1. Imad Khalek, Ph.D., Sr. Program Manager, ikhalek@swri.org
Relationship Among Various Particle Characterization Metrics Using GDI Engine Based Light-Duty Vehicles
Presented by,
Imad Khalek, Ph.D., Sr. Program Manager,
Paper by: Vinay Premnath, Imad Khalek & Peter Morgan (SAE )
PMP 47th Meeting, May 16-17, 2018

2. Acknowledgements
Data analyzed in this work was generated under project CRC 94-2, funded by the Coordinating Research Council (CRC)

3. Background
PM2.5 mass is the only particle metric used in the USA for regulatory emissions limits
Other metrics can be used but require epidemiological studies to establish cause and effect, which could take a very long time
Other countries such as EU, China, India, etc… adapted solid particle number to complement the PM mass regulations
Various metrics such as particle number, size, surface area and mass are important to understand the impact of vehicle emissions on human health
It will be of interest to emissions stakeholders to understand the relationship among the various metrics using a trusted database including various technology engines

4. Objectives
Examine the relationship among various particle emissions metrics analyzed for 12 different light-duty vehicles equipped with GDI engines, running on a chassis dynamometer
Specifically analyze data for:
Three different mass based measurement techniques
Two different number based measurement techniques
Mass and number measurements
Derived mass (from measured size distribution) and measured mass

5. Vehicles and Fuels Used
12 GDI vehicles, operated on 8 different fuels with different PM indices, ethanol content, and octanes numbers (see CRC-94-2 Final report)
Vehicle
Model Year
Displacement
Engine Configuration
Chevrolet Equinox
2011
2.4L
Naturally Aspirated, I4
Chevrolet Malibu
2013
2.0L
Turbocharged, I4
Chevrolet Malibu*
2.5L
Ford F150XL
3.5L
Turbocharged, V6
Honda Accord
Hyundai Santa Fe
Lexus NX200t*
2015
Mazda Mazda6
2014
Mercedes-Benz GLK350*
Naturally Aspirated, V6
Nissan Juke*
2010
1.6L
VW Jetta GLI
2012

6. Test Protocol Eight different fuels were involved in the test campaign
Focus of this presentation is on correlation among various real-time instrumentation that sampled in parallel
Results will be discussed with disregard to vehicle type or fuel properties

7. Experimental Setup LA-92 Drive Cycle LA-92 - Unified Driving Cycle
Cold-start three-phase test
Phase 1 and 3 – 300 sec, Phase 2 – 1135 sec

8. Particle Instrumentation

9. Results
Correlations between different instrumentation will be discussed with disregard to vehicle type or fuel properties
Data was analyzed on a phase-wise basis for all vehicles and fuels tested
Following results will be presented
Correlations between mass based measurements
MSS Soot vs PM vs TC (from OC/EC measurements)
Correlations between number based measurements
EU SPNS vs EEPS-SPSS
Correlations between mass and number based measurements
EU SPNS/EEPS-SPSS vs MSS soot/PM/TC
Correlations between EEPS-SPSS derived mass and mass based measurements

10. Mass-Based Correlation
MSS & TC vs. PM (filter-based)
Slope between TC and PM shows that the two measurement techniques agree to within 1% of each other
Slope of correlation between MSS soot and PM indicate 80 to 90% of PM is soot
MSS soot vs EC slope suggests
Not much difference in soot characteristics among different GDI engines involved
Soot used to calibrate MSS has similar response to soot produced by GDI engines
MSS soot mass vs EC mass

11. Number vs. Mass
Conversion from mass metric to number and vice-versa provides a bridge in estimating two important metrics
Reasonably strong R2 was observed between SPNS PN and different mass based measurements
Slopes influenced by vehicles with high mass with large GNMD (lesser PN)
Spread shows vehicles can have similar number emissions but drastically different mass emissions (and vice-versa)
The slope for number vs soot increased as the PM mass decreased, as shown on this slide and next slide
SPNS (>23 nm) vs TC, MSS soot, PM
SPNS (>23 nm) vs TC, MSS soot, PM for mass < 20 mg/mile

12. Number vs. Mass Down to 1 mg/mi

13. Derived Mass vs. Measured Mass
Strong correlation observed between EEPS derived mass and mass based measurements R2 ~ 0.98
EEPS derived mass under estimated soot mass emissions by ~ 35%
Potential reasons
EEPS ‘default’ inversion matrix assumed spherical particles and unit density
Assumed unit density while converting number distribution to mass distribution
Default Matrix

14. Number-Based Correlation
Good correlation (R2) observed between the two measurement techniques
One-to-one agreement wasn’t observed. Potential reasons include
Fundamental measurement principle
‘Default’ inversion matrix was used
Particle penetration correction factor
Default Matrix
SPNS (>23 nm, TSI CPC 3790) vs EEPS

15. Sub-23 nm as a Percent of Total Emissions
On a nominal basis, fraction of sub 23 nm particles for all GDI vehicles is ~21%
Fraction can reach as high as 70% or as low as 2%
For GDI engine technology, these results indicate that sub 23 nm PN emissions could be significant
This would be true if:
Particle number regulation was extended down to 10 nm
The emission limit kept the same
GPF is not used

16. Summary
Good correlation observed between the three mass based measurements
Typically, soot mass was 10% to 20% lower than PM mass
TC and PM mass showed nearly perfect agreement
One measurement technique can be used to predict values of other two with a SEE < 15%
Good correlation observed between the two number based methods
Particle sizer can be used to predict EU compliant PN
Reasonably good correlation observed between number and mass based measurements below 20 mg/mi
The slope of the relationship is on the order of 2 Trillion particles per 1 mg or 1.5 trillion particles per 1 mg of total PM
EEPS data need further investigation to select the best matrix appropriate for a particular data set
Building a database on the relationship among different metrics can provide a huge benefit to emissions stakeholders assessing the impact of various technology engines on particle emissions characteristics or inferring one metric from the other.