Design and testing of a full-scale 2 MW tidal turbine blade

— The Large Structures Research Group of MaREI, Orbital Marine Power Ltd and ÉireComposites Teo. have designed a full-scale blade for a next-generation 2 MW tidal turbine as part of the H2020 FloTEC project. The 8.5 m long blade will be tested under static load conditions through the H2020 MaRINET2 transnational access programme and fatigue conditions through the OCEAN ERA-NET SEABLADE project. This paper provides an overview of the initial design study which analysed the impact of using a single shear web or two shear webs in the design. The result of this design study led to optimisation of the laminates throughout the blade to reduce the cost of manufacture and, hence, the levelized cost of energy of the device. The finite element analyses were performed using the MaREI@NUI Galway composite blade design software BladeComp. From the results of the analyses a single web design was chosen for the blade. The present work also describes the set-up for the structural tests and an overview of the data acquisition and instrumentation requirements for full-scale static and fatigue blade testing.

Companies will also need to reduce the levelized cost of energy (LCOE) for their production turbines to make tidal energy commercially competitive with onshore wind, offshore wind and solar PV.Structural testing can provide valuable insights into the performance of the blade and blade components, to feed into the iterative design process.
The H2020 FloTEC project is focused on demonstrating an advanced full-scale floating tidal energy system for a low cost of energy [4].As part of this project, the blades of the new Orbital O2 2 MW rotor will be enlarged, and the device operated at a lower rated speed, to capture 50% more energy than the previous design.The project will also focus on reducing the capital and operational costs of the turbine at the design stage.
As a member of the MaREI Centre, the Large Structures Research Group (LSRG) has developed advanced computational design methodologies for tidal current turbine blades [5], [6], performed design and optimisation studies on wind turbine blade structures of several scales [7]- [9], and conducted structural testing of components for a 3/8th scale blade and rotor subsection for a prototype OpenHydro tidal turbine [10].The methodologies developed and the research findings on blade structural design and optimisation have been incorporated into an inhouse blade design software package called BladeComp.BladeComp was one of the primary tools used in the structural design studies for the new blade for the FloTEC project.Additionally, structural testing for static loading (through the MaRINET2 transnational access programme) and for fatigue loading (through the Ocean Energy ERA-NET Cofund) will be conducted at the Large Structures Testing Laboratory in the Alice Perry Engineering Building at NUI Galway.This paper presents the results of an initial design study, analysing the optimum structural layout for blade shear webs.The results of this study are discussed and the details of the upcoming testing programme for the full-scale blade is described.The FE analyses and the detailed test setup description presented in this paper are revised and edited from two reports created by NUI Galway for the FloTEC and SEABLADE projects [11], [12].

II. BLADE AND FE MODEL DEFINITION
The blade geometry is defined by the chord, twist and airfoil distributions along its length.Fig. 1 displays the chord length and hydrodynamic twist angle for the tidal turbine blade assessed in this study.The material used to construct the blade is a glass-fibre epoxy combination known as "powder epoxy" designed by the manufacturing partners, ÉireComposites Teo.The powder epoxy technology has been experimentally investigated for tidal turbine applications [13] and used to manufacture wind turbine blades [14].One of the primary advantages of powder epoxy is its manufacturability.Each section of the blade structure (the upper and lower halves and the shear web) can be partially cured, assembled and then fully cured as a single piece without the need for bond lines, typical of other blade manufacturing procedures.Further details of the manufacturing processes for powder epoxy can be found in Flanagan et al [14].The material properties of unidirectional and biaxial laminates are provided in Table 1.These properties, along with the geometric properties, are used as inputs to BladeComp for generating the blade finite element (FE) models.For the initial analysis of the effect of alternative shear web combinations on the design, a simplified layup for the blade was assumed.The spar and root regions of the blade consisted of layers of biaxial material on the inner and outer surfaces and a solid section of unidirectional laminate in between.The hydrodynamic shells consisted of just biaxial laminates oriented at ±45° to the pitch axis of the blade.Similarly, the webs consisted of just biaxial laminates.Table 2 outlines the distribution of laminate thicknesses along the length of the blade for these three key regions.
Five load cases were considered for the blade, relating to: the maximum flatwise bending moment ( , ), the maximum flatwise shear loading ( ,, ), the maximum pitching moment ( , ), the minimum pitching moment ( , ) and a damage equivalent load case ().The load distribution for each case consists of an edgewise and flatwise distribution oriented parallel and perpendicular to the rotor plane, respectively.Fig. 2 shows the distributions of the two components along the blade for the five load cases.The loads are applied in the FE models as multi-point constraints.The nodes in each blade section are coupled to a reference node for that section.The loads from Fig. 2 are applied as point loads to these reference nodes.Static FE analyses were conducted with a total mesh size of approximately 17,000 nodes.This resulted in a typical element length of approximately 50 mm.Reduced integration S4R elements were applied to the model, after a comparison of preliminary results with higher order S8 elements indicated sufficient mesh density for the analyses.

A. Study overview
The primary variable under consideration in the present study was the number of shear webs used in the blade.
With a single shear web the main structural member of the blade is formed by an I-beam, while with the double web configuration the spar is formed from a box-beam.
The two structures were modelled and analysed for their deflection, rotation and strains throughout the blade.The strains in biaxial and unidirectional materials were inspected for regions including: the spar cap on the lower (pressure) side of the blade, the spar cap on the upper (suction) side of the blade, the leading edge on the upper and lower sides of the blade, the trailing edge on the upper and lower sides of the blade and the shear web(s) (see Fig. 3).BladeComp allows outputs of strain/stress for each of these locations along the blade, processing all the elements in the region to return the maximum/minimum stresses in each.The results of the strain for the three in-plane components were analysed throughout the blade.Additionally, the results of the five load cases were combined to form maximum and minimum strain envelopes for each location in the blade.

B. Study results
The results of the longitudinal strain in several regions of the lower spar cap is shown in Fig. 4. Fig. 5 compares the maximum strain distributions in the shear web(s) for both blade designs.The results of all 5 load cases were combined to find the maximum strain envelopes for the blade for both Fig. 4 and Fig. 5.The strains reported are normalised by the allowable strains for tension and compression for both UD and biaxial materials.
Fig. 6 shows the resultant deflection of both designs at several points along their lengths.

C. Discussion
The figures in the previous subsection demonstrate the effect of using a Single Web (SW) or Double Web (DW) design for the blade.Fig. 4 shows the distribution of the fibre-direction strain along the length of the blade spar caps, comparing the two blade designs.Only the longitudinal strains are reported in this paper.However, the results of the transverse and shear strain distributions were also analysed and presented as supplementary material to the original report [11].The results of the analyses for the seven regions of the blade (Leading Edge Upper and Lower, Trailing Edge Upper and Lower, Spar Caps Upper and Lower and the Shear Webs) and for all three strain components are summarised in Table 3.
The results in the table indicate that the most significant differences between the strains in the two blade designs occur for the shear strains, most notably in the leading edge and spar cap laminates.The summary indicates that the shear strains are distributed differently in the blade between the two designs, indicating increases in the spar cap shear strains in the Double Web design, but increases in the trailing edge shear strains in the Single Web design.
The values of the shear strains are below the allowable limits for UD and biaxial materials, respectively, for all of the laminates except the leading edge laminates in the Single Web model.This location will be a focus for further analysis in the next stage of the blade design process.For the Single Web model the shear strain in the webs exceeds the allowable strain from approximately 4 m to 7 m along the blade.The thicknesses of the webs, however, is quite low at 20 mm for the Single Web blade in this study.
Fig. 6 shows the difference in the deflection of the two blade designs.The overall tip deflection is within approximately 5% for the two designs.The rotation of the tip of both blade designs was less than approximately 1°.Fig. 7 shows contour plots of the longitudinal strain component for both blade designs.The strain in the outermost biaxial plies in the blade is shown.The large deformation scale factor highlights the rotation and deflection of the blade along its length for the two designs.The top view in Fig. 7 shows how the Single Web design has resulted in the concentration of the tensile strain in the spar cap towards the trailing edge of the blade, whereas the strain is more evenly distributed across the width of the spar cap in the Double Web model.The middle and bottom views also highlight the increase in strain in the leading and trailing edge laminates with a Single Web design.The results of this preliminary design study indicated the regions of risk and uncertainty in the design that were addressed in the subsequent blade laminate optimisation task of the FloTEC project.
In this optimisation task, the regions of the blade with high strains (noted in the SW plotlines in Fig. 4 and Fig. 5) were addressed in particular to ensure an even strain distribution across the blade within safety margins.This resulted primarily in increased laminate thicknesses and optimised layup drop rates along the blade.

D. Summary
While the strains in various regions of the blade were redistributed due to the overall change in stiffness of the load-carrying blade spar, the resultant strains were well below the failure strains for the material in all regions except the leading edge laminates in the single web model, which requires further study.The strains retained a factor of safety of 2 for the UD spar cap material at a minimum for both designs.The single web design was, therefore, chosen for further analysis, due to its reduced complexity for manufacturing.

E. Test facility
The facility at NUI Galway for full-scale testing consists of a 375 m 2 high-bay Large Structures Testing Laboratory.The components of the laboratory relevant to the current testing requirements include: • Flexible testing spaces that allow for testing small to large structural and mechanical elements and materials.
• 10 m x 6 m strong floor with anchor points at 750 mm to 1000 mm centres each having a capacity of 500 kN.
• Servo-hydraulic testing machines with capacities ranging from 10 kN to 750 kN.
• Hydraulic ring main that allows convenient, efficient test setup, operation, and maintenance of individual test systems without disturbing other systems (working pressure of hydraulic ring main is max 210 bar and with a flow capacity of 300 l/min).
• Data acquisition systems and sensors for measurement of load, deformation, strain and acceleration (covered in detail in Section F).
Fig. 8 shows a computer-generated image of the test setup identifying the major components and capabilities of the laboratory.

F. Test setup
An overview of the test setup is shown in Fig. 9.The blade is supported at its root on a support frame.The support frame is constructed from a steel cylinder providing adequate stiffness and load transferral to 9 bolts on the strong floor.The blade is loaded via three hydraulic actuators ranging in capacity from 250 kN to 750 kN.In the current setup, the actuators hang from the modular steel frame and apply loads vertically.The three actuators can be controlled separately and in unison for application of complex loading patterns for static and fatigue testing.The load is applied to the surface of the blade through six contact pads and three load introduction mechanisms.The mechanisms split the load from each actuator to two contact pads, thereby doubling the of loading points and increasing the accuracy of the applied bending moment and shear force profiles along the blade.

G. Instrumentation
This section provides details of the instrumentation available for testing, guidelines on the locations for installing the instrumentation, guidelines on the installation procedure, the technical requirements for testing and details of the acquisition requirements of each type of sensor.The instrumentation includes: • Electrical resistance strain gauges applied to the surface of the blade.
• Videometric measurements using Digital Image Correlation (DIC), a 3D laser scanner and a laser scanning vibrometer.
• Load cells at the locations of load application to the blade.
• Accelerometers for natural frequency testing.
Blade deflections of interest include: tip deflection measured in three directions and vertical deflections at up to five other positions along the length of the blade.Measuring the deflections in x-and y-directions at the leading and trailing edges of the blade can also capture the rotational response of the structure.Any deflection of the root support frame or adaptor plate will result in significant parasitic deflections at the blade tip.Hence, it is recommended that the front face of the root attachment fixture is also instrumented.Similarly, the central beam of the reaction frame (upon which the actuators are mounted) should be monitored for vertical deflections.
The non-contact DIC measurement system can provide accurate 3D coordinates, displacements and surface strains.The system consists of two HD stereo cameras.DIC requires a stochastic high contrast pattern to be painted onto the surface of the blade to capture the deformations and strain.Accurate measurements may also require

Fig. 1 .
Fig. 1.Chord length and hydrodynamic twist angle distributions along the length of the blade, normalised by the maximum value of each variable.

Fig. 2 .
Fig. 2. The flapwise (Fx) and edgewise (Fy) loading distributions for the five load cases, normalised by the maximum static flapwise load.

Fig. 3 .
Fig. 3. Naming convention for each region in the blade cross-section.

Fig. 4 (Fig. 4 .
Fig. 4. The maximum longitudinal normalised strain along the blade for (a) the innermost biaxial layer on the pressure side spar cap, (b) the outermost unidirectional layer on the pressure side spar cap, (c) the outermost biaxial layer on the pressure side spar cap.

Fig. 5 .Fig. 6 .Fig. 7 .
Fig. 5.The longitudinal strain along the blade in the highest loaded biaxial layer in the shear web(s) for both the Single and Double Web designs.

Fig. 4 (
b) indicates that the choice of one or two webs has minimal effect on the longitudinal strain in the spar caps of the blade for the UD layers.For the biaxial layers on the inner and outer surface of the blade (Fig.4(a) and (c) respectively) the Double Web design shows a slight increase in the tensile and compressive strains from approximately 4 m along the blade to the tip.The strain values in the figures are normalised by the allowable strains for tension and compression for the UD and biaxial materials.The figures indicate the strains are well within margins of safety for both materials.The driving load case for the analyses was the fatigue damage equivalent load (DEL).

Fig. 5
Fig. 5 shows the strain distributions in the shear webs of both designs.The web towards the leading edge of the blade is designated DW1 and the trailing edge web DW2, in the figure.The Single Web model demonstrates slightly higher strains than the Double Web model along the blade, peaking at the 3 m location.The strain in the DW1 web is nearly equal to the Single Web model from approximately 3 m to the tip.The peak value of strain occurs at approximately the same location as the Double Web model.The DW2 web shows higher strains than the DW1 web up to the 2 m location, after which the strains decrease along the blade length.The values of the longitudinal strains remain significantly lower than the allowable strains.

Fig. 8 .
Fig. 8. Testing capabilities of the MaREI Centre's Large Structures Testing Laboratory at the National University of Ireland Galway.

Fig. 9 .
Fig. 9. 3D rendering of the test setup with the main components highlighted.

TABLE 2 LAYUPS
APPLIED IN EACH SECTION OF THE BLADE MODEL FOR SPAR CAP, OUTER HYDRODYNAMIC SHELL AND SHEAR WEB REGIONS ALONG THE BLADE

TABLE 3 SUMMARY
OF FINDINGS COMPARING THE SINGLE WEB (SW) AND DOUBLE WEB (DW) BLADE DESIGNS FOR ALL FIVE LOAD CASES COMBINED.