Mixing of miscible liquids: Dimensionless scaling for intermediate-to-large density differences in a stirred tank

Michael R. Wagner, Manuela Dubacher, Nikoletta Patsaki, Philipp Eibl, Peter Varun Dsouza, Michael Dekner, Christian Witz, Johan Remmelgas, Stefan Reimann-Zitz, Johannes Khinast* · 2026

Read Mixing of miscible liquids: Dimensionless scaling for intermediate-to-large density differences in a stirred tank by Michael R. Wagner, Manuela Dubacher, Nikoletta Patsaki, Philipp Eibl, Peter Varun Dsouza, Michael Dekner, Christian Witz, Johan Remmelgas, Stefan Reimann-Zitz, Johannes Khinast* (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-06144.txt.

Introduction

Michael Wagner and nine co-authors study how miscible liquids mix in a stirred tank when their densities differ. Blending liquids to a uniform product is basic industrial work, yet the flow physics is complex and product quality hangs on it. Their simulations use an even fifty fifty split of two liquids in the tank.

The team varied the Reynolds and Richardson numbers, which capture flow speed and buoyancy effects, in a systematic sweep. Mixing time rose with the Richardson number, in line with earlier reports, while the Reynolds effect was less clear. Combining power, Froude, and Richardson numbers, they derived an exponential scaling law for dimensionless mixing time. That single law collapsed all of their simulation data onto one master curve.

Catalog record

CREATORMichael R. Wagner, Manuela Dubacher, Nikoletta Patsaki, Philipp Eibl, Peter Varun Dsouza, Michael Dekner, Christian Witz, Johan Remmelgas, Stefan Reimann-Zitz, Johannes Khinast*
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE38 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.06144

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Mixing of miscible liquids: Dimensionless scaling for intermediate-to-large density differences in a stirred tank Michael R. Wagner, Manuela Dubacher, Nikoletta Patsaki, Philipp Eibl, Peter Varun Dsouza, Michael Dekner, Christian Witz, Johan Remmelgas, Stefan Reimann-Zitz, Johannes Khinast* Abstract Mixing of miscible liquids is an essential process in multiple industrial settings, usually with the intent to homogenize the product. This seemingly simple process is in fact a complex hydrodynamic problem that has a direct impact on the product quality. In this study, numerical simulations of a stirred tank were performed with a 50/50 ratio of liquids and systematically varied the Reynolds and Richardson numbers. A positive correlation between the mixing time and the Richardson number was observed, as reported in the literature. The influence of the Reynolds number was not as pronounced and clear. Based on the Power, Froude and Richardson numbers, we were able to derive an exponential scaling for the dimensionless mixing time that collapsed all our data onto one master curve. Keywords Dimensionless analysis, CFD, turbulent mixing, lattice Boltzmann method, stirred tank Introduction Mixing of fluids is one of the most abundant processes in nature, e.g., when fresh water mixes with sea water in a river delta or cold air hits warm air during a change in the weather. Also in industry, mixing of fluids is performed in numerous production processes. Consequently, effective mixing and the optimization of the associated mixing system are of critical importance. Over the centuries, several mixing devices have been developed on various scales, ranging from micromixers (e.g., to produce lipid nanoparticles [ 40 , 36 , 26 , 20 ] ) to large industrial stirred tank reactors with volumes of several hundred cubic meters, e.g., for wastewater treatment [ 13 ] . Additionally, stirred tanks are widely used for cultivating cells in the production of pharmaceutical products [ 11 ] The mixing of miscible liquids is a fundamental unit operation in process engineering, aiming to homogenize a mixture of two or more liquids. This process can occur naturally (e.g., through diffusion or buoyancy-driven convection) or be induced by agitation and convection (e.g., during mixing in the pharmaceutical, chemical, food and waste treatment industries). Inducing turbulence massively enhances the mixing efficiency due to the superimposed random fluctuations of the flow field. In comparison, laminar mixing relies on striation formation via stretching and folding and molecular diffusion, which is slow, compared to turbulent mixing. In stirred vessels, mixing performance is commonly quantified by the mixing time, defined as the time required to reduce concentration inhomogeneities below a prescribed level. Experimentally, it is typically determined from tracer experiments as the time elapsed after tracer injection until the measured concentration at predefined probe locations remains within \pm 5\% or \pm 10\% of its final value. Tracer experiments are widely used because they allow the characterization of mixing in a single homogeneous liquid phase [ 22 , 4 , 28 , 9 , 5 , 29 ] . In simulations, mixing is usually quantified by the coefficient of variation (CoV), defined as the relative standard deviation of the concentration field. The temporal decay of probe concentration fluctuations in experiments is conceptually equivalent to the decay of the CoV, as both describe the reduction of relative concentration inhomogeneities. A small amount of tracer leads to an injection that is effectively instantaneous in time and localized in space, closely resembling a Dirac-type impulse as the initial location. While mixing of a small quantify of fluid with a high-volume bulk is done occasionally, mixing of fluids with comparable amounts is more common in industrial systems. In those systems the liquids may differ in density. Differences in density can induce buoyancy-driven stratification, i.e., the formation of vertically layered density regions in which gravitational forces counteract convective transport and suppress vertical mixing. Consequently, achieving a homogeneous mixture may necessitate increased power input at the impeller or longer mixing times [ 50 ] . A considerable body of research has been dedicated to mixing in baffled stirred tanks [ 23 , 28 , 9 , 27 ] . Baffles allow higher stirrer velocities without the formation of a vortex. Baffles also create strongly turbulent flows above a certain critical Reynolds number, and thus, lead to a shorter mixing time [ 10 ] . Despite extensive studies of the hydrodynamics in stirred tanks using experimental methods and computational fluid dynamics (CFD) simulations [ 48 , 42 ] , deriving generally valid correlations for the mixing time remains challenging, particularly when buoyancy effects are present. To establish a reference, we first consider single-phase tracer mixing in a homogeneous Newtonian liquid under turbulent conditions. Ruszkowski [ 39 ] proposed a correlation for the dimensionless mixing time t_{m}^{\ast} , which is essentially the number of revolutions of the stirrer [ 16 ] : t_{m}^{\ast}=N\cdot t_{m}, (1) N\cdot t_{m}=5.3\cdot N_{p}^{1/3}\left(\frac{T}{D}\right)^{2}, (2) where N is the impeller rotational speed, t_{m} is the mixing time, T and D are the tank and impeller diameters, respectively, and N_{p} is the power number, N_{p}=\frac{P}{\rho\cdot N^{3}D^{5}}, (3) with P denoting the power input and \rho the density of the single liquid. The mixing time defined by Ruszkowski corresponds to the 95% tracer mixing time, i.e., the time required for the tracer concentration to reach and remain within \pm 5\% of its final homogeneous value throughout the vessel. This correlation has been used and extensively cited in both numerical and experimental work [ 33 , 31 , 7 ] . Subsequent experimental and CFD-based studies have corroborated the functional form of Eq. ( 2 ), yielding comparable proportionality constants for fully turbulent, single-phase blending in baffled stirred tanks [ 14 , 51 ] . In the fully turbulent regime, Eq. ( 2 ) implies that the dimensionless mixing time becomes independent of N but not the Reynolds number: Re=\frac{\rho\cdot N\cdot D^{2}}{\eta}, (4) where \eta is the dynamic viscosity and \rho is the density of the liquid. Consequently, further increases in impeller speed do not significantly reduce the dimensionless mixing time once fully developed turbulence is established. When density differences between two liquids induce stratification, buoyancy effects introduce an additional governing parameter [ 45 , 1 , 35 , 4 ] . CFD investigations confirm this trend and highlight the stabilizing effect of buoyancy on the interface between the layers [ 7 , 8 ] if the heavier liquid is at the bottom. Yu et al. [ 50 ] studied blending in a unbaffled stirred tank with a non-flat bottom and an off-center, inclined down-pumping axial impeller, differing from the classical baffled tank configurations typically considered, in the laminar-transitional regime, focusing on stratification effects and buoyancy. The results of their study suggested that buoyancy can be important for stratified systems (i.e., light fluid on top of the heavy fluid), especially when the lighter liquid has a volume fraction of more than a few percent. They found a good fit for the dimensionless mixing time using the relation N\cdot t_{m}=30031\left(\frac{L_{H}}{T}\right)^{2.5}\left(\frac{D}{T}\right)^{-5.6}Re^{-1.3}Ri^{1.3}, (5) where L_{H} denotes the liquid height in the vessel. In their work, the Richardson number is defined using the liquid height as the characteristic vertical length scale (i.e., L_{H}=H ): Ri=\frac{\Delta\rho}{\rho_{h}}\frac{g\cdot H}{(N\cdot D)^{2}}, (6) with \Delta\rho being the density difference between the heavier ( \rho_{h} ) and

Read the complete text