The symposium is sponsored by Vestas Wind Systems A/S and Aalborg University
|9.05-10.00||Carlo L. Bottasso , Politecnico di Milano||Integrated Active and Passive Load Mitigation in Wind Turbines||[slides]|
|10.15-10.45||J.W. van Wingerden, TUDelft||Wind turbine load reduction by rejecting the period loads||[slides]|
|10.45-11.15||Romeo Ortega, LSS/CNRS/SupÚlec||Adaptive Passivity-based Control for Maximum Power Extraction of Stand-alone Windmill Systems||[slides]|
|11.15-11.45||Torben Knudsen, Aalborg University||Detection of Wind Turbine Tower Oscillations Fore-Aft and Sideways||[slides]|
|13.00-14.00||Kathryn Johnson, Colorado School of Mines||The Big Picture: A guided discussion of the present and future of wind turbine control||[slides], [photo]|
|14.00-14.30||Jason Laks, University of Colorado at Boulder||LIDAR-Enabled Set-point Scheduling for Model Predictive Control of Wind Turbines||[slides]|
|15.00-15.30||Martin Evans, Univ. of Oxford/Vestas||Linear stochastic MPC under finitely supported multiplicative uncertainty||[slides]|
|15.30-16.00||Mohsen Soltani, Aalborg University||Lidar-based control of wind turbines||[slides]|
|16.00-16.30||Mahmood Mirzaei, DTU||Uncertainty Assessment and Robust Model Predictive Control of a Wind Turbine||[slides]|
|9.00-10.00||Michel Verhaegen, TUDelft||Cautious Data Driven Fault Detection and Isolation applied the Wind Turbine Benchmark||[slides]|
|10.15-10.45||Peter F. Odgaard, kk-electronics||Results of the Contributions to the Competition on Wind Turbine Fault Detection and Isolation||[slides]|
|10.45-11.15||Damien Castaignet, Ris°||Frequency-weighted model predictive control of trailing edge flaps. Test on a Vestas V27 turbine||[slides]|
|11.15-11.45||Rafael Wisniewski, Aalborg University||Compositional Safety Analys Using Barrier Certificates||[slides]|
|13.00-14.00||Morten H. Hansen, Ris°||Turbine aero-servo-elasticity||[slides]|
|14.00-14.30||Fabiano D. Adegas , Aalborg University||Structured Linear Parameter Varying Control of Wind Turbines||[slides]|
|15.00-15.30||Maryam Soleimanzadeh, Aalborg University||Controller design for a wind farm, considering both power and load aspects||[slides]|
|15.30-16.00||Daria Madjidian, Lund University||A turbine interaction model for choosing operating points in wind farms||[slides]|
|16.00-16.30||Maxim Kristalny , Lund University||Exploiting previewed information in distributed control of large-scale wind farms.||[slides]|
Load reduction techniques for wind turbines can be broadly categorized
into two families: active and passive. The former aims at
mitigating loads by actively controlling the machine (for
example by changing the blade pitch and the generator torque,
or by moving flaps and/or tabs or other active devices), while
the latter is based on the idea of designing a structure that,
when aerodynamically loaded, deforms so as to induce a load
Active individual blade pitch control (IPC) has been shown to be capable of substantially reducing loads, but at the cost of increased actuator duty cycle (ADC); this, in turn, calls for a redesign of the pitch actuator system to meet the increased demand of pitch activity.
On the other hand, passive load alleviation by bend-twist coupling (BTC), obtained by exploiting the anisotropic mechanical properties of composite materials, is very attractive for wind energy applications: in fact there are no actuators which may fail, no moving parts which may wear out, and no need for sensors, all characteristics that are very interesting whenever simplicity, low maintenance and high availability are key to reducing the cost of energy. Furthermore, BTC blades reduce ADC, since the blades self-react to turbulent fluctuations in the wind.
In this paper we explore the possible existing synergies between BTC and IPC. In fact, BTC and IPC can both mitigate loads, but BTC reduces ADC while IPC increases it. The combination of the two technologies can lead not only to improved load reduction performance, but also avoids excessive actuator activity thanks to the passive load control built into the blade design.
To decrease the cost per kWh, the trend in offshore wind turbines is to increase the rotor diameter as much as possible. The increasing dimensions have led to a relative increase of the loads on the wind turbine structure, thus it is necessary to react to disturbances in a more detailed way; for example, each blade separately. The disturbances acting on an individual wind turbine blade are to a large extent deterministic; for instance, tower shadow, wind shear, yawed error, and gravity are depending on the rotational speed and azimuth angle, and will change slowly over time. This paper aims to contribute to the development of individually pitch controlled blades and "smart" blades by proposing a lifted repetitive controller that can reject these periodic load disturbances for modern wind turbines operating above-rated. The performance of the repetitive control method is evaluated on the UPWIND 5MW wind turbine model and our proto-typed "smart" rotor.
This talk addresses the high performance regulation of stand-alone windmill systems consisting of a wind turbine coupled to a generator and a battery charging system, which is a challenging problem for at least two reasons. First, the dynamics of the overall system are described by a highly-coupled set of nonlinear differential equations. Since the range of operating points of the system is very wide, classical linear controllers yield below par performances. Second, in many applications it is desirable to extract from windmill systems their maximum power. This operating point is a nonlinear function of the wind speed, which is hard to measure. In this paper, a nonlinear passivity-based controller that ensures asymptotic convergence to the maximum power extraction point, which is rendered adaptive combining it with a wind speed estimator previously proposed by the authors, is proposed. Detailed computer simulations are presented to validate the approach.
Fatigue loads are important for the overall cost of energy from a wind turbine. Loading on the tower is one of the more important loads, as the tower is an expensive component. Consequently, it is important to detect tower loads, which are larger than necessary. This paper deals with both fore-aft and sideways tower oscillations. Methods for estimation of the amplitude and detection of the cause for vibrations are developed. Good results are demonstrated for real data from modern multi mega watt turbines. It is shown that large oscillations can be detected and that the method can discriminate between wind turbulence and unbalanced rotor.
Wind turbine control is changing rapidly, in large part due to the contributions of people attending this symposium. With audience participation, we will spend this hour discussing the needs, current status, and future directions for wind turbine control in a way that attempts to tie together the expert talks being given at the symposium. I will also present results from research in one future direction: the integration of wind turbines into the set of "active power control" devices for the utility grid.
Over the last five years there has been increasing interest in the use of model predictive control (MPC) in application to wind turbine operation. This technique explicitly handles constraints on control inputs and, because it is based on finite horizon optimization, it has natural extensions that handle time-varying objectives and the use of preview information. The latter feature makes it particularly well suited to incorporation of advanced measurement technologies like light detection and ranging (LIDAR). To our knowledge, there has not been any studies that investigate how MPC performance is affected by the choice of profile for transition between operating regions. In this study, we propose an MPC configuration for use in transitioning between regions 2 and 3, and that is made possible by the availability of preview measurements. More specifically, this study investigates the ability of the proposed MPC configuration to track several different transition profiles through the turbine's operating regions. In addition to dynamic load mitigation, LIDAR measurements are used to schedule the operating set-points along the candidate transition profiles according to wind speed.
Model predictive control in the presence of uncertainty and constraints is an active area of research, but most results available to date concern the case of additive uncertainty or apply constraints in expected value only. In addition, the conventional assumption that model uncertainty is normally distributed prevents the development of suitable guarantees of feasibility and therefore closed loop stability. These two shortcomings compromise the application of MPC to problems such as wind turbine control due to the non-Gaussian multiplicative uncertainties in the dynamics. The novelty in the current work concerns the construction of less conservative terminal sets, the relaxation of constraints through the use of prediction error feedback, and the handling of constraints through sampling and mixed integer programming.
This presentation addresses two issues:
First, a Lidar-based control scheme for wind turbines will be presented. Model-based receding horizon control is used to reduce the structural loads in the transmission system and the tower, as well as provide constant (or at least smooth) power generation. The method is verified against the existing wind turbine control system. The results show significant reductions in both extreme loads and power fluctuations, when compared to a conventional controller.
Next, the required front wind measurements for the above approach will be presented. We also talk about the assumptions in Lidar-based control and challenges or risks associated with those assumptions. Finally, we discuss possible Lidar solutions which are potentially able to provide required information.
In recent years size of wind turbines has been growing almost exponentially. However this trend is not expected to continue as there is a very important limit in the development, namely control systems. Model predictive control has proved beneficial in many application within process control industry and with new findings in the area of fast MPC, now it can be used to control systems with very fast sampling time. One short coming in nominal MPC though, is the assumption of accurate model which is never correct. One way to get around this problem is to employ robust MPC instead. However generally robust MPC problems are either too conservative or computationally too expensive. This difficulty could be resolved by employing robust MPC tailored to our specific problem. Minimax robust MPC approach is used in this work. Nonlinear dy- namics of the wind turbine is derived by combining blade element momentum (BEM) theory and first principle modeling of the flexible structure. Then the nonlinear model is linearized using Taylor series expansion around system op- erating points. As operating points are determined by effective wind speed, an extended Kalman filter (EKF) is employed to estimate it. In addition, a new measurement is introduced in the EKF to give faster estimations. Confidence interval of the estimation is used to assess uncertainties in the linearized model. Significant uncertainties are identified to be in the gain of the system (B matrix of the state space model). Therefore this special structure of the uncertain sys- tem is employed and norm-bounded uncertainty model is used to formulate a minimax model predictive control. The resulting optimization problem is sim- plified by semidefinite relaxation. And the obtained controller is applied on a full complexity, high fidelity wind turbine model. Finally simulation results are presented. First a comparison between PI and robust MPC is given. Af- terwards simulations are done for a realization of turbulent wind with uniform profile based on IEC standard.
Models for Fault Detection and Isolation (FDI) for windturbines may be derived by linearization of nonlinear dynamic first principles models in particular operating points. The description of linearization errors as perturbations and model uncertainty is a difficult issue. To avoid these difficult modeling procedures, we present and illustrate in this lecture a data-driven design method. This method directly produces FDI filters based on data records from the actual system in a fault free operating condition. It will be outlined in this lecture how to cope with the uncertainty in these data driven FDI filters. The derivation of these FDI filters is based on insights borrowed from Subspace identification and are therefore referred to FICSI (Fault Identification connected to Subspace Ide). The novel procedure is illustrated on the benchmark problem presented in Odgaard, et. al, 2009.
Improved reliability of wind turbines are one of key factors to
decrease the cost of energy generated by wind turbines. Reliability
can be addressed by a number of factors of which some are within the
area of control engineering, the most relevant control engineering
areas are Fault Detection and Isolation (FDI) as well as Fault
Tolerant Control (FTC).
In order to draw some research attention to these topic a benchmark model on FDI and FTC of wind turbines has been proposed, and an international competition on the FDI and FTC parts of the benchmark model have been setup. Contributions to the FDI part were presented at the IFAC World Congress 2011 and the results on the FTC part will be presented at IFAC Safeprocess 2012.
In this presentation some of best results of the first part of the competition will be presented together with evaluation results of them on a more detailed model considering additional operational points of the faults as well.
Trailing Edge Flaps on wind turbine blades have been studied in order to achieve fatigue load reduction on the turbine components. We show in this paper how Model Predictive Control can be used to do frequency weighted control of the trailing edge flaps in order to reduce fatigue damage on the blade root. The design model is based on a modal model of the blade structure and a steady state aerodynamic model of the blade airfoils. Depending on the output filter, loads within different frequency range are decreased. A fine tuning of the Kalman filter and of the cost function allows to decrease significantly the blade root loads without damaging excessively the trailing edge flap actuators. Some results from the field test on a V27 wind turbine equipped with a trailing edge flap are also shown.
Reference: Model Predictive Control of Trailing Edge Flaps on a Wind Turbine blade, by Damien Castaignet, Niels K. Poulsen, Thomas Buhl and Jens Jakob Wedel-Heinen, presented at 2011 American Control Conference
This paper proposes a compositional method for verifying the safety of
a system, given as an interconnection of sub- systems. The safety
verification is conducted by the use of barrier certificates; hence,
the contribution of this paper is to show how to obtain compositional
conditions for safety verification.
We show how to formulate the verification problem, as a composition of coupled subproblems, each given for one sub- system. Furthermore, we show how to find the composi- tional barrier certificates via linear and sum of squares pro- grams.
The proposed method makes it possible to verify the safety of higher dimensional systems, than the method for cen- trally computed barrier certificates. This is demonstrated by verifying the safety of an emergency shutdown of a wind turbine.
Modal dynamics of wind turbines are reflected in the time simulations of wind turbines' open- and closed-loop aeroservoelastic response and affect the turbine loads and stability limits. This lecture will first give an introduction to the modal dynamics of the most common three-bladed turbine concept including the structural standstill modes and turbine modes under operation, hereunder the theoretical background of the Coleman transformation as a special case of the Lyapunov-Floquet transformation to eliminate the periodic terms in the system equations enabling computation of the operational turbine modes by eigenvalue analysis. Then, the effects of the aerodynamic forces on the coupled aeroelastic turbine modes are discussed in terms of stability limits and possible aeroelastic instabilities. Finally, open- and closed-loop aeroservoelastic response of a wind turbine with a simple collective and cyclic pitch controller is considered to show the potentials of the new code HAWCStab2 that contains a linear aeroservoelastic model derived analytically from the advanced nonlinear code HAWC2 used for load assessments in wind turbine developments and certifications.
Reference: Aeroelastic Instability Problems for Wind Turbines, by Morten H. Hansen, in Wind Energy 2007; 10
The modeling and controller design procedures presented in this talk are candidates for solving a majority of practical wind turbine control problems. A wind turbine along with common faults and aerodynamic uncertainties is represented as a linear parameter varying (LPV) system. An LPV controller is synthesized tolerant to faults and robust to uncertainties by solving an optimization problem subject to linear matrix inequality (LMI) constraints. The controller structure can be chosen arbitrarily: decentralized, dynamic (full or reduced-order) output feedback, static output feedback are among the possible structures. The numerical example covers the design of a proportional-integral (PI) pitch controller with an active tower damping feedback loop, both gain-scheduled and tolerant to faults on the pitch system. Simulation results of LPV controllers intolerant and tolerant to pitch actuator faults are compared to support a discussion of the consequences of the fault on the closed-loop system as well as fault accommodation.
In this work that we are going to present, a wind farm controller is developed that distributes power references among wind turbines while it reduces their structural loads. The proposed controller is based on a spatially discrete model of the farm, which delivers an approximation of wind speed in the vicinity of each wind turbine. The control algorithm determines the reference signals for each individual wind turbine controller in two scenarios based on low and high wind speed. In low wind speed, the reference signals for rotor speed are adjusted, taking the trade-off between power maximization and load minimization into account. In high wind speed, the power and pitch reference signals are determined while structural loads are minimized. To the best of authors' knowledge, the proposed dynamical model is a suitable framework for control, since it provides a dynamic structure for behavior of the flow in wind farms. Moreover, the controller has been proven exceptionally useful in solving the problem of both power and load optimization on the basis of this model.
Turbines operating in wind farms interact aerodynamically. This interaction causes (an often substantial) deterioration in both power production and fatigue loading at downwind turbines. In order to operate wind farms cost-effectively it is important to understand and address this issue. To this end we introduce a stationary turbine interacton model with a simple intuitive structure. The model maps wind speeds and thrust coefficients of upwind turbines to winds speeds at downwind turbines. Moreover, the wind speed at a downwind turbine is completely determined from information available at its closest upwind neighbor. This makes the model well suited for distributed control algorithms. In an example we show that the total power production of the farm can be increased by coordinating the power production of individual turbines. We also show that the benefits of coordination become larger as the number of turbines in the farm increases, and discuss the mechanisms behind this property.
The current work concerns with the development of techniques for the control of turbines located in large-scale wind farms. Although nowadays turbines in the farms are controlled independently, taking the neighboring turbines into account might be advantageous: 1. The downwind turbines can use their upwind neighbors as sensors, getting access to previewed wind speed measurements; 2. Cooperation between turbines in terms of power production may facilitate loads mitigation by allowing adjustment of individual turbines powers. In this talk I will address these issues within the framework of distributed feedforward control. Potential of this control strategy will be discussed and a number of open problems associated with it will be presented. The first part of the talk will focus on the idea of exploiting previewed wind measurements for the control of an individual turbine. It will be shown that the problem can be conveniently formulated as an H2 model matching optimization with preview and asymptotic behavior constraints. A recent solution of this problem will be presented and followed by simulation results. The second part of the talk will focus on the idea of exploiting possible cooperation between turbines. It will be shown that under a mild simplifying assumption the problem can be formulated as a decentralized model-matching optimization. Challenges associated with this problem will be outlined and possible remedies will be discussed.