Direct Numerical Simulation of MILD Combustion: Mixing and Autoignition from Non-Premixed Streams

This study utilizes direct numerical simulation of a three-stream mixing layer to demonstrate that MILD combustion, characterized by high dilution, is primarily driven by mixing with hot products leading to a premixed-autoignition mode, whereas non-MILD conditions rely more on fuel-air mixing and premixed-deflagrative combustion.

The Gist

Imagine you are trying to bake a giant, perfect cake in a massive industrial oven. The goal is to get the whole thing to rise and cook evenly without burning the edges or leaving raw dough in the middle.

This paper is about a very special way of baking called MILD Combustion (Moderate or Intense Low-oxygen Dilution). Instead of just throwing cold ingredients into a hot oven, MILD combustion is like pre-mixing your cold batter with a huge amount of hot, already-baked cake crumbs before you even turn on the heat. This pre-heats and dilutes the ingredients, so when the "ignition" happens, the whole cake rises gently and evenly, rather than exploding into a fireball.

Here is a breakdown of what the researchers did and what they found, using simple analogies:

1. The Setup: A Three-Lane Highway

The researchers wanted to understand exactly how this "pre-mixing" works. They built a virtual experiment (a computer simulation) that looks like a three-lane highway:

• Lane 1 (The Fuel): Cold gas (hydrogen and methane) trying to get to the party.
• Lane 2 (The Air): Cold oxygen waiting to help the fuel burn.
• Lane 3 (The Hot Products): A stream of super-hot, already-burned gas (like the exhaust from a previous fire).

In real life, these streams mix together chaotically. The researchers created four different "traffic scenarios" to see how the mixing speed changes the outcome.

2. The Experiment: Fast vs. Slow Mixing

They tested two main variables:

• How much hot gas is there? (High Dilution vs. Low Dilution).
• How fast do the fuel and air mix? (Fast Mixing vs. Slow Mixing).

Think of it like stirring a pot of soup.

• High Dilution (MILD): You have a huge pot of hot broth (the hot products) and you slowly stir in a little bit of cold ingredients. The temperature stays moderate, and everything blends smoothly.
• Low Dilution (Non-MILD): You have a small pot of hot broth and you dump in a lot of cold ingredients quickly. You get hot spots and cold spots, leading to a chaotic, uneven cook.

3. The Big Discovery: Two Different Ways to Cook

The study found that the "recipe" for burning changes completely depending on how much hot gas you mix in.

Scenario A: The MILD Way (High Dilution)

When there is plenty of hot gas mixing in:

• The Result: The whole "room" heats up gently. There are no sharp flames or fireballs.
• The Mechanism: It's like spontaneous popping. Because everything is pre-heated and mixed so well, the fuel decides to burn all at once, everywhere, simultaneously.
• The Analogy: Imagine a room full of popcorn kernels that are all heated to the exact same temperature. Suddenly, pop-pop-pop, they all turn into popcorn at the same time. There is no "fire" moving from one kernel to the next; they just all ignite together.
• Key Finding: In this mode, it doesn't matter much how fast the fuel and air mix with each other, because the hot gas is doing all the heavy lifting to prepare the fuel.

Scenario B: The Normal Way (Low Dilution)

When there isn't enough hot gas:

• The Result: You get a traditional fire.
• The Mechanism: It's like lighting a campfire. You have a spark, and the flame spreads out, eating its way through the fuel.
• The Analogy: This is like lighting a match in a cold room. The fire starts at one spot and has to physically travel (propagate) to burn the rest of the wood. It creates hot spots and smoke.
• Key Finding: Here, how fast the fuel and air mix is crucial. If they don't mix well, the fire burns unevenly.

4. Why This Matters

The researchers used a special "microscope" (called CEMA and Flame Index) to look at the tiny details of the burning process. They found:

• In MILD combustion: The fire is actually a wave of spontaneous ignition. It's not a flame traveling; it's chemistry happening everywhere at once. This is great for industry because it produces very little pollution (like Nitrogen Oxides) and is very stable.
• In Normal combustion: The fire is a traveling wave. It relies on the flame front moving through the fuel.

The Takeaway for Engineers

The paper gives engineers a new rule of thumb for designing clean-burning furnaces:

If you want to achieve that "MILD" magic (clean, stable, low-pollution burning), you need to make sure the hot exhaust gas mixes with the fuel faster than the fuel can spontaneously ignite on its own.

If the hot gas mixes slowly, you get a messy, polluting fire. If it mixes fast, you get a gentle, uniform, "popcorn-style" explosion that is perfect for industrial heating.

In short: To get the best, cleanest burn, you don't just need heat; you need to make sure the hot leftovers from the fire mix in before the new fuel gets a chance to light up. It's about timing and mixing, not just temperature.

Technical Summary

1. Problem Statement

Moderate or Intense Low-oxygen Dilution (MILD) combustion is a technology used to achieve stable, low-emission heating in industrial applications. It relies on preheating and diluting reactants with hot combustion products before ignition. However, modeling MILD combustion is challenging because:

• Complex Physics: It lacks a distinct reactive layer, featuring distributed ignition events in space and time rather than a propagating flame front.
• Modeling Gaps: Conventional reduced-order models often struggle to define scalar dissipation and progress variables under these conditions.
• Missing Coupling: Previous Direct Numerical Simulation (DNS) studies have largely focused on either premixed MILD flames or simplified non-premixed setups (e.g., jet-in-hot-coflow) that do not explicitly capture the coupled mixing of three distinct streams: fuel, air, and hot combustion products. This omission limits the understanding of how mixing intensity and stratification govern the transition between MILD and non-MILD regimes in practical non-premixed environments.

2. Methodology

The authors performed a Direct Numerical Simulation (DNS) of a temporally evolving, three-stream mixing layer to mimic the local conditions of a reverse-flow MILD furnace.

• Configuration: The domain consists of three interacting streams:
  1. Central Jet: Fuel (25% H₂, 75% CH₄) at 300 K.
  2. Side Jets: Preheated air at 900 K.
  3. Outer Jets: Hot combustion products (equilibrium state) at 1225 K.
  • The setup allows for independent variation of the mixing time scales between Fuel-Air (FA) and Hot Products-Air (HA).
• Numerical Setup:
  • Solver: In-house finite-difference solver (CIAO) solving reactive, unsteady Navier-Stokes equations in the low-Mach limit.
  • Chemistry: A reduced mechanism (24 species, 251 reactions) derived from C3MechV3.3.
  • Grid: Approximately 1.2 billion cells for high-dilution cases and 0.7 billion for low-dilution cases, resolving the Kolmogorov length scale ($\Delta/\eta < 2$).
• Parametric Study: Four cases were designed by varying the Damköhler numbers ($Da$) for the two mixing processes:
  • Dilution Level: High-Dilution (HD, low $Da_{HA}$) vs. Low-Dilution (LD, high $Da_{HA}$).
  • Fuel Mixing Rate: Fast Fuel (FF, low $Da_{FA}$) vs. Slow Fuel (SF, high $Da_{FA}$).
  • Cases: HD-FF, HD-SF, LD-FF, LD-SF.
• Analysis Tools:
  • Chemical Explosive Mode Analysis (CEMA): To distinguish between autoignition ($|\alpha| < 1$), deflagration ($\alpha > 1$), and extinction ($\alpha < -1$).
  • Flame Index (FI): To differentiate between premixed and diffusion-controlled regions.
  • Scalar Dissipation Rates ($\chi$): Analyzed for both fuel ($\chi_{fuel}$) and hot products ($\chi_{hot}$) to understand mixing-chemistry interactions.

3. Key Contributions

• Novel Dataset: Introduction of a high-fidelity DNS dataset specifically for non-premixed MILD combustion that explicitly resolves the coupled mixing of fuel, air, and hot products.
• Mechanism Identification: Identification of the ratio between hot product mixing time and minimum ignition delay as the governing parameter distinguishing MILD from non-MILD behavior.
• Mode Characterization: Demonstration that under MILD conditions, combustion is dominated by premixed-autoignition, whereas non-MILD conditions exhibit significant deflagrative contributions.
• Mixing Sensitivity: Discovery that MILD combustion modes are sensitive to mixing from both fuel and hot products, whereas non-MILD modes are primarily driven by fuel mixing.

4. Key Results

A. Ignition and Regime Transition

• MILD Regime (High-Dilution):
  • Exhibits low peak temperatures ($T_{max} \approx 1415$ K) satisfying the Cavaliere-de Joannon criterion ($T_{max} - T_{in} < T_{si}$).
  • Ignition is spatially distributed and gradual.
  • Fuel-Air mixing intensity has minimal impact on ignition timing or flame structure once MILD conditions are established.
  • The correlation between fuel and air mixture fractions ($R^2$) approaches unity before ignition, indicating perfect mixing.
• Non-MILD Regime (Low-Dilution):
  • Exhibits high peak temperatures ($T_{max} \approx 2470$ K) and sharp gradients.
  • Ignition is delayed and highly sensitive to fuel-air mixing intensity.
  • Significant thermal and compositional stratification exists prior to ignition.

B. Combustion Modes (CEMA & FI Analysis)

• Heat Release Rate (HRR) Composition:
  • MILD Cases: >88% of HRR comes from premixed-autoignition. Diffusive and deflagrative contributions are negligible (<5%).
  • Non-MILD Cases: Deflagrative contributions increase significantly (up to ~9% premixed-deflagration), indicating the formation of localized reactive layers.
• Comparison with Literature: Unlike previous studies (e.g., Doan et al.) that suggested deflagration plays a larger role in non-premixed MILD, this study shows that shear-driven mixing in the three-stream configuration homogenizes ignition delays, suppressing sustained deflagration and favoring autoignition waves.

C. Scalar Dissipation and Mixing Interactions

• Non-MILD: Combustion mode is driven almost exclusively by $\chi_{fuel}$ (fuel mixing). $\chi_{hot}$ (hot product mixing) has negligible influence. High $\chi_{fuel}$ correlates with deflagration.
• MILD: Both $\chi_{fuel}$ and $\chi_{hot}$ significantly influence the combustion mode. This confirms that the recirculation and mixing of hot products are critical to maintaining the distributed, autoignition-dominated nature of MILD combustion.

5. Significance and Implications

• System Design: The study provides a practical guideline for MILD system design: ensuring the mixing time of hot products is sufficiently fast relative to the minimum ignition delay is the key to achieving and maintaining MILD conditions.
• Model Development:
  • MILD Modeling: Reduced-order models for MILD combustion should focus on autoignition chemistry coupled with multi-stream mixing (both fuel and hot products). They can largely neglect deflagration and diffusion-controlled heat release.
  • Non-MILD Modeling: Models must account for flame propagation and the competition between autoignition and deflagration, which is governed primarily by fuel mixing.
  • Unified Framework: A flexible modeling framework is required to capture the transition between these regimes, accounting for the changing dominance of scalar dissipation rates ($\chi_{hot}$ vs. $\chi_{fuel}$).

In conclusion, this work clarifies that MILD combustion in non-premixed configurations is fundamentally an autoignition-dominated process driven by the intense mixing of hot products, distinct from the stratified, flame-propagation dynamics of conventional combustion.