The air-handling performance of an air-conditioning system is widely governed by two important parameters. Firstly, the operating parameter of the blower and/or, secondly, the geometric design parameters of the ducting system. An optimized ducting design specification can meet a predetermined air-handling performance target for a given sizing of the air-conditioning system. The challenge, however, occurs while carrying forward the duct design to a different platform with a slightly varied air-handling system. The validation for such a design requires time-consuming 3-dimensional Computational Fluid Dynamics (CFD) simulations. The duct design improvement requires optimization of duct cross-sectional area and outlet flow area. And the unit sizing optimization requires finalizing the evaporator core size—pressure drop resistances—and the blower operating speed.
In the current work, the input data for design and unit sizing are fed to a regression model to predict the outflow conditions of vent velocities. The regression is made robust by varying the input data along six different platforms for varying evaporator core sizes and blower operating speeds. The surrogate model is tested against pre-existing simulation results for a long-platform bus model. Also, it is quantitatively validated using the coefficient of determination (R2) and mean square error (MSE), demonstrating its predictive accuracy and reliability. The vent velocities are predicted within the 0.5 m/s range of the simulated 3-dimensional CFD results. This learning-based regression model can be deployed across the bus platforms to predict the vent velocities with an input with respect to duct design parameters and unit operating conditions, as mentioned above.