Probabilistic Neural Network prediction of liquid- liquid two phase flows in a circular microchannel

The present work proposes towards flow pattern prediction in a liquidliquid microchannel flow through a circular channel. Mass transfer in a microchannel mainly depends on the flow regime inside the channel. The liquid-liquid two phase flow regime in a microchannel depends on the flow velocity and the junction characteristics. Hence, the prediction of patterns has a superior role for the characterisation of mass transfer rates. This paper experimentally investigates the flow pattern in an 800 micro meter diameter microchannel with T junction. The slug length variation corresponding to varying inlet flow rate for the aqueous (water) – organic (kerosene) liquids is visualised and measured. A model for the prediction of liquidliquid flow patterns in a circular T-shaped microchannel is designed using Probabilistic Neural Network (PNN). The designed PNN algorithm is explicitly validated by comparing the predicted patterns with the experimentally observed data.