The government has set ambitious targets for renewable energy, including a capacity addition of 60 GW of wind by 2022. The continuous growth of wind power in India is making it increasingly important to obtain accurate wind speed and power forecasts, to ensure smooth integration with the grid and aid in proper scheduling of the power generated. Rapid changes in wind speed and direction can often cause wind farms to deviate significantly from scheduled deliveries and put grid reliability and stability at risk. Accurate wind speed forecasts are necessary to schedule despatchable generation in the day-ahead electricity market, and to plan the operation of other power plants.
Policymakers have been working on improving the regulatory framework for forecasting and scheduling in India. The concept of forecasting and scheduling of wind power generation was first introduced in the country through the Indian Electricity Grid Code [IEGC], 2010 by the Central Electricity Regulatory Commission (CERC). Last year, the CERC and the Forum of Regulators (FoR) came up with regulations to streamline the forecasting and scheduling of wind power.
While the forecasting models have evolved significantly, they have not yet matured in India. Most wind farm owners have been forecasting generation at an individual unit level only, with the objective of keeping up with the requirements of the regulations. However, a project was initiated in Tamil Nadu last year, which aims to provide forecasts for all wind farms in the state and help the state load despatch centre to schedule evacuation in an optimal way.
These developments have taken wind forecasting and scheduling to the next level in India and going ahead, are expected to provide a firmer basis to the country in achieving its long-term wind energy targets.
Need and challenges
In order to balance electricity supply and demand, grid operators schedule power generation one day in advance with power plants under their authority, committing to a 15-minute timetable of when to run and supply electricity, with some deviations allowed. Through this, grid operators can not only prioritise the generation from cleaner resources, but also plan the operation of conventional plants and the required backup generation and transmission capacity. Forecasting, in this regard, helps in achieving higher levels of accuracy in estimating generation from plants, scheduling wind energy and making grid management more effective. The recent increase in wind generation and future projections of a higher share of wind in the total generation portfolio make wind forecasting even more essential for the electricity grid.
The main challenges in wind forecasting are the intermittent nature of wind and the lack of sufficient data. Since wind is a variable and unpredictable energy source, it is relatively difficult to estimate the quantum of power that wind plants will generate the next day. Any fluctuation in wind can cause big fluctuations in power generation. Wind is typically created by small pressure gradients operating over large distances, which are difficult to measure. Wind is also influenced by turbulent processes, air density, temperature, etc., all of which need to be accounted for to ensure better forecasts. In this regard, while weather observations provide some information, there may never be enough data. Moreover, forecasting requires historical generation data as well, which, in the absence of SCADA systems, is either not available or may not be accurate where recorded manually.
In addition to the IEGC, 2010, which was the first step to overcome the difficulties related to the management of infirm wind power, the Renewable Regulatory Fund mechanism was implemented in 2013. This, however, faced many issues, including non-coverage of old plants (commissioned prior to May 2010) and higher financial risks. Moreover, it could not achieve the objective of reducing the deviations in the scheduled and actual generation of wind. Hence, the penalties under the mechanism were suspended by the CERC. However, wind farms were still required to submit their estimates of generation.
In a move to address these issues, the CERC notified the Framework on Forecasting, Scheduling and Imbalance Handling for Variable Renewable Energy Sources (Wind and Solar) at interstate level in August 2015, and introduced amendments in the IEGC, 2010 and the Deviation Settlement Mechanism and Related Matters Regulations, 2014. These regulations have made it mandatory for all existing and upcoming wind/solar power generation plants and load despatch centres to undertake forecasting and scheduling of energy generation. According to the new mechanism, penalties are imposed for a deviation of more than +/-15 per cent of scheduled generation.
The new regulations have also increased the maximum number of daily revisions from 8 to 16, modified the error calculation methodology to include “available capacity” and specified the deviation charges to be determined as a percentage of the power purchase agreement (PPA) rate, among other provisions. With metering, telemetry, and weather data access at each wind farm level as prerequisites under the new regulations, these basic building blocks that were missing till now are likely to lead to better integration of generation from these plants with the grid.
Following the CERC regulations, the FoR formulated the Model Regulations on Forecasting, Scheduling and Deviation Settlement of Wind and Solar Generating Stations at the State Level in 2015, which cover all existing and upcoming wind and solar power producers in India. In line with the FoR’s model regulations, the electricity regulatory commissions of states including Karnataka, Tamil Nadu, Madhya Pradesh and Rajasthan have announced draft regulations on wind/solar forecasting and scheduling. Other state electricity regulatory commissions are likely to follow a similar path which will result in all wind and solar plants across India being covered.
How forecasting works
Forecasting models have evolved from the use of averages of the past few days/week, to correlation-based, to more dynamic-artificial neural networks-based real-time data-centric models. The key inputs in forecasting include static data (latitude/longitude details, past generation data, turbine power curve, dimensions of windmill, etc.) and dynamic data (turbine availability, wind speed and direction, temperature, inverter availability, etc.). The weather data used in forecasting can be obtained from various international and national organisations. Utilities may also have their own local weather monitoring systems. All this data is fed into a forecasting algorithm to forecast hour-ahead/day-ahead/week-ahead generation.
Renewable energy management systems offer a one-point solution in this regard. They consist not only of wind farm data acquisition and management systems to collect real-time data from wind farms, but also a centralised short-term wind power prediction tool. They provide meteorological data, real-time generation data and wind data, and also forecast generation based on this data.
Today, there are a number of private players who assist power plants and utilities in forecasting and scheduling generation. They also provide support in data collection, coordination with load despatch centres and energy accounting.
While the current forecasting technologies used in India are far from perfect, they are highly cost-effective, especially when compared to no forecasting at all. So, as India looks to add wind capacities, it must also work harder on how to forecast, schedule and absorb it. While a fair bit has been done in this regard, there is massive scope for the development of forecasting models that are much more accurate and technologies that are more cost effective. There is scope for the development of better models and more data to be used in the models.