Vault — machine learning
Machine Learning Modeling of the Volumetric Overall Mass Transfer Coefficient in a Novel L-Shaped Pulsed Sieve Plate Extraction Column
This study set out to predict the volumetric overall mass transfer coefficient, a key design parameter in liquid–liquid extraction columns, for a novel L-shaped pulsed sieve plate column. Reliable prediction is hard for non-standard geometries, and the horizontal and vertical sections of this column operate under different hydrodynamic regimes, so no single existing correlation covers both. The researchers trained three machine learning models, an artificial neural network, a random forest, and support vector regression, on experimental data from two liquid–liquid systems. The dataset comprised 36 operating conditions per section. Models were evaluated on a small hold-out set of four test conditions per section. The artificial neural network performed best in both sections, with a test coefficient of determination above 0.98 and an average absolute relative error below 6%. It outperformed published empirical correlations developed for conventional vertical columns. Sensitivity analysis identified dispersed phase holdup and drop size as the dominant predictors in both sections, and residual diagnostics showed no evident systematic bias across predicted values. Readers should note the small dataset and tiny test set, which make the reported accuracy fragile. The model is considered valid for interpolation within the investigated experimental domain only, and its generalization to other conditions is unconfirmed. The abstract does not state its limits beyond these.
The record
- Alert topic
- machine learning
- Authors
- vahid rafiei, sadegh moradi
- Published in
- 2026
- DOI
- 10.21203/rs.3.rs-9940716/v1
- Licence
- CC BY
- Free copy from
- Unpaywall
- Identified by
- Crossref
- Added to the vault
- 2026-11-11
- Link last checked
- 2026-11-11
Getting the paper
Two routes to the same work, and they are not equivalent. The first is the open copy this vault downloaded from, so it is the one that should open without a login, an institutional subscription or a paywall. It was confirmed reachable on 2026-11-11, with no paywall in front of it. The second is wherever the paper officially lives, which is the better citation but may well ask you to pay.
- Free copy via Unpaywall — checked and reachable 2026-11-11, no paywall
- Publisher landing page — may be paywalled
How it got here
A daily literature alert for machine learning surfaced this paper. The trail below is the rest of the story — which service matched the alert’s title to a real bibliographic record, and every host tried while hunting a copy anyone can open. It is on the page for the same reason the licence is: so the provenance can be checked rather than taken on trust.
- Picked up from a daily alert for machine learning
- Identified in Crossref
- arXiv — no open-access copy listed
- OpenAlex — no open-access copy listed
- Unpaywall — ok
More on machine learning
The vault holds 2 papers under this topic, added between 2026-11-04 and 2026-11-11. The ones nearest this one:
About this page
This is a pointer, not the paper. The vault is fed by daily literature alerts: anything scholarly that turns out to have a legally free copy gets a page here, carrying a summary, the licence it was released under, and a link that is re-checked on a schedule so it cannot rot quietly. Nothing is republished — the PDF stays wherever its publisher or repository put it, and every link above points back there.
The summary at the top was written by a language model from the paper’s own abstract and has not been checked by a specialist. Read it as a signpost and the paper as the source. Nothing here is medical, legal, electrical or engineering advice, and a paper appearing in this vault is a record that it exists, not an endorsement of what it concludes.