2024 Volume 37 Issue 6
Article Contents

Shaojie Wang, Mingming Gu, Shengming Yin, Zhongyue Zhou, Liuhao Ma, Fei Qi. In-situ Measurement of CO2 Column Density and Flame Temperature of Single Biomass Particle Combustion by Laser Absorption Spectroscopy†[J]. Chinese Journal of Chemical Physics, 2024, 37(6): 745-753. doi: 10.1063/1674-0068/cjcp2408105
Citation: Shaojie Wang, Mingming Gu, Shengming Yin, Zhongyue Zhou, Liuhao Ma, Fei Qi. In-situ Measurement of CO2 Column Density and Flame Temperature of Single Biomass Particle Combustion by Laser Absorption Spectroscopy†[J]. Chinese Journal of Chemical Physics, 2024, 37(6): 745-753. doi: 10.1063/1674-0068/cjcp2408105

In-situ Measurement of CO2 Column Density and Flame Temperature of Single Biomass Particle Combustion by Laser Absorption Spectroscopy†

  • Corresponding author: E-mail: fqi@sjtu.edu.cn
  • Received Date: 03/08/2024
    Available Online: 27/12/2024
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Figures(10)  /  Tables(1)

Article Metrics

Article views(277) PDF downloads(4) Cited by(0)

Access History

In-situ Measurement of CO2 Column Density and Flame Temperature of Single Biomass Particle Combustion by Laser Absorption Spectroscopy†

Abstract: The measurement of the combustion characteristics of a single biomass particle is essential for studying massive biofuel energy conversion process; however, it is challenging due to the small physical scale of the biomass particle flame. In this work, we report an investigation of the combustion behavior of single lignin and single cellulose particle through CO2 absorption spectroscopy and high-speed photography. The biomass samples are prepared with an initial diameter of about 1.0 mm and are ignited on a heating plate placed inside a multi-pass absorption cell. The shape and size transformation of single biomass particle are recorded through high-speed photography, and the combustion temperature and CO2 column densities are measured through laser absorption spectroscopy at 4.172 μm. We find that the maximum combustion temperatures are 1662 ± 38 K and 1569 ± 26 K for lignin and cellulose particle, respectively. Besides, we find that CO2 generation precedes the generation of visible volatile flame for cellulose particle. While for lignin particle, the CO2 generation is found to be almost at the same time as its volatile combustion stage begins. The measurement technique and results presented in this work are of practical interest for biomass combustion studies and are meaningful for the development of biomass thermal conversion mode.

    • The rapid consumption of fossil fuels has led to detrimental environmental impacts as well as escalating political controversies. As a result, growing interests have been made to researches of finding alternative fuels such as biomass energy source [1]. Biomass holds great promise and has been the subject of extensive research regarding its production and applications. For instance, Fan et al. [2] outlined a method for extracting lignin from tobacco stems, achieving an extraction rate of 85.38% and a purity of 90.21%. Furthermore, biomass is a crucial raw material for synthesizing other important chemicals and intermediates. For example, Jin et al. [3] and Wu et al. [4] have used various synthetic methods to produce benzene and cumene from lignin.

      As the CO2 released during biomass combustion is equal to the amount of CO2 absorbed during its life cycle [5], biomass is considered as a carbon-neutral energy source and an ideal alternative to coal. The use of biomass in existing pulverized coal power plants usually requires biomass samples to be pre-processed with diameters less than 1 mm [6]. Besides, it requires modifications of the combustion facility considering the fact that the ash mass content and volatile content of biomass differ significantly from coal [7]. In order to make better use of biomass fuel, it is necessary to study the combustion behavior of a single biomass particle [8].

      Due to the similarity of pulverized biomass combustion and coal combustion, many experimental methods for measuring coal particles can also be applied to biomass particles. Common techniques include thermogravimetry analysis (TGA) [9], high-speed photography [10], and laser absorption spectroscopy (LAS) [11−13]. Using these methods one can obtain ignition temperature, ignition delay time, burnout time, and some other combustion characteristics of a single biomass particle. The above-mentioned information is important to understand the large-scale combustion phenomenon of biomass particles and valuable to validate the corresponding numerical models [14−18].

      TGA is used as one of the main methodologies to study biomass particle combustion by monitoring its weight loss during a finely-controlled heating process. For example, Liu et al. [19] evaluated the combustion performance of corncob and hardwood samples with diameters of 0.2–0.6 mm and bituminous coal samples by using TGA. They found that biomasses showed shorter ignition delay, faster volatile releasing rate and overall better combustion performance than bituminous coal. Wei et al. [20] also studied the combustion behaviors of coal/biomass under different oxy-fuel condition by TGA method. High-speed photography is another commonly-used technique for biomass particle combustion diagnostics. Using high-speed photography, Shan et al. [10] investigated the transitions between the volatile combustion stage and the subsequent char combustion stage of three types of biomass samples. Riaza et al. [21, 22] evaluated four different pulverized biomasses with diameters of 100–150 μm in air and in O2/CO2 conditions by three-color pyrometry and high-speed photography. They found that the burnout time of volatiles and char residues showed similar trends when the oxygen mole fraction changed. By adding spectral filters, high-speed photography can further be used to image chemiluminescence of CH, C2, Na, etc. for biomass particle combustion processes. For example, Fatehi et al. [23] performed CH chemiluminescence measurement for several different biomass particle flames and obtained more accurate ignition delay times.

      While the above-mentioned techniques have proven useful in measuring biomass combustion properties such as the ignition delay time, burnout time, volatile releasing rate, etc. Volatile combustion temperatures and species are more challenging to measure especially for biomass particles with smaller physical sizes. For example, Hot et al. [24] reported the temperature measurement for biomass pellets (with diameter of 6–8 mm) combustion through detecting mid-infrared degenerate four-wave mixing (IR-DFWM) signal of hydrogen cyanide (HCN). Weng et al. [25] recently reported an one-dimensional gas-phase temperature measurement on a biomass pellet with diameter less than 10 mm through using two-line atomic fluorescence (TLAF) thermometry by injecting indium atoms into the flame. Similarly, Dreizler et al. [26] reported CO and gas-phase flame temperature measurement for single coal particle ( < 100 μm) combustion by using coherent anti-Stokes Raman scattering (CARS) spectroscopy. However, the above-mentioned approaches all required using Nd: YAG laser systems to generate high laser intensity for sufficient signal generation and the required experimental configurations were relatively sophisticated. On the other hand, the experimental setup for LAS is more affordable and flexible in biomass combustion measurement. Apart from that, LAS-based sensor has easy implementation, market penetration and is a well-established technique [27]. For example, Qu et al. [13] measured H2O concentration and temperature with a tunable diode laser at 1.4 μm. The biomass material was fabricated in a cylindrical shape with diameter of 6-8 mm and length of 8–16 mm to increase the effective absorption path length, and a heating wire was used to wrap and ignite the pellet. Weng et al. [28] ignited the cylindrically-shaped biomass pellet on a laminar flame burner, and LAS measurement was performed with a diode laser with center wavelength at 770 nm for potassium detection, the measurements of gaseous species such as H2O, CO and CO2 were not applicable as they were also present in the laminar flame. More recently, Alexey et al. [11] performed an in-situ measurement of potassium, H2O and CO in an entrained-flow reactor for pulverized biomass combustion diagnostics. The physical scale of the biomass flame was about 0.5 m, and the diameters of the biomass particles ranged from 10–1000 μm. Similarly, Wang et al. [29, 30] successfully applied tomographic LAS method to measure the 2D H2O and flame temperature distribution on a pulverized coal burner with an effective optical length of 0.25 m. Though large biomass particle (pellet) or large-scale pulverized biomass flames can increase the absorption signal strength, it could also complicate the interpolation of the measurement results, as effects like thermal conduction, convection etc. could become more significant. Besides, the physical properties of the large biomass particles (e.g., size, heating rate, radial temperature, and chemical compositions) could change during the combustion process. The above-mentioned examples all deal with single biomass pellet flame with large physical size or massive pulverized biomass flame. The capabilities of LAS measurements for small scale biomass flames with only a few biomass particles or even single biomass particle with size comparable or less than 1 mm would thus be of practical interest.

      The physical changes of individual biomass particles during combustion are of significant research importance. While laser absorption spectroscopy (LAS) provides valuable data on temperature and product concentrations, it does not capture the physical transformations of the particles. To enhance the understanding of single biomass particle combustion, this study integrates LAS with high-speed photography to concurrently investigate the combustion behavior of single biomass particles with diameters ranging from 0.8 mm to 1.2 mm. A 10 kHz LAS system is used to obtain the time evolution of volatile combustion temperature and CO2 column densities. To increase the signal-to-noise ratio (SNR), we measure CO2 absorbance at its υ3 fundamental bandhead; compared to the CO2 absorbance at near-infrared regime, the bandhead absorbance is at mid-infrared regime and is 2–3 magnitudes stronger. Besides, a multi-pass absorption cell is built to increase the effective absorption path length. In the meanwhile, a 1 kHz high-speed photography system is used to record the entire combustion process of biomass particle and to help determine different combustion stages. We find that the pattern of CO2 generation and temperature variation differs between lignin and cellulose particle. To the best of our knowledge, this is the first work demonstrating LAS measurement of single biomass particle with diameter comparable or less than 1 mm.

    II.   EXPERIMENTS
    • FIG. 1 shows the schematic of the experimental setup, and it consists of three parts: a multi-pass test cell, a LAS system and a high-speed photography system. Biomass samples (lignin and cellulose) are both grounded and filtered to a particle size range of 0.8–1.2 mm. The cellulose particle shows slightly more homogeneous size distribution than the lignin particle. Before each test, biomass particles drop from the quartz tube in the middle of the absorption cell, trigger the high-speed camera when falling on the heating sheet, and the same camera trigger is used to synchronize the LAS system. The detailed descriptions and functionalities of each part will be introduced in the following context.

    • The main body of the absorption cell is made of aluminum alloy. Two ZnSe windows are mounted on both sides of the absorption cell to allow the mid-infrared laser to pass into and out of the cell. The cell is well-sealed and is refilled with air each time before the biomass particle combustion. At the center of the absorption cell, a quartz tube with an inner diameter of 4 mm is used to drop the biomass particle (as shown in FIG. 1(b)), and a heated silicon carbide ceramic plate is placed below the silica tube to heat the biomass particles. The ceramic plate can be heated up to 700 ºC for a driving voltage of 150 V. Two adjustable cylindrical mirrors with a focal length of 0.8 m are mounted inside and on both sides of the chamber. We keep the distance between these two mirrors to be about 0.8 m, and adjust the mirror angles to allow the laser beam to get reflected 5 times inside the absorption cell. The relative positions of the laser beams with respect to the biomass particle are shown in FIG. 1(c). The distance between the 1st pass laser spot and the 2nd pass laser spot above the biomass particle is 13 mm, and the 5 laser spots are all about 10 mm above the ceramic plate.

    • For the LAS measurement, a function generator (Tektronix AFG3022C) is used to generate a 10 kHz triangle waveform to drive the interband-cascade laser with a center wavelength of 4.172 µm (Nanoplus). The laser outputs are split into three portions and detected by three photovoltaic detectors (PVI-3TE, Vigo System) so that the transmitted laser intensity ${I_{\text{t}}}$ (detector 3), the un-absorbed laser intensity ${I_0}$ (detector 1), as well as the etalon data (detector 2) can be obtained simultaneously (as shown in FIG. 2).

      The free spectral range (FSR) of the etalon used in this work is 0.012 cm−1. The optical absorbance can then be calculated based on the Beer-Lambert law as

    • The high-speed camera (Phantom, VEO 1310L) enabled image-trigger function in which case the drop of the biomass particle within the quartz tube could trigger the camera for data acquisition. The high-speed camera is operated at a frame rate of 1 kHz.

      The data processing procedure of the high-speed images is shown in FIG. 3. To reduce the influence from the natural light, the experimentally measured raw image is first processed with white balance algorithm to enhance its clarity. Then the adaptive histogram method (AHE) algorithm is used to enhance the contrast of the image. We also take pictures of the background when biomass particle is not dropped onto the ceramic plate. After that, we subtract the background from the raw image to include only the pixels position of the volatile flame. The total gray scale value of the extracted pixels from raw images is defined as the luminous intensity and it will be used to represent the overall intensity of the volatile flame.

    III.   THEORY OF ABSORPTION SPECTROSCOPY
    • At 4.172 μm, collisional line-mixing effects are not negligible for CO2 absorbance calculation even for atmospheric flames [31], the corresponding theoretical derivations for line-mixing calculations were described in details in our previous publication [31] and only a brief introduction will be given here.

      To account for collisional line-mixing, a so-called G-matrix formula [32] can be used to calculate the CO2 absorbance spectrum using Eq.(2) for a non-uniform line-of-sight:

      where $n$ (molecule·cm−3) is the total gas number density, ${X_{\rm{abs}}}\left( x \right)$ is the spatially-dependent mole fraction. $ {{\tilde {\boldsymbol{\rho}} }} $ is a N×N diagonal matrix where N is the number of absorption transitions. The diagonal element of $ {{\tilde {\boldsymbol{\rho }}}} $, ${\rho _i}$, is given by the Boltzmann population fraction of the lower state of the transition and its off-diagonal elements are set to zero. The row vector $ {\mathbf{d}} $ and its transpose $ {{\mathbf{d}}^T} $ represent the transition amplitude, with its element ${d_i}$ given by Eq.(3).

      where $ {S_i} $ (cm−1·molecule−1·cm2) is the transition line strength and $ T\left( x \right) $ is the spatially dependent temperature. When the spatial inhomogeneity can be neglected, ${X_{\rm{abs}}}\left( x \right)$ and $ T\left( x \right) $ are considered as constants and the calculation for the absorbance spectrum can be greatly simplified.

      The matrix $ {{\tilde {\bf{G}}}} $ accounts for the collisional broadening and line mixing, and can be given as

      where ${{\tilde {\mathbf{I}}}}$ is the N×N identity matrix, $ {{{\tilde{\boldsymbol{ \nu}} }}_{\mathbf{0}}} $ is a diagonal matrix with its nonzero element $ {\nu _{0,i}} $ (cm−1) corresponding to the transition frequencies, $ P $ (atm) is the pressure, and $ {{\tilde {\bf{W}}}} $ is the collisional-relaxation matrix where its diagonal and off-diagonal elements are given mostly based on the collisional broadening parameter ${\gamma _i}$, lower state rotational energy ${E''_i}$ etc. as provided by HITEMP database [33]. Parts of the most essential spectroscopic parameters for 4.172 μm CO2 absorbance spectrum calculation are summarized in Table I. And R110-R134 transitions refer to specific rotational transitions in the R branch of CO2 molecular spectrum. The spectral fitting routine is similar to our previous publication [31]. In brief, the combustion temperature $ T $ and the spatially-integrated column density of $ X \cdot L $ for CO2 are considered as the free parameters in the spectral fitting routine. Besides, we use the collisional relaxation matrix $ {{\tilde {\bf{W}}}} $ for CO2-Ar collision system as the basis, and we then introduce a constant scaling parameter to adjust the extent of collisional line-mixing. The value of this constant scaling parameter can be estimated by calculating the species-to-species collision rate among different gaseous species. This approach has also been reported and validated in several different previous publications [31, 32, 34].

      In FIG. 4 we show some simulated absorbance spectra at different temperatures. The effective absorption path length is set as 2 cm and the CO2 concentration is set as 5% for all cases. We find that CO2 absorption near 4.172 μm is highly temperature sensitive. The absorption is stronger near the bandhead (2397.2 cm−1) at higher temperatures. The overall absorption strength is relatively weak at 900 K, and it can even be omitted at room temperature, indicating the ambient absorption has negligible effects on the baseline extraction and on flame thermometry measurements. Notice that from Eq.(2) to Eq.(4) we derive the spectroscopic model by considering a non-uniform line-of-sight absorption. In this work, to reduce the complexity of data interpolation, we temporarily assumed homogeneous combustion property distributions along with the laser interaction with the biomass flames.

    IV.   RESULTS AND DISCUSSION
    • Selected sequential images of a lignin particle’s combustion process are shown in FIG. 5. Since lignin’s glass transition temperature is below 200 oC, its particles first melt (at around 75 ms) and then wrap around other reactants to form a nuclear shell structure [35]. The volatile flame is visibly distinguishable at around 200 ms. During the volatile combustion stage, the shape and size of the lignin particle change significantly, which could be as a result of the “lumpy solid” micro morphology of lignin as it could expand at elevated temperatures [36, 37]. The volatile combustion stage ends at around 700 ms, and the subsequent char combustion stage for the lignin particle lasts for more than 3 s, beyond the data acquisition time period of the high-speed photography.

      In FIG. 6, we show an example of the measured CO2 absorbance spectrum for lignin particle combustion as compared to its best-fit spectrum. The fitting residual is also displayed below.

      By considering the line-mixing effects, the agreement between the experimental measurement and the best-fit spectrum is satisfactory. On the other hand, when the magnitude of the measured absorbance is below 0.04, usually near the beginning and near the end of the biomass particle combustion stage, the signal to noise ratio is too low so that the best-fit results are no longer reliable.

      In FIG. 7, we show the flame luminous intensity measured by the high-speed photography as well as the CO2 column density and the combustion temperature obtained from the LAS measurements. The CO2 absorbance is detectable only within the time frame when the volatile flame is visible in bare eye. From 400 ms to 600 ms, the flame temperature and CO2 column density gradually reach their maximum values, and their time-averaged values are 1546 K and 0.118 cm, respectively, and the maximum combustion temperature as well as the maximum CO2 column density are 1662±38 K and 0.156±0.004 cm, respectively. The measured lignin particle combustion temperature is lower than the value (about 1900 K) using three color pyrometry reported by Wang et al. [38, 39]. Notice that three color pyrometry method is usually considered to be more accurate in extracting particle surface temperate than the gas phase temperature. On the other hand, this discrepancy could also be associated with the line-of-sight nature of LAS measurement. Meanwhile, the volume of the multi-pass absorption cell is much greater than the size of lignin particle as well as the volatile flame, thus the CO2 generated during biomass combustion should have negligible effects on the subsequent measurement. During the experiments, there were instances where small particles either bounced off or fell from the heating plate. To address this issue, we conducted repeated experiments, and ultimately obtained five sets of usable results. These results were generally consistent across the repetitions.

    • Similarly, we show the high-speed images of single cellulose particle combustion in FIG. 8.

      At the early stage of the heating, cellulose particles are first heated to melt, and the particle morphology changes into a spherical form due to liquid surface tension, and the surface color changes from white to pale yellow, followed by black patches. The pale yellow substances on the surface of cellulose particles may be widely regarded as “active cellulose” [40]. The volatile flame could be visually detected at about 500–600 ms. Comparatively, the volatile flame for cellulose particle is much less obvious than for lignin particle, indicating the composition of the flame has much less sooty. The visible volatile flame disappears at about 900 ms and the char combustion stage lasts only about 100 ms, which is much shorter than the time period of char combustion stage for lignin particle.

      An example of the best-fit absorbance spectrum for cellulose particle combustion is shown in FIG. 9. And the corresponding flame luminous intensity, combustion temperature, as well as the CO2 gas column densities are shown in FIG. 10. For cellulose particles, repeated experiments also yielded five sets of usable results. Only one set is presented here.

      For cellulose particle, the production of CO2 precedes the appearance of the visible volatile flame. The CO2 generated before 560 ms could be related to the cellulose pyrolysis [41]. Meanwhile, it can still be due to the volatile combustion product of cellulose considering the measured temperature already reached about 1500 K. The trend of combustion temperature and CO2 column density variation for cellulose particle are more stable than for lignin particle. Within the time frame of visible volatile combustion stage (560 ms to 920 ms), the time-averaged combustion temperature and CO2 column density are 1484 K and 0.144 cm, respectively, and the peak combustion temperature and CO2 column density are 1569 ± 26 K and 0.163 ± 0.003 cm, respectively. The difference of the measured flame temperature between lignin particle and cellulose particle is consistent with the work reported by Wang et al. [38, 39] through using pyrometry technique. For both cellulose particle and lignin particle, repetitive measurements have been performed. The measured properties such as flame temperatures, CO2 column densities are in general consistence as they are shown in FIG. 7 and FIG. 10.

    V.   CONCLUSION
    • In conclusion, we develop an in-situ measurement technique for CO2 column density and flame temperature of single biomass particle with its diameter less or comparable than 1 mm through using absorption spectroscopy and high-speed photography. The signal to noise ratio is enhanced by using a multi-pass absorption cell and by targeting the CO2 absorbance near its 4.172 μm bandhead. Meanwhile, since CO2 absorbance has high temperature sensitivity at its absorption bandhead, it is excellent for thermometry extraction and for suppressing the ambient CO2 interferences. For the two biomass samples being investigated, we find that the lignin particle has much longer char combustion time than cellulose particle. Besides, for cellulose particle, the CO2 generation precedes with its visible volatile flame generation (sooty flame). The method presented in this work can be extended to study single biomass particle combustion/pyrolysis even with smaller sizes, for example, by further increasing the optical path length through multi-reflection and by using more sensitive absorption-based technique such as wavelength modulation spectroscopy. However, this method also has some limitations. LAS can provide path-integrated concentration rather than absolute concentrations. Although it is possible to estimate absolute results by calculating the dimensions of the flame, this approach introduces significant errors. Addressing this issue will be an important challenge for future research.

    Figure (10)  Table (1) Reference (41)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return