Continuous separation of liquid-liquid dispersions in a gravity separator is cost-efficient and typical for extraction processes [1]. Fluctuations in feed compositions lead to flooding of the separator with dispersed phase, causing oversized separators to prevent process disruption. Monitoring the flooding points and controlling the separator load can optimize the unit capacity using model predictive control. Yet, adapting to fluctuating feed conditions requires both online measurement data and efficient predictive models.
To this end, a novel temperature-controlled experimental setup was constructed for a pilot-scale DN200 liquid-liquid gravity separator. The setup can perform artificial intelligence-assisted online measurements of separation curves, drop size distributions, and dense-packed zone heights. In our recent study [2], we investigated the relationship between upstream process conditions and flooding points, which are defined as the inlet volume flow until the height of the dense-packed zone reaches a threshold value. We conducted experiments with 1-octanol dispersed in water at dispersed phase fractions of 0.3 and 0.5 and in a temperature range of 20 °C to 50°C. Subsequently, we assessed the prediction accuracy of separator models from the literature [3,4] with our experimental data. The prediction accuracy of the flooding point is around 25 %. This presentation shows the benefit of conducting the necessary online measurements to predict fluctuating flooding points, thereby avoiding costly process disruption.