Smart Energy Storage Integration and Management Platform for Buildings (SESIMP-B)

The SESIMP-B project developed and demonstrated a plug-and-play smart service panel designed to better integrate Battery Energy Storage Systems with building loads and local distributed energy resources. The work aimed to make storage systems easier to install, more coordinated, and more cost-effective for a wide range of buildings, including those in underserved communities. The team tested the platform at several sites across California to evaluate how advanced controls and smart switching improved energy flexibility, reduced peak demand, supported resilience during outages, and extended battery life. The project also included a full data and analytics effort to assess performance and quantify economic benefits such as lower installation costs, shorter payback periods, and improved system longevity. Overall, SESIMP-B addressed key barriers to widespread battery adoption by reducing installation time and complexity, improving operational control, and clarifying the value of storage for building owners and communities.

Berls Contribution: Innovative methods to model flexible building loads and coordinate them with BESS and PV Buildings' heating, ventilation, and air conditioning (HVAC) systems are valuable flexible assets due to their natural thermal inertia. These systems can store energy and only require modified control approaches to leverage this feature. The current approach to deliver demand response with HVAC systems involves increasing the thermostat set point during the cooling season, known as global set point temperature adjustment (GTA). This helps reduce demand during peak times. However, the set point adjustment is often made without knowing the duration of the achievable demand reduction. Moreover, such adjustments can lead to rebound effects that increase cooling demand after the demand response event. This project aims to estimate the duration and magnitude of load reduction and rebound using a GTA approach by employing a physics-informed machine-learning model. The model will incorporate historical load, indoor temperature, cooling set point values, and weather data. The resulting model will be integrated into Gridscape's cloud-based DERMS to modify the load profile and enable smart scheduling of the microgrid systems. This will allow the flexible load of the building's HVAC systems to be fully incorporated into the dispatch strategy.  

Columbia Affiliations
The Department of Mechanical Engineering