The share of Renewable Energy Sources (RES) in global energy production is steadily growing. So is the number and variety of customers using 4RES, our tool for forecasting energy production from renewable sources.
4RES is a forecasting system we’ve been developing for almost 10 years. It has evolved from point forecasts for designated wind farm locations to area forecasts of industrial photovoltaic installations. Moreover, it now also provides production forecasts based on prosumer installations and their impact on the grid.
New forecasting models for higher forecast accuracy
Our systematic cooperation with customers has recently brought several new capabilities to the 4RES system, increasing its attractiveness and competitiveness.
Two seasons ago, we launched a new forecast model supporting the snowfall and the accumulation and disappearance of snow cover on PV panels. The model has significantly improved the forecast accuracy by reducing errors by an average of 25% during periods of precipitation.
Our forecasts support various solar farm designs, so we introduced support for installations with solar tracers and non-standard panel orientations (e.g., aligned east-west). Moreover, our forecasts also take bifacial photovoltaic cell implementation into account. One example is the implementation that includes both-sided energy production from reflected or scattered light.
Models for the latter solutions are based primarily on machine learning and are currently in the testing phase.
The 4RES system has also been enhanced with an API allowing our customers to…
- correct the actual power involved in energy production at a specific time, and resulting from repairs, overhauls, or shutdowns of generating unit sections,
- adjust the production in case of system constraints
- introduce changes in the installed source capacity or add a new source.
Our standard production forecasts for the next day are called D+1 forecasts. Our customers use them mainly for energy trading purposes. In addition, we are starting to provide forecasts in the medium-term horizon, 10 days in advance – the so-called D+9 forecasts. The beneficiaries of the latter are Distribution System Operators and Transmission Network Operators, who use them to plan the National Power System operations.
Each type of forecast is created based on different meteorological models. Often, we use multiple models with different weights to obtain the lowest possible errors.
We are also researching current production measurements to revise ultra-short-term forecasts for the intraday market.
Did you find this article interesting and want to learn more about renewable production forecasting?