Numerical Investigation of Diffusion Flame in Transonic Flow with Large Pressure Gradient

This paper presents a numerical investigation using a finite-volume method to analyze how chemical reactions enhance turbulent transport in transonic diffusion flames with large pressure gradients, demonstrating the viability of turbine-burner concepts through studies of both mixing layers and turbine cascades.

The Gist

The Big Idea: Lighting a Fire Inside a Jet Engine's Turbine

Imagine a jet engine as a two-stage race car. First, there's the combustor (the engine room) where fuel is burned to create a massive explosion of heat. This hot gas spins the turbine (the wheels), which powers the fan at the front.

Usually, by the time the gas hits the turbine, all the fuel is already burned. The turbine just spins in hot air. But what if we could keep burning fuel inside the turbine?

This is the concept of a "Turbine-Burner." Instead of stopping the fire in the engine room, we let it continue into the turbine. This could make the engine lighter, more efficient, and more powerful. However, it's a dangerous game: the gas is moving incredibly fast (supersonic), and the pressure is changing wildly. It's like trying to light a campfire while riding a rollercoaster at 600 mph.

This paper is a computer simulation study asking: "Can we keep a fire burning inside a jet engine's turbine, and what happens if the air entering the turbine is already 'dirty' with exhaust from the first stage?"

---

The Tools: A Digital Wind Tunnel

The researchers built a super-advanced computer code (a "digital wind tunnel") to simulate this.

• The Problem: Chemical reactions (fire) happen much faster than the air moves. In math terms, this is called "stiffness." It's like trying to balance a pencil on your finger while someone is shaking the table violently. Standard math methods crash or take forever to solve this.
• The Solution: They developed a special "splitting scheme." Imagine you are driving a car. Instead of trying to steer and accelerate at the exact same split-second, you do a tiny bit of steering, then a tiny bit of accelerating, then repeat. This allows the computer to handle the fast chemistry and the slow air movement separately but accurately, keeping the simulation stable.

Experiment 1: The Mixing Layer (The "Fire Hose" Test)

Before testing a real engine, they tested a simpler setup: a Mixing Layer.

• The Setup: Imagine two streams of air flowing side-by-side in a narrowing tunnel. The top stream is hot air; the bottom stream is cold fuel. They don't mix immediately; they just slide past each other like oil and water, creating a thin boundary where they touch.
• The Result: When they ignited the fuel, a diffusion flame formed right at that boundary.
  • The Surprise: The fire didn't just sit there. The heat from the fire made the air expand and speed up. This created a "traffic jam" of speed differences (velocity gradients) that actually stirred the air more violently.
  • The Analogy: Think of the fire as a blender. The heat makes the air move faster, which creates more turbulence (swirling), which mixes the fuel and air even better, which makes the fire burn even hotter. It's a self-reinforcing loop.

Experiment 2: Pure Air vs. "Vitiated" Air (The "Exhaust" Test)

In a real engine, the air entering the turbine isn't fresh, clean oxygen. It's "Vitiated Air." This is air that has already passed through the main combustor, so it's full of exhaust gases (CO2, water vapor) and has less oxygen.

• The Comparison: They ran the simulation twice: once with fresh air, once with this "exhaust" air.
• The Outcome:
  • Pure Air: The fire burned hot and bright. The mixing layer was wide and chaotic.
  • Vitiated Air: The fire was weaker and cooler. Because there was less oxygen to start with, the reaction was slower.
  • The Analogy: Imagine trying to light a campfire. With pure air, you have a full bag of oxygen. With vitiated air, you're trying to light it with a bag that's half full of smoke and ash. The fire still burns, but it's smaller, and the "smoke" (the mixing layer) doesn't spread as wide because the fire isn't stirring the air as hard.

Experiment 3: The Real Deal (The Turbine Blade)

Finally, they simulated the gas flowing over an actual turbine blade (a curved wing shape).

• The Challenge: The blade curves the air, creating strong pressure changes. Usually, if you squeeze air (increase pressure), a flame might go out. If you stretch it (decrease pressure), it might blow out.
• The Discovery: Even with the "dirty" vitiated air, the flame held on.
  • The shape of the blade actually helped mix the fuel and air together (like a spoon stirring a pot), which kept the fire alive despite the harsh conditions.
  • The fire burned hotter on the "suction" side (the top curve) of the blade.
  • Aerodynamics: The fire changed how the air pushed against the blade. In the pure air case, the fire was so hot it actually reduced the "push" (lift) on the blade slightly. In the vitiated air case, the push was somewhere in the middle.

The Bottom Line

1. It Works: The "Turbine-Burner" concept is viable. You can keep burning fuel inside the turbine without the flame dying out, even in super-fast, high-pressure flows.
2. Fire is a Mixer: Chemical reactions don't just release heat; they actively create turbulence that helps mix fuel and air, making the fire self-sustaining.
3. Dirty Air is Okay: Even if the air entering the turbine is full of exhaust (vitiated), the fire can still burn, though it will be cooler and less intense.
4. The Future: This proves that we could potentially design smaller, more efficient jet engines by extending the burning process into the turbine. The next step is to build even more complex simulations (3D) to perfect the design.

In short: The researchers proved that you can keep a fire burning inside a jet engine's spinning turbine, even if the air is "dirty," by using the heat of the fire itself to help mix the fuel. It's like a self-stirring, self-fueling engine that could make our planes fly further and cheaper.

Technical Summary

1. Problem Statement

The paper addresses the challenges associated with the turbine-burner concept, where combustion is intentionally extended into the downstream turbine passage of a gas turbine engine to reduce afterburner weight, lower specific fuel consumption, and increase specific thrust.

Key physical challenges identified include:

• Complex Flow Environment: The flow in a turbine passage is compressible, turbulent, and subjected to strong streamwise and transverse pressure gradients caused by blade profiles.
• Flameholding: The rapid acceleration of flow from subsonic to supersonic creates difficulties in maintaining a stable flame.
• Inlet Conditions: Unlike standard simulations that assume pure air, turbine inlets contain "vitiated air" (a mixture of unburned air and combustion products from the primary chamber), which alters thermodynamic and chemical properties.
• Numerical Stiffness: The coupling of stiff chemical reaction source terms with compressible turbulent flow equations requires robust numerical methods to ensure stability and accuracy.

2. Methodology

The authors developed a three-dimensional, finite-volume solver for steady, compressible, reacting, turbulent Navier-Stokes (RANS) equations.

• Governing Equations: The solver handles mass, momentum, energy, and species transport equations. The gas is treated as a perfect gas with variable specific heats based on NASA polynomial fits.
• Turbulence Modeling:
  • The $k-\omega$ Shear-Stress Transport (SST) model (Menter et al.) was used for the primary analysis to capture adverse pressure gradients and separation.
  • The standard $k-\omega$ model (Wilcox) was used for specific comparisons with previous boundary-layer studies.
• Chemistry Model: A one-step global reaction mechanism for methane combustion (Westbrook and Dryer) was employed. The reaction rate follows a modified Arrhenius expression.
• Numerical Scheme:
  • Splitting Scheme: A novel steady-state preserving operator-splitting scheme was developed to handle the stiffness of chemical source terms. Unlike standard Strang splitting, this method integrates the chemical source term and transport terms in a way that guarantees the solution remains at a steady state if the initial condition is a steady state, regardless of the time step size.
  • Solver: The Lower-Upper Symmetric-Gauss-Seidel (LU-SGS) method is used for pseudo-time stepping to accelerate convergence.
  • Grid: Multi-block structured grids were generated. Grid independence was verified, showing second-order accuracy on fine grids.
• Test Cases:
  1. Mixing Layer: A 2D converging-diverging nozzle configuration simulating a transonic mixing layer with a large favorable pressure gradient (-200 atm/m). Cases included laminar and turbulent flows, comparing pure air vs. vitiated air.
  2. Turbine Cascade: A highly-loaded transonic turbine guide vane (VKI LS89) simulating a real turbine environment with a lower pressure gradient (-20 atm/m).

3. Key Contributions

• Development of a Steady-State Preserving Splitting Scheme: The authors introduced a specific splitting algorithm that maintains steady-state consistency while handling stiff chemical source terms, allowing for efficient integration with existing LU-SGS solvers without decoupling the energy equation.
• Full Navier-Stokes vs. Boundary-Layer Comparison: The study validates full 3D RANS simulations against previous 2D boundary-layer approximations, highlighting the limitations of boundary-layer assumptions in regions with strong 2D effects and large pressure gradients.
• Quantification of Vitiated Air Effects: The paper provides a rigorous analysis of how vitiated air (simulating real turbine inlet conditions) impacts flame structure, temperature, and aerodynamic loading compared to pure air.
• Turbulence-Chemistry Interaction: The study elucidates the mechanism by which chemical reaction enhances turbulent transport through increased velocity gradients, leading to high turbulent viscosity in the reaction zone.

4. Key Results

A. Mixing Layer Analysis

• Laminar vs. Turbulent: In the turbulent case, the mixing layer thickness is an order of magnitude larger than in the laminar case. Chemical reaction significantly enhances turbulent transport; the intense production of turbulence (due to high velocity gradients in the reaction zone) leads to large turbulent viscosity, which further strengthens diffusion.
• Ignition Location: The full Navier-Stokes simulation predicted earlier ignition (approx. 5 mm downstream) compared to boundary-layer approximations (approx. 10 mm), attributed to 2D effects near the inlet and the inability of boundary-layer models to capture initial flow curvature.
• Vitiated Air Impact:
  • Combustion: The presence of vitiated air (lower $O_2$ concentration) significantly weakens the chemical reaction, reducing peak flame temperatures and increasing minimum density.
  • Flame Structure: The reaction zone shifts slightly upward (towards the air side) to find sufficient oxygen. The shear layer thickness decreases because weak reactions produce less turbulence, reducing turbulent diffusion.
  • Extinction: In the mixing layer, the flame nearly extinguishes downstream ($x > 75$ mm) in the vitiated air case due to reduced reaction rates and diffusion.

B. Turbine Cascade Analysis (VKI LS89)

• Flame Stability: Two diffusion flames form (one near the suction surface, one in the passage center). Unlike the mixing layer, extinction does not occur in the turbine cascade, even with vitiated air, despite the strong favorable pressure gradient.
• Mechanism: Near the leading edge, local velocity gradients caused by blade curvature enhance molecular and turbulent diffusion, ensuring sufficient mixing to sustain combustion despite the pressure gradient.
• Aerodynamic Performance:
  • Pure Air: Intense combustion increases temperature, reducing pressure diffusion and lowering the pressure difference between the suction and pressure surfaces, thereby reducing aerodynamic loading.
  • Vitiated Air: Due to weaker combustion, the pressure distribution lies between the non-reacting and pure-air reacting cases.
• Methane Consumption: Methane is consumed along the flame interfaces but is not fully depleted by the exit due to the high fuel injection rate; however, the reaction rate increases significantly after the two flame branches merge downstream of the trailing edge.

5. Significance

• Validation of Turbine-Burner Concept: The results demonstrate the viability of the turbine-burner concept, showing that flames can be sustained in highly accelerated transonic flows within turbine cascades, even with vitiated air inlets.
• Design Implications: The study highlights that neglecting vitiated air effects leads to inaccurate predictions of flame temperature, density, and aerodynamic loading. Accurate modeling of the inlet composition is crucial for performance prediction.
• Numerical Advancement: The proposed steady-state preserving splitting scheme offers a robust and computationally efficient approach for simulating stiff reacting flows in complex geometries, bridging the gap between boundary-layer approximations and full 3D simulations.
• Future Work: The authors note that while RANS models provide useful insights, the discrepancies between turbulence models (SST vs. $k-\omega$) suggest that higher-fidelity methods like Large Eddy Simulation (LES) are necessary to fully resolve the complex interaction between turbulence and chemistry in these flows.