My motivation to pursue
doctoral studies comes from my deep-rooted interest for energy systems coupled
with the desire to address the challenges arising with building energy systems
of the future. From the traditional hub and spoke architecture, power systems of
today have evolved into a complex cyber-physical system with heavy reliance on
intermittent distributed energy sources. Such changes necessitate development
of new intricate control & operational methodologies to ensure efficient,
reliable and resilient operation of this system. I am
applying to the Ph.D. program at Stanford University to address these
challenges using an interdisciplinary approach of data analytics, optimization
and energy system engineering.
The need for sustainable
power generation and distribution was the cornerstone for my career in
Electrical Engineering. My undergraduate studies equipped me with the
mathematical and electrical engineering fundamentals vital for building a
strong foundation. Research and internships done during undergraduate career
fueled my interest in energy systems that brought me to Georgia Tech to pursue
a Master’s degree in Electrical Engineering.
In the first year of
Master’s program, I focused my coursework and research in the areas of smart
grids, power electronics applications for power systems, control system design
and optimization. I worked with Professor Santiago Grijalva to build a smart home
energy management system that based its operations on real time electricity
prices. The purpose of this project was to schedule the consumption from the
grid, solar PV and battery storage in a pattern that minimized the total cost
of electricity consumption from the grid. I proposed that data processing and
optimization be done on a cloud server to help increase the system scalability
while expanding it to accommodate several houses supplied by an electric
utility. Working on this project gave me experience in hardware implementation
of the smart grid layers in real time, along with formulation & execution
of optimization problem using mixed integer linear programing.
I worked on a research
project on grid connected smart inverters with Professor Deepak Divan at
Georgia Tech’s Center for Distributed Energy (CDE). The goal of this project
was to understand the drawbacks of traditional inverters in smooth operation of
a grid with heavy inverter interfaced generation and address this issue by
adding smart functionalities to the traditional inverter. For this, I designed a
grid connected H-Bridge inverter and implemented various control layers to add
smart inverter functionality of voltage ride through and dynamic voltage
support. This entire system was then simulated for different types of grid
faults. Based on the observations, I further tuned the control loop to ensure
optimal system performance under any condition. Such smart functionalities
equipped the inverter to provide grid support by staying connected to the grid
during faults, thus speeding up the fault recovery process. This research helped unearth the shortcomings of smart
inverters based on the nature of the system they are connected to and their
asynchronized behavior in systems connected to multiple inverter interfaced
distributed energy resources.
My interest towards
leveraging data analytics for advancing energy systems stems from my graduate
internship with the Advanced Grid Analytics (AGA) group at Landis+Gyr (L+G), a
leading smart grids solutions company. My projects at L+G were focused on distribution
system data analytics, an emerging area under the smart grids domain. I worked on designing and implementing a robust algorithm
that processed the myriad of data available from smart meters for mapping
meters to their corresponding transformers. The algorithm was tested for
effectiveness on different distribution systems belonging to different
utilities and produced visible results. Distribution systems
are equipped with multiple Volt/VAR devices to ensure voltages at every point
in the network lie within the limits specified by American National Standards
Institute (ANSI). One of the pressing problems faced by the electric utilities
is the lack of knowledge on the operational performance of these devices. To
mitigate this issue, I designed a novel tool that utilized the AMI and system
topology data to closely monitor the device performance on a daily basis. The
algorithm factors in the physics that underlie the asset operation and is
computationally efficient, can seamlessly handle frequent system topology
reconfigurations and does not require installation of any additional sensors.
This asset health monitoring algorithm will be the first of its kind and will
be offered to the utilities as a software service or will be integrated as a
new module within L+G’s proprietary AGA platform.
As of today, this tool is under development and testing phase by
validating its results on data pertaining to different utilities and a
preliminary patent application is under review.
Interning at L+G was a steep learning curve in the distribution system data
analytics, utilizing various statistical techniques for analyzing data and
designing algorithms to efficiently process big data.
Working on data analytics piqued my curiosity to explore Machine
Learning. To familiarize myself with machine learning concepts and algorithms, I took few online
courses. I found the decision tree algorithm most suitable for enhancing the
transformer-meter mapping tool and am currently working towards implementing
it. I look forward towards integrating machine learning in my doctoral
Electrification, decentralization and digitization are merging
together to build the next generation electric grid. Some of the challenges
posed by such distributed system architecture are maintaining system stability
along with regulating voltage appropriately in the presence of intermittent
renewable energy sources, smoothly handling bidirectional power flow and
designing a robust distributed control algorithm implementation with minimal
communication infrastructure. With the rise of the age of Internet of Things
(IoT), sensors transmitting vital system data are being placed at different
points in the system. We still lack a systematic data-driven approach to
leverage this information for monitoring, operating and providing diagnostic
& prognostic evaluation of the grid.
As a doctoral student, I wish to be in an institution wherein I
can incubate an idea, dedicatedly work to bring it into reality using
scientific integrity and communicate it to all.
Stanford’s wide array of exceptional engineering programs and faculty are a
perfect match for my multidisciplinary research in Electrical Engineering, Statistics
and Computer Science. I find the work of Prof. Ram Rajagopal (Data driven energy system
modeling) and Prof. Dimitry Gorinevsky (Industrial IoT) interesting and
exceedingly aligned with my research interests. Stanford’s commitment towards
energy & sustainability, eminent faculty, multiple centers/labs for
interdisciplinary research and industrial collaboration makes it the most ideal
place to pursue my graduate career.
I am confident that my
unswerving dedication and strong academic background strengthened by relevant
industrial exposure will help me achieve my goals. After completing my Ph.D.
studies, I hope to pursue a career in industrial Research & Development
sector to contribute towards developing innovative solutions for sustainable
and efficient power generation and distribution.