Entraining gravity currents

Chris Johnson, Andrew J Hogg

Research output: Contribution to journalArticle (Academic Journal)peer-review

45 Citations (Scopus)


Entrainment of ambient fluid into a gravity current, while often negligible in laboratory-scale flows, may become increasingly significant in large-scale natural flows. We present a theoretical study of the effect of this entrainment by augmenting a shallow-water model for gravity currents under a deep ambient with a simple empirical model for entrainment, based on experimental measurements of the fluid entrainment rate as a function of bulk Richardson number. By analysing long-time similarity solutions of the model, we find that the decrease in entrainment coefficient at large Richardson number, due to the suppression of turbulent mixing by stable stratification, qualitatively affects the structure and growth rate of the solutions, compared to currents in which the entrainment is taken to be constant or negligible. In particular, mixing is most significant close to the front of the currents, leading to flows that are more dilute, deeper and slower than their non-entraining counterparts. The long-time solution of an inviscid entraining gravity current generated by a lock-release of dense fluid is a similarity solution of the second kind, in which the current grows as a power of time that is dependent on the form of the entrainment law. With an entrainment law that fits the experimental measurements well, the length of currents in this entraining inviscid regime grows with time approximately as t^0.447. For currents instigated by a constant buoyancy flux, a different solution structure exists in which the current length grows as t^4/5. In both cases, entrainment is most significant close to the current front.
Original languageEnglish
Pages (from-to)477-508
JournalJournal of Fluid Mechanics
Early online date19 Aug 2013
Publication statusPublished - Sept 2013


Dive into the research topics of 'Entraining gravity currents'. Together they form a unique fingerprint.

Cite this