Renewable energy based on solar and wind resources is variable due to its dependency on the weather. The energy output from these sources is quite uncertain as the wind can stop blowing suddenly or a cloud cover can reduce the solar irradiation. If the share of renewables in the total energy mix was small, this intermittency in generation patterns would not have a significant impact. However, with the Indian government targeting 175 GW of renewable energy by 2022 and possibly 450 MW by 2030, the variability of solar and wind needs serious consideration. Such a high share of intermittent energy sources in the electric grids poses immense challenges, especially for utilities and grid operators that have to manage the power demand and supply. This variability not only has technical impacts but also commercial ones. For instance, if the expected solar or wind generation is not available, grid operators would have to opt for any other available power source, which may come at a higher cost. For these reasons, proper forecasting and scheduling of solar and wind generation is of utmost importance.
Global prediction models and applications
Numerical weather models are used to predict the energy generation from solar and wind power plants as they can simulate future atmospheric parameters like wind speeds or solar radiation. Moreover, they can forecast for longer periods that is, several hours or days, unlike observation data, which, though useful, is limited to a few minutes. The key idea behind wind and solar forecasting is using meteorological parameters to estimate the power output. The two major approaches used for this purpose are the statistical approach and the physical approach. The statistical approach depends on mathematical relations between numerical weather predictions and the measured power output. Meanwhile, the physical approach uses power curves to convert meteorological parameters into power output. Often, both these methods are combined for more accurate predictions. Only a few stakeholders have developed their own in-house forecast solution. The majority of the renewable energy generators as well as other stakeholders like grid operators use forecasts provided by commercial forecast service providers. The commercial forecast service providers have the latest and most advanced software and digital tools. They send out power predictions like weather forecasts to their clients on a regular basis. For instance, DNV GL’s short-term energy forecasting service, comprising Forecaster Now, Forecaster Live, Forecaster Plus and Forecaster Solutions, can provide hour-by-hour energy generation predictions up to 15 days ahead, updated as frequently as every five minutes.
Similarly, Underwriters Laboratories offers a range of software products for the development, assessment and operation of renewable power projects. These include the Wind Data Management Dashboard, which allows the monitoring and assessment of meteorological data securely and conveniently, and Windographer for importing, assessing and visualising wind resource data. A French company called Steadysun provides different types of forecasting solutions – SteadyEye for solar generation forecasts based on real-time sky imagery, SteadySat for forecasts based on satellite data, and Steady Met for production forecasts based on meteorological models and simulations. Further, a German company, enercast, uses artificial intelligence and big data for supporting its clients’ decision-making processes. It has self-learning software-as-a-service products for providing power forecasts for wind power and solar plants. A Danish company ENFOR provides a total forecasting service that delivers country-level forecasts of wind and solar power production as well as the electricity load. The service utilises the ENFOR forecast engines, SolarForTM, WindForTM and LoadForTM, in special configurations in order to forecast production or demand for countries.
Regulatory frameworks in India
More than 15 Indian states have released forecasting, scheduling and deviation settlement mechanism (DSM) regulations, with some of these in draft mode. Under these regulations, solar and wind power developers have to pay DSM charges in case of under and over injection as against the scheduled energy generation. Moreover, the Central Electricity Regulatory Commission has defined a framework for forecasting, scheduling and imbalance handling for renewable energy generating stations at the interstate level. According to this, renewable energy generators will provide their forecast to the concerned regional load despatchcentre (RLDC). This may be based on their own forecast or the RLDC’s forecast. The RLDCs then consolidate the estimates and forecast based on certain parameters and weather-related data. The generators may prepare their schedule based on the forecast done by the RLDC or their own forecast. For multiple developers within a solar or wind park, the RLDC is responsible for scheduling, communication and coordination with generators of 50 MW and above connected to the interstate transmission system. Meanwhile, the lead developer is responsible for coordination and communication with the RLDC and other agencies for scheduling of multiple generators with an aggregate capacity of up to 50 MW within the park. The RLDC will upload the day-ahead schedules of forecasted energy generation within 15-minute intervals on its website. Generators can revise their schedule only after giving an advance notice to the concerned RLDC. Any commercial impact on account of deviation from schedule based on the forecast will be borne by the respective generator.
Renewable energy management centres
According to various studies, a centralised approach for forecasting and scheduling is preferred to a decentralised one. In the former, a grid operator or utility receives forecasts for all the renewable energy assets connected to the grid from various forecast service providers and consolidates all the forecasts to accurately predict generation patterns. Meanwhile, a decentralised approach would mean that every asset owner and operator would send its own forecast with discrepancy in data points and parameters, creating further issues for grid operators that rely on schedules based on these forecasts. Realising the importance of centralised forecasting and monitoring of renewable energy generation, 11 renewable energy management centres (REMCs) were recently commissioned in India. This is a landmark development in the effort to meet the country’s ambitious renewable energy targets. The REMCs have been developed by Power Grid Corporation of India Limited (Powergrid) under the central government’s Green Energy Corridors I scheme. These 11 REMCs are co-located with the load despatchcentres (LDCs) in Tamil Nadu, Karnataka, Andhra Pradesh, Maharashtra, Madhya Pradesh, Gujarat, Rajasthan, Bengaluru, Mumbai and New Delhi. Until recently, the LDCs lacked advanced renewable energy forecasting systems. Thus, REMCs were developed to address this issue through renewable energy forecasting systems for long-term planning, operations and reserves management. The REMCs have advanced systems for accurate forecasting and scheduling, such as SCADA systems, visualisation tools, display units, and corresponding hardware and software. REMCs use inputs from commercial forecast service providers and weather service providers to generate more accurate forecasts. These service providers use numerical weather prediction models to generate pooling station-wise forecasts and advanced modelling techniques to convert the meteorological parameters into power output.
Outlook
Forecasting and scheduling do not solely guarantee efficiency in operations. The quality and precision of input data is equally important. While static plant information, such as the location and physical plant characteristics are defined, the time-variant data including weather parameters, satellite images, meter data and project actuals keep changing. It is this time-variant data that needs to be accurately predicted for precise forecasting and scheduling. While many Indian states have adequate regulations for forecasting, scheduling and DSM, in several cases accurate tools are not available for precise predictions and each state has different DSM slabs. Thus, some developers see this as an additional cost. However, the reality is that these regulations are necessary for streamlining energy generation and grid security and will ultimately help in addressing the bigger problem of grid curtailment for renewable energy generators. Forecasting and scheduling is also important to bring renewables on an equal footing with conventional power systems. Going forward, along with more advanced and accurate systems, there is a strong need for uniformity in all states in terms of DSM penalty slabs to bring some respite to developers.
By Khushboo Goyal