New Industry Products

Clir’s Software Aims to Predict Renewable Energy Variability and Production

December 09, 2023 by Jake Hertz

Clir’s software tackles traditional shortcomings in predicting weather variability and renewable energy production forecasting. 

In the dynamic landscape of renewable energy, some of the biggest challenges surround forecasting expected energy production. Influenced by factors such as the weather, which is equally as complex to forecast, renewable production forecasting is a significant challenge for power electronics engineers.


Wind farm in a storm.

Wind farm in a storm. Image used courtesy of NREL


Clir Renewables has developed new software that utilizes a database of over 200 GW of industry data to model the variability of hourly energy production. The software could transform the approach to production forecasting. This article explores the technical background of renewable energy production forecasting, the specifics of Clir’s new technology, and its potential implications for the renewable energy sector.


Challenges in Renewable Energy Production Forecasting

Understanding the nuances of renewable energy production forecasting requires looking into the multifaceted challenges posed by the inherent variability of sources like wind and solar and the limitations of conventional forecasting models. 

This variability, primarily driven by unpredictable environmental conditions, is the crux of the forecasting dilemma. Wind and solar energy outputs are heavily influenced by wind speed, solar irradiance, temperature, and cloud cover. These elements are challenging to predict with precision and are subject to rapid and unpredictable fluctuations, leading to significant variability in energy output.


Forecasting power generation is a difficult task.

Forecasting power generation is a difficult task. Image used courtesy of RE Global


Adding to this complexity is the unique nature of each renewable energy site. Geographic location, topography, and local weather patterns distinctively shape the energy production capabilities of each site. For instance, the energy yield from two geographically close wind farms can differ substantially due to variations in local topography affecting wind patterns. Similarly, solar energy production is highly sensitive to localized cloud cover, which can vary greatly even over short distances and timeframes.


Weather Forecasting Shortcomings

The traditional approach to weather forecasting, heavily reliant on statistical models, often struggles to address these complexities effectively. These models typically utilize historical weather data and energy production records to estimate future output. However, their effectiveness is constrained by several factors. 

First, the reliability of weather forecasts diminishes significantly beyond a two-week horizon, limiting the accuracy of short-term predictions. Additionally, the reliance on historical data as a primary forecasting tool may not fully represent future conditions, particularly in the face of changing climate patterns and evolving weather scenarios.


Many factors impact weather forecasting.

Many factors impact weather forecasting. Image used courtesy of UW-Madison


A critical limitation of these models is their often inadequate consideration of hourly variations in energy production. Such granularity is essential for precision in forecasting, yet many models fail to capture these vital details. This omission can lead to substantial discrepancies in energy production predictions. Moreover, these models tend to overgeneralize, applying broad patterns to predict energy output without adequately accounting for individual sites' unique characteristics and conditions. 

This generalized approach can result in significant inaccuracies, especially in environments with diverse and complex weather systems.


Clir Renewables' Innovative Software Solution

Addressing these challenges, Clir Renewables has introduced a technology that leverages machine learning and an extensive database of over 200 GW of operational data. This software significantly enhances the accuracy of modeling the shape and volume of energy production. It utilizes a machine learning model trained on hourly gross energy data from wind or solar farms and corresponding climate data. The model is then tested against actual production data to refine its accuracy.


Clir’s energy production prediction model

Clir’s energy production prediction model. Image used courtesy of Clir


The software generates a time-series distribution of hourly gross energy production, incorporating over 20 years of data, resulting in a detailed simulation of energy production uncertainty bands for any given time frame. Furthermore, Clir’s software accounts for expected energy losses at the farm level, using peer loss data based on various operational factors. This advanced modeling provides a more accurate representation of net energy production.

A key application of Clir’s software is in the realm of power production hedges. These hedge contracts, essential for revenue certainty in fluctuating market price scenarios, can be optimized using Clir’s accurate energy production forecasts. By analyzing expected production across multiple assets and considering diversification impacts, the software also enables optimization of portfolio-wide hedges.


Empowering Operators

The introduction of Clir Renewables' production volume modeling platform is a significant milestone in the renewable energy sector. It empowers infrastructure investors and operators with a tool to unlock millions in revenue potential. As tests with major wind investors demonstrated, this technology provides a more accurate basis for production expectations, leading to substantial financial benefits.

This innovation helps to enhance the economic feasibility of projects involving renewable energy and contributes to reducing reliance on fossil fuels. Such advancements are set to play a pivotal role in shaping future renewable energy optimization and off-take strategies.