3.1. Calibration and Characterization Results
The temperature and velocity profiles at the height of the fibres (in 10 mm distance from the nozzle outlet) are shown in
Figure 4. The nozzle diameter is 12.5 mm, and the radial coordinate
is determined by placing the prandtl-tube at the edge of the 16 mm diameter outlet tube. The radial coordinate is zero at the nozzle axis, positive in one radial direction away from the nozzle axis and negative in the other direction. The measurable distance of the xy-stage is not sufficient to measure the complete distance in the negative
direction, and there may be some misalignment between the real radial coordinate of the nozzle
and the radial coordinate
.
The left diagram in
Figure 4 shows that the flow can be described as a plug flow. Flow velocity seem to decrease in the negative
direction, but this may be caused by a shift in the differential-pressure measurement device with time, even without flow and a slight angle of the prandtl-tube to the nozzle axis. The temperature of the gas jet shown on the right side of
Figure 4 can be higher than the tube temperature at the outlet of heater 1 (
) and the nozzle temperature (
) because the corresponding thermocouples are not positioned in the gas flow and are some distance apart from the heaters (in colder regions). The temperature is not constant over the radial coordinate of the hot air jet. The temperature in the middle of the air jet can be more than 100 °C lower than at the outside regions, which may be caused by heater 2, which heats the gas flow in the tube from the outside. This temperature difference can be decreased by increasing the tube temperature at the outlet of heater 1 to 400 °C and decreasing the nozzle temperature to 300 °C as shown by the red curve with rectangles in the right diagram of
Figure 4. However, the nozzle temperature has to be higher than 300 °C to provide sufficient temperature for soot oxidation (about 350 °C). The temperature and velocity profiles are stable and do not show the flow instabilities reported for downward heated flows in the literature [
10,
11].
A correction of the measured temperature regarding heat radiation was not performed because Roberts et al. [
12] reported an error for a sheathed thermocouple of 38.3 °C at a thermocouple reading of 591.8 °C, which is below 10% of the measured value. The error decreases for lower temperatures and is thus negligible in this study with temperatures up to 470 °C.
The homogeneity of the LLS is measured by placing a fibre perpendicular to the LLS at different positions. The average photomultiplier anode voltage for the different fibre positions is measured to be 425 mV, and the maximum deviation of the photomultiplier anode voltage is 22% at a photomultiplier amplification voltage of 356 V. This LLS homogeneity is in accordance with Wurster et al., who measured a maximum deviation of 20 % [
5].
The calibration for the dependence of the photomultiplier signal voltage (difference between peak voltage and average noise) on the particle size and material is shown in
Figure 5. Each data point in the diagram represents 20 measurements.
Figure 5 shows that the photomultiplier signal of the glass spheres and hard coal particles of similar number median Feret diameter differ significantly. This is due to the different optical and geometrical properties of the particles. The illumination with red light and the presence of the fibre array in the measurement chamber does not systematically change the calibration curve. This is due to the optical filter, which hinders most of the red light from reaching the photomultiplier and due to the definition of the signal voltage as the difference between the peak voltage and the average noise, which cancels out differences caused by the presence of the fibre array. The calibration curves shown in
Figure 5 at an increased temperature seem to be slightly lower than the other calibration curves. The hot air jet, which transports the test particles through the measurement chamber, may serve as a thermal gas lens because hot air has a lower refractive index than cold air [
13], while the refractive index, and thus the reflectance, of glass increases with temperature [
14] and therefore may cancel out the gas lens effect. The size reduction of the coal particles due to oxidation is negligible at these temperatures and residence times. However, the measurements at high temperature are within the standard deviation of the measurements at low temperature, and hence the temperature’s influence on the photomultiplier signal voltage is not significant.
The measurement results are fitted according to the following equation:
where the photomultiplier anode number median signal voltage is
, the number of the median Feret diameter is
and the fit parameters are
and
. Equation (1) can serve as a calibration curve to determine the particle size to a corresponding photomultiplier signal. The fit parameters of the measurements shown in
Figure 5 are summarized in
Table 1.
The fit parameters in
Table 1 show an almost quadratic dependence of the measurement signal on the diameter for the glass spheres. This is in accordance with theory because spherical particles have a projection area proportional to the square of their diameter, and the projection area is proportional to the intensity of scattered light if the particles are much larger than the wavelength of the used light [
15,
16]. The fit parameters for glass spheres show only a slight variation under the different measurement conditions mentioned in
Table 1.
The coal particles show only a weak dependence of the photomultiplier anode number median signal voltage on the number of the median Feret diameter. The examination of a microscopic image of those particles provided in
Figure 6 shows that those particles have a highly irregular shape and inhomogeneous light reflectance properties. Because it is not sure if the size, the shape or the inhomogeneous optical properties of those coal particles are the dominant factor for light scattering, those coal particles are not suited for calibration of measurement devices at all. However, the measurement results of these particles are used in the following paragraph to discuss the measurement of heterogeneous agglomerates containing soot and glass.
The size of the agglomerates containing both soot and glass spheres simultaneously is hard to determine because both components have different optical properties, and thus the calibration curves for both components are far apart from each other, as shown in
Figure 5. Additionally, the weak dependence of the signal voltage on the diameter of the coal particles shows that the size of optically inhomogeneous objects is hard to determine using this measurement technique. The signal voltage of soot-glass agglomerates may strongly depend on the position of the glass spheres in the agglomerate, relative to the laser-light-sheet and the photomultiplier.
Because of these difficulties in size determination for soot-glass agglomerates, a different measurement procedure is necessary to investigate the detachment of agglomerates from single fibres. In this measurement procedure, the soot in the particle structures deposited on the fibres is oxidized at low flow velocities and thus without detachment. Afterwards the flow velocity is increased to observe the detachment of solely glass sphere agglomerates.
The device’s particle size measurement range is limited by the noise measured by the photomultiplier for small particles and the maximum voltage or the size of the LLS for large particles. The extrapolation of the calibration data results in a measurement range of 257 to 1523 µm for glass spheres and 403 to 15200 µm for coal particles. With this measurement range, neither the soot agglomerates (volume median diameter of 0.352 µm) nor the single glass spheres (volume median diameter of 6.8 µm), used to initially build the particle structures on the fibre, can be measured directly. This means that the particle structure on the fibre needs to be large enough, to allow the detachment of measurable agglomerates. The detachment of small particles is still possible also with large particle structures on the fibres, however, the large detached agglomerates contribute more to the change in the particle structure of the fibre (see
Figure 1 for comparison) and are thus of primary interest of this project.
The LLS-OPC improves detection of detached particle structures over pure visual observation of the deposit on the fibre by video recording and subsequent image analysis. By image analysis of the structures on the fibre array alone, it is difficult to determine if an agglomerate detached from the fibre or just rearranged. The LLS-OPC, in contrast, reliably detects and counts detached structures above the specified minimum size.
3.2. First Detachment Experiments
Figure 7 shows a detachment experiment at elevated temperature.
The velocity in
Figure 7 (black solid line) is the velocity calculated with the nozzle temperature (blue dashed line) at 1.013 bar (absolute) and the jet temperature at fibre array (blue triangles) is derived from calibration measurements, in which a thermocouple was placed at the position of the fibre array. The accumulated detached volume is calculated by the volume of glass spheres (equal to the step height in
Figure 7) with a feret-diameter according to the measurement signal. With increasing flow velocity, the jet temperature at the fibre array first increases and then decreases because first some flow is needed in order of hot air reaching the fibre array and then further increase of the flow rate decrease the residence time of the air in the heaters and thus the air temperature. The first detached particle agglomerate was detected at a flow velocity of 2.5 m/s and its calculated volume is 0.03 mm
3. This flow velocity is rather high, because typical filtration velocities in fibrous depth filters are commonly lower than 0.5 m/s in order to prevent particle bounce during deposition [
17]. One aim of this research project is to investigate if significant detachment is possible at such low flow velocities, depending on the reactive-inert ratio, the size and the morphology of the particle structure and the detachment conditions used.
The images on the right side of
Figure 7 show images of the three parallel fibres at the start (0 s) and at the end (1045 s) of the detachment experiment. The images show that the particle structure becomes smaller during the experiment and rearrangement in the flow direction. Thus, the rearrangement/ deformation of the particle structures can be observed with the novel apparatus. It is difficult to make a connection between the detected agglomerates and changes in the particle structure on the fibre because the different particle structures overlap and the particulate structures change on different parts of the fibre simultaneously. In addition, only a part of the deposit structure on the fibres is in focus, and the focus shifts along the fibre axis between the individual fibres due to the 45° arrangement of the fibre array with respect to the observation direction.
Thus, only part of the deposit structure can be clearly observed. Here, another advantage of the LLS-OPC becomes apparent, as it captures the entire area of the deposits on the array and thus detects all detached agglomerates above a certain size limit, regardless of their spatial origin.
This experiment shows that detached particle agglomerates can be detected directly downstream of the fibre array using the LLS-OPC simultaneously with camera observation of the particle structures on the fibre array at high temperatures up to 430 °C. Therefore, this apparatus is suited to investigate the influence of the reaction, and thus the removal of particle structure components, on the detachment of large agglomerates from a single fibre or a fibre array under different conditions such as reactive content, particle size and temperature. A limitation of the current apparatus is that only glass sphere agglomerates with a light scattering equivalent diameter of single glass spheres larger than 257 µm can be detected, and thus the investigation of the detachment phenomena is limited to large agglomerates.