Which Equations Of State Predict Multiple Bubble Points? Avoid Simulation Errors

Multiple bubble-point behavior is a subtle but operationally dangerous phase-equilibrium phenomenon that can appear in gas-hydrocarbon separation processes. For predicting these complex phase envelopes in pilot plant studies, the BWR-11 equation of state, the Chao-Seader correlation, and the Grayson-Streed modification are suitable models. The critical simulation limitation is not in the equations themselves, but in the standard flash algorithms of commercial process simulators, which routinely converge to only a single bubble point and require manual intervention to uncover the complete phase behavior.

Multiple bubble points—where a single mixture shows two distinct bubble pressures at one temperature—can trap a pilot plant operator if ignored. The thermodynamic models capable of describing this behavior (BWR‑11, Chao‑Seader, Grayson‑Streed) exist, but simulation software must be guided manually; automatic flash routines will almost always miss the second bubble point, making physical understanding and manual initialization essential.

Why Multiple Bubble Points Matter in Pilot Plants

The Underlying Physics

When a light, sparingly soluble gas like nitrogen or carbon dioxide is dissolved in a heavy oil, the mixture’s phase envelope can develop a retrograde loop. In these systems, lowering the pressure at constant temperature can first bring the liquid to boiling, then drop it back into a single‑phase liquid, and finally bring it to a second, lower‑pressure bubble point. This is not a calculation artifact—it is a real thermodynamic consequence of highly asymmetric mixtures.

If a pilot plant’s separation train is designed around a single bubble‑point estimate, unexpected vapour generation in a “liquid” line or pump cavitation can easily occur. Accurate mapping of the full phase boundary is therefore a safety and performance necessity, not an academic exercise.

Where These Models Excel

The three referenced models—BWR‑11, Chao‑Seader, and the Grayson‑Streed extension—were specifically developed for light‑gas‑heavy‑liquid systems. They handle the low‑temperature, high‑pressure non‑idealities that cause the retrograde S‑curves in the phase diagram.

BWR‑11 uses a multi‑parameter virial‑type expansion that faithfully captures the PVT behaviour of asymmetric mixtures.
Chao‑Seader and Grayson‑Streed combine liquid‑phase activity coefficient logic with an equation of state for the vapour, making them particularly adept at systems where the liquid is highly non‑ideal and the gas is nearly pure.

These models can, in principle, reproduce the complete double‑bubble‑point curve—providing the simulation is driven correctly.

The Real Barrier: Flash Algorithm Limitations

How Standard Flash Routines Work

Every commercial process simulator contains a flash algorithm that, given a temperature and pressure, solves the isothermal‑isobaric equilibrium equations. The algorithm searches for a vapour fraction (V/F) that satisfies the material balance and equal fugacity criteria.

The problem? The algorithm is essentially a root‑finding solver. When a mixture exhibits two bubble points at the same temperature, the function it searches has two roots. A conventional solver, which starts from a neutral initial guess and marches downhill, will latch onto the nearest root—almost always the first bubble point it encounters.

The Automatic‑Scan Blind Spot

Commercial software can generate a “phase envelope” by stepping through temperatures and automatically finding the bubble pressure at each step. Because the double‑root region requires a non‑monotonic search, this automatic scan will trace only the outer edge of the envelope. The inner, retrograde bubble‑point curve remains invisible.

The practical consequence: an operator looking at a simulator‑generated phase envelope may believe the mixture is always liquid below a certain pressure, when in fact a second two‑phase region exists at even lower pressure. In a pilot plant with deep vacuum or high‑pressure‑ratio letdown valves, this hidden region can become active.

The Mandatory Manual Fix

To capture the complete behaviour, you must manually initialize the flash at multiple starting vapour fractions or temperature‑pressure points. Setting (V/F = 0) or (1) from within the two‑phase region, and using physical insight about where the second bubble appears, forces the solver to converge to the alternative root. This iterative, user‑guided mapping is the only reliable way to build the full phase diagram in current simulators.

Understanding the Trade‑offs of Different Equations of State

The Breadth of Available Models

In pilot‑plant education and research, cubic equations like Soave‑Redlich‑Kwong (SRK) and Peng‑Robinson (PR) are standard tools for hydrocarbon separations. They are fast, require only critical properties and acentric factors, and work well for demethanizers, deethanizers, and gas compressors. Their limitations, however, become apparent with retrograde multiple bubble‑point systems.

PR assumes a fixed critical compressibility factor (Z_c = 0.307), while real substances in these units often have (Z_c) between 0.2 and 0.3, degrading accuracy near the critical region.
SRK shares similar volumetric weaknesses and, without carefully regressed binary interaction parameters, cannot reproduce the strong asymmetry needed for double bubble points.

Models like the Starling‑Han BWR (SH BWR) and the Lee‑Kesler correlation introduce additional parameters to improve compressibility representation. While they do better in the dense‑fluid region, they still rely on binary interaction parameters (e.g., (M_{ij})) that can vary with temperature and must be fitted from experimental data. A 10 % variation in (M_{ij}) can translate into a significant error in bubble pressure, making pilot‑plant validation essential.

The Polar‑Mixture Frontier

If your pilot‑plant fluid contains water or CO₂, standard cubic equations become even less predictive for three‑phase equilibria and hydrate formation. In these cases, specialized activity‑coefficient‑based models or the modified Van der Waals (M‑VDW) equation, which introduces symmetry and energy empirical parameters, can match liquid molar volume to within ±2 % and vapour pressure to within ±1 %. However, these models trade simplicity for fidelity and still cannot circumvent the flash‑algorithm root‑finding limitation.

Making the Right Choice for Your Pilot‑Plant Study

The correct tool depends on what you are trying to achieve and which phase behaviour you suspect. Treat the choice as a three‑tier decision.

If your primary focus is to map a suspicious, asymmetric mixture that could show double bubble points: Start with the BWR‑11, Chao‑Seader, or Grayson‑Streed models. Do not trust an automatic bubble‑point trace; manually initialize multiple flash calculations to uncover the hidden root. Validate the predicted second bubble point with a physical pilot‑plant observation wherever possible.
If your primary focus is a conventional hydrocarbon separation (methane‑ethane‑propane) with no heavy‑oil asymmetry: SRK or PR will usually give accurate results for standard dew‑ and bubble‑point calculations, with much faster convergence. However, always perform a quick sensitivity analysis—vary temperature manually near the estimated critical region to check for any retrograde hints that might lie outside the automatic trace.
If your primary focus involves polar components or three‑phase formation: Move beyond cubic equations. Use activity‑coefficient‑based models or the M‑VDW approach with carefully regressed binary parameters, and recognize that even the best equation is useless if the flash solver is not guided to the physically correct root.

The equations of state that can describe multiple bubble points are well established; the challenge is that your simulator will not find them without your engineering judgment. Always pair the right thermodynamic model with a deliberate, manually guided flash strategy—this is the only way to protect pilot‑plant operations from the hidden retrograde vapour region.

Summary Table:

Thermodynamic Model	Best Application	Simulation Limitation
BWR-11	Light-gas-heavy-liquid asymmetric systems	Standard flash solvers converge to a single root; requires manual initialization.
Chao-Seader / Grayson-Streed	Systems with highly non-ideal liquids and nearly pure gases	Auto-scans trace only the outer envelope; misses retrograde curves.
Cubic EoS (SRK / PR)	Standard hydrocarbon separations (methane, ethane)	Degrades near critical regions; fails to reproduce double bubble points.

Achieve Precision in Chemical Engineering Separation

Validate your thermodynamic simulations with real-world physical data. LABPARK provides high-quality Educational and Vocational Unit Operations Pilot Plants in chemical engineering, bioprocess & biotech, and environmental & water treatment for universities, research institutes, and enterprises.

Ensure your separation systems are safe from unexpected vapor generation and retrograde behavior. Contact us today to explore how our pilot plants can support your research and vocational training needs!