Optimizing space heating efficiency in sustainable building design a multi criteria decision making approach with model predictive control.

Abstract

Efficient space heating is vital for sustainable building design, offering opportunities to reduce energy consumption and costs while maintaining thermal comfort. This study examines the optimization of space heating in a nearly-zero energy building (nZEB) located in Oslo, Norway, under cold climatic conditions. The research question explores how advanced control strategies can balance heating costs and thermal comfort efficiently. A novel Model Predictive Control (MPC) framework integrates Long Short-Term Memory (LSTM) neural networks for energy demand prediction and the Ant Nesting Algorithm (ANA) for multi-objective optimization. Dynamic predictions for indoor temperature and heating requirements, based on EnergyPlus simulations and real weather data, guide the system in minimizing heating costs (HC) and comfort penalties (CP) simultaneously. The MPC framework incorporates constraints aligned with ASHRAE Standard 55 adaptive comfort theory, ensuring efficient control of temperature setpoints between 20 °C and 22 °C. Pareto set analysis evaluates optimization outcomes for selected winter days and electricity price scenarios ($0.328/kWh vs. $0.493/kWh), with results demonstrating up to 17% daily heating cost savings compared to conventional methods while maintaining comparable thermal comfort levels. The implications of the research indicate that the suggested framework has the potential to be integrated into automated systems for real-time predictive control, offering a promising tool for building managers and designers. While the proposed framework shows potential as a valuable approach to sustainable heating optimization, it represents one of several methods that can contribute to improving energy efficiency and comfort in sustainable building design, particularly in nearly-zero energy buildings located in cold climates.