Decisions about the viability of renewable energy developments must be data-driven and based on robust, reliable modelling. Alan Yates, Head of Energy Applications at the Institute for Environmental Analytics explores how effective strategies for renewable power production benefit from advanced energy modelling.
The ultimate goal of the energy transition is the complete decarbonisation of energy systems. Integrating plants that capture solar, wind and tidal energy into the energy infrastructure involves complex data-driven decision-making. As it is neither viable nor realistic to effect an immediate switchover, large-scale adoption of renewables into an established grid has to be a phased process to be sustainable.
A fundamental principle for electricity systems is that generation must balance with demand across all time periods, from micro-seconds to hours and throughout seasons. This doesn’t change when renewable energy sources are integrated, but increasing the proportion of variable renewable energy (VRE) in the mix does introduce some major new challenges.
The intermittent nature of renewables increases the need for flexibility in the power system to avoid blackouts, brownouts and other power quality issues. Understanding how the weather might behave, and how this impacts generation potential, is thus relevant for almost every decision about increasing the penetration of renewable energy. Modelling these impacts becomes even more significant when you factor in climate change. The need to de-risk energy supply and climate investments, and incentivise funding calls for informed, persuasive, scenario-based and data-driven modelling and insights.
Assess potential renewables resources
The availability and characteristics of VRE resources vary widely from location to location. The need to properly account for weather variability and seasonality is crucial for planning, whether that’s for an individual site, a portfolio of installations or a complete system. This calls for suitable datasets, models and tools.
The Institute for Environmental Analytics (IEA) has developed customised methods to merge numerical weather models with satellite observations that make it possible to downscale wind, temperature and irradiance estimates to a 1km spatial resolution and 10-minute timestep.
Once the expected resource has been quantified, it’s possible to estimate likely power yields and levelised costs. As a renewable energy project progresses, the level of detail required to reduce uncertainties, prove viability and make the financial case increases. Again, this necessitates appropriate data, modelling techniques and technology.
Forecast renewable production
Increasing the proportion of VRE in the energy mix also increases the importance of accurate forecasts of expected production. Current best practice involves running a combination of forecasting techniques up to the event. For example, if you are forecasting power production at 10:00 am on a given day, numerical weather prediction models provide sufficient forecasting skill up to 24 hours ahead.
From 24 hours up to 1 hour before the event, the best forecasting skill for solar comes from using satellite observations to predict the movement of clouds using a technique known as Cloud Motion Vectors.
From 1 hour before and up to the event, the best forecasting skill comes from statistical models combining forecasting methods with on-site observations from sky cameras, anemometers, and on-site power production.
Combining these prediction methods can give the system operator valuable data and help the site owner to schedule planned maintenance at times of low- or no-generation, to minimise costs, optimise revenues and maximise CO2 emission reduction.
As with all VRE estimates, the accuracy of the forecast is likely to be localised, dependent on weather regimes and potentially variable. In stable conditions forecasting accuracy could be very close to 100%, but in intermittent conditions forecasting accuracy can drop significantly.
Weather variability across timescales
It is possible to produce effective production profiles by modelling future mixes of energy generation across a range of timescales. This involves navigating through large, high-resolution time-series data to find typical and extreme conditions and requires significant modelling power.
Understanding variability across multiple timescales provides the key to understanding likely weather behaviour and the potential for, and impact on, renewable energy generation. Variability between years and months is important for feasibility analysis, planning and design activities. Understanding variation on a daily, hourly, minute or even second-by-second basis is crucial for making operational decisions to manage the grid.
For a solar plant, overcast conditions will maximise the need for dispatchable generation to compensate for the lack of VRE. Highly intermittent generation must be balanced either by storage or flexible dispatchable generation. Clear sky conditions will minimise the need for dispatchable generation.
At times of excess generation from wind or solar, the system operator may need to curtail energy production, store energy in batteries, increase demand – for example, encourage EV charging or industrial consumption – or convert it into a fuel such as hydrogen.
A tool that enables generation profiles to be visualised with increasing granularity shows how significant short-term variability can be on any given day. At an individual site, up to 80% of solar generation can be lost in as little as 30 seconds as clouds obscure the sun.
Robust weather behaviour modelling shows the impact on power generation of both typical and extreme weather conditions. This modelling informs the strategies that governments, operators, generators and investors will need to adopt to ensure a stable, efficient and economically sound renewable energy system.
Of course, every case is different. From scenario analysis to project design through to operational energy generation, models must be customised to the relevant specifics to highlight and quantify the uncertainty and limitations.
Effective decision-making around renewable energy production calls for a tool that enables quick evaluation of multiple different generation scenarios to plan effectively for a successful energy transition.
The Institute for Environmental Analytics has developed EnergyMetric, a web-based application to enable efficient development of modelled scenarios to inform decision-making for prospecting, pre-feasibility and feasibility analyses. The application is designed to help planners and investors create and explore potential future VRE generation scenarios as power systems evolve through transition phases.