% IMPORTANT: The following is UTF-8 encoded. This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.
@ARTICLE{Zhai:1052072,
author = {Zhai, Song and Bartkowiak, Niklas and Sibirtsev, Stepan and
Jupke, Andreas},
title = {{E}xperimental determination and model-based prediction of
flooding points in a pilot-scale continuous liquid-liquid
gravity separator},
journal = {Separation and purification technology},
volume = {377},
issn = {1383-5866},
address = {Amsterdam [u.a.]},
publisher = {Elsevier Science},
reportid = {FZJ-2026-00742},
pages = {134177 -},
year = {2025},
note = {Funding: Deutsche Forschungsgemeinschaft (DFG, German
Research Foundation) – 466656378 – within the Priority
Programme “SPP 2331:Machine Learning in Chemical
Engineering”},
abstract = {Liquid-liquid gravity phase separation is crucial in
chemical, biotechnological, metallurgical, and recycling
processes. However, fluctuations in the feed stream
conditions of the separator significantly affect the
coalescence of dispersed drops, leading to the accumulation
of a dense-packed zone (DPZ) and flooding. In this study, we
investigate the relationship between feed stream conditions,
such as temperature, and flooding points in a pilot-scale
DN200 liquid–liquid gravity separator. A
temperature-controlled experimental setup enabling a
temperature range of 20°C to 50 °C was constructed with
artificial-intelligence-assisted online measurements of
separation curves, drop size distributions, and DPZ heights.
Experiments were conducted with 1-octanol dispersed in water
at dispersed phase fractions of 0.3 and 0.5. Experimental
data show that temperature-dependent coalescence parameters,
Sauter mean diameter , and phase fraction primarily
influence flooding points. Further, we evaluated the
prediction accuracy and consistency of two models from the
literature, a lumped zero-dimensional model and the
established Henschke model, which require solely feed stream
data, geometry data, and physical properties. Both models
underestimate experimental flooding points by a mean
absolute percentage error and relative standard deviation
MAPE ± RSD of (21.5 ± 12.2) $\%$ and (24.8 ± 14.8) $\%$
for the Henschke and 0D model, respectively. Considering the
experimental relative standard error of 8.2 $\%$ accounting
for 95 $\%$ confidence, the prediction accuracy and
consistency of both models are reasonable. This study
suggests batch settling experiments and endoscope
measurements in the feed stream of the liquid–liquid
separator to predict its flooding point due to fluctuations
in the feed.},
cin = {IBG-2},
ddc = {540},
cid = {I:(DE-Juel1)IBG-2-20101118},
pnm = {2172 - Utilization of renewable carbon and energy sources
and engineering of ecosystem functions (POF4-217)},
pid = {G:(DE-HGF)POF4-2172},
typ = {PUB:(DE-HGF)16},
doi = {10.1016/j.seppur.2025.134177},
url = {https://juser.fz-juelich.de/record/1052072},
}