2. Evaluate the redesigned GRC gearbox and Alstom’s drivetrain solutions for minimizing the impacts of non-torque loads using modeling and experimental approaches jointly;

3. Compare the drivetrain responses between the original GRC, the redesigned GRC and Alstom’s Pure Torque drivetrains.

2. GRC THREE-POINT SUSPENSION DRIVETRAIN AND ITS RELIABILITY ISSUES

2.1. Drivetrain configuration

The NREL GRC gearbox is mounted on a three-point suspension drivetrain as shown in Figure 1 and detailed by Musial et al.2 and Bodas and Kahraman.7 The power rating of the GRC drivetrain is 750 kW. A spherical roller main bearing sup- ports the main shaft. Two elastomeric trunnions in the torque arms support the gearbox. The hub center and the rotor’s cen- ter of gravity (COG) are upwind of the main bearing, resulting in a pitching moment on the main shaft in addition to aerodynamic loads.

Figure 4. Flow charts of non-torque load path of the (a) original GRC, (b) modified GRC and (c) Alstom’s Pure Torque drivetrains.

The gearbox has three stages with an overall gear ratio of 1:81.491. It is composed of one low-speed planetary stage and two parallel shaft stages. The planetary stage accommodates three-planet gears to achieve the best load sharing as recom- mended in design standards.12,13 A three-planet planetary gear allows a greater gear ratio compared with the four-planet system because of larger assembly space. In addition to load sharing, other benefits of utilizing three planets include low vibration, low noise and ease of assembly. The three-planet planetary system of the GRC gearbox suppresses the transla- tional vibration and resonances of translational modes. This observation is based on the mesh phasing rules (a vibration reduction model) discussed in the work of Parker,14 which was analytically derived based on the mesh load symmetry. By arranging a set of planet and tooth numbers of the sun gear, a particular harmonic of mesh frequency is suppressed in the dynamic response. As a result, the vibro-acoustic propagation of gearbox dynamics can be significantly reduced if gear vibration is minimized.

Torque is applied to the planetary carrier that is supported by upwind and downwind cylindrical roller bearings as shown in Figure 5. The sun gear is the output of the planetary gear stage, which is  a  floating  design.  A spline connection at the downwind end of the sun shaft allows the self-adjusting and radial motion of the sun gear to try to equalize the planetary load sharing. The ring gear is bolted to the front and rear gearbox housing. Double- row cylindrical roller bearings support the planet gears.4,9 The gearbox is connected  to  the  generator  through  a coupling  with  flexible  elastomeric links.

2.2. Study approaches

The GRC gearbox was tested both in the 2.5 MW dynamometer facility at the NWTC (shown in Figure 6) and in a wind turbine at Xcel Energy’s Ponnequin wind farm in northern Colorado for hundreds of operating hours. Gearbox internal loads and responses subjected to torque and non-torque loads were the focus of the measurements.4,15 Major instrumenta- tions and measurements included the following:

1. Relative displacement of planet carrier rim to gearbox housing;

2. Planet load share and ring gear tooth load distribution;

3. Main shaft azimuth angle to sync to bearing and gear strain  gauges;

4. Planet bearing loads and load zone;

5. High-speed shaft locating bearing axial load distribution;

6. Planet gear motion relative to carrier;

7. Sun pinion radial and axial motion;

8. Gearbox housing acceleration;

9. Gearbox displacement and misalignment at trunnion mounts.

Figure 5. NREL GRC original gearbox design cutaway view.

Figure 6. GRC gearbox in the dynamometer testing facility at NWTC.

Two modeling approaches were also used to perform sensitivity studies on the effects of non-torque loads on gearbox internal loads, including a computational model of the drivetrain and an analytical model of the planetary gear. The com- putational model of the GRC drivetrain in this work was established in SIMPACK version 8.9.16 The flexible main shaft, planetary carrier, housing and gear shafts were modeled using reduced degrees of freedom through modal condensation. The gear model accounted for tooth profile and lead modifications, tooth contact loss and fluctuating mesh stiffness. Bearings were modeled using diagonal stiffness matrices with clearances as shown in Figure 7. Trunnion mounts were modeled using diagonal stiffness matrices extracted from the measured load–displacement curve.4 The analytical model calculated the plane- tary gear load sharing and loads for three-point suspension drivetrains considering gravity, pitching moment, bearing clearance and input torque.17 Both approaches were validated against the experimental results collected during the field testing.