the fundamental mode may be inadequate for modern large-diameter monopiles.

Pile-group foundations generally exhibit smaller scour-induced seismic amplification than single piles. Shaking-table tests showed that the pile-head acceleration of a rock-socketed pile group increased gradually and almost linearly with scour depth, whereas that of a comparable single pile rose abruptly once the anchorage ratio reached about 4.6 [136]. The pile group also exhibited greater liquefaction resistance and more stable deformation, with the difference becoming more pronounced as scour deepened. Although deeper scour advanced the onset of liquefaction, it also accelerated excess pore-pressure dissipation and subsequent soil reconsolidation by reducing the overburden pressure.

For jacket and tripod foundations, scour has only a limited effect on the fundamental frequency. For example, the frequency of a tetrapod piled jacket decreased by less than 1% at a scour depth of two pile diameters, compared with reductions exceeding 8% for monopiles [125]. Nevertheless, the seismic response increased substantially: the same jacket exhibited peak displacement, bending moment, and acceleration about 1.8–1.9 times those of the unscoured condition. The critical vulnerabilities were concentrated in local components. Under seismic loading, windward and leeward piles underwent alternating uplift and settlement, producing transient global tilting with small, residual displacement, while the leeward piles consistently developed larger bending moments [137]. For semi-rigid pile jackets, punching-shear failure at the pile tip became more pronounced under deep scour, and the coupled wind–wave–earthquake condition produced the most severe response, leading to the recommendation of emergency turbine shutdown during earthquakes [138]. Tripod foundations exhibited a similar frequency response, but combined loading caused foundation rotation, with one pile in compression and the others in uplift, and both horizontal displacement and rotation increased nonlinearly with scour depth.

Wide-shallow bucket foundations exhibit the opposite trend. Owing to their higher fundamental frequency than comparable monopiles, scour may reduce the seismic response by weakening bucket–soil kinematic interaction and increasing energy dissipation. For a suction-bucket foundation in clay, a scour depth of 0.2 D reduced the peak seismic bending moment and tower-top displacement by roughly a fifth to a quarter, while modestly increasing the rotation and lateral displacement under wind–wave loading [139]. This opposite response reflects the spectral characteristics of the earthquake. Liquefaction is mainly concentrated outside the buckets because lateral confinement suppresses pore-pressure build-up within them, and increasing soil density reduces the liquefied zone by up to about 70% [124]. This benefit is not universal: a wide-shallow multi-bucket jacket still exhibited increases of about 30%, 97%, and 58% in maximum displacement, lateral acceleration, and bending moment under scour [124]. This contrast is instructive rather than contradictory, as the apparent seismic benefit is not intrinsic to bucket foundations but reflects where the fundamental frequency sits relative to the dominant spectral band, and is thus contingent on configuration, soil, and input motion. As each finding rests on a single numerical study of one geometry, the reported response reduction for wide-shallow buckets should be treated as a case-specific possibility rather than a design rule. Foundation configuration also influences performance, one study suggests that a five-bucket layout may provide greater overturning and seismic resistance and lower scour fragility while using markedly less steel than a tripod–bucket configuration [140].

Overall, the seismic response of a scoured foundation depends on its position within the ground-motion spectrum and its load-transfer mechanism. Consistent with the spectral mechanism discussed in Section 6.1, scour may reduce the seismic response of wide-shallow bucket foundations but generally amplifies that of pile-supported foundations. The critical vulnerable components also vary with foundation type, involving overall structural response in monopiles, pile–soil interfaces in pile groups and jackets, leeward piles in multi-leg foundations, and external soil liquefaction in bucket foundations. These differences indicate that seismic safety should be assessed using fully coupled wind–wave–earthquake analyses rather than changes in natural frequency alone.

Across the preceding sections, a foundation’s sensitivity to scour is governed largely by a single variable—its structural redundancy and global stiffness. Multi-leg and composite systems resist capacity loss through load redistribution (Section 3.3), retain their natural frequency because it is set by the global truss stiffness rather than by the soil restraint on a single pile (Section 4.3), show smaller coupled deformation amplification than monopiles (about 35–44% versus 58–104%, Section 5.1), and, for wide-shallow buckets, may even see their seismic response reduced (Section 6.3). This robustness, however, carries a hazard specific to non-monopile foundations: the same stiffness that stabilizes the global frequency decouples it from the local demand. A tripod whose fundamental frequency falls by only 0.9–1.4% can still sustain a ~20% increase in below-mudline deflection and a marked loss of fatigue life [85], and a tetrapod jacket frequency changes by less than 1% while its peak local response reaches 1.8–1.9 times the unscoured value [125]. Frequency tracking, a viable scour screen for monopiles (Section 4), may therefore be systematically blind to scour damage in multi-leg foundations, where the critical degradation lies in local stress and fatigue rather than in global modal properties. Scour assessment for these foundations must rely on direct capacity, deformation, and fatigue based checks.

6.4. Seismic Fragility Analysis

Because both scour development and earthquake intensity are inherently uncertain, deterministic analyses based on a single scour scenario cannot fully capture the seismic risk of OWTs. Fragility and reliability analyses, which quantify structural safety in terms of the probability of exceeding a specified limit state, have therefore become an important complement to the deterministic assessments discussed above.

Two elements are central to such analyses. The first is the selection of an appropriate seismic intensity measure. Conventional peak ground acceleration (PGA) has limited capability in predicting the seismic demand of scoured structures, whereas broadband spectral acceleration, which accounts for higher-mode contributions, provides a more reliable indicator of structural response [141]. This finding is consistent with the spectral mechanism discussed in Section 6.1. The second is the probabilistic treatment of scour depth and earthquake intensity. By considering both as random variables, structural safety can be represented by a fragility surface defined in the scour–earthquake space, forming the basis of multi-hazard reliability frameworks that jointly account for scour and seismic loading [142]. The large computational demand of probabilistic analyses has motivated the development of efficient assessment methods, including lumped-parameter models, the dynamic stiffness method, and rapid fragility-estimation techniques, which substantially reduce computational cost while maintaining acceptable accuracy [128,143,144].

Such probabilistic analyses reveal two findings that are obscured by deterministic methods. First, the effect of scour on seismic safety is strongly nonlinear, with the failure probability rising sharply once the scour depth reaches about 0.85 D in one fragility analysis [141]; this suggests that a foundation considered safe at moderate scour may deteriorate rapidly with only a small further increase, although the specific threshold is likely to be case-dependent. Second, the serviceability limit state is more sensitive to scour than the ultimate limit state [138], indicating that scour-induced deterioration tends to manifest first as excessive deformation rather than as a loss of bearing capacity. This echoes the conclusion of Section 5 that scour erodes the serviceability margin before the ultimate one, and suggests that serviceability criteria may govern the seismic safety of scour-affected turbines.

Unlike the progressive degradation under long-term cyclic loading, the effect of scour on seismic safety appears mainly as a transient amplification of the structural response, an amplification that is strongly nonlinear and spectrally dependent, so that frequency change alone is insufficient to judge seismic safety. Notably, under both long-term cyclic and seismic loading scour erodes the serviceability margin first, a consistency across sections that points to the priority of serviceability criteria in safety assessment.

These insights notwithstanding, the seismic evidence shares several methodological gaps. Foremost is the soil constitutive model, as most studies adopt simplified formulations, such as Mohr–Coulomb, linear-elastic, or total-stress models, that cannot reproduce cyclic pore-pressure accumulation, hydro-mechanical coupling, or liquefaction, so the scour–liquefaction interaction discussed in Section 6.2 remains only partially captured. The evidence is further constrained by scale and modeling limits, with centrifuge tests restricted by container size and numerical models by computational cost, leaving extrapolation to the very-large-diameter and multi-leg foundations now entering service uncertain. Although probabilistic and fragility-based methods (Section 6.4) are beginning to address the uncertainty in scour and seismic demand, structural-parameter uncertainties are not yet systematically incorporated. Advanced effective-stress soil models capable of representing liquefaction, validation across foundation scales, and fuller uncertainty quantification are therefore key priorities.

Beyond addressing these modeling and validation gaps, a complementary line of research seeks to mitigate the seismic demand on scoured foundations directly. A substantial body of work has used structural-side devices—tuned-mass dampers, tuned liquid column dampers, and hybrid isolation–damping systems—to reduce the acceleration, shear, and bending-moment demand of monopile- and jacket-supported turbines under seismic and combined environmental loading [145,146,147]. Comparatively little of this work, however, has considered vibration control under scour. This gap matters because scour alters the foundation’s dynamic characteristics, so that progressive scour gradually detunes passive control devices and erodes their effectiveness [148]; the same frequency shift, as shown above to be an unreliable indicator of safety, therefore also undermines the tuning on which passive dampers depend. The few available studies show that scour affects both the optimal design of tuned-mass dampers and the resulting fatigue life [118], that scour-induced frequency shifts can drive adaptive tower-damping control [149], and that semi-active devices may retain effectiveness as scour develops [148]. This interaction nevertheless remains insufficiently understood, and represents a promising direction complementary to the geotechnical scour-protection measures discussed in Section 7.

7. Effect of Scour Protection on Structural Safety

Previous sections have shown that scour can progressively impair the bearing capacity, dynamic performance, and seismic safety of OWTFs. To mitigate these adverse effects, a range of scour countermeasures have been developed. These can be broadly grouped into armoring measures that cover the seabed around the foundation to resist erosion, such as rock armor and protective mattresses, and ground-improvement measures that modify the surrounding soil to suppress scour and strengthen the pile–soil interaction, such as grouting and solidified-soil backfilling [22,23,150]. However, existing studies have primarily focused on the effectiveness of these measures in reducing scour, while their ability to restore structural performance and safety margins remains insufficiently understood. This section therefore evaluates the role of scour protection in restoring structural performance and safety margins from the perspective of structural safety.

Among design standards, monopile scour protection is addressed most directly by DNV-RP-0618 [151], the dedicated recommended practice for rock protection. It provides a systematic framework covering armor-layer external stability (via the Shields parameter), internal stability and filter criteria, falling-apron flexibility, edge-scour prediction, and sinking estimation. Notably, its listed purposes for protection—maintaining lateral soil capacity, keeping the natural frequency above a design threshold, and limiting pile deflection—are structural rather than purely erosion-related, aligning with the structural-safety perspective adopted here.

7.1. Recovery of Bearing Capacity and Reliability Through Scour Protection

The benefit of scour protection is most directly evident in the recovery of static bearing capacity, with a range of measures able to compensate effectively for the performance lost to scour (

Table 7. Recovery of foundation bearing performance by scour protection.

Protection Measure	Improved Indicator	Magnitude	Ref.
Rock armor (2 D layer, 30 kPa surcharge)	Lateral displacement reduced	~41%	[152]
Rock armor (5 D, dense sand)	Lateral capacity increased	>30%	[153]
Grouting	Ultimate lateral capacity increased	~228.5%	[154]
Solidified-soil backfill	Bearing capacity restored	Up to or above pre-scour level	[155]
Vertical protection layer	Vertical capacity increased	5–23% (with width and thickness)	[156]
Rock armor (2 m thick)	Failure probability reduced	~7.6% to 0.026%	[157]
Solidified soil	Capacity loss limited	<15% (vs. ~22.2% unprotected)	[158]

).

The mechanisms differ in emphasis. Rock armor restores lateral capacity by compensating for the scour-induced loss of embedment and enhancing the soil restraint around the pile [159]. Grouting and solidified soil raise capacity by improving the surrounding soil properties and suppressing scour development; in one study, grouting reduced the ultimate scour depth to roughly one-sixth to one-third of the unprotected value [160], and notably, even when the downstream scour depth of a grouted pile is similar to that of an ungrouted one, its lateral capacity remains markedly higher, indicating that the benefit of grouting arises not only from the reduced scour but also from the altered pile–soil interaction [154]. For foundations already damaged by scour, solidified-soil backfilling can restore the embedment and soil resistance, allowing the capacity to recover to or even exceed the pre-scour level [155].

More importantly, the recovery of bearing capacity translates directly into improved structural reliability and design margin. A 2 m thick rock armor layer has been shown to reduce the failure probability of a monopile from about 7.6% to 0.026%, with a marked increase in the reliability index [157]; moreover, by exploiting the favorable effect of protection, a shorter pile can be used while maintaining the same bearing performance, with the L/D reduced from five to 4.6, equivalent to gaining a larger safety margin [152]. Protection also curbs the long-term degradation of bearing performance, as the vertical capacity and stiffness of an unprotected monopile may fall by about 22.2% and 68.7% respectively over a 20-year service period, whereas solidified-soil protection limits the capacity loss to less than 15% and substantially improves stiffness retention [158]. Scour protection therefore not only restores the instantaneous bearing capacity but also maintains the bearing safety margin of the foundation throughout its service life.

7.2. Mechanisms and Scale Dependence of Scour-Protection Effects

Although scour protection generally improves foundation bearing performance, the magnitude of the improvement varies considerably and is jointly governed by the underlying mechanism, the foundation scale, and the timing of deployment.

Two explanations have been proposed for the benefit of scour protection. One attributes the improvement to the increase in effective stress caused by the additional surcharge, the “stress mechanism”, whereas the other emphasizes the direct resistance provided by contact between the protection material and the pile, the “material mechanism” [152,156]. Li et al. [152,156] established two representative finite-element models, as shown in Figure 13.

The two lead to noticeably different predictions, as in these representative models, the material mechanism gave an average stiffness increase on the order of 20% greater than the stress mechanism, and the associated reliability analyses indicated higher reliability indices when the protection layer was in close contact with the pile [152,157]. Other studies, however, found that the stiffness enhancement arose mainly from the increase in ground stress due to the self-weight of the armor layer rather than from the material stiffness or shear strength, and that this enhancement persisted even after a gap formed between the pile and the layer [161]. This divergence indicates that the assessment of protection benefits depends strongly on the assumed contact mechanism, since neglecting the material pile contact may underestimate the actual contribution, whereas over-relying on it may overestimate the long-term benefit once a gap develops.

Weighing the evidence on each side, the balance of currently available studies leans toward the stress mechanism, although the question cannot yet be regarded as settled. Flume experiments distinguished the two explicitly: reversing the loss of lateral stiffness requires the protection layer to be stiff and firmly compacted against the pile (a contact-dependent effect), whereas reversing the loss of effective stress requires only the surcharge provided by the protection material [162]. The stress-based effect has been reproduced consistently across independent studies—centrifuge tests [153], large parametric numerical analyses [163], and finite-element models with model-test validation [161,164]—all attributing the stiffness gain primarily to the increased confining stress from the weight of the protection layer. Centrifuge evidence further showed that, under cyclic loading in sand, densification prevented a gap from forming at the pile–protection interface [153], so the contact-dependent contribution is difficult to rely on in the long term. The material–contact mechanism, quantified through a contact coefficient in finite-element models [152], therefore appears to be an additional condition-dependent contribution rather than the primary source of benefit. Independent field-scale validation nonetheless remains limited [158] and is a priority before either contribution can be adopted with confidence in design.

The protection parameters also affect the various performance indicators unequally. For solidified-soil remediation, the repair depth primarily governs the ultimate capacity, while shallow remediation mainly improves the initial stiffness, with the two being dominated by the material strength and the elastic modulus, respectively [165]. Protection should therefore be designed for the target performance rather than by simply enlarging its extent.

A more contentious issue is the scale dependence of the stiffness enhancement. Some studies, drawing on one year of field monitoring across 72 turbines at an operational wind farm combined with finite-element modeling, have argued that the contribution of scour protection to monopile stiffness has long been underestimated, and that it can serve as an effective retrofit for existing foundations [166]. In contrast, studies on large-diameter monopiles have indicated that a noticeable stiffness gain appears only with heavy rock armor and relatively large protection heights, where the increase is on the order of 10%, and that the primary function of protection remains scour mitigation [163]. The effectiveness also depends on the timing of deployment. Preventive protection can effectively suppress scour development and preserve structural performance, whereas remedial measures generally offer only limited recovery for already damaged foundations. Remedial rock backfilling often restores only part of the lost stiffness, is highly dependent on the backfill thickness, and, in some cases, proves almost ineffective [162,167].

7.3. Dual Effects of Scour Protection on Structural Safety

Although scour protection generally enhances structural safety, its effects are not universally beneficial. Some measures alter the dynamic characteristics of the structure and thereby introduce new adverse responses or safety concerns. Scour usually lowers the natural frequency and shifts it toward 1P, whereas pre-installed rock armor continuously stiffens the soil under cyclic loading and can instead raise the frequency—by a few percent in one study—so that it may move toward 3P; solidified-soil grouting appears more favorable for maintaining frequency stability, although its cracking resistance still requires improvement [17]. Under seismic loading, one study found that rock armor may slightly increase the natural frequency while also raising the peak pile bending moment by roughly 10–20% [168]. Measures optimized for static performance therefore do not necessarily improve dynamic or seismic performance. This is consistent with the conclusions of Section 4 and Section 6 that neither a stable natural frequency nor a frequency change alone is sufficient to evaluate structural safety.

This also prompts a reconsideration of how the benefits of scour protection should be evaluated. Ma et al. [164] reported that scour protection produced only modest improvements in natural frequency, ultimate capacity, and fatigue life, yet reduced the serviceability limit state pile-head rotation by about 12.3%, as shown in Figure 14. This single result suggests that the primary value of protection may lie not in improving individual performance metrics but in mitigating the critical responses that govern structural safety. As shown in Section 5 and Section 6, scour typically erodes the serviceability margin before the ultimate margin; consequently, improving serviceability performance directly protects the most vulnerable part of the structural safety margin.

In summary, scour protection is an effective means of restoring foundation safety margins. It can compensate for scour-induced capacity loss and improve structural reliability, although its effectiveness depends on the protection mechanism, foundation type, and timing of implementation. More importantly, protection is not inherently beneficial, as measures optimized for static performance may compromise dynamic and seismic performance. Its primary value therefore lies not in maximizing individual performance metrics, but in preserving the serviceability margin that is typically lost first. Consequently, scour protection should be designed from a whole-life perspective, with the objective of maximizing the overall structural safety margin rather than simply minimizing scour depth.

Despite this progress, research on scour protection shares two main limitations. First, the protection layer is usually represented in a highly simplified way—as an equivalent surcharge, a solid body, or a Winkler idealization—rather than through the grain-scale contact and spatial coupling that govern its real behavior, which is also the root of the unresolved stress versus material mechanism debate discussed above. Second, protection performance is largely evaluated under monotonic or static loading, whereas the wind and wave actions on OWTFs are cyclic; the behavior of protected foundations under long-term cyclic and coupled vertical–lateral loading, including the possible degradation of the protection layer itself, remains poorly studied. Grain-scale and coupled modeling, together with evaluation under realistic cyclic loading, are therefore key directions for future work. This gap is mirrored in current guidance: DNV-RP-0618 [151] centers on external and internal stability and sinking, and does not address the protection layer’s feedback on dynamic or seismic response or its degradation under long-term cyclic loading. The dynamic and coupled effects reported above—such as pre-installed armor raising the frequency toward 3P or increasing peak seismic bending moments—thus lie beyond present design provisions, marking a direction in which research is ahead of the standards.

8. Conclusions and Future Perspectives

8.1. Conclusions

This review has examined, from the unifying perspective of structural safety, how scour erodes the safety margin of fixed-bottom OWTFs and how scour protection can restore it. Tracing this single thread through the scour mechanism, bearing capacity, natural frequency, long-term cyclic response, seismic performance, and protective measures yields several conclusions that are not apparent when these aspects are viewed in isolation.

• Scour represents not merely a loss of embedment but a fundamental change in the soil mechanical state. Its consequences differ in direction between sand and soft clay, so that simplified treatments risk being either overconservative or unconservative.
• Scour depth is the dominant geometric parameter in most loading regimes, whereas scour width and side-slope angle exert generally secondary effects that can become influential within certain frequency ranges. The distinction between local and global scour, however, remains consequential; treating three-dimensional scour holes as one-dimensional embedment loss leads to systematic errors in bearing-capacity, natural-frequency, and seismic assessments.
• Natural frequency is a necessary but insufficient indicator of structural safety. A small frequency shift can mask substantial degradation in stress, deflection and fatigue, and its seismic consequence is spectral rather than purely stiffness-related. Frequency-based monitoring should therefore be complemented by direct capacity- and deformation-based checks.
• Scour, long-term cyclic loading, and seismic action are coupled rather than additive, and their combined effect cannot be captured by superimposing the separately computed effects. For scour and cyclic loading, the two act on pile–soil stiffness in opposing ways, and the scour depth itself evolves under cyclic loading, so the coupled response can substantially exceed the single-factor estimate (Table 5). For scour and seismic action, the coupling is mechanistic: scour modifies the embedment, stiffness, and modal characteristics that govern the seismic response, so its effect cannot be obtained by superimposing a separate scour effect on a no-scour seismic response. Single-factor or sequential assessment therefore does not reliably predict the coupled response over the service life of scour-affected foundations.
• Across the studies reviewed here, scour tends to breach the serviceability margin before the ultimate one under both long-term cyclic and seismic loading. This recurring pattern suggests that serviceability criteria, rather than ultimate capacity, may often govern the safety assessment of scour-affected turbines, and that monitoring and intervention thresholds should generally be defined at serviceability level.
• Scour protection restores safety margins but is not unconditionally beneficial. Measures optimized for static performance may adversely shift dynamic and seismic responses. The primary value of protection therefore lies not in maximizing individual performance metrics but in preserving the serviceability margin that is the first to be lost, which calls for a whole-life design perspective rather than a scour-depth-minimizing one.

8.2. Implications for Design and Practice

The findings synthesized above carry several practical implications for the design, monitoring, inspection, and selection of scour-affected OWTFs.

• Design practice. Because scour tends to breach the serviceability limit before the ultimate limit under both cyclic and seismic loading, scour-affected design is better governed by serviceability than by ultimate capacity criteria, with monitoring and intervention thresholds defined at serviceability level. As scour, cyclic, and seismic effects are coupled rather than additive, these actions should be assessed through coupled analysis rather than by superimposing separately computed effects. Empirical scour-depth allowances (about 1.3–2.5 D) remain useful as conservative first estimates, but site-specific, time-dependent assessment is preferable for the very-large-diameter monopiles now entering service, and three-dimensional scour geometry should not be reduced to a one-dimensional loss of embedment.
• Monitoring. The natural frequency offers a global, non-intrusive screening indicator for scour, but the small scour-induced shift must be separated from larger operational and environmental variations; frequency tracking is therefore best used to flag candidate changes for targeted inspection rather than to measure scour depth directly. Because a stable frequency can mask substantial degradation in stress, deflection, and fatigue, frequency-based monitoring should be complemented by capacity- and deformation-based checks.
• Inspection. Inspection effort is best prioritized where the evidence base is weakest and the consequences most severe—the post-scour bearing-capacity, cyclic, and seismic behaviors, which lack full-scale validation—and at locations prone to protection-layer instability. Where frequency monitoring flags a change, targeted underwater or geotechnical inspection can confirm its cause, since scour must be distinguished from other sources of stiffness change, such as grout degradation.
• Foundation selection. Scour sensitivity depends strongly on foundation type: multi-leg jackets exhibit markedly smaller coupled amplification than monopiles under combined scour and cyclic loading (about 35–44% versus 58–104%), suggesting greater robustness to scour where site conditions permit, whereas the higher static stiffness of multi-leg systems can mask local stress and fatigue degradation and warrants direct checking. For existing foundations, scour protection can serve not only as a preventive measure but also as a retrofit that partially restores stiffness and capacity.

8.3. Future Perspectives

Building on these conclusions, several directions emerge as priorities for future research.

• Toward coupled, probabilistic, and standardized assessment. Scour, cyclic, and seismic effects should be evaluated through fully coupled, multiphysics analyses that propagate a time-varying scour state, rather than by superimposing separately computed effects. Because scour depth, soil parameters, and loading are inherently uncertain, such analyses should also move to a probabilistic basis—propagating these uncertainties through capacity, frequency, and fatigue analyses and expressing outcomes as failure probabilities or fragility surfaces, which would also help reconcile the conflicting effect sizes across studies. Ultimately, these advances need to be translated into standardized, verifiable procedures that reflect the coupled, three-dimensional, and time-dependent nature of scour. This gap is concrete: present standards represent scour as a fixed-multiple depth or a uniformly lowered seabed [34,35], address protection through stability criteria alone [151], and leave its coupling with cyclic and seismic action unquantified.
• Quantitative criteria for scour–liquefaction interaction. The coupling between scour and seismic liquefaction remains the least quantified of the interactions considered here, and the available evidence points to a complex, non-monotonic effect rather than a uniformly severe one: some studies report that liquefaction reduces the relative influence of scour [135], while others show scour altering the timing and extent of liquefaction in opposing ways [123,136]. Its direction appears to depend on initial soil density and its magnitude on foundation type. Systematic criteria spanning different soils and foundation configurations are still lacking and represent an important gap for the increasingly large turbines being deployed in seismically active and soft-soil regions.
• Size effects for next-generation large-diameter foundations. Several key mechanisms are influenced by foundation diameter, including the transition from inertial-dominated to kinematic-dominated seismic response, the diminishing relative effectiveness of scour protection, and the decay of natural frequency with increasing structural size. Much of the existing knowledge derives from small-diameter flexible piles, and its extrapolation to the very large monopiles and multi-leg systems now entering service is uncertain. Dedicated studies of scale effects are needed to avoid both unsafe and uneconomical designs at the upper end of the size range.
• Full-scale field validation remains one of the central challenges. Corroboration of the reviewed findings is limited and uneven. Continuous field and full-scale evidence is largely confined to the scour–natural-frequency relationship [16,90] and to recent field-monitoring data on long-term lateral behavior [166], supplemented by in situ scour measurements at a multi-leg (tripod) site [82]; the post-scour bearing-capacity, cyclic, and seismic findings rest almost entirely on numerical and reduced-scale evidence. This shortfall is most acute for multi-leg, suction-bucket, and composite foundations, whose evidence is sparse and typically limited to a single configuration; cross-type experimental and field programs are needed before the foundation-type contrasts drawn here can be treated as design-relevant rather than indicative. Addressing this gap through systematic field monitoring, an issue distinct from size effects as discussed above, is essential before the reported effect sizes can be adopted as design values.
• A closed-loop framework linking monitoring, assessment, and protection. The natural frequency emerges in this review with a dual character, where it is both a passive consequence of scour and an active, non-intrusive indicator for scour monitoring. Building on the accelerometer-based frequency-tracking approaches and their field limitations discussed in Section 4, and given that research on monitoring, performance assessment, and protective intervention remains largely compartmentalized, integrating real-time scour monitoring with the safety-assessment frameworks discussed above, including digital-twin approaches that continuously assimilate monitoring data and using the resulting information to guide protection decisions, would enable a closed-loop safety-management strategy. Such an approach is particularly valuable because preventive protection has been shown to be substantially more effective than remedial intervention.
• The interaction between scour and structural-side vibration control warrants systematic investigation. Because scour progressively alters the foundation’s dynamic characteristics, passive control devices tuned to the pristine structure gradually lose effectiveness as scour develops [118,148]; future devices should therefore be designed to remain robust across the full range of scour depths expected over the service life, for example through semi-active or adaptive damping strategies [148,149]. More broadly, geotechnical scour protection and structural-side control are still largely studied in isolation, yet they address the same safety problem from two complementary directions. Integrated designs that combine scour protection with vibration-mitigation devices, together with monitoring schemes that track scour-induced frequency changes to retune the control system in service, represent a promising but largely unexplored direction for safeguarding scour-affected OWTFs.