Why is SRK unsuitable for liquid-phase gas solubility, and which model should be chosen? Henry’s law guide

For dilute gases dissolved in a liquid, standard SRK is usually the wrong primary model. Cubic equations of state such as Soave–Redlich–Kwong are mainly calibrated for vapor-phase PVT behavior, hydrocarbon phase equilibria, and fugacity calculations. They do not reliably represent the liquid-phase chemical potential of a dilute solute, especially in strongly non-ideal systems such as oxygen or carbon dioxide in water; an SRK calculation can therefore produce errors of more than an order of magnitude and even predict the wrong temperature trend.

Use Henry’s law for dilute gas solubility in a liquid, particularly for aqueous systems. Use an activity-coefficient model such as NRTL, UNIQUAC, or UNIFAC when the liquid mixture is strongly non-ideal or when broader vapor-liquid equilibrium behavior must be represented.

Why SRK Fails for Liquid-Phase Gas Solubility

SRK’s primary design purpose

SRK describes fluid behavior through an equation relating pressure, temperature, and molar volume. Its parameters account for molecular attraction, molecular size, temperature dependence, and, for mixtures, mixing rules.

This structure is highly useful for vapor-phase properties, hydrocarbon mixtures, high-pressure gas processing, and vapor-liquid phase calculations where component fugacities can be represented adequately.

Gas solubility is a liquid-phase chemical-potential problem

The solubility of a gas is governed by the equality of the gas and liquid chemical potentials at equilibrium. For a dilute solute, this requires an accurate description of the solute’s partial fugacity or activity in the liquid phase.

A conventional SRK model does not automatically provide that description. Its liquid-root calculation is not equivalent to a validated model of how an isolated gas molecule interacts with a solvent such as water.

Water is strongly non-ideal

Water has polarity, hydrogen bonding, and a highly structured liquid phase. Dissolved gases such as oxygen, nitrogen, and carbon dioxide interact with water very differently from the non-polar hydrocarbons for which cubic equations are commonly effective.

A standard SRK parameterization generally lacks the solvent-specific information needed to represent these interactions. Treating the liquid as simply another cubic-EOS phase can therefore give a physically misleading solubility prediction.

A liquid root does not guarantee a correct liquid model

Cubic equations can produce multiple volume roots, which are often interpreted as vapor-like and liquid-like states. Selecting the liquid root solves a mathematical root-selection problem, but it does not prove that the predicted liquid fugacity is accurate.

This distinction is important: a mathematically valid liquid root is not necessarily a thermodynamically reliable liquid-phase activity model.

What Goes Wrong in Aqueous Solubility Calculations

Large magnitude errors

For oxygen dissolved in liquid water, a direct SRK calculation can differ from observed solubility by more than an order of magnitude. Such an error is unacceptable when the result controls absorber sizing, stripping performance, reactor calculations, or material balances.

The problem is not merely numerical convergence. A solver can converge cleanly to a result generated by an unsuitable thermodynamic representation.

Incorrect temperature dependence

The same calculation may predict that gas solubility changes with temperature in the wrong direction. For many gases in water at ordinary pressures, solubility decreases as temperature increases, although the exact behavior depends on the solute, solvent, pressure, and temperature range.

An incorrect temperature trend indicates that the model is misrepresenting the underlying liquid-phase thermodynamics, not simply that the input data are slightly inaccurate.

Poor treatment of dilute components

In a gas-solubility problem, the dissolved gas may be present at very low concentration while the solvent dominates the liquid phase. The model must represent the limiting behavior of the solute as its liquid mole fraction approaches zero.

Henry’s law is designed for this dilute-solute limit. A conventional cubic EOS is not necessarily parameterized to reproduce it.

Which Model Should Be Chosen?

Henry’s law for dilute gas-in-liquid solubility

For a dilute gas dissolving in a liquid, Henry’s law is generally the appropriate starting point:

[ f_i^{V} = H_i x_i ]

where (f_i^{V}) is the gas-phase fugacity of solute (i), (x_i) is its liquid-phase mole fraction, and (H_i) is the Henry’s law constant under the selected convention.

At low pressure, gas fugacity may sometimes be approximated using partial pressure. At higher pressure, fugacity corrections, pressure corrections to the Henry constant, and an appropriate gas-phase EOS may be required.

Use temperature-specific Henry constants

Henry’s law is empirical or semi-empirical, so its accuracy depends on the quality and range of the available parameters. The Henry constant must be correlated with temperature and, where relevant, pressure and solvent composition.

Using a value outside its validated temperature or pressure range can introduce significant error even when Henry’s law is the correct model family.

Activity-coefficient models for non-ideal liquid mixtures

For systems involving alcohols, organic acids, bases, electrolytes, or other strongly non-ideal liquid mixtures, use a liquid-phase activity-coefficient model. Common choices include:

• NRTL or UNIQUAC when reliable binary interaction parameters are available.
• UNIFAC when group-contribution estimates are needed and the chemical groups are covered adequately.
• Electrolyte activity models when ions and dissociation materially affect the equilibrium.

These models describe liquid-phase non-ideality through activity coefficients, which are directly relevant to phase equilibrium.

Combine models when the system requires it

A practical vapor-liquid equilibrium model may use different tools for different phases. For example, a gas-phase SRK or Peng–Robinson EOS can calculate vapor fugacity, while Henry’s law or an activity-coefficient model represents the liquid phase.

This gamma-phi or hybrid approach is often more defensible than forcing one conventional cubic EOS to describe both phases.

When SRK or Peng–Robinson Is Appropriate

Hydrocarbon-rich and non-polar mixtures

SRK and Peng–Robinson remain valuable for natural gas, LNG, and hydrocarbon separation systems. They are widely used for vapor-liquid equilibria, gas and liquid fugacities, density, enthalpy, and phase behavior in these applications.

They also offer fast calculations, modest data requirements, and convenient derivatives for iterative flowsheet simulations.

High-pressure gas processing

For demethanizers, deethanizers, compressors, and related gas-processing equipment, a cubic EOS is often a suitable process model. The relevant components and phases are commonly within the range where the EOS has been parameterized and validated.

The presence of a small amount of water does not automatically make SRK invalid, but water-containing systems require appropriate binary interaction parameters and, in some cases, a dedicated water or hydrate treatment.

As the gas-phase component of a hybrid model

Even when SRK is unsuitable for the liquid solubility calculation, it may still be useful for calculating the gas-phase fugacity (f_i^V). The liquid-side relation can then be supplied by Henry’s law or an activity-coefficient model.

The correct question is therefore not always “EOS or no EOS.” It is which model should represent each phase and each non-ideal interaction?

Understanding the Trade-offs

Speed versus liquid-phase accuracy

Cubic EOS models are computationally efficient and require relatively few inputs, often including critical properties, acentric factors, and interaction parameters. This makes them attractive for large flowsheets and pilot-plant simulations.

That efficiency does not compensate for a poor liquid-phase representation when dilute gas solubility is the key output. A faster model that predicts the wrong equilibrium can lead to incorrect equipment and operating decisions.

Data requirements versus predictive capability

Henry’s law can be highly accurate when a reliable Henry constant and temperature correlation are available. Its limitation is that it is usually specific to a solute-solvent pair and a defined range of conditions.

Activity-coefficient models can cover broader liquid-mixture behavior, but they require interaction parameters or group-contribution information. Complex or electrolyte systems may need specialized parameter sets and validation data.

Generic EOS parameters are not universal

Critical properties and acentric factors are sufficient for many vapor-phase estimates, but they do not encode every solvent-specific interaction in a liquid solution. Standard mixing rules may also be inadequate for polar and associating compounds.

Adding fitted binary interaction parameters can improve an EOS, but it does not automatically make the model reliable at infinite dilution or across all temperatures and compositions.

Numerical convergence can conceal model error

Cubic EOS calculations involve nonlinear equations and may yield multiple roots. Poor initial guesses can cause convergence to an inappropriate root, but even correct root selection cannot fix an unsuitable thermodynamic model.

Model validation against measured solubility, equilibrium ratios, and temperature trends is therefore essential.

Making the Right Choice for Your Goal

Select the model according to the dominant chemistry and the property that controls the unit operation.

• If your primary focus is dilute gas solubility in water: Use Henry’s law with validated temperature and pressure corrections, and use an EOS separately for gas-phase fugacity when necessary.
• If your primary focus is strongly non-ideal liquid mixtures: Use an activity-coefficient model such as NRTL, UNIQUAC, or UNIFAC, selecting parameters appropriate to the components and operating range.
• If your primary focus is hydrocarbon-rich gas separation: Use SRK or Peng–Robinson, with suitable binary interaction parameters and phase-equilibrium validation.
• If your primary focus is a water-containing high-pressure process: Consider a hybrid EOS/activity model or a specialized water, electrolyte, or hydrate model rather than relying on unmodified SRK.

The reliable choice is the simplest model that reproduces the relevant liquid-phase equilibrium data over the actual operating range.

Summary Table:

Need reliable equipment for gas solubility, absorption, stripping, or other chemical engineering studies? Contact LABPARK to discuss your requirements. LABPARK provides Educational and Vocational Unit Operations Pilot Plants for chemical engineering, bioprocess and biotechnology, and environmental and water treatment applications. Our systems help universities, research institutes, and enterprises conduct practical training, validate thermodynamic models, evaluate operating conditions, and scale experimental findings with application-focused pilot-plant solutions.