Himanshu Sharma is a Research Engineer at PNNL. He is currently working to develop and integrate advanced optimization and control methods for energy and power system applications. Earlier, he worked on developing a data-driven Perron-Frobenius operator-based framework for sensor placement for monitoring indoor air quality under uncertainty. His work is geared toward solving challenging problems of complex-physical systems such as energy & smart buildings by leveraging state of the art simulation and data-driven methods.

Previously, Himanshu was a Post-Doctoral Scholar at Argonne Leadership Computing Facility ( ALCF) Argonne National Laboratory, whereas a member of the data science team, he primarily worked on distributed training of Bayesian Neural Network on high performance computing cluster to analyze computational challenges and limitations for training uncertainty capturing Neural Networks. He carried his Ph.D. under the guidance of Dr. Baskar Ganapathysubramanian and Dr. Umesh Vaidya. His Master thesis was conducted under the guidance of Dr. Murali Damodaran at IIT-Gandhinagar.


  • Machine Learning
  • Dynamical Systems & Controls
  • Numerical Optimization
  • Computational Fluid Dynamics
  • Uncertainty Quantification
  • Building energy systems


  • Ph.D in Mechanical Engineering, 2019

    Iowa State University

  • M.Tech in Mechanical Engineering, 2014

    Indian Institute of Technology, Gandhinagar

  • B.E in Mechanical Engineering, 2012

    Devi Ahilya Vishwavidyalaya


Integrating Deep Learning with Computational Fluid Dynamics Solvers

The work aims to demonstrate capabilities doing in-situ co-simulations strategies of Machine Learning and Computational Fluid Dynamics Solver (OpenFOAM)

Bayesian Neural Networks Distributed Training Performance Analysis at Scale

The aim of the work is to perform distributed training of Bayesian Neural Networks on High Performance Computing Clusters.

Indoor Air Quality Sensor Placement Accounting for Airflow Uncertainty

The sensor placement algorithm accounting for building airflow uncertainty based on Perron-Frobenius Operator.

Construction of Perron Frobenius Operator from Computational Fluid Dynamics (CFD) Data

A data-driven approach is developed to construct the Perron-Frobenius operator on-line using the OpenSource CFD tool OpenFOAM.

Surrogate Modeling for Integrating High-Fidelity Data with Energy Simulations Tools

The work demonstrate the co-simulation strategy of coupling Building Simulation Tool(FATM) with surrogate model constructed with high fidelity CFD data.

Recent & Upcoming Talks

High Fidelity Data Surrogates for Enhancing Co-Simulation Capabilities of Energy Simulation Tools

Energy modeling tools are used by engineers and researchers to evaluate the impact of different design features, products, and …

Scientific Efforts in Response to COVID-19 Pandemic

The talk aims to provide details about the on-going scientific efforts at National Labs, Universities in United States to combat …

Integrating Data-Driven and High Fidelity Computational Models for Buildings Energy Systems

Residential and commercial buildings account for 41% of total energy consumption in the United States. Optimizing energy consumption …


  • 902 Battelle Blvd, Richland, WA 99353
  • Monday-Friday 9:00 to 6:00 (PST)