Until just a few years ago, millions of people in rural India did not have reliable access to electricity, according to a report issued by the World Bank (link resides outside of ibm.com). There was no singular reason for the lack of electrification. Rather, it was a combination of issues — an aging infrastructure, distribution inefficiencies, an ever-growing population and a relentless gap between supply and demand — that left households in the dark.
To combat this problem, India launched its 24/7 Power For All program in 2016, aimed at providing every household, business and other electricity-consuming entity with round-the-clock access to power. As a joint initiative between the central and state governments, the program also imposed new and strict regulations related to energy supply and demand.
Rachit Kumar Agarwal is the Managing Partner at Mercados Energy Markets India Private Limited, India (Mercados EMI), a consultancy firm that specializes in solutions for the energy sector. He recalls the challenge: “States had to procure power beforehand. This meant they had to know precisely how much power was required every day, every hour and in 15-minute time blocks. It was a problem.”
For state governments, the key to balancing supply is accurately predicting demand. But, until recently, officials calculated predictions manually using spreadsheets. They analyzed years’ worth of historical demand data, yet calculations were off. Predictions came in too high or too low. Lacking precision, distribution companies bought too much power, leading to waste, or too little power, contributing to financial losses.
Enhanced Predictions
Predicts electricity demand with error margins of 1.8% – 6% compared to 10% – 15% in the past
Enhanced Outcomes
Frequency of weather updates 15 minutes out to seven hours
The problem only grew more complicated when India’s electricity regulator, the Central Electricity Regulatory Commission (CERC), narrowed the operating frequency band for the national grid to the range of 49.9 hertz to 50.05 hertz. “To maintain the frequency within this range, the utilities have to accurately estimate the demand,” explains Mr. Agarwal, “and then draw the exact amount of power based on that demand. If you under-draw or over-draw, there may be penalties.
“The bottom line is you have to know how much power you will consume so that you buy only as much as you require.”
In early 2020, the states of Uttar Pradesh and Bihar, located in a region of extreme summers and winters and having some of the largest consumer bases in India, turned to Mercados EMI for help. In addition to seeking a solution that would infuse AI into the prediction process, officials wanted to better understand all of the factors that impact demand — not just historical data.
Turns out, predicting energy consumption has a lot to do with the weather.
“What came out of our detailed discussions with the utilities over six months was that accurate weather data plays a very important role in predicting demand,” recalls Mr. Agarwal. Armed with new insights, Mercados EMI created an AI-based demand forecasting solution — jouleOS® — based on Environmental Data Services, sourced from The Weather Company®, an IBM Business. The technology helps forecasters predict day-ahead energy requirements.
The IBM solution has since expanded. Now, organizations can access Weather Data APIs through the IBM Environmental Intelligence Suite, a suite of applications that combine and integrate The Weather Company Data APIs with geospatial analytics, dashboards for visualization and alerting capabilities.
To build its forecasting solution, Mercados EMI first developed a model to accurately understand demand. The model combined the state’s historical demand data with historical weather pattern data from The Weather Company History on Demand data package.
“We already had the demand data from previous years. What we required was historical weather patterns,” adds Mr. Agarwal. “So, for example, if demand is 5,000 megawatts at a particular instant, what were the weather parameters at that instant? And when the weather parameters changed, how did the demand change?”
To answer these questions and, ultimately, meet the government’s data resolution requirements, Mercados EMI applied the forecast engines of The Weather Company Enhanced Forecast data and the IBM Environmental Intelligence Suite platform. “Weather data is where we gained immense help from IBM,” adds Mr. Agarwal. “It not only provided us the return in terms of time — hourly and at 15-minute increments out to seven hours — but also within a geographic radius of 500 x 500 meters.” The 15-minute forecasts represent a significant improvement: previously, the states relied on weather data with a timeframe of three hours from publicly available open-source databases.
The models, combined with Mercados EMI’s proprietary AI engine, help distribution companies better forecast electricity demand a day ahead. Mercado EMI now offers its jouleOS technology under its Power Portfolio Optimization suite of solutions.
Today, both the states of Uttar Pradesh and Bihar use the system, with Madhya Pradesh and a state in Southern India soon to follow.
The accuracy of the state’s demand predictions improved almost immediately. When the process was manual-based, error margins ran 10% – 15%. With IBM’s weather data as a key input, Mercados EMI achieved an initial error rate of only 5% – 6%. On some days, it’s as low as 1.8%, equating to a 98.2% accuracy rate.
Mr. Agarwal points to weather data for the solution’s success. “We have other inputs, like holidays, special days and Sundays. But it’s predominantly weather — the coefficients of the demand equation showed that weather plays a very important role in forecasting.”
In addition to day-ahead demand forecasting, Mercados EMI uses weather data from The Weather Company for real-time demand forecasting. The real-time market was launched by the power exchanges in India in June 2020 to trade power a few time blocks in advance. In response, Mercados EMI successfully deployed India’s first real-time market engine for the Uttar Pradesh Power Corporation in January 2021. The engine helps distribution companies in Uttar Pradesh predict demand at specific delivery times and to buy and sell electricity in the real-time market.
“It means if at 10:17 AM I want to predict demand at 2:00 PM, I can,” concludes Mr. Agarwal. “It’s demand forecasting at a granular level, up to seven hours into the future with very sharp accuracy. This will be a first-in-India solution and we have quoted IBM as our weather source, officially.”
Delhi-based Mercados EMIExternal Link (link resides outside of ibm.com) is an internationally-recognized consultancy firm that specializes in solutions for the electricity, oil and gas, and renewable energy sectors. Its services and expertise cover the areas of power market design, demand forecasting, power portfolio optimization, policy and regulation, utility management, markets simulation, business strategy and grid modeling. Mercados EMI was founded in 2008.
Footnotes
© Copyright IBM Corporation 2021. IBM Corporation, The Weather Company, New Orchard Road, Armonk, NY 10504
Produced in the United States of America, January 2021.
IBM, the IBM logo, and ibm.com are trademarks of International Business Machines Corp., registered in many jurisdictions worldwide. Other product and service names might be trademarks of IBM or other companies. A current list of IBM trademarks is available on the web at “Copyright and trademark information” at www.ibm.com/legal/copytrade.shtml.
The Weather Channel®, The Weather Company®, and Weather Underground® are trademarks or registered trademarks of TWC Product and Technology, LLC, an IBM Company.
This document is current as of the initial date of publication and may be changed by IBM at any time. Not all offerings are available in every country in which IBM operates.
The performance data and client examples cited are presented for illustrative purposes only. Actual performance results may vary depending on specific configurations and operating conditions. THE INFORMATION IN THIS DOCUMENT IS PROVIDED “AS IS” WITHOUT ANY WARRANTY, EXPRESS OR IMPLIED, INCLUDING WITHOUT ANY WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND ANY WARRANTY OR CONDITION OF NON-INFRINGEMENT. IBM products are warranted according to the terms and conditions of the agreements under which they are provided.