The ever-increasing complexity of renewable energy projects and their supporting technological and financial frameworks is accelerating the need to have a sophisticated understanding of the energy produced.
In the past, the total quantity of energy generated was the only real consideration for many owners and operators. However, in today’s market, renewable projects are increasingly having to consider the delivery profile of that energy.
Historically, when projects could rely on a fixed price for their energy, such as through a government support mechanism, owners didn’t worry so much about the variable market price. However, as the penetration of renewable energy has been growing in markets throughout the world, projects are becoming increasingly exposed to market pricing (e.g. acting as merchant projects) and the generation profile is now just as important as the quantity produced. Furthermore, the development of hybrid power plants powered by renewables means that additional clarity is needed on the profile and the certainty of supply. This is true when trying to efficiently combine wind and solar power generation with the requirements of storage units or with electrolysers producing “green” hydrogen.
For a naturally variable source of power production like renewables, all of the above market developments pose a challenge. At DNV GL, we have been developing increasingly advanced methodologies to understand the production of renewable projects at specific times. This provides valuable insights for those taking on the risk of variable generation in a world of variable prices.
The simplest way of assessing the shape of production is to look on a monthly basis. Monthly analysis is sufficiently aggregated to be easy to model and low risk to use. It can also be expanded to consider daily variations in monthly production. Nonetheless, this resolution is becoming more coarse relative to the resolution of revenue generated, so the next step is to move into the domain of hourly production analysis. How often will the asset produce above or below certain thresholds? How much will it produce when energy prices are high or low? How well will a storage system or back-up energy supply cope with the variable levels of demand and generation?
It’s worth highlighting that the accurate modelling of hourly production is not trivial. In particular, energy loss factors can no longer be applied as annual averages but must be added as a realistic time series. Also, each loss factor has a different hourly characteristic and poses a different problem. Incorrectly accounting for these can easily skew any resulting analysis. For example, does turbine downtime impact every hour of the year equally? Are icing losses being included in summer months?
Additionally, there remains one final problem once you have identified a realistic power time series: your time series is based on one (or more) historical year of meteorological data and doesn’t represent what your asset will actually produce in the future. We cannot predict hourly production beyond a few weeks into the future, so history is our best and only guide. However, it represents one of an infinite number of possible futures, not the only one. Should you really hedge your bet on a single set of numbers that you know are not going to be repeated?
Luckily, the latest developments we have undertaken in time series modelling offer a solution and it’s all down to the stochastic engine. The stochastic engine analyses a set of concurrent data (wind speed, irradiance, temperature, price, demand, etc), understands the statistical interdependencies and then generates infinite new sets of variables which retain the same statistical characteristics. In other words, the stochastic engine allows us to create thousands of plausible energy, load or revenue time series to take a big data look at time series problems. Want to know the P90 production in any hour? Want to see the P99 revenue over a year? These are the kinds of questions we can now answer with the stochastic engine, giving renewable projects the tools needed to realistically model future scenarios.
Off-takers already value the ability to know what their operational output will look like and soon so will savvy investors. Better understanding means better financing arrangements and better credibility for the renewables industry. Now that’s something worth modelling!