Research

Current

2019 BENEFIT FOA: Develop publicly available, high-fidelity datasets that measure the holistic load flexibility performance of a suite of commonly used commercial building HVAC and thermal storage equipment under a variety of testing conditions. This work is being conducted in collaboration with Drexel University and Texas A&M University.

Application of Reinforcement Learning to HVAC controls: Explore the use of RL in HVAC controls at different levels to assess the challenges and opportunities of the approach. To date we have investigated a simple RL controller that determines the position of a valve used to control an air temperature. We have also developed an RL controller based on a theoretical controls approach that determines the setpoint for a VAV reheat coil used to control a zone temperature. This latter controller has been implemented in the IBAL.

Thermal storage control approaches: Investigate control approaches for incorporating thermal ice storage to meet building loads. Control decisions include when to build ice, when to use ice for cooling, and which chiller to use to build ice. The decisions need to incorporate weather and load forecasts and utility rates on time scales of days, but also on time scales of hours.

Machine learning in HVAC: Build models of HVAC equipment and provide guidance on how to evaluate and compare models. Validate the models not only on data generated in the IBAL, but also on data from outside sources.

Support for the development of semantic models for HVAC: ASHRAE Standard 223P is being developed to standardize machine-readable semantic models for representing building system information for analytics, automation, and control. These semantic models will enhance automation for different building applications including:

a. automated fault detection and diagnostics,
b. control system configuration,
c. building system commissioning,
d. digital twins,
e. optimization of energy use,
f.  energy audits, and
g. smart grid interactions.

Part of this work is being performed by parastoo.delgoshaei [at] nist.gov (NIST researchers )who are developing a semantic model of the IBAL as a test case for the standard development.

Outside Collaborations

 We work with outside collaborators through grants and cooperative agreements. Contact us (amanda.pertzborn [at] nist.gov (amanda[dot]pertzborn[at]nist[dot]gov)) for information about possible opportunities.