Desempenho térmico de edificações residenciais: análise da influência de sistemas de parede, padrões de uso e operação e métodos de modelagem para simulação

Abstract

The thermal performance of a building depends on factors such as climate, construction, form, use, operation and method of analysis. This work analyzed the influence of wall systems, use and operation schedules and modeling methods on the thermal performance of residential buildings in Brazil with simulations. Three wall systems were adopted, which are ceramic brick masonry, concrete wall and light steel/wood framing. Regarding the types of use, a standard use, an intensified use and a reduced use of housing units were analyzed. Regarding the operation of the units, the influence of selective ventilation and daytime ventilation was analyzed. Also, five modeling methods were analyzed, which are complete model, ground, intermediate and top floor model, top two floors model, top floor model and critical housing units model. The results of the analysis of wall systems presented variations depending on the climate. The light framing system was the system with the best thermal performance in cold cities and in hot cities when there is no use of natural ventilation; concrete wall presented the best results in hot cities in scenarios with natural ventilation; ceramic brick masonry presented average results among the other systems. The results regarding the usage indicated that intensified use tends to increase indoor temperatures, which is good in colder climates, but will cause thermal discomfort in warmer climates; reduced use has the opposite effect. Daytime natural ventilation tends to cause thermal discomfort in cold cities, but it can be an alternative in warmer climates, especially when combined with scenarios of reduced use at the night period. Finally, the analysis of the modeling methods presented good results in the ground, intermediate and top floor model and the top two floors model, which are indicated for model simplifications; the other models showed considerable losses in similarity results in relation to the reference model, which means they are not recommended.