SmartCool

AI-Powered Cooling Plant Management for Energy and Cost Savings

SmartCool is an AI-powered software solution that optimizes multi-chiller cooling plants in large buildings. It transforms existing BMS/SCADA and weather data into safe, explainable operating recommendations and automated control actions, reducing dependence on manual sequencing without requiring major hardware retrofits. By combining load forecasting, physics-guided modeling, constrained optimization, and continuous adaptation, SmartCool helps building operators reduce energy use and cost while maintaining cooling demand and operational safety.

Team Members

Prof. Walter Zhe WANG — Principal Investigator
Mr. Shuhao LI — Lead Researcher and Developer
Ms. Qiqi HUANG — Researcher
Mr. Siqi LI — Researcher
Dr. Mingchen LI — Researcher

Core Innovation & Technology

SmartCool integrates building-system data, physics-informed AI, and operational optimization into a unified software workflow. It works with existing BMS/SCADA infrastructure and common sensors, enabling rapid deployment without major hardware replacement.

Load forecasting and performance modeling: Uses plant, building, and weather data to predict cooling demand and model chiller efficiency while accounting for physical relationships and uncertainty.
Safety-constrained optimization: Determines suitable chiller combinations and load allocation while respecting cooling demand, equipment limits and operating constraints.
Continuous adaptation and scalable integration: Updates its performance models as weather, load, and equipment conditions change, while standardized data mapping supports deployment across different building systems.

End-to-End Intelligent Optimization System Architecture — An overview of the five core modules driving SmartCool—from automated LLM-based load forecasting (AUTOLFM) and thermodynamic chiller modeling (PI-BNN) to dynamic programming optimization, self-evolving online learning, and plug-and-play semantic model deployment.

Impact & Metrics

SmartCool is applicable to large commercial complexes, offices, campuses, hospitals, airports, industrial parks, hotels, district cooling systems and other multi-chiller facilities.

At a 180,000 m² commercial complex in Shanghai, SmartCool was evaluated through an alternating field test against the existing rule-based control strategy from July to October 2024. The system achieved 11.52% energy savings, equivalent to 384 MWh and 204.5 tCO₂e avoided annually, without equipment failure during the test.

At HKUST, SmartCool was deployed on a hybrid freshwater–seawater cooling plant comprising nine chillers. An alternating-week field test conducted from April to June 2026 achieved 7.57% energy savings, while maintaining stable cooling operation.

Impact & Proven Metrics — *Real-world impact and field test results of the SmartCool system.*

Watch how SmartCool brings intelligence and sustainability to enterprise climate control:

Milestones & Status

Patents

China invention patent (202410909230.3). Chiller operating sequence optimization method, system, storage medium and product.
China invention patent (202511426158.X). Method for controlling chillers and corresponding control system, device, storage medium and program product.
Hong Kong patent application (42026122281.4). Chiller sequencing optimization based on data-driven model predictive control with dynamic programming.

Recognition & Awards

Second Place, 2023 NeurIPS CityLearn Challenge.
Gold Award, Global AI Challenge for Building E&M Facilities.
Carbon Award 2026, Green Tech Enterprise Category.
Global Top 6 Venture, Lightning Pitch & Peer Hack at the Schwarzman Scholars 10th Anniversary Reunion.
HK$200,000 Lo Kwee Seong Technopreneurship Tech-Ship Fund.
Second Prize, 12th Hong Kong University Student Innovation and Entrepreneurship Competition.
Silver Award, Hong Kong Techathon+ 2026 Sustainability & ESG Track.