Development of a wind turbine gearbox virtual load sensor using multibody simulation and artificial neural networks

A current development trend in wind energy is characterized by the installation of wind turbines (WT) with increasing rated power output. Higher towers and larger rotor diameters increase rated power leading to an intensification of the load situation on the drive train and the main gearbox. However, current main gearbox condition monitoring systems (CMS) do not record the 6‑degree of freedom (6-DOF) input loads to the transmission as it is too expensive. Therefore, this investigation aims to present an approach to develop and validate a low-cost virtual sensor for measuring the input loads of a WT main gearbox. A prototype of the virtual sensor system was developed in a virtual environment using a multi-body simulation (MBS) model of a WT drivetrain and artificial neural network (ANN) models. Simulated wind fields according to IEC 61400‑1 covering a variety of wind speeds were generated and applied to a MBS model of a Vestas V52 wind turbine. The turbine contains a high-speed drivetrain with 4‑points bearing suspension, a common drivetrain configuration. The simulation was used to generate time-series data of the target and input parameters for the virtual sensor algorithm, an ANN model. After the ANN was trained using the time-series data collected from the MBS, the developed virtual sensor algorithm was tested by comparing the estimated 6‑DOF transmission input loads from the ANN to the simulated 6‑DOF transmission input loads from the MBS. The results show high potential for virtual sensing 6‑DOF wind turbine transmission input loads using the presented method.

     

Integration of electromagnetic finite element models in a multibody simulation to evaluate vibrations in direct-drive generators

This paper introduces a novel electromechanical model for calculating electromagnetic excited structural vibrations and structure borne acoustics for gearless wind turbines. Therefore, the wind turbine model structure is explained and a drivetrain model is derived to investigate the drivetrain decoupled from the aerodynamic excitations. The drivetrain model is fed with results from an electromagnetic finite element model of the generator considering air gap width changes and the wind turbine torque and speed characteristics. Furthermore, an exemplary ramp-up of the drivetrain is simulated. It can be seen, that generator structure oscillations are excited during certain rotational speeds, which may be relevant for the acoustic behavior of the turbine.

     

Reduction of torsional vibrations of drivetrains of machines and plants using impulsive torques and characterization of related energy transfer effects

The reduction of torsional vibrations of drivetrains of machines and plants is of considerable interest in various fields of engineering, as they can have a negative influence not only on the drivetrain itself, but also on the driven machinery. Hence, a variety of countermeasures were developed to minimize such vibrations. In the present contribution, the possibilities of reducing torsional vibrations of drivetrains by introducing impulsive torques are investigated. The impulsive strength is chosen in a way that fully elastic impacts of a point mass with a rigid wall are mimicked, i.e., energy is neither fed to nor extracted from the mechanical system by the impulses. It is shown that therewith a transfer of vibration energy to higher modes is possible, where it is dissipated more effectively utilizing the enhanced damping capabilities of higher modes. A modal damping amplification factor is introduced allowing to characterize the energy transfer based on a mapping of the state-vector from one instant of time where an impulse is applied, to the next. It is demonstrated with a numerical example that the damping amplification factor allows to analyze modal energy transfer effects in mechanical systems exhibited to impulsive torques in an efficient manner.

     

Simulative investigation of wind turbine gearbox loads during power converter fault

Three phase short circuit power converter faults in wind turbines (WT) result in highly dynamic generator torque reversals, which lead to load reversals within the drivetrain. Dynamic load reversals in combination with changing rotational speeds are, for example, critical for smearing within roller bearings. Therefore, an investigation of the correlation between three phase short circuit converter faults and drivetrain component failures is necessary.

Due to the risk of damage and the resulting costs, it is not economically feasible to extensively investigate three phase short circuit converter faults on test benches. Valid WT drivetrain models can be used instead. A WT drivetrain model, which has been developed and validated in a national project at the CWD, is used and a three phase short circuit converter fault is implemented. In this paper, the resulting torque load on the drivetrain for a three phase short circuit converter fault in rated power production is presented. This converter fault leads to a highly dynamic reversing electromagnetic torque which exceeds the rated torque by a factor of three. As a result the load on the rotor side high speed shaft (HSS) bearing oscillates and increases by around 15 per cent compared to rated power production. Simultaneously the rotational velocity of the HSS oscillates with an amplitude of 10 rpm. Therefore an increase in the risk of smearing is expected.

     

Calculation of structure-borne sound in a direct drive wind turbine

In this publication, the methods will be presented that are deployed to formulate a multi-physical system model of a direct drive wind turbine in order to calculate structure borne sound. The model includes excitation effect as well as sound radiating behaviour. The mechanical structure as a medium partner between excitation and radiation will be formulated through a multi-body simulation model in the time domain. In the multi-body simulation model, all relevant drivetrain components are considered with their structural eigenmodes in the frequency range of interest. The electromagnetic forces of the multi-pole ring generator are calculated and introduced into the mechanical structure at each stator tooth, rotor pole and various axial positions individually. Similarly, the modelling of the bearings is investigated for a range of available methods. Sound emission is evaluated at the large outer surface structures like tower, blades and nacelle cover. To minimize computational effort, the surface accelerations are not calculated for each surface node, instead a modal approach is used. Through a combination of mode shapes with mode participation factors of the respective structures, the surface accelerations can be regained during a post-processing step. Those results are used as input for airborne sound calculations. Nevertheless, the high number of modal and spatial degrees of freedom results in high computing costs.

     

Reconstruction of operating loads in wind turbines with inertial measurement units

Wind turbines are a major source of renewable energy. Load monitoring is considered to improve reliability of the systems and to reduce the cost of operation. We propose a load monitoring system which consists of inertial measurement units. These track the movement of rotor blade, hub and tower top. In addition, wind turbine states, e.g. yaw angle, pitch angle and rotation speed, are recorded. By solving a navigation algorithm with a Kalman Filter approach, the raw sensor data is combined with an error model to reduce the tracking error. In total, five inertial measurement units are installed on the research wind energy converter AD 8–180 on the test site in Bremerhaven. Results show that tracking the blade movement in full operation is possible and that loads can be estimated with a model-based approach. In comparison to simulations, the blade deflections can be approximated by an aeroelastic model. The presented approach can be used as basis for comprehensive load monitoring and observer system with additional increase of system robustness by measurement redundancy.

     

Methodology for the systematic design of conical plain bearings for use as main bearings in wind turbines

With the possibility to replace sliding segments on the tower without disassembling the drivetrain, the use of segmented plain bearings with conical sliding surfaces as main bearing in wind turbines has a great potential to reduce the maintenance costs and thus the levelized cost of energy (LCOE). Furthermore, the short axial design leads to lower investment costs. Since this design is totally new and no design guidelines are available so far, the objective of this paper is to investigate the influence of the geometric parameters on the hydrodynamic pressure distribution of the bearing. In this context a parameter screening is performed using a suitable test field according to Plackett and Burman in order to determine the most relevant parameters. With the help of the simulations carried out after this test field, correlations between the geometric parameters and the hydrodynamic pressure distribution are evaluated. To be able to quantitatively analyze the three-dimensional pressure distribution, several key values are defined in this paper that describe the pressure distribution. The content of this paper is part of a methodology with the goal of developing a design guideline for conical plain bearings.

     

Identification of relevant acoustic transfer paths for WT drivetrains with an operational transfer path analysis

The aim of a transfer path analysis (TPA) is to view the transmission of vibrations in a mechanical system from the point of excitation over interface points to a reference point. For that matter, the Frequency Response Functions (FRF) of a system or the Transmissibility Matrix is determined and examined in conjunction with the interface forces at the transfer path. This paper will cover the application of an operational TPA for a wind turbine model. In doing so the path contribution of relevant transfer paths are made visible and can be optimized individually.

     

Advanced method of including housing stiffness into calculation of gear systems

One of the central goals during the design of helical gear systems is the achievement of a well-distributed contact load in the gear mesh. An equal load distribution is a key factor for a high load carrying capacity, the economic use of materials and a long lifetime. Mesh misalignment can be caused by tooth deflections, manufacturing deviations or elastic deformation of the shaft-bearing system and the gearbox housing. Those deformations have to be taken into account during the design process of adequate tooth-flank geometry. Elastic deformations of gearbox housings can be significant, especially in the case of automotive applications with aluminium cases. This paper presents an advanced method of including housing stiffness into the calculation of gear systems. A validation of the approach is carried out by comparing the calculated deformations with measurements of a static test rig of a hypoid gearbox.

Many calculation programs offer the opportunity to analyse the deformation behaviour of the shaft-bearing-housing system. Most of the components in these programs are described by analytic approaches. However, components that are geometrically more complex, like the housing or planet carriers cannot be represented as easily as that by analytic expressions. There are several alternatives to take into account the elasticity of those objects. One way is to model the stiffness of the bores using imported stiffness matrices. These matrices contain the elasticity of the bores itself as well as crossover influences between the bearings. The reduced stiffness matrices may be the result of a static reduction of the geometry using the finite element method (FEM). As state of the art, the reduction is mostly carried out at the centre points of the bearing bores. The proposed advanced method uses the static reduction of geometries on several points at the bores, distributed over the circumference. This approach offers a more detailed modelling of the elastic behaviour of complex geometries within the analytic deformation calculation of gear systems. To validate the advanced approach, the calculation results of the elastic deflections of the shaft-bearing-housing system is compared with measurements of a static test rig. In the course of these comparisons, the influence of different modelling methods of gearbox housings on the accuracy of the calculation results is discussed.

     

Usability of SCADA as predictive maintenance for wind turbines

Wind energy is an essential source of renewable energy. However, to compete with conventional energy sources, energy needs to be produced at low costs. An ideal situation would be to have no costly, unscheduled maintenance, preferably. Currently, O&M are half of the yearly expenses. The O&M costs are kept low by scheduled and reactive maintenance. An alternative is predictive maintenance. This method aims to act before any critical and costly repair is required. Additionally, the component is used to its full potential. However, such a strategy requires a damage indication, similar to one provided by a condition monitoring system (CMS). This paper investigates if Supervisory Control and Data Acquisition (SCADA) can be used as a damage indicator and CMS. Since 2006, every wind turbine is obliged to use such a SCADA-system. SCADA records a 10-minute average, maximum, minimum, and standard deviation of multiple technical information channels. Analytics can use those data to determine the normal behavior and a prediction model of the wind turbine. The authors investigated statistical and data mining methods to predict main bearing faults. The methods indicated a defect of up to 6 months before its maintenance.

     

Robust two-degrees of freedom linear quadratic gaussian position control for the front axle actuator of a steer-by-wire system

The Steer-by-Wire (SbW) system is a key technology for highly automated driving. For automated lateral vehicle guidance, the precise position control of the SbW Front Axle Actuator is an essential prerequisite. This contribution presents the modeling, control design, nominal performance, and stability analysis as well as the robustness analysis of the position control for the SbW Front Axle Actuator. Based on a nonlinear model of the plant a simplified linear system model is derived. This model yields the basis for the design of a Two-Degrees of Freedom Linear Quadratic Gaussian Control (2DOF LQG control), which allows an independent design of the command and the disturbance response. Besides an evaluation of the nominal control behavior, μ-analysis is applied to assess the robustness of performance and stability. Finally, real vehicle tests for different driving maneuvers are presented to verify simulation results.

     

Enhanced damage classification accuracy on a transmission by extending existing datasets with generative adversarial networks

In many areas of drive technology, condition monitoring of transmissions and drive systems is becoming an increasingly important discipline. Condition monitoring systems are used in many cases in combination with machine learning algorithms. The generation of a sufficient amount of data per condition class is relevant to ensure training stability and accuracy of the applied algorithms. Especially in early development phases a sufficient data generation is not often given. In the scope of this paper, a Generative Adversarial Network is applied to generate synthetic data and therefore extend existing measurement data sets. Acceleration data in three different condition classes is used, that has been collected on a gearbox as part of the PHM Data Challenge 2009. In order to highlight relevant features and reduce the number of data points, data is pre-processed via appropriate signal analysis techniques, in this case with the spectral kurtosis. It is shown, that in this use case the synthetically generated data via a Generative Adversarial Network has the same feature characteristics as the real measured data sets. The augmentation of the existing data set also improves the detection accuracy with artificial neural networks for the classification of different system states.

     

Early stage white etching crack identification using artificial neural networks

White Etching Cracks (WEC) in gearbox bearings is a major concern in the wind turbine industry, which can lead to a premature failure of the gearbox. Though many hypotheses regarding the generation of WEC have been proposed over the decades, the answer is still disputable. To trace back the failures to earlier stages before they occur, an innovative sensor-set has been utilized on a test rig to monitor the influencing factors that lead to WEC. This paperwork seeks to recognize abnormal patterns from recorded sensor data and derive statements of sensible sensor combinations in WEC early detection. A Long Short Term Memory (LSTM) network-based autoencoder is proposed for the anomaly detection (AD) task. Employing an auto-associative sequence-to-sequence predictor, a model is trained to reconstruct the normal time series data without WEC. The reconstruction error of testing time series data is evaluated for the determination of its anomaly. The results show that the specified LSTM autoencoder framework can qualitatively distinguish anomalies from collected multivariate time series data. Moreover, the anomaly score evaluated via reconstruction-error-based metrics can discriminate normal and abnormal behaviors in the study. This investigation’s results entail a significant step towards early WEC risk detection and more cost-efficient wind turbine technology if this approach can be further applied on stream data with plausible thresholds in monitoring system.

     

Sensor selection for cost-effective virtual torque measurements on a wind turbine gearbox

Forschung im Ingenieurwesen - Measuring the operational torques during the lifetime of a wind turbine gearbox offers a valuable source of information for design, monitoring,...     

Full-scale field test of a model predictive control system for a 3 MW wind turbine

Model predictive control (MPC) is a strong candidate for modern wind turbine control. While the design of model predictive wind turbine controllers in simulations has been extensively investigated in academic studies, the application of these controllers to real wind turbines reveals open research challenges. In this work, we focus on the validation of a linear time-variant MPC system for a 3 MW wind turbine in a full-scale field test. First, the study proves the MPC’s capability to control the real wind turbine in the partial load region. Compared to the turbine’s baseline PID controller, the MPC system offers similar results for the electrical power output and for the occurring mechanical loads. Second, the study validates a previously proposed, simulation-based rapid control prototyping process for a systematic MPC development. The systematic development process allows to completely design and parameterize the MPC system in a simulative environment independent of the real wind turbine. Through the rapid control prototyping process, the MPC commissioning in the wind turbine’s programmable logic controller can be realized within a few hours without any modifications required in the field. Thus, this study establishes the proof of concept for a linear time-variant MPC system for a 3 MW wind turbine in a full-scale field test and bridges the gap between the control design and field testing of MPC systems for wind turbines in the multi-megawatt range.

     

A probabilistic model for discretionary lane change proposals in highway driving situations

This contribution presents a novel probabilistic approach for the generation of discretionary lane change proposals with a focus on highway driving situations. The developed model is based on the quantification of the utility of driving lanes. It generates a lane change proposal if the current driving lane is unsatisfactory in the sense that the desired velocity of the automated vehicle is undershot because of a slow preceding vehicle. A driving simulator study was conducted to create a dataset for the optimization of the model parameters. The optimization goal is to accurately match the timings of the lane change intentions of all participants. Finally, the applicability of the model is shown on real data from a test vehicle.

     

Development of test methodologies for experimental lifetime investigations of tidal turbines

As a regenerative energy source, tidal energy can significantly contribute to greenhouse gas reduction, even though the potentially achievable energy output is lower than that of wind or solar energy. The decisive advantage of tidal turbines lies in the simply and reliably predictable energy output. However, their commercial use has so far been impeded by the fact that on the one hand complex mechanical systems are required to convert energy of tidal currents and on the other hand multi-axial loading conditions caused by turbulent ocean currents act on the turbine. For this reason, field tests on prototypes are an essential part of the development strategy to ensure operational reliability. However, in-field tests do not allow for accelerated lifetime testing, so that test bench experiments are becoming an increasingly important alternative. Today, established procedures for testing the turbines main bearings and gearing system are already available, both for setting up the required test configuration and for determining the corresponding test loads. However, the use of advanced calculation methods, such as the finite element method for stress calculation, requires a deep understanding of the examined components and hinders the transfer of the approaches to other components.

To simplify the process of test loads determination, a general methodology is presented, which relies exclusively on standardized empirical calculation rules. Doing this, fatigue equivalent loads can be determined for any component in a simple process. It was shown that the achieved reduction in complexity opens further potential for test acceleration, since several components can be tested simultaneously.

     

Temperature influence on involute gear measurements

Forschung im Ingenieurwesen - Involute gears are high-stress components used in a variety of drivetrain applications. Small manufacturing tolerances require low measurement uncertainties...     

Characterisation of the conveying effect of turned radial shaft seal counter-surfaces using a simplified hydrodynamic simulation model

A typical sealing system for rotating shafts consists of the radial shaft sealing ring (RSS), the lubricant and the shaft counter-surface (SCS) of the rotating shaft. The properties of the machined surface of the SCS have an impact on the sealing system. The structural pattern of the SCS influences the lubricant flow along the axial direction. In this paper, a simplified micro scale hydrodynamic simulation model is presented in order to study and determine the axial flow of the lubricant induced by the SCS of the sealing system, isolated from the effects induced by the seal, to allow for a rating of the shaft surface. The influence of the seal was neglected to allow for a simplified simulation. Simulated shaft surfaces corresponding to different machining parameters of machined SCS are used as input. These variants of SCS were created using a kinematic model which simulates an ideal surface machining process of the shaft. A micro scale hydrodynamic simulation model is used to investigate the influence of machining parameters on the lubricant flow along the axial direction across the tribo-contact. From this investigation, the connection between parameters applied for machining of the SCS and conveying effects can be estimated. The simulation model is also validated with experimental results of hard turned shafts of different machining parameters. Differences between manufactured real surfaces and kinematically simulated surfaces are the cause of deviations between the results.

     

Holistic simulation of wind turbines with fully aero-elastic and electrical model

Requirements for the design of wind turbines advance facing the challenges of a high content of renewable energy sources in the public grid. A high percentage of renewable energy weaken the grid and grid faults become more likely, which add additional loads on the wind turbine. Load calculations with aero-elastic models are standard for the design of wind turbines. Components of the electric system are usually roughly modeled in aero-elastic models and therefore the effect of detailed electrical models on the load calculations is unclear. A holistic wind turbine model is obtained, by combining an aero-elastic model and detailed electrical model into one co-simulation. The holistic model, representing a DFIG turbine is compared to a standard aero-elastic model for load calculations. It is shown that a detailed modelling of the electrical components e.g., generator, converter, and grid, have an influence on the results of load calculations. An analysis of low-voltage-ride-trough events during turbulent wind shows massive increase of loads on the drive train and effects the tower loads. Furthermore, the presented holistic model could be used to investigate different control approaches on the wind turbine dynamics and loads. This approach is applicable to the modelling of a holistic wind park to investigate interaction on the electrical level and simultaneously evaluate the loads on the wind turbine.