Improving short-term wind power forecasts during ramp events by including nacelle wind speed observations

<p>Wind energy is and will be one of the key technologies for a transition to green electricity. However, the smooth integration of the generated wind energy into the electrical grid depends on reliable power forecasts. Rapid changes in power generation, so-called ramps, are not always reflected properly in NWP data and pose a challenge for power predictions and, therefore, grid operation. While contributions to the topic of ramp forecasting increased in the recent years, this work approaches the mitigation of deviations from the forecast more directly.</p> <p>The power forecast tool used here is based on an artificial neural network, trained and evaluated on multiple years of data. It is applied in comparison to power generation data for a 44 MW wind farm in Brandenburg. For short-term wind power forecasts, NWP wind speeds in this power forecast tool are replaced with recent Doppler Lidar wind profiles and nacelle wind speed observations from ultra-sonic anemometers, aiming to provide an easy-to-implement way to reduce negative impacts of ramps. Compared to NWP input data, this persistence approach with observational data aims to improve the forecast quality especially during the time of wind ramps.</p> <p>Different ramp definitions and forecast horizons are explored. In general, the number of ramps detected increases dramatically when using wind speed observations instead of the (too smooth) NWP model data. In addition, the mean deviation between power forecast and actual power generation around ramp events decreases, indicating a reduced need for balancing efforts.</p>

Statistical impact of wind-speed ramp events on turbines, via observations and coupled fluid-dynamic and aeroelastic simulations

Via 11 years of high-frequency measurements, we calculated the probability space of expected offshore wind-speed ramps, recasting it compactly in terms of relevant load-driving quantities for horizontal-axis wind turbines. A statistical ensemble of events in reduced ramp-parameter space (ramp acceleration, mean speed after ramp, upper-level shear) was created to capture the variability of ramp parameters and also allow connection of such to ramp-driven loads. Constrained Mann-model (CMM) turbulence simulations coupled to an aeroelastic model were made for each ensemble member, for a single turbine. Ramp acceleration was found to dominate the maxima of thrust-associated loads, with a ramp-induced increase of 45 %–50 % for blade-root flap-wise bending moment and tower-base fore–aft moment, plus ∼ 3 % per 0.1 m/s2 of bulk ramp-acceleration magnitude. The ensemble of ramp events from the CMM was also embedded in large-eddy simulation (LES) of a wind farm consisting of rows of nine turbines. The LES uses actuator-line modeling for the turbines and is coupled to the aeroelastic model. The LES results indicate that the ramps, and the mean acceleration associated with them, tend to persist through the farm. Depending on the ramp acceleration, ramps crossing rated speed lead to maximum loads, which are nearly constant for the third row and further downwind. Where rated power is not achieved, the loads primarily depend on wind speed; as mean winds weaken within the farm, ramps can again have U < Vrated. This leads to higher loads than pre-ramp conditions, with the distance where loads begin to increase depending on inflow Umax⁡ relative to Vrated. For the ramps considered here, the effect of turbulence on loads is found to be small relative to ramp amplitude that causes Vrated to be exceeded, but for ramps with Uafter < Vrated, the combination of ramp and turbulence can cause load maxima. The same sensitivity of loads to acceleration is found in both the CMM-aeroelastic simulations and the coupled LES.

Statistical impact of wind-speed ramp events on turbines, via observations and coupled fluid-dynamic and aeroelastic simulations

Via 11 years of measurements, we calculated the probability space of expected offshore wind speed ramps, recasting it compactly in terms of relevant load-driving quantities for horizontal-axis wind turbines. A statistical ensemble of events in reduced ramp-parameter space (ramp acceleration, mean speed after ramp, upper-level shear) was created, to capture the variability of ramp parameters and also allow connection of such to ramp-driven loads. Constrained Mann-model (CMM) turbulence simulations coupled to an aero-elastic model were made for each ensemble member, for a single turbine. Ramp acceleration was found to dominate the maxima of thrust-associated loads, with a ramp-induced increase of 45–50 % for blade-root flap-wise bending moment and tower base fore-aft moment, plus ~3 % per 0.1 m s−2 of bulk ramp acceleration magnitude. The ensemble of ramp events from the CMM was also embedded in large-eddy simulation (LES) of a wind farm consisting of rows of nine turbines. The LES uses actuator-line modelling for the turbines and is coupled to the aero-elastic model. The LES results indicate that the ramps, and the mean acceleration associated with them, tend to persist through farm. Depending on the ramp acceleration, ramps crossing rated speed lead to maximum loads, which are nearly constant for the third row and further downwind. Where rated power is not achieved, the loads primarily depend on wind speed; as mean winds weaken within the farm, ramps can again have U < Vrated. This leads to higher loads than pre-ramp conditions, with the distance where loads begin to increase depending on inflow Umax relative to Vrated. For the ramps considered here, the effect of turbulence on loads is found to be small relative to ramp amplitude that causes Vrated to be exceeded, but for ramps with Uafter < Vrated, the combination of ramp and turbulence can cause load maxima. The same sensitivity of loads to acceleration is found in both the the CMM-aeroelastic simulations and the coupled LES.

Characterisation of intra-hourly wind power ramps at the wind farm scale and associated processes

One of the main factors contributing to wind power forecast inaccuracies is the occurrence of large changes in wind power output over a short amount of time, also called “ramp events”. In this paper, we assess the behaviour and causality of 1183 ramp events at a large wind farm site located in Victoria (southeast Australia). We address the relative importance of primary engineering and meteorological processes inducing ramps through an automatic ramp categorisation scheme. Ramp features such as ramp amplitude, shape, diurnal cycle and seasonality are further discussed, and several case studies are presented. It is shown that ramps at the study site are mostly associated with frontal activity (46 %) and that wind power fluctuations tend to plateau before and after the ramps. The research further demonstrates the wide range of temporal scales and behaviours inherent to intra-hourly wind power ramps at the wind farm scale.

A simple methodology to detect and quantify wind power ramps

Knowledge about the expected duration and intensity of wind power ramps is important when planning the integration of wind power production into an electricity network. The detection and classification of wind power ramps is not straightforward due to the large range of events that is observed and the stochastic nature of the wind. The development of an algorithm that can detect and classify wind power ramps is thus of some benefit to the wind energy community. In this study, we describe a relatively simple methodology using a wavelet transform to discriminate ramp events. We illustrate the utility of the methodology by studying distributions of ramp rates and their duration using 2 years of data from the Belgian offshore cluster. This brief study showed that there was a strong correlation between ramp rate and ramp duration, that a majority of ramp events were less than 15 h with a median duration of around 8 h, and that ramps with a duration of more than a day were rare. Also, we show how the methodology can be applied to a time series where installed capacity changes over time using Swedish onshore wind farm data. Finally, the performance of the methodology is compared with another ramp detection method and their sensitivities to parameter choice are contrasted.

Wind Power Ramp Event Forecasting Based on Feature Extraction and Deep Learning

In order to improve the accuracy of wind power ramp forecasting and reduce the threat of ramps to the safe operation of power systems, a wind power ramp event forecast model based on feature extraction and deep learning is proposed in this work. Firstly, the Optimized Swinging Door Algorithm (OpSDA) is introduced to detect wind power ramp events, and the extraction results of ramp features, such as the ramp rate, are obtained. Then, a ramp forecast model based on a deep learning network is established. The historical wind power and its ramp features are used as the input of the forecast model, thereby strengthening the model’s learning for ramp features and preventing ramp features from being submerged in the complex wind power signal. A Convolutional Neural Network (CNN) is adopted to extract features from model inputs to obtain the coupling relationship between wind power and ramp features, and Long Short-Term Memory (LSTM) is utilized to learn the time-series relationship of the data. The forecast wind power is used as the output of the model, based on which the ramp forecast result is obtained after the ramp detection. Finally, the wind power data from the Elia website is used to verify the forecast performance of the proposed method for wind power ramp events.