Search

Green roofs are widely used in hot or cold climates mainly because they are capable to improve the energy efficiency of buildings and, when implemented at a large scale, reducing air pollution and the urban heat island effect (UHI) in urban contexts. Artificial Neural Network (ANN) black-box algorithms are a valid alternative to studying complex systems. However, the literature highlights - quite surprisingly – none of the available research refers to coupling ANNs and green roofs in the Mediterranean area, where green roofs are instead considered one of the most suitable technologies to reduce the high cooling demand. Therefore, the objective of this research work is to create and validate an ANN for the prediction of the monthly green roof’s internal and external surface temperatures and the monthly internal air temperature, starting from different green roof parameters and climatic variables. Specifically, the ANN was created with reference to a Mediterranean climate considering an existing green roof on a building of the University of Palermo characterized by a cooling demand predominance; 180 green roof configurations, obtained by varying the characteristic parameters of vegetation (plant height, leaf area index and leaf reflectivity) and the substrate thickness and thermophysical properties (lightweight and heavyweight), were dynamically simulated on an hourly basis to build the training dataset. In addition, other 72 green roof configurations were simulated to generate the dataset for the validation purpose of the ANN accuracy. The optimal ANN-related architecture consists of 90 neurons with one hidden layer and guarantees very high accuracy predictions. The outcomes of this research represent a useful tool to determine the thermal response of green roofs and their impact on the building energy demand and indoor thermal comfort and UHI mitigation.

The rapid improvement in the standard of living requires more detailed and sophisticated methods of evaluating comfort conditions. But, maintaining thermal comfort conditions in confined environments may require complex regulation procedures and the proper management of heating, ventilating and air conditioning (HVAC) systems. In turn, the requirements for indoor thermal comfort do not necessarily coincide with those of energy saving purposes, which in the last years are becoming a crucial issue owing to the enactment of the European Energy Performance of Buildings Directive (EPBD). The aim of this work is to compare different indoor control thermal comfort strategies in view of the evaluation of the energy cost of quality in buildings. In particular, a new PID-fuzzy controller is presented and compared with a classic ON–OFF controller. The performances of the two controllers are quantified and compared by means of two cost functions that are based on the quadratic forms of the overall energy required by the thermal fluid and of the deviation from the preferred set point of the predicted mean vote (PMV). It is found that the application of the PID-fuzzy controller results in lower costs of energy input and lower deviation from set point of PMV.