influence analysis of blade chord length on the performance of a four bladed wollongong wind turbine

Influence analysis of blade chord length on theperformance of a four-bladed Wollongong wind turbine ZhaoyongMao,aWenlongTian,and ShaokunYan School of Marine Science and Technology, North

Wollongong wind turbine

Zhaoyong Mao, Wenlong Tian, and Shaokun Yan

Citation: Journal of Renewable and Sustainable Energy 8, 023303 (2016); doi: 10.1063/1.4943093 View online: http://dx.doi.org/10.1063/1.4943093

View Table of Contents: http://aip.scitation.org/toc/rse/8/2

Published by the American Institute of Physics

Influence analysis of blade chord length on the

performance of a four-bladed Wollongong wind turbine ZhaoyongMao,a)WenlongTian,and ShaokunYan

School of Marine Science and Technology, Northwestern Polytechnical University,

710072 Xi’an, China

(Received 26 September 2015; accepted 19 February 2016; published online 3 March 2016)

The Wollongong wind turbine is a new kind of vertical axis wind turbine (VAWT) with its blades rotated by only 180 for each full revolution of the main rotor A computational study on the effect of blade chord length on the turbine output performance of a four-bladed Wollongong turbine has been conducted using the commercial computational fluid dynamics (CFD) code ANSYS 13.0 A validation study was performed using a Savonius turbine and good agreement was obtained with experimental data Both rotating and steady CFD simulations were conducted

to investigate the performance of the VAWT Rotating two-dimensional CFD simu-lations demonstrated that a turbine with a blade length of 550 mm has the highest power curve with a maximum averaged power coefficient of 0.3639, which is almost twice as high as that of a non-modified Savonius turbine Steady two-dimensional CFD simulations indicated that the Wollongong turbine has a good self-starting capability with an averaged static torque coefficient of 1.09, which is about six times as high as that of a Savonius turbine.V C 2016 Author(s) All article content, except where otherwise noted, is licensed under a Creative Commons Attribution 3.0 Unported License [http://dx.doi.org/10.1063/1.4943093]

I INTRODUCTION

In recent years, people have shown increasing attention on renewable energy with increas-ing environmental pollution, risincreas-ing energy demand, and depletincreas-ing fossil fuel resources Wind energy utilization has become a research hotspot because it is economical and experimental Wind turbines can be classified into two groups: horizontal axis wind turbine (HAWT) and ver-tical axis wind turbine (VAWT), depending on the relative direction between the turbine rota-tional axis and the wind direction VAWTs rotate around an axis perpendicular to the wind direction VAWTs are less efficient than HAWTs, but they operate effectively in the presence

of highly unstable, turbulent wind flow patterns1and are more suitable for power generation in places with a complex terrain, such as remote rural areas where erratic wind flow patterns are quite common Moreover, VAWTs can operate regardless of the flow direction2and are suitable for small scale, distributed power generation

Previous research on VAWT mainly focused on two types of turbines,3 Darrieus turbine and Savonius turbine The Darrieus turbine is one of the best-known VAWTs, which has three

or four straight airfoils to create lift and is the most efficient type of VAWT Considerable effort has been made to model the dynamic forces on the Darrieus turbine.4,5The Darrieus tur-bine has also been studied extensively using CFD and experimental methods to optimize the performance.6,7 Daroczyet al studied the effects of turbulence models in the simulation of H-Darrieus rotors.8Variable pitch control mechanism,9channelling device,10twisted blades,11and new airfoil shapes12 have been adopted to further improve starting torque and efficiency, and

a)

Author to whom correspondence should be addressed Electronic mail: maozhaoyong@nwpu.edu.cn

reduce shaking of the Darrieus turbine Recently, Chowdhury et al.13 studied the performance

of the Darrieus wind turbine in upright and tilted configurations

The Savonius turbine generates torque through the combined effects of drag and inside forces and typically has two or three bucket-shaped blades.14 The Savonius turbine has been studied experimentally and numerically to examine the effects of various design parameters such as the rotor aspect ratio, the overlap, the number of buckets, the rotor endplates, and the influence of bucket stacking.14–16In addition, many researchers have worked to improve the ef-ficiency and the starting torque characteristics of the Savonius rotor Some of these include add-ing guide vanes or deflector plates in front of the rotor preventadd-ing the negative torque opposite the rotor rotation.17,18As for the novel blades of the Savonius turbine, researchers have studied numerically the performance of the turbine with arc-type blades19 and blades developed from Myring equations.20 A numerical study was also carried out to investigate the interaction between multi-turbines.21

Cooper and Kennedy developed a novel VAWT with actively pitched blades, called a Wollongong turbine.22Fig.1(a) shows the 3-D schematic of a four-bladed Wollongong turbine

FIG 1 Schematic of a four-bladed Wollongong turbine: (a) 3-D view and (b) plan view.

The turbine has four blades with their axes held parallel to the main rotor axis The blades were simple flat plates, the pitch of which was controlled by a transmission chain of bevel gears so that the blades rotated about their mid-chord axis by only 180 for each full revolution

of the main rotor A detailed description of mechanical transmission principle can be found in Ref 22 The Wollongong turbine has good self-starting performance and generates relatively high torque However, few accurate theoretical models or CFD analysis can be found in the existing literature

This research aims to explore the output performance and self-starting capability of a four-bladed Wollongong turbine, and to investigate the effect of blade chord length on the perform-ance of the turbine The analysis has been performed with a two-dimensional computational fluid dynamics (CFD) method A sliding mesh model, which uses a time averaged solution to determine the turbine performance, and has been proven to give relatively accurate results for a two-dimensional rotating rotor, was used in the rotating simulations to investigate the dynamic torque and power characteristics of the turbine Steady simulations have also been conducted to predict the self-starting capability of the turbine

II NUMERICAL METHOD

The simulations were performed using the commercial code Fluent 13.0,23 which is based

on the finite volume method and has been widely used in wind turbine simulations The compu-tational domain discretization was generated with the mesh tool in ANSYS Workbench A slid-ing mesh method was applied to perform the unsteady simulation

A Simplified physical model

Since the straight blades have the same cross section, the blade span effect can be ignored and two-dimensional simulations are chosen In two-dimensional simulations, the blade has a unit span of 1 m Fig.2illustrates the situation where the turbine rotates with an constant angu-lar velocity, x, in a flow with an inlet velocity,U The main parameters of the simplified physi-cal model are shown in TableI

B Governing equations

The governing equations are given by the incompressible form of the Navier-Stokes equa-tions, including the continuity equation and momentum equation, as shown below:23

@q

@

@tðq~vÞ þ r q~ð v~vÞ þ q 2x*

 ~vþ x*

 x*

 ~v

¼ rq þ rsþ F*; (2)

where q is the density of wind, v is the relative velocity, x is the angular velocity, s refers to the stress tensor, andF*stands for external body forces

To predict the turbulence effects in the transient predictions, a standard two-equation k e model was used,20 which is based on the transport equations for the turbulence kinetic energy,

k, and its dissipation rate, e The standard k e model has the advantages of robustness, econ-omy, and reasonable accuracy, and is widely used and able to simulate many flow regimes:

@ðqkÞ

@t þ@ðqkuiÞ

@xi

¼ @

@xj

lþli

ak

@xj

þ Gkþ Gb qe  YMþ Sk; (3)

@ð Þqe

@t þ@ðqeuiÞ

@x lþli

a

@x

þ C1e

e

kðGkþ C3eGbÞ  C2eqe

2

k þ Se; (4)

whereGkis the generation of turbulent kinetic energy due to mean velocity gradients, Gb is the generation of turbulent kinetic energy due to buoyancy, YM represents the contribution of the fluctuating dilatation of the overall dissipation rate, ak and ae are the turbulent Prandtl numbers fork and e

C Computational domains, boundary conditions, and grid generation

The computational domain was a rectangle with a width of 6 m and a length of 12 m, the turbine was placed in the symmetry axis of the top and bottom boundary and at a distance of 3

m from the left boundary (see Fig 3) The overall domain is split into six subdomains, includ-ing an external stationary domain, an internal rotational domain, and four subdomains that con-tain the four blades

The boundary conditions employed consist of a velocity inlet on the left side, a pressure outlet on right, and two sliding walls on top and bottom No-slip boundary conditions were imposed at the surface of the blades Siding interfaces exist between the external stationary do-main and the internal rotational dodo-main, allowing the transport of the flow properties Sliding interfaces also exist between the four subdomains and the internal rotational domain

The above method of domain division is reasonable because the velocities of both the internal rotational domain and subdomains can be defined individually using the User Defined Function (UDF) The rotational motion of the internal rotational domain represents the motion of the turbine, and the motion of the four subdomains represents that of the four blades

The domain is discretized with rectangular elements, with a total number of about 80 000 Grids closest to the profiles of the blades were refined with rectangular boundary elements to

FIG 2 Simplified physical model.

TABLE I Main parameters of the simplified physical model.

describe with sufficient precision the boundary layer flow The height of the first elements above the wall surface was set such that the yþ value was between 30 and 100, depending on the rotation velocity of the rotor and the position of the elements on the blade Grid node den-sity was higher in the subdomains than in the external stationary domain and internal rotational domain Moreover, to precisely present the flow field inside the turbine, grid node density is higher near the blades The grids were created so that the control volumes are finer near the blades and coarser towards the boundaries (Fig.4)

In this paper, all simulations were carried out with a constant wind velocity, 6 m/s, at the inlet of the domain The outlet pressure is set at standard atmospheric pressure,P0¼ 1  105Pa The air was considered incompressible with a density of 1.2084 kg/m3and a dynamic viscosity of 1:7979 105Pa

Blades of different chord lengths, from 300 mm to 600 mm, were studied to find the opti-mal chord length For each case, several simulations were studied, for flow coefficient k ranging between 0.2 and 0.6 The flow coefficient represents the ratio of blade rotating speed to free wind speed, and has the following expression:

k¼xR

The rotating simulations were conducted for three rotor revolutions with a time step of 2/ step and in each step, 100 convergences were determined by the order of magnitude of the residuals The drop of all scaled residuals below 105 was employed as convergence criterion

D Numerical method validation

In order to validate the accuracy of the sliding mesh method used in this paper, we calcu-lated the averaged torque of a two-bladed Savonius turbine for a range of rotation velocities The numerical results are then compared with wind tunnel experimental data from Hayashi’s work.24Fig 5illustrates the schematic of the Savonius turbine and detailed geometric parame-ters can be found in Ref.24

Fig.6shows the averaged torque of the Savonius turbine at different rotation velocities It can be seen from Fig.7that the numerical results agree well with the experimental data, espe-cially when the rotation velocity exceeds 30 rad/s The curve of the numerical results is a little higher than that of the experimental data This phenomenon may be explained by the fact that experimental conditions are not perfectly ideal and two-dimensional simulations do not consider the torque loss at the end of the blades

FIG 3 Computing domains and boundary conditions.

III RESULTS AND DISCUSSION

Performance of a wind turbine can be characterized by the manner in which the two main indicators—power coefficient (CP) and torque coefficient (Cm)—vary with k For analysis, the following relations have been used:

CP¼ P

where M is the dimensional torque, P is the dimensional power, q is the air density, U is the inflow speed, andR is the rotor radius The swept area of the rotor, S, is given by the relation-shipS¼ ð2R þ 0:5cÞH, where c is the blade chord length and H is the rotor height Since only 2D simulations were performed, the unit heightH¼ 1m was used

FIG 4 Computational grid of: (a) external stationary domain and (b) internal rotational domain and subdomains.

A Torque characteristic analysis

Fig.7shows the variation of averaged torque coefficient with respect to k for seven config-urations with chord length ranging from 300 mm to 600 mm It can be seen in Fig.7that chord length can significantly affect the averaged torque coefficient of the turbine A turbine with a larger chord length has a higher torque curve when the chord length is smaller than 550 mm However, further increasing the chord length will lower the torque curve The averaged torque coefficient decreases as k increases There exists an inflection point on each averaged torque curve, and as the chord length increases the inflection point moves forward After the inflection point, the torque coefficient shows a nearly linear decrease as the k increases, and this behav-iour is quite similar to other drag type turbines The effect of the chord length on the averaged torque coefficient was further investigated by computing the dynamic torque on a single blade throughout one revolution

Fig 8shows the dynamic torque coefficient of three blades with different length throughout one revolution at k¼ 0:5 The peak positive torque coefficient occurs at approximately u ¼ 25 and the peak negative torque coefficient occurs near u¼ 270 In the region u¼ ½120; 270,

FIG 5 Schematic of the Savonius turbine.

FIG 6 Averaged torque versus rotation velocity.

the torque coefficient curves first rise and then fall with small fluctuation A large region of posi-tive torque exists in u¼ ½40; 110 (torques are the same in the regions u ¼ ½40; 0 and

u¼ ½320; 360 due to the periodical rotating of the turbine), and a relatively small region of negative torque exists in u¼ ½230; 320 Blades with larger chord length generate larger posi-tive torque, but they also generate larger negaposi-tive torque

Fig.9shows the pressure contours of the turbine with chord length c¼ 550 mm at k ¼ 0:5 and u¼ 0, 20, 40, 60 Because two adjacent blades are 90apart, Fig.9illustrates the pres-sure contours of 16 blade positions When u¼ 20 (see the blade at the bottom left in Fig 9(b)), the blade has the most obvious pressure drop between the upwind surface and the down-wind surface This position corresponds to the maximum torque point in Fig 8 Fig 10 shows the velocity vectors of the turbine with chord length c¼ 550 mm at k ¼ 0:5 and u ¼ 40, 60 It can be seen from Fig 10 that the two downstream blades lie in the wake flow of the two

FIG 7 Averaged torque coefficient of the turbine with respect to k.

FIG 8 Dynamic torque coefficients of three blades with different length throughout one revolution at k ¼ 0:5.

upstream blades Because the wake flow is irregular turbulent flow, torque generated on the two downstream blades will fluctuate, corresponding to that in the region u¼ ½120; 270 in Fig.8 The wind slows down as it passes the upstream blades Due to the wind velocity loss, the pressure drop on the both sides of the downstream blades, as can be seen in Figs 9(c) and 9(d), is not as obvious as that of the upstream blades This illustrates that the torque coefficient

in the region u¼ ½120; 270 is far smaller than in u ¼ ½40; 110

B Power characteristic analysis

Fig.11shows the effects of chord length on the averaged power coefficient with respect to

k It can be seen from the figure that for a specific chord length, the averaged power coefficient

FIG 9 Pressure contours at k ¼ 0:5 with chord length c ¼ 550 mm (a) u ¼ 0  , (b) u ¼ 20  , (c) u ¼ 40  , and (d) u ¼ 60 

FIG 10 Velocity vectors at k ¼ 0:5 with chord length c ¼ 550 mm (a) u ¼ 40  and (b) u ¼ 60