Multi-Objective Optimization, Building retrofi, Occupant behavior
Building retrofit design aims at achieving a certain low energy target at a minimum cost. However, these buildings tend to be less comfortable than expected, prone to overheating, poor air quality, and less resilient towards different user profiles. Even when more accurate simulation models are used to calculate the energy demand, occupant behavior is usually oversimplified as a static schedule or rule-based model which often does not depend on comfort conditions and does not represent the actual occupants’ reactions to manage indoor comfort. This can cause a significant gap between the simulated and the measured building performance. To address this gap, we have compared the performance indicators of optimal retrofit solutions obtained through a multi-objective optimization of a reference case and recalculated using different occupant-behavior models for the daily building operation – i.e. opening/closing windows. Recalculations of the optimal solutions have been performed with the generally used static schedules, a rule-based adaptive model, and an innovative probabilistic approach. The results have been analyzed through Pareto difference metrics to quantify the influence of behavioral models on energy consumption, cost, and comfort. Two referent scenarios – in a heating and cooling dominated climate – have been tested to observe the results under different boundary conditions. The findings demonstrate that the performance indicators vary strongly with each behavioral model severely compromising on the competing objectives of energy demand, i.e. cost, indoor air quality, and thermal comfort. The importance of realistic user behavior modeling is highlighted to prevent misleading conclusions on optimal solutions in the assessment of energy efficiency measures. It is pointed out that probabilistic behavior models are highly sensitive to variations of operating conditions, even leading to a positive rebound effect for certain retrofit strategies.