Skip to content

Proof of Work

Applying NLP to building energy data since 2019; large language models since 2023.

Energy efficiency measures are the basic unit of building energy work: the recommendations in an audit, the alternatives in an energy model, the line items in a utility program. But every source names and groups them its own way. The same measure appears under different names across technical reference manuals, audit tools, standards and software, so data from one source can't be combined with another, and portfolio-level questions (what was recommended, what was installed, what saved energy) are hard to answer.

ASHRAE Research Project 1836-RP set out to fix that. Working with Dr. Amanda Webb, I compiled and text-mined about 3,500 measures from 16 sources, and we developed a three-level categorization system, a controlled vocabulary for naming, and 38 characterization properties. In 2024 we showed that large language models can match measures across naming systems. The custodians of two U.S. Department of Energy data standards have put the work to use: LBNL for the BEDES dictionary (2021), and NLR (then NREL) for BuildingSync's water conservation measures (2022). The work has been published and presented at the venues below.

Speaking the Language of Efficiency: Guidance for Naming & Categorizing EEMs · ASHRAE Journal 67(4)

Cover technical feature translating the 1836-RP research for practitioners.

Hot Air Podcast, Episode 28 · ASHRAE Journal

Categorizing energy efficiency measures, with Dr. Amanda Webb.

ASHRAE Homer Addams Award · ASHRAE

Awarded for my contribution to the completed ASHRAE research project 1836-RP.

Can LLMs Understand EEMs? Using Large Language Models To Manage Building Energy Efficiency Measure Data · IBPSA-USA SimBuild 2024 · Denver, CO

One of the first studies applying large language models to energy efficiency measure data.

Leveraging Natural Language Processing and Large Language Models to Understand, Categorize, and Standardize Building Energy Efficiency Measures · Ph.D. dissertation, University of Cincinnati

Ph.D. dissertation on NLP and large language models for energy efficiency measure data. Open access.

Developing a standardized categorization system for energy efficiency measures (1836-RP) · Science and Technology for the Built Environment 30(1), 1–16

The three-level categorization system, naming vocabulary, and characterization properties from 1836-RP.

Developing a Standardized Categorization System for Energy Efficiency Measures · ASHRAE Research Project 1836-RP Final Report

Final research report for ASHRAE Research Project 1836-RP.

AI Research for a LEED Course · University of Cincinnati

Research assistant to Dr. Hazem Elzarka on AI and machine learning topics for the LEED course he taught.

Seminar 68, ASHRAE 2023 Annual Conference · ASHRAE · Tampa, FL

Presented the 1836-RP research on categorizing and naming energy efficiency measures.

An EEM by Any Other Name: Best Practices for Naming Energy Efficiency Measures · ASHRAE Transactions 129, 100–108

Best practices for naming energy efficiency measures, with NLR's (then NREL's) water conservation measures as the case study.

What we talk about when we talk about EEMs: using text mining and topic modeling to understand building energy efficiency measures (1836-RP) · Science and Technology for the Built Environment 29(1), 4–18

Text mining and topic modeling of about 3,500 energy efficiency measures from 16 sources.

Water Conservation Measures for BuildingSync · University of Cincinnati · NLR (then NREL)

Ran NLR's draft water conservation measures through the 1836-RP method and tool, at the laboratory's request.

ASHRAE 1836-RP: Standardizing Energy Efficiency Measures · University of Cincinnati · ASHRAE

ASHRAE-funded research on a standardized categorization system for about 3,500 energy efficiency measures.

Graduate Research Assistant · University of Cincinnati

Graduate researcher on ASHRAE 1836-RP. NLP on building energy data from 2019, large language models from 2023.

Energy Poverty Research (RECS 2015) · University of Cincinnati · M.Eng. Capstone project

Analyzed Residential Energy Consumption Survey data to identify drivers of fuel poverty in the U.S.

Publications

  1. Speaking the Language of Efficiency: Guidance for Naming & Categorizing EEMs

    Khanuja, A., & Webb, A. (2025). ASHRAE Journal 67(4)

  2. Can LLMs Understand EEMs? Using Large Language Models To Manage Building Energy Efficiency Measure Data

    Khanuja, A., & Webb, A. (2024). IBPSA-USA SimBuild 2024 · Denver, CO

    Paper (IBPSA) →SimBuild 2024 proceedings →

  3. Leveraging Natural Language Processing and Large Language Models to Understand, Categorize, and Standardize Building Energy Efficiency Measures

    Khanuja, A. (2024). Ph.D. dissertation, University of Cincinnati

    Read on OhioLINK (open access) →

  4. Developing a standardized categorization system for energy efficiency measures (1836-RP)

    Webb, A. L., & Khanuja, A. (2024). Science and Technology for the Built Environment 30(1), 1–16

    DOI →Code and data →

  5. Developing a Standardized Categorization System for Energy Efficiency Measures

    Webb, A., & Khanuja, A. (2023). ASHRAE Research Project 1836-RP Final Report

    ASHRAE store listing →

  6. An EEM by Any Other Name: Best Practices for Naming Energy Efficiency Measures

    Khanuja, A., & Webb, A. (2023). ASHRAE Transactions 129, 100–108

    Table of contents, Transactions vol. 129 pt. 2 →Code and data →

  7. What we talk about when we talk about EEMs: using text mining and topic modeling to understand building energy efficiency measures (1836-RP)

    Khanuja, A., & Webb, A. L. (2023). Science and Technology for the Built Environment 29(1), 4–18

    DOI →Code and data →

Cited by

  • EPlusOut-Model Context Protocol (MCP): Large Language Model (LLM)-enabled simulation results

    Sweeney, Meyer & Goel (2026). IBPSA-USA SimBuild 2026, Minneapolis. Cadence OneFive and West Monroe Partners.

    Cites the SimBuild 2024 paper as prior work on AI-enabled workflows.

  • Analysis of performance improvement items for existing buildings using ASHRAE RP-1836

    Kim, Jang, An & Chae (2025). Korean conference paper (DBpia). Gachon University, Republic of Korea.

    Reviews and introduces 1836-RP to inform a future Korean framework for energy efficiency measures.

  • Exploring Large Language Model as an Interactive Sports Coach

    Lee (2025). arXiv:2509.26593. Yonsei University.

    Cites the SimBuild 2024 paper on using LLMs to map measures across taxonomies.

  • Application of Machine Learning Techniques for Predicting Heating Coil Performance in Building Heating Ventilation and Air Conditioning Systems

    Nassif, Dharmasena & Nassif (2025). Energies 18(9), 2314. University of Cincinnati.

    Cites STBE 2023 as related work.

  • Towards a Deeper Characterization of Positive Energy Districts Using Topic Modelling: Are Experts or Machines Better?

    Shah, Han & Zhang (2025). SSRN preprint. Dalarna University.

    Follow-on topic-modelling work that cites STBE 2023.

  • Perspectives of Machine Learning and Natural Language Processing on Characterizing Positive Energy Districts

    Han, Canli, Shah, Zhang, Dino & Kalkan (2024). Buildings 14(2), 371. Dalarna University (Sweden) and Middle East Technical University (Türkiye).

    Discusses the STBE 2023 text-mining and topic-modeling method in a review article.