Intelligent Co-simulation to Consider Urban Climate in Building Energy Simulation

The microclimate in urban environments could be very different from those obtained from the weather data commonly employed by current building energy simulation (BES) tools, being potentially a great source of inaccuracy in the building performance assessment. Hence, this paper proposes a solution based on a cosimulation strategy, where the urban microclimatic condition is taken into account by the BES tool using a CFD model. Particularly, the convective heat transfer coefficients and near wall air temperature values are simulated by the CFD model considering the local urban geometry and physical parameters. Following a recently proposed co-simulation strategy, called intelligent cosimulation, the BES-CFD coupling is performed only during a short simulation period, for training an Artificial Neural Network (ANN) model. For the remaining simulation period, the ANN model is employed, keeping a great level of accuracy while reducing drastically the computational time normally required by CFD models. The proposed methodology is applied on a BESTEST like building, within a dense urban environment. The cosimulation is implemented in a Python script, using DOMUS as the BES tool and ANSYS-CFX as the CFD model.