Accurate Forecasting of Renewable Generation a Tough Task for Solar & Wind Developers

With India’s target to install 175 GW of renewables by 2022, the share of renewables is only going to grow, making the grid integration of these intermittent renewable resources into the country’s energy mix critical and at the same time challenging. Due to the intermittent nature of wind and solar energy, forecasting and scheduling are essential for stable and efficient grid management. According to Mercom India Solar Project Tracker, India has a cumulative solar installation of  33.8 GW as of Q3 2019.

Renewable energy generation, especially from wind and solar power, is variable, due to its dependency on the weather. The most critical scheduling input comes from weather forecasting agencies. Without accurate weather prediction, it is impossible for a renewable generator to submit accurate schedules. Energy forecasts are arrived at by combining plant availability with the weather forecasts for the location. Based on modeling, historical data, machine learning, and weather forecast, the generators need to predict their demand and generation for the next day. These details have to be sent to the scheduling entity, which could be the state or regional load despatch centers.

The day ahead scheduling process is the backbone of grid operation, both at the intra-state and inter-state network levels. The process aims at optimal scheduling and dispatch of the electricity for the next day based on the data provided by the generators.

With the seasonal variations, the generation frequency of renewables changes drastically in a shorter period, which the grid operators do not have the visibility into since until recently, the wind and solar generators were exempt from any forecasting and scheduling responsibilities. The grid operators are responsible for large-scale grid integration of solar and wind generating stations while maintaining grid stability and security.

At a panel discussion moderated by Mercom India at the Renewable Energy India (REI) Expo 2019 conference in September, one of the panelists enunciated how difficult forecasting and scheduling can be with an apt example. Earlier this year, the Gujarat government anticipated a storm on June 13, 2019, and nearly 250,000 people were evacuated from the location ahead of time. However, until June 16, 2019, there was no sign of the upcoming storm on Gujarat’s shores. Then the next day, on June 17, the storm did make its landfall, but it was 250 miles away from the location that had been forecasted before.

Ironically, neither Indian Space Research Organization (ISRO) nor international agencies, which are well-equipped with advanced technologies, could predict the exact time and date the storm would hit the Indian shores.

Now, on the other hand, renewable developers under the Deviation Settlement Mechanism (which includes machine learning, scheduling, and weather forecasting), are expected to predict accurately the exact generation of power in an area of 5X5 km in 15 time-blocks. If they fail to do so, they are liable for a penalty.

The central and several state electricity regulatory commissions have issued forecasting guidelines for the industry. These guidelines provide a methodology in case there is a deviation in the generation, and developers are required to pay penalties due to these deviations. This process is called deviation settlement mechanisms (DSM) through which developers compensate electricity grid infrastructure providers for errors in forecasting and scheduling of power generated by their projects.

The Central Electricity Regulatory Commission (CERC) has established a fee for errors based on a 15-minute time block and charges for deviation payable or receivable to/from regional deviation settlement mechanism DSM pools by renewable generators. If the error is more than 15%, then additional charges for deviation will be levied along with the fixed rate.

According to a report titled ‘GIZ-India Green Energy Corridors IGEN-GEC-Report on Forecasting, Concept of Renewable Energy Management Centers and Grid Balancing’, the current forecasting scenario in India for generation forecasting is still at its nascent stage.  However, it adds that the deviation due to “incorrect load forecast and conventional power projects not adhering to a schedule is higher than the variability due to the renewable energy sources.”

What do renewable developers have to say? 

According to a solar developer, Mercom spoke to; the CERC governs all the entities connected to the central grid and state electricity regulatory commissions (ERCs) issue regulations for the individual states.

‘So, each state has a different deviation settlement mechanism, and the percentage for the deviation is also different,” he said. Citing an example of Andhra Pradesh, the developer said that in this southern state of India, one could pool solar and wind separately, but it is not the same in other states. In Rajasthan, forecasting and scheduling allow pooling, but the SLDC was not in favor of implementing it.”

Regarding solar and wind, the latter generates more power during the night-time, whereas solar generates power during the daytime, but due to the variation in the wind, predictability is very difficult.

He said that in the case of solar, the predictability of the cloud movement is very difficult. There are several forecasting agencies, but they are unable to forecast the time when the cloud will cross a specific project. Typically when the cloud passes a solar project,  it reduces power generation by nearly 40 to 50%.

Asked how cloud movement affects the developers in forecasting, he said, “Penalties are affected because of the cloud movement. Even though solar forecasting and scheduling are a bit easy because you can easily track the curve with sun irradiation and temperature expectation, if heavy clouds are moving, then it will be difficult for a model to identify, and this leads to penalization.”

“Regulation is necessary, but penalization should be minimum,” he emphasized.

Aanal Pathak, a senior manager at REConnect Energy, said, “The charges levied for the forecasting errors caused are distributed across various bands. These bands differ from state to state and broadly fall under the range of 8% to 35%. Further, these charges will be in the range of ₹0.5 ($0.0007)/kWh to ₹1.5 ($0.021)/kWh.

Further, she said, “It is a known fact that maintaining grid stability and simultaneously providing power quality is a tedious task with its own legacy of issues. Variable generation from renewable energy such as wind and solar together pose significant technical difficulties for grid management. To gauge future projections of a higher share of renewables, it is required to have a good forecast and appropriate balancing action. Aggregation is an ideal solution, but it requires that the grid condition is favorable. The best-case scenario will only be implemented when the forecasting and scheduling practice is maintained at pooling sub-station levels as well.”

When asked about the non-standard regulations across several states and uncertainty of DSM charges, Pathak said, “The DSM payment cycle is very short in some states, especially where the settlement is on a weekly basis. De-pooling is a major concern for many generators as the regulations have left it to the generators and their qualified coordinating agencies (QCAs) to decide de-pooling methodology, and in many instances, there is no common consensus between generators on de-pooling methodology to be used”.

Pathak further added, “The five-minute time block may result in revisions being more accurate compared to the 15-minute time block, only if a greater number of revisions are allowed under the five-minute interval scheduling mechanism.”

Interestingly, during the Confederation of Indian Industry (CCI) conference held in Chennai in October 2019, the senior deputy general manager at the southern regional load despatch center, Power System Operation Corporation (POSOCO), Sharmeena Verghese, talked about the importance of grid discipline, forecasting, and scheduling in integration of renewable energy into the grid. She also added that the government is also contemplating migration from 15-minute time-block to 5-minute time-block for the deviation settlement mechanism.

Forecasting and Scheduling in other countries:

According to the ‘GIZ-India Green Energy Corridors IGEN-GEC-Report, the Anemos platform used by the Australian Wind and Solar Energy Forecasting Systems (AWEFS) and ASEFS, could be used as a reference for an India-specific forecasting system.

The Anemos-based Australian forecasting system allows for the incorporation of many renewable energy forecast models from various providers.

The AWEFS runs locally in the despatch centers, so it doesn’t have to depend on internet availability or get access to external prediction providers. The system also includes algorithms to deal with poor quality or missing measurement data.

The grid operation in Australia is similar to India, which is delicate due to the vast distances covered by transmission lines, and grid congestions play a significant role in handling renewable energy production, states the report.

Forecasting and scheduling have now become indispensable for efficient renewable energy integration for all the major renewable energy markets, and India is no exception. Considering that the future is higher renewable energy generation, forecasting of renewable energy seems to be a price that is inevitable to enable higher renewable offtake and minimize curtailment. While the importance of forecasting and scheduling is undisputed, the solar and wind generators expect clear and uniform regulations in the states and center and smoother implementation with room for flexibility in genuine cases where weather forecast has been close to impossible due to the impact of climate change.

Several developers have also mentioned that the deviation mechanism in India is skewed and difficult because of monsoons. Forecasting in other months is less of an issue, but the deviation charges are heavy during the monsoon season. A developer told Mercom that there is almost a ₹0.50-0.60 impact on tariff during monsoons.

“Deviation charges are adding to project costs, which will eventually get passed on. There needs to be some consideration for on-the-ground realities. By the time government agencies realize a regulation is flawed and rectify it, the industry loses another six months, and the goal of installing 100 GW of solar by 2022 gets that much more difficult,” said Raj Prabhu, CEO of Mercom Capital Group.

Image credit: LONGi Solar