Improve decision-making process and reduce risks in the energy retrofit of existing buildings through uncertainty and sensitivity analysis

Abstract

Buildings energy efficiency has a key role in achieving the objectives set in the energy-environment field. In energy renovation, the performance goal is generally defined based on design assessments, supported by simulation tools. However, due to the uncertainty relating to input variables, the actual performance of the building may be different from the expected value. The uncertainty of the input parameters depends on several factors, including insufficient knowledge in the design phase, the actual quality of construction and plant systems, the unpredictability of external climatic conditions, the modes of use of the building by users. To adequately describe the energy behaviour of the building, it is necessary to define a range of variation within which the real performance value occurs. This range is estimated through the Uncertainty and Sensitivity Analysis (UA/SA), which considers the simulation output as “probabilistic” rather than “deterministic”. In this study, the UA/SA was applied to plan renovation interventions on an energy-intensive public residential building located in the Mediterranean area. The sensitivity analysis has shown that the main parameters affecting the annual energy demand are the efficiency of the cooling system, the set-point temperature and the photovoltaic efficiency. When risk mitigation actions are neglected, the most frequent result is within the range 17,780–19,573 kWh in terms of annual primary energy demand, and the chance to meet the design target is only 22%. Conversely, when appropriate risk mitigation measures are considered for the mentioned parameters, the possibilities to meet the set goal increase to 82%, and the annual primary energy need reduces down to 5489–6446 kWh. Therefore, the UA/SA is a reliable tool to support legislators in meeting the objectives set by the energy policies promoting the attainment of the planned results.