1 Introduction

Current turbines convert the kinetic energy of moving water, such as rivers or tidal currents, to mechanical power. These turbines have the potential to generate renewable electricity at scales ranging from distributed sensing to grid-scale electricity [1, 2]. While the working fluid varies between wind and water, the fundamental fluid dynamics behind the conversion processes are the same. For these reasons, current turbines have borrowed heavily from developments in the more mature wind energy sector, though some considerations, such as cavitation and biofouling, are unique to current turbines [3].

Two main types of current turbines are being developed for commercial applications: axial-flow turbines have an axis of rotation parallel to the direction of fluid flow, while cross-flow turbines have an axis perpendicular to the flow. While axial-flow turbines have achieved greater commercialization in the wind industry, recent efforts have demonstrated improvements to cross-flow turbine individual device efficiency [4] and array efficiency [5]. Further, cross-flow turbines have a number of desirable attributes such as vertical orientation, which allows generator placement at ground level and obviates the need for a yaw drive [6]. In water, their rectangular cross-section allows them to achieve higher blockage in confined flows, thus boosting overall efficiency [7]. For these reasons, the present work focuses primarily on cross-flow turbines.

Biofouling poses a performance risk for current turbine deployments, as for any long-term marine application where performance depends on boundary layer hydrodynamics (e.g., vessel drag). Any surface that is submersed in the ocean undergoes this colonization process by which marine organisms can accumulate substantially on the surface over time [8].

While significant biofouling has been observed on long-term marine turbine deployments (e.g., Fig. 1), there has been limited quantification of performance consequences, particularly for cross-flow turbines. It is well-known that antifouling coatings can be used to minimize biofouling, but trade-offs in cost, effectiveness, and toxicity mean that not all turbines or surfaces are likely to be treated, and presently available solutions may be sub-optimal for marine current turbine applications [9]. To quantify the cost–benefit trade-off of antifouling measures, further knowledge of turbine performance and load changes as a consequence of biofouling, and better quantification of fouling growth rates and species composition would be beneficial (this is an active area of research [10]). This is particularly relevant for long-term and autonomous deployments [1] where some degree of biofouling is probable between maintenance interventions.

Fig. 1
figure 1

Ducted axial-flow tidal current turbine showing substantial biofouling after 3 years of deployment in the Race Rocks Ecological Reserve, British Columbia Canada. (Note: Photo from: http://www.racerocks.ca/tidal-energy-turbine-removal/. Lester B. Pearson College. Garry Fletcher photographer, racerocks.ca)

Full size image

For the purposes of this study, we consider representative biofouling for a hypothetical current turbine deployed in Puget Sound, an estuary with multiple sites suitable for tidal energy development [11, 12]. Biofouling in Puget Sound is dominated by barnacles, algae, and mussels, with species composition and abundance varying across water depth, location, and season. Barnacles, such as Balanus crenatus, are found over a wide range of depths (e.g., 6–15 m [13, 14], and 55–75 m [15]). Evidence of barnacle fouling at depths suited to tidal current generation, combined with their relatively large mature size and growth rate [16], suggest that barnacles have the potential to significantly impact turbine operation. Barnacle settlement is most probable when water temperatures exceed 10 °C [13] and the primary barnacle settlement period in the Pacific Northwest is from April to November [14] with peak growth occurring in summer and early fall [16].

Relatively few studies have quantified current turbine power degradation due to fouling and most of those have focused on axial-flow current turbines. Orme et al. [17] experimentally investigated the effect of fouling on the lift and drag of an airfoil at constant angles of attack using extruded plastic cones to resemble barnacles. Significant declines in the peak lift to drag ratio were observed for both an increase in fouling height and density. Using blade element momentum theory, these changes in airfoil performance were estimated to reduce the maximum power coefficient (Eq. 2) between 20 and 70%. Another experimental study by Walker et al. [18] investigated the effect of increased blade surface roughness on axial-flow turbine performance by applying contact cement to the blades as a proxy for calcareous fouling. This reduced turbine power and thrust by 20%. Concurrent numerical investigations also revealed that roughness along the leading 10% of the blade chord had a similar effect on lift and drag as roughness across the entire chord.

The most similar prior work to our study is a numerical investigation of cross-flow turbine fouling performed by Rivier et al. [19]. Two-dimensional simulations showed that barnacle colonization of the blades significantly altered the vorticity field, with attendant implications for performance. As for Walker et al., the proximity of fouling to the leading edge of the foil was shown to be significant. Both a realistic and a triangular barnacle profile were considered, with each profile producing different velocity fields around the blades. This suggests that the roughness profile is important, and a realistic fouling shape should be used for accurate experimental testing. Rivier et al. also noted the need for higher fidelity modeling at a lower computational cost, as results were sensitive to the choice of turbulence model and three-dimensional simulations were unsatisfactory due to the coarse mesh required for tractable computation. Little could be concluded regarding sensitivity to three-dimensional effects and no estimations of turbine power or thrust coefficients were made. Rivier et al. highlights the difficulty of accurately modeling cross-flow turbine response to biofouling and motivates our experimental study.

The only experimental investigation of cross-flow turbine surface roughness was by Howell et al. [20] who evaluated cross-flow turbine performance at multiple Reynolds numbers for smooth and rough blade surfaces. That study revealed that the maximum torque for the rough blades was achieved at lower tip-speed ratios than for the smooth blades. It was also found that turbine power output was enhanced by the roughened blade surface below a critical Reynolds number, but degraded above that value. This is indicative of the general sensitivity of cross-flow turbine performance to Reynolds number [21, 22].

Here, we evaluate the changes in turbine power output and structural loads as barnacle fouling progresses from initial colonization to maturity. Further, we seek to understand when fouling would likely become observable in performance data (i.e., for condition health monitoring of a turbine) and how these performance changes are related to the size and concentration of fouling on blade surfaces. Our efforts build on prior computational work by Rivier et al. for barnacle fouling and experimental work by Howell et al. on uniform surface roughness.

2 Background

2.1 Cross-flow turbine hydrodynamics

The relative velocity and angle of attack encountered by the blades on a cross-flow turbine vary substantially with the blade’s azimuthal position. This is in contrast with axial-flow turbines where these quantities are invariant with azimuthal position. Consequently, for a cross-flow turbine, instantaneous torque varies periodically, and blades can experience deep dynamic stall and interact with their own wake [23,24,25]. As observed by Rivier et al. [19], resolving these unsteady boundary layer interactions in simulation often comes at high computational cost.

Cross-flow turbine performance is characterized by the relationship between rotational speed and non-dimensional power. The tip-speed ratio (λ), represents how fast the turbine rotor moves compared to the free stream velocity as

$$\lambda = \frac{R\omega }{{U_{0} }}$$
(1)

where R is the turbine radius, ω is the rotation rate of the turbine, and \(U_{0}\) is the free stream velocity. The power coefficient (CP), represents the efficiency with which the turbine converts the kinetic power incident over its projected frontal area to mechanical power, as

$$C_{P} = \frac{{Q_{h} \omega }}{{\frac{1}{2}\rho AU_{0}^{3} }}$$
(2)

where Qh is the hydrodynamic torque produced by the turbine, ρ is the density of the fluid, and A is the projected area of the turbine rotor. Additionally, the non-dimensional thrust coefficient (CT) is given by

$$C_{T} = \frac{T}{{\frac{1}{2}\rho AU_{0}^{2} }}$$
(3)

where T is the thrust force acting on the turbine.

2.2 Effects of fouling on blade hydrodynamics

Biofouling of turbine rotor blades increases the surface roughness, altering the flow field near the blade surface (i.e., the boundary layer). At moderate to high Reynolds numbers, elevated surface roughness decreases lift and increases drag [26], which decreases time-average hydrodynamic torque, Qh. The magnitude of this decrease depends on the roughness height, chord-wise location, fouling density, and foil shape. Surface roughness increases boundary layer mixing and can lead to earlier and longer turbulent transitions along an airfoil [27]. Further, as roughness height increases, so does the viscous drag due to a higher local velocity shear rate at the blade surface [26].

The effect of roughness on airfoil performance depends on the roughness height, k, relative to the undisturbed boundary layer thickness, δ. Roughness elements that extend beyond the boundary layer thickness (i.e., k/δ > 1), such as a mature barnacle, produce separated flow in their wake, which alters the pressure distribution that develops along the blade surface. This local separation increases the pressure drag and decreases the lift developed by the foil, as well as causing static stall at lower angles of attack [27, 28]. Surface roughness has also been shown to affect dynamic stall, as demonstrated by Huebsch and Rothmayer [29] for icing on an aircraft wing.

The chord-wise location of roughness elements also has a significant effect on airfoil performance, with roughness near the leading edge having a larger effect than roughness near the trailing edge, though the location of maximum performance loss varies with the roughness height [28]. This effect is due to reductions in peak suction pressure at the leading edge and the higher local velocity at this location which elevates drag from skin friction. Foil profiles and angles of attack that yield relatively large leading edge suction peaks are, therefore, more affected by surface roughness. Consequently, wind turbine design emphasizes foil profiles less sensitive to leading-edge surface roughness [30]. Finally, as the density of roughness elements on an airfoil increases, performance decreases [17]. This effect is coupled to the chordwise location, as the combination of these parameters determines the spatial extent of disruption to the boundary layer.

The effect of biofouling on turbine hydrodynamics depends on species composition and maturity. While biofilms as thin as 0.1 mm have been shown to increase skin friction on a flat plate [31], barnacle heights can exceed 10 mm [32]. At this size, barnacles would cause local flow separation on a turbine blade and thus can be expected to affect viscous drag, pressure drag, lift, and dynamic stall.

3 Methods

The effects of barnacle growth on a cross-flow turbine were investigated at laboratory scale. An interpretation of results in the context of a larger turbine is presented in Sect. 5.2. “Clean” and “fouled” turbine blades and blade support attachments were fabricated using additive manufacturing. Turbine performance was assessed over a range of tip-speed ratios at a single Reynolds number. Experimental data is available at http://hdl.handle.net/1773/43775.

3.1 Turbine and test facility

The experimental turbine measured 17.2 cm in diameter and 25.3 cm in height. The turbine rotor consisted of two straight blades with a NACA 0018 profile, a chord length of 2 cm, and a preset pitch angle of 6° relative to a quarter-chord mount point. Blades were connected to the center shaft at each end by two foil-shaped struts (NACA 0030 profile). Definitions of turbine azimuthal position, forces, and torque are shown in Fig. 2. Turbine geometry represented a 1:5 scale model of a turbine capable of producing power outputs in the kilowatt range, which would be relevant for powering persistent sensing or recharging underwater vehicles [1, 2].

Fig. 2
figure 2

Definition of turbine azimuthal position, forces, and torque

Full size image

Experiments were conducted in the Alice C. Tyler recirculating water flume at the University of Washington. The test section of this flume measured 75 cm in the cross-stream direction and tests were run at a dynamic depth of 56 cm. During experiments, a constant inflow velocity of 1 m/s and temperature of 34 °C were maintained. Turbulence intensity was approximately 4% and the turbine blockage ratio was 9.8%. The experimental set up is shown in Fig. 3. For the upper load cell, a 6-axis Mini45 (ATI Industrial Automation) was used, and for the lower load cell a 6-axis Nano25 (ATI Industrial automation) was used. A Yaskawa servomotor (SGMCS-05B3C41) equipped with a 218 count encoder was used to control the turbine’s rotation at a constant angular velocity [33]. Turbine performance data was acquired using a National Instruments 6351 DAQ at a sampling rate of 1000 Hz with each tip-speed ratio maintained for 60 s. Inflow velocity was measured five turbine diameters upstream of the turbine at the vertical and cross-stream midpoint of the flow using an Acoustic Doppler Velocimeter (Nortek Vector) sampling at 64 Hz.

Fig. 3
figure 3

Schematic of cross-flow turbine experimental testing apparatus with measured and applied torques indicated [33]

Full size image

Turbine performance metrics (Eqs. 2–3) were calculated as instantaneous quantities. From the instantaneous data, cycle averages were calculated for each rotation, time-averages were calculated over all rotations, and phase-averages were calculated as a function of azimuthal blade position in 6° azimuthal bins. Because it is difficult to temporally correlate free stream turbulence and its interaction with the turbine rotor, the denominator in all the performance metrics is a time average of velocity over an integer number of turbine rotations. The variability in time-average and phase-average performance coefficients was quantified by the inter-quartile range (IQR) (i.e., interquartile range of cycle averages and values underpinning the phase average in each azimuthal bin, respectively).

The Reynolds number for these experiments is defined in terms of the blade chord length, c, as

$${\text{Re}}_{c} = \frac{{U_{0} c}}{\nu }$$
(4)

where ν is the kinematic viscosity of water. For all experiments, Rec was 2.7 × 104, which is likely in a transitional regime [21, 22].

3.2 Fabrication of fouled turbines

Turbine blades were “fouled” using a barnacle model derived from a three-dimensional scan of a real barnacle taken from the shores of Puget Sound.Footnote 1 This barnacle model was geometrically scaled to three different heights and randomly patterned across the surface of the blades at three different densities. In choosing a random spatial distribution, we assumed that there would be no preference for barnacle colonization with respect to chord or span-wise position and ignore the natural clustering tendency of barnacles. As a consequence of this random distribution, the average distance of fouling from the leading edge was equal to half the chord length. Illustrations of the barnacle model, fouled blade, and fully assembled turbine are shown in Fig. 4. All experiments used clean struts to focus on blade hydrodynamics.

Fig. 4
figure 4

a Three-dimensional barnacle scan used to generate fouling model b CAD model of “fouled” turbine blade covered in barnacles c fully assembled turbine showing blades and struts fabricated using additive manufacturing

Full size image

Barnacle size was quantified by the non-dimensional roughness height k/c, where k was the measured barnacle height. Barnacle density was quantified by the number of barnacles per blade, though researchers have employed various density metrics to quantify the extent of barnacle fouling. For example, Orme et al. [17] represented the density as the number of barnacles per chord length-squared area, while Demirel et al. [34] used the percent of blade surface area covered by barnacles. Some biofouling surveys [14] have employed the dimensional barnacles per square centimeter. Because our results are only quantitatively suited for comparison with a geometrically-scaled turbine, the total number of barnacles per blade is the most relevant density metric.

Clean and fouled turbine blades were fabricated from Somos WaterShed XC 11122 using stereolithography with a resolution of 0.05 mm. Blades were structurally reinforced by key stock (3/32″ x 3/32″ x 12″ Grade 18-8 Stainless Steel) and bonded to the end plates using epoxy. Even with reinforcement, blades deflected substantially in the radial direction during testing, which may have altered performance. However, this deflection was consistent across all turbines tested and should not affect relative comparisons. Additionally, the extent of testing was limited by structural failures in the 3D printed plastic at the joint connecting blade to strut (e.g., tests at multiple Reynolds numbers were originally planned).

As-manufactured barnacle heights and densities are shown in Table 1. Values were selected based on biofouling composition surveys conducted in Puget Sound [13, 14] and prior investigation of barnacle fouling of vessel hulls [34,35,36]. This range in fouling severity approximated turbine performance degradation at different stages of barnacle maturity for a given initial colonization pattern. Barnacle dimensions were geometrically scaled 1:5 to be consistent with the turbine model. Roughness scaling is discussed further in Sect. 5.2.

Table 1 Experimental test matrix of barnacle heights and densities, showing the percent of the blade surface area covered for each combination. Note: density illustrations are for large barnacles, and height illustrations do not represent the experimental densities
Full size table

As-printed barnacle heights were measured using an electronic height gage “NAME HERE” with an accuracy of 0.000 mm. The physical heights of the printed barnacles were markedly smaller than the CAD models used for production, and the aspect ratio of the barnacles was, therefore, greater than intended. This is to say that additive manufacturing is a powerful technique for exploring new topics, but is not without its limitations.

4 Results

Time-average turbine performance is summarized in Fig. 5. Without exception, an increase in either barnacle height or density decreases the maximum power coefficient (Fig. 5a, b). The magnitude of this performance loss becomes substantial as barnacle size increases, and with large barnacles at medium and high densities, the turbines do not produce net power at any tip-speed ratio. Additionally, barnacle size alters the optimal tip-speed ratio (tip-speed that maximizes the power coefficient). As an exception to the overall trend, we note that the power coefficient for the small barnacles increases compared to the clean blade at lower tip-speed ratios.

Fig. 5
figure 5

Time-average power coefficient and thrust coefficient as a function of tip-speed ratio for different barnacle heights and densities. Insets show the region of interest for power generation in greater detail and error bars represent the inter-quartile range (IQR) of cycle-to-cycle variations in performance coefficients

Full size image

Barnacle fouling also slightly decreases the thrust coefficient (Fig. 5c, d). Trends are less consistent than for the power coefficient, as an increase in barnacle density yields a general, but modest decrease in turbine thrust across tip-speed ratios. However, barnacle height does not have a consistent effect on the thrust coefficient, particularly at tip-speed ratios corresponding to peak power coefficient and many of the median trends fall within the inter-quartile range of cycle-to-cycle variation (Fig. 5d).

The effects of barnacle height on phase-average turbine performance is shown in Fig. 6. When barnacle height increases, the effect is nearly an offset in power coefficient, with the turbine producing less power at all azimuthal positions. The magnitude of this offset varies with barnacle size (Fig. 6a). In contrast, barnacle fouling decreases the amplitude of the oscillation in phase-average thrust coefficient (Fig. 6b). Barnacle fouling does not appear to cause significant phase shifts in power or thrust. This may be a consequence of speed-regulated control, which maintains blade kinematics across the cases [33]. Phase shifts might occur under other control strategies (e.g., torque-regulated control).

Fig. 6
figure 6

Effect of increasing barnacle height on phase-average turbine performance coefficients. Results are shown for medium density at λ = 2.88, near the peak of the power performance curve

Full size image

5 Discussion

5.1 Performance

The presence of barnacles on the blades and mounting structure of a cross-flow turbine can dramatically reduce power production. Although this result is intuitive, the present work suggests that beyond a relatively limited threshold of barnacle height and density, a cross-flow turbine can be rendered incapable of producing power. Fortunately, barnacle fouling on blades does not increase structural loads, such that blade fouling is primarily a performance, not a survival, concern. This is consistent with prior experimental results for axial-flow turbines [18].

As illustrated in Fig. 7, reductions in maximum CP are a function of both barnacle height and density. As barnacle density increases, the marginal change in CP with density varies with barnacle size (Fig. 7a). For small barnacles, the performance degradation is nearly invariant with density (i.e. a few barnacles are almost as detrimental as many). This is likely because once barnacles are distributed spanwise along the blade, the aggregate disruption to the boundary layer is substantial. For medium and large barnacles (i.e., barnacles closer to mature size), the marginal change in CP with density also decreases, but is still substantial in absolute terms. Conversely, as barnacle height increases (Fig. 7b), the marginal reduction in CP increases with height. In other words, once barnacles colonize the surface of a cross-flow turbine blade in any substantive density, the performance degradation will accelerate as they mature.

Fig. 7
figure 7

Trends in maximum power coefficient with increasing barnacle height and density. For the cases of large barnacles at medium and high density, the power coefficient is negative over all tip-speed ratios and “maximum” CP is taken at the same tip-speed ratio as maximum CP for the low density case

Full size image

5.2 Scaling of results

While these results suggest that biofouling poses a significant performance risk to marine current turbines, they correspond to a single turbine geometry and Reynolds number. Here we discuss the relevance and limitations of quantitatively extending these experimental results to turbines with a similar geometry operating at different Reynolds numbers.

In general, cross-flow turbine performance has been shown to increase with blade-chord Reynolds number up to a critical value, beyond which performance is independent of Reynolds number. Bachant and Wosnik [21] found that for their three-bladed turbine, this critical value was around Rec = 2 × 105. While turbine geometry varies between our study and Bachant and Wosnik, this strongly suggests that our experiments (Rec = 2.7 × 104) take place at a transitional Reynolds number, and clean turbine performance would likely increase with Reynolds number. Miller et al. [22] suggests that Reynolds independence is not achieved for some cross-turbines until Rec = 1.5 × 106, such that even a kW-scale turbine could still be operating in a transitional regime. Regardless, as the Reynolds number increases, foil stall occurs at higher angles of attack due more efficient momentum transfer within the boundary layer, and the maximum lift to drag ratio is thus increased. Due to the angle of attack variation experienced by a cross-flow turbine blade, this increases lift at some azimuthal positions and increases cycle-average power output.

The effect of blade roughness on cross-flow turbine performance also depends on Reynolds number. For example, Howell et al. [20] reported that, at lower Reynolds numbers, rough blades (k/c = 0.005) outperformed smooth blades, but, at higher Reynolds numbers, smooth blades had superior performance. A similar effect can be seen in Fig. 5 when comparing the performance of clean blades to blades with small barnacles (k/c = 0.006). At lower tip-speeds, turbine performance increases with fouling. Because the roughness of the fouled blade trips the boundary layer into turbulence at a lower Reynolds number, this likely causes the flow to stay attached longer and delays the onset of stall to a higher angle of attack, thus increasing the performance for some rotation rates. Although the viscous drag is increased by the small barnacles, the pressure drag and thus overall drag is reduced. At higher tip-speed ratios, the azimuthal range in angle of attack is reduced and the increase in viscous drag likely dominates. For the larger barnacle sizes, the increase in pressure drag caused by the expansion of the boundary layer is sufficient to outweigh the benefit of delayed stall. In summary, beyond Reynolds independence, we would expect performance at all tip-speed ratios to decrease due to fouling, but below Reynolds independence, the onset of fouling may increase performance at some tip-speeds.

The effect of surface roughness on airfoil performance has been investigated in the airfoil icing literature and this analogue provides insight into experimental scaling of surface roughness. Airfoil performance is sensitive to Reynolds number, k/δ (roughness height normalized by boundary layer thickness), and the location of roughness with respect to the leading edge. As discussed in Sect. 2.2, different roughness heights affect airfoil performance in different ways. Small roughness (k/δ < 1), primarily increases the skin friction along the blade, and performance changes are relatively sensitive to Reynolds number [37]. Conversely, large roughness (k/δ > 1), causes local flow separation, but the performance changes are relatively insensitive to Reynolds number [28].

As a consequence, numerous studies have found that geometric scaling of the roughness height (i.e., maintaining k/c across geometric scale) provides accurate airfoil performance during subscale testing [28, 37, 38]. However, for small ice profiles with k/δ ≤ 1, geometric scaling has been found to be generally less accurate at representing full scale performance effects [38]. Broeren et al. [39] found that geometric scaling for simulated runback-ridge ice (for which k/δ ~ 1 [39]) amplified the performance losses measured at full-scale. This suggests that, for this roughness type, the appropriate scaling factor is greater than geometric scaling (i.e., performance-equivalent sub-scale roughness is shorter than a geometric scaling of full-scale roughness).

To evaluate the appropriateness of geometric scaling for our experiments, k/δ, was estimated for the experimental turbine (details provided in “Appendix”). k/δ was found to range from near zero (Reynolds-dependent regime) in regions close to the trailing edge of the blade, to greater than 10 (Reynolds-independent regime) for the large barnacles along the leading edge. For the small barnacles, k/δ < 1 across most of the blade. This suggests geometric scaling of this roughness would likely lead to an over-prediction of effects at larger scale. Conversely, because the large barnacles exceed the estimated boundary layer thickness for most azimuthal positions, geometric scaling is likely to yield accurate results. The medium barnacles are likely in a transitional k/δ regime.

Our performance results are only quantitatively valid for turbines that are geometrically similar to the experimental turbine due to the strong influence of turbine geometry on power output [40]. As the number of blades, solidity, foil profile, aspect ratio etc. are changed, the absolute effect of fouling will likely also change. These results are, however, likely qualitatively representative, with barnacle fouling likely to reduce maximum CP and associated CT.

5.3 Implications for field-scale operation

Experimental turbine geometry and barnacle size represent a 1:5 scale model of a kW-scale turbine. If geometric scaling is accepted as valid, this suggests that maximum CP declines by 15% when low-density barnacles are only 0.5 mm tall. This would be barely perceptible by touch and marginally visible by eye but would be detectable in turbine power output and would likely be detected by a condition monitoring system. Once barnacle height reaches 1.5 mm, CP declines by 30–50%, and by a height of 4 mm, the turbine would cease to produce power. The pace of these performance declines would also depend on the colonization density. Barnacle growth rates can vary between 0.02 and 0.10 mm per day depending on myriad factors including species, location, and season [16, 41]. At these growth rates, barnacle height would increase from 0.5 to 1.5 mm in 10–50 days and to 4 mm in 25–125 days. This is a relatively limited window for reactive maintenance and emphasizes the benefits of durable antifouling coatings.

The shift in the shape of the power performance curve seen in the data suggest that biofouling could also have implications for the control of cross-flow turbines. For example, λoptimal shifts by 5% between clean blades and blades with small barnacles. A change of this magnitude could lead to additional power losses of 1–3% [42] if a non-linear controller (e.g., kω2) is used to maximize power [43]. This suggests that there may be marginal benefits to implementing adaptive control schemes.

5.4 Effect of fouling on support struts

While not the focus of this study, it is informative to consider how fouling of other components, such as the blade support struts, would affect turbine power performance. To this end, an experiment was conducted with fouled blades and struts. Fouled struts were fabricated with large barnacles at a medium density. Specifically, the density was 216 barnacles per strut which corresponded to the 10% surface coverage of the large barnacle, medium density fouled blades. We note that this fouling case is one where the turbine does not produce power at any tip-speed ratio, even with clean struts (Fig. 5a), but is still informative.

To estimate the relative contribution to performance loss caused by fouling of the blades versus fouling of the mounting structure, measured turbine performance was decomposed as for Strom et al. [44] as

$$C_{P } = C_{P,blades} + C_{P,mount} + C_{P,secondary}$$
(5)

where CP,blades represents the net non-dimensional power produced by the blades, CP,mount represents the non-dimensional performance loss from the struts structure, and CP,secondary encompasses the effect of the mounting structure on the blades (e.g., reduction in tip loss, span-wise flow), as well as the effect of the blades on the mounting structure (e.g., induction). Based on Strom et al. [44], the secondary effects are likely negligible, such that the blade performance can be estimated from measurements of CP and measurements of CP,mount (taken by rotating the support structure without blades).

As illustrated in Fig. 8, turbine power loss is substantially greater when fouling occurs on the blades compared to an equivalent fouling size and density on the support structure. This difference may be less significant for a turbine with higher support structure losses (e.g., disc end plates).

Fig. 8
figure 8

Comparison of degradation in CP for blade fouling versus fouling of the struts shown for large barnacles at medium density (Eq. 5)

Full size image

6 Conclusions

The performance effects of biofouling on a cross-flow turbine were measured for three different sizes and densities of barnacle fouling. The presence of barnacles on a turbine rotor affects both power and thrust. Experiments demonstrated a significant decrease in maximum CP, with declines ranging from 15% for the small barnacles (k/c = 0.006) to greater than 100% for the large barnacles (k/c = 0.035). This indicates that barnacle fouling can render a cross-flow turbine incapable of producing power. Measurements of power loss for fouled struts demonstrate that performance reductions are likely to be dominated by blade fouling, but that strut fouling is significant. Due to the likelihood of substantial loss in power production, antifouling coatings should be applied to all cross-flow turbine components. In contrast, relatively small declines in CT of 2–5% were measured, demonstrating that while biofouling severely degrades turbine power output, it does not greatly affect structural loading. Quantitative extrapolation of these results to larger-scale turbines is challenging, as the boundary layer changes associated with fouling may scale differently with Reynolds number than the blade kinematics that give rise to turbine power and thrust. This could be an interesting line of future research, as could the effect of barnacle location and clustering relative to the random colonization pattern used in this study.