Predicting Uniform Droplet Formation in Microchannels Using Time-Domain Analysis

The merge of two droplets at a T-junction in a microfluidic device causes small fluctuations in the flow which leads to a cascading effect that induces perturbations and missed mergers. We study the effect of flow fluctuations and inlet geometry on microfluidic drop mergers by utilizing numerical simulations and in-depth time domain analysis of two-phase fluid flow velocities. Furthermore, we demonstrate the successful use of time-domain signal decomposition to predict events downstream by monitoring the signal changes at a single position over time. The variational mode decomposition method utilized is specifically suited for non-linear and non-stationary signals as opposed to the traditional methods for signal decomposition such as Fourier transform and wavelet transform. This method allows us to decompose the signal into components that can be used to eliminate noise of certain frequencies in real-time and predict dynamics in the device. We propose metrics to predict the formation of monodisperse or satellite droplets by analyzing the time-domain signal within the inlet and prior to the droplet pinch-off.