Hemanth Hariharan
Hemanth Hariharan
Welcome! I am Hemanth Hariharan, a recent graduate of the master’s program in Sustainable Design and Construction at Stanford University. I did my undergraduate studies in Civil Engineering at the Indian Institute of Technology, Madras, and am fascinated by the natural and man-made world. In addition to my passion for the sciences, I am a Tabla player (Indian percussion drummer), a content creator for my YouTube channel Scientific Blunders, and an avid sportsperson!
Resume / LinkedIn
Projects (Click the project title to see more!)
Energy analytics and financial modeling
During my summer internship with COI Energy, I worked on studying historical energy consumption data for customer selection and financial modeling to determine customer payback periods. My model generates data, computes payback period for multiple scenarios (Monte Carlo analysis), and can also be used for specific buildings.
Deep learning model for carbon storage to study brine leakage
As a research assistant at the Stanford Center for Carbon Storage, I worked with Dr. Sally Benson and Dr. Sarah Saltzer on developing an interactive tool to study brine leakage into drinking aquifers due to pressure buildup from CO2 storage. My work uses CCSNet, a deep-learning modeling suite, to run hundreds of thousands of trials on the backend, and develop a nomogram and phase diagram to compute the ‘Area-of-review’ for various permeabilities, injection rates, thicknesses, depths, and brine densities.
Machine Learning for Wind Turbine Output Prediction
As part of a team of 3 in Dr. Andrew Ng’s course ‘Machine Learning’, we performed an exploratory data analysis on a wind-energy SCADA dataset comprising features including wind speed, direction, and energy generated. We used an ensemble machine learning method comprising ARIMA, gradient-boosted decision trees (XGBoost), convolutional neural networks (CNN), and long short-term memory (LSTM) neural networks to perform time-series forecasting of wind energy.
Crane Data Analysis
As part of a team of 3 in Dr. Fischer’s course ‘Managing Fabrication and Construction’, we performed data analysis and visualization of crane data from two building projects. Our goal was to determine production rates and pour cycle times for key activities, assess productivity for both projects, and propose recommendations for future projects.
ERCOT Interconnection Queue Analytics
During my internship with Cypress Creek Renewables, I analyzed and visualized trends in publicly available historical data from the ERCOT interconnection queue and used machine learning to predict study times and project outcomes. My research helps solar developers make more informed decisions on where to greenfield projects.
Low Carbon Fuel Standards - 24/7 carbon-free charging
I conducted an independent study with my MS advisor, Prof. Rajagopal on decarbonizing Stanford University’s electric bus fleet. I applied California’s Low Carbon Fuel Standards (LCFS) to calculate potential monetary benefits for Stanford University through both charging and capacity pathways.
Systems modeling of growth and pollution of a city
As part of a team of 4 in Prof. Rishee Jain’s course ‘Urban Systems Engineering’, we created a systems model to predict the variation of certain urban parameters (pollution, building, coal, and solar resources) with time. A sensitivity and policy analysis was also conducted to model the impact of policy interventions on the system.
Developing a roadmap for a 100% wind-water-solar (WWS) California
In Prof. Jacobson’s course ‘100% clean, renewable energy and storage for everything’, I quantified end-use demand in California from EIA-published data and converted it to WWS energy to arrive at an energy mix containing rooftop and utility-scale solar, onshore and offshore wind, and existing geothermal and hydroelectric power.
Financial modeling of HVDC undersea cable project
In Prof. Bennon’s course ‘Global Project Finance’, we worked on building a financial model of an HVDC cable project to utilize excess renewable energy capacity. We performed sensitivity analyses on construction timelines, exchange rate fluctuations, operating costs, and potential outages and made recommendations to improve project valuation. Monte Carlo Analysis was also performed using the ChanceCalc library to estimate the probability distribution of the IRR (Internal Rate of Return) and ROE (Return on Equity).
Life Cycle Assessment of Carbon Condo vs. US average condo
As a team of five, we conducted an LCA comparison of a carbon nanotube (CNT) building with the US average building. The goal of the LCA was to provide city and building professionals with quantitative decision-making tools for the implementation of CNT as a building material to make efficient use of energy resources, eliminate pollution, and reduce climate change.
Net-zero energy home design in Wolfeboro, NH
As a team of two, we redesigned an existing summer home in Wolfeboro, NH into an NZE home using passive solar design strategies, rooftop solar, and a geothermal heat pump. An iterative process was employed to optimize the R-values of the building envelope and minimize shading losses. We achieved a projected NPV of savings of ~$30k over 20 years.
Energy@Stanford & SLAC Summer Course
Attended summer conference (week-long multi-disciplinary session on Energy) and presented solutions for a < 2C future using the En-ROADS simulator.
Machine learning to characterize the behavior of masonry spandrels
As part of my B.Tech thesis at IITM with Dr. Arun Menon, I developed a tool for the parametric study of existing masonry strength formulations. I characterized the lateral resistance of a masonry wall based on geometry, boundary conditions, strength, and stiffness. Parametric analysis was done through non-linear FE modeling on TNO-DIANA and I formulated a predictive equation of spandrel strength using the machine learning toolbox of MATLAB.
Inelastic buckling of concrete-filled tube (CFT) columns
During my summer research internship at Purdue University with Dr. Amit Varma, I developed a GUI on MATLAB with a pre-processor and a post-processor for an existing MATLAB code to analyze CFT columns. Column curves and interaction curves were generated iteratively and I used higher-order interpolation to develop a tool that provided column designs. In addition to my computational research, I also worked at the Bowen Lab in setting up test specimens of composite walls.
