Learning and Decision-Making for Energy Storage Systems under Renewable Uncertainty

Title of the Talk: Learning and Decision-Making for Energy Storage Systems under Renewable Uncertainty
Speakers: Vivek Deulkar
Host Faculty: Dr.Abhijit Das
Date: Dec 30, 2025
Time: 11:00 am.
Venue: CS-105

Abstract: Harnessing renewable energy comes with the challenge of dealing with its intermittencies. The generating patterns of renewable generation, like wind or solar, are highly intermittent, which makes the reliable operation of meeting load difficult. This talk addresses the modeling of these uncertainties in renewables along with the use of battery storage for a target reliable operation of meeting load. Specifically, the uncertainties are characterized via the Markov model, while the battery storage is used to store the excess renewable generation, which then can be used to serve the load in times of deficit. The mathematical model is analyzed using Markov-modulated fluid queue (MMFQ), which is subsequently used to estimate the battery size required for target reliability of meeting load.

The first half of this talk focuses on the planning timescale, where we address the issue of optimally sizing the battery storage, subject to reliability constraints. The second half focuses on the operations timescale, where we consider the problem of optimally scheduling the battery storage (in the presence of renewable uncertainties). Specifically, a dynamic energy exchange mechanism between two battery-equipped renewable generators is proposed which results in improvement of loss of load for both the agents operating these generators. The last segment of this talk is about efficient cycling of multiple battery units via reinforcement learning, with the objective of reducing the cycling degradation of battery units.

Speaker’s Bio: Vivek Deulkar is an Assistant Professor at Plaksha University, Mohali, Punjab, where he is affiliated with Robotics and Autonomous Systems as well as with Indorama Ventures Center for Clean Energy. Prior to that, he was a Postdoctoral Research Associate at Manning College of Information and Computer Sciences, University of Massachusetts, Amherst. His research focuses on sequential decision-making under uncertainty, with an emphasis on stochastic modeling, optimization, and learning-based methods for energy and sustainability applications, particularly in the context of renewable integration and battery storage. He did his Ph.D. in the department of Electrical Engineering at the Indian Institute of Technology Bombay. His Ph.D. research was in the space of smart grid managing uncertainties in renewable generation using battery storage, drawing on tools from stochastic modeling, applied probability, sequential decision-making, optimization, control, and reinforcement learning. Vivek has done an M.Tech in Control and Computing from Electrical Engineering, IIT Bombay, and a B.Tech in Electronics and Communication Engineering from Visvesvaraya National Institute of Technology (VNIT), Nagpur. Vivek also has experience working with the Engineering Research Centre, Tata Motors Ltd., Pune, where he worked in the Auto-Electric Embedded Systems design group towards software development of embedded controllers, which are used for various automotive applications.

Meeting link: https://meet.google.com/qav-paas-qmb