Objective & Background
In 2017, according to the International Renewable Energy Agency (IRENA), the record of 167 GW of increment in global renewable installed capacity has been achieved, compared to 2016. This record is mainly due to photovoltaic installations. Despite the environmental benefit, the exponential growth of renewables, causes new problems to deal with, because of the weather dependence and non-programmable nature of some sources, like wind and solar. Operationally, these characteristics can lead to instability on the voltage profile and increasing uncertainty on energy reserve scheduling. The Transmission System Operator (TSO) every day provides orders on energy dispatch on extended areas, defined mainly on the basis of power transmission limitations. The expected production from renewables over the whole market area must be considered by TSO, to improve dispatching and reserve management. In this study a probabilistic forecasting tool of the hourly solar output for the six Italian market zones is presented and its performance is analysed. The forecast system relates to solar production, because this source is expected to grow above all other Renewable Energy Sources (RES) in Italy in the near future.
Method
The forecast system is based on the statistical post-processing technique Analog Ensemble (AnEn). AnEn is able to estimate the power forecast on the basis of the meteorological projection and no other information on the installed production plants. The method estimates the probability distribution of the future generation using the measured production registered in the past when the weather scenario looked very similar to the forecasted one. The AnEn input is made up of the forecast of the meteorological variables affecting the solar production from two different regional Numerical Weather Prediction (NWP) models: Weather Research and Forecasting model (WRF) and Regional Atmospheric Modeling System (RAMS). For training the AnEn, a period of two years of measurements of hourly regional power on each bidding zone, together with the corresponding weather forecast on the same areas, has been used. The method has been tested on each Italian bidding zone for a period of one year, by means of the regional solar output supplied by the Italian TSO. The AnEn, based on the multi-model forecast has been compared with persistence and another AnEn configuration, based on the weather forecast of a single NWP model.
Principal Findings
The AnEn with a multi-model approach is able to estimate with accuracy the solar production on a large zone. The statistical approach enables to obtain the power forecast with no need of technical information on the installations, very often difficult to obtain, especially on a large number of small installations.
Conclusions
The multi-model approach performs best on respect to other reference methods. AnEn is a probabilistic method, therefore also a confidence level for the forecast is provided.