AI-ML models can help India’s discoms be sustainable in scorching summer

Making discoms sustainable must be balanced by meeting peak load demand arising from abrupt weather changes. AI-ML prediction models can help.

As India prepares to survive another scorching summer with predictions of maximum temperature going above 47 degrees Celsius, the heat is on for its power supply grid. 

India’s discoms face a dual challenge: meeting changing peak load demand while also moving towards a decarbonisation target set by the central government as part of their renewable energy purchase obligations (RPOs) for purchasing power from renewable energy sources.  

Last year, the peak demand of Tata Power, a private power distribution company in Delhi, crossed 2250 MW post noon on a hot summer day. A sudden change in peak load demand means urgently buying power to meet that demand and ensure uninterrupted power supply. 

To predict such demand surges, discoms need a state-of-the-art peak load forecasting method for each consumer, which is also responsive to daily weather changes at a fifteen-minute interval. 

Predicting power demand 

Recently, researchers conducted a temperature, rainfall and precipitation sensitivity analysis of peak load consumer demand for Chennai’s Tamil Nadu Power Distribution Corporation Limited (TNPDCL) based on the last 100 years of weather data. 

The analysis showed that with a one percent change in temperature, rainfall and other daily weather variables, the peak load demand for a fifteen-minute interval could change by more than 2.5 percent. 

India’s discoms are not yet prepared for such sharp changes in peak load demand due to future climate variability in most Indian cities. 

Across the world, accurate prediction models are being developed for discoms and similar entities using AI-ML modelling structures to help them meet unexpected power demand. 

In India, some research centres have developed their own AI-ML based synthetic load model using a fifteen-minute prediction of consumer peak load demand. Within the model, each fifteen-minute demand of different consumers has been taken as an input for each day to predict peak demand. Such models can help discoms prepare for their power procurement in advance and plan for the short and long term by knowing changes in peak demand.

The renewables dilemma

The country’s discoms must also move from a 22 percent renewable energy purchase portfolio to more than 42 percent by 2047. The challenge is that peak load demand—which is constantly changing due to weather conditions—must be managed by buying more renewable energy-based power as per their RPOs. 

This requires an increase in the financing and investment requirements of India’s public and private discoms. 

With state discoms in financial crisis and receiving bailouts, the challenge is how to mobilise the extra funds. This can be done by imposing prudent Goods and Service Tax (GST) or through direct fund transfer from the government. 

Estimates emerging from a predictive model by this researcher and their colleagues show that private discoms may have to manage a three-fold rise in investment needs for an RPO increase of two times in 2032-35. 

The challenge is whether in the short run discoms can enhance their renewable energy procurement portfolio while meeting a 15-minute rise in peak demand owing to sudden temperature changes in cities. 

This means enhancing discoms’ ability to move towards renewable energy sources while balancing rising peak demand and meeting the minimum return on investment for procuring power. 

Private discoms like Tata Power are already meeting their renewable energy procurement obligations by buying power from renewable energy generation sources through existing power purchase agreements (PPAs). However, this is not a long-term solution as these PPAs will lapse by 2032, and return on investments need to be sustainable in the long term. 

The way ahead

To meet new renewable energy demand from consumers or discoms, energy needs to be stored in batteries that can then be supplied to decentralised energy demand centres of discoms. 

If the country’s power grid does not allow new renewable energy power to be transmitted due to unpredictable power supply from solar and wind, battery storage solutions have to be worked out at a decentralised level. 

Positive stories are already emerging from the states of Rajasthan and Gujarat. Some discoms have been able to meet peak demand using solar energy stored through battery storage systems during daytime, which gets reused during peak load in the evening. However, new RPO commitments must come for private discoms beyond 2032. 

A heatwave-led summer can add to discoms’ worries in the short term. A well-processed AI-ML model that visualises a world where different degrees of peak load demand can arise from climate variabilities can help Indian discoms meet their RPOs and plan for resource adequacy in the short and long term. 

Peaking discom decarbonisation is a matrix of parallel realities for all current discoms of India now. Intelligent prediction scenario-based prediction models can help to solve such conflicting realities with an eye on the past, present and future. 

Anandajit Goswami is a Research Fellow at the Ashoka Centre For People Centric Energy Transition (ACPET), Ashoka University.

Prodyut Mukherjee from EnGenuity and Renuka Sane of TrustBridge have co-authored the research that led, in part, to this article. Vaibhav Chowdhary, Director, ACPET, has reviewed the article.

Originally published under Creative Commons by 360info™.