Predictive HVAC
Industry
Government
Client
Örnsköldsvik Municipality
Team
PILOT
Year
2025
We were hired to design an AI control layer that cuts a water park’s energy use and costs without sacrificing visitor comfort.
Challenge
Paradiset water park in Örnsköldsvik serves over 200,000 visitors annually but faces substantial energy costs maintaining optimal water and air temperatures year-round. Traditional HVAC systems struggle with handling weather forecasts and visitor pattern data, leading to energy waste and inability to optimize heating, ventilation, and water filtration systems dynamically. The municipality sought a data-driven way to cut consumption and costs while maintaining visitor comfort.
Approach
Developed predictive AI models using Long Short-Term Memory (LSTM) networks and Auto-Regressive Integrated Moving Average (ARIMA) algorithms to analyze weather forecasts, visitor trends, and energy price data for dynamic HVAC optimization
Direct BMS integration via EcoStruxure™ Building Operation with BACnet/Modbus, enabling real-time control of HVAC, lighting, and filtration.
Sustainability-focused cloud design (low-carbon EU servers, on-demand compute) so operational costs remain below achieved energy savings.
Deployed low-carbon cloud servers in Europe to align with environmental standards while ensuring system scalability for potential adaptation to hotels, manufacturing plants, schools, and other facilities.
Created comprehensive data processing pipeline analyzing historical temperature patterns, visitor count fluctuations, and energy pricing trends to inform device regulation strategies balancing energy conservation with visitor comfort.
Architecture/Backend
Weather data ingestion, visitor trends, energy prices; real-time + historical analytics.
AI/ML models
Time-series forecasting with LSTM and ARIMA to predict optimal settings.
Infrastructure/Deployment
Cloud deployment on low-carbon EU servers with on-demand compute.
Key results
15–20% reduction in HVAC energy use; up to 15% less water consumption; ~10% improvement in labor efficiency through automated control.
Up to 1,163 MWh in annual energy savings; ~€3,000–€4,000 monthly cost reduction (SEK 35,600–47,500; exchange rate noted on Nov 6, 2024).
Comfort preserved by dynamically adapting setpoints to weather and visitor loads rather than static schedules.
Anticipated labor efficiency improvement of approximately 10% by automating temperature and system adjustments previously requiring manual oversight.
