Two-Angle Ratio Scattering (STAR) Method for Real-Time Measurement of Agglomerate Soot Concentration and Size: Experimental Measurements

Abstract

The Scattering by Two-Angle Ratio (STAR) light scattering method described in this (and companion) article has been developed and tested on a range of gas turbine and diesel engines. Research literature on optical parameters has been used (without resorting to arbitrary calibrations) to predict values of the gas phase soot mass concentration, which are in good agreement with gravimetric measurements. More than 90 measurements for gas turbine and diesel engines are shown for both in situ and sampling configurations of STAR. Results are obtained for transient concentrations ranging from less than 1 μg/m³ to 100 mg/m³ at data rates up to 10 Hz. Absolute concentration comparisons with gravimetric measurements agree with an R ² correlation of 97% and have a precision of better than ±5%. These experimental results are consistent with the assumption that primary particle soot properties are nearly invariant for a wide range of engine operating conditions.

1. INTRODUCTION

In a companion article (Holve 2011), a rationale and theory for using Scattering by Two-Angle Ratio (STAR) have been developed to more accurately determine the size and mass concentration of agglomerate soots based on light scattering (patent no. 7782459). The analysis gives two simple expressions that determine the mean agglomerate diameter, d_go , and mass concentration as a function of the scattering signals measured at two distinct angles, θ ₁ and θ ₂:

where

• R_{θ 2/θ1} is the scattering signal ratio at two angles,

• is the second scattering moment ratio of agglomerate distribution, and

• is the ratio of structure functions at two values of q (scattering wave vector, cm⁻¹) at each angle.

The scattering ratio measurement gives a unique implicit expression for d_go (mean gyration diameter) as a function of the measured ratio, R_{θ 2/θ1}. The relationship for the mass concentration C_m is as follows:

where

• ρ_p is the primary particle density (kg/m³),

• d_p is the primary particle diameter (m),

• λ is the illumination wavelength (m),

• f( m ) = |( m ² − 1)( m ² + 2)|² is the scattering refractive index function,

• m is the complex refractive index of the primary particle,

• D_f is the soot fractal dimension (= 1.8),

• P_l is the laser power (watts),

• l_s is the laser beam scattering sample volume length (m),

• Ω is the receiver lens solid angle aperture for scattered light (Sr),

• P_Tθ = V_θ /η_θG_θ is the scattering power measured at each detector angle (watts),

• V_θ is the scattering signal from gas (volts),

• η_θ is the light transmission efficiency through window/hot mirror/lens optics,

• G_θ is the overall detector and electronics gain (V/w), and

• C_m # is the unique dimensionless function based on structure and moment functions and dependent on the scattering ratio R_{θ 2/θ1} only.

Equations (1) and (2) are based on well-known Rayleigh-Debye-Gans (RDG) and particle fractal aggregate (PFA) theories for light scattering of particles in which the phase shift parameter ρ = (2πd_g /λ)|m−1| _eff < 1, where |m – 1| _eff is the effective agglomerate refractive index (Sorensen 2001). Sorensen shows that for soot particles or any ramified (branched) particle with fractal dimension less than 2, ρ is in fact less than unity for any size of fractal aggregate, because the effective refractive index of the agglomerate decreases faster with increasing d_g . Sorensen also shows that RDG is accurate within 10% of detailed numerical computations for the fractal dimension D_f = 1.8 and ρ < 1.0.

Equations (1) and (2) allow explicit estimation of light scattering measurement uncertainties due to uncertainties in the knowledge of the instrument and soot properties (Holve 2011). The resulting equations are generally applicable to soot from both diesel engines and gas turbines (GTs). Measurements on a variety of fuels, laminar or turbulent research flames, and diesel engines confirm that the fractal dimension, D_f , of soot agglomerates in the large particle limit is near constant (1.7–1.9). It is also known that the agglomerate size distributions in the exhaust flow are self-preserving, forming agglomerate clusters of primary particles that have a comparatively narrow range of distribution widths (Sorensen 2001). There is an exception due to a largely thermophoretic growth/supply mechanism for wall-bound material that can become significantly larger (Friedlander 2000).

Equation (2) shows that the soot mass concentration, C_m , can be expressed as a function of instrument and primary particle soot properties and the near-invariant C_m #, which is dependent only on the measured scattering ratio at two angles. In effect, the measured scattering signals are proportional to the mass concentrations with only minor corrections for variations in agglomerate and primary particle size. Comparisons of STAR with gravimetric measurements in Section 3 confirm that STAR measurements on diesel and GT engines are consistent with assuming that values of the primary particle soot parameter S_p [(the first term in brackets of Equation (2) = {[ρ_p /f(m)]/(d_p /λ)^3−Df }] are near constant over a wide range of engine operating conditions.

There is a caveat, and this relates to the fuel/air equivalence ratio, Φ, which must not be too fuel rich, i.e., Φ must be less than ≈4, according to DeCarlo et al. (2004). Under extremely fuel-rich conditions, occurring less frequently in commercial engines, the fractal dimension can increase significantly, to a value of 2.95, decreasing the C_m # by a factor of 2. Almost all practical diesel engine combustion systems are tuned to operate at fuel/air equivalence ratios of near unity, i.e., near stoichiometric or fuel lean values.

In the following, a range of soot measurements from both diesel and GT engines by STAR are shown, demonstrating the ability to measure the mean agglomerate size and concentration with

• 1. fast response time: up to 10 Hz,

• 2. sensitivity and resolution: 1 μg/m³,

• 3. precision: better than 5%,

• 4. rapid, online calibration verification using Rayleigh scat- tering measurements of air, and

• 5. good agreement with gravimetric measurements (R ² = 97%), using the fundamental scattering analysis mentioned earlier, combined with standard literature values for the refractive index, density, and size of primary particles. This measured agreement is well within the intrinsic uncertainty of the primary particle properties (±20%).

2. IN SITU AND SAMPLING VERSIONS OF STAR

2.1. STAR Design

Figure 1 shows a scaled layout of the sampling instrument design for STAR (known as Siris), which uses sealed window cartridges that can be removed for cleaning, if necessary. We have obtained measurements with both 90° and 120° large angle detector versions. The particle flow field must pass through the laser sample volume of length l (typically less than 10 mm). The detector slits, geometry, and lens magnification define the respective sample volume lengths. In addition to this basic optical design, measurements of the pressure and temperature must be obtained for reference to normal temperature and pressure (NTP) conditions. Accurate measurements of stray light can be obtained at near vacuum conditions, allowing a direct calibration of instrument response from air scattering (Section 2.1).

FIG. 1 An example design layout of the STAR instrument, Siris.

For in situ measurements of GT exhaust, a separate instrument with an optical path length of approximately 1 m was also designed with the same effective optical geometry as in Figure 1. This 1-m box frame was fabricated using 4^″ × 6^″ aluminum box tubing, which provided sufficiently rigid alignment for the measurements. Given the high level of background light for an open in situ system, a higher power solid-state diode laser (25 mW), backlight dumps, and a light chopper, combined with interference filters and lock-in amplifiers for each detector, were used to extract the low light levels of scattering compared with the background stray light. The in situ version of STAR eliminates potential sample line losses.

2.2. Other Measurement Methods

A brief list and description of comparable instrument methods are described as follows:

The two-angle ratio method combined with absorption has been described and used previously to provide in situ measurements of soot agglomerate size and concentrations at high values of order 0.1–10 g/m³ for a small (10 mm diameter) ethylene diffusion flame (De luliis et al. 1998). This work showed good promise for the method, but the range was limited to high concentrations in order to obtain accurate absorption measurements.

In addition, there are two commercial light scattering instruments designed primarily for ambient aerosol monitoring. The Dustrak instrument manufactured by TSI is based on classical nephelometry and collects light at 90°. It has been used for both ambient dust sampling and soot measurements, but generally relies on specific calibration correlations to obtain particle concentrations. Using one light detector, it cannot provide size information from light scattering. The Dataram instrument manufactured by Thermo Scientific uses light collection at θ = 60° combined with two light-emitting diode (LED) illumination wavelengths (typically λ = 660 and 880 nm). This gives a variation in the scattering wave vector q [ = (4π/λ) sin(θ/2)] from 9.51 to 7.13, a variation of 33%. Although this range is satisfactory for the measurement of ambient aerosol mean sizes that have small values of the complex refractive index (near transparent, Lilienfeld 2000), measurement applications to soot require a broader range of q (≈5 and 13.5 provided by STAR) to accurately measure the mean size of soot from 50 to 400 nm and to provide large values of q that give near constant values of the C_m # (Holve 2011).

Laser-induced incandescence (LII) has been available for more than a decade and has been implemented as a sampling and in situ instrument (Schulz et al. 2006). Although fast (20 Hz), the method is complex and has a lower concentration measurement limit of 10 μg/m³.

The photoacoustic sampling method measures soot particle absorption in the gas phase (Beck et al. 2003) with minimum detection levels of 2 μg/m³ and a response time of 1 Hz. Other optical absorption methods use filter collection to concentrate and enhance the measured opacity. The MAAP and Aethelometer instruments measure light absorption of black carbon samples collected on a moving filter. To obtain ±1 μg/m³ detectability requires approximately 100 s of measurement time, compared with 0.1 s for STAR. This sensitivity and rapid response time are needed for rapid transient diesel engine cycle measurements following partial dilution measurements as shown in Section 3. The TEOM measures soot mass directly but requires approximately 1000 s to obtain ±1 μg/m³ resolution.

There are a number of fast scanning mobility measurement methods (EEPS and Cambustion) that have 0.1 s response time. The Dekati Mass Monitor combines mobility and inertial classification. Mass computations for these methods are based on correlations with filter mass measurements or correlations of agglomerate density with size.

2.3. STAR Calibration

The STAR system sensitivity is sufficient so that the optical and electronic gain can be measured directly from ambient air Rayleigh scattering. This provides a simple and convenient check on the instrument performance that can be conducted without the need for special calibration gases (Sorensen 2001). The total scattering cross section for air integrated over 4π Sr at NTP is σ_g = 2.51 × 10⁻²⁷ cm² at λ = 635 nm, ±2% (Miles et al. 2001; Sneep and Ubachs 2005). Using formulae for gas scattering cross sections from Van de Hulst (1981), the following expression is derived for the combined product of instrument parameters on the right-hand side in terms of the measured net scattering voltage at each detector, V_θ , and known Rayleigh ratio at NTP:

where

• [3N(∂σ_g /∂Ω)/8π] = Rayleigh ratio (m⁻¹Sr⁻¹),

• N is the number of molecules/m³, and

• (∂σ_g /∂Ω) is the differential gas scattering cross section (m²/Sr).

Repetitive calibration measurements show that the instrument parameters product (η_θG_θP_ll_θ Ω _θ ), used in Equation (2), can be measured within ±3% for both sampling and in situ versions of STAR. A larger scattering cross-section gas (e.g., propane) must be used for the in situ version because of the smaller optical numerical aperture and higher background light levels. Measurements were obtained at varying pressures and confirmed linearity with N. To cross-check and verify this method, measurement of absolute scattering at the detectors from a single-mode fiber was also obtained. The fiber tip with known radiant emittance was traversed through the sample volume, combined with individual measurements of the other instrument parameters, and gave results within the combined uncertainty of all parameters, differing from gas scattering by less than 8%. Rayleigh scattering was judged to be more accurate and provides a convenient online method for verifying instrument responsivity and accurate offset corrections for STAR as instrument components age and vary over time.

In addition to using the same response calibration method mentioned above, the in situ STAR instrument uses Scitec Instruments, Ltd. PC card lock-in amplifiers coupled with a 100-Hz chopper to minimize the effects of high background light from fluorescent test-cell lights or even higher background light from outdoor testing. While it is possible to improve the signal to background discrimination by a factor of 1000 or more using lock-in amplifiers, it is still important to use aperture cones for all detectors and the laser beams to shield optical elements from direct sunlight.

3. STAR MEASUREMENTS

3.1. Gas Turbines

Measurements were obtained with the 90° large angle detector versions of an in situ STAR and a sampling version of STAR on a turboprop engine model T56-A15 at the Rolls Royce Rework facility, Oakland, CA. Engine conditions varied from idle to full power, and at the end of each test, a fuel additive was injected to give a near three-fold reduction in the soot emissions levels, providing a wider concentration range for evaluating STAR. Figure 2 shows a comparison of the concentration (up to 100 mg/m³) and diameters (300–350 nm), which remain in good agreement throughout variations in concentration for both the sampling and in situ instruments. This agreement on the size measurements for the two instruments, combined with the precision and accuracy of the scattering response calibrations for the in situ and sampling version of STAR, suggested that the lower concentration measurements for STAR are due to sample line losses.

FIG. 2 Comparison of in situ STAR and sampled STAR mass concentrations and mean diameters over a portion of a GT engine test cycle. The decrease and subsequent increase in mass concentration correspond to the period of injection of a soot suppression fuel additive. The mean agglomerate size for both STAR instruments agree and show a small increase during the additive injection.

Subsequent measurement comparisons in our laboratory with a short sample line confirmed this result by using a conventional Volkswagen diesel exhaust. We compared the effects of line loss and nonisokinetic sampling by using the STAR in combination with short (5 feet) and long (33 feet) ¼^″ diameter heated (approximately 100°C) sample lines. There is a small decrease (5%) with increasing sample rates varying from 3 to 15 Lpm for both lines, which is consistent with superisokinetic sampling at the probe nozzle (exhaust flow velocity was not measured). The overall sample loss in the longline was 26% compared with the short line, which has negligible line loss, and there was no measurable change in the median agglomerate diameter. The longline loss is similar to the concentration differences measured by the longline sampling STAR instrument in comparison with the GT in situ measurements (20%–30%) and confirms that the differences are primarily due to sample line wall losses. Long sample lines are unavoidable for large-scale turbine measurements with STAR, but line losses can be determined by performing similar short and longline comparisons.

A subsequent test series was performed on a marine GT (Kawasaki 501-KF) measuring soot concentration and particle size with STAR and obtaining parallel gravimetric mass concentrations. This is a newer engine design than the turboprop engine mentioned earlier and gives approximately 10 times lower mass concentration values. Both STAR and comparative filter mass measurements were obtained during this test series. The results are discussed and compared with diesel engine results (Figure 3) in the following sections.

3.2. Diesel Engines

3.2.1. Methodology

The application of STAR for the measurement of exhaust emissions is subject to stringent requirements based on EPA 40CFR Part 1065 (henceforth “1065”), ISO16183, and others. For these conditions, we used the Siris geometry shown in Figure 1 with a large angle detector of 120 degrees that uses a different value of C_m # as described by Holve (2011). The Siris instrument was integrated with the BG-3 partial flow sampling system (PFSS, patent no. 5058440) manufactured by Sierra Instruments, operating in a heated filter enclosure maintained at 47°C as defined within 1065. The BG-3 is able to follow both transient test cycles at 80 Hz, as well as quasi-steady flows, varying exhaust dilution ratios from typical values of 6 or more. The BG-3 employs a radial inflow dilution tunnel that substantially reduces particulate losses by boundary layer control (Graze 1994). The tunnel is closely coupled to the exhaust stack to mitigate particle losses in the undiluted exhaust stream. Once diluted, the exhaust is transported through a short (one-foot) insulated and electrically grounded line after which it enters the heated filter enclosure housing the Siris module. This configuration addresses aforementioned transfer line losses, while providing a fully transient-capable dilution system platform with proven correlation to the CVS systems that comprise the EPA reference devices (Shade et al. 2009).

This integrated system BG-3 using the STAR concept is known as Soot Trak (ST) with Siris as the two-angle light scattering instrument, with a response rate of approximately 10 Hz. ST was used to measure two nonroad diesel engines at a major engines manufacturer (EM), over a wide range of operating conditions, including quasi-steady and transient test cycles. Comparisons of gravimetric filter measurements were obtained, using the in-house filter weighing lab, with capabilities of ±2.5 μg accuracy. In addition, EM performed measurements of the soluble organic fraction (SOF) for most samples to assess the degree of organics adsorption on the filter itself. All following measurements were obtained by EM technicians, and in most cases, a minimum of three repeat runs for a given test was obtained.

ST and filter comparison measurements are expressed in g/h at the engine exhaust, taking the sampled ST concentration or filter mass at measured temperature and pressure, and combining with the engine exhaust flow (m³/h at NTP) and the PFSS dilution ratio. All flows and dilution ratios of the BG-3 are known within ±2% and calibrated weekly.

FIG. 3 An overview of all data comparisons of ST versus gravimetric measurements of PM and PM minus SOF. The R ² value for PM is 95% and improves to R ² = 97% for PM-SOF.

3.2.2. Overall Comparison of ST with Gravimetric Filter Results

Previous measurements with GTs and conventional diesel vehicles at Process Metrix (PMC) showed reasonable agreement (±25%), although filter weighing facilities at PMC were limited, and there was no capability for measuring the percentage of SOF, or %SOF. Our first objective has been to confirm the fundamental basis of analysis in Holve (2011) with the goal of validating both accuracy and near invariance of the soot property parameter, fixed at S_p = 141, over a wide range of engine-operating conditions. In addition, Holve (2011) has shown little variation (<15%) in S_p for soot that has adsorbed liquid hydrocarbons up to 50% of the total mass, i.e., ST measures the total mass of “organic carbon (OC) coated soot.”

All measurements shown in Figure 3 for both ST and gravimetry (except GT PM) are taken using the BG-3 PFSS. The GT measurements are compared in units of concentration (mg/m³) at NTP. Most diesel measurements are derived from a single engine (no. 1) with a maximum rating up to about 300 hp, along with a second diesel engine (no. 2) with a maximum rating of around 250 hp. Engine 1 employed a passive CCRT (Catalyzed Continuously Regenerating Technology, Johnson-Matthey) with an approximate soot removal efficiency of over 99%. In order to evaluate ST over a wide range of concentrations using both steady-state and transient test cycles, four different switchable, fixed diameter bypass (BP) spools were designed into the CCRT assembly to allow simulation of exhaust PM between approximately 0.007 and 0.025 g/hp-h on the United States Environmental Protection Agency (USEPA) transient on-highway test cycle (FTP). The diameters of the spools were 0.25^″, 0.5^″, 0.75^″, and 1.00^″; it should be noted that not all spool sizes were used for all tests.

Figure 3 shows high emissions rates without the CCRT, while Figure 4 shows low emissions rates with the CCRT and various levels of BP. Correlation between ST and PM-SOF is improved for all measurements, subtracting filter adsorption of gas phase hydrocarbons (positive mass artifact). All measurements, unless otherwise noted, were taken with TX-40 filter media. More than 90 measurements are shown for quasi-steady (low to high loads) and transient test cycles. Examples of transient cycle are discussed in Section 3.3, which examines the repeatability and precision of fast ST measurements.

FIG. 4 Expanded scale of Figure 3, showing the correlation at low-level emissions conditions. Eight-mode (ISO8178-4 C-1) total emissions at average 147 hp were repeated four times at four different CCRT BP levels and ST correlates well with filter results. Four levels of spool BP were used to acquire measurements at low emissions rates. The percentage variance of ST measurement ranges from 5% to 22%, while PM-SOF measurement variances range from 24% to 107%, both values increasing with decreasing PM level. NRTC (average 32 hp) and RMC (average 71 hp) transient cycles show increasing filter artifact mass with decreasing CCRT BP.

Figure 3 shows two primary sets of results, one that includes the total filter mass (PM) and the other with SOF subtracted. In both cases, the R ² values are above 94%, although it is clear that the gravimetric PM results are increasingly higher than the ST results at lower emissions rate. Correction for SOF significantly improves the agreement at lower levels. Also, note that the GT measurements obtained by PMC without SOF removal are in general agreement with the diesel PM results, although the scatter is greater.

The absolute agreement of ST with PM-SOF is within 8% of the theoretical value for S_p (which has a literature uncertainty of ±20%). It is important to note that the measurements occur over all engine-operating conditions from idle at low horsepower (<10 hp) up to a maximum of 270 hp, showing that there are not significant variations in the soot property parameters with engine operating conditions.

Although some of the SOF removal occurs on the soot itself, the majority of SOF appears to be captured by the filter. It was found that above 25 hp (9% of the full load) operating conditions, the SOF fraction ranged from 15% to 25% of the total measured PM, increasing rapidly at near idle (<1% of the full load) operating conditions to 50% and higher. Some filter measurements used two TX-40 filters, where the first captured solid “soot” plus adsorbed gas phase species and small (<10 nm) droplets, while the second filter collected only the gas phase components and droplets. Although generally lower, the second filter method produced results comparable to the extractive SOF removal method, indicating that both filters adsorb and approach equilibrium with the gas phase hydrocarbons. Measurements of total hydrocarbons over a test cycle show that the gas phase mass can be 10–100 times greater than that collected by the filters, thereby exposing both filters to the same concentration. Similar results have been obtained in a study by Khalek (2007), discussed further below.

Figure 4 more clearly shows the importance of filter adsorption and collection of gas phase species for a broad range of operating conditions. First, the agreement with 8-mode operating conditions at relatively high (147-cycle-weighted hp) engine powers is good down to 0.03 g/h. The 8-mode cycle consists of 8 quasi-steady measurements at 5-min intervals, ranging from <10 to 270 hp for engine 1. Although individual filter measurements were taken for each individual 5-min steady-state condition, the total of all 8 modes was added and is compared in this figure for four BP conditions around a diesel particulate filter (DPF), as previously described. The uncertainty bars (based on four repeat measurements) show the percentage of variance of both ST and filter measurements, with the filter measurement variability being significantly greater than that of the ST results, increasing to more than 100% coefficient of variation (1 SD/mean) at the lowest emissions rate of 0.03 g/h. These conditions tend to weight the higher horsepower conditions, where the fraction of elemental carbon (EC) is higher than OC. Measurements were also taken with Teflon filters, giving close agreement. Analysis of the zero BP and 1.0^″ BP results shows that CCRT soot removal efficiency is significantly greater than 95%.

Results were also obtained for two other types of transient cycles: the nonroad transient cycle (NRTC) using a 20-min cycle time with an average power of 32 hp and the ramped mode cycle (RMC) using a 30-min cycle time with an average power of 71 hp for engine 1. In both these cases, the 80% and 100% BP conditions show good agreement with the overall correlation functions established in all the other measurements. However, at low emissions rates, both the NRTC and RMC values of PM-SOF are significantly greater for the filter results than for ST. The primary reason for the difference between the 8-mode quasi-steady measurements and transient results appears to be the lower average horsepower conditions for the transients, which weight idle conditions with higher levels of gas phase constituents compared to EC. Although SOF has been removed from the filter mass measurements, there can be other significant contributions (e.g., sulfates, nitrates, etc.) that become significant at low concentrations as indicated by Khalek (2007) and Andersson et al. (2002). We note that the ST measurements of NRTC and RMC test cycles are consistent with the estimate of significantly greater than 95% removal of solids by the CCRT at zero BP conditions, consistent with the 8-mode results.

FIG. 5 Comparison of particle mass and number of agglomerate particles above 30 nm for ST, EEPS, and filter for mode 5 at 0.25^″ BP. The EEPS results assume a uniform density of 1 g/cm³ for agglomerates.

Measurements by Khalek (2007) and colleagues show a similar behavior at low emissions following a DPF. Their results show factors of 5 or more mass from filter measurements than for real-time mobility measurement instruments, and they attribute the bulk of the discrepancy to filter mass artifacts derived from condensation of gas phase species on the filter. They conclude that gas phase species and small droplets are adsorbed onto the filter, which has a much larger surface area than the collected “soot,” which itself may already be in equilibrium with the gas phase hydrocarbons. The detailed mechanisms for this “filter artifact” are not entirely understood, but other workers have independently confirmed these results, e.g., Maricq et al. (2006) and Andersson et al. (2002).

ST results were compared with mobility measurements (EEPS) of mass and agglomerate particle number above 30 nm for three repeat runs of mode 5 with 0.25^″ BP as shown in Figure 5. The EEPS results in this case assume a uniform density of 1 g/cm³, and ST uses the mean gyration diameter, combined with an assumed distribution shape, to compute the total agglomerate particle number. ST also computes the number and surface area of primary particles, based on the fact that soot agglomerates are ramified and the majority of primary particles is exposed to surface contact. The average number of primary particles per agglomerate ranged from 60 to 120, based on the measured mean gyration diameters (240–340 nm). For the same test condition, although two of the measurement results are comparable, one result is approximately 3 times larger than the other. Both ST and EEPS tracked this anomaly along with the filter measurement. This deviation was eventually traced to an engine hardware failure. Given the density approximations for EEPS and the assumed distribution shape for ST, the agreement on both number and mass is consistent with the filter measurement uncertainties.

3.3. Diesel Test Cycle Transient Measurements

Figure 6 shows time-correlated results of four transient NRTC tests at various levels of flow BP around the DPF. The corresponding high and low emissions rates versus time validate the detailed repeatability of transient response at widely varying soot loadings generated by the various BP levels. The cumulative emissions rates shown in Figures 3–5 were obtained by integrating these transient measurements over the entire test cycle.

FIG. 6 ST measurements showing NRTC transient replication for a passive DPF at four BP flows. The results for 0.75 BP are nearly the same as for 1.0 BP and are mostly covered by 1.0 BP.

Figure 7 shows four repeat measurements for the RMC test cycle obtained over three separate days giving cumulative cycle concentrations ranging from 3.3 to 3.5 μg/m³. The running cumulative cycle concentration is based on the time-weighted concentration integrated over the entire test cycle and divided by the total cycle time. The total cycle concentration can then be compared with the total cycle filter concentration measurements. Comparison of gravimetric measurements gave values ranging from 10 to 16 μg/m³ consistent with the filter artifact mass results described in Figure 4. These overall concentration levels are quite low and illustrate the difficulties of obtaining statistically relevant filter measurements at filter mass loading levels below about 15 μg. In comparison, the ST measurements can provide a detailed insight into the effects of test cycle transitions at soot levels below most commercially available soot measurement devices.

FIG. 7 ST comparison of four repeat runs of RMC test cycle conditions following a DPF at 0.25^″ BP.

4. SUMMARY AND CONCLUSIONS

The primary objective of this work has been to develop a fast response and robust optical instrument based on fundamental RDG and PFA theories that can quantitatively measure gas phase engine exhaust soot concentrations and agglomerate particle size. The measurements of mass concentration, agglomerate size, and agglomerate and primary particle number presented here are based on the analysis model developed by (Holve 2011), combined with the assumption that the soot primary particle parameter, S_p , is near invariant. Soot properties are based on recent measurements of the refractive index, which give consistent values for both scattering and absorption. The model and assumptions have been confirmed by comparisons with gravimetric measurements for diesel engines (R ² correlation = 0.97) over a wide range of operating conditions.

A simple method for instrument verification and calibration based on measuring the Rayleigh scattering of air is used. A range of measurements have been performed on GTs using both an in situ and a sampling STAR method. A comparison of in situ and sampling measurements is able to show the effects of sample line loss, which are approximately 20%–30% for a 10 m sample line.

A more recent version of the STAR method has been implemented in combination with a PFSS (BG-3) and is known as soot track. In addition to the absolute accuracy of ±10% based on fundamental soot properties described earlier, the precision (±5%) and speed of 10 Hz for the STAR method are demonstrated. It has been shown that the absolute concentration resolution for STAR is approximately 1 μg/m³, which is important for diesel engine exhaust measurements following a CCRT combined with subsequent dilution using a PFSS. Comparisons with a mobility method (EEPS) also show comparable results for both mass and agglomerate particle number above the minimum primary particle size, i.e., 30 nm.

Agreement with gravimetric measurements is shown at concentration and emissions rates varying by more than a factor of 1000. However, at very low concentrations following a CCRT, the adsorption of gas phase species by the gravimetric filter can generate apparent values of mass that exceed ST results by factors of 2–5. Other workers have described this additional mass as an artifact of the filter itself, and not a true measure of gas phase PM. Indeed, ST measurements of PM are consistent with CCRT solids’ collection efficiencies of greater than 95% for all measurements obtained.

Acknowledgments

This work was supported in part by the Navy and Air Force SBIR programs. Linda Riedlinger of Caterpillar squeezed in many extra hours of data acquisition and processing. Her dedication is gratefully acknowledged. Thanks to my partners and Alan Alsing for their skill in implementing Siris.