Professional Documents
Culture Documents
Renewable Energy
journal homepage: www.elsevier.com/locate/renene
a r t i c l e i n f o a b s t r a c t
Article history: Portable micro-hydrokinetic turbines are designed and characterized using computational fluid dy-
Received 22 February 2014 namics (CFD) simulations. The two equation keu shear-stress transport (SST) turbulence model is
Accepted 12 November 2014 employed to predict quasi steady flow structures for a wide range of tip-speed ratios. Seven input design
Available online 29 November 2014
parameters selected a priori are used to create preliminary turbine rotor designs by using a hydraulic
design methodology. Various blade designs are characterized and compared in terms of torque and
Keywords:
thrust over a range of operating conditions. Performance characteristics of two, three, and four blade
Micro-hydrokinetic
designs are shown to be similar. The results indicate that a maximum power coefficient of 0.43 with a
Portable turbine
Turbulence
73.7% efficiency relative to Betz limit is achieved. The portable hydrokinetic turbines, designed and
CFD characterized here, do not require large civil structures, making this technology an attractive alternative
to conventional hydropower.
© 2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.renene.2014.11.032
0960-1481/© 2014 Elsevier Ltd. All rights reserved.
W.C. Schleicher et al. / Renewable Energy 76 (2015) 234e241 235
sections and thus can affect the power output from the rotor. the blades. Hayati et al. [18] investigated the effect of rake angle on
Hwang et al. [10] studied a vertical axis turbine that actively marine propeller performance and concluded that increasing the
controlled blade attack to maximize power output and improved rake angle improved the thrust performance of conventional pro-
self-start. They showed that by individually controlling each blade's pellers. Even though these propellers are imposing energy onto the
attack based on the oncoming flow that there was a 25% fluid and not absorbing the energy, it is possible that adjusting the
improvement in performance compared to pure cycloidal motion rake angle may improve thrust performance in the energy
for the same operating conditions. absorbing case.
The design principles used for wind turbines, marine propellers,
and propeller turbines can be used in hydrokinetic designs. Mas- 2. Hydraulic design
souh and Dobrev [11] studied the vortex wake behind a horizontal
axis wind turbine in a wind tunnel and compared the results to CFD Some design parameters must be assumed a priori to the design
analysis. Their results showed that the tip vortices are not limited to process. These input variables are shown in Table 1.
a cylindrical surface as predicted from linear propeller theory, and First, the tip diameter (Dt), hub diameter (Dh), and mean
expand radially as they move downstream. This increases the diameter (Dm) are calculated. A rough estimate of the required tip
diameter of the stream-tube that encompasses the turbine. Ver- diameter (D*t ) is calculated as shown in equation (1). This rela-
meer et al. [12] also studied the wake characteristics behind wind tionship is derived from the fluid's power flux through the rotor
turbines in the near and far wake regions. Alexander et al. [13] have blade. This equation assumes that there is no hub, therefore the
studied axial-flow, flat blade propeller turbines that can be man- result must be rounded up to account for the area lost by the hub.
ufactured in underdeveloped countries to provide sustainable po- Once the rounded tip diameter (Dt) is selected, the hub diameter is
wer generation for communities. They have shown that simplifying
the blade geometry on propeller turbines can still produce a sig-
nificant amount of power and can be easier to manufacture for Table 1
locations where advanced machining may not be possible. The Input design variables selected a priori.
work of Alexander et al. [13] was validated and compared with an Input variable Description
Archimedean screw turbine by Schleicher et al. [14], who have P Designed mechanical power output [W]
studied different micro-hydro systems [15,16]. Singh and Nestmann CP Designed power coefficient [e]
[17] experimentally studied the part-load performance on small U Designed free-stream velocity [m s1]
axial-flow propeller turbines and found that modifying the exit tip U Designed rotation rate [rad s1]
ZB Designed number of blades [e]
region on their studied propellers consistently showed an increase
s Designed solidity [e]
in flow and output shaft power and thus the hydraulic efficiency of b Designed blade thickness [m]
236 W.C. Schleicher et al. / Renewable Energy 76 (2015) 234e241
predicted as depicted in equation (2). The selected hub and tip Table 2
diameters are used to calculate the mean diameter. The mean Preliminary turbine design variables studied.
diameter is where the blade angles will be prescribed for the pre- Input design variables
liminary design. Multiple diameters between the hub and tip can P 500 [W] s 0.83 [e]
be used if more control of the blade angles is desired. CP 0.4 [e] ZB 2 [e]
U 2.25 [m s1] b 0.0127 [m]
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi U 15.708 [s1]
8P Output design variables
D*t y (1) Dt 0.5334 [m] j 72.26 [ ]
CP prU 3 Dh 0.0635 [m] s 0.5882 [m]
Dm 0.3745 [m] c 0.4882 [m]
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi x 1.3072 [e] Dq 142.29 [ ]
8P b 52.58 [ ] Dm 0.1488 [m]
Dh y D2t (2) b0 72.26 [ ]
CP prU 3
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pDm
1 2
s¼ (7)
Dm ¼ Dt D2h (3) ZB
2
Once the mean diameter has been selected, the relative flow
c ¼ ss (8)
angles to the turbine's rotating reference frame can then be
determined. A simplifying assumption used is that the relative flow
angles entering and leaving the turbine are only functions of radial
Dm ¼ c cos j (9)
distance. This means that the relative flow angle incident to the
leading edge is the same as the deviation of the flow from the 2c
Dq ¼ sin j (10)
trailing edge (b1 ¼ b2 ¼ b). This leaves the local tip-speed ratio (x) Dm
and relative flow angle to be calculated from equations (4)e(6). The preliminary design studied is pictured in Fig. 1. The corre-
sponding design variables are listed in Table 2.
1D U
m Further optimization of the design could be achieved using a
x¼2 (4)
U response surface methodology. This method provides a systematic
design strategy by changing all variables at the same time, allowing
for optimum conditions to be found regardless of initial values.
b ¼ tan1 x (5)
Changing one variable at a time while holding other variables
constant is a common approach to optimization, however, it is
b0 ¼ b þ 24:874x0:876 (6) inefficient and usually unsuccessful at arriving at an optimized
design. A central composite design of experiments will be used to
The relative blade angles can then be evaluated. Since the
determine the simulations needed to model a second-order
relative incidence and deviation flow angles are equal, the leading
regression of the results. From this regression analysis, minimum
edge and trailing edge relative blade angles are equal as well
and maximum quantities can be obtained for performance char-
ðb01 ¼ b02 ¼ b0 Þ, thus the relative blade angle is equal to the stagger
acteristics of the turbine-diffuser system, helping to narrow in on
angle (b0 ¼ j). Cebria n et al. [19] empirically related the relative
an optimum hydraulic design. The present authors employ this
blade angle to the relative flow angle and the local tip-speed ratio
method to optimize a system consisting of a hydrokinetic turbine
for maximum pressure loading in flat plate cascades as seen in
and a diffuser. The results of such optimization will be presented in
equation (6).
a full paper.
Finally, the mean chord length, meridional blade length, and
wrap angle are determined. The circumferential spacing between
blades (s) is calculated as shown in equation (7). The solidity chosen 3. Modeling, meshing, and numerical method
a priori is used to calculate the mean chord length. The meridional
blade length and wrap angle are determined once the mean chord Pictured in Fig. 2a is an overview of the computational domain
length is calculated. mesh. The domain is made up of two regions: the outer region and
a cylindrical region interior to the mesh encompassing the turbine frame was used in the vicinity of the turbine which transformed
(See Fig. 2d). These two regions are not conformal between meshes flow from an unsteady inertial frame to a steady non-inertial frame
in order to generate a structured hexahedral mesh needed for these with the inclusion of the centrifugal and Coriolis forces into the
computations. The outer domain is channel-like and has a diameter transport equations shown below
of 11.4Dt and length of 14.6Dt. These dimensions were determined
to be large enough that the far-field boundary conditions had little vUr;i
to no effect on the turbine's performance characteristics. The inner ¼0 (11)
vxi
cylindrical domain is 2Dt in diameter and extends 6Dt downstream
and 5.7Dt upstream of the turbine. The centerline of the turbine sits vUr;i vUr;i 1 vp v2 Ur;i
midstream in the channel and 2.3Dt below the slip wall. Fig. 2b þ Ur;j ¼ 2εilq Ul Ur;q εilq εqst Ul Us xt þ n
vt vxj r vxi vxj vxj
depicts the surface mesh on the inlet to the computational domain
(12)
and Fig. 2c shows the surface mesh on the turbine blades. Special
attention was paid to resolve the boundary layers adequately for Here, Ur is the velocity relative to the rotating frame of reference.
the implemented turbulence model. The position is represented by x and time by t. Static pressure is
The computations were carried out with the extended library of depicted as p, density by r, angular rotation rate of the frame of
OpenFOAM [20] using the steady-state SIMPLE segregated solver reference by U, ε is the permutation symbol and the subscripts i, j, l,
method with multiple frames of reference. A rotating reference q, s, and t are index placeholders.
Fig. 2. A) Overview of domain mesh B) mesh on the inlet C) mesh on the turbine blades D) clip through the mesh showing the placement of the rotor.
238 W.C. Schleicher et al. / Renewable Energy 76 (2015) 234e241
improved prediction of adverse pressure gradients in the near wall f ¼ torque [Nm] f ¼ thrust [N]
region as compared to the standard keu and keε models by N1, N2, N3 1,188,542, 5,929,864, 14,607,868 1,188,542, 5,929,864, 14,607,868
incorporating Bradshaw's observation that turbulent shear-stress is r21 1.709 1.709
proportional to the turbulent kinetic energy in the wake region of r32 1.351 1.351
the boundary layer [22]. The equations for kinematic eddy viscosity, f1 34.5242 644.4971
f2 35.3426 636.9606
turbulent kinetic energy, and specific dissipation rate are written as
f3 35.2616 636.3088
p 1.25 1.02
a1 k f21 33.6621 654.8748
nT ¼ (13) ext
maxða1 u; SF2 Þ e21
a 2.4% 1.2%
e21
ext 2.6% 1.6%
" # GCI21fine 3.1% 2.0%
vk vk vU v vk
þ Uj ¼ tij i b* ku þ ðn þ sk nT Þ (14)
vt vxj vxj vxj vxj
(GCI21
fine ) were calculated. The results show that there is a 2e3%
error band on the calculated quantities of torque and thrust due to
" #
vu vu v vu discretization.
þ Uj ¼ aS2 bu2 þ ðn þ su nT Þ Pictured in Fig. 3 are the flow field results for the N2 mesh on the
vt vxj vxj vxj
meridional plane with a free-stream velocity of 2.25 m/s and con-
1 vk vu stant rotation rate of 150 rpm. The results presented in Fig. 3 have
þ 2ð1 F1 Þsu2 (15)
u vxj vxj been normalized as
0.3
P
P 1
CP ¼ p (20) 0.2
8 rU Dt
3 2
0.1 0.5
T
CT ¼ p (21)
8 rU 2 D2
t 0 0
0 2 4 0 2 4
Here, R is the blade tip radius, T is the thrust force, TSR is the tip- TSR [ − ] TSR [ − ]
speed ratio, CP is the power coefficient, and CT is the thrust
coefficient. Fig. 4. Characterization of a two, three, and four blade preliminary design.
240 W.C. Schleicher et al. / Renewable Energy 76 (2015) 234e241
A) B) C)
6 6 6
4 4 4
2 2 2
[−]
press 0 0 0
C
−2 −2 −2
−4 −4 −4
−6 −6 −6
0 0.5 1 0 0.5 1 0 0.5 1
Normalized Chord [−]
Fig. 5. Pressure coefficient as a function of normalized chord length for A) near the hub radius B) at the mean diameter and C) near the blade tip.
The results indicate that there is no significant improvement in be conducted to improve the loading on the blade by more uni-
power generation with the addition of blades, however, the addi- formly distributing the pressure acting on the blades.
tion of blades does appear to significantly affect the thrust gener- Presented in Fig. 6 is a comparison of power coefficient with
ated from the blades. As the number of blades increases, the net general performance of various wind turbine designs [27]. The
thrust on the propeller turbine increases. The preliminary design preliminary design is in compliance with the Betz limit and
methodology predicted the power performance at the designed Glauert's ideal prediction of efficiency. The design also has a higher
tip-speed ratio within 7% of the CFD results, with a power coeffi- power coefficient compared to wind turbines operating in a similar
cient of 0.43 at the turbine's best efficiency point. This results in a tip-speed ratio range; however, the data used to compile the wind
73.7% relative efficiency to the Betz limit (CP ¼ 0.593), the theo- turbine plots may be dated and have higher power coefficients.
retical limit to the power coefficient.
The thrust coefficient trend observed is similar to that of Batten
et al. [5e7]. Fig. 4b shows that the initial increase in thrust coeffi- 5. Conclusions
cient occurs quicker at lower tip-speed ratios and peaks around a
tip-speed ratio of 0.5. Thrust coefficient decreases as tip-speed in- A preliminary design methodology was presented for a person-
creases above 0.5, as shown in Fig. 4b for the two bladed design. The portable micro-hydrokinetic turbine. This system will be used in
calculated thrust coefficient is also higher than the results reported military temporary encampment situations to provide enough
by Batten et al. It is believed these differences are a result of the energy to run necessary equipment. The design will be interfacial
propeller-type design as opposed to the axial fan designs studied by with the US Marine's GREENS solar modules.
Batten et al. The keu SST turbulence model was employed with a steady-
Plots of the pressure coefficient near the hub, mean, and tip radii state SIMPLE segregated solver with multiple frames of reference
are shown in Fig. 5. Here, pressure coefficient is defined as to characterize the preliminary design. A spectral convergence
Cpress ¼ (p p∞)/(rU2/2), where p∞ is the free-stream pressure and study was conducted, and a 5,929,864 cell mesh was chosen for the
U is the free-stream velocity magnitude. The normalized chord at calculations, providing a GCI of 3.1% for torque and 2.0% on thrust.
the 0 station corresponds to the leading edge and 1 station at the This was a compromise between computational resources and
trailing edge. The hub plot in Fig. 5a depicts a well loaded segment spectral independence for the computations. The detailed spatial
of the blade. The mean radius in Fig. 5b indicates that the loading on structure of velocity, pressure, and vorticity fields is characterized
this portion of the blade has peaked. Fig. 5c is near the tip radius of both upstream and downstream of the turbine. The flow field in the
the turbine and shows that the pressure loading is beginning to wake of the turbine is governed by pairs of double helix vortex
approach a stalled condition. Future optimization simulations will structures.
0.6
0.5
Betz Limit
C [−]
0.4
Glauert Ideal
P
Fig. 6. Comparison of the preliminary design's performance with existing wind turbine designs (adapted from Fox et al. [27]).
W.C. Schleicher et al. / Renewable Energy 76 (2015) 234e241 241
The preliminary design method and computational results agree [8] Mukherji SS, Kolekar N, Banerjee A, Mishra R. Numerical investigation and
evaluation of optimum hydrodynamic performance of a horizontal axis hy-
within 7% for the best efficiency design point at 2.25 m/s and a
drokinetic turbine. J Renew Sustain Energy 2011;3:1e18.
150 rpm rotation rate. The results indicate a maximum power co- [9] Myers L, Bahaj AS. Power output performance characteristics of a horizontal
efficient of 0.43 with a relative efficiency of 73.7% compared to the axis marine current turbine. Renew Energy 2006;31:197e208.
Betz limit. The results match well with general trends for various [10] Hwang IS, Lee YH, Kim SJ. Optimization of cycloidal water turbine and the
performance improvement by individual blade control. Appl Energy 2009;86:
wind turbine designs [27]. It also exceeds the performance of wind 1532e40.
turbine designs that operate in the same tip-speed ratio range. [11] Massouh F, Dobrev I. Exploration of the vortex wake behind of wind turbine
Future design optimization will be conducted with the addition rotor. J Phys Conf Ser 2007;75:1e9.
[12] Vermeer LJ, Sørensen JN, Crespo A. Wind turbine wake aerodynamics. Prog
of a diffuser to enhance power generation characteristics. Optimi- Aerosp Sci 2003;39:467e510.
zation will also occur at slower fluid speeds to broaden the appli- [13] Alexander KV, Giddens EP, Fuller AM. Axial-flow turbines for low head
cable range of the system. Once an optimized design is found, a microhydro systems. Renew Energy 2009;34:35e47.
[14] Schleicher WC, Ma H, Riglin JD, Kraybill Z, Wei W, Klein R, et al. Characteristics
prototype will be constructed and the numerical results will be of a micro-hydro turbine. J Renew Sustain Energy 2014;6:1e14.
validated with experimental results at the David Taylor Model [15] Schleicher WC, Riglin JD, Kraybill ZA, Oztekin A. Design and simulation of a
Basin. micro hydrokinetic turbine. In: 1st marine energy technology symposium,
Washington, D.C.; 2013.
[16] Riglin JD, Schleicher WC, Kraybill Z, Klein RC, Oztekin A. Computational fluid
Acknowledgment dynamics and structural finite element analysis of a micro hydro turbine. In:
ASME 2013 international mechanical engineering congress and exposition,
San Diego; 2013.
The authors would like to gratefully acknowledge funding from [17] Singh P, Nestmann F. Exit blade geometry and part-load performance of small
the Office of Naval Research for this work under Award N00014-12- axial flow propeller turbines: an experimental investigation. Exp Therm Fluid
Sci 2010;34:798e811.
M-0050. [18] Hayati AN, Hashemi SM, Shams M. A study on the effect of the rake angle on
the performance of marine propellers. J Mech Eng Sci 2012;226:940e55.
[19] Cebrian D, Ortega-Casanova J, Fernandez-Feria R. Lift and drag characteristics
References of a cascade of flat plates in a configuration of interest for a tidal current
energy converter: numerical simulations analysis. J Renew Sustain Energy
[1] Conti JJ, Bearmon JA, Napolitano SA, Schaal AM, Turnure JT. Annual energy 2013;5:1e19.
outlook 2013 with projections to 2040. Washington, DC: U.S. Energy Infor- [20] ESI-OpenCFD. OpenFOAM documentation. 5 February 2014 [Online]. Avail-
mation Administration; 2013. able: http://www.openfoam.org/docs/index.php.
[2] Ravens T, Cunningham K, Scott G. Assessment and mapping of the riverine [21] Menter F. R. Zonal two equation k-u turbulence models for aerodynamic
hydrokinetic resource in the continental United States. Palo Alto, CA: Electric flows. [AIAA Paper 93-2906].
Power Research Institute; 2012. [22] Menter FR. Two-equation eddy-viscosity turbulence models for engineering
[3] Jenkins D. Renewable energy systems: the earthscan expert guide to renew- applications. AIAA J 1994;32(8):1598e605.
able energy technologies for home and business. Florence: Taylor and Francis; [23] Celik I, Chen CJ, Roache PJ, Scheurer G. Symposium on quantification of un-
2013. certainty in computational fluid dynamics. In: FED-ASME, New York; 1993.
[4] U.S. Marine Corps. Guide to employing renewable energy and energy efficient [24] Roache PJ. A method for uniform reporting of grid refinement studies. In:
technologies. Quantico: Marine Corps Warfighting Laboratory; 2012. Proceedings of the 11th AIAA computational fluid dynamics conference,
[5] Batten WM, Bahaj AS, Molland AF, Chaplin JR. Hydrodynamics of marine Orlando; 1993.
current turbines. Renew Energy 2006;31:249e56. [25] Roache PJ. Perspective: a method for uniform reporting of grid refinement
[6] Batten WM, Bahaj AS, Molland A, Chaplin JR. The prediction of the hydrody- studies. ASME J Fluids Eng 1994;116(3):405e13.
namic performance of marine current turbines. Renew Energy 2008;33: [26] Roache PJ. Quantification of uncertainty in computational fluid dynamics.
1085e96. Annu Rev Fluid Mech 1997;29:123e60.
[7] Batten WMJ, Bahaj AS, Molland AF, Chaplin JR. Experimentally validated nu- [27] Pritchard PJ, Fox RW, McDonald AT. Fox and McDonald's introduction to fluid
merical method for the hydrodynamic design of horizontal axis tidal turbines. mechanics. Hoboken: John Wiley & Sons, Inc.; 2011.
Ocean Eng 2007;34:1013e20.