top of page
Anchor 1
  • nikolaosperakis

Numerical and experimental investigation of hydrocarbon flames

Updated: May 30, 2022

Carbon dioxide molar concentration (contour plot) and heat-release iso-surface (white jet) for a piloted premixed ethylene jet flame.

In this collaborative study between the University of California, Los Angeles (UCLA), the University of Southern California (USC) and Stanford University, a comparative experimental and numerical study of three hydrocarbon fuels was carried out. Novel experimental measurement methods based on laser absorption tomography as well as high-fidelity Large-Eddy Simulations (LES) were employed in an effort to quantitatively describe the differences between ethylene, n-heptane and toluene.

I performed the numerical combustion simulations of the piloted premixed jet flame configurations under the guidance of Prof. Matthias Ihme in Stanford and the results were published in two Combustion and Flame journal papers:

1) "Carbon oxidation in turbulent premixed jet flames: A comparative experimental and numerical study of ethylene, n -heptane, and toluene"

2) "Turbulence-induced bias in time-averaged laser absorption tomography of correlated

concentration and temperature fields with a first-order correction"

In the second paper, the bias introduced by the time-averaged measurements was estimated and a first-order correction was employed based on the results of the large-eddy simulation.

Turbulent combustion has been the focus of extensive research efforts over the last several decades, with particular attention devoted to the investigation of hydrogen and light hydrocarbon fuels such as methane. Although these studies provide valuable knowledge about highly turbulent flames, relatively few investigations have assessed the importance of finite-rate chemistry in the context of fuel specific effects, particularly those caused by the variety of functional groups encountered in practical fuels. Considering that many energy conversion devices still rely on turbulent combustion of liquid fuels comprising numerous high molecular weight components, the investigation of fuel effects is of particular importance.

In this study, the thermochemical structure of turbulent jet flames of ethylene, n -heptane, and toluene, was experimentally captured using a piloted premixed jet flame burner. Laser absorption tomography (LAT) was utilized to provide two-dimensional temperature and mole fraction measurements of CO and CO2 using mid-infrared semi-conductor lasers. These carbon oxides were chosen for their roles as critical combustion intermediates and products and their relevance in determining a boundary of heat release associated with the kinetically slow oxidation of CO to CO2 . The novel experimental dataset was accompanied by a series of LES using finite-rate chemistry models to examine the predictive accuracy of current models in capturing fuel effects in these flames, as well as to quantify the influence of turbulent flow-field behavior on the measurements.

The canonical piloted premixed jet flame burner configuration is widely used for turbulent combustion model validation and a modified version was chosen for this study. The burner consists of a central jet tube and a pilot and outer co-flow to stabilize the high-velocity central jet. Experiments were performed at a single jet Reynolds number of 50,000.

Cross-section of the piloted premixed jet burner (PPJB) used for this study along with a chemiluminesence image of a representative flame depicting the radial and axial axes. All measurements are in mm.

As far as the measurement technique is concerned, in this study, we implemented multiple measurement planes of a turbulent jet flame to construct two-dimensional images of temperature and gas composition. A scanned-wavelength direct-absorption method was employed with a tunable quantum cascade laser (QCL) and a tunable inter-band cascade laser (ICL) to spectrally resolve select ro-vibrational transitions in the fundamental vibrational bands of CO and CO2 near 4.9 and 4.2 μm, respectively. During the measurement, the optomechanical assembly was translated horizontally via an automatic translation stage, and the encoder signals of its stepper motor were used to resolve the spatial location of the measurements in time. A manual vertical stage translated the entire assembly to repeat the measurements at different heights downstream of the jet exit. The lasers were current-modulated to scan in wavelength at 1 kHz, and the signals were temporally and spatially averaged to yield an overall spatial resolution in the radial direction of 0.5 mm (the same as the beam diameter), while the resolution in the vertical direction was similar but sampled step-wise at 20 mm intervals.

Top-down schematic of PPJB facility with optomechanical translation stage system. The central jet is surrounded by a co-flow H2/air flame. The lasers, optics, and detectors are mounted and move together while the burner remains stationary.

The measurements were complimented by large-eddy simula-tions. For this, a finite-rate combustion model using reduced chemical models was utilized for the simulation of all three fuels. For the subgrid-scale turbulence-chemistry interaction, the dynamic thickened-flame model was employed, and the Vreman model was used to represent the turbulent subgrid stresses. All chemical kinetic mechanisms employed in these simulations were DRG-reduced and validated against calculations of 1D laminar flames with laminar flame speed, temperature profiles, and major species profiles as reduction targets. For ethylene (C2H4) , a DRG reduced model based on USC Mech II was used. For both n-heptane (n-C7H16) and toluene (C6H5CH3), reduced-order models were based on JetSurF 2.0.

The LES computations provide spatially-resolved instantaneous thermochemical properties (temperature, mole fractions, reaction rates) for the different flames under investigation. Although the CO and CO2 concentrations near the pilot flame region are similar for all of the flames, fuel-specific effects are immediately notable from the instantaneous images; regions of the flows downstream of the pilot flame exhibit local CO and CO2 mole fraction levels which are highest for the ethylene-air flame, next-highest for the n-heptane-air flame, and lowest for the toluene-air flame. For adequate comparison with the experimental laser absorption tomography measurements—which represent time- and azimuthally-averaged thermochemistry—the simulations were run for five convective flow-through times and statis- tical flow-field results were obtained by averaging both in time and in azimuthal direction.

Left: Instantaneous (non-time-averaged) LES predictions of CO and CO 2 mole fraction. Right: Two-dimensional CO mole fraction for each fuel from both experimental results and LES predictions

Two-dimensional CO 2 mole fraction for each fuel from both experimental results (left sides) and LES predictions (right sides).

The LES predictions for all flames show generally good quantitative agreement with measurements, with larger discrepancies observed in upstream regions of the flames for the larger-molecular-weight fuels examined, namely n -heptane and toluene. A thermochemical state-space analysis was conducted, revealing potential discrepancies in the turbulent mixing and residual deficiencies in the low-temperature chemical model, representing opportunities for further investigation and model refinement. More information can be found in the paper:

Download PDF • 3.08MB

It is important to recognize that turbulence-induced thermochemical fluctuations in both the experiments and numerical simulations—resulting in correlated flowfield scalars —can influence our ability to make direct comparisons of these averaged values. This was also the motivation behind the follow-up study which was published in Combustion and Flame and which aimed at understanding the bias associated with flow-field turbulence. In this paper, spatio-temporally resolved temperature and species mole fraction profiles predicted by large eddy simulations (LES) were used to synthetically generate time-resolved line-of-sight absorption measurements at very short time scales (microsecond) to reflect the unsteady nature of a canonical jet burner across various transverse measurement planes. Inversion methods were employed on the time-averaged line-of-sight data to produce radially-resolved temperature and mole fraction profiles, analogous to those produced by laser absorption tomography performed on a time-averaged axisymmetric flowfield. Bias in the measurements compared to true time-averaged scalar fields is a function primarily of temperature dependence in absorptivity and non-zero correlation between temperature and species concentration scalars. A first-order correction to tomography measurements was proposed to account for the bias based on estimated correlations and the known spectroscopic parameters of the probed absorption transitions.

Laser absorption spectroscopy exploits resonance with discrete energy modes of gas molecules to ascertain thermochemical properties of flow fields from light absorption. When applying the method for the estimation of molar fractions and temperature, it is assumed that the time-averaged correlation between the turbulent fluctuations in temperature and mole fraction are zero. This however implies that the two are independent, which is not the case, especially when CO and CO2 are used as markers.

Left: 2D spectral absorption coefficient field for an instantaneous timestep in a turbulent flame; Middle: 1D array of spectral absorbance; Right: spectral absorbance as a function of wavenumber for several timesteps (gray) alongside time-averaged value (red).

A first-order correction based on the the correlation term of temperature and species mole fractions estimated by temporally and spatially-resolved LES predictions was used to reduce the bias of the experimental method. Better agreement with LES predictions were achieved after the correction. Notably, for the plane at x/D = 17.1 the disagreement in temperature near r/D = 0.3 is over 200 K without correction, and reduced to 20 K after applying the first-order correction. Similarly, for the plane at x/D = 27.4 the temperature disagreement is reduced from 100 K to 20 K near r/D = 0.5. This highlights the general usefulness and applicability of the first-order correction on real complex turbulent flowfields. Notably, the biases induced by the correlated scalar fluctuations could be misinterpreted as experimental

or modeling errors if not corrected. As such, the first-order correction analysis presented in this work enables more direct comparison between experimental measurements and numerical simulations.

Temperature profiles obtained from LAT measurements and corresponding simulation predictions and first-order correction at the plane of x/D = 17.1 and x/D = 27.4.

More information can be found in the paper:

Download PDF • 2.17MB



bottom of page