- Email: [email protected]
Application of bend-twist coupled blades for horizontal axis tidal turbines
Renewable Energy 50 (2013) 541e550
Contents lists available at SciVerse ScienceDirect
Renewable Energy journal homepage: www.elsevier.com/locate/renene
Application of bend-twist coupled blades for horizontal axis tidal turbines R.F. Nicholls-Lee*, S.R. Turnock, S.W. Boyd Fluid Structure Interactions Research Group, University of Southampton, Southampton, Hampshire, SO17 1BJ, UK
a r t i c l e i n f o
a b s t r a c t
Article history: Received 2 December 2011 Accepted 25 June 2012 Available online 16 August 2012
The blades of a horizontal axis tidal turbine are required to operate in a harsh subsea environment over a long life cycle with minimal need for maintenance. The concept of using passively adaptive, bend-twist coupled spars for horizontal axis tidal turbine blades has been identified as a potential method of improving energy capture. In this work a structural analysis is coupled with a fluid dynamic model to perform a full fluid-structure interaction analysis of a range of composite, bend-twist coupled blades. Blade element momentum theory is used to assess the presence of stall and corroborate the performance data attained from the fluid analysis. This paper discusses the individual analyses and the manner in which they are coupled. Several example problems were analysed using the design tool. The results compare well to the preliminary studies and indicate that a decrease of up to 12% in thrust and an increase of up to 5% in power capture could be achieved through the use of properly designed, bendtwist coupled blades. Ó 2012 Elsevier Ltd. All rights reserved.
Keywords: Adaptive composites Bend-twist coupling Composite blade design Fluid-structure interactions Renewable energy Tidal turbine
1. Introduction The oceans are a large, renewable, resource of untapped energy. There are many marine renewable energy sources however tidal energy has the advantage of being highly predictable and less susceptible to climate change than most [1e7]. Due to environmental concerns regarding potential devices it is thought that a breakthrough will occur in the area of kinetic energy devices; however, technology is at an early stage of development and further research into the field is required to advance the concepts, improve the feasibility of maintenance and make devices more efficient and economic. The blades of a Horizontal Axis Tidal Turbine (HATT) are the only means of extracting energy from the tidal flow and therefore the efficiency, and consequently annual energy capture, of a device could be increased by improving the blade design. With design for through-life performance and decommissioning becoming ever more prevalent, turbines are required to withstand the aggressive subsea environment for many years whilst being environmentally disposable come the end of their life. This has a large impact on fatigue loading, with an HATT typically experiencing in the order of 1 108 rotational cycles over a 20 year life span. Offshore maintenance is already a costly procedure and this, coupled with the fact that tidal arrays are typically located in remote regions at sea in areas of high flow velocity, promotes a need for minimal * Corresponding author. E-mail address: [email protected] (R.F. Nicholls-Lee). 0960-1481/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.renene.2012.06.043
maintenance over long periods. Optimisation of the blade design could increase the amount of energy capture, reduce structural loading and also minimise the need for maintenance. Composites offer several advantages over metals such as superior fatigue characteristics, high stiffness to weight ratio, ease of manufacture of structures with complex curvature and a reduction in inertial loading. The fundamental assumption is that the design methodology for a composite lifting surface, such as a turbine blade, should in general follow that for conventional isotropic materials that use sufficient stiffness to maintain a given optimum design shape. This conventional approach to lifting surface design relies on their being a fixed optimum shape. As the flow regime experienced is rarely steady, however, every shape will deflect under a given load condition and will dynamically respond to the transient fluid load fluctuations. Composites, however, offer the potential for hydroelastic tailoring; with the use of adaptive composites having been identified as a potential method for load reduction, increased efficiency and enhanced control of wind turbine blades [8e16] and propellers [17e20]. The concept of using adaptive, bend-twist coupled, composite blades in order to improve energy capture but also decrease design complexity has been considered. Preliminary analysis suggests that a 2.5% increase in annual energy capture and a 10% decrease in thrust loading could be expected through the use of a bend-twist coupled adaptive HATT blade [21e25]. There are many parameters that require optimisation in isotropic HATT blade design; not least diameter, section shape, thickness/chord ratio, pitch, skew, rake. Integrating adaptive composite materials into the blade also
542
R.F. Nicholls-Lee et al. / Renewable Energy 50 (2013) 541e550
requires that other variables are optimised such as the material properties, number of plies and the ply angle. Ultimately the blade analysis becomes complex and involves many time consuming iterations of Blade Element Momentum Theory (BEMT), Finite Element Analysis (FEA) and Computational Fluid Dynamics (CFD) so as to produce an optimal design. Following on from the preliminary investigations, the aim of this work is to gain a more detailed knowledge of the performance of a composite, bend-twist coupled HATT blade by performing a full fluid-structural interaction analysis. This includes the development of a design tool for adaptive HATT blades incorporating fluid and structural analyses and BEMT in a full Fluid Structural Interaction (FSI) analysis. 2. Adaptive composites An adaptive textile composite is a structure tailored to exhibit desirable elastic deformation behaviour not necessarily proportional to the imposed load. An example of such a structure would be a box beam so tailored that an imposed cantilever load results in twisting as well as bending, although no torsional load was imposed. Such a structure is said to exhibit bend-twist coupling. Composite blade elastic coupling can not only be tightly controlled but also varied over the span by appropriate selection of ply angles, thicknesses and spanwise layup. A 60e70% reduction in production costs was seen in the marine propeller industry after research into coupled composite propeller blades, along with smoother power take up, reduced blade vibration, reduced noise, and better fatigue performance [26]. Much research has been undertaken in this field in the wind industry [8e16,27e29]. Wind turbines carry loads primarily by twisting and bending, much like tidal turbines. The level of load reduction depends on the twist distributed along the blade length, which is controlled by the amount of bend-twist coupling. This, in turn, depends on the blade cross sectional geometry, the level of anisotropy in the structural material, and the material distribution [27]. For conventional laminated composites constructed of orthotropic layers, the level of anisotropy is determined by the fibre orientation with respect to the primary loading direction. Kooijman [28] found that the mirror layup, Fig. 1, is required for bend-twist coupling to be exhibited; with a 20 fibre angle orientation achieving a maximum level of coupling [30]. 3. Fluid structure interactions Fluid-structure interactions (FSI) occur when a fluid interacts with a solid structure, exerting pressure that causes deformation in the structure subsequently altering the flow of the fluid itself. If a situation involving structure flexure is to be analysed it is highly beneficial to couple both the fluid dynamics and the structural analysis programs to produce iterative solutions for complex problems. In operation an HATT is subjected to hydrostatic pressure, and as it is a lifting body with rotational motion it is subjected to further hydrodynamic forces e thrust, torque, cavitation etc. Due to the
Fig. 1. Layup for bend-twist coupling illustrating that the composite plies in the top and bottom surface of the central area of the blade must mirror each other for coupling to occur.
complex loading scenario experienced by a tidal turbine, knowledge of the hydroelastic behaviour of the blades, hub, nacelle, and also the support structure under this regime could lead to a more thorough understanding of structural constraints and how performance of the turbine could be improved. Whilst being a highly informative technique for the design analysis of a standard fixed blade device (assumed to be stiff), FSI is essential when considering adaptive composite blades as the pressure loading alters the shape more significantly when compared to standard blades. There are three methods of joint fluid structural modelling in the time domain which involve solving the governing equations in a coupled, uncoupled or integrated manner [31e35]. In this work, a loosely coupled, modular approach is used; the flow problem and structural problem are solved successively until the change is smaller than the convergence criterion. This method has the benefit that the two domains are discretised to better suit the problem; as the mesh for the fluid analysis will tend to require greater refinement in different areas of the geometry than that for the structural analysis, and vice versa. The program alternately manipulates both the surface pressure distribution yielded from the fluid dynamics program and the displacement data output from the structural analysis, and feeds them back into the next relevant stage of the process. This is illustrated schematically by the flow chart in Fig. 2. Similar methods to this have been shown to be successful by Turnock and Wright [31] for the analysis of rudder propeller interactions. 3.1. Fluid dynamics analysis The surface panel code used in this work was PALISUPAN (PArallel LIfting SUrface PANel) [36]. This code has previously been used to model the behaviour of a representative tidal turbine and good comparisons were obtained with published data, although problems were experienced in getting low twist sections to work due to the low pitch of the wake sheet [37,38]. Previous studies indicated that an optimum panel distribution can be achieved that maintains the accuracy of the result obtained with a finer distribution, but reduces the calculation time to around 15 min when using a standard desktop PC with one quad core Intel Xeon processor and 12 GB of RAM [37]. In comparison a full Reynolds Averaged Navier Stokes simulation of a similar tidal device carried out over 128 quad core Intel Nehalem processors on the 152 node Redrock computer cluster at the National Renewable Energy Laboratory would take up to 24 h depending on the mesh density [39]. Parameterisation and optimisation of surface panel codes is relatively simple, due to the low process times when implementing multiple runs e over 30 at a time e being highly feasible. Using a frozen wake model it is possible to reproduce the helical wake characteristic of tidal turbines. The number and distribution of panels in the wake is also very important for accurate modelling of device performance. Codes have been developed that generate panel distributions over complex shapes, such as a propeller or tidal turbine, and the associated wake panel arrangement. Such a code, Propgen [40] was used in this case to create the input blade geometry. Blade section data, section offsets, panel number and distribution for both blade and wake were input and the geometry is constructed from a set of section curves. The 2-D section is then mapped onto a cylindrical surface according to the specified variables using a transformation matrix. The appropriate pitch of the wake can be found through consideration of the axial and radial momentum changes in the flow due to the tidal turbine. Once the blade geometry was generated it is then exported to Adaptflexi [41], a program which enables geometry manipulation and definition of the fluid analysis domain and flow variables, in which a surface is lofted through the sections to generate the blade. Additional variables are now defined for the HATT scenario such as
R.F. Nicholls-Lee et al. / Renewable Energy 50 (2013) 541e550
543
Fig. 4. Panel distribution over blade and wake.
Fig. 2. Flow diagram of the fluid structure interaction simulation process. This illustrates the loosely coupled analysis with the fluid and structural solvers on the left of the blue line operating separately from each other, but being coupled by the Matlab based code on the right of the blue line. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
flow velocity, RPM, number of blades, water density, and panel discretisation; and subsequently a surface mesh created consisting of quadrilateral panels. Good mesh distribution definition, over both blade and wake, is important for accurate modelling of device performance [38,42]. Fig. 4 illustrates a single blade (colour) and wake panel distribution for analysis in the CFD code of a 20 m
Fig. 3. Validation study for an HATT in Palisupan.
diameter, three bladed HATT; there are in excess of 2800 panels in the blade of the turbine alone. A panel sensitivity analysis was carried out for a 20 m diameter, 3 bladed HATT [37]. The number of panels on the blade was varied until the total force and momentum differed by less than 2.5% as the number of panels was increased. This was found to require of the order of 2800 panels on the blade with approximately 40 panels in the chordwise direction and 75 in the spanwise direction. This ratio of panels is comparable with that typically found both for low aspect foils (rudders/keels) [42] but also the much lower aspect ratio propeller blades [40]. A further variable which is the Kutta condition for convergence was maintained at a value of 0.01. Fig. 3 presents a comparison of the performance of the BEMT code, Cwind [43], and the lifting surface panel code, Palisupan, alongside the original experimental data of Barnsley et al. [43] for a tip pitch setting of 2 . The BEMT section was represented by a single series of NACA 63-2xx Cl and Cd performance data. Following the work of Barnsley et al. [43] the influence of threedimensional behaviour is to prolong stall, and give much higher working CL before stall is initiated. This extra increment of lift was included through modification of the BEMT code. In comparing the potential flow based analysis from Palisupan with the BEMT excellent agreement is observed in the range of turbine operation 4 < TSR < 6. For this turbine, at TSR > 6, stall behaviour dominates which is captured by the BEMT but not by the potential flow analysis, which excludes stall [42]. In the full fluid-structural interaction analysis presented in this paper, the BEMT code is used to check the validity of the CFD results such that it can be determined when stall occurs, and the effect on turbine performance. This is discussed in more detail later. In the immediate wake a finer panel distribution can be observed when compared to the far wake downstream. This captures the flow more precisely leading to a more accurate solution for the fluid dynamic analysis. The turbine is rotationally symmetric and therefore the body coefficients are rotated and copied twice more to simulate the whole entity. The output from Adaptflexi is in the form of a .uns file which contains the flow data for the simulation along with a list of panels throughout the turbine and wake model and the corresponding corner points. The problem is then solved using Palisupan [36] and the resulting pressure loading across the surface of the blade was then calculated for input to the structural model, along with turbine performance data in the form of power coefficient, thrust and data information.
544
R.F. Nicholls-Lee et al. / Renewable Energy 50 (2013) 541e550
b 6.8 Max. Spar induced twist (o)
Max. Spar Bend (m)
a 1.16 1.14 1.12 1.10 1.08 1.06 1.04 1.02 1.00 0.98 0
50000
100000
150000
200000
250000
No. of Elements
6.7 6.6 6.5 6.4 6.3 6.2 6.1 6.0 5.9 5.8 0
50000
100000
150000
200000
250000
No. of Elements
Fig. 5. Solution convergence for increasing mesh density, a) Bend, b) Induced twist.
3.2. Structural analysis The blade was modelled in ANSYS 12.1 [44]. The model was meshed using the SHELL 281, 8 node, finite strain, shell element and solved using a large displacement static non-linear analysis. The SHELL 281 element is suitable for analysing thin to moderatelythick shell structures, i.e. the ratio between surface area and thickness of a single element must be greater than ten. It is wellsuited for linear, large rotation, and/or large strain nonlinear applications. Each layer of the laminate is defined by the material properties and thickness e modelled from the inner skin out to the outer roving. Each model consisted of in excess of 200,000 quadrilateral elements. This was a very fine mesh and was such that discretisation issues were not significant [38]. An example of a 20 m diameter HATT was solved structurally in ANSYS 12.1 for ten different mesh densities, ranging from under 25,000 elements to over 225,000 elements. The sensitivity of solution for both the maximum bend and induced twist over the entire length of the central spar at different mesh densities are illustrated in Fig. 5a and b respectively. Both solutions converged as the mesh density increased, with the optimum number of elements for both accuracy of solution and minimising computational time being 200,000. The computer used was a standard desktop PC with four Intel Xeon processors and 24 GB of RAM. The geometry created for the fluid analysis is imported into ANSYS 12.1, and a central spar fitted through the blade, Fig. 6. This ensured that nodes existed at the corner points of each panel of the mesh used for the fluid analysis. The spar was fitted such that it supported the blade at one third of the local chord distance from the leading edge; this is where the centre of effort of the blade typically acts. The resulting geometry was then meshed and the pressure loads from the fluid dynamic analysis applied. This was achieved directly, without requiring interpolation, by virtue of the FEA geometry having the same areas mapped out as the surface panel fluids model; the pressure at the centre point of each panel is output from Palisupan and applied to the corresponding panel in the structural model. The blade structural mesh was created in the smallest areas first to ensure that it was adequately dense in these regions, resulting in a higher quality of mesh overall. The coupled spar was theoretically “laid up” in the manner shown in Fig. 7. The structural deformation problem was solved using a static non-linear analysis such that large deflection effects were taken into account [23,24]. The maximum stress seen in the blade is monitored with the criteria that it must not exceed a third of the yield stress of the beam material; thereby incorporating a safety factor of three in to the design. Experimental analysis on a bendtwist coupled, double box beam of the same configuration as Fig. 7 has been undertaken in order to corroborate the numerical results [32]. In this work the partial safety factor method, commonly used in the wind turbine design industry [45], has not yet been integrated
and therefore a safety factor of three was chosen in an effort to account for the through-life stresses present on the turbine. If this value is exceeded the number of plies in the mid-layer of the bendtwist coupled beam is altered, and the Finite Element Analysis (FEA) rerun for the same pressure loading until an acceptable stress level is reached. The deflections are output at the nodes corresponding to the panel corner points in the fluid analysis geometry. The deflection of the blade is then remapped into a new .uns input file for the fluid dynamic analysis and the problem rerun. This simulation is continued in an iterative manner until the deflection of the blade between successive runs is less than 10 mm; i.e. it has converged. The displacement value of 10 mm was chosen as this was small (0.05%) when compared to the turbine diameter of 20 m. 3.3. Coupling design Tools Coupling of both the structural and fluid dynamic analyses has been carried out in Matlab 2009b. Fig. 8 shows a flow chart of the process. Initially tables of data for the range of design parameters under investigation are generated; these variables include design RPM, diameter, flow velocity, section shape and number of blades. The relevant data for the first run is then extracted from these tables and the Propoptions and Propeller input files for Propgen created. Propgen is then run to create the blade, hub, and wake geometry and a script is generated to pass the geometry through Adaptflexi. The output required from the fluid analysis is governed through this script; pressure information, panel discretisation, and overall turbine loadings. Adaptflexi is then used to generate the .uns file and .cmd file and the case run in the fluid analysis code. Structural variables are then introduced in the form of ply angle, number of plies and material data. The intention being to create a design which maximises induced twist whilst minimising bend. The coupling code then takes the geometry data from the fluid dynamic analysis and generates a blade within ANSYS that has a coupled box beam as the main central spar. The required lay-up information and the pressure loading are then applied to the
Fig. 6. Geometry of blade showing the central spar (a) and surrounded by the blade skin (b).
R.F. Nicholls-Lee et al. / Renewable Energy 50 (2013) 541e550
Fig. 7. Lay-up of central, bend-twist coupled, spar illustrating the extruded aluminium mould at the centre which was removed post-cure.
corresponding areas and the problem solved. Once a solution is achieved the stress check detailed earlier is then performed. The deformed positions of the loaded blade are then calculated through knowledge of the displacements of each node; and bend and induced twist determined.
545
The nodal displacements are then transferred to the .uns file for the relevant panel corner points. The panel list in the .uns file is divided into trailing edge body panels, main body panels, wake panels attached to the trailing edge, mid wake panels and far wake panels. The displacements of each panel on the trailing edge of the main body are known from the FEA and are evaluated as axial, radial and angular displacements in cylindrical co-ordinates and then applied to the body keypoints in the .uns file. These displacements are then translated back through the wake panels to give a deformed wake to match the warped blade. The problem is then run through the fluid analysis program again to determine a new surface pressure loading for the deflected blade. This loading is applied to the deflected blade in ANSYS and the structural analysis carried out. These stages are repeated until convergence. In order to perform a final check on the data obtained, once convergence has occurred the input files for the BEMT code are created from the main blade parameters and performance
Fig. 8. Flow chart detailing the coupling process.
546
R.F. Nicholls-Lee et al. / Renewable Energy 50 (2013) 541e550
Table 1 Example case blade design data.
Table 2 Material mechanical properties [47].
r/R
Chord (m)
P/D
t/c
Coupled spar SE84LV HEC 300/400
Blade skins SE84LV RC200T
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
2.304 2.033 1.798 1.576 1.367 1.170 0.986 0.814 0.655
0.3365 0.2263 0.1593 0.1166 0.0866 0.0678 0.0556 0.0458 0.0358
0.0858 0.0753 0.0653 0.0616 0.0581 0.0546 0.0511 0.0476 0.0441
Property
Value
Property
Value
Fibre weight Cured ply thickness Ex EY Ez
300 g/m2 0.281 mm 129.2 GPa 8.76 GPa 8.76 GPa 0.335 0.0172 0.0172 5.76 GPa 5.76 GPa 5.76 GPa
Fibre weight Cured ply thickness Ex EY Ez
195 g/m2 0.214 mm 65.2 GPa 65.2 GPa 8.76 GPa 0.05 0.05 0.05 5.76 GPa 5.76 GPa 5.76 GPa
characteristics calculated for the device. This data is compared to that gained through the final iteration of the fluid analysis to ensure that the thrust and torque estimates are similar and the turbine has not stalled. In this manner a composite, bend-twist coupled blade can be designed for use on an HATT. 3.4. Example case In order to illustrate the FSI blade design method, an example design problem was considered [38]. The design spring peak current is 2.5 m/s, with both the ebb and flood stages of the tide assumed to have similar time history, i.e. they have been
nxY nYZ nxZ
GxY GYZ GxZ
nxY nYZ nxZ
GxY GYZ GxZ
approximated as a double sinusoid. The turbine has a diameter of 20 m, a hub-diameter ratio of 0.2 and three bend-twist coupled blades. The section shape to be used for the blade is the NACA 63815 which was shown previously to perform well [46]. The blade is of the form detailed in Table 1. A three-dimensional representation of the example turbine is illustrated in Fig. 9. The bend-twist coupled spar is constructed of SE84LV UD carbon pre-preg, and the blade skins made from SE84LV RC200T twill weave carbon pre-preg (Table 2) both from Gurit [47]. The blade central spar initially has 20 plies in the mid-layer at 20 , 30 plies in the inner skin at 45 and 5 plies in the outer roving at 90 e each ply has a thickness of 0.000281 m. The number of plies in the mid-layer are then increased in steps of five plies until the minimum stress criteria is achieved (sy/3). The maximum displacement is at the tip of the blade, with a value of 1.48 m. For this particular turbine, the number of mid layer plies was increased to 35 in order to meet the maximum stress criterion, and the solution converged in 16 iterations of the FSI loop. There is good correlation between the results from the final iteration of the fluid dynamic study and the BEMT analysis, with less than 1% difference in thrust coefficient and 2% difference in power coefficient [38]. The main properties of the turbine, and performance results, are detailed in Table 3. 3.5. Discussion of results The design tool was used to analyse a range of turbines with bend-twist coupled blades in which the number of blades, flow velocity and diameter were varied whilst rotational speed was kept constant at 12 RPM. The blade design for each turbine e section, chord, P/D etc. e is the same as detailed in Table 1. Table 4 details the results of this study indicating bend, induced twist, and CT and Cpow from both the fluid dynamic and BEMT analyses. It is apparent that the main limitation of BEMT, the assumed relation between disc and far wake induction, has a strong effect at Table 3 Case study turbine principal particulars and performance data. Current velocity Diameter Number of blades Revolutionary speed Mid-layer ply angle Number of mid-layer plies Thrust Torque Bend Induced twist Maximum stress Thrust coefficient Power coefficient Absorbed power
Fig. 9. 3-D representation of the example HATT.
2.5 20 3 12 20 40 724.5 792.3 1.48 8.6 28.1 0.72 0.40 1248.9
(m/s) (m) (blades) (RPM) (o) (plies) (kN) (kNm) (m) (o) (GN/m2) e e (kW)
Table 4 Variation of performance parameters for a selected set of adaptive tidal turbine designs over a range of blade numbers, diameters and tip speed ratios. It is assumed that all designs are operating at mid-depth of a channel of depth 2D. The blade tip cavitation number for each device does not fall below 1.0 and therefore the effect on performance can be neglected [42]. Case no.
2 2 2 2 2 2 2 2 2 2 2 2 3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4
V
f
(m/s)
1.5 1.5 1.5 1.5 2 2 2 2 2.5 2.5 2.5 2.5 1.5 1.5 1.5 15 2 2 2 2 2.5 25 2.5 2.5 1.5 1.5 1.5 1.5 2 2 2a 2 2.5 2.5 2.5 25
D (m)
10 15 20 25 10 15 20 25 10 15 20 25 10 15 20 25 10 15 20 25 10 15 20 25 10 15 20 25 10 15 20 25 10 15 20 25
TSR
4 8
4
6 4
6 2
4.189 6.283 8.378 10.472 3.142 4.712 6.283 7.854 2.513 3.770 5.027 6.283 189 6.283 378 10.472 3.142 712 6.283 7.854 2.513 3.770 5.027 283 189 6.283 8.378 10.472 3.142 4.712 283 7.854 513 3.770 5.027 6.283
Bend (m)
0.08 0.639 2.513 4.823 0.061 0.487 1.912 3.589 0.05 0.396 1.548 2.902 0.071 0.571 2.25 4.234 0.054 0.435 1.709 3.212 0.044 0.353 1.382 2.594 0.064 0.515 2.033 3.832 0.049 0.392 1.541 2.901 0.04 0.317 1.245 2.34
Ind twist ( )
0.94 4.98 14.41 21.53 0.72 3.81 11 06 16.36 0.59 3.09 9.00 13.36 0.84 4.46 12.95 19.10 0.64 3.40 9.91 14.72 0.52 2.76 8.05 11.98 0.75 4.02 11.74 17.39 0.57 3.06 8.95 13.35 0.47 2.48 7.26 10.84
Fixed
Adaptive
CT CFD
Cpow CFD
CT BEMT
Cpow BEMT
CT CFD
Cpow CFD
CT BEMT
Cpow BEMT
0.194 0.286 0.777 1.132 0.217 0.971 1.073 0.860 0.453 0.766 0.670 0.535 0.140 0.231 0.269 0.302 0.135 0.378 1.041 1.186 0.172 0.834 1.030 0.805 0.091 0.170 0.231 0.251 0.124 0.174 0.256 0.670 0.119 0.191 0.973 1.102
0.046 0.107 0.154 0.379 0.077 0.314 0.438 0.404 0.139 0.378 0.336 0.279 0.033 0.044 0.052 0.051 0.041 0.112 0.221 0.399 0.038 0.197 0.424 0.375 0.003 0.040 0.026 0.044 0.025 0.024 0.047 0.133 0.026 0.032 0.226 0.420
0 161 0.200 0.543 0.821 0.179 0.809 0.751 0.602 0.374 0.638 0.468 0.374 0.116 0.161 0.188 0.219 0.111 0.315 0.728 0.829 0.142 0.695 0.720 0.563 0.705 0.119 0.161 0.182 0.102 0.145 0.179 0.468 0.098 0.159 0.680 0.770
0.023 0.057 0.140 0.355 0.043 0.283 0.402 0.381 0.088 0.375 0.320 0.274 0.017 0.023 0.048 0.048 0.023 0.101 0.203 0 377 0.024 0.195 0.404 0.367 0.002 0.021 0.023 0.041 0.014 0.022 0.043 0.125 0.016 0.032 0.215 0.412
0.172 0.256 0.693 1.018 0.196 0.877 0.973 0.779 0.413 0.697 0.614 0.490 0.125 0.207 0.240 0.270 0.122 0.344 0.942 1.074 0.158 0.765 0.944 0.739 0.081 0.153 0.208 0.225 0.113 0.158 0.233 0.609 0.109 0.175 0.885 1.015
0 0 0 0 0
0 0 0 0 0
0.024 0.58 0.146 0.372 0.044 0.291 0.418 0.400 0.090 0.386 0.333 0.287 0.017 0.024 0.050 0.050 0.023 0.104 0.211 0.395 0.024 0.201 0.420 0.386 0.002 0.022 0.024 0.043 0.014 0.023 0.045 0.132 0.017 0.033 0.224 0.433
0 0 0 0
047 110 160 398 078 0.322 0.451 0.418 0.142 385 346 293 034 0.045 0.054 0.054 0.041 0.115 0.228 0.413 0.039 0.200 0.437 0.394 0.003 0.041 0.027 0.046 0.025 0.025 0.049 0.137 0.026 0.033 0.233 0.442
147 178 495 738 162 0.728 0.678 0.544 0.338 0.570 0.420 0.339 0.106 0.144 0.172 0.196 0.100 0.284 0.657 0.750 0.128 0.622 0.645 0.510 0.069 0.106 0.147 0.163 0.092 0.130 0.162 0.424 0.088 0.142 0.610 0.699
Change in CT (%)
Change in Cpow (%)
-12.23 -11.86 -12.11 -11.23 -10.62 -10.74 -10.25 -10.50 -9.65 -9.77 -9.05 -9.17 -11.98 -11.61 -12.23 -11.48 -10.25 -10.01 -10.50 -10.38 -9.29 -9.05 -9.17 -8.81 -11.23 -11.36 -11.11 -11.36 -9.53 -9.77 10.01 -9.89 -9.17 -8.81 -9.89 -8.58
2.15 2.82 3.66 4.58 1.48 2.44 2.92 3.29 1.86 1.62 2.94 4.85 2.18 2.87 3.73 4.66 1.54 2.49 2.97 3.35 1.93 1.66 2.98 4.91 2.24 2.88 3.74 4.67 1.55 2.51 3.00 3.36 1.94 1.67 2.99 4.92
R.F. Nicholls-Lee et al. / Renewable Energy 50 (2013) 541e550
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 I8 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
No. of blades
547
548
R.F. Nicholls-Lee et al. / Renewable Energy 50 (2013) 541e550
high induction factors and thrust coefficients; indicated in Table 4 by the lower prediction of CT from the BEMT. The fluid analysis predicted slightly higher values of Cpow than the BEMT. This was due to the fact that the iterative boundary layer prediction method was not included in the Palisupan calculations. In practice a performance loss would be present due to the onset of stall; however, the cases at lower flow velocities in which stall was not occurring show better correlation between the BEMT and fluid analysis results. The solidity of a blade is defined as total blade area divided by the swept area e this can be altered by changing the number of blades or altering the blade chord [48]. It can be seen from Table 4 that Cpow changes less over a wide TSR for the two blade turbines (low solidity), cases 1e12. The higher solidity turbines tend to perform better at lower TSRs, but are more sensitive to TSR changes. In general turbines require a minimum of three blades to be selfstarting at low flow velocities, with two bladed devices requiring a greater flow velocity to start up thus negating the broader capture area under the Cpow curve that the low solidity creates. CT can be observed to increase with an increase in solidity, although this is also dependent on the turbine diameter and resulting TSR. At higher TSR the lower solidity turbines experience greater thrust force. Changing the diameter of the turbine alters the swept area of the device; an increase in diameter increases the amount of power available for extraction due to the increased rotor area. CT and Cpow generally increase with an increase in diameter, but effectively it is the absorbed power of the device that is increasing which is reflected in the non-dimensional coefficients. The four bladed devices follow this trend, however the two and three bladed devices can be observed to have a reduction in CT and Cpow, and hence absorbed power, at the highest flow velocities. The peak performance of the two bladed devices analysed occurred at a flow velocity of 2 m/s with a diameter of 20 m (case 7). The 25 m diameter rotor then underperformed at this, and greater flow velocities for the lower solidity devices. The final two columns of Table 4 show the percentage change of both CT and Cpow between the initial fixed blade and the adaptive blades after the final run of the design tool. There is a general trend for a decrease in CT, and an increase in Cpow with the use of the adaptive blades. This is in the same order as the preliminary work [21]; however, there are some interesting trends appearing with the more holistic analysis using the design tool. Generally the reduction in CT is greater at lower flow velocities (cases 1e4, 13e16 and 25e28) and for lower solidity turbines. Conversely the improvement in Cpow increases slightly with increasing flow velocity. Previous work showed that while the adaptive blades did not increase the maximum Cpow of the turbine greatly, the performance curve broadened indicating the device was more efficient at higher TSR, and thereby capturing more energy over the whole tidal cycle. This, more detailed analysis, confirms this trend; albeit with a greater increase in Cpow at higher TSR. The larger diameter devices exhibit a greater improvement in Cpow, in the region of 4%. This is attributed to the increased amount of coupling present in the blades due to the additional length. Comparing the bend and induced twist of the 10 m and 25 m, three bladed turbines at a flow velocity of 2.5 m/s (cases 21 and 24 respectively) it can be seen that the bend increases from 0.044 m to 2.594 m and the induced twist increases from 0.172 to 11.98 . This is expected as the majority of torque, and hence thrust, to turn the turbine is developed from the outer third of the rotor diameter e the tip of the blade. The larger diameter turbine will therefore not only experience a greater thrust force due to the size increase, but also a significantly bigger bending moment thereby increasing the effect of
the bend-twist coupling inherent in the blade. Due to the manner in which the analysis has been carried out the number of plies in the mid-layer of each of these blades does differ in order to meet the maximum stress criterion which also contributes to the change in degree of coupling; the 10 m device had 35 plies and the 25 m device 45 plies. In this analysis the number of mid-layer plies were optimised for each individual condition (vf, D, number of blades), whereas in reality for a specific number of blades and turbine diameter the composite layup will be the same as the tidal velocity changes. Fig. 10 illustrates three sets of power data for the three bladed device with zero degrees pitch at 12 RPM: the initial fixed bladed device, the preliminary adaptively coupled device, and the coupled device analysed through the design tool. Each device has the same blade chord, section and pre-twist distribution. The results gained through the design tool show good agreement with those of the preliminary analysis. The maximum Cpow from the device analysed through the design tool is higher and occurs at a lower TSR than that for the preliminary device. Fig. 11 illustrates the relationship between CT and TSR for the same three cases. The results attained through the use of the design tool agree with the preliminary adaptive blade results for higher TSRs, but at lower TSR an increased drop in thrust coefficient can be observed. This decrease in thrust loading reduces structural fatigue, potentially increasing turbine life-span whilst reducing the frequency of maintenance required. This has positive implications for the through-life cost of the device, with maintenance being costly in the aggressive subsea environment; a reduction in the level of maintenance required should correspond to a reduction in through-life cost of the turbine. The annual energy capture of HATT can be approximated using the method detailed in NichollseLee et al [22]. Applying this method to the analysis of the example 20 m diameter, 3 bladed example turbine in Section 3.4, the annual energy capture of the adaptive device is 2.5% greater than that of a similar device with fixed blades. The flow regime into a tidal device at any one time is not constant, with variations in the vertical water column apparent alongside effects from turbines in arrays, and thus the blades experience changing inflow depending upon where they positioned in the area of the rotor disk. This transience should be reflected in both the fluids and structural analyses in the design tool and is an area of required further work.
Fig. 10. Relationship between TSR and Cpow for the fixed blade turbine, the preliminary analysis coupled turbine and the coupled turbine analysed using the design tool.
R.F. Nicholls-Lee et al. / Renewable Energy 50 (2013) 541e550
Fig. 11. Relationship between TSR and CT for the fixed blade turbine, the preliminary analysis coupled turbine and the coupled turbine analysed using the design tool.
4. Conclusions Adaptive composite technology has been in use in the aerospace industry for some time, and is now also being integrated into blades for wind turbines; however there are limited applications in the marine field. Preliminary analysis of a new conceptual blade with an integrated bend-twist coupled composite spar showed that such a blade has the potential to increase annual energy capture by around 2.5%. A decrease in thrust loading of up to 10% was also observed, thereby increasing the fatigue life of the turbine and associated structure. A tool for the design of passively adaptive, composite HATT blades has been developed. A surface panel code is used to predict the pressure loading on an initial HATT blade. This data is then used in FEA to assess the structural response of a composite beam. The deformed shape is then fed back into the fluid analysis code and a new pressure distribution calculated. Likewise these forces are imported back into the structural solver and so forth in an iterative manner until a suitable level of convergence of blade deformation is reached. The performance data is then checked using the BEMT code and the design of an adaptive blade realised. Several example problems were analysed using the design tool. The results compare well to the preliminary studies, and indicate that aimproved decrease of up to 12% in CT and an increase of up to 5% in Cpow could be achieved through the use of properly designed bend-twist coupled blades. This implies that structural fatigue loading on the turbine is decreased; hence potentially decreasing through-life maintenance costs and increasing the life-span of the device. Acknowledgements This work was part funded through the EPSRC Project - EP/ I009876/1: Tailored Composites for Tuned Deformation Response to Unsteady Fluid Loading. References [1] Kirke B. Developments in ducted water current turbines; 2005. 12. [2] Charlier RH. A "Sleeper" awakes: tidal current power. Renewable and Sustainable Energy Reviews 2003;7:515e29. [3] Callaghan J, Boud R. Future marine energy. Carbon Trust; 2006. [4] ENTEC and BWEA. Marine renewable energy - state of the industry report. In: UK R, editor; 2009. [5] VanZwieten J, Driscoll FR, Leonessa A, Deane G. Design of a prototype ocean current turbine - part I: mathematical modeling and dynamics simulation. Ocean Engineering 2006. [6] Harrison GP, Wallace AR. Climate sensitivity of marine energy. Journal of Renewable Energy 2005;30(12):1801e17.
549
[7] Pearce N. Worldwide tidal current energy developments and opportunities for Canada’s Pacific coast. International Journal of Green Energy 2005;2:365e86. [8] Karaolis N, Mussgrove P, Jeronimidis G. Active and passive aeroelastic power control using asymmetric fibre reinforced laminates for wind turbine blades. In: Proceedings of the 10th British wind energy conference; 1988. London, U.K. [9] Lobitz DW, Laino DJ. Load mitigation with twist coupled HAWT blades. In: 37th aerospace sciences meeting and exhibit and 18th ASME wind energy symposium. Reno, Nevada, USA: American Institute of Aeronautics and Astronautics; 1999. [10] Lobitz DW, Veers PS. Aeroelastic behavior of twist-coupled HAWT blades. In: 36th aerospace sciences meeting and exhibit and 17th asme wind energy symposium. Reno, Nevada, USA: American Institute of Aeronautics and Astronautics; 1998. [11] Veers P, Bir G, Lobitz D. Aeroelastic tailoring in wind turbine blade applications. In: Windpower ’98, American wind energy association meeting and exhibition. Bakersfield, California, USA: Sandia National Laboratories; 1998. [12] Eisler GR, Veers PS. Parameter optimisation applied to use of adaptive blades on a variable speed wind turbine. Sandia National Laboratories; 1998. [13] Griffin DA. Evaluation of design concepts for adaptive wind turbine blades. Sandia Report. Sandia National Laboratories; 2002. [14] Maheri A. A new approach in simulation of wind turbines utilising bend-twist adaptive blades. Bristol: University of the West of England; 2006. [15] Kooijman H. Bending-torsion coupling of a wind turbine rotor blade; 1996. [16] Lobitz DW, Veers PS, Eisler GR, Laino DJ, Migliore PG, Bir G. The use of twistcoupled blades to enhance the performance of horizontal axis wind turbines. Sandia Report. Sandia National Laboratories; 2001. [17] Young YL. Fluid-structure interaction analysis of flexible composite marine propellers. Journal of Fluids and Structures 2008;24:799e818. [18] Young YL. Dynamic hydroelastic scaling of self-adaptive composite marine rotors. Composite Structures 2010;92:97e106. [19] Liu Z, Young Y. Utilization of bend-twist coupling for performance enchancement of composite marine propellers. Journal of Fluids and Structures 2009;25:1102e16. [20] Khan AM, Adams DO, Dayal V, Vogel JM. Effects of bend-twist coupling on composite propeller performance. Mechanics of Composite Materials and Structures 2000;7:383e401. [21] Nicholls-Lee R, Turnock S. Enhancing performance of a horizontal axis tidal turbine using adaptive blades. In: Oceans ’07. Aberdeen, Scotland: IEEE; 2007. [22] Nicholls-Lee R, Turnock S, Boyd S. Performance prediction of a free stream tidal turbine with composite bend-twist coupled blades. In: 2nd international conference on ocean energy (ICOE 2008); 2008. Brest, France. [23] Nicholls-Lee R, Boyd S, Turnock S. Development of high performance composite bend-twist coupled blades for a horizontal axis tidal turbine. In: 17th international conference on composite materials (ICCM-17); 2009. Edinburgh, Scotland. [24] Nicholls-Lee R, Boyd S, Turnock S. Obtaining experimental data suitable for validation of a numerical design tool for an adaptive composite free stream tidal turbine blade. In: The 11th international symposium on practical design of ships and other floating structures (PRADS 2010); 2010. Rio de Janeiro, Brazil. [25] Nicholls-Lee RF, Turnock SR, Boyd S. Tidal turbine blade selection for optimal performance in an array. In: 30th international conference on ocean, offshore and arctic engineering; 2011. Rotterdam, The Netherlands. [26] Young YL, Michael TJ, Seaver M, Trickey MT. Numerical and experimental investigations of composite marine propellers. In: 26th symposium on naval hydrodynamics; 2006. Rome, Itlay. [27] Locke J, Hidalgo IC. The implementation of braided composite materials in the design of a bend-twist coupled blade. Sandia Report. Sandia National Laboratories; 2002. [28] Kooijman H. Bending-torsion coupling of a wind turbine rotor; 1996. [29] Lobitz DW, Veers PS, Migliore PG. Enhanced performance of HAWTs using adaptive blades. In: Wind energy ’96, ASME wind energy symposium; 1996. Houston, U.S.A. [30] Ong CH, Tsai SW. Design, manufacture and testing of a bend-twist D-Spar. In: S.N. Laboratories, editor. Stanford, California: Department of Aeronautics and Astronautics, Stanford University; 1999. [31] Turnock SR, Wright AW. Directly coupled fluid structural model of a ship rudder behind a propeller. Marine Structures 2000;13(1):53e72. [32] Nicholls-Lee RF. Adaptive composite blades for horizontal axis tidal turbines. In: School of Engineering Sciences. Southampton: University of Southampton; 2011. p. 218. [33] Guruswamy GP. Unsteady aerodynamic and aeroelastic calculations for wings using Euler equations. AIAA Journal 1990;28(3):461e9. [34] Guruswamy, G.P. and Byun, C., Fluid-Structural interactions using NavierStokes flow equation coupled with shell finite element structures, AIAA, Editor. 1993, AIAA. [35] Guruswamy GP, Byun C. Wing-body aeroelasticity on parallel computers. Journal of Aircraft 1996;33(2):421e8. [36] Turnock SR. Palisupan user guide; 2000. [37] Turnock SR, Nicholls-Lee RF. Design of three bladed tidal turbine blades for bidirectional fixed pitch or azimuthing variable pitch operation. In: Logþ1/ Alstom/WUMTIA, editor. Economic viability of a simple tidal stream energy capture device; 2006. [38] Nicholls-Lee RF, Turnock SR, Boyd S. A method for analysing fluid structure interactions on a horizontal axis tidal turbine. In: The 9th European Wave and tidal energy Conference; 2011. Southampton, UK. [39] Lawson MJ, Li Y, Sale DC. Development and verification of a computational fluid dynamics model of a horizontal axis tidal current turbine. In: 30th
550
[40]
[41] [42] [43]
R.F. Nicholls-Lee et al. / Renewable Energy 50 (2013) 541e550 international conference on ocean, offshore and arctic engineering. Rotterdam, The Netherlands: ASME; 2011. Pashias C. Propeller tip vortex simulation using adaptive grid refinement based on flow feature identification. In: School of engineering sciences. Southampton: University of Southampton; 2005. Rycroft N, Turnock S. Three dimensional multiblock grid generation: fleximesh. Southampton: Univeristy of Southampton; 1997. Molland AF, Turnock S. Marine rudders and control surfaces. ButterworthHeinemann; 2007. 448. Barnsley MJ, Wellicome JF. Dynamic models of wind turbines - aerodynamic model development; 1993. Final report on contract jour 0110 for the
[44] [45] [46]
[47] [48]
commission of the european communities Directorate-General XII science, research and development. Inc., A., ANSYS 12.1 Help. 2009, Canonsburg, Pennsylvania. Commission, I.E. Wind turbine generator systems - part 1: safety requirements. Geneva: International Electrotechnical Commission; 1999. Molland AF, Bahaj AS, Chaplin JR, Batten WMJ. Measurements and predictions of forces, pressures and cavitation on 2-D sections suitable for marine current turbines. Proceedings of the Institute of Mechanical Engineers 2004;218(M):127e38. Gurit, SE84LV - Low temperature cure epoxy prepreg system, Gurit. p. 6pp. Burton T, Sharpe D, Jenkins N, Bossanyi E. Wind energy handbook. Chichester: John Wiley & Sons Ltd; 2004.
Contents lists available at SciVerse ScienceDirect
Renewable Energy journal homepage: www.elsevier.com/locate/renene
Application of bend-twist coupled blades for horizontal axis tidal turbines R.F. Nicholls-Lee*, S.R. Turnock, S.W. Boyd Fluid Structure Interactions Research Group, University of Southampton, Southampton, Hampshire, SO17 1BJ, UK
a r t i c l e i n f o
a b s t r a c t
Article history: Received 2 December 2011 Accepted 25 June 2012 Available online 16 August 2012
The blades of a horizontal axis tidal turbine are required to operate in a harsh subsea environment over a long life cycle with minimal need for maintenance. The concept of using passively adaptive, bend-twist coupled spars for horizontal axis tidal turbine blades has been identified as a potential method of improving energy capture. In this work a structural analysis is coupled with a fluid dynamic model to perform a full fluid-structure interaction analysis of a range of composite, bend-twist coupled blades. Blade element momentum theory is used to assess the presence of stall and corroborate the performance data attained from the fluid analysis. This paper discusses the individual analyses and the manner in which they are coupled. Several example problems were analysed using the design tool. The results compare well to the preliminary studies and indicate that a decrease of up to 12% in thrust and an increase of up to 5% in power capture could be achieved through the use of properly designed, bendtwist coupled blades. Ó 2012 Elsevier Ltd. All rights reserved.
Keywords: Adaptive composites Bend-twist coupling Composite blade design Fluid-structure interactions Renewable energy Tidal turbine
1. Introduction The oceans are a large, renewable, resource of untapped energy. There are many marine renewable energy sources however tidal energy has the advantage of being highly predictable and less susceptible to climate change than most [1e7]. Due to environmental concerns regarding potential devices it is thought that a breakthrough will occur in the area of kinetic energy devices; however, technology is at an early stage of development and further research into the field is required to advance the concepts, improve the feasibility of maintenance and make devices more efficient and economic. The blades of a Horizontal Axis Tidal Turbine (HATT) are the only means of extracting energy from the tidal flow and therefore the efficiency, and consequently annual energy capture, of a device could be increased by improving the blade design. With design for through-life performance and decommissioning becoming ever more prevalent, turbines are required to withstand the aggressive subsea environment for many years whilst being environmentally disposable come the end of their life. This has a large impact on fatigue loading, with an HATT typically experiencing in the order of 1 108 rotational cycles over a 20 year life span. Offshore maintenance is already a costly procedure and this, coupled with the fact that tidal arrays are typically located in remote regions at sea in areas of high flow velocity, promotes a need for minimal * Corresponding author. E-mail address: [email protected] (R.F. Nicholls-Lee). 0960-1481/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.renene.2012.06.043
maintenance over long periods. Optimisation of the blade design could increase the amount of energy capture, reduce structural loading and also minimise the need for maintenance. Composites offer several advantages over metals such as superior fatigue characteristics, high stiffness to weight ratio, ease of manufacture of structures with complex curvature and a reduction in inertial loading. The fundamental assumption is that the design methodology for a composite lifting surface, such as a turbine blade, should in general follow that for conventional isotropic materials that use sufficient stiffness to maintain a given optimum design shape. This conventional approach to lifting surface design relies on their being a fixed optimum shape. As the flow regime experienced is rarely steady, however, every shape will deflect under a given load condition and will dynamically respond to the transient fluid load fluctuations. Composites, however, offer the potential for hydroelastic tailoring; with the use of adaptive composites having been identified as a potential method for load reduction, increased efficiency and enhanced control of wind turbine blades [8e16] and propellers [17e20]. The concept of using adaptive, bend-twist coupled, composite blades in order to improve energy capture but also decrease design complexity has been considered. Preliminary analysis suggests that a 2.5% increase in annual energy capture and a 10% decrease in thrust loading could be expected through the use of a bend-twist coupled adaptive HATT blade [21e25]. There are many parameters that require optimisation in isotropic HATT blade design; not least diameter, section shape, thickness/chord ratio, pitch, skew, rake. Integrating adaptive composite materials into the blade also
542
R.F. Nicholls-Lee et al. / Renewable Energy 50 (2013) 541e550
requires that other variables are optimised such as the material properties, number of plies and the ply angle. Ultimately the blade analysis becomes complex and involves many time consuming iterations of Blade Element Momentum Theory (BEMT), Finite Element Analysis (FEA) and Computational Fluid Dynamics (CFD) so as to produce an optimal design. Following on from the preliminary investigations, the aim of this work is to gain a more detailed knowledge of the performance of a composite, bend-twist coupled HATT blade by performing a full fluid-structural interaction analysis. This includes the development of a design tool for adaptive HATT blades incorporating fluid and structural analyses and BEMT in a full Fluid Structural Interaction (FSI) analysis. 2. Adaptive composites An adaptive textile composite is a structure tailored to exhibit desirable elastic deformation behaviour not necessarily proportional to the imposed load. An example of such a structure would be a box beam so tailored that an imposed cantilever load results in twisting as well as bending, although no torsional load was imposed. Such a structure is said to exhibit bend-twist coupling. Composite blade elastic coupling can not only be tightly controlled but also varied over the span by appropriate selection of ply angles, thicknesses and spanwise layup. A 60e70% reduction in production costs was seen in the marine propeller industry after research into coupled composite propeller blades, along with smoother power take up, reduced blade vibration, reduced noise, and better fatigue performance [26]. Much research has been undertaken in this field in the wind industry [8e16,27e29]. Wind turbines carry loads primarily by twisting and bending, much like tidal turbines. The level of load reduction depends on the twist distributed along the blade length, which is controlled by the amount of bend-twist coupling. This, in turn, depends on the blade cross sectional geometry, the level of anisotropy in the structural material, and the material distribution [27]. For conventional laminated composites constructed of orthotropic layers, the level of anisotropy is determined by the fibre orientation with respect to the primary loading direction. Kooijman [28] found that the mirror layup, Fig. 1, is required for bend-twist coupling to be exhibited; with a 20 fibre angle orientation achieving a maximum level of coupling [30]. 3. Fluid structure interactions Fluid-structure interactions (FSI) occur when a fluid interacts with a solid structure, exerting pressure that causes deformation in the structure subsequently altering the flow of the fluid itself. If a situation involving structure flexure is to be analysed it is highly beneficial to couple both the fluid dynamics and the structural analysis programs to produce iterative solutions for complex problems. In operation an HATT is subjected to hydrostatic pressure, and as it is a lifting body with rotational motion it is subjected to further hydrodynamic forces e thrust, torque, cavitation etc. Due to the
Fig. 1. Layup for bend-twist coupling illustrating that the composite plies in the top and bottom surface of the central area of the blade must mirror each other for coupling to occur.
complex loading scenario experienced by a tidal turbine, knowledge of the hydroelastic behaviour of the blades, hub, nacelle, and also the support structure under this regime could lead to a more thorough understanding of structural constraints and how performance of the turbine could be improved. Whilst being a highly informative technique for the design analysis of a standard fixed blade device (assumed to be stiff), FSI is essential when considering adaptive composite blades as the pressure loading alters the shape more significantly when compared to standard blades. There are three methods of joint fluid structural modelling in the time domain which involve solving the governing equations in a coupled, uncoupled or integrated manner [31e35]. In this work, a loosely coupled, modular approach is used; the flow problem and structural problem are solved successively until the change is smaller than the convergence criterion. This method has the benefit that the two domains are discretised to better suit the problem; as the mesh for the fluid analysis will tend to require greater refinement in different areas of the geometry than that for the structural analysis, and vice versa. The program alternately manipulates both the surface pressure distribution yielded from the fluid dynamics program and the displacement data output from the structural analysis, and feeds them back into the next relevant stage of the process. This is illustrated schematically by the flow chart in Fig. 2. Similar methods to this have been shown to be successful by Turnock and Wright [31] for the analysis of rudder propeller interactions. 3.1. Fluid dynamics analysis The surface panel code used in this work was PALISUPAN (PArallel LIfting SUrface PANel) [36]. This code has previously been used to model the behaviour of a representative tidal turbine and good comparisons were obtained with published data, although problems were experienced in getting low twist sections to work due to the low pitch of the wake sheet [37,38]. Previous studies indicated that an optimum panel distribution can be achieved that maintains the accuracy of the result obtained with a finer distribution, but reduces the calculation time to around 15 min when using a standard desktop PC with one quad core Intel Xeon processor and 12 GB of RAM [37]. In comparison a full Reynolds Averaged Navier Stokes simulation of a similar tidal device carried out over 128 quad core Intel Nehalem processors on the 152 node Redrock computer cluster at the National Renewable Energy Laboratory would take up to 24 h depending on the mesh density [39]. Parameterisation and optimisation of surface panel codes is relatively simple, due to the low process times when implementing multiple runs e over 30 at a time e being highly feasible. Using a frozen wake model it is possible to reproduce the helical wake characteristic of tidal turbines. The number and distribution of panels in the wake is also very important for accurate modelling of device performance. Codes have been developed that generate panel distributions over complex shapes, such as a propeller or tidal turbine, and the associated wake panel arrangement. Such a code, Propgen [40] was used in this case to create the input blade geometry. Blade section data, section offsets, panel number and distribution for both blade and wake were input and the geometry is constructed from a set of section curves. The 2-D section is then mapped onto a cylindrical surface according to the specified variables using a transformation matrix. The appropriate pitch of the wake can be found through consideration of the axial and radial momentum changes in the flow due to the tidal turbine. Once the blade geometry was generated it is then exported to Adaptflexi [41], a program which enables geometry manipulation and definition of the fluid analysis domain and flow variables, in which a surface is lofted through the sections to generate the blade. Additional variables are now defined for the HATT scenario such as
R.F. Nicholls-Lee et al. / Renewable Energy 50 (2013) 541e550
543
Fig. 4. Panel distribution over blade and wake.
Fig. 2. Flow diagram of the fluid structure interaction simulation process. This illustrates the loosely coupled analysis with the fluid and structural solvers on the left of the blue line operating separately from each other, but being coupled by the Matlab based code on the right of the blue line. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
flow velocity, RPM, number of blades, water density, and panel discretisation; and subsequently a surface mesh created consisting of quadrilateral panels. Good mesh distribution definition, over both blade and wake, is important for accurate modelling of device performance [38,42]. Fig. 4 illustrates a single blade (colour) and wake panel distribution for analysis in the CFD code of a 20 m
Fig. 3. Validation study for an HATT in Palisupan.
diameter, three bladed HATT; there are in excess of 2800 panels in the blade of the turbine alone. A panel sensitivity analysis was carried out for a 20 m diameter, 3 bladed HATT [37]. The number of panels on the blade was varied until the total force and momentum differed by less than 2.5% as the number of panels was increased. This was found to require of the order of 2800 panels on the blade with approximately 40 panels in the chordwise direction and 75 in the spanwise direction. This ratio of panels is comparable with that typically found both for low aspect foils (rudders/keels) [42] but also the much lower aspect ratio propeller blades [40]. A further variable which is the Kutta condition for convergence was maintained at a value of 0.01. Fig. 3 presents a comparison of the performance of the BEMT code, Cwind [43], and the lifting surface panel code, Palisupan, alongside the original experimental data of Barnsley et al. [43] for a tip pitch setting of 2 . The BEMT section was represented by a single series of NACA 63-2xx Cl and Cd performance data. Following the work of Barnsley et al. [43] the influence of threedimensional behaviour is to prolong stall, and give much higher working CL before stall is initiated. This extra increment of lift was included through modification of the BEMT code. In comparing the potential flow based analysis from Palisupan with the BEMT excellent agreement is observed in the range of turbine operation 4 < TSR < 6. For this turbine, at TSR > 6, stall behaviour dominates which is captured by the BEMT but not by the potential flow analysis, which excludes stall [42]. In the full fluid-structural interaction analysis presented in this paper, the BEMT code is used to check the validity of the CFD results such that it can be determined when stall occurs, and the effect on turbine performance. This is discussed in more detail later. In the immediate wake a finer panel distribution can be observed when compared to the far wake downstream. This captures the flow more precisely leading to a more accurate solution for the fluid dynamic analysis. The turbine is rotationally symmetric and therefore the body coefficients are rotated and copied twice more to simulate the whole entity. The output from Adaptflexi is in the form of a .uns file which contains the flow data for the simulation along with a list of panels throughout the turbine and wake model and the corresponding corner points. The problem is then solved using Palisupan [36] and the resulting pressure loading across the surface of the blade was then calculated for input to the structural model, along with turbine performance data in the form of power coefficient, thrust and data information.
544
R.F. Nicholls-Lee et al. / Renewable Energy 50 (2013) 541e550
b 6.8 Max. Spar induced twist (o)
Max. Spar Bend (m)
a 1.16 1.14 1.12 1.10 1.08 1.06 1.04 1.02 1.00 0.98 0
50000
100000
150000
200000
250000
No. of Elements
6.7 6.6 6.5 6.4 6.3 6.2 6.1 6.0 5.9 5.8 0
50000
100000
150000
200000
250000
No. of Elements
Fig. 5. Solution convergence for increasing mesh density, a) Bend, b) Induced twist.
3.2. Structural analysis The blade was modelled in ANSYS 12.1 [44]. The model was meshed using the SHELL 281, 8 node, finite strain, shell element and solved using a large displacement static non-linear analysis. The SHELL 281 element is suitable for analysing thin to moderatelythick shell structures, i.e. the ratio between surface area and thickness of a single element must be greater than ten. It is wellsuited for linear, large rotation, and/or large strain nonlinear applications. Each layer of the laminate is defined by the material properties and thickness e modelled from the inner skin out to the outer roving. Each model consisted of in excess of 200,000 quadrilateral elements. This was a very fine mesh and was such that discretisation issues were not significant [38]. An example of a 20 m diameter HATT was solved structurally in ANSYS 12.1 for ten different mesh densities, ranging from under 25,000 elements to over 225,000 elements. The sensitivity of solution for both the maximum bend and induced twist over the entire length of the central spar at different mesh densities are illustrated in Fig. 5a and b respectively. Both solutions converged as the mesh density increased, with the optimum number of elements for both accuracy of solution and minimising computational time being 200,000. The computer used was a standard desktop PC with four Intel Xeon processors and 24 GB of RAM. The geometry created for the fluid analysis is imported into ANSYS 12.1, and a central spar fitted through the blade, Fig. 6. This ensured that nodes existed at the corner points of each panel of the mesh used for the fluid analysis. The spar was fitted such that it supported the blade at one third of the local chord distance from the leading edge; this is where the centre of effort of the blade typically acts. The resulting geometry was then meshed and the pressure loads from the fluid dynamic analysis applied. This was achieved directly, without requiring interpolation, by virtue of the FEA geometry having the same areas mapped out as the surface panel fluids model; the pressure at the centre point of each panel is output from Palisupan and applied to the corresponding panel in the structural model. The blade structural mesh was created in the smallest areas first to ensure that it was adequately dense in these regions, resulting in a higher quality of mesh overall. The coupled spar was theoretically “laid up” in the manner shown in Fig. 7. The structural deformation problem was solved using a static non-linear analysis such that large deflection effects were taken into account [23,24]. The maximum stress seen in the blade is monitored with the criteria that it must not exceed a third of the yield stress of the beam material; thereby incorporating a safety factor of three in to the design. Experimental analysis on a bendtwist coupled, double box beam of the same configuration as Fig. 7 has been undertaken in order to corroborate the numerical results [32]. In this work the partial safety factor method, commonly used in the wind turbine design industry [45], has not yet been integrated
and therefore a safety factor of three was chosen in an effort to account for the through-life stresses present on the turbine. If this value is exceeded the number of plies in the mid-layer of the bendtwist coupled beam is altered, and the Finite Element Analysis (FEA) rerun for the same pressure loading until an acceptable stress level is reached. The deflections are output at the nodes corresponding to the panel corner points in the fluid analysis geometry. The deflection of the blade is then remapped into a new .uns input file for the fluid dynamic analysis and the problem rerun. This simulation is continued in an iterative manner until the deflection of the blade between successive runs is less than 10 mm; i.e. it has converged. The displacement value of 10 mm was chosen as this was small (0.05%) when compared to the turbine diameter of 20 m. 3.3. Coupling design Tools Coupling of both the structural and fluid dynamic analyses has been carried out in Matlab 2009b. Fig. 8 shows a flow chart of the process. Initially tables of data for the range of design parameters under investigation are generated; these variables include design RPM, diameter, flow velocity, section shape and number of blades. The relevant data for the first run is then extracted from these tables and the Propoptions and Propeller input files for Propgen created. Propgen is then run to create the blade, hub, and wake geometry and a script is generated to pass the geometry through Adaptflexi. The output required from the fluid analysis is governed through this script; pressure information, panel discretisation, and overall turbine loadings. Adaptflexi is then used to generate the .uns file and .cmd file and the case run in the fluid analysis code. Structural variables are then introduced in the form of ply angle, number of plies and material data. The intention being to create a design which maximises induced twist whilst minimising bend. The coupling code then takes the geometry data from the fluid dynamic analysis and generates a blade within ANSYS that has a coupled box beam as the main central spar. The required lay-up information and the pressure loading are then applied to the
Fig. 6. Geometry of blade showing the central spar (a) and surrounded by the blade skin (b).
R.F. Nicholls-Lee et al. / Renewable Energy 50 (2013) 541e550
Fig. 7. Lay-up of central, bend-twist coupled, spar illustrating the extruded aluminium mould at the centre which was removed post-cure.
corresponding areas and the problem solved. Once a solution is achieved the stress check detailed earlier is then performed. The deformed positions of the loaded blade are then calculated through knowledge of the displacements of each node; and bend and induced twist determined.
545
The nodal displacements are then transferred to the .uns file for the relevant panel corner points. The panel list in the .uns file is divided into trailing edge body panels, main body panels, wake panels attached to the trailing edge, mid wake panels and far wake panels. The displacements of each panel on the trailing edge of the main body are known from the FEA and are evaluated as axial, radial and angular displacements in cylindrical co-ordinates and then applied to the body keypoints in the .uns file. These displacements are then translated back through the wake panels to give a deformed wake to match the warped blade. The problem is then run through the fluid analysis program again to determine a new surface pressure loading for the deflected blade. This loading is applied to the deflected blade in ANSYS and the structural analysis carried out. These stages are repeated until convergence. In order to perform a final check on the data obtained, once convergence has occurred the input files for the BEMT code are created from the main blade parameters and performance
Fig. 8. Flow chart detailing the coupling process.
546
R.F. Nicholls-Lee et al. / Renewable Energy 50 (2013) 541e550
Table 1 Example case blade design data.
Table 2 Material mechanical properties [47].
r/R
Chord (m)
P/D
t/c
Coupled spar SE84LV HEC 300/400
Blade skins SE84LV RC200T
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
2.304 2.033 1.798 1.576 1.367 1.170 0.986 0.814 0.655
0.3365 0.2263 0.1593 0.1166 0.0866 0.0678 0.0556 0.0458 0.0358
0.0858 0.0753 0.0653 0.0616 0.0581 0.0546 0.0511 0.0476 0.0441
Property
Value
Property
Value
Fibre weight Cured ply thickness Ex EY Ez
300 g/m2 0.281 mm 129.2 GPa 8.76 GPa 8.76 GPa 0.335 0.0172 0.0172 5.76 GPa 5.76 GPa 5.76 GPa
Fibre weight Cured ply thickness Ex EY Ez
195 g/m2 0.214 mm 65.2 GPa 65.2 GPa 8.76 GPa 0.05 0.05 0.05 5.76 GPa 5.76 GPa 5.76 GPa
characteristics calculated for the device. This data is compared to that gained through the final iteration of the fluid analysis to ensure that the thrust and torque estimates are similar and the turbine has not stalled. In this manner a composite, bend-twist coupled blade can be designed for use on an HATT. 3.4. Example case In order to illustrate the FSI blade design method, an example design problem was considered [38]. The design spring peak current is 2.5 m/s, with both the ebb and flood stages of the tide assumed to have similar time history, i.e. they have been
nxY nYZ nxZ
GxY GYZ GxZ
nxY nYZ nxZ
GxY GYZ GxZ
approximated as a double sinusoid. The turbine has a diameter of 20 m, a hub-diameter ratio of 0.2 and three bend-twist coupled blades. The section shape to be used for the blade is the NACA 63815 which was shown previously to perform well [46]. The blade is of the form detailed in Table 1. A three-dimensional representation of the example turbine is illustrated in Fig. 9. The bend-twist coupled spar is constructed of SE84LV UD carbon pre-preg, and the blade skins made from SE84LV RC200T twill weave carbon pre-preg (Table 2) both from Gurit [47]. The blade central spar initially has 20 plies in the mid-layer at 20 , 30 plies in the inner skin at 45 and 5 plies in the outer roving at 90 e each ply has a thickness of 0.000281 m. The number of plies in the mid-layer are then increased in steps of five plies until the minimum stress criteria is achieved (sy/3). The maximum displacement is at the tip of the blade, with a value of 1.48 m. For this particular turbine, the number of mid layer plies was increased to 35 in order to meet the maximum stress criterion, and the solution converged in 16 iterations of the FSI loop. There is good correlation between the results from the final iteration of the fluid dynamic study and the BEMT analysis, with less than 1% difference in thrust coefficient and 2% difference in power coefficient [38]. The main properties of the turbine, and performance results, are detailed in Table 3. 3.5. Discussion of results The design tool was used to analyse a range of turbines with bend-twist coupled blades in which the number of blades, flow velocity and diameter were varied whilst rotational speed was kept constant at 12 RPM. The blade design for each turbine e section, chord, P/D etc. e is the same as detailed in Table 1. Table 4 details the results of this study indicating bend, induced twist, and CT and Cpow from both the fluid dynamic and BEMT analyses. It is apparent that the main limitation of BEMT, the assumed relation between disc and far wake induction, has a strong effect at Table 3 Case study turbine principal particulars and performance data. Current velocity Diameter Number of blades Revolutionary speed Mid-layer ply angle Number of mid-layer plies Thrust Torque Bend Induced twist Maximum stress Thrust coefficient Power coefficient Absorbed power
Fig. 9. 3-D representation of the example HATT.
2.5 20 3 12 20 40 724.5 792.3 1.48 8.6 28.1 0.72 0.40 1248.9
(m/s) (m) (blades) (RPM) (o) (plies) (kN) (kNm) (m) (o) (GN/m2) e e (kW)
Table 4 Variation of performance parameters for a selected set of adaptive tidal turbine designs over a range of blade numbers, diameters and tip speed ratios. It is assumed that all designs are operating at mid-depth of a channel of depth 2D. The blade tip cavitation number for each device does not fall below 1.0 and therefore the effect on performance can be neglected [42]. Case no.
2 2 2 2 2 2 2 2 2 2 2 2 3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4
V
f
(m/s)
1.5 1.5 1.5 1.5 2 2 2 2 2.5 2.5 2.5 2.5 1.5 1.5 1.5 15 2 2 2 2 2.5 25 2.5 2.5 1.5 1.5 1.5 1.5 2 2 2a 2 2.5 2.5 2.5 25
D (m)
10 15 20 25 10 15 20 25 10 15 20 25 10 15 20 25 10 15 20 25 10 15 20 25 10 15 20 25 10 15 20 25 10 15 20 25
TSR
4 8
4
6 4
6 2
4.189 6.283 8.378 10.472 3.142 4.712 6.283 7.854 2.513 3.770 5.027 6.283 189 6.283 378 10.472 3.142 712 6.283 7.854 2.513 3.770 5.027 283 189 6.283 8.378 10.472 3.142 4.712 283 7.854 513 3.770 5.027 6.283
Bend (m)
0.08 0.639 2.513 4.823 0.061 0.487 1.912 3.589 0.05 0.396 1.548 2.902 0.071 0.571 2.25 4.234 0.054 0.435 1.709 3.212 0.044 0.353 1.382 2.594 0.064 0.515 2.033 3.832 0.049 0.392 1.541 2.901 0.04 0.317 1.245 2.34
Ind twist ( )
0.94 4.98 14.41 21.53 0.72 3.81 11 06 16.36 0.59 3.09 9.00 13.36 0.84 4.46 12.95 19.10 0.64 3.40 9.91 14.72 0.52 2.76 8.05 11.98 0.75 4.02 11.74 17.39 0.57 3.06 8.95 13.35 0.47 2.48 7.26 10.84
Fixed
Adaptive
CT CFD
Cpow CFD
CT BEMT
Cpow BEMT
CT CFD
Cpow CFD
CT BEMT
Cpow BEMT
0.194 0.286 0.777 1.132 0.217 0.971 1.073 0.860 0.453 0.766 0.670 0.535 0.140 0.231 0.269 0.302 0.135 0.378 1.041 1.186 0.172 0.834 1.030 0.805 0.091 0.170 0.231 0.251 0.124 0.174 0.256 0.670 0.119 0.191 0.973 1.102
0.046 0.107 0.154 0.379 0.077 0.314 0.438 0.404 0.139 0.378 0.336 0.279 0.033 0.044 0.052 0.051 0.041 0.112 0.221 0.399 0.038 0.197 0.424 0.375 0.003 0.040 0.026 0.044 0.025 0.024 0.047 0.133 0.026 0.032 0.226 0.420
0 161 0.200 0.543 0.821 0.179 0.809 0.751 0.602 0.374 0.638 0.468 0.374 0.116 0.161 0.188 0.219 0.111 0.315 0.728 0.829 0.142 0.695 0.720 0.563 0.705 0.119 0.161 0.182 0.102 0.145 0.179 0.468 0.098 0.159 0.680 0.770
0.023 0.057 0.140 0.355 0.043 0.283 0.402 0.381 0.088 0.375 0.320 0.274 0.017 0.023 0.048 0.048 0.023 0.101 0.203 0 377 0.024 0.195 0.404 0.367 0.002 0.021 0.023 0.041 0.014 0.022 0.043 0.125 0.016 0.032 0.215 0.412
0.172 0.256 0.693 1.018 0.196 0.877 0.973 0.779 0.413 0.697 0.614 0.490 0.125 0.207 0.240 0.270 0.122 0.344 0.942 1.074 0.158 0.765 0.944 0.739 0.081 0.153 0.208 0.225 0.113 0.158 0.233 0.609 0.109 0.175 0.885 1.015
0 0 0 0 0
0 0 0 0 0
0.024 0.58 0.146 0.372 0.044 0.291 0.418 0.400 0.090 0.386 0.333 0.287 0.017 0.024 0.050 0.050 0.023 0.104 0.211 0.395 0.024 0.201 0.420 0.386 0.002 0.022 0.024 0.043 0.014 0.023 0.045 0.132 0.017 0.033 0.224 0.433
0 0 0 0
047 110 160 398 078 0.322 0.451 0.418 0.142 385 346 293 034 0.045 0.054 0.054 0.041 0.115 0.228 0.413 0.039 0.200 0.437 0.394 0.003 0.041 0.027 0.046 0.025 0.025 0.049 0.137 0.026 0.033 0.233 0.442
147 178 495 738 162 0.728 0.678 0.544 0.338 0.570 0.420 0.339 0.106 0.144 0.172 0.196 0.100 0.284 0.657 0.750 0.128 0.622 0.645 0.510 0.069 0.106 0.147 0.163 0.092 0.130 0.162 0.424 0.088 0.142 0.610 0.699
Change in CT (%)
Change in Cpow (%)
-12.23 -11.86 -12.11 -11.23 -10.62 -10.74 -10.25 -10.50 -9.65 -9.77 -9.05 -9.17 -11.98 -11.61 -12.23 -11.48 -10.25 -10.01 -10.50 -10.38 -9.29 -9.05 -9.17 -8.81 -11.23 -11.36 -11.11 -11.36 -9.53 -9.77 10.01 -9.89 -9.17 -8.81 -9.89 -8.58
2.15 2.82 3.66 4.58 1.48 2.44 2.92 3.29 1.86 1.62 2.94 4.85 2.18 2.87 3.73 4.66 1.54 2.49 2.97 3.35 1.93 1.66 2.98 4.91 2.24 2.88 3.74 4.67 1.55 2.51 3.00 3.36 1.94 1.67 2.99 4.92
R.F. Nicholls-Lee et al. / Renewable Energy 50 (2013) 541e550
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 I8 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
No. of blades
547
548
R.F. Nicholls-Lee et al. / Renewable Energy 50 (2013) 541e550
high induction factors and thrust coefficients; indicated in Table 4 by the lower prediction of CT from the BEMT. The fluid analysis predicted slightly higher values of Cpow than the BEMT. This was due to the fact that the iterative boundary layer prediction method was not included in the Palisupan calculations. In practice a performance loss would be present due to the onset of stall; however, the cases at lower flow velocities in which stall was not occurring show better correlation between the BEMT and fluid analysis results. The solidity of a blade is defined as total blade area divided by the swept area e this can be altered by changing the number of blades or altering the blade chord [48]. It can be seen from Table 4 that Cpow changes less over a wide TSR for the two blade turbines (low solidity), cases 1e12. The higher solidity turbines tend to perform better at lower TSRs, but are more sensitive to TSR changes. In general turbines require a minimum of three blades to be selfstarting at low flow velocities, with two bladed devices requiring a greater flow velocity to start up thus negating the broader capture area under the Cpow curve that the low solidity creates. CT can be observed to increase with an increase in solidity, although this is also dependent on the turbine diameter and resulting TSR. At higher TSR the lower solidity turbines experience greater thrust force. Changing the diameter of the turbine alters the swept area of the device; an increase in diameter increases the amount of power available for extraction due to the increased rotor area. CT and Cpow generally increase with an increase in diameter, but effectively it is the absorbed power of the device that is increasing which is reflected in the non-dimensional coefficients. The four bladed devices follow this trend, however the two and three bladed devices can be observed to have a reduction in CT and Cpow, and hence absorbed power, at the highest flow velocities. The peak performance of the two bladed devices analysed occurred at a flow velocity of 2 m/s with a diameter of 20 m (case 7). The 25 m diameter rotor then underperformed at this, and greater flow velocities for the lower solidity devices. The final two columns of Table 4 show the percentage change of both CT and Cpow between the initial fixed blade and the adaptive blades after the final run of the design tool. There is a general trend for a decrease in CT, and an increase in Cpow with the use of the adaptive blades. This is in the same order as the preliminary work [21]; however, there are some interesting trends appearing with the more holistic analysis using the design tool. Generally the reduction in CT is greater at lower flow velocities (cases 1e4, 13e16 and 25e28) and for lower solidity turbines. Conversely the improvement in Cpow increases slightly with increasing flow velocity. Previous work showed that while the adaptive blades did not increase the maximum Cpow of the turbine greatly, the performance curve broadened indicating the device was more efficient at higher TSR, and thereby capturing more energy over the whole tidal cycle. This, more detailed analysis, confirms this trend; albeit with a greater increase in Cpow at higher TSR. The larger diameter devices exhibit a greater improvement in Cpow, in the region of 4%. This is attributed to the increased amount of coupling present in the blades due to the additional length. Comparing the bend and induced twist of the 10 m and 25 m, three bladed turbines at a flow velocity of 2.5 m/s (cases 21 and 24 respectively) it can be seen that the bend increases from 0.044 m to 2.594 m and the induced twist increases from 0.172 to 11.98 . This is expected as the majority of torque, and hence thrust, to turn the turbine is developed from the outer third of the rotor diameter e the tip of the blade. The larger diameter turbine will therefore not only experience a greater thrust force due to the size increase, but also a significantly bigger bending moment thereby increasing the effect of
the bend-twist coupling inherent in the blade. Due to the manner in which the analysis has been carried out the number of plies in the mid-layer of each of these blades does differ in order to meet the maximum stress criterion which also contributes to the change in degree of coupling; the 10 m device had 35 plies and the 25 m device 45 plies. In this analysis the number of mid-layer plies were optimised for each individual condition (vf, D, number of blades), whereas in reality for a specific number of blades and turbine diameter the composite layup will be the same as the tidal velocity changes. Fig. 10 illustrates three sets of power data for the three bladed device with zero degrees pitch at 12 RPM: the initial fixed bladed device, the preliminary adaptively coupled device, and the coupled device analysed through the design tool. Each device has the same blade chord, section and pre-twist distribution. The results gained through the design tool show good agreement with those of the preliminary analysis. The maximum Cpow from the device analysed through the design tool is higher and occurs at a lower TSR than that for the preliminary device. Fig. 11 illustrates the relationship between CT and TSR for the same three cases. The results attained through the use of the design tool agree with the preliminary adaptive blade results for higher TSRs, but at lower TSR an increased drop in thrust coefficient can be observed. This decrease in thrust loading reduces structural fatigue, potentially increasing turbine life-span whilst reducing the frequency of maintenance required. This has positive implications for the through-life cost of the device, with maintenance being costly in the aggressive subsea environment; a reduction in the level of maintenance required should correspond to a reduction in through-life cost of the turbine. The annual energy capture of HATT can be approximated using the method detailed in NichollseLee et al [22]. Applying this method to the analysis of the example 20 m diameter, 3 bladed example turbine in Section 3.4, the annual energy capture of the adaptive device is 2.5% greater than that of a similar device with fixed blades. The flow regime into a tidal device at any one time is not constant, with variations in the vertical water column apparent alongside effects from turbines in arrays, and thus the blades experience changing inflow depending upon where they positioned in the area of the rotor disk. This transience should be reflected in both the fluids and structural analyses in the design tool and is an area of required further work.
Fig. 10. Relationship between TSR and Cpow for the fixed blade turbine, the preliminary analysis coupled turbine and the coupled turbine analysed using the design tool.
R.F. Nicholls-Lee et al. / Renewable Energy 50 (2013) 541e550
Fig. 11. Relationship between TSR and CT for the fixed blade turbine, the preliminary analysis coupled turbine and the coupled turbine analysed using the design tool.
4. Conclusions Adaptive composite technology has been in use in the aerospace industry for some time, and is now also being integrated into blades for wind turbines; however there are limited applications in the marine field. Preliminary analysis of a new conceptual blade with an integrated bend-twist coupled composite spar showed that such a blade has the potential to increase annual energy capture by around 2.5%. A decrease in thrust loading of up to 10% was also observed, thereby increasing the fatigue life of the turbine and associated structure. A tool for the design of passively adaptive, composite HATT blades has been developed. A surface panel code is used to predict the pressure loading on an initial HATT blade. This data is then used in FEA to assess the structural response of a composite beam. The deformed shape is then fed back into the fluid analysis code and a new pressure distribution calculated. Likewise these forces are imported back into the structural solver and so forth in an iterative manner until a suitable level of convergence of blade deformation is reached. The performance data is then checked using the BEMT code and the design of an adaptive blade realised. Several example problems were analysed using the design tool. The results compare well to the preliminary studies, and indicate that aimproved decrease of up to 12% in CT and an increase of up to 5% in Cpow could be achieved through the use of properly designed bend-twist coupled blades. This implies that structural fatigue loading on the turbine is decreased; hence potentially decreasing through-life maintenance costs and increasing the life-span of the device. Acknowledgements This work was part funded through the EPSRC Project - EP/ I009876/1: Tailored Composites for Tuned Deformation Response to Unsteady Fluid Loading. References [1] Kirke B. Developments in ducted water current turbines; 2005. 12. [2] Charlier RH. A "Sleeper" awakes: tidal current power. Renewable and Sustainable Energy Reviews 2003;7:515e29. [3] Callaghan J, Boud R. Future marine energy. Carbon Trust; 2006. [4] ENTEC and BWEA. Marine renewable energy - state of the industry report. In: UK R, editor; 2009. [5] VanZwieten J, Driscoll FR, Leonessa A, Deane G. Design of a prototype ocean current turbine - part I: mathematical modeling and dynamics simulation. Ocean Engineering 2006. [6] Harrison GP, Wallace AR. Climate sensitivity of marine energy. Journal of Renewable Energy 2005;30(12):1801e17.
549
[7] Pearce N. Worldwide tidal current energy developments and opportunities for Canada’s Pacific coast. International Journal of Green Energy 2005;2:365e86. [8] Karaolis N, Mussgrove P, Jeronimidis G. Active and passive aeroelastic power control using asymmetric fibre reinforced laminates for wind turbine blades. In: Proceedings of the 10th British wind energy conference; 1988. London, U.K. [9] Lobitz DW, Laino DJ. Load mitigation with twist coupled HAWT blades. In: 37th aerospace sciences meeting and exhibit and 18th ASME wind energy symposium. Reno, Nevada, USA: American Institute of Aeronautics and Astronautics; 1999. [10] Lobitz DW, Veers PS. Aeroelastic behavior of twist-coupled HAWT blades. In: 36th aerospace sciences meeting and exhibit and 17th asme wind energy symposium. Reno, Nevada, USA: American Institute of Aeronautics and Astronautics; 1998. [11] Veers P, Bir G, Lobitz D. Aeroelastic tailoring in wind turbine blade applications. In: Windpower ’98, American wind energy association meeting and exhibition. Bakersfield, California, USA: Sandia National Laboratories; 1998. [12] Eisler GR, Veers PS. Parameter optimisation applied to use of adaptive blades on a variable speed wind turbine. Sandia National Laboratories; 1998. [13] Griffin DA. Evaluation of design concepts for adaptive wind turbine blades. Sandia Report. Sandia National Laboratories; 2002. [14] Maheri A. A new approach in simulation of wind turbines utilising bend-twist adaptive blades. Bristol: University of the West of England; 2006. [15] Kooijman H. Bending-torsion coupling of a wind turbine rotor blade; 1996. [16] Lobitz DW, Veers PS, Eisler GR, Laino DJ, Migliore PG, Bir G. The use of twistcoupled blades to enhance the performance of horizontal axis wind turbines. Sandia Report. Sandia National Laboratories; 2001. [17] Young YL. Fluid-structure interaction analysis of flexible composite marine propellers. Journal of Fluids and Structures 2008;24:799e818. [18] Young YL. Dynamic hydroelastic scaling of self-adaptive composite marine rotors. Composite Structures 2010;92:97e106. [19] Liu Z, Young Y. Utilization of bend-twist coupling for performance enchancement of composite marine propellers. Journal of Fluids and Structures 2009;25:1102e16. [20] Khan AM, Adams DO, Dayal V, Vogel JM. Effects of bend-twist coupling on composite propeller performance. Mechanics of Composite Materials and Structures 2000;7:383e401. [21] Nicholls-Lee R, Turnock S. Enhancing performance of a horizontal axis tidal turbine using adaptive blades. In: Oceans ’07. Aberdeen, Scotland: IEEE; 2007. [22] Nicholls-Lee R, Turnock S, Boyd S. Performance prediction of a free stream tidal turbine with composite bend-twist coupled blades. In: 2nd international conference on ocean energy (ICOE 2008); 2008. Brest, France. [23] Nicholls-Lee R, Boyd S, Turnock S. Development of high performance composite bend-twist coupled blades for a horizontal axis tidal turbine. In: 17th international conference on composite materials (ICCM-17); 2009. Edinburgh, Scotland. [24] Nicholls-Lee R, Boyd S, Turnock S. Obtaining experimental data suitable for validation of a numerical design tool for an adaptive composite free stream tidal turbine blade. In: The 11th international symposium on practical design of ships and other floating structures (PRADS 2010); 2010. Rio de Janeiro, Brazil. [25] Nicholls-Lee RF, Turnock SR, Boyd S. Tidal turbine blade selection for optimal performance in an array. In: 30th international conference on ocean, offshore and arctic engineering; 2011. Rotterdam, The Netherlands. [26] Young YL, Michael TJ, Seaver M, Trickey MT. Numerical and experimental investigations of composite marine propellers. In: 26th symposium on naval hydrodynamics; 2006. Rome, Itlay. [27] Locke J, Hidalgo IC. The implementation of braided composite materials in the design of a bend-twist coupled blade. Sandia Report. Sandia National Laboratories; 2002. [28] Kooijman H. Bending-torsion coupling of a wind turbine rotor; 1996. [29] Lobitz DW, Veers PS, Migliore PG. Enhanced performance of HAWTs using adaptive blades. In: Wind energy ’96, ASME wind energy symposium; 1996. Houston, U.S.A. [30] Ong CH, Tsai SW. Design, manufacture and testing of a bend-twist D-Spar. In: S.N. Laboratories, editor. Stanford, California: Department of Aeronautics and Astronautics, Stanford University; 1999. [31] Turnock SR, Wright AW. Directly coupled fluid structural model of a ship rudder behind a propeller. Marine Structures 2000;13(1):53e72. [32] Nicholls-Lee RF. Adaptive composite blades for horizontal axis tidal turbines. In: School of Engineering Sciences. Southampton: University of Southampton; 2011. p. 218. [33] Guruswamy GP. Unsteady aerodynamic and aeroelastic calculations for wings using Euler equations. AIAA Journal 1990;28(3):461e9. [34] Guruswamy, G.P. and Byun, C., Fluid-Structural interactions using NavierStokes flow equation coupled with shell finite element structures, AIAA, Editor. 1993, AIAA. [35] Guruswamy GP, Byun C. Wing-body aeroelasticity on parallel computers. Journal of Aircraft 1996;33(2):421e8. [36] Turnock SR. Palisupan user guide; 2000. [37] Turnock SR, Nicholls-Lee RF. Design of three bladed tidal turbine blades for bidirectional fixed pitch or azimuthing variable pitch operation. In: Logþ1/ Alstom/WUMTIA, editor. Economic viability of a simple tidal stream energy capture device; 2006. [38] Nicholls-Lee RF, Turnock SR, Boyd S. A method for analysing fluid structure interactions on a horizontal axis tidal turbine. In: The 9th European Wave and tidal energy Conference; 2011. Southampton, UK. [39] Lawson MJ, Li Y, Sale DC. Development and verification of a computational fluid dynamics model of a horizontal axis tidal current turbine. In: 30th
550
[40]
[41] [42] [43]
R.F. Nicholls-Lee et al. / Renewable Energy 50 (2013) 541e550 international conference on ocean, offshore and arctic engineering. Rotterdam, The Netherlands: ASME; 2011. Pashias C. Propeller tip vortex simulation using adaptive grid refinement based on flow feature identification. In: School of engineering sciences. Southampton: University of Southampton; 2005. Rycroft N, Turnock S. Three dimensional multiblock grid generation: fleximesh. Southampton: Univeristy of Southampton; 1997. Molland AF, Turnock S. Marine rudders and control surfaces. ButterworthHeinemann; 2007. 448. Barnsley MJ, Wellicome JF. Dynamic models of wind turbines - aerodynamic model development; 1993. Final report on contract jour 0110 for the
[44] [45] [46]
[47] [48]
commission of the european communities Directorate-General XII science, research and development. Inc., A., ANSYS 12.1 Help. 2009, Canonsburg, Pennsylvania. Commission, I.E. Wind turbine generator systems - part 1: safety requirements. Geneva: International Electrotechnical Commission; 1999. Molland AF, Bahaj AS, Chaplin JR, Batten WMJ. Measurements and predictions of forces, pressures and cavitation on 2-D sections suitable for marine current turbines. Proceedings of the Institute of Mechanical Engineers 2004;218(M):127e38. Gurit, SE84LV - Low temperature cure epoxy prepreg system, Gurit. p. 6pp. Burton T, Sharpe D, Jenkins N, Bossanyi E. Wind energy handbook. Chichester: John Wiley & Sons Ltd; 2004.