We performed a detailed validation study on a full scale Darrieus H-type VAWT

The wind turbine stands on a tubular steel tower, with a base diameter of 1.9 m. The drive train generator operates at 1200 rpm, while the rotor spins at a nominal speed of 55 rpm. The Micon 65/13M wind turbine was used for the Long-Term Inflow and Structural Testing program at the USDA-ARS test facility in Bushland, Texas. This project was initiated by Sandia National Laboratories in 2001 to explore the use of carbon fiber in wind turbine blades. The wind turbine is equipped with CX-100 blades, those structural model used in current FSI simulations was validated in Section 3.4.1. FSI simulations of the full Micon 65/13M wind turbine are carried out at realistic operational condition. A constant inflow wind speed of 10.5 m/s and fixed rotor speed of 55 rpm are prescribed. These correspond to the operating conditions reported for the field tests in [84]. The air density and viscosity are 1.23 kg/m3 and 1.78×10−5 kg/, respectively. Zero traction boundary conditions are prescribed at the outflow and nopenetration boundary conditions are prescribed at the top, bottom, and side surfaces of the outer computational domain. No-slip boundary conditions are prescribed at the rotor, nacelle, and tower, and are imposed weakly. Figure 4.3 shows the computational domain and Figure 4.4 mesh used in this study. The mesh consists of 5,134,916 linear elements, grow racks which are triangular prisms in the rotor boundary layers and tetrahedra everywhere else in the domain. The mesh is refined in the rotor and tower regions for better flow resolution near the wind turbine.

The size of the first element in the wall-normal direction is 0.002 m, and 15 layers of prismatic elements were generated with a growth ratio of 1.2. Figure 4.4 shows a 2D blade cross-section at 70% span wise station to illustrate the boundary-layer mesh used in the computations. The computations were carried out in a parallel computing environment. The mesh is partitioned into subdomains using METIS, and each subdomain is assigned to a compute core. The parallel implementation of the methodology may be found in [95]. The fluid and structural equations are integrated in time using the Generalized-α method with the time-step size of 3.0 × 10−5 s for all cases. In each time step, block-iterative FSI coupling is employed, which is efficient and stable for the application considered here.In Figure 4.5 the time history of the aerodynamic torque is plotted. As can be seen from the plot, using FSI, we capture the high frequency oscillations caused by the bending and torsional motions of the blades. In the case of the rigid blade the only high frequency oscillations in the torque curve are due to the trailing-edge turbulence. For the rigid blade case the effect of the tower on the aerodynamic torque is more pronounced, while in the case of FSI it is not as visible due to the relatively high torque oscillations. The ’dips’ in the aerodynamic torque can be seen at 60◦ , 180◦ , and 300◦ azimuthal angle, which is precisely when one of the three blades is passing the tower. The computed values of the aerodynamic torque are plotted together with field test results from. The upper and lower dashed lines indicate the aerodynamic torque bounds, while the middle dashed line gives its average value. Both the aerodynamic and FSI results compare very well with the experimental data.We present a preliminary, ongoing FSI simulation of a 5MW offshore wind turbine undergoing yawing motion. The wind turbine is equipped with 61 m blades designed by Sandia.

The structural model of a blade used in current FSI simulations was validated in Section 3.4.2. The wind turbine rotor is positioned at 80 m above ground and is tilted by 5◦ to avoid the blade hitting the tower as the rotor spins. Furthermore, the wind turbine rotor plane is initially placed at 15◦ relative to the wind direction. A fixed yawing rotational speed is applied to the gearbox to slowly turn the rotor into the wind at 0.03 rad/s . The inflow wind speed is set to 11.4 m/s. The initial rotor speed is set to 12.1 rpm, and the rotor is allowed to spin freely during the prescribed yawing motion. The structural mechanics mesh of the full turbine has 13,273 quadratic NURBS shell elements and two quadratic NURBS beam elements. The aerodynamics mesh has a total of 5,458,185 linear elements. Triangular prisms are employed in the blade boundary layers, and tetrahedral elements are used elsewhere in the aerodynamics domain.The size of the first boundary-layer element in the wall-normal direction is 1 cm, and the time step of 0.0001 s is employed in the computation. Snapshots of the structure deformed configuration are shown in Figure 4.10, while isosurfaces of vorticity colored by flow speed are shown in Figure 4.11. Figures 4.12 and 4.13 show the time history of the axial component of the aerodynamic torque and angular speed. Both are slowly increasing as the rotor turns into the wind, as expected. The level of the computed aerodynamic torque is consistent with the earlier simulationsfor this wind turbine operating under similar wind- and rotor-speed conditions .We present an FSI simulation of a 1.2 kW VAWT, which is a three-bladed, medium-solidity Darrieus turbine designed by Windspire Energy. The details of wind turbine geometry together with aerodynamic validation using a field-test data are presented in Section 2.3.2. The structural model is presented in Figure 4.14.

The rotor and struts are made of aluminum, and the tower is made of steel. Quadratic NURBS are employed for both the beam and shell discretizations. The total number of beam elements is 116, and total number of shell elements is 7,029. As a part of FSI simulations, we perform a preliminary investigation of the startup issues in VAWTs using the FSI methodology described earlier and the structural model of the Windspire design. We fix the inflow wind speed at 11.4 m/s, and consider three initial rotor speeds: 0 rad/s, 4 rad/s and 12 rad/s. Of interest is the transient response of the system. In particular, we will focus on how the rotor angular speed responds to the prescribed initial conditions, and what is the range of the tower tipdisplacement during the VAWT operation. The VAWT is allowed to spin freely and accelerate under the action of the ambient wind. The time step in the computations is set to 2.0 × 10−5 s. The mesh moving technique described in Section 4.2 is applied to this case in a straightforward fashion. The radius and height of the inner cylindrical domain that encloses the rotor are 1.6 m and 7 m, respectively. That is, the cylindrical domain extends 0.5 m above and below the rotor blades. The rotor axis direction nrot is defined according to Eq. , where the points xori and xtip are located at the bottom and top intersections of the tower beam and shell, respectively. The instantaneous rotor angular velocity is computed from Eq. , the spinning component is removed as per Eq. , and the two angular velocities are used to update the sliding-interface mesh positions. We fluid mesh was adopted from the aerodynamics simulations presented in Section 2.3.2 The time history of rotor speed is shown in Figures 4.15–4.17. For the 0 rad/s case the rotor speed begins to increase suggesting this configuration is favorable for self-starting. For the 4 rad/s case, grow table the rotor speed has a nearly linear acceleration region followed by a plateau region. In [16] the plateau region is defined as the regime when the turbine operates at nearly constant rotational speed. From the angular position of the blades in Figure 4.16 it is evident that the plateau region occurs approximately every 120◦ when one of the blades is in a stalled position. It lasts until the blade clears the stalled region, and the lift forces are sufficiently high for the rotational speed to start increasing again. As the rotational speed increases, the angular velocity is starting to exhibit local unsteady behavior in the plateau region. While the overall growth of the angular velocity for the 4 rad/s case is promising for the VAWT to self start, the situation is different for the 12 rad/s case . Here the rotor speed has little dependence on the angular position and stays nearly constant, close to its initial value. It is not likely that the rotor speed will reach to the operational levels in these conditions without an applied external torque, or a sudden change in wind speed,which is consistent with the findings of [17]. Figure 4.18 shows, for a full turbine, a snapshot of vorticity colored by flow speed for the 4 rad/s case. Figure 4.19 zooms on the rotor and shows several flow vorticity snapshots during the rotation cycle.

The figures indicate the complexity of the underlying flow phenomena and the associated computational challenges. Note the presence of quasi-2D vortex tubes that are created due to massive flow separation, and that quickly disintegrate and turn into fine-grained 3D turbulence further downstream. Figure 4.20 shows the turbine current configuration at two time instances during the cycle for the 4 rad/s case. The displacement is mostly in the direction of the wind, however, lateral tower displacements are also observed as a result of the rotor spinning motion. The displacement amplitude is around 0.10-0.12 m, which we find reasonable given the tower height of 9 m, and one of the VAWT design objectives being that the structure is not too flexible. This is also the case for the 0 rad/s and 12 rad/s cases.In this dissertation more advanced FSI simulations of wind turbines, such as rotor yawing for HAWTs, and full-machine FSI of VAWTs were targeted. A structural model of wind turbines design was constructed and discretized using the recently proposed isogeometric rotation-free shell and beam formulations. This approach presents a good combination of accuracy due to the structural geometry representation using smooth, higher-order functions, and efficiency due to the fact that only displacement degrees of freedom are employed in the formulation. By constructing a detailed material model of wind turbine blade with non-symmetric, multilayer layup we were able to reproduce the experimentally measured eigenfrequencies of the CX-100 blade of Micon 65/13M HAWT. To our knowledge, this is the first full-scale validation of the IGA-based thin-shell composite formulation. The ALE-VMS technique for aerodynamics modeling was augmented with an improved version of the sliding interface formulation, which allows the interface to move in space as a rigid object and accommodate the global turbine deflections in addition to the rotor spinning motion. The pure aerodynamics computation produced good agreement with reported wind tunnel and field-test data. A simulation of two side-by side wind turbines was also performed. Using novel mesh moving techniques we were able to simulate a large scale 5MW HAWT undergoing yawing motion. We also present FSI simulations of full-scale Micon 65/13M wind turbine with the CX-100 blades mounted on its rotor. The results of the aerodynamic and FSI simulations shows a good agreement with field test data for this wind turbine. The FSI simulation captures high-frequency oscillations in the aerodynamic torque, which are caused by the blade structural response. In the future work we plan to explore methods and devices to mitigate such high-frequency rotor vibrations. Dynamic FSI modeling of VAWTs in 3D and at full scale were reported for the first time in this dissertation with investigation of turbine start-up issues. From the FSI computations we see that for given wind conditions the rotor naturally accelerates at lower values of angular speed. However, as the angular speed grows, the rotor may encounter a dead band region. That is, the turbine self-starts, but then it is trapped in a lower rotational speed than is required for optimal performance, and some additional input is required to get the rotor to accelerate further. There may be multiple dead band regions that the turbine needs to overcome, with external forcing applied, before it reaches the target rotational speed. In the future, to address some of these issues, we plan to couple our FSI formulation with an appropriate control strategy to simulate more realistic VAWT operation scenarios. The numerical examples presented in this dissertation illustrate the successful application of the proposed techniques to the FSI simulation of wind turbines at full scale.It has been reported that bacteria loads associated with enormous amount of animal waste produced in the U.S. are the leading cause of impairment for rivers and streams.