Paper Conference

Proceedings of ASim Conference 2016: 3rd Asia conference of IBPSA-China, Japan, Korea

     

A NEW LOCAL PRESSURE LOSS COEFFICIENT MODEL OF DUCT TEE JUNCTION APPLIED IN TRANSIENT SIMULATION OF HVAC AIR-SIDE SYSTEM

Qiujian Wang, Zhizhong Huang, X. Peng

Abstract: Most of the local pressure loss coefficient (LPLC) models for duct fittings used in heating ventilation and air conditioning (HVAC) air-side system transient simulations are the simplified ones. The LPLCs are defined as a constant value at the rated flow condition or as a polynomial function of flow ratio (or velocity ratio). However, literature reviews give the evidence that the LPLC is affected by the inlet Reynold number, which makes the usability of the two simplified LPLC models (Constant-value model, Polynomial function model) questionable. To figure out the influence of these simplifications, this study took the diverging tee junction as an example. First we used CFD calculations to generate a new LPLC dataset and trained a data-driven model from it with the method of feature weighted support vector regression combined with particle swarm optimization (PSO-FWSVR). Then we compared this new model with the polynomial model in the LPLC calculation on the test set. The mean absolute percentage error of new model is less than 3%, which is much better than polynomial model. At last we applied all the three LPLC models in a simple air-side system simulation to analyze the impact of LPLC simplifications on the simulated operation condition of a system. It is found that the constant-value LPLC model can cause a large error of the simulation results of system flow conditions. It is too simplified to reflect the dynamic behavior of air distribution system correctly. And the result from polynomial model is comparatively closer to PSO-FWSVR model, but the bias may be unacceptable when simulating a system with high control precision.
Keywords: HVAC air-side system, Transient simulation, local pressure loss coefficient, uniform design, feature weighted SVR
Paper:
asim2016_321