ETH 2002 1 09/19/02 Stefan Carli, Volkswagen AG, D-38436 Wolfsburg Ulrich Matter; Matter Engineering AG, CH-5610 Wohlen Sampling methodology influences on modern Diesel particle number size distribution measurements Abstract: The set up and the conditions under which gravimetric particulate measurements of diesel vehicles are conducted are regulated. This is not the case for particle number size measurements namely SMPS and ELPI measurements. It will be shown, how the sampling locations (tailpipe CVS), parameters such as dilution ratio, material (connecting / sealing material), temperatures, residence time and the measurement history of connecting tubes influence the measured quantities. In case of the nanoparticle problematic it was investigated to use an one step high dilution instead of a thermodesorber to avoid nucleation particle artifacts. Engine load and fuel sulphur influences were also investigated in respect to the various parameters mentioned above. Introduction The set up and the conditions under which gravimetric particulate measurements of diesel vehicles are conducted is regulated and properly defined to make comparable evaluations possible. This is not the case for particle number size measurements. In a project conducted together with Matter Engineering AG the various influences on number size distribution measurements by set up, sampling conditions, engine load and fuel quality and the measurement instruments themselves were investigated. Many results were already presented at the last years Nanoparticle Conference (2001): Sample probe design has no significant influence on nanoparticle NSD measurements Tubing with electrostatic loading ability lose significant numbers of nanoparticles Measuring NSD at higher load the formation of nucleation particles occurs, especially with high sulphur fuel Already little variation of CVS-DR has a strong effect on particle number and size (no reproducible measurements possible) - Provoke nucleation: Low DR, cold humid dil-air, high sulphur fuel - Avoid nucleation: High DR > 50, hot dil-air, sulphur free fuel The combination oxidation catalyst, high sulphur fuel content and high engine load seems to be the main cause for nucleation particles No significant nucleation particle effect observed from transfer line (Diesel) Relationship between aerodynamic and mobility diameter depends on particle morphology (density), morphology changes dependent on load and fuel quality High discrepancy in particle number and size comparing SMPS s instruments only good for relative measurements as long as no reliable standard exists The 2002 presentation is about an additional investigation to evaluate connector and sealing material particle effects from the CVS transfer line our experience with alternative exhaust dilution systems for particle measurements. At Volkswagen we made bad experiences with silicone material connectors between the end of the vehicle tailpipe and the CVS tranferline. As the silicon material gets hot we observe a huge increase in the particle number.
ETH 2002 2 09/19/02 Figure 1: Experimental set up Silicone connector exhaust Chassis dyno SMPS+temperatur Dilution air SMPS+tempertur Dilution tunnel (CVS) Samling line <52 C Composition 75 % soot + ash 20 % hydrocarbons + lube oil 5 % sulphuric acid Dr.Carli: ETH-Conf 3 We build up an adapter were material samples of silicon, teflon and rubber material from testing facility in Biel was introduced into a perforated tube and mounted this adapter between the end of the tailpipe and the transferline to CVS. Figure 2: Adapter with samples between tailpipe and transferline neu Swagelok 12mm Verschlusskappe ca 50 AD 12/ ID 10mm Auspuff Befestigungs Manschette Metallschlauch zu CVS Dichtungsmaterial in zyl. Drahtgitter verpackt Entnahmesonde Temp.- Messung vorhanden Anschluss MD19-2E-Verdünner Dr.Carli: ETH-Conf 4 The Temperature was measured before the sample and at the end of the transfer line. The measurements were conducted with two SMPS s in parallel (see figrure 1). A gasoline vehicle was used to generate hot exhaust. When we did the first test with the hot exhaust from the petrol vehicle without any sample to measure the background particle level we observed a unexpected huge
ETH 2002 3 09/19/02 number of particles. The transferline which was in normal use before was identified to be the source because after one hour burn out at 400 C this effect vanished. Figure 3 "dirty" transferline 1.E+10 1.E+09 1.E+08 250 C before sample 297 C before sample 358 C before sample 402 C before sample 180 C after sample 225 C after sample 279 C after sample 330 C after sample Figure 4 Silicon connection Teflon material connetion 1.E+03 250 C before sample 180 C before sample 308 C before sample 236 C after sample 372 Cbefore sample 303 C after sample 400 C before sample 342 C after sample 1.E+03 266 C before sample 206 C after sample 314 C before sample 247 C after sample 369 C before sample 298 C after sample 413 C before sample 343 C after sample connection material from FH Biel 1.0E+07 without sample 1.E+03 269 C before sample 208 C after sample 316 C before sample 248 C after sample 366 C before sample 292 C after sample 410 C before sample 350 C after sample 1.0E+06 1.0E+05 1.0E+04 1.0E+03 253 C before sample 182 C after sample 313 C before sample 241 C after sample 371 C before sample 297 C after sample 407 C before sample 339 C after sample Figure 4 shows the signal curves: Without sample backgroud Silicone sample significant generation of particles Teflon and additional sample no signifcant higher particle emission under this conditions
ETH 2002 4 09/19/02 Concluding statements: All connection and sealing material used in the hot sampling train should be tested for particle effects. The transferline is potentially a source for artefact particles if the exhaust temperature niveau raises. Commonly used dilution system for direct exhaust dilution is the two stage ejector diluter (first stage heated, dil. Ratio 1:100) directly mounted at the tailpipe. Principle: The compressed air creates a pressure drop at the nozzle and this way exhaust is sucked through the nozzle into the mixing chamber. Figure 5: Ejector diluter principle (direct exhaust dilution) heating ejector diluter diluted exhaust dilution ratio 1 : 10 exhaust compressed air overflow nozzle Dr.Carli: ETH-Conf 13 As can be seen from the figure 6 that there is the danger of high dilution errors by nozzle fouling. Figure 6 New unused diluter nozzle Ejector diluter nozzle after 24 h non-stop raw exhaust gas dilution front side back side
ETH 2002 5 09/19/02 Better experience we have with a partial flow dilution tunnel mounted with a short heated connection line to the tailpipe. Variable dilution ration from 1:6 to 1:50 Diluted exhaust flow of 400 i/min Figure 7: Diesel engine test bench with partial flow dilution Dr.Carli: ETH-Conf 5 We experienced that it was sufficient to raise the dilution rate to 1:50 to let the nucleation particle mode disappear even at high load and high sulphur fuel conditions. Result was a quite good correlation between mass measurement and number (SMPS) for different engine loads (50/100/120 km/h; TDI 85 kw engine) and fuel sulphur content (300, 25 ppm). Figure 8 Dependency between Mass and Number 1,400E+08 1,200E+08 Total Number [1/cm3] 1,000E+08 8,000E+07 6,000E+07 4,000E+07 2,000E+07 0,000E+00 0,00 20,00 40,00 60,00 80,00 100,00 120,00 Particulate Matter [mg/m3]
ETH 2002 6 09/19/02 Summary/Conclusions It was shown, how parameters such as material (connecting / sealing material), temperatures and the measurement history of connecting tubes influence the measured quantities. It was investigated to use an one step high dilution (partial stream dilution tunnel) instead of two stage ejector diluter to avoid unstable dilution conditions because of nozzle fouling. A one step dilution of 1:50 was sufficient to avoid nucleation resulting into good correlation between mass measurement and number even for under high load and high fuel sulphur conditions. The new particle measurement instruments can as yet and in future only be applied for the qualitative assessment of particle number and size distribution as a relative comparison because of the nature of particle number behaviour and the various possibilities of measurement influences. Acknowledgement Thank you to Dr. Ulrich Matter and Thomas Mosimann for the excellent work and the very good collaboration.