Performance predictions using Artificial Neural Network for isobutane flow in non-adiabatic capillary tubes

• A non-adiabatic capillary tube model has been derived and validated.
• An Artificial Neural Network is capable of reproducing data generated by the 1d model.
• The Artificial Neural Network is validated by experiments from literature.
• The model is able to evaluate mass flow rates under non-choked conditions.
• The model is able to evaluate mass flow rates under two phase inlet conditions.

This work presents an Artificial Neural Network (ANN) model of non-adiabatic capillary tubes for isobutane (R600a) as refrigerant. The basis therefore is data obtained by a 1d homogeneous model which has been validated by own measurements and measurements from literature. With this method it is possible to account for choked, non-choked, and also for two-phase inlet conditions, whereas most of the correlations reported in literature are not capable of predicting mass flow rates for non-choked and two-phase inlet conditions. The presented models are valid for a broad range of input parameters in respect to domestic applications – the mass flow rates range from 0 to 5 kg h−1, inlet pressure is from saturation pressure at ambient conditions up to 10 bar, the inlet quality is from 0.5 (capillary) and 0.7 (suction line) to 0 and subcooling (capillary) and superheating (suction line) from 0 K to 30 K.