Last updated: 8 April 2021
Bio-inspired surface modification has shown potential as an antifouling strategy. The surface textures can act on the ability of fouling micro-organisms to find secure attachment to the substrate but also alters fluid induced stresses over the organisms. The two effects combine to reduce initial settlement and improve the self-cleaning properties of the surface by altering cell adhesion in turbulent flows.
This offers the prospect of improved antifouling properties without the usual environmental risk associated to the toxicity of products often used to combat fouling in marine applications. Part of the NEMMO project has been dedicated to designing an effective texture tailored to the NEMMO turbine. Geometries inspired by the growth rings of the brill fish Scophthalmus rhombus found in marine water of Mediterranean region, Norway and Iceland (See Figure 3) have been studied.
The designs considered mimic some of the features of the micro-ridges with inter-ridge gaps selected to block out the smallest micro-organisms. Typographies with certain characteristics similar to those studied here are known to disrupt the wall eddies reducing their scale in the streamwise directions. In the transitionally rough regimes, this effect is confined to the inner part of the boundary layer and the impact on the outer layer is limited to setting the length and velocity scales of turbulence due to an increase in wall shear stress.
This in turn can affect boundary layer transition and separation in adverse pressure gradients so that texturing of the NEMMO turbine blade could alter its drag and lift characteristics. Part of the research on micro-texturing for anti-fouling has been dedicated to improving our understating of these effects. The aim was to assess potential detrimental effects on power conversion for the hydro-turbine. A two stage computational fluid dynamics study has been performed to achieve this.
Scale resolving turbulent flow simulations were carried out with an LES simulation methodology. The objective was to estimate the equivalent sand roughness of the candidate surface textures. Channel flow simulation were performed with periodic streamwise and spanwise boundaries initially at a friction Reynolds number Re_τ=1,890. The dynamic k-Equation LES solver was shown to reproduce self-sustaining turbulence with channel dimensions in wall units of 1,500 and 3,000 in the streamwise and spanwise directions respectively.
A mesh convergence analysis confirmed that the simulation parameters provided consistent trends by comparison with DNS predictions from A. Lozano-Durán, 2014 (see meshes M1 and M2 in Figure 4 and Figure 5). The two types of textures considered were applied to the full lower surface of the channel with a lower Re_τ=395 . The mesh needed to capture the textures was obtained by refining the smooth channel wall mesh M1.
A manageable computational load was achieved resolving each cavity between textures with 8, 4 and 5 cells in the stream-wise, span-wise and wall normal directions respectively and increasing the gap between textures to Δz^+≅17.6. The rest of the domain was meshed so that y^+ at wall boundaries was kept lower than 2. The total mesh count was approximately 4.1 million cells with a time step defined to maintain a Courant number lower than 1. The result provided estimates of the shift in the log law profile and in turn estimates of the equivalent sand roughness.
Although results suggest that the micro-textures can be expected to have a benign increase in wall stress at worst and an actual drag reduction in some cases (see Figure 6), these were obtained with under resolved gaps between textures due to computational resource limitations. An additional sensitivity analysis was carried out to complement these findings by testing the effect of higher surface roughness operating in the transitionally rough regime and applied of a turbine section.
The two types of surface typographies studied for their anti-fouling properties were considered when estimating the likely range of equivalent sand roughness from existing correlations. The research in this case focussed on a dimensional analysis of the NEMMO turbine blade. A single 2D blade section at radial to turbine radius ratio r/R=0.84 was simulated under steady external flow using the transitional four equation RANS k-ω SST turbulence model of ANSYS Fluent version 2019 . Inlet conditions were based on a turbine operating at an angle of attack of 5.46° in a 2.5m/s tidal stream at a tip speed ratio of 6.7. The wall unit was found to be within the range of 1 to 4.8 μm with a friction Reynolds number Re_τ between 5×10^4 and 3.5×10^5.
Roughness was accounted for in terms of its effect on transition and through boundary conditions for the turbulent kinetic energy k, the specific dissipation rate ω and the wall shear stress. The transition model relies on a correlation for a modified rough transition momentum thickness Reynolds number Re ̃_(θ_t ) defined in terms of the geometric roughness height [2] while wall boundary conditions are determined from the equivalent sand roughness height. With these rough wall modifications to the boundary conditions for the turbulent quantities k and ω, hydraulically smooth regime is recovered below k_s^+=2.25.
Results reported here considered two physical roughness height k={10,28} μm and six equivalent sand roughness k_s={0,3,5,10,20,28} μm. LES estimates suggest that k_s is likely to remain within or near the hydraulically smoot limit of 2.25 but the analysis was extended to include the physical maximum height of the textured, i.e. 28 μm. In addition, four inlet turbulence were tested to cover artificially low inlet turbulence to more realistic conditions defined to match measurements taken at a Scottish site in 55m water depth .
The sensitivity analysis confirms that the roughness settings can have a significant impact. From Figure 7, it is evident that the free stream turbulence has a major bearing results. At artificially low turbulence, the effect is exacerbated by triggering early boundary layer transition to turbulence. At the higher inlet turbulence, it remains moderate with single digit percentage increase in drag with a maximum of 8% at the larger (k,k_s )=28 μm compared to the smooth wall reference value. The corresponding increase in lift compared to the smooth results is smaller at approximately 3.4%. These results suggest that the 10 μm deep textures would not be expected to increase drag by more than 1% and decrease lift by more than 2%.