Open Access
Add to BibSonomy
Abstract
Stationary source emissions of key industries, such as thermal power plants, have become the central consideration in environmental protection programs. Existing photoelectric sensors at stationary sources usually use a single wavelength laser to measure the total mass concentration of the particulate matter, bearing inherent errors due to the changing particle size distribution (PSD). However, the total mass concentration cannot comprehensively estimate the air pollution caused by the stationary sources. Therefore, it is required to measure both the mass concentration and PSD of the aerosols emitted by the stationary sources, based on which we can get a distributed mass concentration. To implement this, in this study, we designed a novel three-wavelength photoelectric sensor and tested its performance. Results showed that the prototype correctly determines the mean particle size and standard deviation of the PSDs and consequently adjusts the coefficient for measuring the mass concentration from light intensity, providing a comprehensive assessment of the pollutants.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Nowadays, the Air Quality Index (AQI) is extensively used as a quantitative indication for measuring and monitoring the harmful effects of particulate matter (PM) [1] on the environment. Among the various types of PM, PM2.5 has attracted considerable attention. This atmospheric particle is usually measured by the optical methods [2–4]. The PM usually contaminated with bacteria and viruses [5,6]. Studies have shown that as the size of the PM decreases, it becomes exceedingly harmful to the human body owing to the high penetrability of the smaller particles through the alveoli into the blood [7,8]. For example, PM2.5 [9] can enter the alveoli and PM1 [10] can even directly enter into the bloodstream, while PM10 is blocked from entering the lungs. Presently, the main sources of air pollution that the Environmental Protection Administration focuses upon are stationary sources, pollution sources that unable to move, such as petrochemical plants and thermal power plants. A case study of a thermal power plant in China shows [11] that: 1) Following electrostatic precipitation, the mass concentration of PM2.5 and PM10 are 0.69 mg/m3 and 0.81 mg/m3, respectively, and the mass concentration of PM2.5 accounts for 85.23% of that of PM10; 2) after wet desulfurization, the mass concentrations of PM2.5 and PM10 becomes 1.41 mg/m3 and 2.18 mg/m3, respectively, and the mass concentration of PM2.5 accounts for 64.75% of that of PM10; 3) after wet electrostatic precipitation, the mass concentrations of PM2.5 and PM10 are 0.22 mg/m3 and 0.26 mg/m3, respectively, and the mass concentration of PM2.5 accounts for 85.94% of that of PM10. It is obvious that PM2.5 is the primary PM in the emissions of the stationary sources. Therefore, for continuous online monitoring of the stationary source PM emissions, the measurement of PM2.5 must be considered in particular, instead of just measuring the total mass concentration of the PM.
Majority of the mass concentration measurement techniques in the continuous emission monitoring systems (CEMS) at stationary sources [12,13] are based on the optical method. Determination of the PM mass concentration using photoelectric sensors is advantageous owing to their fast response and high sensitivity [14,15]. Thus, the photoelectric detection method is widely used in various fields, such as modeling of the multiple scattering properties [16,17], measurement of the scattering coefficient [18] and aerosol optical depth [19–21], determining the concentration of PM2.5 [16,22], and dust storm monitoring [23].
However, the photoelectric PM measurement techniques and equipment that are currently used at stationary sources [24] measure only the total mass concentration. The mass concentration of PM2.5 cannot be obtained from the total mass concentration without the knowledge of particle size distribution (PSD) of the aerosol. Therefore, it is impossible to comprehensively assess the harmful effects of the pollution sources on the human health. Although there are some optical methods that can measure the PSD, they either use too many wavelengths [25,26] or employ expensive and complex optical instruments [27–29]. Such as GRIMM Mini-WRAS [30], which measures particle size by making particles pass the measuring area one by one, does not have enough measuring range of mass concentration for emission from the stationary sources. Despite the optical methods, Differential Mobility Analyzer (DMA) is also a common technology to measure PSD. However, its measuring range of particle size is usually smaller than 1µm since it needs extremely high voltage separate particles by size. As a result, the corresponding equipment become very complicate and fragile, and thus, these methods cannot be applied in the high temperature and humidity environment of the stationary sources.
In this study, we designed and implemented a reliable photoelectric sensor based on the three-wavelength laser light sources to measure both the mass concentration and PSD of the aerosols from the emissions of a stationary source. The objective of the study was to obtain a distributed mass concentration, and thus comprehensively evaluate the harmful effects of PM. Further, the mass concentration of the aerosol along with the Sauter mean diameter and standard deviation of the PSD were calculated by the three scattered light intensities.
Note that, the three-wavelength measurement cannot be accomplished by simply combining three laser sources together. In fact, when three laser beams irradiate the measuring area and are scattered by the particles, the interference between the laser beams becomes a major problem. In order to solve this problem, the lasers in the present sensors are usually arranged around the measuring area to avoid the interference between the scattered light of each laser. However, this arrangement significantly increases the active area of the sensor when the lasers need to be away from the measuring area. In this study, a novel optical structure is carefully designed, in which the three laser beams do not interfere with each other within 90 mm × 90 mm × 300 mm space.
The contributions of this study are as follows:
- • We proposed a retrieval method using a novel optical sensor, which can comprehensively measure the mass concentration of PM1, PM2.5, and PM10, to replace the currently available monitoring techniques involving complex equipment, which merely measure the total mass concentration.
- • We designed the optical structure of the sensor, which can protect photoelectric system from high temperature and humidity and most importantly, avoid interference between the laser beams.
- • We tested the sensor in various practical applications, and the results showed that the sensor can correctly measure the static parameters of the PSD.
The paper is organized as follows: Section 2 introduces the proposed three-wavelength method with a scattering intensity model. Based on the method, section 3 further describes the three-wavelength sensor design in detail. Section 4 shows the numerous test results of the prototype sensor for the detection of monodisperse and real dust aerosols, followed by the conclusions in section 5.
2. Proposed three-wavelength based measurement method
In this section, we describe a method that can retrieve the PSD by assuming a certain distribution function. First, we discuss our previous investigations on the measurement of the Sauter mean diameter. Next, we develop a scattering intensity model to obtain the statistical parameters of the PSD using the three-wavelength based measurement method.
2.1. Optical measurement of the Sauter mean diameter
Optical sensors are widely used for measuring particle mass concentration or PSD. However, they can measure only the total mass concentration and are too expensive and vulnerable for continuous monitoring of stationary source emissions. Therefore, we propose an optical sensor to measure both the mass concentration and statistic parameters of the aerosol PSD.
It has been validated through several investigations that the emissions of a stationary source, which is used in a greater than 300 MW power plant, show a typical characteristic particle spectrum in which the PM2.5 number concentration exhibit a single-peak distribution [11,31,32]. Further, the peak of the number concentration is in the range of 700–800 nm. Therefore, we need to obtain the median particle size of the aerosol for calculating the single-peak distribution to determine the mass concentrations of PM2.5 and PM10.
To calculate the Sauter mean diameter, we used the Mie theory [33], which is a classical method. Generally, the PM is assumed as a sphere, and the non-spherical particles are considered as optically equivalent spheres such that the particle size is indicated by the diameter of the spheres. The Sauter mean diameter (SMD), ${D_S}$, a statistic parameter [34] for characterizing the PSD in aerosols, is defined as:
where ${C_V}$ is the volume concentration of the particulates and ${C_S}$ is the surface area concentration, respectively given by: where ${C_N}$ is the number concentration of the particulate aerosol, $f(x )$ is the function of the PSD in the aerosol, and x is the particle size.Based on the Mie theory, when incident light with wavelength $\lambda$ passes through the aerosol with a PSD of $f(x )$, its scattered light power P can be expressed as follows:
where $q({x,m,\lambda ,\theta } )$ is the scattering intensity defined as the intensity of a monochromatic light scattered by a single particle into a receiving aperture, m is the refractive index of the particulate aerosol, and $\theta$ is the observing angle.To describe the general relationship between the particle size, wavelength of the incident light, and the scattering power, Gebhart [35] developed the “three-region” law:
- I) When $x < \lambda, q({x,m,\lambda ,\theta }) = {T_I} \cdot {x^6}$, Rayleigh scattering occurs;
- II) When $x \approx \lambda, q({x,m,\lambda ,\theta}) = {T_{II}} \cdot {x^3}, q({x,m,\lambda ,\theta})$ is proportional to the particle volume;
- III) When $x > \lambda, q({x,m,\lambda ,\theta}) = {T_{III}} \cdot {x^2}, q({x,m,\lambda ,\theta})$ is proportional to the particle surface area,
For the appropriate incident light wavelength ${\lambda _V}$, which is close to the particle size $x$, the scattering light power is proportional to the particle volume concentration, as shown below [36]:
For the appropriate incident light wavelength ${\lambda _S}$, which is smaller than the particle size $x$, the scattering light power is proportional to the particle surface concentration, as shown below:
According to the definition of SMD, the expression of ${D_S}$ is given by [37]:
Moreover, the mass concentration of the aerosol ${C_m}$ is expressed as:
where $\rho$ is the particle density.According to Eq. (6), the relationship between the scattering power and mass concentration is expressed as:
where ${C_m}$ is proportional to ${P_V}$ when the incident light wavelength and particle sizes are nearly equivalent. However, when the measured particle size changes, the coefficient parameter ${T_m}$ changes accordingly. This indicates that the mass concentration measured by a single wavelength, as in the existing apparatus used at the stationary sources, are not accurate owing to changes in the measured PSD, i.e., it is unknown whether it is a small group of large particles or a large group of small particles, generating the same scattering intensity. However, using several wavelengths, the change in the PSD can be measured and calibrated, as the mass concentration depends on the PSD.2.2. Retrieval of particle spectrum based on three-wavelength scattered light
In this subsection, we propose a method to retrieve the PSD. The schematic diagram of the retrieval method is illustrated in Fig. 1.
Fig. 1. Schematic diagram of retrieving the distribution function $f(x )$ based on standard log-normal distribution.
As described in subsection 2.1, the SMD is obtained by the dual-wavelength scattered light method. However, two wavelengths are not sufficient to determine the PSD of aerosols in emissions. To this end, actually more parameters of the distribution function must be calculated. According to the available literature [11,31,32], it is reasonable to assume that the PSD of aerosol emissions conforms to the log-normal distribution. This assumption is based on the fact that PM2.5 is the primary PM in the emissions.
The distribution function of a standard log-normal is given as:
Thus, Eq. (4) can be rewritten as:
Because the measured aerosol is known, we can assume that the refractive index m is a known quantity. Even if the refraction changes, its effect is reduced to minimum, as we use the ratios of light intensity of different wavelengths for the measurement. Furthermore, based on the structure introduced in the following section 3, the scattering angle $\theta$ is likewise a known quantity. Therefore, there are only three variables $\mu$, $\sigma$, and $\lambda$ that must be solved in Eq. (14).
According to the “three-region theory,” by choosing two appropriate wavelengths, the SMD of the aerosol can be determined, which is equivalent to the median particle size $\mu$. Therefore, we deduce that using three wavelengths, we can obtain both $\mu$ and $\sigma$.
In order to validate the relationship between scattering light intensity, $\mu$, and $\sigma$, we performed simulations based on the Mie theory. In the simulations, aerosols of different median particle sizes $\mu$ and standard deviations $\sigma$ are investigated, with $\mu$ ranging from 100 to 5000 nm at intervals of 100 nm, and $\sigma$ ranging from 1.1 to 1.5 at intervals of 0.1, which represent most scenarios of a stationary source. Therefore, there were 250 types of aerosols with a log-normal distribution ${f_{ij}}(x )$ given by:
Considering the wide range of the particle diameters for measuring the surface area and volume concentration of the aerosols under investigation, the wavelengths of the laser sources are selected based on the “three-region theory” combined with the available practical components. Therefore, 450, 940 and 1550 nm lasers were chosen as the incident light sources. We define the scattering light power of each incident light through aerosols with log-normal distribution ${f_{ij}}\left( x \right)$ as $P_{1550}^{ij}$, $P_{940}^{ij}$, and $P_{450}^{ij}$ for the three incident wavelengths, respectively, ($P_{1550/940}^{ij} = P_{1550}^{ij}/P_{940}^{ij}$; $P_{1550/450}^{ij} = P_{1550}^{ij}/P_{450}^{ij}$; and $P_{940/450}^{ij} = P_{940}^{ij}/P_{450}^{ij}$) to cancel the effect of the number concentration ${C_N}$ and the refractive index m. Figure 2 shows the ratios of the scattering light power, where different colors represent the corresponding standard deviations and each point represents the median particle size.
Fig. 2. Simulated ratios of scattering light power (median particle size range from 100 nm to 5 µm).
Figure 2 shows that for each ${f_{ij}}\left( x \right)$, there would be a certain point determined by the ratio values of the three scattering light powers. Because $P_{1550/450}^{ij} = P_{1550/940}^{ij}/P_{940/450}^{ij}$, we can simply use the two ratios $P_{1550/940}^{ij}$ and $P_{940/450}^{ij}$ to represent ${f_{ij}}(x )$ on a plane, as shown in Fig. 3.
Therefore, the scattering light power’s ratios $\left( {\begin{array}{cc} {P_{1550/940}^{ij},}&{P_{940/450}^{ij}} \end{array}} \right)$ can be listed for each ${f_{ij}}(x )$ detected by the three wavelengths of the incident light. Upon measurement of the scattering light, the ratios are determined, and we can retrieve ${\sigma _j}$ and ${\mu _i}$ by comparison with the ratio in the ratio map of Fig. 3. Hence, the PSD function ${f_{ij}}(x )$ is obtained. To minimize the measurement error, we choose five points from $\left( {\begin{array}{cc} {P_{1550/940}^{ij},}&{P_{940/450}^{ij}} \end{array}} \right)$ that are closest to the measured ratio $\left( {\begin{array}{cc} {P_{1550/940}^{ij},}&{P_{940/450}^{ij}} \end{array}} \right)$ by calculating ${D_{ij}}$:
Thus, a set of the five smallest values of ${D_{ij}}$, $\min \left\{ {{D_{ij}}} \right\}$, is provided for calculating the measured ${\mu_m}$ and ${\sigma _m}$ as:
The change in the median particle size and standard deviation of the aerosol can be determined according to the change in the ratios.
3. Design of the three-wavelength optical sensor
We designed a sensor prototype based on the three-wavelength method to measure the mass concentration, SMD, and PSD of the aerosol from the emissions of a stationary source. We specially designed a compact structure to solve the interference problem caused by the multiple laser sources. The prototype consisted of four units: the optical path, air purge, circuit control, and signal processing.
A diagram of the prototype is shown in Fig. 4(a). Part 1 serves as a plug-in base, where laser sources A, B, and C are inserted. Part 2 is a block, where quartz rod $\textrm{A}^{\prime}$, $\textrm{B}^{\prime}$, and $\textrm{C}^{\prime}$ transmit the scattering lights corresponding to laser sources A, B, and C. Part 3 is a nozzle for the entrance of clean air, which forms a small positive pressure at each outlet to ensure the surfaces of all the optical components are clean. Part 4 is the shell of variable length (according to practical demand) to protect the optical path and part 5 depicts the measuring area, through which the detected aerosol passes.
The internal structure is shown in Fig. 4(b). Part 6 includes three apertures used to control the width of each incident laser beam to eliminate stray light. Each incident laser beam passes through the aperture and enters the measurement area (part 5), where it is scattered by aerosol particles. Part 7 depicts the ring-like reflector used to reflect the scattered light from incident laser rays A, B, and C to the corresponding quartz rods $\textrm{A}^{\prime}$, $\textrm{B}^{\prime}$, and $\textrm{C}^{\prime}$, which transmit the scattered light to the photodiodes in the circuit unit.
Part 8 is a tube used to place and fix the quartz rods. On this tube, part 9 depicts a window below the ring-like reflector for the reflected scattering light to pass through. Parts 10 and 11 are two baffles used for fixing the structure and preventing the particles to enter into the interior of the prototype, whereas incident laser beam and scattered light pass through freely. Parts 12 and 13 depict the optical filter and optical trap, respectively, used to eliminate the laser beam passing through the center of the ring-like reflector and prevent its return to the measuring area.
During the measurement, the laser light is scattered by the aerosol particles, and the scattered light with certain scattering angles is reflected by a ring-like reflector into the quartz rods, as shown in Fig. 5. The light is transferred within the quartz rods by total reflection. To understand the interaction of the three light paths passing into the quartz rods from the reflector, we simulate the different optical paths reflected by reflector. Figure 6 shows the irradiance map, where the three colors depict the three different wavelength of laser beams, and the circles represent the surfaces of the three quartz rods. By comparison and analysis, the scattering lights from the three laser beams scattered by the particles will be almost incident on the quartz rods at different area, such that each quartz rod receives about 83% of the laser light of the corresponding wavelength. The interference between the lasers of different wavelengths is less than 1% of the scattering lights, which is insignificant.
The recognition accuracy and detection sensitivity of the prototype require pure working gas and accurate control of the flow rate. To ensure this requirement is met, the air inlet/outlet ports are added in the prototype to protect key optical components such as reflectors, quartz rods, etc.
The three photodetectors receive the 450, 940, and 1550 nm scattered light signals and convert them into electrical signals, sending them to the signal conditioning circuit. The signal conditioning circuit amplifies the electrical signal to an appropriate voltage and transmits it to the analog-to-digital (A/D) interface of the microcontroller unit (MCU). Finally, the MCU transmits the data to the computer terminal for subsequent processing through its serial port circuit. After receiving the data, the computer terminal processes, analyzes, and displays the results in real time. A photograph of the sensor prototype is shown in Fig. 7.
The software system performs online monitoring for smoke and dust emissions from the stationary source, integrating the mass concentration and PSD. According to the relationships between different PSDs and the concentration of particles with different wavelengths of incident light that have been discussed in section 2, the particle size spectrum and mass concentration of the output particles can be calculated. Measurement results are displayed on the software interface in the form of charts and figures. The diagram and interface of the software system are illustrated in Fig. 8 and Fig. 9, respectively. In addition to PSD and the total mass concentration, the software system also displays the amounts of PM10, PM2.5, and PM1, which are not provided by other measurement systems.
4. Experimental results
The experiment comprises two stages. The first stage verifies the simulation results, and the second stage investigates the actual dust environment. In the verification stage, di-ethyl hexyl sebacate (DEHS) particles were used as aerosol material to investigate the relationship between the particle size and ratio of the scattering intensity at different incident wavelengths. In the dust testing environment, ultra-fine test dust (A1 dust), coarse test dust (A4 dust), and Japanese industrial standard class 11 dust (Kanto loam) were used as aerosol materials to test the equipment under actual operating conditions.
4.1 Verification of simulation results
The experiment platform uses TSI Model 3475 Condensation Monodisperse Aerosol Generator (MAG) to generate mono-disperse DEHS aerosol with a refractive index of 1.4 in the standard log-normal distribution. The DEHS aerosol is pumped into the smoke chamber, where our prototype sensor measures the scattering intensity. The reference instruments TSI Model 3936 Scanning Mobility Particle Sizer (SMPS) and TSI Model 3321 Aerodynamic Particle Sizer (APS) are used to measure the PSD of the aerosols sampled from the smoke chamber. An airbag embedded in the chamber balances the air pressure during sampling. A block diagram and photograph of the experimental platform are shown in Fig. 10.
By measuring the scattering intensity of the three different wavelengths incident light, two groups of ratios $P_{1550/940}^{ij}$ and $P_{940/450}^{ij}$ are obtained. The comparison is shown in Fig. 11, where the lines depict the simulation results, while the dots depict the experimental results.
Fig. 11. Experimental and simulation results of (a) $P_{1550/940}^{ij}$ and (b) $P_{940/450}^{ij}$ by DEHS with refractive index 1.4.
Due to the limitations in reference instruments, the range of the tested particle size is 800 to 4000 nm, which is controlled by saturator temperature. After calculating the ratio of scattering light intensities between 1550, 940, and 450 nm, the experimental results consistently follow the simulation curve (Fig. 10). The average error of $P_{1550/940}^{ij}$ between experiment and simulation is 18.4% and that of $P_{940/450}^{ij}$ is 23.4%. The experimental error mainly originates from the aerosol loss in different paths between our prototype and the reference instruments such as SMPS and APS.
4.2 Dust measurement
The dust measurement employs the TSI Model 3400A Fluidized Bed Aerosol Generator (FBAG) to generate aerosols of A4 coarse test dust, A1 ultra-fine test dust, and JIS test powders (Kanto loam). The well dispersed dust is pumped into a dust chamber from its top, where our prototype measures the scattering intensity of the dust aerosols. The reference instrument TSI Model 3321 APS is used to measure the PSD of the dust aerosols sampled from the dust chamber, while the Met One ES-642 dust monitor is used to measure its mass concentration. The block diagram and photograph of the dust experiment platform is shown in Fig. 12.
In the dust chamber, the dust aerosol enters from the top and disperses down slowly. The dust aerosol settles down to the bottom within hours while the prototype and dust monitor perform measurements simultaneously in the middle of the chamber. Thus, the mass concentration first reaches a peak value, and subsequently decreases slowly until most of the dust settles at the bottom. The resulting mass concentrations are shown in Fig. 13.
Fig. 13. Comparison of mass concentration measurement by Met One ES-642 dust monitor and three wavelengths of the prototype sensor (a) for A1 coarse test dust, (b) for Kanto loam and (c) for A4 ultra-fine test dust.
Results show that the mass concentration ${C_m}$ measured by the prototype agrees with the results obtained from dust monitor. The error can be described as the relevant standard deviation $RS{D_{{C_m}}}$:
Table 1. $RS{D_{{C_m}}}$ for three sample dust aerosols between ES-642 and light intensity of three wavelengths
According to the $RS{D_{{C_m}}}$ values listed in Table 1 and the curves depicted in Fig. 13, we must choose different wavelengths to measure the mass concentration for different dust aerosols. For A4 dust and Kanto loam, the 450 nm response will fit the result from the ES-642 dust monitor, whereas for A1 dust, the 940 nm response will be more appropriate. Therefore, no single wavelength laser can accurately measure the mass concentration of different aerosols whose PSD varies as the coefficient parameter ${T_m}$ definitely changes as described in section 2. Thus, the present apparatus, which measures the mass concentration of aerosol emissions with a single wavelength contains principal errors. To precisely measure the mass concentration, the wavelength of incident light must be chosen based on the PSD of the aerosol.
The $\mu $ (SMD) and $\sigma$ can be calculated based on the ratios $P_{1550/940}^{ij}$ and $P_{940/450}^{ij}$. The simulation results presented in section 3 establish the ratio map of the coordinate values, ($\left( {\begin{array}{cc} {P_{1550/940}^{ij},} &{P_{940/450}^{ij}} \end{array}} \right)$, ($i = 1,2, \cdots ,50$; $j = 1,2,3,4,5$.). Subsequently, pairs of coordinate values of $\left( {\begin{array}{cc} {P_{1550/940}^{ij},} &{P_{940/450}^{ij}} \end{array}} \right)$ obtained from the measuring result, will be compared with each simulation pair $\left( {\begin{array}{cc} {P_{1550/940}^{ij},} &{P_{940/450}^{ij}} \end{array}} \right)$ to determine the closest set of $\mu$ and $\sigma$ of the dust aerosols under investigation according to the method described in section 3.
Figure 3 in section 3 shows that when the standard deviation increases from 1.2 to 1.5, the ratio changes gradually. Therefore, we show only two lines in Fig. 14 for clarity of the measurement ratio results. Figure 14 shows the ratio map of $\left( {\begin{array}{cc} {P_{1550/940}^{ij},}&{P_{940/450}^{ij}} \end{array}} \right)$, ($i = 1,2, \cdots ,50$; $j = 1,2,3,4,5$.) depicted by the two lines and $\left( {\begin{array}{cc} {P_{1550/940}^{ij},}&{P_{940/450}^{ij}} \end{array}} \right)$ depicted by the points. The square points depict the A1 dust, the triangle points depict the Kanto loam and dots depict the A4 dust. For Kanto loam and A4 dust, the points are more concentrated at the line for $\sigma = 1.2$. In contrast, for A1 dust, the points are more dispersed and fall between the two lines for $\sigma = 1.2$ and $\sigma = 1.5$.
Fig. 14. Ratio map of simulation and measuring results. The line connecting the diamonds represents the aerosol with $\sigma = 1.2$, and the line with dots $\sigma = 1.5$. Squares: A1 dust; triangles: Kanto loam; dots: A4 dust.
The $\mu$ and $\sigma$ values measured by our prototype and reference instruments are listed in Table 2 and Table 3, respectively. Because the concentration of the dust aerosol in the dust chamber is stable for a period of time, $\mu$ and $\sigma$ values in the two tables represent the time-averaged results. The $RSD_{\mu}$ and $RSD_{\sigma}$ are calculated by comparing the results obtained by APS and the prototype by calculating $RSD_{C_{m}}$ in Eq. (19).
Table 2. µ values for APS and prototype comparison
Table 3. σ values for APS and Prototype comparison
In Table 2, the measurements obtained by APS show that $\mu$ increases in the order of A4 dust, Kanto loam and A1 dust. The prototype’s results follow the same trend; however, the values are larger than those obtained by APS. In Table 3, $\sigma$ measured by the prototype is smaller than the one obtained by APS, indicating that the distribution calculated based on the prototype measurement is more concentrated. $RS{D_\mu }$ is below 23.45% and $RS{D_\sigma }$ is below 20.73%, which is acceptable for optical measurement. Because the laser wavelength is fixed, the prototype is unlikely to detect particles larger than 5 $\mu m$, which explains why $\sigma$ measured by prototype is consistently smaller than that of the reference instruments.
These results also show that the particle size of A1 dust is larger than that of Kanto loam and A4 dust. Because of this the 450 nm laser beam measured the mass concentration of A4 dust and Kanto loam with a lower RSD compared to the 940 nm laser beam, as shown in Table 1. In contrast, the 940 nm laser beam measured the A1 dust more accurately. Therefore, our sensor can accordingly select the scattering intensity of a more appropriate wavelength with a different conversion coefficient of mass concentration ${T_m}$ with the aim to reduce the measurement error.
5. Conclusion
We proposed a method to measure the mass concentration, SMD, and PSD of aerosols in stationary source emissions. To this end, we developed a prototype sensor based on the three-wavelength technology. The prototype sensor is designed to meet the requirements of high temperature and high humidity working conditions in the power plant, and its operation is demonstrated in such conditions using the sensor at a thermal power plant for the experiment. The experimental results with DEHS aerosols indicate that the mean particle size and standard deviation of the aerosol PSD are effectively determined by the prototype. Furthermore, the experiments conducted on dust aerosols indicate that our prototype correctly measures the mass concentration and PSD of aerosol samples in real time. The measurement errors in the mass concentration of dust aerosols are 14.13% for A1 dust, 11.01% for A4 dust, and 11.53% for the Kanto loam. The measurement errors for the mean diameter and standard deviation are below 23%. In comparison with the existing apparatus employed in stationary sources that only measures the mass concentration using a single wavelength laser beam, our prototype yields not only mass concentration, but also the mean diameter and standard deviation. Therefore, the PSD of the aerosol can be determined, which is crucial for the quantitative measurement of the concentration of PM2.5 and PM1. However, it is difficult to measure the PSD of the aerosol emissions at a thermal power plant by the reference instruments mentioned in section 4 or other available experimental apparatus. This drawback of the reference instruments limits the performance and efficiency evaluation of our proposed sensor for determining the PSD of the aerosol emissions at stationary sources.
Funding
National Key Research and Development Program of China (2016YFC0201101); National Natural Science Foundation of China (61873322); National Natural Science Foundation of China (62071189).
Disclosures
The authors declare no conflicts of interest.
References
1. Z. Wang, L. Chen, J. Zhu, H. Chen, and H. Yuan, “Double decomposition and optimal combination ensemble learning approach for interval-valued AQI forecasting using streaming data,” Environ. Sci. Pollut. Res. 27(30), 37802–37817 (2020). [CrossRef]
2. W. Wang, T. Lin, C. Liu, C. Su, and S. Lung, “Fusion of Environmental Sensing on PM2.5 and Deep Learning on Vehicle Detecting for Acquiring Roadside PM2.5 Concentration Increments,” Sensors 20(17), 4679 (2020). [CrossRef]
3. J. Renard, V. Michoud, and J. Giacomoni, “Vertical Profiles of Pollution Particle Concentrations in the Boundary Layer above Paris (France) from the Optical Aerosol Counter LOAC Onboard a Touristic Balloon,” Sensors 20(4), 1111 (2020). [CrossRef]
4. W. Delp and B. Singer, “Wildfire Smoke Adjustment Factors for Low-Cost and Professional PM2.5 Monitors with Optical Sensors,” Sensors 20(13), 3683 (2020). [CrossRef]
5. J. Choi, S. C. Hong, W. Kim, and J. H. Jung, “Highly Enriched, Controllable, Continuous Aerosol Sampling Using Inertial Microfluidics and Its Application to Real-Time Detection of Airborne Bacteria,” ACS Sens. 2(4), 513–521 (2017). [CrossRef]
6. P. Du, R. Du, W. Ren, Z. Lu, Y. Zhang, and P. Fu, “Variations of bacteria and fungi in PM2.5 in Beijing, China,” Atmos. Environ. 172, 55–64 (2018). [CrossRef]
7. R. Lu, Y. Li, W. Li, Z. Xie, C. Fan, P. Liu, and S. Deng, “Bacterial community structure in atmospheric particulate matters of different sizes during the haze days in Xi'an, China,” Sci. Total Environ. 637-638, 244–252 (2018). [CrossRef]
8. H. Gou, J. Lu, S. Li, Y. Tong, C. Xie, and X. Zheng, “Assessment of microbial communities in PM1 and PM10 of Urumqi during winter,” Environ. Pollut. 214, 202–210 (2016). [CrossRef]
9. X. Hu, Q. Li, S. Shao, Q. Zeng, S. Jiang, Q. Wu, and C. Jiang, “Potential lung carcinogenicity induced by chronic exposure to PM2.5 in the rat,” Environ. Sci. Pollut. Res. 24(23), 18991–19000 (2017). [CrossRef]
10. H. Yang, S. Chien, M. Chao, and Y. Wang, “Size Distribution of Particulate Polycyclic Aromatic Hydrocarbons in the Diluted Four-stroke Motorcycle Exhausts,” Aerosol Air Qual. Res. 5(2), 225–235 (2005). [CrossRef]
11. YaFeng Wu, “Comparison about Particles Samplers for Coal-fired Pollution Sources and its Application in Power Plants,” Master thesis, (Chinese Research Academy of Environmental Sciences, 2017).
12. Y. Hu, C. Wang, X. Wu, and M. Li, “Analyzing on Monitoring Technology Situation of Fine Particulate Matters Emitted from Stationary Source,” Environ. Sci. Technol. 29(1), 69–74 (2016).
13. K. Yang, G. Zhou, Q. Wang, Q. Zhong, and E. Teng, “The Current Technical Situation and Development Tendency of Continuous Emission Monitoring System,” Environmental Monitoring in China 26(5), 18–26 (2010).
14. Y. Xu, “Improvement of Monitoring Measures on Atmospheric Particulates from Exhaust Gas by Stationary Sources in China,” Environmental Monitoring in China 33(1), 54–60 (2017).
15. Y. Liang, “Monitoring Technology Situation and Thinking on Low Concentration Particulate Matters Emitted from Stationary Sources,” Environmental Monitoring in China 29(5), 161–164 (2013).
16. M. Kahnert and R. Scheirer, “Multiple scattering by aerosols as seen from CALIPSO - a Monte-Carlo modelling study,” Opt. Express 27(23), 33683–33699 (2019). [CrossRef]
17. G. Xu, P. G. Stegmann, S. D. Brooks, and P. Yang, “Modeling the single and multiple scattering properties of soot-laden mineral dust aerosols,” Opt. Express 25(24), A990–A1008 (2017). [CrossRef]
18. K. Markowicz and M. Chiliński, “Evaluation of Two Low-Cost Optical Particle Counters for the Measurement of Ambient Aerosol Scattering Coefficient and Ångström Exponent,” Sensors 20(9), 2617 (2020). [CrossRef]
19. C. Bassani, R. Cavalli, and S. Pignatti, “Aerosol Optical Retrieval and Surface Reflectance from Airborne Remote Sensing Data over Land,” Sensors 10(7), 6421–6438 (2010). [CrossRef]
20. F. Seidel, D. Schläpfer, J. Nieke, and K. Itten, “Sensor Performance Requirements for the Retrieval of Atmospheric Aerosols by Airborne Optical Remote Sensing,” Sensors 8(3), 1901–1914 (2008). [CrossRef]
21. M. He, Y. Hu, J. P. Huang, and K. Stamnes, “Aerosol optical depth under “clear” sky conditions derived from sea surface reflection of lidar signals,” Opt. Express 24(26), A1618–A1634 (2016). [CrossRef]
22. K. Cai, Q. Zhang, S. Li, Y. Li, and W. Ge, “Spatial–Temporal Variations in NO2 and PM2.5 over the Chengdu–Chongqing Economic Zone in China during 2005–2015 Based on Satellite Remote Sensing,” Sensors 18(11), 3950 (2018). [CrossRef]
23. R. Albarakat and V. Lakshmi, “Monitoring Dust Storms in Iraq Using Satellite Data,” Sensors 19(17), 3687 (2019). [CrossRef]
24. Kevin Goohs, “Thermo Scientific PM CEMS: A dual-sensing technology,” (White Paper, Thermo Scientific, 2014).
25. T. Cheng, Y. Wu, X. Gu, and H. Chen, “Effects of mixing states on the multiple-scattering properties of soot aerosols,” Opt. Express 23(8), 10808–10821 (2015). [CrossRef]
26. H. Di, Z. Wang, and D. Hua, “Precise size distribution measurement of aerosol particles and fog droplets in the open atmosphere,” Opt. Express 27(12), A890–A908 (2019). [CrossRef]
27. J. A. Grant-Jacob, B. S. Mackay, J. A. G. Baker, D. J. Heath, Y. Xie, M. Loxham, R. W. Eason, and B. Mills, “Real-time particle pollution sensing using machine learning,” Opt. Express 26(21), 27237–27246 (2018). [CrossRef]
28. A. Chen, S. Wang, Y. Liu, S. Yan, J. Zeng, and T. Deng, “Light Scattering Intensity Field Imaging Sensor for In Situ Aerosol Analysis,” ACS Sens. 5(7), 2061–2066 (2020). [CrossRef]
29. Malvern Instruments, “Increased Dynamic Range and resolution with the Mastersizer 3000,” Technical Notes, Malvern Panalytical, 2015.
30. Grimm Aerosol Technik GmbH & Co. KG, “Portable Wide Range Aerosol Spectrometer MINI-WRAS 1371,” Manual.
31. Jikun Zhao, “Design of Dust Generation and Concentration Measurement Device Based on Light Scattering Method,” Master thesis, (Taiyuan University of Technology, 2017).
32. Xiaoyu Liu, “Characteristics of Particulate Emission from Typical Stationary Combustion Sources,” Master thesis, (Chinese Research Academy of Environmental Sciences, 2007).
33. P. Baron and K. Willeke, Aerosol Measurement: Principles, Techniques, and Applications, 2nd ed. Wiley-Interscience, 2005.
34. J. Sauter, “Die Grössenbestimmung der in Gemischnebeln von Verbrennungskraftmaschinen vorhandenen Brennstoffteilchen,” VDI-Forschungsheft Nr. 279 (1926).
35. L. Armbruster, H. Breuer, J. Gebhart, and G. Neulinger, “Photometric Determination of Respirable Dust Concentration without Elutriation of Coarse Particles,” Part. Part. Syst. Charact. 1(1–4), 96–101 (1984). [CrossRef]
36. T. Deng, S. Wang, and M. Zhu, “Dual-wavelength optical sensor for measuring the surface area concentration and the volume concentration of aerosols,” Sens. Actuators, B 236, 334–342 (2016). [CrossRef]
37. S. Wang, X. Xiao, T. Deng, A. Chen, and M. Zhu, “A Sauter mean diameter sensor for fire smoke detection,” Sens. Actuators, B 281, 920–932 (2019). [CrossRef]
