An Oscillation-Free Real Fluid Quasi-Conservative Finite Volume Method for Transcritical and Phase-Change Flows

This paper presents a new Real Fluid Quasi-Conservative (RFQC) finite volume method that eliminates spurious pressure oscillations in transcritical and phase-change flows by locally linearizing the equation of state and evolving auxiliary variables, thereby ensuring accurate and robust shock capturing with minimal energy conservation errors.

The Gist

Imagine you are trying to simulate a high-speed jet engine on a computer. Inside this engine, fuel is being sprayed into a stream of super-hot, supersonic air. The fuel isn't just a simple gas or a simple liquid; it's in a weird, "in-between" state called transcritical. It's so hot and pressurized that it's trying to boil, but the pressure is so high it can't quite turn into a gas yet. It's like water that is simultaneously liquid and steam, but behaving in a very chaotic, non-linear way.

The problem is that when scientists try to simulate this on a computer, the math often breaks. The computer starts seeing "ghosts"—fake, wild swings in pressure that don't exist in reality. It's like trying to balance a stack of Jenga blocks where the blocks keep changing shape and size randomly. The simulation crashes, or it produces garbage data.

This paper introduces a new, smarter way to do the math, called the RFQC method (Real Fluid Quasi-Conservative). Here is how it works, explained simply:

1. The Problem: The "Rigid" Calculator

Traditional computer models treat the fuel like a rigid, perfect robot. They try to conserve energy perfectly at every single step. But because the fuel is so weird (changing from liquid to gas instantly), the math gets confused. When the computer averages the energy in a small box, it calculates a pressure that is wrong. This creates those "ghost" pressure waves that ruin the simulation.

2. The Solution: The "Freezing" Trick

The authors realized that instead of fighting the complexity, they should simplify it locally.

Imagine you are trying to draw a perfect circle on a piece of paper, but the paper is crumpled and moving. Instead of trying to draw the whole circle at once, you freeze the paper for a split second, draw a straight line, and then move to the next spot.

The RFQC method does something similar:

• The Freeze: At every tiny moment in the simulation, the computer "freezes" the complex rules of the fuel. It pretends the fuel is a simple, predictable fluid (like a stiffened gas) just for that one tiny step.
• The Evolution: It lets the simulation run forward using these simple, frozen rules. This prevents the "ghost" pressure waves from appearing because the math is now stable.
• The Re-Projection (The Magic Step): Once the step is done, the computer "thaws" the fuel. It looks at the result and says, "Okay, that was a simplification. Now, let's force the numbers to match the real physics of the fuel again." It adjusts the energy slightly to make sure everything is consistent with the real, complex rules.

3. The Analogy: The "Translator"

Think of the simulation as a conversation between two people who speak different languages:

• Person A (The Physics): Speaks "Real Fluid," a complex language with weird grammar and exceptions.
• Person B (The Math Solver): Speaks "Simple Linear," a language that is easy to calculate but can't handle exceptions.

Old Methods: Person B tries to translate Person A's complex sentences directly. They get confused, misunderstand the meaning, and the conversation turns into a shouting match (pressure oscillations).

The RFQC Method:

1. Freeze: Person A pauses and says, "For this next sentence, I will speak in Simple Linear."
2. Translate: Person B easily understands and processes the sentence.
3. Re-Project: Immediately after, Person A says, "Actually, I meant it in Real Fluid." The system quickly adjusts the meaning to ensure the real intent is preserved, without losing the stability of the translation.

4. Why This Matters

The authors tested this method on some of the hardest scenarios possible:

• Flash Evaporation: Fuel suddenly turning to gas (like opening a soda can at supersonic speed).
• Shock Waves: High-speed air hitting a fuel droplet.
• Phase Changes: Liquid turning to gas and back again.

In these tests, the old methods either crashed or gave wrong answers. The new RFQC method stayed calm, accurate, and didn't produce any "ghost" waves. It proved that you can simulate these chaotic, real-world engine conditions without the computer breaking down.

The Bottom Line

This paper gives engineers a new, robust tool to design better rockets and jet engines. It solves a decades-old math problem by admitting that sometimes, to get the right answer, you have to take a "shortcut" (freezing the rules) for a split second, and then immediately correct yourself to stay true to reality. It's a "quasi-conservative" approach: it's almost perfectly conservative, but just flexible enough to handle the chaos of real fluids.

Technical Summary

1. Problem Statement

The simulation of real-fluid flows involving transcritical transitions and phase changes (e.g., in scramjet combustors, high-pressure engines, and rocket turbopumps) presents significant numerical challenges.

• Spurious Pressure Oscillations: Traditional fully conservative finite volume schemes often generate non-physical pressure oscillations at contact discontinuities or interfaces. This arises from the inconsistency between the linear averaging of conserved variables (density, internal energy) in the numerical scheme and the highly non-linear relationship between internal energy and pressure in real fluids (governed by complex Equations of State, or EoS).
• Limitations of Existing Methods:
  • Pressure-Based (PB) methods: Avoid oscillations but violate energy conservation by discarding the total energy equation.
  • Double Flux (DF) methods: Maintain formal conservation but introduce spatial flux inconsistencies at interfaces, limiting energy conservation accuracy to first-order and causing errors in flows with severe property variations.
  • Hybrid methods: Rely on ad-hoc shock sensors and are computationally expensive.
  • Fully Conservative schemes: Often fail to capture shocks and interfaces simultaneously without oscillations.

There is a critical need for an algorithm that balances pressure stability, energy conservation, algorithmic simplicity, and computational cost for arbitrary real-fluid EoS.

2. Methodology: Real Fluid Quasi-Conservative (RFQC)

The authors propose a Real Fluid Quasi-Conservative (RFQC) finite volume method. This approach extends the classic quasi-conservative framework (originally for two-phase flows with stiffened gas EoS) to arbitrary real fluids.

Core Strategy:

1. Local Linearization: Instead of evolving the full non-linear EoS directly, the method locally linearizes the relationship between internal energy ($e$) and pressure ($p$) at each grid point and time step.
   • The relation is decomposed as: $\rho e = p\xi + E_0$.
   • $\xi$ (reciprocal of the Grüneisen coefficient, $\Gamma$) acts as the linearization coefficient.
   • $E_0$ is the linearization remainder.
2. Advection of Coefficients: The coefficients $\xi$ and $E_0$ are treated as passive scalars and evolved via advection equations ($\partial_t \xi + u \partial_x \xi = 0$ and $\partial_t E_0 + u \partial_x E_0 = 0$). This ensures they are transported with the flow velocity without generating flux inconsistencies at interfaces.
3. Oscillation-Free Reconstruction: At the end of each time step, the pressure is reconstructed using the evolved coefficients and the updated internal energy: $p = (\rho e - E_0) / \xi$. This guarantees pressure equilibrium across contact discontinuities.
4. Thermodynamic Re-projection: To maintain consistency with the true real-fluid EoS, the conserved variables (specifically total energy) are recalculated ("re-projected") using the reconstructed pressure and the exact EoS. This step introduces a controlled energy conservation error but ensures thermodynamic consistency.

Algorithm Flow:

1. Initialization: Compute $\xi$ and $E_0$ from initial primitive variables.
2. Evolution: Solve Euler equations and advection equations for $\xi$ and $E_0$ using a Riemann solver (e.g., HLLC) with frozen coefficients.
3. Recovery: Reconstruct primitive variables (including oscillation-free pressure) using the evolved coefficients.
4. Re-projection: Recalculate total energy and update coefficients based on the exact EoS to ensure the next step starts with thermodynamically consistent states.

3. Key Contributions

• Generalization to Arbitrary EoS: Unlike previous methods restricted to specific EoS forms (like stiffened gas), RFQC works with any EoS (e.g., Peng-Robinson, tabulated, or neural network-based) by locally freezing thermodynamic coefficients.
• Elimination of Flux Inconsistency: By evolving $\xi$ and $E_0$ via advection equations rather than freezing them globally or using double-flux approximations, the method avoids the spatial flux inconsistencies that plague the Double Flux method, achieving higher conservation accuracy.
• Theoretical Error Analysis:
  • Smooth Regions: The energy conservation error is shown to be second-order with respect to the time step ($O(\Delta t^2)$), dominated by the non-linear curvature of the EoS.
  • Discontinuous Regions (Shocks): The error is determined by the entropy production rate ($\Delta s$), making it consistent with the inherent truncation error of shock-capturing schemes (first-order).
  • This proves the method does not introduce artificial low-order spatial errors beyond those required for shock capturing.
• Robustness: The method is designed to handle extreme conditions, including shock-interface interactions and phase transitions, where traditional conservative schemes often diverge.

4. Numerical Results

The method was validated using n-dodecane with the Peng-Robinson (P-R) EoS across several test cases:

• Smooth Density Wave Advection:
  • Verified second-order convergence of energy conservation error in smooth flows crossing the saturation line.
  • Demonstrated effective suppression of pressure oscillations.
• Transcritical Riemann Problem:
  • RFQC showed excellent agreement with exact solutions for density, velocity, and pressure.
  • Outperformed Pressure-Based (PB) methods (which had temperature deviations) and Double Flux (DF) methods (which slightly underestimated temperature).
• Flash Evaporation Riemann Problems:
  • Tested both complete and incomplete evaporation scenarios involving non-classical expansion shocks and rarefaction wave splitting.
  • RFQC captured shock velocities and intermediate states more accurately than DF and PB methods.
  • DF and PB methods exhibited non-physical temperature rises near rarefaction wave heads.
• Subcritical Liquid Impingement:
  • RFQC matched exact solutions perfectly.
  • DF and PB methods produced non-physical artifacts (density dips and temperature spikes).
• Shock-Interface Interaction:
  • Critical Success: In a high-speed shock impacting a vapor-liquid interface, RFQC provided stable, accurate results.
  • Failure of Competitors: Both DF and PB methods suffered from numerical divergence (severe oscillations and negative temperatures) under the same conditions, rendering their results unusable.

5. Significance

The RFQC method represents a significant advancement in computational fluid dynamics for advanced propulsion systems.

• Reliability: It provides a robust framework for simulating complex real-fluid phenomena (transcritical injection, cavitation, spray jets) where traditional conservative schemes fail due to oscillations and non-conservative schemes fail due to energy loss.
• Efficiency: It adds minimal computational overhead (only one additional projection step per time step) compared to standard conservative schemes, making it suitable for large-scale 3D simulations.
• Versatility: Its ability to handle arbitrary EoS makes it applicable to a wide range of industrial and scientific problems beyond the specific cases tested, including supercritical fluid power cycles and cryogenic systems.

In conclusion, the RFQC method successfully bridges the gap between numerical stability (oscillation-free) and physical fidelity (energy conservation), offering a superior alternative for simulating real-fluid flows with strong thermodynamic non-linearities.