Search

New wind turbine technologies and designs are being explored in order to reduce the cost of energy from offshore wind farms. Two potential routes to a lower cost of energy are the X- Rotor Concept (XRC) and Multi-Rotor System (MRS) turbines. A key cost saving for both Novel concepts included in this paper is operation and maintenance (O&M) costs savings. The major component replacement cost for conventional horizontal axis, XRC and MRS turbines are examined and the benefits of the concepts are provided in this paper. A review on existing decision support systems for offshore wind farm O&M planning is presented with a focus on how applicable these previous models are to novel turbine concepts, along with analysis of how the influential factors can be modified to effectively model XRC and MRS.

A hierarchical Wind Farm Control (WFC) approach was previously developed that uses Power Adjusting Controllers (PACs) on each wind turbine in a wind farm. The PACs can be retrofitted to existing assets with no knowledge of, or change to, the wind turbine full envelope controller (FEC). However, knowledge of the wind turbine aerodynamics is required and is not usually directly available from the Original Equipment Manufacturer (OEM), necessitating estimation. In this letter, estimated aerodynamic properties are obtained via a scanning LIDAR that directly measures the shape of a 2.5MW commercial wind turbine’s blades. The impact of the resulting aerodynamic uncertainty on the PAC tuning and the accuracy of the change in power output from the PAC at a turbine level and at a wind farm level is assessed. It is shown that it is possible to tune a stable PAC using aerodynamic information estimated via blade scanning. Although the requested turbine change in power suffers from some inaccuracy, the slow integral action at a WFC level causes the impact on the accuracy of the change in wind farm power output to be negligible. As such, the application of a WFC methodology utilising PACs without prior knowledge of the turbine aerodynamics is shown to be possible by using blade scanning to estimate the aerodynamic coefficients. Hence it is practical to retrofit the methodology to wind farms when aerodynamic information from the OEM is not available.

The quantity and size of wind farms continue to grow as countries around the world strive to meet ambitious targets for renewable electricity generation such as the UK government’s Net Zero target of increasing offshore wind energy from current levels (circa 6 GW) to circa 75 GW by 2050. With increasing size and quantity of wind farms, there is a growing requirement to use wind farm level control both to help with grid integration and to minimise the loads on the turbines in the farm. In this paper, a methodology of distributing power set points through a wind farm to minimise the loads on the turbines whilst meeting a delta power set point for the farm is presented. The methodology in this paper uses a hierarchical control structure, in which a network wind farm controller calculates the required change in wind farm power and then passes this value on to a distributed controller that defines the change in power required from each wind turbine. The network wind farm controller calculates a delta change in wind farm power that the wind farm holds in reserve. The distributed controller allocates the reductions in power output by first setting a baseline reduction that considers the steady state tower loads. The baseline is then adjusted to meet the required change in power, distributing the additional change in one of two ways; either proportional to the square of each turbines estimated wind speed or proportional to the initial baseline. Performance is assessed using the StrathFarm simulation tool. The wind turbine models incorporated into StrathFarm are sufficiently detailed to provide the tower and blade loads and the wind field model is sufficiently detailed to represent turbulence, wind shear, tower shadow and wakes and their interaction. The performance of the proposed wind farm controllers are assessed for a range of wind conditions for two 4x4 wind farms of 5MW wind turbines, one closely spaced (500m) and one less closely spaced (1000m). Both the accuracy of the change in power output from the wind farm and the change in turbines DELs are discussed. Depending on the wind conditions, the approach is found to reduce the tower and blade loads by about 10% more than the case in which each turbine is simply allocated the same change in power. There is good accuracy in the change in power at higher wind speeds. Below rated wind speed, wake effects reduce the accuracy of the change in power.

“Blue Growth” and “Blue Economy” is defined by the World Bank as: “the sustainable use of ocean resources for economic growth, improved livelihoods and jobs, while preserving the health of ocean ecosystem”. Multi-purpose platforms (MPPs) can be defined as offshore platforms serving the needs of multiple offshore industries (energy and aquaculture), aim at exploiting the synergies and managing the tensions arising when closely co-locating systems from these industries. Despite a number of previous projects aimed at assessing, from a multidisciplinary point of view, the feasibility of multipurpose platforms, it is here shown that the state-of-the-art has focused mainly on single-purpose devices, and adopting a single discipline (either economic, or social, or technological, or environmental) approach. Therefore, the aim of the present study is to provide a multidisciplinary state of the art review on, whenever possible, multi-purpose platforms, complementing it with single-purpose and/or single discipline literature reviews when not possible. Synoptic tables are provided, giving an overview of the multi-purpose platform concepts investigated, the numerical approaches adopted, and a comprehensive snapshot classifying the references discussed by industry (offshore renewables, aquaculture, both) and by aspect (technological, environmental, socio-economic). The majority of the multi-purpose platform concepts proposed are integrating only multiple offshore renewable energy devices (e.g. hybrid wind-wave), with only few integrating also aquaculture systems. MPPs have significant potential in economizing CAPEX and operational costs for the offshore energy and aquaculture industry by means of concerted spatial planning and sharing of infrastructure.

Renewable energy sources are foreseen as a provider of full range of ancillary services. An innovative concept of alignment between renewable power generation elements and synchronous generators is proposed: Renewables Association with Synchronous generators (RAS). It mitigates the dependence on direct frequency measurements, which are prone to noise and lack of accuracy, and enables perfect coordination between the responses of renewable and conventional power plants. RAS relies on a leader synchronous generator, connected at the point of common coupling of the renewable power plant or close to it. This synchronous generator is able to provide ancillary services (e.g. frequency support and reactive compensation). The renewable power plant is controlled to provide such services similar to the leader synchronous generator, but scaled down/up to match the rating of the renewable power plant by integrating supplementary controllers that are associated with the synchronous generator response. Two approaches are proposed to provide voltage support, besides a supplementary frequency support controller. These RAS-based voltage and frequency support methods are compared to other methods proposed in the literature. Results show the positive impact of RAS concept on the provision of active power and reactive compensation to tackle frequency and voltage events respectively, following the response of the leader synchronous generator. DIgSILENT PowerFactory is the applied simulation environment.

A gain-scheduled feedforward controller, based on pseudo-LIDAR (light detection and ranging) wind speed measurement, is designed to augment the baseline feedback controller for wind turbine′s load reduction in above rated operation. The pseudo-LIDAR measurement data are generated from a commercial software – Bladed using a designed sampling strategy. The nonlinear wind turbine model has been simplified and linearised at a set of equilibrium operating points. The feedforward controller is firstly developed based on a linearised model at an above rated wind speed, and then expanded to the full above rated operational envelope by employing gain scheduling strategy. The combined feedforward and baseline feedback control is simulated on a 5 MW industrial wind turbine model. Simulation studies demonstrate that the proposed control strategy can improve the rotor and tower load reduction performance for large wind turbines.

Modelling and performance analysis of wind turbine control systems with aeroelastic tailoring blades (ATBs) are investigated. An industrial scale horizontal axis wind turbine (HAWT) model with rigid blades is firstly developed as the baseline model using the blade element momentum (BEM) theory. Designed twist angle variation distributions along blades are then introduced to the baseline model to characterise the ATB nature. The developed ATB wind turbine models are analysed by employing a baseline control system. The performances of the ATB wind turbine systems are compared with that of the baseline turbine using both nonlinear models and linearised models at selected wind speeds. The impacts of ATB design can be clearly observed from the simulation studies. Preliminary results suggest that with ATB design in wind turbines, the blade fatigue loads and the pitching activities can be reduced for large turbines without compromising the energy capture performance.

A process for optimising both the design and operation of the generator for a large offshore vertical axis wind turbine (VAWT) is developed. The objectives of the optimisation process are to minimise additional costs and losses in the generator to allow for a fair evaluation of the impact of the VAWT environment on the powertrain. A spectrum of torque control strategies was tested based on the ratio, q, of the allowed electrical torque variation to the inherent mechanical torque variation. Equations relating q to the generator losses were established. The effect of q on the energy extracted by the rotor was also investigated and incorporated into the optimisation process. This work shows that a variable q strategy with respect to wind speed can improve turbine performance across the range of operational wind speeds depending on the torque loading from the rotor blades. In turn, this also allows for the torque rating of the generator to be reduced from the peak torque rating that would otherwise be expected, creating an opportunity to downscale the generator size, reducing costs. The optimisation of powertrain design and operation should be carried out at as high level as is possible, ideally using the fully factored Cost of Energy (COE) to guard against unexpected losses due to excessive focus in one COE factor (for example reducing upfront cost but in turn reducing availability).

Model predictive and linear quadratic Gaussian controllers are designed for a 5MW variable-speed pitch-regulated wind turbine for three operating points – below rated wind speed, just above rated wind speed, and above rated wind speed. The controllers are designed based on two different linear dynamic models (at each operating point) of the same wind turbine to study the effect of utilising different control design models (i.e. the model used for designing a model-based controller) on the control performance. The performance of the LQG controller is enhanced by improving the robustness, achieved by replacing the Kalman filter with a modified Luenberger observer, whose gain is obtained to minimise the effect of uncertainty and disturbance.

This paper presents an optimised yaw algorithm for the novel yaw system whereby the thrust of individual rotors of a Multi Rotor System (MRS) are varied in order achieve yaw torque. The control system is built upon a previous paper where a non-optimised yaw algorithm is shown to be capable of providing stable yaw control for an MRS. The optimised yaw algorithm provides increased energy capture and improved performance compared to that of the non-optimised system. The MRS model is also developed further from the previous paper which simplifies the MRS by clustering together groups of rotor and power conversion (RPC) systems. All 45 RPC systems are modelled to more accurately represent the MRS and to allow feedback from each RPC system. Both an estimated wind speed and system of flags are fed back to the optimised yaw algorithm to calculate the RPC systems with greatest potential change in yaw torque and to ensure that each RPC system does not operate outside a defined safe operational envelope. The performance of the energy capture for the optimised yaw controller is analysed and compared to the basic non-optimised controller. The optimised controller shows improved yaw control at each wind speed. The reduction in energy capture for the non-optimised controller is 0.61 %, 1.9% and 0.89% of the total energy captured for 8m/s, 11m/s and 15m/s respectively. This compares to 0.45%, 0.67% and 0.66% for the optimised controller.

This project presents a holistic wind farm control approach that enables wind power plant to provide the full range of ancillary services including synthetic inertia at the wind farm level rather than single turbine level. In order to detect a power system event and select the magnitude of the service provision from the wind farm, a fully instrumented small/medium generator is used. By slaving the wind farm output to the generator natural response during power system events, the wind farm is able to provide a stable scaled-up range of ancillary services without relying in delayed or noisy grid frequency measurements.

In this paper it is shown that measured data in a wind turbine, available to the controller, can be formulated into a polynomial regression problem in order to estimate the turbine's maximum efficiency power coefficient, C p,max, and drivetrain losses, assuming the latter can be well approximated as being linear. Gaussian process (GP) machine learning is used for the regression problem. These formulations are tested on data generated using the Supergen Exemplar 5 MW wind turbine model, with results indicating that this is a potential low cost method for detecting changes in aerodynamic efficiency and drivetrain losses. The GP approach is benchmarked against standard least-squares (LS) regression, with the GP shown to be the superior method in this case.

When the installed capacity of wind power becomes high, the power generated by wind farms can no longer simply be that dictated by the wind speed.With sufficiently high penetration, it will be necessary for wind farms to provide assistance with supply-demand matching. The work presented here introduces a wind farm controller that regulates the power generated by the wind farm to match the grid requirements by causing the power generated by each turbine to be adjusted. Further benefits include fast response to reach the wind farm power demanded, flexibility, little fluctuation in the wind farm power output and provision of synthetic inertia.

To investigate potential improvement in wind turbine control employing LIDAR measurement, pseudo-LIDAR wind speed data is produced with Bladed using a designed sampling strategy, and assessed with preliminary frequency-domain analysis. A model-inverse feed-forward controller is adapted to combine with feedback control so as to enhance pitch control performance at high wind speed. This controller is applied to an industrial-scale 5MW wind turbine model and the control performance is compared with a baseline feedback controller. Simulation study demonstrates that the combined feedforward/feedback control scheme has improvements in reducing pitch angle variation and reduction of load relevant metrics.

This paper presents a fast selective harmonic current mitigation strategy for inverters with active power filter capabilities based on synchronous reference frames and two degrees-of-freedom internal model controllers. The advantage of this control strategy over the conventional proportional integral (PI) control solution is a significant increase in the speed of harmonic detection and mitigation. Furthermore, this control strategy reduces the computational burden when applied in a digital controller. These characteristics make this strategy desirable for applications, where fast/harmonic detection and mitigation are needed. Mathematical analysis and simulations are presented to corroborate the performance of the proposed controller strategy. Finally, the results of this proposal are verified in a 1 kW three-phase multifunctional inverter with harmonic compensation capabilities up to the 17th harmonic.

Wind energy sector has gained the highest attention among all other renewable energy resources. The UK government has a target of 15% of its energy to be produced from renewable sources [1]. In 2016 wind energy contributed up to 10% to the UK’s total electricity supply [2]. Over the last 30 years wind technology has improved, and various wind turbine concepts and generators have been developed and enhanced. The increasing use of wind energy in the UK grid imposed the requirements for wind farms to comply with current Grid Codes as conventional power plants and contribute to the network support and operation. The future of the large scale wind farms may lay offshore and comprehensive studies in terms of controllability and reliability are required [3, 4]. Nowadays the leading generator technologies in large wind turbine development are the double-fed induction generators (DFIG) and the fully rated converter (FRC) [5]. From the controlled development point of view, both technologies share the same mechanical system dynamics, and similar controllers can be applied to each technology to attain control over the speed of the turbine. However, during AC faults, the dynamics of each wind turbine technology are different due to the distinct fault response nature of each machine as well as the different type of power conversion interface to the ac grid. During a fault condition the power electronics of the DFIG wind turbine are disconnected in order to protect partially-rated converter and control over the machine is lost. The DFIG also acts as an induction generator and consumes reactive power which does not comply with fault ride through capabilities. This work analyses the behavior of the DFIG and FRC under voltage dip. The two technologies are combined into a hybrid wind plant and a control method is proposed to support DFIG wind turbine during voltage dip by providing reactive power via FRC wind turbine. Fig. 1 shows simplified power system configuration of hybrid wind farm.

Anomalies in the wind field and structural anomalies can cause unbalanced loads on the components and structure of a wind turbine. For example, large unbalanced rotor loads could arise from blades sweeping through low level jets resulting in wind shear, which is an example of anomaly. The lifespan of the blades could be increased if wind shear can be detected and appropriately compensated. The work presented in this paper proposes a novel anomaly detection and compensation scheme based on the Extended Kalman Filter. Simulation results are presented demonstrating that it can successfully be used to facilitate the early detection of various anomalous conditions, including wind shear, mass imbalance, aerodynamic imbalance and extreme gusts, and also that the wind turbine controllers can subsequently be modified to take appropriate diagnostic action to compensate for such anomalous conditions.

Wood-pole overhead conductor lines are known to be subject to different wind induced motions, some which lead to conductor fatigue over long periods and others that can cause severe damage over short, isolated incidences. These lines are often located in rural areas and, with the ever increasing deployment of onshore wind farms, the potential impact of the wind turbine wakes on these conductor motions is a matter of interest. Discussion of the potential effects of conductor fatigue on wood-pole overhead conductor lines is important as this could impact the planning, modelling and siting of onshore wind farms. This paper presents a literature analysis of both the mechanism by which different conductor motions are excited and wind turbine wake characteristics with specific interest at low levels akin to wood-pole height. The archival value of this paper is that experimental data from an active wind farm, at wood-pole overhead line heights, is analysed and discussed for the first time and the results show that wake effects at these heights are likely to be minimal. However turbulent buffeting in particular is cited as a problem of interest and the potential effects of wood-pole motions are outlined.

The following paper provides an overview of a novel wind turbine concept known as the X-Rotor Offshore Wind Turbine. The X-Rotor is a new wind turbine concept that aims to reduce the cost of energy from offshore wind. Cost reductions are achieved through reduced capital costs and reduced maintenance costs. The following paper includes results from an early feasibility study completed on the concept. In the feasibility study exemplary designs were created and structural analyses were carried out. Turbine capital costs and maintenance cost of the X-Rotor concept were then roughly estimated. X-Rotor turbine costs and O&M costs were compared to four existing wind turbine types to investigate potential cost savings from the X-Rotor concept. Results show that the X-Rotor has potential to reduce O&M costs by up to 55% and capital costs by up to 32%. The combination of the capital cost and O&M cost savings show potential to reduce the CoE by up to 26%.

This paper investigates the effect of wind flow direction on the fatigue loads on wind turbine within a wind farm, to determine the most critically loaded machines within the wind farm. The fatigue loads are calculated according IEC standards. In this study, the two layouts are considered. Additionally, this paper includes the effect of wind flow direction on the power efficiency of the wind farm for the two layouts. The simulations performed by StrathFarm, which is an in-house developed wind farm modeling tool, for the below and above rated wind speed flow with turbulence. In the regular and installed regular layouts at zeros degree of the prevailing wind flow direction, the spacing among turbines is eight diameters. At 90 degrees of wind flow direction the spacing between turbines is five diameters. The simulations demonstrate that the majority of the maximum fatigue loads occur at the range 40 and 70 degrees for the two layouts. However, there is small number of machines, which are experienced the highest fatigue loads at 90 degrees of wind flow direction. For the two layout the lowest power efficiency occurs at 10 degrees of wind flow direction. The regular layout is experienced same power efficiency at five and eight diameters spacing among turbine in the wind flow direction. The highest power efficiency occurs at wind flow angles, which produce the highest fatigue loads.

Lifetime extension is receiving increasing attention because of ageing asset bases, the need for efficient use of capital budgets, and the optimistic lifetime assumptions used at the project design stage. Based on the industrial attention and the overall observable reduction in subsidies for new investments and repowering, especially in onshore wind energy, lifetime extension is expected to become essential in the future. This contribution presents a methodology for life extension assessment of individual onshore wind turbine towers, as the key structural components, using the joint aeroeastic-finite element analysis and taking account of wind directionality and stress magnification around the tower door. The results demonstrate that the spread in the wind rose provide the potential for tower lifetime extension, however, the stress concentration around the tower door and site-wide variations of wind characteristics have to be also taken into account. The outcomes of this paper indicates that the wind rose dispersion can also provide a lifetime extension potential in addition to a more benign weather and operational conditions.

The present thinking and trend for connection of large offshore wind farms, dispersed over wide areas, is to use multi-terminal HVDC networks rather than point-to-point DC transmission systems. The aim behind this approach is to improve the security of supply and minimise the loss of generation during scheduled maintenance or unexpected disturbances in any part of the power network. This paper describes various models of multi-terminal HVDC networks connecting offshore wind farms to a number of mainland AC grids which have been developed in MATLAB-Simulink with the main objective of facilitating numerous studies such as steady-state power flow, optimal power dispatch analysis, transient stability, and provision of ancillaries.

The integration of large-scale offshore wind farms into electricity networks without causing stability hazards is one of the future challenges for interconnecting large offshore wind farms into national grids. This paper presents enhanced control strategies for offshore wind farm arrays interconnected via an HVDC link. To improve fault ride- through capabilities of the HVDC link and wind farms, a frequency controller is proposed and the third harmonic injection technique is applied as a means to improve the reliability of the offshore arrays. This new controller can easily coordinate the power flow from different sources (i.e. large-scale wind farms and conventional power stations). The power production from the offshore wind farms will depend upon the type of wind turbines installed. Thus, the performance of wind farms based on DFIG and FRC wint turbines is compared aiming to give a better understanding and to identify areas where control improvements can be introduced to optimise offshore wind power transmission.

This paper presents a novel model-based detection scheme capable of detecting and diagnosing gusts. Detection is achieved by monitoring the innovations error (i.e., the difference between the estimated and measured outputs) of an extended discrete Kalman filter. It is designed to trigger a detection/confirmation alarm in the presence of wind anomalies. Simulation results are presented to demonstrate that both operating and coherent extreme wind gusts can successfully be detected. The wind anomaly is identified in magnitude and shape through maximum likelihood ratio and goodness of fit, respectively. The detector is capable of isolating extreme wind gusts before the turbine over speeds.

A gain-scheduled feedforward controller employing pseudo-LIDAR wind measurement is designed to augment the baseline feedback controller for wind turbine load reduction during above rated operation. The feedforward controller is firstly designed based on a linearised wind turbine model at one specific wind speed, then expanded for full above rated operational envelope with gain scheduling. The wind evolution model is established using the pseudo-LIDAR measurement data which is generated from Bladed using a designed sampling strategy. The combined feedforward and baseline control system is simulated on a 5MW industrial wind turbine model developed at Strathclyde University. Simulation results demonstrate that the gain scheduling feedforward control strategy can improve the rotor and tower load reduction performance for large wind turbines.

This paper introduces two novel metrics to judge the capability and influence of wind power to provide virtual inertia response (i.e. frequency support). The first metric considers the generation unit (i.e. wind turbine generator (WTG)/wind farm (WF) vs. synchronous generator). This metric is applied to compare between three different methods of provision of frequency support. The second metric assess the improvement or hindering in frequency response at the point of common coupling (PCC) between a WF and a synchronous area. This metric is critical especially to WFs that are connected via High Voltage Direct Current (HVDC) or Low-frequency AC links. Both metrics are universal so that they could be applied to any support method, and any power system. The first metric is applied to assess the virtual inertia response of an offshore WF, which is considered as a power plant along with the HVDC transmission link. Results assure the positive impact of the provision of frequency support by wind power. This impact is quantified could be used to tune frequency support controllers, and optimize system planning. It is verified that no obstacles are implied by the HVDC link to integrating frequency support methods, as the WF dominates the support process.

A step-by-step coordinated design procedure for power system stabilisers (PSSs) and automatic voltage regulators (AVRs) in a strongly coupled system is described in this paper. It is shown that it is possible to separate the design of individual PSSs and separate the design of individual AVRs. Thereby, the designs of AVR and PSS devices at a given machine can be coordinated to achieve near-optimal overall power system stability performance, including oscillation stability performance and transient stability performance. The proposed coordinated PSS/AVR design procedure is established within a frequency-domain framework and serves as a most useful small-signal complement to established large-signal transient simulation studies.

Maximum likelihood estimation (MLE) of hyperparameters in Gaussian process regression as well as other computational models usually and frequently requires the evaluation of the logarithm of the determinant of a positive-definite matrix (denoted by C hereafter). In general, the exact computation of logdetC is of O(N3) operations where N is the matrix dimension. The approximation of logdetC could be developed with O(N2) operations based on power-series expansion and randomized trace estimator. In this paper, the accuracy and effectiveness of using uniformly distributed seeds for logdetC approximation are investigated. The research shows that uniform-seed based approximation is an equally good alternative to Gaussian-seed based approximation, having slightly better approximation accuracy and smaller variance. Gaussian process regression examples also substantiate the effectiveness of such a uniform-seed based log-det approximation scheme.

This paper describes the modelling and performance effects of induction generator-based wind farms on the oscillation stability of power systems. For groups of 5, 10 and 20 induction generators, it is established that the basic dynamic properties of each group of generators are similar to those of a third-order equivalent machine model whether the generators are evenly loaded or unevenly loaded. The paper analyses the impact of induction-based wind generation on oscillation stability for multi-machine systems as applied to the standard IEEE 30 bus system. The effects of load model representation, system loading, generator output and transmission line network parameters are considered. In general, the multi-machine system is shown to be well damped under all normal operating conditions.

The power converter is one of the most vulnerable components of a wind turbine. When the converter of an offshore wind turbine malfunctions, it could be difficult to resolve due to poor accessibility. A turbine generally has a dedicated controller that regulates its operation. In this paper, a collective control approach that allows a cluster of turbines to share a single converter, hence a single controller, that could be placed in a more accessible location. The resulting simplified turbines are constant-speed stall-regulated with standard asynchronous generators. Each cluster is connected by a mini-AC network, whose frequency can be varied through a centralised AC–DC–AC power converter. Potential benefits include improved reliability of each turbine due to simplification of the turbines and enhanced profit owing to improved accessibility. A cluster of 5 turbines is assessed compared to the situation with each turbine having its own converter. A collective control strategy that acts in response to the poorest control is proposed, as opposed to acting in response to the average control. The strategy is applied to a cluster model, and simulation results demonstrate that the control strategy could be more cost-effective than each turbine having its own converter, especially with optimal rotor design.

Two tradeoffs in the effectiveness of automatic voltage regulators (AVRs) and power system stabilizers (PSSs) are investigated together for the first time. The first is the effect of a high-gain fast response AVR on decreasing power system oscillation stability as well as increasing transient stability, and vice versa. The second is that a PSS can reduce transient stability by overriding the voltage signal to the exciter as well as increasing oscillation stability, and vice versa. In essence, the actions of the AVR and PSS devices are dynamically interlinked. A novel Bode frequency response framework for dynamic analysis of AVR and PSS performance and tradeoffs is presented. Bode frequency response also assists with the determination of suitable generator locations for PSSs and the assessment of robustness under changing power system operating conditions.

Maximum likelihood estimation of hyperparameters in Gaussian processes (GPs) as well as other spatial regression models usually requires the evaluation of the logarithm of the matrix determinant, in short, log det. When using matrix decomposition techniques, the exact implementation of log det is of O(N 3) operations, where N is the matrix dimension. In this paper, a power-series expansion-based framework is presented for approximating the log det of general positive-definite matrices. Three novel compensation schemes are proposed to further improve the approximation accuracy and computational efficiency. The proposed log det approximation requires only 50N 2 operations. The theoretical analysis is substantiated by a large number of numerical experiments, including tests on randomly generated positive-definite matrices, randomly generated covariance matrices, and sequences of covariance matrices generated online in two GP regression examples. The average approximation error is ∼9%.

—The main objective of this work is to explore the feasibility of using LIght Detection And Ranging (LIDAR) measurement and develop feedforward control strategy to improve wind turbine operation. Firstly the Pseudo LIDAR measurement data is produced using software package GH Bladed across a distance from the turbine to the wind measurement points. Next the transfer function representing the evolution of wind speed is developed. Based on this wind evolution model, a model-inverse feedforward control strategy is employed for the pitch control at above-rated wind conditions, in which LIDAR measured wind speed is fed into the feedforward. Finally the baseline feedback controller is augmented by the developed feedforward control. This control system is developed based on a Supergen 5MW wind turbine model linearised at the operating point, but tested with the nonlinear model of the same system. The system performances with and without the feedforward control channel are compared. Simulation results suggest that with LIDAR information, the added feedforward control has the potential to reduce blade and tower loads in comparison to a baseline feedback control alone.

A Wind farm controller has been proposed with the objective of maximizing generated power whilst reducing the fatigue loads on the turbines. Since the loads on a wind turbine depend on both the wind-field/rotor interaction and the turbine full envelope controller, the simulation models must include the most significant turbine structural modes, the full envelop controller as well as a representation of the wind-field. In this paper, a wind farm Simulink model that captures all these relevant dynamic elements is reported. In addition, wind farm control is investigated. The objective is to maintain maximum wind farm power output whilst redistributing the fatigue loads between turbines.

With the development of large offshore wind farms and attainment of high wind power penetration, it is no longer satisfactory for wind farms to be passive providers of generated power. Instead, offshore wind farms must become virtual generation plant that behave similarly to conventional generation. The power must be adjusted as required by the operators. To do so, requires flexible operation of the individual turbines and a wind farm controller to match power output to demand. In addition to adjusting the power output, the wind farm controller could enable the wind farm to provide ancillary services such as curtailment, frequency support, voltage support, etc. Furthermore, there is extensive information regarding the local environment and conditions, including SCADA information, environmental information (wind direction, time of year, sea state etc., maintenance and repair logs information, wind farm layout information, condition monitoring and turbine health information) as well as individual wind turbine control information from nearby turbines. The potential to exploit this information through the wind farm controller to enable operators to make the most of their assets is substantial. In its most sophisticated form, the wind farm controller could control the operation of the individual wind turbines to achieve the most effective short and long term operation of all the assets in the wind farm. Accordingly, the general objectives for the wind farm controller is to maximise wind farm generated power, provide ancillary services, including curtailment, frequency support and voltage support, and minimise O&M costs. An approach to wind farm control, that is hierarchical, decentralise and scalable, is presented.

Improving offshore wind turbine reliability is a key industry goal to improve the availability of this renewable energy generation source. The semiconductor devices in the wind turbine power converter are traditionally considered as the most sensitive and important components to achieve this and managing their thermomechanical stressing is vital, since this is one of their principal long-term aging mechanisms. Conventional deterministic reliability prediction methods used in industrial applications are not suitable for wind turbine applications, due to the stochastic nature of the wind speed. This paper develops an electrothermal model of the power devices, which is integrated with a wind turbine system model for the investigation of power converter thermal cycling under various operating conditions. The model has been developed to eliminate the problems of pulse width modulation switching, substantially reducing simulation time. The model is used to improve the current controller tuning method to reduce thermal stresses suffered by the converter during a grid fault. The model is finally used to design a control method to alleviate a key problem of the doubly fed induction generator—severe thermal cycling caused during operation near synchronous speed.

This paper investigates new ways of inferring nonlinear dependence from measured data. The existence of unique linear and nonlinear sub-spaces which are structural invariants of general nonlinear mappings is established and necessary and sufficient conditions determining these sub-spaces are derived. The importance of these invariants in an identification context is that they provide a tractable framework for minimising the dimensionality of the nonlinear modelling task. Specifically, once the linear/nonlinear sub-spaces are known, by definition the explanatory variables may be transformed to form two disjoint sub-sets spanning, respectively, the linear and nonlinear sub-spaces. The nonlinear modelling task is confined to the latter sub-set, which will typically have a smaller number of elements than the original set of explanatory variables. Constructive algorithms are proposed for inferring the linear and nonlinear sub-spaces from noisy data.

Gaussian process (GP) regression is a Bayesian non-parametric regression model, showing good performance in various applications. However, it is quite rare to see research results on log-likelihood maximization algorithms. Instead of the commonly used conjugate gradient method, the Hessian matrix is first derived/simplified in this paper and the trust-region optimization method is then presented to estimate GP hyperparameters. Numerical experiments verify the theoretical analysis, showing the advantages of using Hessian matrix and trust-region algorithms. In the GP context, the trust-region optimization method is a robust alternative to conjugate gradient method, also in view of future researches on approximate and/or parallel GP-implementation.

Gaussian process (GP) is a Bayesian nonparametric regression model, showing good performance in various applications. However, its hyperparameter-estimation procedure suffers from numerous covariance-matrix inversions of prohibitively O(N3) operations. In this paper, we propose using the quasi-Newton BFGS O(N2)-operation formula to update recursively the inverse of covariance matrix at every iteration. As for the involved log det computation, a power-series expansion based approximation and compensation scheme is proposed with only 50 N2 operations. A number of numerical tests are performed based on the 2D-sinusoidal regression example and the Wiener-Hammerstein identification example. It is shown that by using the proposed implementation, more than 80% O(N3) operations are eliminated, and the speedup of 5 ~9 can be achieved.

When data contains components with different characteristics and it is required to identify both, standard Gaussian regression, based on a model with a single stochastic process, is inadequate. In this paper, a novel adaptation of Gaussian regression, based on models with two stochastic processes, is presented. In both the prior and posterior joint probability distributions, the Gaussian processes for the two components are independent. The effectiveness of the revised Gaussian regression method is demonstrated by application to wind turbine time series data.

While there exists a substantial literature on the identification of Hammerstein and Wiener models, the identification of Wiener-Hammerstein models has received considerably less attention yet this is a model class of very great practical importance. This paper proposes an elegant approach to estimating Wiener-Hammerstein systems from measured data.

A Matlab/Simulink based tool for developing fast simulation wind farm models has been proposed. The wind farm development tool accepts user defined parameters such as mean wind speed, turbulent intensity wind farm layout and etc. to develop a wind farm model. The wind farm model contains the 5MW Supergen wind turbine dynamics suitable for controller design, a wind-field model based on the Veers algorithm for generating turbulent wind-field, and wake models based on the Frandsen analytical wake modelling technique for representing the wake propagation through the wind farm. The developed wind farm model is suitable for active power control and developing control algorithms for primary frequency response.

Failure of offshore wind turbine components is one of the main contributors to the O&M cost and in turn the cost of energy from offshore wind. In order to try and reduce the O&M costs a clear understanding of what drives these failures must be obtained. This paper aims to find a correlation between weather conditions and turbine failures. Site conditions examined include, amongst others, wind speed, turbulence intensity and turbine positioning in the wind farm. Exact turbine and wind farm numbers, rated power and blade size cannot be provided for confidentiality reasons. The findings of this analysis show that, weather conditions are a key factor to offshore turbine failures.

Recent years have witnessed an enormous growth of interest in dynamic systems that are characterised by a mixture of both continuous and discrete dynamics, commonly referred to as hybrid or switching systems. Nevertheless, there is a notable lack of suitable benchmark problems on which to assess and compare competing analysis and design methods. The present paper provides a collection of detailed benchmark design and analysis tasks which, while somewhat simplified in nature, reflect the scale and complexity of the tasks encountered in at least one important application domain, namely wind turbine regulation.