For the simulation of service loads and of their effect on the whole turbine the wind turbine manufacturers use program systems whose particular strengths lie in the wind load simulation at the rotor, in the rotor dynamics as well as in the control-technological operation of the whole turbine. The complex dynamic behavior of the drive train, consisting of the rotor, the rotor shaft, the main gearbox, the brake, the coupling and the generator, is represented as a two-mass oscillator. This simplification, which certainly is necessary within the framework of the wind load simulation programs, is by no means sufficient for the exact description of the dynamics of the more and more complex drive trains with capacities up to 5 MW. At first, the extension to a multimass torsional vibration model seems to be useful for the exact determination of the torsional vibrations in the drive train. However, in the turbines of all manufacturers there have been found forms of damage on drive train components (high axial loads in bearings, high coupling loads, radial loads on generator bearings) that cannot be explained even on the basis of a torsional vibration analysis. Moreover, in measurements on drive trains natural frequencies in the signals occurred that can no longer be explained by the torsional vibration behavior alone. Consequently, a real multibody simulation becomes necessary, for which also radial and axial vibrations can be taken into account, in addition to torsion, since these influence the torsional vibration behavior considerably. These dependences become already clear in an analysis of natural frequencies. This is illustrated by the example of a 700-kW turbine as well as by a planetary gearing for a 3-MW turbine. Especially in the dimensioning of the off-shore turbines with several MW output power, which are being planned, the use of multibody simulation will be advantageous, since the testing of turbine prototypes of this order of magnitude under the corresponding operating conditions are surely more cost-intensive and risky than the virtual testing with well validated simulation models.
展开▼
