A simulation-based method for air loop balancing and fan sizing using uncertainty and sensitivity analysis
commissioning, air loop balancing, fan sizing, uncertainty analysis, sensitivity analysis, Modelica modeling
Two main tasks of commissioning an air-side HVAC system are to verify the fan capacity and to balance the air loop. Simulations can be of assistance to these two tasks. However, the models used in the simulation will inevitably have some uncertainties, especially for the models of the pressure loss components. This paper proposes to use uncertainty analysis to obtain the adjustment instructions for tuning the dampers and the fan pressure value for sizing the fan on a virtual testbed implemented in Modelica. An air-side system with eight terminals, ten dampers and seven junctions is taken as the use case. 24 correction factors for the pressure loss coefficients (PLC) (10 for the dampers, 14 for the junctions) are taken as the inputs for the uncertainty analysis. 1000 samples of the correction factors are generated by using the Latin Hypercube Sampling method. The proportional balancing method is adopted to determine the positions of the damper so that the designed terminal flow rates could be met. The fan pressure value is also determined accordingly. The distributions of the dampers’ positions and fan pressure can be used to guide the balancing work and fan sizing in practice. In addition, the sensitivity analysis reveals that the position adjustments and fan pressure results are more sensitive to the uncertainties of the dampers’ PLCs. When the uncertainty level of the dampers’ PLC is reduced from ±40% to ±10%, the ranges of the damper’s positions will be significantly narrowed down to less than 15%, and the 95th percentiles of the fan pressure will drop from 116Pa to 38Pa, which shows the practicality and benefit of the proposed method.
Tsinghua University Press
Qiujian Wang, Yiqun Pan, Yangyang Fu et al. A simulation-based method for air loop balancing and fan sizing using uncertainty and sensitivity analysis. Build Simul, 2019, 12(2): 247–258.