ComStock / ResStock
Website: https://www.nrel.gov/buildings/comstock, https://resstock.nrel.gov/
The ComStock and ResStock analysis tools are helping states, municipalities, utilities, and manufacturers identify which building stock improvements save the most energy and money.
– Modeled national energy use of commercial buildings by using statistical sampling methods, large-scale building energy simulations, and high performance computing
– Led the development of ComStock surrogate models using deep learning algorithm for uncertainty quantification
– Applied ComStock to analyze the impact of COVID-19 mitigation strategies from ASHRAE Epidemic Task Force on national commercial building energy use and load profile. Applied ComStock to analyze whether and how 100% renewable energy goal can be achieved in 2045 in Los Angeles
Bridging Gaps Towards a Sustainable Energy Future by Applying Human Factors in Transportation-Building Integration
Sponsor: RII 2024 UArizona National Labs Partnerships Grants.
Spearheaded/lead by Prof. Alyssa Ryan and co-lead by Prof. Liang Zhang, this award-winning project is set to deliver human behavior models, particularly in the integrated domains of building and transportation energy use modeling, as part of the broader endeavor toward decarbonization.
Enhance Data Infrastructure for Building Decarbonization with Large Language Models
Sponsor: Technology and Research Initiative Fund (TRIF) Impact, UArizona
The project aims to address the inefficiencies in leveraging big data for building decarbonization studies by utilizing Large Language Models (LLMs) for data management and interpretation. These LLMs are designed to streamline the handling of diverse datasets, organize data in contextually sensitive ways, and simplify access to complex information for users. The project also involves creating a comprehensive data corpus related to building decarbonization and developing a scalable method called ‘semantic modeling’ for advanced greenhouse gas emission analysis. The initiative has far-reaching societal and educational implications, as it aids in fast-tracking decarbonization in the building sector by offering stakeholders easy access to critical data.
Empirical Validation of Energy Simulation
Website: https://www.energy.gov/eere/buildings/empirical-validation-energy-simulation-frp-iunit-and-nzertf
– Designed validation-grade experiments in iUnit, the standardized experiment chamber at NREL, to quantify the systematic bias and uncertainty of building energy modeling tools including EnergyPlus and TRNSYS
– Spearheaded a framework for sequential calibration of timeseries building energy modeling
Sensor Impact Evaluation and Verification on Fault Detection and Diagnostics (FDD)
Website: https://www.energy.gov/eere/buildings/articles/sensor-impact-evaluation-and-verification
– Streamlined a systematic framework to evaluate sensor impact on the building energy efficiency and thermal comfort from FDD perspective. Published a Python package in PyPi to automate the analysis for distribution
– Collaborated with Oak Ridge National Laboratory on developing EnergyPlus faulty building models based on the Flexible Research Platforms
– Co-led the project in staff planning, deliverable quality control, and communications
Py-CoSim: An Open-Source Python-EMS-Based EnergyPlus Co-Simulation Platform
Py-CoSim is an innovative, open-source co-simulation platform developed to streamline and enhance building energy modeling by integrating EnergyPlus with Python-based control and analysis tools. Py-CoSim addresses the high technical barriers traditionally associated with building energy co-simulations, which often require expertise in multiple complex simulation tools.
Intelligent Campus Program
Video: https://www.youtube.com/watch?v=0LIbq2570Sw
NREL’s Intelligent Campus program enables researchers and energy managers to study the integration of renewable energy and energy efficiency technologies, make operational decisions that minimize emissions or enhance resiliency, and support a variety of research projects.
Task Description: Developed a probabilistic sequence-to-sequence deep learning algorithm to predict building energy load and PV power generation for (1) advanced building and community operation control and (2) fault detection and diagnostics.