Search

International audience; Uncertainty in the upcoming production of photo-voltaic (PV) plants is a challenge for grid operations and also a source of revenue loss for PV plant operators participating in electricity markets, since they have to pay penalties for the mismatch between contracted and actual productions. Improving PV predictability is an area of intense research. In real-world applications, forecasts are often needed for different time frames (horizon, update frequency, etc.) and are derived by dedicated models for each time frame (i.e. for day ahead and for intra-day trading). This can result in both different forecasted values corresponding to the same horizon and discontinuities among time-frames. In this paper we address this problem by proposing a novel seamless probabilistic forecasting approach able to cover multiple time frames. It is based on the Analog Ensemble (AnEn) model, however it is adapted to consider the most appropriate input for each horizon from a pool of available input data. It is designed to be able to start at any time of day, for any forecast horizon, making it well-suited for applications like continuous trading. It is easy to maintain as it adapts to the latest data and does not need regular retraining. We enhance short-term predictability by considering data from satellite images and in situ measurements. The proposed model has low complexity compared to benchmark models and is trivially parallelizable. It achieves performance comparable to state-of-the-art models developed specifically for the short term (i.e. up to 6 hours) and the day ahead. The evaluation was carried out on a real-world case comprising three PV plants in France, over a period of one year.

International audience; The valorization of photovoltaic (PV) energy generation involves several decision making processes at different times with different objectives. For example, a PV power plant coupled with a Battery Energy Storage System (BESS) has to provide bids in the day-ahead electricity market, but can also provide ancillary services. On the delivery day, it can also participate in intra-day trading sessions, and must decide which quantity to charge or discharge from the BESS in real-time. These successive decision-making processes all require forecasts of the energy production level for different forecast horizons.However, the models and the inputs used for the different forecast horizons are often different. A common result is that in situ measurements are more accurate for very-short term forecasts (real-time to one hour ahead forecasts), satellite data is better for short-term forecasts (up to 6 hours ahead), and Numerical Weather Predictions (NWP) are better for long-term forecasts (day-ahead and longer). Models also vary, with auto-regressive approaches being commonly used for very-short term forecasts, while longer forecast horizons use a wide range of machine learning models.The RES producers have thus to develop and maintain numerous forecasting models for the different decision-making processes they are involved in, usually fitted for each power plant. This increases further the complexity of the decision-making processes and can create problems regarding the continuity of the forecasts.In this work we propose a forecasting model for PV power generation that can use all the inputs mentioned before, and weights them according to the forecasting horizon. It can thus operate from very short-term to day-ahead forecast horizons with state-of-the-art performance. It can also directly provide probabilistic forecasts for an aggregation of power plants, thus allowing having a single forecasting model for managing a virtual power plant. The model follows the “lazy learning” paradigm, where generalization from the training set is only computed when a forecast is requested. Thus, the model is resilient to changes in the neighborhood of the plant (surrounding environment, partial outage, soiling, etc.)The model is based on the Analog Ensemble (AnEn) method. However it is structurally expanded to allow the method to use an arbitrary large number of inputs. Each input is then weighted depending on the forecast horizon. As an example, for a given input for one-hour ahead, the weight is computed based on the Mutual Information (MI) between the input and the PV power generation observed one hour later. This allows dynamically selecting the most relevant inputs depending on the horizon. The model is evaluated for short-term and day-ahead forecasts, and compared with a Quantile Regression Forest (QRF) for day-ahead forecasts, and a linear Auto-Regressive Integrated Moving Average (ARIMA) model for the short term forecasts. Results show that the AnEn model is competitive with the QRF model in day-ahead forecasting. It is also consistently better than the ARIMA model for short-term forecasting.

International audience; The increasing share of photovoltaic (PV) power in the global energy mix presents a great challenge to power grid operators. In particular, PV powers intermittency caused by varying weather conditions can cause mismatches between energy production and expectation. Battery Energy Storage Systems (BESS) are often put forward as a good technological solution to these problems, as they are able to mitigate PV power forecast errors. However, the investment cost for such systems is high, which makes their use in operational contexts difficult. In this paper, we compare several strategies to manage a PV power plant coupled with a BESS in a market environment. They are obtained by stochastic optimization using a Model Predictive Control (MPC) approach. This paper proposes an approach that takes into account the ageing of the BESS, both at the day-ahead level and in the real-time control of the BESS, by modeling the cost associated with BESS usage. As a result, the BESS arbitrates between compensating forecast errors and preserving its own life expectancy, based on both PV production and price scenarios derived from probabilistic forecasts. A sensitivity analysis is also carried out to provide guidelines on the optimal sizing of the BESS capacity, depending on market characteristics and BESS prospective costs.

International audience; Accurate analysis of meteorological and pyranometric data for long-term analysis is the basis of decision-making for banks and investors, regarding solar energy conversion systems. This has led to the development of method-ologies for the generation of Typical Meteorological Years (TMY) datasets. The most used method for solar energy conversion systems was proposed in 1978 by the Sandia Laboratory (Hall et al., 1978) considering a specific weighted combination of different meteorological variables with notably global, diffuse horizontal and direct normal irradiances, air temperature, wind speed, relative humidity. In 2012, a new approach was proposed in the framework of the European project FP7 ENDORSE. It introduced the concept of " driver " that is defined by the user as an explicit function of the pyranometric and meteorological relevant variables to improve the representativeness of the TMY datasets with respect the specific solar energy conversion system of interest. The present study aims at comparing and benchmarking different TMY datasets considering a specific Concentrated-PV (CPV) system as the solar energy conversion system of interest. Using long-term (15+ years) time-series of high quality meteorological and pyranometric ground measurements, three types of TMY datasets generated by the following methods: the Sandia method, a simplified driver with DNI as the only representative variable and a more sophisticated driver. The latter takes into account the sensitivities of the CPV system with respect to the spectral distribution of the solar irradiance and wind speed. Different TMY datasets from the three methods have been generated considering different numbers of years in the historical dataset, ranging from 5 to 15 years. The comparisons and benchmarking of these TMY datasets are conducted considering the long-term time series of simulated CPV electric production as a reference.

International audience; This work analyses the long-term trend of the daily global (GHI) and diffuse (DHI) irradiations received on a horizontal plane for four cities in China: Harbin, Beijing, Wuhan and Guangzhou, located from North to South. Measurements of GHI and DHI between 1990 and 2013 have been retrieved from GEBA and WRDC networks. During this period, the yearly mean of the GHI increases for most of the sites (0.1 to 0.7% per year) except for Harbin for which it decreases (-0.4% per year) while the yearly mean of the DHI increases for all sites (0.2 to 0.9% per year). The effects of the aerosol optical depth at 550 nm and the cloud cover on such changes have been investigated. It has been found that aerosols have a direct impact on GHI in clear-sky conditions, especially for Beijing and Wuhan, and that the correlation is strong between the GHI measurements for all-sky conditions and aerosol optical depth at 550 nm. Expectedly, the correlation is much more significant between the GHI measurements and the cloud cover.

International audience; This paper explores the possibilities provided by satellite-based surface solar irradiation databases, for the attention of the photovoltaic actors, from the engineering and the financing to the operating of photovoltaic systems. First, the problematic of using meteorological ground stations for the determining of solar dataset or time-series is addressed with the example of the French national meteorological network and the focus made on the South-West part of France: the heterogeneity in terms of spatial and temporal representativeness along with the relevance of the delivered measurement are questioned. Then, an innovative synthesis of 16 satellite-based ISS databases available so far is presented through the distribution of their corresponding features within three categories: spatial and temporal representativeness, data type in terms of component and format, and finally operating mode for the retrieving of the data in terms of price and accessibility. The results of this review shall help photovoltaic actors making the correlation between the available satellite-based databases and their specific needs.

International audience; This study explores the possibility to calibrate the estimation of the global horizontal irradiation provided by HelioClim-3, a satellite-based surface solar irradiation database (available at www.soda-is.com). The main objective of this work is to refine such an estimation whose performances differ from one site to another. A first processing of the long-term measurements provided by nine weather stations located in Provence-Alpes-Côte d'Azur Region (South France) leads to the characterization of the clearness index error variability for that Region: this parameter is made up of a bias, a drift and 3 sinusoids with periods respectively equal to the astronomical year, half a year and one third of a year. We show that the phase of the dominant frequency (365 days) is similar whatever the tested site. We propose a simple calibration procedure based on a linear regression whose performances, in terms of mean bias error and root mean square error, depend on the beginning and the duration of the measurement campaign; to illustrate this point, the mean bias error on the global horizontal irradiation for nine sites considered systematically goes below 3% when considering a 6-month measurement campaign starting in May. We also show that the performances of the proposed calibration are also applicable to another site in the same Region for which the initial error exceeds 13%. A graphical representation allows visualizing the characterization of these measurement campaigns depending on the expected accuracy.

International audience; This paper presents the benchmarking of different Typical Meteorological Year (TMY) datasets applied to a Concentrated-PV (CPV) system. Using 18-years of high quality meteorological and pyranometric ground measurements, five types of TMY datasets were generated using variable time period and following different methods: the standard Sandia method or only considering the Direct Normal Irradiation (DNI) or a more sophisticated DNI-based driver considering the characteristics of the CPV system. The results show that the Sandia method is not suitable for CPV systems. The TMY datasets obtained using dedicated drivers are more representative to derive TMY datasets from limited long-term meteorological dataset.