Академический Документы
Профессиональный Документы
Культура Документы
FIV2018-161
ABSTRACT INTRODUCTION
Parametric excitation is a classical non-autonomous
Wind turbine blades are subjected to oscillatory loads problem and is observed when at least one of the pa-
due their rotation. The gravity compresses or tensions the rameters of the equation of motion depends explicitly on
blade depending on its position along the cycle. Due to time. Particularly, if the stiffness is written as k(t) =
sheared flow, the wind load also changes in a cyclic way k̄ + ∆k sin Ωt, the equation of motion assumes the form
as the blade rotates. This kind of load can lead to fa- of the Mathieu’s equation.
tigue damage. It is possible to analyse some of the funda- The stability of the trivial solution depends on the pa-
mentals of this physical phenomenon considering a two- rameters that govern the parametric excitation (i.e, ∆k and
dimensional cross section of the blade, which is an airfoil. Ω). Particularly, the condition in which the parametric ex-
This airfoil is allowed to respond in the heaving di- citation has twice the natural frequency of the structure is
rection, subjected to oscillatory flow and parametric exci- a favorable scenario for the instability of the trivial solu-
tation, due to the variation of the structural characteris- tion and is herein named as principal parametric instabil-
tics of the blade because of the alternation of compression ity condition. The Strutt’s diagram is a common way to
and tension. check the stability of the trivial solution in the space of
parameters ∆k × Ω. Further details regarding fundamen-
In this work, we use numerical simulations to inves- tal aspects of the parametric excitation can be found in
tigate the oscillating flow around an airfoil allowed to the textbooks [1], [2] and [3].
translate in the heaving direction, mounted on an elastic Parametric excitation is commonly investigated in the
base of varying stiffness. The flow and stiffness vary with ocean engineering scenario, particularly in risers’ dynam-
the same frequency. The goal is to assess the effect of ics. In this application, the motion of the floating units in-
parametric excitation on the response of the structure to duces a time-dependent normal force and, consequently, a
the oscillating aerodynamic load. time-modulated geometric stiffness. Examples of works
1
with uniform properties are shown in equations (1) and
(2), in which v̄ j are the averaged velocity components,
v0j are the components of the velocity fluctuations, x j are
the space directions, t is time, ρ is the fluid density, p̄ is
averaged the pressure and ν the fluid kinematic viscosity.
2
TABLE 2: Velocity inflow parameters
Parameters Value
Section radius [m] 60
Tower height [m] 80
Wind Velocity at 80 m [m/s] 10
Shear coefficient 0.1
FIGURE 2: Mesh employed in the simulations
Rotor frequency(ω) [rpm] 12
Parameters Number blade rotation and the wind shear effect. In this work, we
studied a section located at 60 m from the root blade in a
Points 11410 wind turbine with tower height of 80 m, under the action
Cells 11155 of a wind with speed of 10 m/s at 80 m, with shear coeffi-
cient of 0.1. The wind turbine rotor rotation was constant
and equal to 12 rpm. The airfoil mass was considered to
be 88.5 kg/m. The parameters for the inflow velocity are
M denotes the airfoil mass, C the structural damping (as- shown in table 2.
sumed 0 in this work), K is the spring stiffness coefficient, We have investigated a total of 15 different cases,
F are the aerodynamic forces, all per unit length. In turn varying the mean stiffness of the structure, k̄, and the am-
y, ẏ and ÿ are body displacement, velocity and accelera- plitude of the stiffness variation, ∆k. The values p of the
tion respectively. parameters were chosen so as to have n = ω/ k̄/m =
The coupling between fluid and structure is two-way, 1, 1.23, 1.41, 1.5 and 2 and ∆k = 0, 0.1k̄ and 0.2k̄.
because the flow forces the airfoil, generates an acceler-
ation of the body, which is then used as a source term in
the flow momentum equation. The flow boundary condi- RESULTS
tions at the airfoil surface are also changed, since the flow Table 3 shows the maximum displacement and phase
there has the same velocity of the airfoil. angle (difference between the phases of the main compo-
The numerical simulations were carried out using the nents of force and displacement) obtained for all the cases
open source software OpenFOAM. The solver used for investigated. It is clear that the parametric excitation has a
the solution of the equations (1), (2) and (3), fluid part, very important impact on the dynamic response of the air-
was the pimpleDyMFoam and for the rigid body, equa- foil. The maximum displacements obtained for cases in
tion (4), we employed the library sixDoFRigidBodyMo- which ∆k 6= 0 are significantly higher than those obtained
tion with the solver symplectic. for ∆k = 0, and the largest displacements were observed
The mesh used had 70 m of height and 60 m of length. for the highest ∆k. The phase angle, on the other hand,
Five refinement regions were built around the airfoil, hardly varied in the cases investigated.
using triangles on intermediate refinement regions and One of the most interesting result of this analysis was
quadrilateral elements in the rest of the mesh. Next to that the variation of the maximum displacement with re-
the airfoil wall, we built ten layers of refinement with spect to the variation of the natural frequency was not
growing rate of 1.2, where the first layer has 0.002 m, monotonic, as can be seen in figure 3. The maximum
to discretize the boundary layer region and resulted in a displacement increased from n = 1 to n = 1.41, then ex-
maximum y+ value of 190. The table 2, summarizes the hibited a sudden drop for n = 1.5 followed by a slight de-
mesh statistics and figure 2 shows the mesh. crease for n = 2 when ∆k 6= 0 and increase when ∆k = 0.
Velocity and angle changed with the time, due to the This difference in behaviour for cases with and without
3
TABLE 3: Maximum displacement and phase angle for
each case calculated
4
FIGURE 6: Displacement, force and stiffness time series,
k̄ = 70.07 N/m/m, ∆k = 0.2k̄.
FIGURE 8: Lift time series, k̄ = 140.14 N/m/m, ∆k =
0.2k̄.
5
ical insight about the phenomenon. We intend to proceed
with the work, improving the model so as to get results
that will be more relevant to real applications.
ACKNOWLEDGEMENTS
The third author is grateful to the Brazilian National
Council of Research (CNPq) for the grant 310595/2015-
0.
REFERENCES
[1] Nayfeh, A. H., 1973. Perturbation Methods. John
Wiley & Sons.
[2] Nayfeh, A. H., and Mook, D. T., 1979. Nonlinear
oscillations. John Wiley & Sons.
[3] Meirovitch, L., 2003. Methods of Analytical Dynam-
FIGURE 10: Static pressure contours along one revolu-
ics. Dover Publications.
tion, k̄ = 140.14 N/m/m, ∆k = 0.2k̄. [4] Patel, M. H., and Park, H. I., 1991. “Dynamics of
tension leg platform tethers at low tension. Part I -
Mathieu stability at large parameters”. Marine Struc-
tures, 4, pp. 257–273.
direction, subjected to oscillatory inflow and parametric [5] Franzini, G. R., Pesce, C. P., Salles, R., Gonçalves,
excitation. This model is a simplification of a wind tur- R. T., Fujarra, A. L. C., and Mendes, P., 2015. “Ex-
bine blade section turning subjected to sheared inflow. perimental investigation with a vertical and flexible
The results were obtained from numerical simulations us- cylinder in water: response to top motion excitation
ing the finite volume method and Spalart-Allmaras tur- and parametric resonance”. Journal of Vibration and
bulence model. We have tested fifteen different cases, Acoustics, 137 (3), pp. 031010–1 – 031010–12.
varying the mean and amplitude of the stiffness of the [6] Franzini, G. R., and Mazzilli, C. E. N., 2016. “Non-
airfoil. The system exhibited an interesting behaviour, linear reduced-order model for parametric excitation
showing a that the maximum displacement increased with of vertical and immersed slender rod”. International
increasing stiffness amplitude. However, the variation of Journal of Non-linear Mechanics, 80, pp. 29–39.
the maximum displacement with the mean stiffness was [7] Ramakrishnan, V., and Feeny, B. F., 2012. “Reso-
not monotonic, and presented a peak when the forcing nances of a forced mathieu equation with reference
frequency was 1.41 times the natural frequency of the to wind turbine blades”. Journal of Vibration and
blade. Also interesting was the frequency content of the Acoustics, 134, pp. 064501–1–064501–5.
lift signal. For the cases with no parametric excitation, the [8] Franzini, G. R., Santos, R. C. S., and Pesce, C., 2017.
lift signal clearly shows one strong component with the “A numerical study on piezoelectric energy harvest-
same frequency as the angular frequency of the turbine ing by combining transverse galloping and paramet-
and another component with frequency equals to twice ric instability phenomena”. Journal of Marine Sci-
the angular frequency of the turbine. This is due to the ence and Application, 16, pp. 465–472.
combination of the variation of velocity because of the [9] Spalart, P. R., and Allmaras, S. R., 1992. “A one-
sheared inflow and because of the airfoil displacement. equation turbulence model for aerodynamic flows”.
Increasing the stiffness amplitude, and consequently the In 30th Aerospace Sciences Meeting and Exhibit,
airfoil velocity in the heave motion, made the signal al- AIAA.
most monochromatic.
This is a first study about parametric excitation in
wind turbines, employing a simplified model to get phys-