Results from the Pre-Tests on a HDV at the Chassis Dynamometer at TU-Graz

Transcription

1 INSTITUTE FOR INTERNAL COMBUSTION ENGINES AND THERMODYNAMICS A-8010 GRAZ (AREA CODE [++43/316]) Inffeldgasse 25 Tel.: Fax HEAD: Univ.-Prof. Dipl.-Ing. Dr. Rudolf PISCHINGER GRAZ UNIVERSITY OF TECHNOLOGY Erzherzog-Johann-Universität Results from the Pre-Tests on a HDV at the Chassis Dynamometer at TU-Graz Internal report for PARTICULATES Draft Version 9/2002 Elaborated by: Edim Bukvarevic D.I. Dieter Engler D.I. Dr. Stefan Hausberger Bericht Nr.: ISO 9001 Pre-Tests_Report_Draft3.doc certified by TÜV Management Service page 1 of 50 Erstellt von Inst. f. VKM u. THD

2 CONTENTS : 1. ABSTRACT INTRODUCTION MEASUREMENT SET-UP Vehicle and engine specifications Laboratory equipment set-up Particle measurement set-up Dilution systems Number size distribution sampling units TEST PARAMETERS Driving conditions Steady-state tests Transient tests Fuel quality TEST RESULTS AND ANALYSIS Dilution ratio Transient dilution tests Number size distribution Steady-state tests Scanning Mobility Particle Spectrometer (SMPS) Electrical Low Pressure Impactor (ELPI) Double Differential Mobility Particle Spectrometer (DDMPS) Comparison of size distributions with different sampling systems Steady-state tests summary Transient tests Comparison of TrDMPS and ELPI results Morphology Sampling procedure Results Summary CONCLUSIONS Dilution ratio Number size distribution Morphology Draft Version 2/50

3 1. ABSTRACT This report documents evaluation of the pre-tests particle emissions measurements on a HDV performed at the chassis dynamometer at TU-Graz. In these tests various properties were measured with an increased focus on particle number size distribution in both steady-state and transient tests. The effects of several test parameters were investigated, such as measurement set-up (primary dilution units, particle sampling systems), driving conditions (engine load, test cycles) and fuel quality (sulphur content). All dilution and sampling systems were capable of producing reproducible measurements of number size distributions, especially for accumulation mode particles. Nucleation mode particles were confirmed to be much more sensitive to measurement set-up and sampling conditions. Low sulphur fuel was confirmed to reduce nucleation mode, but accumulation mode was much less affected by the sulphur content. Draft Version 3/50

4 2. INTRODUCTION Diesel exhaust particle emissions have become the focus of increased attention due to the concern of potential health effects. While total mass particle emissions are well studied and documented, less data concerning other, perhaps more significant physical and chemical properties (size, number, composition, etc.) are available. It is known that measurement and characterization of particle number and size is more complex than total mass measurement, especially for transient tests, and compatibility of different measurement techniques and systems is not as good as for total mass, measured by standard filter method. Consequently, further work is required on the sampling, calibration and standardization of the different measurement techniques. Recently, attention has focused on the number-based size distribution of exhaust particles. Diesel particles are a complex mixture of elemental carbon, a variety of hydrocarbons, sulphur compounds and other species. Many of these species are volatile and may be in the gaseous or the particulate phase, depending on temperature, dilution ratio and other parameters at location of sampling. Depending on these parameters different amounts of volatile species will be measured as particle emissions. The size distribution is generally bimodal, with peaks below 30 nm (nucleation mode) and above nm (accumulation mode). Sometimes a third mode can be observed coarse mode (1-10 µm), which contains re-entrained material (previously deposited in the exhaust system) and is not very reproducible. Nucleation mode consists mainly of volatile material, which makes it highly sensitive to sampling conditions. In many cases particles in the nucleation mode can be completely removed by heating. The nucleation mode contains only small percentage of particle mass, but may contain more than 90% of their number. Accumulation mode, also named soot mode, consists of solid material (elemental carbon, ash) and usually is stable and reproducible. In this context the objectives of the HDV Pre-Tests performed at the TU-Graz were: Testing of the prototype of Particulates dilution sampling system on a HDV according to Deliverable 4; Increase of knowledge and understanding of particle emissions from motor vehicles; Comparison of different dilution systems and analysis of their effects on measured emissions; Comparison of different particle size distribution sampling systems and their effect on the results; Evaluation of the effects of fuel quality (sulphur content) on particle emissions; Evaluation of the influence of driving conditions (engine load and speed, driving cycle). Draft Version 4/50

5 3. MEASUREMENT SET-UP 3.1. Vehicle and engine specifications The measurements were conducted on a Heavy Duty Vehicle with following specifications: Table 3-1 Vehicle and engine specifications Model SCANIA R124 LA 4X2 NA 420 (EURO 3) Engine Capacity cm 3 Max. Power 309 kw at 1900 rpm Max. Torque 2000 Nm at rpm Bore to Stroke 127/154 mm Compression Ratio 18: Laboratory equipment set-up All measurements were performed on the HDV chassis dynamometer at the Institute for Internal Combustion Engines and Thermodynamics (Technical University Graz TUG). HDV Chassis Dynamometer The mechanical test stand unit is built in form of a steel frame construction, in that the modules roller set, flywheel and electrical brake unit are installed. The test stand frame is based on a steel frame integrated in the building and is connected with the building structure by anti-vibration elements. The brake is a thyristor-controlled d.c. machine, which can be driven as generator (brake operation) and motor (motoring operation). The brake control is appropriate for stationary and transient driving. The determination of the traction force at the point of tyre-contact takes place via measurement of torque at that oscillating supported brake machine by means of a load cell, which operates according to the DMS principle. The simulated vehicle speed is recorded by measuring the roller speed. The test stand is equipped with a wind simulator, to achieve comparable thermal engine cooling conditions as in real driving. Table 3-2 HDV Chassis Dynamometer technical specifications Max. Traction Force Max. Braking Power Max. Drag Power Max. Speed Vehicle Mass Roller Diameter Max. Axle Weight 27 kn 360 kw 290 kw 120 km/h t 0.5 m 12 t Draft Version 5/50

6 CVS System The CVS (constant volume sampling) system, together with the exhaust gas analysing system, is a complete measuring system to record the exhaust emissions of diesel engines. The system can be used for steady state and transient conditions. The system is automatically controlled by the software TORNADO from Kristl&Seipt Engineers. For measurement of the gaseous emissions an AVL CEB II bench is used. Technical specifications: Heating facility to control the inlet air temperature to 25 +/- 5 C Dilution air filter container consisting of five units (coarse filter, active coal filter and fine filter) Heated probe and transfer pipe for continuous HC and Knox measurement at the end of the tunnel Probe and transfer pipe to the secondary dilution tunnel for measurement of particulates Heat exchanger for temperature control of the air - exhaust gas mixture to 50 +/- 11 Three parallel venturi nozzles with a nominal flow rate of 30, 50 and 60 m³/min, and three valves for the choice of 30, 60, 90 or 120 m³/min CVS flow rate Centrifugal blower The chassis dynamometer and the main parts of the CVS system are given in Figure 3-1: fresh Filterkästen air filter fresh Frischluft air roller Rollen - Beobachtungs control - unit raum exhaust Abgasgas driver Fahrerbildschirm screen dilution Verdünnungstunnel tunnel roller Rollensatz set CVS CVS control - Steuerung unit fan Fahrtwindsimulator analyzers Abgasanalyse heated Beheizte Leitung line (HC, NOx) particle Partikel Probe sample electric elektrische brake Bremse venturi Lavalnozzles - Düsen ventilator Ventilator bag Probenentnahme sample Beutel cyclone Zyklon heat Wärmetauscher exchanger exhaust Abgaskamin Figure 3-1 Schematic of the HDV Chassis Dynamometer and CVS System Draft Version 6/50

7 3.3. Particle measurement set-up The measurements in the Pre-Tests Programme consisted of performing steady state point and transient tests with various combinations of different primary diluter and particle sampling systems, in order to characterize the effects of measurement configuration parameters on the test results. Simultaneously with the size distribution measurement equipment, in the most tests the CVS system was used for measurement of total particle mass with standard filter method. Additionally, two NOx analysers were used and placed before and after the respective diluter, in order to measure NOx concentrations in the undiluted and diluted flow and thereby calculate and observe the dilution rate as ratio of the two concentrations. Figure 3-2 shows a schematic of the basic sampling configuration used in the pre-tests. Exhaust gas from tailpipe CVS Smart- Sampler ELPI SMPS NOx 2 Dilution air PARTICULATE Dilution DEKATI Dilution Smart- Sampler SMPS NOx 1 NOx 2 DDMPS TrDMPS Figure 3-2 Schematic of the particle measurement configuration with modifications The different combinations of diluters and particle sampling systems used are given in Table 3-3. Table 3-3 Measurement matrix Measurement equipment Dilution System ELPI SMPS DDMPS TrDMPS 2 NOx PARTICULATE Dilution tests DEKATI Smart Sampler DDMPS (closed system, steady state only) TrDMPS (closed system, used transient only) Dilution tests Dilution tests Draft Version 7/50

8 3.3.1 Dilution systems Five different dilution units were used: DEKATI primary diluter, Particulates primary diluter, Smart Sampler, DDMPS dilution (closed system with its own dilution unit) and TrDMPS dilution (closed system with its own dilution unit) Number size distribution sampling units Following systems were used: Scanning Mobility Particle Spectrometer (SMPS), Electrical Low Pressure Impactor (ELPI), Dual Differential Mobility Particle Spectrometer (DDMPS) and Transient Differential Mobility Particle Spectrometer (TrDMPS). ELPI was always used at the dry branch (i.e. after thermodenuder) of the Particulate and DEKATI Dilution system. When used with the Particulates Dilution SMPS was either used at the wet or also at the dry branch during some measurements. The comparisons between ELPI and SMPS which are shown in Figure 5-15 through Figure 5-18 are from the measurements where ELPI and SMPS were located at the dry branch. Draft Version 8/50

9 4. TEST PARAMETERS The following testing parameters that could affect the test results were varied: measurement set-up (technique, equipment, dilution parameters, etc.), driving conditions and fuel quality. Variations of the measurement set-up were shown in chapter Driving conditions In the pre-tests measurement programme both steady-state tests and transient test cycles were used. Each transient test cycle was repeated twice Steady-state tests Following steady-state points were involved: 0% load, 500 rpm (idle) 10% load, 1060 rpm 25% load, 1060 rpm 50% load, 1270 rpm Full load could not be driven long enough due to overheating of wheels Transient tests Three real world cycles and an additional transient dilution test (used to synchronize the two NOx analysers and to measure and observe real dilution ratio fluctuations with different dilution systems) were used: Transient dilution test cycle, Urban real world cycle, Rural real world cycle and Highway real world cycle Fuel quality Two different fuels were used: Diesel fuel D-2 EN (sulphur content: 300 ± 50 mg/kg) Diesel fuel D-5 EN Swedish Class 1 (sulphur content 10 mg/kg max) The most tests were performed with the low sulphur D-5 fuel; only one series of steady-state tests was performed with the D-2 fuel. Draft Version 9/50

10 5. TEST RESULTS AND ANALYSIS 5.1. Dilution ratio The dilution ratio plays an important role in measurements of exhaust particle emissions. In the pre-tests measurements the stability of dilution ratio during transient tests was of special interest. Three different dilution systems (Particulates, DEKATI and Smart Sampler), used as primary dilution units, were tested and compared. Two NOx analysers were used to measure NOx concentrations in undiluted and in diluted exhaust gas flow, in order to obtain a measured dilution ratio. This measured dilution ratio was calculated as the ratio of the two NOx concentrations. While the dilution ratio during steady-state tests, as expected, was very stable, the transient tests deserve a closer investigation Transient dilution tests A transient test cycle, the dilution test, was performed in order to synchronize the measured NOx concentrations and to investigate the fluctuations of the measured dilution ratio for all dilution systems used. The test includes several steady state operations at idling, 10 km/h and 70 km/h to assess the delay times of the analysers (Figure 5-1) Vehicle Speed, km/h Time, s Figure 5-1 Dilution Test Cycle The dilution ratio was set to a start value before each test. The calculation of the dilution ratio DR(NOx) resulted in considerable fluctuations even after the best possible correction of the time shift between the two sensors. Since the time resolution of the sensors was 1 s, a considerable share of these fluctuations is likely to be result of the remaining time shift. In order to obtain a dilution ratio, which would better match the real dilution, various smoothened dilution rates were calculated. The one that was calculated as a ratio of averaged NOx concentrations, to compensate for measurement related fluctuations, DR(NOx_m), was assumed to be more realistic than DR(NOx). The results are given in Figure 5-2 through Figure 5-4, with: DR(NOx) = NOx/NOx dil Draft Version 10/50

11 DR(NOx_m) = NOx_m/NOx dil _m NOx_m = (NOx i-1 + NOx i + NOx i+1 )/3 NOx, NOx dil Concentrations in undiluted and diluted exhaust flow, v Vehicle speed, n Engine speed, F Braking force of the chassis dynamometer. Dilution Test Cycle - Smart Sampler Dilution n v DR(NOx) DR(NOx_m) v [km/h] F [x10 N] n [rpm] DR, v n, F F DR Time [sec] Figure 5-2 Dilution ratio with Smart Sampler in dilution test cycle Dilution Test Cycle - DEKATI Dilution n v DR(NOx) DR(NOx_m) v [km/h] F [x10 N] n [rpm] DR, v n, F F DR Time [sec] Figure 5-3 Dilution ratio with DEKATI primary diluter in dilution test cycle Draft Version 11/50

12 n Dilution Test Cycle - Particulates Dilution v DR(NOx) DR(NOx_m) v [km/h] F [x10 N] n [rpm] Dilution Rate F DR n, F Time [sec] Figure 5-4 Dilution ratio with Particulates primary diluter in dilution test cycle Another problem with the measured dilution ratio DR(NOx) is that at lower and negative engine loads (deceleration) NOx concentrations decrease drastically (almost zero), which leads to inaccuracies in calculating dilution as ratio of concentrations. In Figure 5-2 through Figure 5-4, it can be observed that under overrun conditions during deceleration, the fluctuations of this ratio increase and it results in very high peaks, which are probably unrealistic. In spite of these difficulties, we found the dilution ratio calculated as ratio of NOx concentrations to be a useful variable. The smoothening of these concentrations reduces the effect of the remaining time shift and results in a more realistic dilution ratio. The effect of inaccuracy at low NOx concentrations remains and cannot be reduced. All three dilution systems show very similar behaviour. Although the dilution units do not seem to be able to keep the dilution ratio stable during dynamic conditions in a transient test, a considerable fraction of these fluctuations might be result of the above-discussed effects. Draft Version 12/50

13 5.2. Number size distribution Steady-state tests All steady-state tests were run for 10 minutes with warm engine. Each test series, consisting of four steady-state points, was carried out using three particle sampling systems simultaneously (SMPS, ELPI and DDMPS). SMPS and ELPI were combined with various dilution systems as follows: 1. test series: SMPS + Smart Sampler; ELPI + DEKATI primary diluter; 10 ppm S fuel 2. test series: SMPS and ELPI + Particulates primary diluter; 10 ppm S fuel 3. test series: SMPS and ELPI + Particulates primary diluter; 300 ppm S fuel Scanning Mobility Particle Spectrometer (SMPS) During each steady-state test three to five scans were made with the SMPS system. Stability of the scans was found to be very good in all tests, independent of set-up. This can be seen in Figure 5-5, which shows five SMPS scans in one of the steady-state tests with Smart Sampler diluter. SMPS Particle Size Distribution - Dilution: Smart Sampler (12.5) Idle (500 rpm, 0% load) (LKW ) Mean Value Dp [nm] Figure 5-5 Stability of the SMPS scans in steady-state tests In Figure 5-6 effects of both dilution system and fuel quality (sulphur content) can be seen. Draft Version 13/50

14 SMPS Particle Size Distribution rpm, 10% load 1.00E E E+00 Smart Sampler DR = 12.5; 10 ppm S Fuel Particulates DR = 14.3; 10 ppm S Fuel Particulates DR = 14.3; 300 ppm S Fuel Dp [nm] Figure 5-6 Effects of dilution system and fuel quality on SMPS particle size distribution SMPS Particle Size Distribution Idle (500 rpm, 0% load) Mean 1060 rpm, 10% load Mean 1060 rpm, 25% load Mean 1270 rpm, 50% load Mean Dp [nm] Figure 5-7 Effect of engine load on SMPS particle size distribution Two different effects of the sulphur content in the fuel could be observed. For particle diameters smaller than 20 nm (nucleation mode) low sulphur fuel reduced particle number. This effect is most significant at idling, decreases with engine load and disappears at high engine loads (no nucleation mode). Particle diameters larger than 20 nm (accumulation Draft Version 14/50

15 mode) were affected in a way that low sulphur fuel reduced particle number at idling but increased at higher engine loads. However this effect is very small and might be result of other unknown parameters. SMPS measured slightly higher particle numbers with Smart Sampler than with Particulates diluter, which could be observed especially at higher loads, but this effect was also small and might be due to changed ambient conditions and not the dilution unit itself. Mean values of all 3 test series for each steady-state point were calculated to illustrate the effect of engine load on particle size distribution (Figure 5-7). Nucleation mode decreased with engine load and disappears completely at high loads. Accumulation mode increased with increasing engine load Electrical Low Pressure Impactor (ELPI) ELPI has a time resolution of 1 s and it can also be used in transient tests. In steady-state tests it also made one scan per second within 10 minutes of each test. The mean value and stability of the consecutive scans (standard deviations) can be seen in Figure 5-8. ELPI Number Size Particle Distribution - 10 ppm Sulphur Fuel 1270 rpm, 50% load - Particulates Dilution DR = E+02 Mean Value 1.00E E E-01 Dp [nm] Figure 5-8 Typical deviation of the ELPI consecutive scans from the mean value in a steady-state test. The stability of the ELPI scans in steady-state tests was good, however for particle diameters above 1 µm there were zero values measured, which might have affected test results. These zero values are comprehensible because at the larger diameters, particle number concentrations are very low compared to smaller diameters. The influence of fuel quality and dilution systems can be seen in Figure 5-9. Effects of fuel quality are very similar to those observed with SMPS system. A distinct increase of nucleation mode with high sulphur fuel is discernable. This effect decreased with increasing engine load and reversed at high loads. On the contrary to SMPS, ELPI measured nucleation mode even at higher engine loads. Accumulation mode was also increased with high sulphur Draft Version 15/50

16 fuel, but only at idling. At higher engine loads high sulphur fuel reduced both nucleation and accumulation mode. Particle size distributions were slightly higher with DEKATI dilution than with Particulates dilution, especially at higher loads. It is not clear whether this was result of dilution unit itself or of changed ambient and dilution conditions (measured DR with DEKATI primary diluter increased with engine load). ELPI Particle Size Distribution Idle (500 rpm, 0% load) 1.00E E+01 DEKATI DR = 14.6; 10 ppm S Fuel Particulates DR = 14.1; 10 ppm S Fuel Particulates DR = 15.2; 300 ppm S Fuel 1.00E Dp [nm] Figure 5-9 Effects of dilution system and fuel quality on ELPI particle size distribution ELPI Particle Size Distribution 1.00E E E+01 Idle (500 rpm, 0% load 1060 rpm, 10% load 1060 rpm, 25% load 1270 rpm, 50% load 1.00E Dp [nm] Figure 5-10 Effect of engine load on ELPI particle size distribution Draft Version 16/50

17 Double Differential Mobility Particle Spectrometer (DDMPS) DDMPS is a closed system with its own dilution unit. Each steady-state point was measured with two different dilution ratios (11 and 26) and for each DR three consecutive scans were made. An example of three consecutive scans in one steady-state point (idling) is presented in Figure DDMPS Number Size Particle Distribution - 10 ppm sulphur fuel idle (500 rpm, 0% load) - Dilution Ratio 1: E mean value Dp [nm] Figure 5-11 Stability of three consecutive DDMPS scans in steady-state tests 1.00E+09 DDMPS Particle Number Size Distribution - 10 ppm sulphur fuel 1060 rpm, 25% load - Dilution Ratio 1:26 day 1 day Dp [nm] Figure 5-12 Repeatability of the results on different days. Draft Version 17/50

18 Dilution ratio and engine load had little effect on stability of the scans. Only in some tests the scans were more scattered for unknown reasons. Repeatability form different days was found to be good (Figure 5-12). DDMPS Number Size Particle Distribution - Idle (500 rpm, 0% load) 1.00E+09 1:11, 300 ppm S fuel 1:26, 300 ppm S fuel 1:11, 10 ppm S fuel 1:26, 10 ppm S fuel Dp [nm] Figure 5-13 Effects of fuel quality and dilution ratio on DDMPS particle size distribution Effect of fuel quality was very similar to those of SMPS and ELPI. Dilution ratio had some effect, but no distinct trends were discernable (Figure 5-13). DDMPS Particle Size Distribution 1.00E rpm, 0% load mean 1060 rpm, 10% load mean 1060 rpm, 25% load mean 1270 rpm, 50% load mean Dp [nm] Figure 5-14 Effect of engine load on DDMPS particle size distribution. The Effects of engine load on accumulation mode are similar to other systems. Accumulation mode is not only increased but also shifted to smaller diameters with increasing engine load. Draft Version 18/50

19 The nucleation mode is also shifted to smaller diameters with increasing engine load(figure 5-14) Comparison of size distributions with different sampling systems Figure 5-15 through Figure 5-18 show the measured size distributions of the different systems for different steady state points. Only the tests with the low sulphur fuel are considered. DDMPS distributions are mean values of two tests series and two respective dilution ratios (total of 12 scans). SMPS and ELPI values are results from different tests with different dilution systems. Idle (500 rpm, 0% load) - 10 ppm Sulphur Fuel 1.00E E E E+00 DDMPS ELPI (DEKATI DR = 14.6) ELPI (Particulates DR = 14.1) SMPS (Smart Sampler DR = 12.5) SMPS (Particulates DR = 14.1) Dp [nm] Figure 5-15 Comparison of number size distributions of different systems at 500 rpm and 0% load (idle) 1060 rpm, 10% load - 10 ppm Sulphur Fuel 1.00E E E E+00 DDMPS ELPI (DEKATI DR = 15.9) ELPI (Particulates DR = 14.3) SMPS (Smart Sampler DR = 12.5) SMPS (Particulates DR = 14.3) Dp [nm] Figure 5-16 Comparison of number size distributions of different systems at 1060 rpm and 10% load Draft Version 19/50

20 1060 rpm, 25% load - 10 ppm Sulphur Fuel 1.00E E E E+00 DDMPS ELPI (DEKATI DR = 16.7) ELPI (Particulates DR = 12.5) SMPS (Smart Sampler DR = 12.5) SMPS (Particulates DR = 12.5) Dp [nm] Figure 5-17 Comparison of number size distributions of different systems at 1060 rpm and 25% load 1270 rpm, 50% load - 10 ppm Sulphur Fuel 1.00E E E E+00 DDMPS ELPI (DEKATI DR = 18.3) ELPI (Particulates DR = 11.9) SMPS (Smart Sampler DR = 12.5) SMPS (Particulates DR = 11.9) Dp [nm] Figure 5-18 Comparison of number size distributions of different systems at 1270 rpm and 50% load It is important to note that the test with SMPS and Smart Sampler at 1270 rpm and 50% load (Figure 5-18) did not function properly (sample flow rate was not constant), which caused deviation in SMPS result at this steady-state point. Excluding this error, it can be observed that the steady-state size distributions of DDMPS, SMPS and ELPI are in close agreement in the accumulation mode range. Draft Version 20/50

21 The nucleation mode distributions were much more affected by the sampling systems and the differences increased with increasing engine load. SMPS measured no nucleation mode at higher loads. Figure 5-19 through Figure 5-22 illustrate the mean values of all tests with the 10 ppm sulphur fuel compared to the test results with the 300 ppm sulphur fuel. Idle (500 rpm, 0% load) 1.00E E E+01 DDMPS, 10 ppm Sulphur Fuel DDMPS, 300 ppm Sulphur Fuel ELPI 10 ppm Sulphur Fuel ELPI 300 ppm Sulphur Fuel SMPS 10 ppm Sulphur Fuel SMPS 300 ppm Sulphur Fuel 1.00E E E E+02 Dp [nm] Figure 5-19 Effects of fuel quality on size distributions at 500 rpm and 0% load (idle) 1.00E rpm, 10% load 1.00E E+01 DDMPS, 10 ppm Sulphur Fuel DDMPS, 300 ppm Sulphur Fuel ELPI 10 ppm Sulphur Fuel ELPI 300 ppm Sulphur Fuel SMPS 10 ppm Sulphur Fuel SMPS 300 ppm Sulphur Fuel 1.00E E E E+02 Dp [nm] Figure 5-20 Effects of fuel quality on size distributions at 1060 rpm and 10% load Draft Version 21/50

22 1060 rpm, 25% load 1.00E E E+01 DDMPS, 10 ppm Sulphur Fuel DDMPS, 300 ppm Sulphur Fuel ELPI 10 ppm Sulphur Fuel ELPI 300 ppm Sulphur Fuel SMPS 10 ppm Sulphur Fuel SMPS 300 ppm Sulphur Fuel 1.00E E E E+02 Dp [nm] Figure 5-21 Effects of fuel quality on size distributions at 1060 rpm and 25% load 1270 rpm, 50% load 1.00E E E+01 DDMPS, 10 ppm Sulphur Fuel DDMPS, 300 ppm Sulphur Fuel ELPI 10 ppm Sulphur Fuel ELPI 300 ppm Sulphur Fuel SMPS 10 ppm Sulphur Fuel SMPS 300 ppm Sulphur Fuel 1.00E E E E+02 Dp [nm] Figure 5-22 Effects of fuel quality on size distributions at 1270 rpm and 50% load Draft Version 22/50

23 Summary As expected, all dilution systems were able to keep dilution ratio stable in steady-state tests, but the measured dilution ratios DR(NOx) were different for every steady-state point. The number size distributions were calculated by multiplying raw data with the measured dilution ratios. Dilution systems had little effect on SMPS and ELPI size distributions. SMPS measured slightly higher concentrations with Smart Sampler than with Particulates primary diluter, especially at higher engine loads. At idling nucleation mode was increased with Smart Sampler. ELPI measured slightly higher concentrations with DEKATI than with Particulates primary diluter, but only for larger particle diameters. However, these effects might be result of changed conditions such as ambient temperature, humidity and other parameters between different days and not the dilution unit itself. Although SMPS, DDMPS and ELPI measure different physical properties, respectively the electrical mobility diameter and the aerodynamic diameter, the steady-state particle size distributions are in close agreement. Deviations generally increased with increasing engine load. With SMPS no nucleation mode was measured at higher loads. Effects of sulphur content in fuel were very similar with all systems. Low sulphur fuel reduced particle concentrations at idling and low load. Especially the nucleation mode was reduced. At higher loads this effect reversed, i.e. at the 50% load concentrations were even higher with low sulphur fuel. Draft Version 23/50

24 5.2.2 Transient tests Only two of the used particle sampling systems are capable of measuring size distributions in transient tests: ELPI and TrDMPS. As already discussed in section 5.1, strong fluctuations of DR occurred in all transient tests, however, a considerable share of these fluctuations may be due to the measurement related factors (eg. remaining time shift, inaccuracy of the NOx sensors at the lower NOx concentrations). On the other hand there must be real reasons for the instability of dilution ratio such as dilution air flow fluctuations associated with pressure fluctuations or with a limited accuracy of the flow controller. The fact that real dilution ratio is unknown causes difficulties in transient tests, since the raw data need to be multiplied with primary dilution ratio to obtain end results. Various smoothened dilution ratios were calculated in order to dampen unrealistic fluctuations of DR(NOx). In Figure 5-23 the measured and the smoothened DR are illustrated in first 300 sec of a city cycle with Particulates dilution system. City Cycle - Particulates Dilution DR(NOx) DR(NOx_m) n v v [km/h] F [x10 N] n [rpm] DR, v n, F F DR Time [sec] Figure 5-23 Calculated dilution ratios in a city cycle with Particulates primary diluter ELPI raw data were multiplied with constant dilution ratio and with different calculated DR in order to investigate effects of their fluctuations on transient particle number counts. Figure 5-24 shows ELPI particle number counts for diameter of 78 nm obtained by multiplying with the constant (mean value for the whole test) and the measured dilution ratio. The measured one represents the non-smoothened DR(NOx), which has the strongest fluctuations. Additionally DDMPS values are given for comparison. Particle diameter of 78 nm is close to typical accumulation mode peak, which is known to be very stable and reproducible. Draft Version 24/50

25 City Cycle - TrDMPS and ELPI Particle Number Counts (78 nm) Particulates Dilution 1.00E+09 ELPI TrDMPS n n, F 1.00E E E+00-5 ELPI 77.3 nm (constant DR) F ELPI 77.3 nm (DR(NOx)) -10 TrDMPS 78.5 nm -15 F [kn] n [x100 rpm] Time [sec] Figure 5-24 TrDMPS and ELPI particle number counts for diameter of 78 nm (accumulation mode) It can be seen that deviations even between the two extreme ELPI results (constant DR and DR(NOx)) are still not very significant. Considering this and the fact that the considerable share of dilution ratio fluctuations is probably result of the measurement related factors, it can be concluded that calculating with constant DR (mean value) results in a satisfactory level of accuracy. On the other hand it can be seen that there is a significant disagreement between TrDMPS and ELPI results (ELPI number count is higher). This can be observed in different degrees for all diameters and in all transient tests. It is not clear whether this has measurement related reasons or it is due to an incorrect secondary dilution ratio. Both systems show very similar trends and a clear correlation with engine load (chassis force) for this particle diameter. Figure 5-25 through Figure 5-27 show the real world cycles used in the transient pre-tests. Draft Version 25/50

26 60 City Cycle Vehicle Speed, km/h Time, s Figure 5-25 City Cycle Rural Cycle Vehicle Speed, km/h Time, s Figure 5-26 Rural Cycle Highway Cycle 90 Vehicle Speed, km/h Time, s Figure 5-27 Highway cycle Draft Version 26/50

27 In Figure 5-28 and Figure 5-29 ELPI and TrDMPS particle number size distributions for the first 180 sec of a city cycle are shown. City cycle - ELPI Size Distributions E E E E E E E E E E+00 dn/dlndp, cm Dp, nm 0 60 Time, s Figure 5-28 ELPI particle number size distributions during first 180 s of a city cycle City Cycle - TrDMPS Size Distributions E E E E+00 dn/dlndp, cm Dp, nm Time, s Figure 5-29 TrDMPS particle number size distributions during first 180 s of a city cycle Draft Version 27/50

28 It can be seen that ELPI size distributions show clear and stable bimodal pattern independent of changes of engine load and speed in a transient test. This can be observed in all test cycles. Nucleation mode is always present and usually higher than accumulation mode. For particles >1 µm, which are beyond the TrDMPS size range and can be measured only by ELPI, increasing concentrations can also be seen. These microparticle peaks are always ahead of the nucleation and accumulation peaks. They are also much shorter and are likely to be result of abrupt increase of engine load (acceleration). TrDMPS nucleation mode is much less stable. It decreases and shifts to smaller diameters with increasing engine load. At very high loads it disappears completely. TrDMPS and ELPI size distributions in transient tests, their comparison and correlation with engine speed and load is discussed in the next section Comparison of TrDMPS and ELPI results In order to compare the two sampling systems, two particle diameters are selected 15 and 131 nm, which approximately represent nucleation and accumulation mode particles. Figure 5-30 and Figure 5-31 show the comparison of TrDMPS and ELPI number concentrations for the two diameters during the first 180 s of a city cycle. ELPI 2 µm concentration is shown in Figure 5-32, representing larger microparticles (coarse mode). City Cycle - TrDMPS and ELPI (Particulates Dilution) Particle Number Concentrations 1.00E ELPI TrDMPS n n, F 1.00E+02 F E E ELPI 15 nm TrDMPS 15 nm F [kn] n [x100 rpm] Time [sec] Figure 5-30 TrDMPS and ELPI 15 nm particle concentrations (representing nucleation mode) in a city cycle Draft Version 28/50

29 City Cycle - TrDMPS and ELPI (Particulates Dilution) Particle Number Concentrations 1.00E+09 ELPI TrDMPS n F n, F 1.00E E E ELPI 131 nm TrDMPS 131 nm F [kn] n [x100 rpm] Time [sec] Figure 5-31 TrDMPS and ELPI 131 nm particle concentrations (representing accumulation mode) in a city cycle City Cycle - ELPI 2000 nm Particle Number Concentration - Particulates Dilution 4.50E E E nm F [kn] n [x100 rpm] E E E E E E Time [sec] n, F Figure 5-32 ELPI 2µm particle concentrations (representing coarse mode) in a city cycle It can be seen that, apart from differences in absolute numbers, ELPI and TrDMPS have similar trends. Especially the 131 nm particle concentrations (accumulation mode) are in good agreement tendency-wise and correlate with chassis dynamometer force (engine load). Draft Version 29/50

30 The nucleation mode (15 nm) particle concentrations are not in such an agreement. ELPI 15 nm concentration is always higher than 131 nm concentration and also correlates with engine load. ELPI nucleation mode is stable and its peak is also always higher than accumulation mode peak. The situation is similar to the one in steady-state tests where ELPI also always measured nucleation mode independent of engine load. TrDMPS nucleation mode particle concentration also increases with engine load to a certain degree, but then, at higher loads it drops to very low levels, probably due to higher exhaust temperatures. ELPI 2 µm particle concentration is somewhat unpredictable, but the peaks always appear during abrupt increases of engine load and speed. These coarse mode microparticle peaks are very short and might be result of combustion dynamics or of re-entrained particles, previously deposited in the exhaust system. In Figure 5-33 average number concentrations and mean particle diameters of both ELPI and TrDMPS in a city cycle are compared. Since the two systems measure in different diameter ranges, only trends can be compared - not the absolute numbers. City Cycle - TrDMPS and ELPI (Particulates Dilution) Average Particle Number Concentrations and Mean Particle Diameters, Dpm [nm] 1.00E E E E+00 ELPI TrDMPS n F ELPI TrDMPS Time [sec] n, F ELPI average particle number concentration TrDMPS average particle number concentration TrDMPS mean particle diameter ELPI mean particle diameter F [kn] n [x100 rpm] Figure 5-33 Comparison of ELPI and TrDMPS average particle number concentrations and mean particle diameters in a city cycle It can be seen that ELPI and TrDMPS average concentrations have similar trends and correlate with chassis dynamometer force. The fluctuations of TrDMPS mean particle diameter are associated with its instable nucleation mode, i.e. at high engine loads TrDMPS nucleation mode is reduced and mean particle diameter is shifted to higher diameters. The ELPI mean particle diameter is generally more stable, due to its stable nucleation mode, but short, high peaks up to 700 nm can be observed. These peaks are not very reproducible, but they are related to abrupt increases of engine load and speed, when coarse mode concentrations rapidly increase. Draft Version 30/50

31 5.3. Morphology In the framework of the HDV pre-tests measurements an analysis of particulate morphology was carried out. Particle samples were collected and characterized by a high-resolution Transmission Electron Microscope (TEM). Additionally, a chemical analysis using Energy- Dispersive X-Ray spectroscopy (EDX) was performed to obtain some information on particle elemental composition. EDX is a microanalytical technique that uses the characteristic spectrum of x-rays emitted by the specimen after excitation by high-energy electrons to obtain information about its elemental composition. Elements of low atomic number are difficult to detect by EDX (hydrogen for example can not be detected). EDX does not provide accurate quantitative chemical analysis Sampling procedure The particle samples were collected at several steady-state points using a Differential Mobility Analyser (DMA), operating at a fixed voltage and thus selecting particles of a single size class within a very narrow range. The size classes were selected according to the size distribution at the related steady-state point. Usually the size class with the maximum concentration in the size distribution was selected. In some cases, one or two additional size classes were selected and the test was repeated at the same steady-state point in order to cover a larger particle size range and to investigate the differences in the morphology. The collection was performed at a constant dilution ratio of 5:1 with the low sulphur fuel (10 ppm S). The particles were collected directly onto copper grids, coated with an amorphous carbon film. Draft Version 31/50

32 5.3.2 Results In the following sections the respective samples are described together with the related TEM images and EDX plots. The different TEM images within the same sample represent different areas with typical and atypical structures found in the sample. The related particle number size distributions are also shown. During A-test series the particle size distributions were measured with the ELPI. The particle size distributions during B-test series were not measured, but the DDMPS size distributions from the other tests at the same engine loads are exemplary presented. Sample A1 Selected size class: 15 nm Engine speed and load: 1270 rpm, 50% load Sampling time: 15 min Volume flow rate: 2 l/min ELPI Number Size Particle Distribution rpm, 50% load DEKATI Dilution (DR = 9.7) - 10 ppm Sulphur Fuel 1.00E nm 1.00E E E Dp [nm] Figure 5-34 ELPI size distribution at 1270 rpm and 50% load during the test for the Morphology - Sample A1 In Figure 5-35 through Figure 5-38 some of the TEM images with various magnifications are exemplary presented. Draft Version 32/50

33 Figure 5-35 Figure 5-36 Figure 5-37 Figure 5-38 Although the diluted exhaust aerosol is classified into a narrow particle size fraction (15 nm), the film was covered mostly with large clusters of agglomerated spherical primary particles. Both the spherical primary particles (spherules) and the agglomerated clusters and chains are typical diesel soot structures. In this sample no atypical particles could be found. As already mentioned, all structures are much larger than the selected size class. Draft Version 33/50

34 FELMI - TU Graz Cursor: 0.000keV = 0 ROI THU 13-DEC-01 08:30 C Cu O Cu Si Cu CX4133 A7311 PROBE A1 UEBERBLICK Figure 5-39 EDX plot related to the sample A1 (overview) From the EDX plot (see Figure 5-39) it can be seen that the particles in this sample consist mainly of elemental carbon. It should be noted that the Cu peaks in the EDX plots stem from the copper grid, while the Si signal is from the C-film and is an artifact of the manufacturing process. Also, the carbon film contribute to the carbon peak therefore no quantitative conclusion can be made about these elements. The EDX spectrum shown in Figure 5-39 with the signal peaks from carbon, oxygen, silicon and copper is typical for the whole sample and, as in the next sections will be seen, for all particles with the typical morphology (i.e., agglomerated clusters of primary soot spherules). Draft Version 34/50

35 Sample A3 Selected size class: 20 nm Engine speed and load: 1060 rpm, 25% load Sampling time: 20 min (volume flow 2 l/min) ELPI Number Size Particle Distribution rpm, 25% load DEKATI Dilution (DR = 17.9) - 10 ppm Sulphur Fuel 1.00E nm 1.00E E E Dp [nm] Figure 5-40 ELPI size distribution at 1060 rpm and 25% load during the test for the Morphology - Sample A3 Figure 5-41 Figure 5-42 Draft Version 35/50

36 Figure 5-43 Figure 5-44 This sample was similar to the A1 sample, but it was covered more with particles. There were also smaller clusters and even single non-agglomerated spherules. No atypical particles could be found. The elemental composition was similar to the composition in the sample A1. Draft Version 36/50

37 Sample A5 Selected size class: 40 nm Engine speed and load: 1060 rpm, 25% load Sampling time: 20 min (volume flow 2 l/min) ELPI Number Size Particle Distribution rpm, 25% load DEKATI Dilution (DR = 12.9) - 10 ppm Sulphur Fuel 1.00E nm 1.00E E E Dp [nm] Figure 5-45 ELPI size distribution at 1060 rpm and 25% load during the test for the Morphology - Sample A5 Figure 5-46 Figure 5-47 (see Figure 5-50) Draft Version 37/50

38 Figure 5-48 (see Figure 5-51) Figure 5-49 This sample was more covered than the A3 and atypical particles were found (see Figure 5-46 through Figure 5-49). These atypical particles could be re-entrained material, previously deposited in the exhaust system. Also small spherical particles (10 20 nm) that look different than the usual soot spherules (more like droplets) could be found (see Figure 5-49). The related EDX plots can be seen in Figure 5-50 and Figure FELMI - TU Graz Cursor: 0.000keV = 0 ROI THU 13-DEC-01 08:34 Cu C O Si Cu Cu CX4138 A7311 PROBE A5 3BF Figure 5-50 EDX plot related to the particle in Figure 5-47 Draft Version 38/50

39 FELMI - TU Graz Cursor: 0.000keV = 0 ROI THU 13-DEC-01 08:35 Fe Cu C Cu Fe Fe O Si S Cl Mn Cu CX4139 A7311 PROBE A5 4BF Figure 5-51 EDX plot related to the particle in Figure 5-48 The particles in Figure 5-46 and Figure 5-47 have similar elemental composition like the typical agglomerated soot spherules (see Figure 5-50). The particle in Figure 5-48, on the other hand, has somewhat different composition (see Figure 5-51). In addition to typical C, O, Si and Cu signals, there are peaks from Fe, S, Cl and Mn. Draft Version 39/50

40 Sample A7 Selected size class: 80 nm Engine speed and load: 1060 rpm, 25% load Sampling time: 20 min (volume flow 2 l/min) ELPI Number Size Particle Distribution rpm, 25% load DEKATI Dilution (DR = 16.2) - 10 ppm Sulphur Fuel 1.00E nm 1.00E E E Dp [nm] Figure 5-52 ELPI size distribution at 1060 rpm and 25% load during the test for the Morphology - Sample A7 Figure 5-53 Figure 5-54 Draft Version 40/50

41 Figure 5-55 Figure 5-56 This C-film was covered very well (more than the A5). Some atypical particles occurred (see Figure 5-54 and Figure 5-55). The elemental composition of these particles was inconspicuous, i.e. it was similar to those of the typical agglomerates. Draft Version 41/50

42 Sample B2 Selected size class: 10 nm Engine speed and load: 500 rpm, 0% load (idle) Sampling time: 20 min (volume flow 2 l/min) DDMPS particle size distribution - Idle (500 rpm, 0% load) 1.00E nm 1.00E E E Dp [nm] Figure 5-57 Average DDMPS size distribution at 500 rpm, 0% load (mean value from 2 tests 12 scans) Figure 5-58 (see Fig. 5-60) Figure 5-59 Draft Version 42/50

43 The carbon film was damaged in this test. Neither typical soot spherules nor agglomerates could be found. Some large, sulphur containing particles occurred (see Figure 5-58). The elemental composition of this atypical particle is shown in Figure It can be seen that it contains sulphur and traces of chlorine and iron. FELMI - TU Graz Cursor: 0.000keV = 0 ROI THU 13-DEC-01 08:43 Cu Cu C Cu O Si S Cl Fe CX4140 A7311 PROBE B2 1BF Figure 5-60 EDX plot related to the particle in Fig Draft Version 43/50

44 Sample B4 Selected size class: 10 nm Engine speed and load: 1060 rpm, 25% load (idle) Sampling time: 20 min (volume flow 2 l/min) 1.00E+09 DDMPS particle size distribution rpm, 10% load 10 nm 1.00E E E Dp [nm] Figure 5-61 Average DDMPS size distribution at 1060 rpm, 10% load (mean value from 2 tests 12 scans) Figure 5-62 (see Figure 5-64) Figure 5-63 Draft Version 44/50

45 FELMI - TU Graz Cursor: 0.000keV = 0 ROI THU 13-DEC-01 08:47 C Cu O Cu Si Cu CX4164 PROBE B4 2BF Figure 5-64 EDX plot related to the particle in Fig The C-film was almost empty few atypical particles (Figure 5-62) and only one agglomerate (Figure 5-63) could be found. The elemental composition of the particle in Figure 5-62 was inconspicuous (see Figure 5-64). Draft Version 45/50

46 Sample B6 Selected size class: 50 nm Engine speed and load: 1060 rpm, 25% load (idle) Sampling time: 20 min (volume flow 2 l/min) 1.00E+09 DDMPS particle size distribution rpm, 10% load 50 nm 1.00E E E Dp [nm] Figure 5-65 Average DDMPS size distribution at 1060 rpm, 10% load (mean value from 2 tests 12 scans) Figure 5-66 (see Figure 5-68) Figure 5-67 (see Figure 5-69) Draft Version 46/50

47 FELMI - TU Graz Cursor: 0.000keV = 0 ROI THU 13-DEC-01 08:49 Cu C O Cu Na Si S Cl K Ca Fe Cu CX4165 PROBE B6 3BF Figure 5-68 EDX plot related to the particle in Fig This was a very unusual sample. The C-film was covered very well and also some atypical, sulphur containing particles were found (see Figure 5-66), but the rest of the sample looked different as well. The elemental composition is also different with signals from Na, Cl, K, Ca and Fe (Figure 5-68 and Figure 5-69). It is not clear what caused these dissimilarities. FELMI - TU Graz Cursor: 0.000keV = 0 ROI THU 13-DEC-01 08:51 Cu C O Cu Na Cl K Ca Fe Cu CX4167 PROBE B6 4BF Figure 5-69 EDX plot related to the particle in Fig Draft Version 47/50

48 5.3.3 Summary Appart from atypical particles, which occur only sporadically, exhaust particles morphologically characterized in this study apper to be agglomerated structures (clusters, chains) formed of spherical primary particles (spherules) with a diameter estimated between 10 and 40 nanometers. The size of the agglomerates themselves is difficult to estimate since, tightly packed on the C-films, they may have further agglomerated to very large structures, but they appear to reach sizes of up to a few micrometers. Classifying diluted exhaust aerosol into different size fractions (monodisperse fractions) did not appear to make significant difference in the TEM images. It seems as if this classification did not work properly, i.e. even when very small sizes were selected (10, 15 nm), which should have excluded almost all solid soot particles, spherules of all sizes (up to 40 nm) could be observed. The agglomerates seen in the TEM images are also much larger than the selected size, but they could have agglomerated after the classification and especially on the C-film. Also the lump-looking atypical particles are much larger than the selected particle size. The only effect that could be observed with the changed selected size is that with increasing size the C-films were more covered. One possible explanation is that the most particles are indeed classified properly, but occasionally some particles outside the selected size come through and are collected on the C-film. This would mean that a significant share of the particles seen in the TEM images are such non-classified particles. Engine load and speed had no discernable effect on the particle morphology in the TEM images. The EDX plots suggest that the typical agglomerates contain mainly elemental carbon (soot). The atypical particles, appart from having the atypical morphology, i.e. looking more like a solid mass of material, usually have somewhat different elemental composition. They may contain sulphur and some other elements (Na, Cl, K, Ca, Fe, Mn). It should be noted again that EDX cannot give any information about elements of low atomic number such as hydrogen and is also limited in providing accurate information about quantitative elemental composition. The fact that the atypical particles usually have different elemental composition suggests that they may have a different origin than the typical agglomerates. They could result from the re-entrained material or from the incomplete combustion of lubrication oil. The atypical particles range in size from around 100 nm up to a few micrometer. Draft Version 48/50