Currents - CUSD

Buildings waste enormous amounts of energy conditioning spaces that are sitting empty for much of the day. Currents addresses this by connecting a user's calendar to a building's HVAC system, predicting when a room will be occupied and adjusting the temperature accordingly, no new hardware required. This semester, the team finalized a fully functional mobile app, sharpened their thermal prediction models using real building data, and secured a pilot deployment at Cornell's Upson Hall, marking the project's first real-world test.

Project Overview

Stakeholders Upson Hall Facilities (confirmed pilot partner); Cornell building management systems; prospective corporate clients including hotels and office spaces

Disciplines / Majors Software engineering, physics, machine learning, business

Team Overview Currents is a student-led CUSD project developing an intelligent HVAC management system that reduces building energy consumption without requiring new hardware. The system uses user calendar data and thermal prediction models to optimize heating and cooling schedules for single-occupancy rooms. The team is organized into five subteams: Mobile, Backend, Modeling, Predictions, and Business.

Problem Statement Buildings waste significant energy heating and cooling unoccupied rooms. Existing systems don't adapt to user schedules, leading to energy use during empty periods and uncomfortable conditions upon arrival. A solution is needed that works within existing infrastructure and is easy for occupants to use.

Approach Users interact with a React Native mobile app to input calendar data and temperature preferences. The backend processes this information and communicates with building management and HVAC systems to execute real-time adjustments. The Modeling team develops thermodynamic equations to predict room conditioning time, while the Predictions team applies regression models and machine learning to improve accuracy. The Business team manages client outreach and develops cost-benefit analyses.

Key Accomplishments This Semester The modeling team improved thermal equations by incorporating solar radiation and air leakage terms. The mobile team prepared a fully functional app for pilot testing. Backend team automated data export processes and debugged prediction pipelines. The Predictions team improved model accuracy using advanced regression techniques. The business team successfully negotiated a pilot deployment in Upson Hall.

Next Steps Launch the Upson Hall pilot, gather real-world usage data, and refine models based on live performance. Business team to expand outreach to corporate clients such as hotels and office buildings. Technical teams to continue optimizing all system components.

Risks & How They Were Addressed System reliability is critical ahead of pilot launch. The team has addressed this by automating backend processes and continuously testing the prediction pipeline against real data from Upson Hall. Integration across five subteams requires strong coordination, which the team has managed through regular cross-team meetings.