Propagation of uncertainty through building simulation software
When utilizing forward based energy simulations and network airflow models for prediction of thermal performance, a deterministic approach to modeling are used. The model parameters are determined to be a single value, which produces a single value or time series of the quantities of interest. However, some model parameters can fluctuate during the prediction period like occupancy, lighting, wind speed for natural ventilation, and internal heat sources. In cases like these, utilizing probabilistic methods for the network airflow model or the energy simulation provide better insight to the dynamics of the building. The results provide a variance of the quantities of interest, displays the best and worse case performance, and metrics of robustness for the building design.
This study focused on a naturally ventilated space where the internal energy load and the wind speed are considered to be stochastic variables. The internal sources were broken up into randomly fluctuating variables and stochastic process that describe the process of people entering or leaving the space, occupant’s metabolic rate, office equipment source terms, and lighting. Realizations for the number of people in the space based can be created by the interface below.
The uncertainty in the inputs are propagated through the natural ventilation model to produce probabilistic outputs which identifies the percentage of time the natural ventilation system is in the buoyancy dominated state or the wind driven state.
Anthony D. Fontanini
- Fontanini, U. Vaidya, B. Ganapathysubramanian, “A stochastic approach to modeling the dynamics of natural ventilation systems”, Energy and Buildings, 2013, 63, pg. 87 – 97.