Power and renewables

Forecasting, the story so far and the route ahead

Forecasting, the story so far and the route ahead

Contact us:

Ayumu Suzuki

Ayumu Suzuki

Technical Lead, Forecasting

Related services:

Forecaster

Over the last decade there has been an enormous 400 GW of wind capacity installed. During this time, wind power forecasting capabilities have advanced significantly. But how will wind power forecasting continue to advance the wind industry and help maximize the value of renewable energy?

The wind doesn’t always blow, and the sun doesn't always shine. Unlike conventional fossil fuel energy plants which can schedule their generation, the output of a renewable power plant is as changeable as the weather. A wind farm can’t generate energy without a breeze. Therefore, being able to understand and predict weather patterns can considerably impact the profitability of a wind turbine or wind farm project.

As more renewable sources are connected to the grid, accurate short-term forecasting becomes essential. Accurate forecasts bring significant financial and social benefits for stakeholders throughout the energy value chain, enabling reliable grid operations and better decision-making for traders. As forecasting contributes to making renewables a viable and economical source of energy, it plays a key role in contributing to the energy transition and improving air quality for a healthier environment for society.

In 2010, global wind farm capacity was around 200 GW . But over the course of the last decade there has been an enormous 400 GW installed. This significant rise in wind farm installations has ensured that more energy is produced from clean, renewable energy sources each year and less is generated from fossil fuel power plants, such as coal and gas. This is particularly good news for countries keen to meet legally binding climate reduction targets by 2050, as coal produces around 980 grams of carbon dioxide per kilowatt-hour and natural gas around 465 grams of carbon dioxide per kilowatt-hour. So, increased wind usage over the last ten years has saved billions of greenhouse gas emissions and make the air that little bit cleaner for us all to breathe.

There have been some incredible advances within the industry over this time, such as innovative large offshore turbine technology, improvements in foundation design and new installation techniques. But one of the lesser appreciated developments has come during the operational phase of a wind farm project, in the form of forecasting energy output. Wind power forecasting is not a new concept, it has been around for over 15 years. However, its capabilities to help owners and operators predict and maximize the value of energy production has been accelerated over the last decade, as the energy markets are maturing to integrate renewables.

Global forecasting market
The key focus for short-term power forecasting has traditionally been on the accuracy of day-ahead forecasts in Europe and North America. With more accurate day-ahead forecasts, traders can minimize the imbalance costs and maximize the value of the energy they are trading.

With the development of the single electricity market across Europe, an increasing number of countries are now part of the single day-ahead market. In 2018, we saw the Single Electricity Market on the island of Ireland joining the market, and Greece has been the most recent joiner in December 2020. We expect to see further geographical expansion of the market across Eastern Europe. As more renewables are connected to the grid, market participants are becoming more interested in trading in the intraday markets. Due to the variability of renewables, traders can choose bidding strategies to reduce risks by taking part on the intra-day market as opposed to day-ahead, and the European market is evolving to provide a single EU cross-zonal intraday electricity market to accommodate for this shift in requirements. Across other parts of the world, such as North America and Australia, the shift of focus towards intra-day forecasts with higher granularity is also being observed.

In countries where renewables are growing but electricity trade markets are yet to be established, mostly across Asia and Africa, we are observing an increase in grid operators requesting the power producers to provide day-ahead and intra-day forecasts. In some countries there are penalties associated with inaccurate forecasts, and this provides an incentive for accurate forecast provision to minimize the imbalance costs for the power producers.

Looking ahead, we are expecting to see energy storage play a key part in the integration of renewables as they become an increasingly economically viable option to solve the variability problems associated with renewables. Throughout this transition, we would expect to see an increase in demand for economic dispatch scheduling for storage solutions, where accurate short-term generation forecasts, coupled with price forecasts, are expected to be crucial inputs.

What’s changed in wind power forecasting?
Over the last decade, wind power forecasting capabilities have advanced significantly. Although the multi-model ensemble approach which combines weather model data, applied downscaling and optimization remains relevant, advancements in computer technology have enabled weather models to be run at much higher resolutions and updated more frequently. This provides more accurate inputs to power forecasting models. Meteorological agencies have also acknowledged the growth of the wind power industry and started to provide 100m above ground level forecasts which are more relevant for the purpose of wind power forecasting. This was not commonly available a decade ago.

In recent years, advancements in machine learning models have had a significant impact on forecasting methodologies, by helping to better capture variations in wind and solar power generation, especially for sites which were difficult to forecast accurately using traditional statistical methodologies. DNV GL’s latest development of machine learning models has resulted in an improvement of up to 2% Mean Absolute Error (MAE as % of installed capacity), in comparison to DNV GL’s traditional forecasting methodology. Based on a case study carried out in 2019, DNV GL discovered that the potential benefit of a 2% MAE improvement in day-ahead forecasts can be in the region of 1 GBP/MWh of additional revenue on the Great Britain market. This can be an extra 1.3 million GBP annually for a 500 MW offshore wind farm. This added value has been acknowledged by several traders.

Looking ahead, we would expect to see technical advancements in areas where the energy market’s key requirements are shifting, such as higher frequency intra-day forecasts. As the focus shifts towards real-time (sub 5 minutes), efficient and reliable data exchange will be key to ensure highly accurate forecasts are delivered to the end users.

Accurate forecasts have supported the integration of renewables over the last decade, and we expect them to continue to play a key role in the energy transition over the decade to come.

Contact us:

Ayumu Suzuki

Ayumu Suzuki

Technical Lead, Forecasting

Related services:

Forecaster