Authors: Ricardo Castillo, Yeqin Wang, Travis Monk, Stephanie Vasquez,. Suhas Pol, Beibei Ren, Andy Swift, Fazle Hussain, Carsten H. Westergaard.
PIV measurements in the wake of a dynamically controlled model wind turbine Authors: Ricardo Castillo, Yeqin Wang, Travis Monk, Stephanie Vasquez, Suhas Pol, Beibei Ren, Andy Swift, Fazle Hussain, Carsten H. Westergaard
Presenter: Carsten Westergaard,
Professor of Practice, Mechanical Engineering/ National Wind Institute Windfarms 2016, May 23-25, 2016, Dallas, TX
INTRODUCTION
Vision: Develop a scaled model wind farm simulator with
functional and programmable model scale turbines enabling fast controls innovation 600 times faster than real time Objectives of this work:
Develop a turbine platform emulating full
scale turbine controls capabilities Demonstrate wake control using different controls capabilities • Static yaw – wind misalignment • Dynamic wake control with dynamic yaw
TURBINE CONTROL CAPABILITES
+
Wind
+
PWM
DC generator 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝜔𝜔𝑔𝑔 𝑝𝑝𝑝𝑝𝑝𝑝 Blade Optimal control
Buck DC/DC converter 𝜔𝜔𝑟𝑟
-
Speed controller Pitch controller Yaw control
𝑉𝑉𝑑𝑑𝑑𝑑
𝐼𝐼𝑅𝑅 ,
R
VDC
Yaw control • Rate: 4 deg per rev.
Speed control • TSR: 0 to destruction or auto rotation
Local controller
• Rate of change RPM: ~ “instant”
Pitch control Ct for pitch -2 to +10
• Collective rate: ~4 deg per rev. (or ~ 0.3 points in Ct change per rev.) • Cyclic rate: one per rev.
0
2
TSR
6
8
10
Control authority in one rotation
TURBINE CONTROL
EXPERIMENTAL SETUP
PIV setup • Laser: 150 mJ Quantel, 10 Hz • Four LaVision PIV cameras • Field of view: 1.8 x 6.6 D • ~840 pixel per rotor diameter
Tunnel • Cross section: 1.2m X 1.8m • Wind speed: 10 m/s
Turbine • Nominal RPM ~ 5000 • TSR ~ 6.5 • REtip ~ 50,000 • Cl/Cdmax~25 (XFOIL)
PERFORMANCE AT YAW AND SPEED CHANGE
Turbine test • TSR 6.0 to 7.2 • Pitch 3 to 6 degrees
Reasonable agreement
with BEM simulation in Qblade from TU Berlin Power drops approximately with Cos2(θ) as seen in many other publications
AVERAGE PIV FLOW FIELD, STATIC YAW OFFSET
Ensemble average of 500 instances for each Yaw angle Takes only 50 sec. to obtain data Demonstrates wake steering using static yaw offset U=10.85 m/s, RPM=4300, TSR=6.2
NACELLE SIZE INFLUENCE NEAR WAKE SIGNIFICANTLY 1.1
1
0.9 yaw=-25
0.8
yaw=-20
Illustrate nacelle at -25, 0 and 25 degree
yaw=-15 yaw=-10
0.7
yaw=-5
U/U
yaw=0
0.6
yaw=5 yaw=10 yaw=15
0.5
yaw=20 yaw=25
0.4
x/D=1
0.3
0.2
0.1 -1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
Cross stream distance y/D
Clearly non-symmetrical profiles for both the “free wake” and
the part disturbed by the nacelle Small speedup outside wake
FAR WAKE
1
0.9
yaw=-25 yaw=-20 yaw=-15 yaw=-10
0.8
yaw=-5 U/U
yaw=0 yaw=5
0.7
yaw=10 yaw=15 yaw=20
0.6
yaw=25
x/D=7
0.5
0.4 -1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
Cross stream distance y/D
Asymmetrical wake, even at small yaw angles Clear asymmetry in wake profiles at large yaw angle Minor speedup outside wake
COMPARE DEFICIT TO JENSEN PARK MODEL
Observed k~0.033 for Yaw~0 degree Far wake agreement with Jensen for Ct ~ 1 and
Ct ~ 0.9 in yawed condition U=10.85 m/s, RPM=4950, TSR=6.6
WAKE DEFICIT Variation of wake minimum velocity as a function of yaw angle
0.8
0.7
0.6
x/D=1 x/D=2
Wake minimum velocity
0.5
x/D=3 x/D=4 x/D=5
0.4
x/D=6 x/D=7
0.3
0.2
0.1 -25
-20
-15
-10
-5
0
5
10
15
20
Yaw angle (degrees)
As angle increase, Ct decrease and deficit decrease
25
WAKE TRAJECTORY
0.6
0.4
yaw=-25
0.2
yaw=-20 yaw=-15
Cross stream distance y/D
yaw=-10 yaw=-5
0
yaw=0 yaw=5 yaw=10 yaw=15
-0.2
yaw=20 yaw=25
-0.4
-0.6 1
2
3
4
Downwind distance x/D
5
6
7
8
WAKE DEFLECTION at x/D =7
All data scaled linearly to Ct~1 and x/D~7 “according” to
the DTU linear model
𝑦𝑦 𝑥𝑥 = 0.24 𝐶𝐶𝑡𝑡 tan 𝜃𝜃 + 𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜 𝐷𝐷 𝐷𝐷
DYNAMIC WAKE STEERING WITH YAW CONTROL
10 cycles recorded (only two shown) at 10 Hz Yaw rate ~ 2.5 deg/sec U=10.85 m/s, RPM ~5400, TSR ~7.3
INTANTANEOUS VERSUS AVERAGE
Arrow indicate direction of wake motion
CONCLUSION
Dynamic control of model turbine wake agrees with expectations and wake
steering has been demonstrated for static yaw +/-25 degree and dynamics yaw +/-25 degree Average and dynamic flow fields have been satisfactorily captured Wake deflection observations can be collapsed with DTU linear model and is insensitive to blade layout and absolute size Nuances may exist, but for controls, it is sufficient. Estimating wake deflection with DTU model (proportional to Ct, tan(θ) and x/D) is a good first order approximation for controls These test results are in agreement with NREL 5MW reference SOFWA simulations, DTU Ellipsys simulations, Qblade simulations of a V27 rotor and NTU model test and 1.5 years average observations on 67 x 1.5 MW turbine in the mid-West form Sandia CREW project
FUTURE WORK
Future work • Improve the mechanical design of turbine, in particular nacelle impact, pitch accuracy and blade design • Increase the quality and accuracy of the data collection system, in particular measurement of Ct and increase field of view (modify windows in wind tunnel) • Refine wake steering strategies for deployment on DOE/Sandia SWiFT facility in conjunction with LIDAR installations Thank you to the Governors Office in the State of Texas and Binational Industrial Research and Development (BIRD) Foundation for supporting this work