Mass Transfer Coefficients and Correlations

In chemical engineering and related fields, the rate at which a component moves from one phase to another—such as oxygen dissolving into water or a solute being extracted from a liquid—is often the slow, controlling step in a process. To design and scale up equipment like absorption towers, distillation columns, or reactors, you need a reliable, quantitative way to predict this rate. Mass transfer coefficients provide this essential link, translating complex fluid dynamics and concentration gradients into a single, usable number for engineering calculations.

Defining the Mass Transfer Coefficient

At its core, a mass transfer coefficient, often denoted as $k$, quantifies the rate of mass transfer per unit area per unit driving force. The driving force is the difference in concentration (or partial pressure) that motivates a species to move. The most common expression is the flux equation:

$$N_A=k\Delta C_A$$

where $N_A$ is the molar flux of species A (mol/(m²·s)), $k$ is the mass transfer coefficient (m/s), and $\Delta C_A$ is the driving force concentration difference (mol/m³). This simple form hides significant complexity because the value of $k$ depends on everything from fluid properties and velocity to the specific geometry of the system.

You will encounter two primary classifications: individual and overall coefficients. An individual mass transfer coefficient describes the resistance to transfer within a single phase (e.g., $k_g$ for the gas film, $k_l$ for the liquid film). When mass transfer occurs between two phases in contact, like gas and liquid, the resistances in both films combine. For a process like gas absorption following Henry's Law, the total driving force is expressed using bulk phase concentrations, leading to an overall mass transfer coefficient, $K$. The relationship is analogous to electrical resistances in series: the inverse of the overall coefficient ($1/K$) equals the sum of the inverses of the individual resistances ($1/k_g+1/(H{k}_l)$), where $H$ is Henry's constant.

Foundational Theories: Film, Penetration, and Renewal

To predict why $k$ has a certain value and how it depends on fluid motion, several physical models have been developed. These theories provide the conceptual backbone for more practical correlations.

Film theory is the simplest model. It proposes that a stagnant, laminar film or boundary layer exists at the interface between two phases. All resistance to mass transfer is concentrated within this film, outside of which the fluid is perfectly mixed. Transfer across the film occurs solely by molecular diffusion. While useful for its simplicity, this theory often underestimates the impact of turbulence and predicts that $k$ is proportional to the diffusivity $D$ to the first power ($k\propto D^1$), which isn't always accurate in dynamic flows.

Penetration theory, developed by Higbie, offers a more dynamic view. It models a fluid element exposed to the interface for a short, fixed contact time before being replaced by fresh bulk fluid. During this brief exposure, mass transfer occurs via unsteady-state diffusion into the element. This model more realistically describes processes like gas absorption into falling liquid droplets and predicts $k\propto D^{0.5}$, which aligns better with many experimental observations.

Surface renewal theory, by Danckwerts, refines penetration theory by proposing that fluid elements at the interface are randomly replaced. Instead of a fixed contact time, there is a distribution of times governed by a surface renewal rate. This stochastic approach often provides the best agreement with data for turbulent interfaces, maintaining the proportionality $k\propto D^{0.5}$.

Dimensionless Correlations for Prediction

Since first-principles calculation of $k$ is intractable for real systems, engineers rely on empirical dimensionless correlations. These correlations group variables into dimensionless numbers, allowing experimental data from lab-scale equipment to be scaled up to industrial plants. The three most critical numbers for convective mass transfer are:

• Sherwood number ($Sh$): The primary output of the correlation. It represents the ratio of convective to diffusive mass transfer. $Sh=\frac{kL}{D}$, where $k$ is the mass transfer coefficient, $L$ is a characteristic length (like pipe diameter), and $D$ is the molecular diffusivity. A higher $Sh$ indicates more efficient convective transport.
• Schmidt number ($Sc$): The ratio of momentum diffusivity (kinematic viscosity) to mass diffusivity. $Sc=\frac{\nu }{D}$. It is a fluid property, indicating how thick the velocity boundary layer is relative to the concentration boundary layer. For gases, $Sc\approx 1$; for liquids, $Sc\gg 1$.
• Reynolds number ($Re$): The ratio of inertial to viscous forces. $Re=\frac{\rho uL}{\mu }$. It characterizes the flow regime (laminar or turbulent) and is the primary driver for the correlation.

A typical correlation for forced convection inside a pipe takes the form:

$$Sh=aR{e}^{b}S{c}^c$$

where $a$, $b$, and $c$ are constants determined from experiments. For example, the widely used Chilton-Colburn analogy relates mass transfer to fluid friction and heat transfer. It defines a mass transfer Stanton number ($S{t}_m=Sh/(Re\cdot Sc)$) and states $S{t}_m\cdot S{c}^{2/3}=j_D$, where $j_D$ is the Chilton-Colburn $j$-factor for mass transfer, which is equal to the friction factor $f/2$ under certain conditions. This analogy is powerful for estimating mass transfer rates from more easily measured friction or heat transfer data.

The geometry and flow condition dictate which correlation you use. Common ones include:

• Flow inside pipes: The Gnielinski correlation or Dittus-Boelter-type equations adapted for mass transfer.
• Flow over flat plates: Laminar and turbulent boundary layer correlations.
• Flow around spheres or cylinders: Correlations like the Frössling equation for single particles, crucial for modeling packed or fluidized beds.
• Liquid films flowing down a wall: Relevant for wetted-wall columns.

Common Pitfalls

1. Misapplying Driving Force Units: The units of the mass transfer coefficient $k$ are directly tied to the driving force used. If your driving force is a partial pressure difference (Pa), you have $k_g$ in mol/(m²·s·Pa). If it's a liquid concentration difference (mol/m³), you have $k_l$ in m/s. Using a $k$ value with the wrong driving force expression is a dimensional error that will give a nonsensical flux. Always check that the product $k\cdot \Delta C$ yields units of flux (mol/(m²·s)).

1. Ignoring the Difference Between $k$ and $K$: Confusing an individual coefficient with an overall coefficient is a major source of error. You cannot use an overall $K_L$ (based on liquid driving force) in an equation that requires the film coefficient $k_l$ to calculate the interfacial concentration. This mix-up leads to incorrect analysis of which phase's resistance is controlling the process.

1. Using a Correlation Beyond Its Validity: Every dimensionless correlation has stated ranges for $Re$ and $Sc$. Applying a laminar flow correlation to a turbulent system, or using a pipe flow correlation for a packed bed, will produce highly inaccurate predictions. Always note the geometry and flow regime assumptions in the correlation's source.

1. Forgetting the Chilton-Colburn Analogy's Limits: While incredibly useful, the Chilton-Colburn analogy assumes $Sc$ and $Pr$ (Prandtl number) are not extremely small or large. It works well for gases and ordinary liquids but can break down for liquid metals (very low $Pr$) or highly viscous polymers (very high $Sc$).

Summary

• Mass transfer coefficients ($k$ or $K$) are pragmatic engineering parameters that relate mass transfer flux to a driving force, encapsulating the complex effects of hydrodynamics and diffusion.
• The film, penetration, and surface renewal theories provide physical explanations for how $k$ depends on diffusivity and flow, evolving from a simple stagnant layer model to more realistic dynamic interface models.
• Practical design relies on dimensionless correlations (e.g., $Sh=aR{e}^{b}S{c}^c$) and analogies like the Chilton-Colburn analogy, which allow for the scaling of experimental data across different geometries and flow conditions.
• Accurate application requires careful attention to the definition of driving forces (individual vs. overall coefficients) and strict adherence to the stated limits of empirical correlations.