Measurement of Soot Concentration in Burner Diffusion Flames through Emission Spectroscopy with Particle Swarm Optimization

Zizhen Li, Ni Wan, Xiangchen QianSensors 2024 · DOI

TL;DR

This paper introduces an improved emission spectroscopy (IES) method for measuring soot concentration in diffusion flames, enhanced by an adaptive particle swarm optimization (APSO) algorithm. This new approach significantly improves accuracy and reduces computational complexity compared to standard methods, offering a more reliable tool for understanding and mitigating soot formation.

What Problem Does This Paper Address?

Soot, a primary environmental pollutant from hydrocarbon combustion, contributes to greenhouse effects, acid rain, and reduced visibility. Accurately measuring soot concentration in flames is crucial for understanding its formation mechanisms and developing effective abatement strategies. Traditional contact-based methods, like the thermocouple particle density (TPD) method, are limited by point measurements and potential flame interference. Optical methods, such as emission spectroscopy (ES), offer non-invasive measurement but face challenges due to the complex nature of flame radiation, including the assumption of grey-body behavior and the neglect of self-absorption, which can lead to significant reconstruction errors, especially in optically thick flames. Existing optimization algorithms for these methods can also be slow or prone to getting stuck in local optima.

Methods

The core of this work is an Improved Emission Spectroscopy (IES) method that builds upon the standard ES technique. The IES method incorporates an Adaptive Particle Swarm Optimization (APSO) algorithm to address the challenges of flame radiation reconstruction.

Here's a breakdown of the key methodological aspects:

  • Image Preprocessing: Flame images are first enhanced using histogram equalization and a dark-channel-based image defogging algorithm to improve contrast and eliminate scattered light, crucial for accurate temperature and soot concentration calculations.
  • Emission Spectroscopy (ES) Foundation: The method relies on analyzing the radiation characteristics of gas and solid particles in the flame. It models the flame as a series of circles and uses the path integral of its characteristic field to determine radiation intensity.
  • Addressing Self-Absorption: A significant improvement over standard ES is the explicit inclusion of the self-absorption term in the radiative transfer equation. This accounts for the radiation absorbed by the flame itself, which is critical for accurate measurements in flames with larger optical thicknesses.
  • Adaptive Particle Swarm Optimization (APSO): To efficiently solve the complex, multi-peak optimization problem arising from the radiative transfer equations (especially with self-absorption), the paper employs APSO. This algorithm is an enhancement of the standard Particle Swarm Optimization (PSO). APSO introduces a "disturbance factor" that dynamically adjusts the inertia weight, allowing for better control over the balance between global and local search capabilities. This leads to faster convergence and reduced computational complexity compared to standard PSO.
  • Two-Dimensional Reconstruction: By combining the optical measurements with the APSO-driven reconstruction, the method achieves a one-dimensional tomographic reconstruction of the flame's characteristic field, enabling the visualization of soot concentration distribution.

Datasets

The study utilizes experimental data collected from a laboratory-scale combustion rig.

  • Experimental Data: The authors conducted experiments on diffusion flames generated by burning methane at controlled volumetric flow rates (0.6 to 0.9 L/min). Flame images were captured using a high-speed digital camera (Photron FASTCAM Mini UX50) at 250 frames per second with a resolution of 1280 × 1024 pixels.
  • Reference Data: For comparison and validation, the study also references data from:
  • The Thermocouple Particle Density (TPD) method, used as a ground truth for soot concentration.
  • Chemkin simulation results for soot concentration under similar methane flow rates.
  • Previously published experimental data from other studies (e.g., reference [33]).

No new benchmark datasets were explicitly introduced, but the experimental setup and data collection protocol are detailed for reproducibility.

Key Insights and Ideas

  • Self-Absorption is Key: The paper strongly emphasizes that neglecting the self-absorption term in emission spectroscopy leads to significant errors, particularly in larger or denser flames. Explicitly modeling this phenomenon is crucial for accurate soot concentration measurements.
  • APSO for Efficiency: The adaptive nature of APSO provides a substantial advantage in optimizing the complex, multi-peak functions involved in flame reconstruction. It achieves faster convergence and better global search than standard PSO, making the IES method more practical.
  • IES Outperforms ES: The improved IES method, by incorporating self-absorption and APSO, demonstrates a marked improvement in accuracy and repeatability over the conventional ES method.
  • TPD as a Benchmark: The TPD method serves as a reliable reference, and the IES method's results show a much closer agreement with TPD than the standard ES method.
  • Soot Distribution Patterns: The experimental results reveal characteristic soot distribution patterns in diffusion flames, with concentrations peaking in specific regions and varying with flame height and radial position.

Key Results

  • Accuracy Improvement: The IES method achieved a maximum absolute deviation of 0.09 ppm from the TPD method's measurements, a significant improvement over the standard ES method's deviation of 0.84 ppm.
  • Repeatability and Consistency: The IES method demonstrated better repeatability and generally lower standard deviations compared to the TPD method, indicating more concentrated and less fluctuating data.
  • Computational Efficiency: While the paper notes that APSO can increase computational complexity due to a large number of particles, the specific implementation in this study achieved faster reconstruction times compared to standard PSO. For instance, the average computation time for APSO was 9.06 × 10⁻⁸ s, compared to 9.74 × 10⁻⁸ s for PSO under specific conditions.
  • Agreement with Simulation: The IES results showed good agreement with Chemkin simulation data for soot volume fraction, especially when compared to the ES results.
  • Two-Dimensional Soot Mapping: The method successfully reconstructed and visualized the two-dimensional distribution of soot concentration in diffusion flames.

Limitations and Future Work

  • Limitations:
    • The study was conducted on a laboratory-scale facility with methane diffusion flames. The applicability to larger industrial flames or different fuel types might require further validation.
  • While APSO improves efficiency, the paper acknowledges that considering self-absorption makes the calculation more complex and time-consuming than ignoring it.
  • The paper mentions that the standard PSO algorithm can become trapped in local optima, implying that while APSO mitigates this, it might not entirely eliminate the possibility.
  • Future Work:
    • The authors suggest that the proposed algorithm can predict the distribution of soot concentration in small-size diffusion flames and help understand combustion reaction mechanisms. This implies further research into applying the method to more complex combustion scenarios.
  • Further investigation into optimizing the APSO parameters and exploring its application to other combustion-related measurement problems could be beneficial.
  • Extending the method to other fuel types or different combustion regimes (e.g., premixed flames) would broaden its applicability.

Summary updated

BibTeX
@article{li2024measurement,
  author = {Zizhen Li and Ni Wan and Xiangchen Qian},
  title = {{Measurement of Soot Concentration in Burner Diffusion Flames through Emission Spectroscopy with Particle Swarm Optimization}},
  journal = {Sensors},
  year = {2024},
  doi = {10.3390/s24041292},
}
Figure