Building performance optimization through sensitivity Analysis, and economic insights using AI

Abstract

Optimizing building designs for energy efficiency and occupant comfort presents significant challenges due to the complex and often conflicting requirements of various stakeholders. Consequently, this study conducts a multifaceted sensitivity and economic impact analysis that aims to improve building performance in terms of energy efficiency and occupant comfort by implementing machine learning techniques. Using a broad dataset comprising of 12,000 energy simulation runs for Tvedestrand Upper Secondary School in Norway, several machine learning models were employed with Multi-Layer Perceptron outperforming the others. In addition, several sensitivity analysis methods were used to explore the influence of individual parameters on building performance. The analysis reveals that ventilation rate, room depth, U-value of the facade, and heat gains significantly affect energy consumption. Economic impact analysis was also carried out to compare the cost-effectiveness of traditional HVAC systems with Building Management System (BMS) HVAC solutions. The BMS HVAC system shows significantly lower operational costs over time, with investment costs averaging around 1200 Norwegian kroner (NOK)/m² and operational costs of approximately 150 NOK/m² per year. Sensitivity analysis under different economic scenarios highlights the economic viability of the BMS HVAC system. This study identifies optimal building parameters that balance energy efficiency and thermal comfort, achieving total energy consumption between 11.05 and 22.51 kWh/m² and zero discomfort hours (h > 26 °C). In sum, the findings offer valuable insights for stakeholders, enabling informed decisions about sustainable building design and energy efficiency improvements, ensuring both technical soundness and financial viability under a wide range of conditions, while using the tested tools.