About

Do Gyun Kim is an Assistant Professor of Visual Communication Design at the Rueff School of Design, Art, and Performance at Purdue University. He is a graphic designer working on typography, motion graphics, human rights design, and experimental texture. Kim received his M.A. in Graphic Design from Oklahoma City University and holds an M.F.A. in Graphic Design from Oklahoma State University. He has taught undergraduate graphic design courses in Mississippi and Missouri, and has lectured on digital design as a graduate instructor at Oklahoma State University. Kim's passions include designing for social change and history, and he believes that design has the power and ability to play a critical role in society. His research process is based on texture and revolves around themes of politics, human rights, and historical events. His design works have received over 20 graphic design awards in national and international competitions in the USA and Korea, and his work has been published in the USA, Germany, and Korea.

Research topics

• Computer Science
• Artificial Intelligence
• Control engineering
• Engineering
• Risk analysis (engineering)
• Software engineering
• Reliability engineering
• Systems engineering

Selected publications

• Virtual refrigerant charge sensor for variable-speed heat pumps based on feature selection

  Applied Thermal Engineering · 2026-02-18 · 1 citations

  article
  PublisherDOI

• Hybrid modeling approach for better identification of building thermal network model and improved prediction

  SSRN Electronic Journal · 2025-01-01

  preprintOpen accessSenior author
  PublisherDOI

• Reinforcement Learning Control for Buildings Co-Optimizing Energy, Comfort, and Indoor Air Quality: An Annual Assessment

  2025-08-25

  articleSenior author

  Efficient control of Heating, Ventilation, and Air Conditioning (HVAC) systems is crucial for optimizing energy use and maintaining indoor comfort in buildings. Traditional control methods, such as PID control, cannot handle energy use trade-offs among multiple components in the building energy system at a supervisory level. Reinforcement learning (RL) presents a promising solution, offering adaptive and data-driven control strategies that optimize performance over time. However, RL also faces several challenges, including the conflicts encountered in co-optimizing energy savings, occupant comfort, and indoor air quality, and the requirement for extensive interactions with the environment in training. We proposed a flexible simulation platform that integrates a hybrid model for RL training and designed an RL agent to control the entire central HVAC system, focusing on co-optimizing energy consumption, thermal comfort, and indoor air quality (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{CO}_{2}$</tex> and PM2.5 concentrations). Finally, we evaluated the RL agent's performance over an annual cycle. Our findings indicate that the RL agent can effectively manage the HVAC system with 14.7 % energy savings annually and balance multiple objectives, which demonstrates significant potential for improving HVAC system control and sustainability in buildings.

  PublisherDOI

• Domain-aligned LLM framework for trustworthy scientific Q/A via query reformulation retrieval-augmented generation

  ChemRxiv · 2025-11-27

  articleSenior author

  Large language models (LLMs) are broadly useful but lack domain-specific knowledge and often generate hallucinations in scientific applications. To address this challenge, we present a domain-aligned and trustworthy LLM framework that integrates query reformulation retrieval-augmented generation (QR-RAG) with a structured database derived from scientific literature. As a proof-of-concept, the framework is applied to water-splitting catalysis, a representative domain characterized by complex experimental conditions and heterogeneous data. For domain-specific Q/A, we combine LLM-guided query decomposition and optimization with hybrid retrieval, replacing a baseline that uses vector search based solely on cosine similarity. In quantitative tests, the framework significantly increases answer accuracy to 85.6% compared with 21.3% for a raw database with a conventional RAG (C-RAG) and reduces operating cost by 39%. Moreover, the qualitative evaluation using the Retrieval-Augmented Generation Assessment (RAGAS) confirms that the proposed system generates answers more faithfully grounded in retrieved evidence, with fewer hallucinations. The proposed framework bridges the gap between general-purpose LLMs and trustworthy domain-specific Q/A, marking a practical breakthrough for applying LLMs in specialized research domains.

  PublisherDOI

• Designing reinforcement learning algorithms for building HVAC control: From experimental observation to simulation comparisons

  Applied Thermal Engineering · 2025-03-18 · 6 citations

  article
  PublisherDOI

• Hybrid modeling approach for better identification of building thermal network model and improved prediction

  ArXiv.org · 2025-12-05

  preprintOpen accessSenior author

  The gray-box modeling approach, which uses a semi-physical thermal network model, has been widely used in building prediction applications, such as model predictive control (MPC). However, unmeasured disturbances, such as occupants, lighting, and in/exfiltration loads, make it challenging to apply this approach to practical buildings. In this study, we propose a hybrid modeling approach that integrates the gray-box model with a model for unmeasured disturbance. After reviewing several system identification approaches, we systematically designed the unmeasured disturbance model with a model selection process based on statistical tests to make it robust. We generated data based on the building model calibrated by real operational data and then trained the hybrid model for two different weather conditions. The Hybrid model approach demonstrates the reduction of RMSE approximately 0.2-0.9C and 0.3-2C on 1-day ahead temperature prediction compared to the Conventional approach for mild (Berkeley, CA) and cold (Chicago, IL) climates, respectively. In addition, this approach was applied for experimental data obtained from the laboratory building to be used for the MPC application, showing superior prediction performances.

  PublisherOA PDFDOI

• Virtual Refrigerant Charge Sensing Method for Next-Generation Refrigerant in Residential Heat Pumps

  2025-08-04

  article
  PublisherDOI

• Seasonal performance evaluation of zero-superheat active refrigerant charge control for variable-speed heat pumps

  Energy and Buildings · 2025-11-20 · 1 citations

  article
  PublisherDOI

• Tensorized Interior Radiative Heat Transfer for a Scalable and Calibrated Building Energy Simulator

  2025-11-11

  articleOpen access

  Building energy simulation is a critical tool for developing and testing advanced control strategies, such as Reinforcement Learning (RL), to provide demand flexibility and affordable energy costs. The recently introduced Smart Buildings Control Suite (sbsim) provides a lightweight, scalable, and data-calibrated simulation environment based on a 2D finite-difference model. However, the initial model primarily focused on conductive and convective heat transfer, neglecting the significant impact of long-wave radiative heat exchange between interior surfaces. This paper presents a significant extension to the sbsim framework by incorporating a physically-grounded model for interior radiative heat transfer. Our primary contribution is the development and integration of a fully tensorized radiative heat transfer module, which preserves the computational efficiency and scalability of the original simulator. This was achieved by developing a pipeline for view factor calculation, including an algorithm to identify directly seeing surfaces within complex floor plans, and formulating the net radiation equations for efficient execution on modern hardware accelerators. We validate the numerical accuracy of our tensorized implementation by comparing its results against a traditional iterative approach, demonstrating identical outcomes. This enhancement increases the physical fidelity of sbsim, enabling more accurate training of RL agents for building energy optimization.

  PublisherOA PDFDOI

• Field Implementation of MPC for Heat Pump-Based Duel Fuel Systems in Small Commercial Buildings for Decarbonization

  2024-04-25 · 2 citations

  articleOpen access

  In the transition from fossil fuel to electrified heating, several areas of the US are seeing a concerning pattern. After adding heat pumps (HPs), commercial building owners leave their gas-based units in place, creating hybrid (dual-fuel) systems that are difficult to integrate and control. Causes include a lack of trust in HPs, capacity constraints in certain climate zones, additional uses for gas, and progressive but partial equipment replacement based on end-of-life considerations. Current control products available on the market are unable to address the diversity and complexity of these systems. For example, infrared (IR) remote-controlled mini-splits are common in small-medium commercial buildings (SMCBs) but are especially difficult to integrate with each other or with existing equipment due to limited interoperability among other devices and poor control access. The poor control integration of the original gas-based systems and HP units, and the complexity of optimizing these systems, cause high greenhouse gas emissions and energy costs. This paper describes an open-source control application utilizing model predictive control (MPC) to coordinate and optimize operations of heat-pump and gas-fired (GF) heating dual-fuel systems while maintaining optimal comfort for the occupants in small commercial buildings. Model predictive control is designed and implemented to minimize greenhouse gas emissions by shifting peak load via pre-heating while considering the trade-off between the degradation of HP performance during cold weather and the high emission of the gas-fired boiler. The control application we have designed has been deployed in a small commercial building in New York to manage five IR remote-controlled ductless heat pump mini-splits and a thermostatically controlled furnace. This deployment fully utilizes low-cost IoT devices for both metering and control. The developed MPC and Baseline controls were implemented for 2 months of the winter heating season by alternating each control day by day, and the test results showed MPC reduced 27% of cost and 14% of electricity peak demand while completely eliminating GF usage via shifting 23.4% of the thermal load from occupied-peak time to non-occupied-non-peak time.

  PublisherOA PDFDOI

Frequent coauthors

• James E. Braun

  Purdue University West Lafayette

  40 shared

• Jie Cai

  Nanjing Forestry University

  12 shared

• Jianghai Hu

  11 shared

• Jiacheng Ma

  Anhui University of Technology

  9 shared

• Sang woo Ham

  8 shared

• David Blum

  University of Tübingen

  6 shared

• Michael Wetter

  6 shared

• Vamsi Putta

  Purdue University West Lafayette

  6 shared

Education

• PhD, Mechanical Engineering

  Purdue University

  2015

• MS, Mechanical Engineering

  Korea Advanced Institute of Science and Technology

  2010

• BS, Mechanical Engineering

  Pusan National University

  2008