Measured and predicted energy demand of a low energy building: Important aspects when using building energy simulation

Three different simulation tools were used to simulate a low energy terraced house in the south of Sweden. The software tools all use dynamic models to calculate, for example, the energy demand for heating and the indoor temperatures. The aim of this paper is to discuss the relative importance to the annual energy demand of different energy aspects of a Swedish low-energy house. Both measured and simulated values are considered and compared. The focus is on the impact of choice of software, the habits of the tenants, and the relative impact of different design parameters such as ventilation rates. The measured values for total electricity demand range from about 6000 kWh to over 12000kWh, the average demand being 8020kWh. The annual predicted total energy demand using three different simulation software tools deviated by about 2%. The energy use deviation due to airflow control was about 10%, and the deviation due to differences in heat exchanger efficiency was about 20% and the deviation in annual energy use due to differences in internal gains due to differences in tenant habits, assumed in the models, was 7%. Furthermore, when comparing the predicted energy use during the design process of the low-energy building with actual measurements after the tenants have moved in, these differ about 50% in average for this specific case. Practical application: Building energy simulation software is often used to make predictions of how different construction materials, design principles and operation influence the energy balance and indoor thermal comfort. It is therefore important that the output of these software tools is trustworthy and accurate. This paper discusses the importance of accurate input data during the design process in order to achieve a valid prediction of energy use with emphasis on tenants' behaviour. It was shown that the deviations in a parametric study were larger than the deviations in the comparison between the results from the three simulation tools. This indicates a need for more accurate models for modelling tenant behaviour and habits rather than more accurate building component models.