Brake Wear Particle Emission Rates
and Characterization

£%	United States

Environmental Protect
Agency


-------
Brake Wear Particle Emission Rates
and Characterization

Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency

Prepared for EPA by
ERG, Eastern Research Group, Inc.
EPA Contract No. 68HE0C18C0001
Work Assignment No. 1-04

This technical report does not necessarily represent final EPA decisions
or positions. It is intended to present technical analysis of issues using
data that are currently available. The purpose in the release of such
reports is to facilitate the exchange of technical information and to
inform the public of technical developments.

NOTICE

4>EPA

United States
Environmental Protection
Agency

EPA-420-R-22-024
September 2022


-------
Acknowledgment

ERG would like to thank the participating staff members from EPA National Fuel,
Vehicle, and Emissions Laboratory (NVFEL) and California Air Resources Board
(CARB) for their joint cooperation and contributions to this work.

ERG would like to acknowledge the work performed by Caltrans on the 2010-2012
California Household Travel Survey. The results of that survey are hosted and
presented for public use by the National Renewable Energy Laboratory (NREL), and
ERG would like to acknowledge the benefit to researchers by NREL maintaining and
hosting the data. ERG would also like to thank the project subcontractor LINK
Engineering, Inc. staff for their efforts and dedication to this project.

ERG and LINK wish to acknowledge the significant contributions of Dan Berletchick
from RAYLOC providing business intelligence and detailed vehicle and brake
information; of Gregory Vyletel from Federal Mogul for review of the vehicle selection to
validate it against common industry practices; of Jerry Forystek from Brake Parts
International for the proper use of the replacement rate index.

i


-------
Table of Contents

Acknowledgment	i

Table of Contents	ii

List of Figures	iv

List of Tables	viii

Executive Summary	x

Background	x

Objectives and Methods	x

Results	xi

Conclusions	xi

Main Report	1

Introduction	1

Materials and Methods	3

Representative Test Vehicle and Friction Material Selection	3

Track Testing and the Brake Temperature Model	14

Test Cycle Development	19

Test Matrix	37

LINK Test Laboratory Setup	42

Teflon Filter Weighing Interlaboratory Evaluation	61

Test Procedures and Quality Assurance Processes	62

Results	65

Operational Parameter Results	65

Batch Gravimetric Results	67

Vehicle-Level Mass Results	73

Results by Speed Segment	75

Emission Mass by Vehicle Weight	80

Emission Mass and Component Mass Loss During Testing	81

Particle Counts	82

Particle Size Distributions	84

Tunnel Blanks	91

Trends in Individual Brake Events	92

WTLP-Brake Tests	101

TEM Grid Loading	102

Discussion	103

ii


-------
Instrument Agreement in Mass Measurements	104

Evaluation of the Burnish Procedure	104

Evaluation of the Prius Regen Simulation	111

Reference Tests	113

Issues encountered	114

Comparison of Results to Literature	115

Potential MOVES Emissions Factors	116

Summary and Conclusions	121

Recommendations	123

Realistic Emissions Factors	123

Heavy Duty Vehicles	125

Tire Wear	126

References	129

Glossary of Terms, Abbreviations, and Symbols	131

Appendices

Appendix A. Vehicle and Friction Material Selection Supporting Data
Appendix B. Heating and Cooling Matrix for Track Testing
Appendix C. Derivation of the Generalized Coastdown Curve and Road Load
Coefficients

Appendix D. The ERG Vector Collinearity Cycle Building Approach
Appendix E. Distributions of Parameters of Interest for the Vector Method's (New

CBDC) 3 Speed Segments
Appendix F. Test Matrix and Test Dates
Appendix G. CVS Flow Setting Results
Appendix H. Tabulated Test Result Summary
Appendix I. LINK Test Result Reports
Appendix J. Teflon Filter Masses and Weight Gains

Appendix K. Vehicle-Level Particle Number Emission Rates by Speed Segment
Appendix L. EEPS Particle Size Distributions
Appendix M. Zero Blank Results

iii


-------
List of Figures

Figure 1. General trend in pad material mix by vehicle age	5

Figure 2. Overall mix of friction material quality by age	6

Figure 3. Overall Mix of Friction Materials, Quality Grade, and Vehicle Age	7

Figure 4. Brake Thermocouples and wireless hub transmitter installed on the

Camry test vehicle	15

Figure 5. Histograms of temperatures of front rotors (top) and rear rotors/drums

(bottom) over the WLTP-Brake cycle operated on the test track	16

Figure 6. The modeled right front outer rotor temperature and the corresponding

track-test measured temperature	18

Figure 7. Illustration of microtrips and braking events within a speed trace	22

Figure 8. Illustration of a braking event, extracted from the original microtrip
speed trace (blue), and inserted into the test cycle between

engineered segments (red)	26

Figure 9. The percentage of the number of microtrips within each bin of average

speed	27

Figure 10. Speed trace of the braking cycle developed for this work	28

Figure 11. The Speed traces for the WTLP-Brake and the concatenated UC/SCC .29
Figure 12. Distribution of brake event durations for the candidate cycles and the

Caltrans data	31

Figure 13. Distribution of vehicle speeds for the candidate cycles and the

Caltrans data	32

Figure 14. Distribution of braking event (negative) acceleration rates for the

candidate cycles and the Caltrans data	32

Figure 15. Distribution of modeled brake temperatures for the candidate cycles

and the Caltrans data	33

Figure 16. Distribution of brake event relative power for the candidate cycles and

the Caltrans data	34

Figure 17. Dynamometer speed and estimated Camry front rotor temperature

traces for the CBDC burnish cycle	36

Figure 18. Schematic of LINK Laboratory Setup	43

Figure 19. A brake rotor installed in the LINK test enclosure	44

Figure 20. Cumulative distribution of brake temperature over ERG's New CBDC

(Modeled) and the 10 trips of the WLTP-Brake (on track)	45

Figure 21. ProLINK screen capture of regenerative braking parameters	49

Figure 22. Plot of retarding torque (blue), friction brake torque (red), regenerative
torque (green), wheel speed (black), and regenerative power (purple)

during a regenerative-equipped braking event	50

Figure 23. Dynamometer-to-track comparison of velocity, brake pressure, and

wheel torque for example regenerative braking events	51

Figure 24. TSI 100S4 MOUDI	52

Figure 25. TSI QCM MOUDI	53

Figure 26. TSI CPC	53

Figure 27. TSI APS	54

Figure 28. TSI EEPS	54

Figure 29. Sample line schematic for this program	55

iv


-------
Figure 30. Detail schematic of Sample Line 4	56

Figure 31. Dual-stage stainless steel filter holder for PM10 sampling	57

Figure 32. Brake assembly for PM system evaluation with Arizona dust	58

Figure 33. Mass collection efficiency results of two tests of the Arizona dust

experiment as measured by the 100S4	59

Figure 34. Velocity contour of cooling air resulting from CFD simulation	59

Figure 35. Simulated and measured cumulative particle count at different

locations of PM sampling system	60

Figure 36. Graphical plot of the PM response to the parameters of the number of
sampling nozzles, airflow rate, brake rotational speed, and brake

rotational direction during the Arizona dust evaluation	61

Figure 37. LINK Laboratory vs. NVFEL measurements for the 30 filters used in

the interlaboratory evaluation	62

Figure 38. LINK software example quality review parameters and pass/fail

indication	64

Figure 39. The average rotor temperature over the CBDC, averaged by model,

axle and test weight. Error bars represent the 95% confidence interval

of the mean of tests	66

Figure 40. The peak rotor temperature during the CBDC, averaged across tests
by model, axle and test weight. Error bars show the 95% confidence

interval of the mean of tests	66

Figure 41. Average brake torque measured during the CBDC, averaged over all

tests of each model, axle, and test weight combination	67

Figure 42. Single-wheel PM Mass Emission Rates for Camry as measured by

100S4 (Blue) and 47mm Teflon filter (Green)	68

Figure 43. Single-wheel PM Mass Emission Rates for Civic as measured by
100S4 (Blue) and 47mm Teflon filter (Green). Note the Civic rear

brake is a drum system	69

Figure 44. Single-wheel PM Mass Emission Rates for F-150 as measured by

100S4 (Blue) and 47mm Teflon filter (Green)	70

Figure 45. Single-wheel PM Mass Emission Rates for Prius as measured by

100S4 (Blue) and 47mm Teflon filter (Green)	71

Figure 46. Single-wheel PM Mass Emission Rates for Rogue as measured by

100S4 (Blue) and 47mm Teflon filter (Green)	72

Figure 47. Single-wheel PM Mass Emission Rates for Sienna as measured by

100S4 (Blue) and 47mm Teflon filter (Green)	72

Figure 48. Vehicle-level PM10 mass emission rates for each vehicle, test weight,

and friction material combination	74

Figure 49. Overall trend in single-wheel mass emission rates over the three
different speed segments making up the CBDC overall test cycle

(averaged for all tests)	77

Figure 50. The PM2.5 mass fraction of PM10 for the different speed segments,

averaged across all tests	77

Figure 51. Speed correction factors for the three speed bins. These factors can
be multiplied by the overall cycle emission rate to estimate the
emissions of operation within each speed bin	78

v


-------
Figure 52. Total braking energy (proportional to kJ) within each speed segment
and the F-150 Reference Test Average Total-Cycle PM2.5 emission

mass for each speed segment	79

Figure 53. Total PM2.5 emission divided by total braking time in each speed

segment, averaged for the F-150 reference tests	79

Figure 54. Total vehicle test cycle PM mass emissions vs simulated vehicle test

weight, categorized by pad material	80

Figure 55. Total test-cycle PM10 emissions plotted against the total mass loss of
the pad and rotor during the burnish and test cycle. Each point

represents one test	81

Figure 56. Total cycle PM10 emissions relative to mass lost for pads, rotors, or
the sum of pads and rotors. R-squared values are presented for linear

fits to each	82

Figure 57. Vehicle-level particle number emission rates for each vehicle and

friction material combination	83

Figure 58. Overall average single-wheel particle number emission rate across the

three speed ranges	84

Figure 59. Size distribution of Camry front brake PM as measured by APS	85

Figure 60. Size distribution of Camry rear brake PM as measured by APS	85

Figure 61. Size distribution of Civic front brake PM as measured by APS	86

Figure 62. Size distribution of Civic rear brake PM as measured by APS	86

Figure 63. Size distribution of F-150 front brake PM as measured by APS	87

Figure 64. Size distribution of F-150 rear brake PM as measured by APS	87

Figure 65. Size distribution of Prius front brake PM as measured by APS	88

Figure 66. Size distribution of Prius rear brake PM as measured by APS	88

Figure 67. Size distribution of Rogue front brake PM as measured by APS	89

Figure 68. Size distribution of Rogue rear brake PM as measured by APS	89

Figure 69. Size distribution of Sienna front brake PM as measured by APS	90

Figure 70. Size distribution of Sienna rear brake PM as measured by APS	90

Figure 71. Size distribution of Tunnel Blanks as measured by APS	92

Figure 72. Trends in CPC count against braking event average speed,

categorized by friction material	94

Figure 73. Trends in QCM brake event PM mass against braking event average

speed, categorized by friction material	94

Figure 74. Trends in total braking event CPC count against average rotor

temperature, categorized by friction material	95

Figure 75. Trends in QCM brake event PM mass against braking event average

rotor temperature, categorized by friction material	96

Figure 76. Trends in CPC count against braking event duration, categorized by

friction material	97

Figure 77. Trends in CPC count against braking event total braking energy,

categorized by friction material	98

Figure 78. Trends in QCM braking event PM mass against braking event total

energy, categorized by friction material	99

Figure 79. Trends in CPC count against braking event average braking power,

categorized by friction material	100

vi


-------
Figure 80. Trends in QCM brake event PM mass against braking event average

power, categorized by friction material	101

Figure 81. Comparison between the CBDC average results to replicate tests of
the WLTP-Brake for the OES-NAO materials for the Camry and F-150

front axles	102

Figure 82. Partector-indicated saturation level plotted against the duration of

loading time during the CBDC	103

Figure 83. Agreement between 100S4 mass measurements (Y-axis) and 47mm

PTFE mass measurements (X-axis) for all tests and tunnel blanks	104

Figure 84. Cumulative CPC Particle Count Measured during a burnish of the

Camry rear Aftermarket LM pads	105

Figure 85. Cumulative CPC Particle Count Measured during a burnish of the

Civic front OES-NAO pads	106

Figure 86. Calculated brake effectiveness (proportional to coefficient of friction)

during a burnish of the F-150 front aftermarket metallic pads	107

Figure 87. Calculated brake effectiveness (proportional to coefficient of friction)

during a burnish of the Prius front aftermarket NAO pads	108

Figure 88. Cumulative CPC Particle Count Measured during a burnish of the

Prius rear OES-NAO pads	108

Figure 89. Particle size data during the burnish cycle of the Rogue front

aftermarket-NAO pads. The upper plot presents the APS result and

the lower presents the EEPS	109

Figure 90. Particle size data during the burnish cycle of the Sienna rear OES-
NAO pads. The upper plot presents the APS result and the lower

presents the EEPS	110

Figure 91. Speed and Brake Pressure traces for a selected point of the

dynamometer test cycle for a test of the Camry and a test of the Prius 112
Figure 92. Temperature Traces of the Front Brakes of the Camry and Prius over
WLTP-Brake Trip 10, operating on the test track and brake

dynamometer	113

Figure 93. Temperature Traces of the rear Brakes of the Camry and Prius over
WLTP-Brake Trip 10, operating on the test track and brake

dynamometer	113

Figure 94. Reference trends by date for the 47mm PTFE and 100S4 PM10 mass

measurements (left axis); and CPC particle count (right axis)	114

Figure 95. Various literature values for brake emissions, with the ranges from this

study overlaid for comparison	116

Figure 96. Particle number emission rates vs brake event average VSP for all
braking events in the test program. Power fits are shown by friction

material type	119

Figure 97. QCM-measured PM2.5 mass emission rate vs brake event average
VSP for all braking events in the test program. Power fits are shown

by friction material type	120

Figure 98. Total vehicle test cycle PM mass emissions vs simulated vehicle test

weight, categorized by pad material	122

vii


-------
List of Tables

Table 1. Counts of Make, Series, Model, and Model Year	4

Table 2. Estimated Pad Material Breakdowns by Combined Models	6

Table 3. Series, Model, Model Year, and FMSI Code for Most Common Vehicles .... 8

Table 4. Consolidating Top Series by Model Year and FMSI	9

Table 5. Counts of the Top 25 FMSI-Grouped Vehicles in California	10

Table 6. Top FMSI-Grouped Series, with BWh	11

Table 7. Top FSMI-Grouped Series, with BWI2	12

Table 8. The 6 Specific Makes, Series, and Model Years Selected for Testing	13

Table 9. Temperature Model Coefficients (based on the Toyota Camry)	17

Table 10. Statistics on data logged with the 3 different logger types in the

Caltrans Survey Data	20

Table 11. The percentage of Caltrans-survey and new test cycle total distance

traveled by microtrips within each average speed range	28

Table 12. Relevant parameters of the three candidate brake test cycles	30

Table 13. Brake Dynamometer Test Matrix Parameter Summary	41

Table 14. The CVS cooling/sample flow settings for each vehicle/axle

combination	46

Table 15. Brake Torque Split Percentages Based on SAE Standard J2789	47

Table 16. Results of SAE J2951 Analysis of the Sienna Front Assembly

Operating over the CBDC Test Cycle	48

Table 17. Updated parameters used for the simulation of the Prius regenerative

braking system on the dynamometer	52

Table 18. The various filter media types and the respective analyses for each	56

Table 19. The particulate filter types used during testing	57

Table 20. LINK Arizona Dust Test Parameters	58

Table 21. Estimated balance of friction materials by model for vehicles models at

3 and 11 years old	75

Table 22. Measured in-use brake emission rates by model, estimated for 7 year

old vehicles	75

Table 23. Slopes and intercepts for linear fits by material for PM mass emission

rate (mg/mi) versus vehicle tested weight (kg)	80

Table 24. Selected statistics for single-wheel particle number counts for all tests

(#/mi)	82

Table 25. PM counts and mass measurements for the two tunnel blank

procedures as well as the averages for all brake emissions tests	91

Table 26. Comparison of Overall Average Measured Brake Torque for Camry

and Prius	111

Table 27. Ranges of vehicle-level PM emission rates (mg/mi), summarized in

literature and for the vehicle models in this study	115

Table 28. Represented time and represented distances for the overall CBDC and

each speed segment	117

Table 29. Estimated in-use brake emission rates per unit time of braking by

model	118

Table 30. Estimated Deterioration Rates Based on Friction Material Trend with

Vehicle Age	121

viii


-------
Table 31. Estimated in-use brake PM emission rates by model (7 years old)	122

Table 32. Estimated in-use brake emission rates per unit time of braking by

model	123

Table 33. Selected light-duty tire and roadway PM emission factors from

literature	128

ix


-------
Executive Summary

Background

As tailpipe emissions of PM from the light-duty fleet have decreased significantly, non-
tailpipe PM emissions, such as brake and tire wear have become more relevant and
may be significantly impacting air quality near roadways. This research project was
jointly funded by EPA and CARB to measure and analyze particulate matter (PM)
emitted during light-duty vehicle braking to allow for updating inventory model emission
factors as well as to better understand the vehicle operational conditions associated
with varying levels of brake PM emissions. This study utilized a LINK Engineering
(LINK) brake dynamometer (in which the brake components of a single wheel are
mounted and operated electronically) for the measurement of PM emissions over a
prescribed driving cycle.

Objectives and Methods

Based on literature search, Eastern Research Group (ERG) and LINK determined that 4
parameters of a test cycle would be most likely to affect braking PM emissions: vehicle
speed, deceleration rate, brake component temperature, and the duration of a given
brake event. These four parameters formed the basis of the development of a new test
cycle and procedure to test vehicle brake assemblies on the brake dynamometer.

Six test vehicles (with common cross-platform brake components) were selected to
represent the range of vehicle types in the light-duty fleet. These vehicles were subject
to track tests in which brake system temperatures were measured during standardized
driving cycles. This temperature data was used to both inform a light-duty brake
temperature model that ERG developed (temperature as a function of vehicle speed)
and to select the cooling level used during brake dynamometer tests.

A new test cycle was developed to represent the operation of real-world vehicles in
California based on the 2010-2012 Caltrans Household Travel Survey. This survey
included speed data logged from over 2,000 vehicles operating throughout the state.
ERG used the developed temperature model to estimate the distribution of brake
temperatures encountered by these actual California trips. ERG developed a new test
cycle to be as similar as possible to the speeds, deceleration rates, temperatures, and
braking durations encountered by real vehicles. This new cycle is composed of three
distinctive operation patterns characterized with 3 speed bins such that emission rates
could be resolved across different trip average speed ranges.

The LINK lab site included a constant volume sampling (CVS) system with an integrated
brake dynamometer. The airflow through the CVS provides the medium for transferring
brake particles to the point of sampling and brake cooling. Cooling flowrates were
selected for the front and rear brake for each model to match temperatures encountered
during track testing. Measurements were made in batch and continuously by a variety of
instruments including gravimetric sampling in parallel on coated aluminum impactors
(TSI 100S4) as well as on 47mm Teflon filters. Instrumentation was also installed to

x


-------
measure particle size distributions, particle counts, and continuous particle mass. A test
matrix was developed consisting of 85 single-day tests of different test parameters:
brake friction materials, vehicle test weights, and the 6 selected vehicles. All test
parameter combinations were replicated in at least two different tests. One of the
vehicles was equipped with regenerative braking and the dynamometer was
programmed to simulate that regenerative braking function during testing of that model.

Results

Figure ES-1 presents the vehicle-level PM10 emissions results for each of the 6 tested
models by three different pad materials: Original Equipment Service non-asbestos
organic (OES-NAO), aftermarket NAO, and aftermarket Low-Metallic (LM). Vehicle-level
emissions for each pad type are calculated by doubling the average front and rear
single-wheel emission rates and summing.

Correlations were also found
between PM emissions and
other test parameters. In
general, for a given friction
material type, emission rates
increased linearly with vehicle
tested weight. Particle count
emissions generally trended
with overall mass, however
the observed range from the
lowest to highest emitter was
reduced as compared to the
range of measured masses.

ERG also analyzed emissions measurements for the purposes of comparison to
existing emissions factors in EPA's Motor Vehicle Emissions Simulator (MOVES), which
includes an operating mode (Opmode) bin specifically for braking. In MOVES2014, this
bin includes a PM2.5 emission rate of 0.558 g/hr during braking. The PM2.5 test results
from this program, when averaged (and not weighted by model prevalence in-use), had
an average of 0.71 g/hr during braking. These values could be different because the
vehicles in the program do not necessarily completely represent the prevalence of all in-
use vehicles. Also, braking is partially represented in the coasting Opmode bins so,
taken together with the emissions of the braking bin, the total modeled emissions may
will be closer to the average for this work.

Conclusions

The test results were sensitive to the various parameters in the test matrix. When in-use
pad material rates are accounted for, the estimated vehicle-level PM2.5 emission rates
vary from 1.9 mg/mi to 5.4 mg/mi depending on model. Metallic pad materials tended to
emit higher PM masses and tended to have larger particles in their emitted size
distributions.

40

2 —	¦ After-N AO
I jE 30

O- BO

£	HLW: Heavily Laden Weight

S 20

_i c

¦s E 10
> m

I OES-NAO
I After-NAC
After-LM

til

Camry Civic F-150 F-150 Prius Rogue Sienna Sienna
HLW	HLW

Figure ES-1. Vehicle level braking emissions by
model and friction material

xi


-------
Main Report

Introduction

This report presents the findings of a research study to measure and analyze particulate
matter (PM) emitted during light-duty vehicle braking. Research was conducted by
Eastern Research Group, Inc. (ERG) and subcontractor LINK Engineering (LINK) and
funded in cooperation between the US Environmental Protection Agency (EPA) and the
California Air Resources Board (CARB). The project was initiated by CARB Project
17RD016 for the purposes of updating the emission factors in the Emission Factor
inventory model (EMFAC). EPA contributed additional funding to broaden the types of
measurements made during this work with the aim of potentially updating the emissions
factors used in the Motor Vehicle Emissions Simulator (MOVES). This report covers all
aspects of the project and, where appropriate, presents results in a manner comparable
to MOVES emission factors.

EPA provided additional funding to this project to expand the types of measurements
planned in the original CARB study. These additions involved the parallel gravimetric
measurement of PM emissions using Teflon (PTFE) filters in a manner consistent with
the requirements for PM sampling described in 40 CFR 1065. Additionally, the project
funding allowed for the use of the sampling of PM onto transmission electron
microscope (TEM) grids during many of the tests.

As tailpipe emissions of PM from the light-duty fleet have decreased significantly, non-
tailpipe PM emissions, such as brake and tire wear have become more relevant and
may be significantly impacting air quality near roadways. This research was conducted
to inform the relative level of PM emissions from braking in the light-duty fleet,
characterize PM emission factors to potentially allow for emissions model updates, and
generally better understand the variables most significant in affecting brake emissions.

Light-Duty Vehicle Braking Systems

Typical light-duty vehicle braking systems rely on hydraulically activated friction
between two component surfaces, one that is generally stationary with respect to the
vehicle's suspension, and the other that rotates with the vehicle's wheel. During braking,
the friction generated between brake components generates heat and abrades the
components at their frictional interface resulting in particles being released into the
atmosphere.

Most modern vehicles rely on pad and rotor combinations for braking; the rotor rotates
with the wheel, and the pads are mounted in a caliper stationary to the suspension. The
driver's application of force to the brake pedal provides hydraulic pressure within the
caliper to squeeze the pads against opposite sides of the rotor and the resulting
frictional torque slows the vehicle. Less common in modern light-duty vehicles are drum
brakes, in which a cylindrical drum rotates with the wheel and hydraulic force presses
stationary shoes against the inside of the rotating drum to provide deceleration torque.

1


-------
Hybrid and modern electric vehicles also employ a relatively new method of slowing the
vehicle, regenerative braking, that operates in parallel with the hydraulic brake system
and uses the electric drive motor to slow the vehicle and provide energy back to the
vehicle's electric storage. These vehicles use complex blending strategies to manage
the amount of braking between the regenerative system and the hydraulic system such
that driver perceives the system as operating like a conventional vehicle.

LINK Brake Dynamometer Experience and Capabilities

ERG partnered with LINK at the proposal stage of CARB project 17RD016. LINK has
been designing, manufacturing, and performing testing with brake dynamometers for
decades. Their organization is a part of the Particle Matter Program (PMP) overseen by
the EU's Joint Research Centre (JRC). They have been involved in recent years with
the development of the state of the art for the measurement of PM emissions from
braking. LINK has the capability to:

•	conduct particulate matter emissions sampling of different brake configurations
using a brake dynamometer and representative drive cycle.

•	sample particulate matter emissions between 6 nm and 20 |jm in size,

•	collect particle characterization measurements of the particles, including particle
number, particle mass, and size distribution

•	collect continuous particulate matter measurements (g/sec), with the ability to
correlate the continuous measurements to gravimetric filter measurements

LINK operates in accordance with all sections of the ISO 17025:2017 standard, titled
General Requirements for the Competence of Testing and Calibration Laboratories1,
and management is committed to continually improving the quality of all operations. A
familiarity with, implementation of and compliance with the ISO 17025:2017 standard is
a mandatory requirement for individuals at all levels of the LINK organization.

ERG initiated the program by reviewing two existing test cycles relevant to this work, the
EMFAC Unified Cycle (UC) and its associated Speed Correction Cycles (SCCs), as well
as the World-Harmonized WLTP-Brake cycle, developed in Europe in cooperation with
the JRC.2 The UC/SCCs were designed for exhaust emissions testing, while the WLTP-
Brake was designed specifically for use on brake dynamometers. ERG also reviewed
available literature to begin developing a list of vehicle operational parameters most
relevant to brake emissions. ERG selected the following four parameters as the initial
assumption of the operational parameters with the most relevance to brake PM
emissions:

•	Vehicle speed during braking event

•	Deceleration rate

•	Brake component temperature

1	International Standards Organization "General requirements for the competence of testing and
calibration laboratories", https://www.iso.org/standard/66912.html

2	Mathissen, M. et. al., A novel real-world braking cycle for studying brake wear particle emissions, Wear,
Volumes 414-415, 2018

2


-------
• Brake event duration

The test program was designed to evaluate the PM emission responses to the above
parameters as well as to perform testing over a representative range of values for each.

Materials and Methods

Representative Test Vehicle and Friction Material Selection

ERG and LINK first determined the list of representative test vehicle models that would
be used to develop an understanding of brake thermal regimes during in-use vehicle
operation. Brake system components from these vehicle models would then be used
during PM emissions tests on the brake dynamometer. The on-track vehicle testing
involved operating each vehicle over the WLTP-Brake cycle (adapted for track testing)
followed by a heating and cooling matrix cycle of standardized stops and cruises. By
measuring brake temperatures during these on track operations, the laboratory testing
could be designed to replicate actual vehicle temperatures within reason and provide
relevant measurement and characterization of brake emissions. Also, the temperature
ranges encountered during track testing informed a new ERG-developed model of
operational temperatures of the driving in an in-use vehicle speed dataset. The
measurement of brake emissions using repeatable and reproducible systems and
isokinetic constant volume sampling will further improve the estimation of emissions
inventories for light duty vehicles.

To guide the selection of vehicle models, LINK adapted the concept of the brake wear
index (BWI), which is a representation of the total material present in the in-use fleet
that can be emitted as particulate. There are several assumptions used to determine a
BWI to guide the vehicle selection. First, the ranking assumes proportionality between
the number of registered vehicles and the amount of PM becoming airborne during and
after braking. Second, the larger the wearable mass of the foundation brakes, the larger
the potential for contribution of PM. The wearable mass is assumed to be a direct
function of the friction material volume before reaching the service thickness. In
addition to the friction material, the wearable mass includes the volume of the mating
disc or drum with its volume up to its service thickness. To estimate the mass, the
method uses nominal and typical material density for the predominant compositions
(non-asbestos organic/ceramic versus semi-metallic or low-metallic for the friction
material) with the estimated market share as a function of the vehicle age range. The
wearable mass uses grey cast iron for the disc and drums. The third assumption to
complete the computation of a BWI, combines the number of registered vehicles with
the total wearable mass as well as the replacement rate of the friction material and the
disc or drum. The main sources of information used by LINK in this analysis were: a)
registration counts by make, model, and year of manufacture provided by CARB for
2016; b) brake dimensions obtained from the yearly publication and the online catalog
from the Friction Materials Standards Institute (FMSI®) and from consulting with
aftermarket friction manufacturers; c) material densities from consulting with friction

3


-------
materials formulators, and d) replacement rates as the fraction of vehicles that have had
brake materials replaced during a given year for vehicles in a given age range.
The selection of six vehicles to characterize brake emissions representative of the light-
duty vehicle market in California was completed in multiple steps. The first step was to
review the statistics for vehicles registered in California for the year 2017. The top-25
vehicles based on registration counts included 'make' and 'series' of different 'model
year' and 'model'. Therefore, LINK prescreened vehicles of different series from the
original statistics. The column "Top-25 Reg" in Table 1 illustrates an extract of this pre-
screening step. The approach is to treat all models of Toyota Corolla as a single entry in
the development of the top-25 vehicles list. The next vehicle with a series different than
Corolla is Toyota Camry. Camry was selected as the second vehicle in the top-25
vehicles list. Only the first few vehicles are included here for brevity, but the complete
list is presented later in this section.

Table 1. Counts of Make, Series, Model, and Model Year

Entry #

Model
Year

Make

Series

Model

Count

Top-25
Reg

1

2016

Toyota

Corolla

L

58637

1

2

2015

Toyota

Corolla

L

55315

1

3

2010

Toyota

Corolla

BASE

54362

1

4

2014

Toyota

Camry

L

53570

2

5

2015

Toyota

Prius



53318

3

Next, the friction material and backing plate identification codes ('FMSI' codes) of front
and rear axle friction materials were determined for more than 100 different entries of
different model years, make, series, and models. The number of vehicles having the
same FMSI codes were then summed together and this new count was taken for each
model, grouped together by all series and model years having the same FMSI codes.
The FMSI database and industry surveys were used to determine the dimensions of
disc or drum and friction material. These dimensions, along with representative material
density values were used to estimate the wearable mass of the brake parts using the
following equation. The actual values used during this process for each of the top-25
vehicles are included in Appendix A.

Estimated wearable mass =

(Wearable volume of friction material) ¦ (Density of friction material)

+ (Wearable volume of front axle disc) ¦ (Density of disc)

+ (Wearable volume of rear axle disc/drum) ¦ (Density of disc/drum)

Density of friction material depends on the type of friction material used in the vehicle of
interest. For example, density of non-asbestos organic (NAO) material is 2.9 g/cm3 and
density of semi- or low-metallic (LM) material is 3.75 g/cm3. Thus, a brake lining made
of semi-metallic material contains higher wearable mass than an equivalent lining made
of NAO material.

4


-------
Interviews and surveys with technical specialists resulted in percent population of
friction material formulations as a factor of vehicle age. Figure 1 illustrates the overall
general trend regarding the relative presence of NAO and Semi-metallic/Low-metallic
(SM/LM) friction materials for light-duty vehicles in use. Except for special applications,
most street service passenger car and light truck Original Equipment Manufacturer
(OEM) and Original Equipment Service (OES) pads in North America use NAO friction
materials. As vehicles age, SM/LM pads become more common as they tend to cost
less than NAO pads and are therefore more likely to be selected as vehicles depreciate.

100%

90%

80%

70%

60%

50%

40%

30%

20%

10%

0%

3 years	7 years	11 years

¦ NAO ¦ LM/SM
Figure 1. General trend in pad material mix by vehicle age

Using these market statistics based on vehicle age, LINK generated Table 2, which
indicates the estimated mix of metallic versus NAO for the six vehicles selected. The
wearable mass for the friction material prorates the density of the friction material (NAO
v. Semi- or Low-Metallic) and the estimated market share. Note that the estimates
presented are based on vehicle age range and general vehicle type only and are not
adjusted for specific vehicle characteristics (as that level of detail was not available from
industry surveys). For example, the Prius may have a different level of low/semi-metallic
market share as a result of its regenerative braking systems.

5


-------
Table 2. Estimated Pad Material Breakdowns by Combined Models

Make

Model

MY

Production

Est. %

Est. %







Years

NAO

Metallic

HONDA

CIVIC LX

2012-2015

3-6

77%

23%

TOYOTA

PRIUS REGULAR

2010-2016

2-8

82%

18%

NISSAN

ROGUE S

2014-2016

2-4

88%

12%

TOYOTA

CAMRY (BASE, L, LE)

2009-2016

2-9

82%

18%

TOYOTA

SIENNA LE

2011-2015

3-7

77%

23%

FORD

F150 SUPERCREW

2015-2016

2-3

87%

13%

In addition, it is important to note the fact that, in aftermarket supplier parlance, there
are three broad commonly-used categories of aftermarket friction materials of "good,
better, and best." Good friction materials provide a good performance and relatively
quiet braking at a reasonable price in NAO and semi-metallic formulations. Better
friction materials are the most extensive line of aftermarket friction materials and also
come in NAO and semi-metallic formulations. They are designed to last longer and wear
better and perform well at the mid-price range. Better friction materials feature
chamfers, slots, and anti-noise shims in many applications, and provide smooth pedal
feel and proper fit. Best aftermarket friction materials are the closest to the OEM/OES
friction material in terms of dimensional quality, hardware kits, performance, comfort,
and product life. Figure 2 illustrates the general mix of friction material quality in relation
to the vehicle age. In the chart, the total percentages for NAO and SM sum to 100%
within each vehicle age group.



70%



60%

Q_



3

o

50%

5



CD
CuO

40%

<



c



+-»

30%

c




-------
When all the above factors are combined in one graph, it becomes apparent how wide
the range of friction materials is that can be applied to a given vehicle during a given
brake job. Figure 3 illustrates the combined effect of vehicle age on friction material type
and formulation equipped on in-use vehicles. The figure is based on combining the data
presented in Figure 1 and Figure 2.

100%

90%

80%

70%

60%

50%

40%

30%

20%

10%

0%

3 years	7 years	11 years

I best NAO ¦ better NAO good NAO SS better LM/SM good LM/SM
Figure 3. Overall Mix of Friction Materials, Quality Grade, and Vehicle Age

Other factors which influence the ultimate friction material used for the specific vehicle
include vehicle aging behavior, demographics, market dynamics among dealers, large
retailers, and private branding, etc.

Based on aforementioned steps to determine registration count as well as the wearable
mass (including pad material type), a first iteration index, BWh was determined for the
top-registered vehicles as follows.

BWh = (Registration Count) ¦ (Wearable mass)

To complete the numerical assessment of which vehicles were relevant, representative,
and available for rental to conduct the proving ground test track measurements, LINK
conducted a technical survey to determine the replacement rates of brakes. The survey
provided by the brake suppliers and manufacturers resulted in replacement rates as a
function of vehicle age. A second BWI index (BWI2) was then determined as follows:

BWI2 = (BWh) ¦ (Replacement rate by vehicle age)

7


-------
In the end, three levels of ranking were established among the top 25 vehicles. The total
registration count of vehicles for a given make, series, and model resulted in the first
phase of ranking. The BWh, with the total wearable mass resulted in a second phase of
ranking. The BWI2, includes an additional factor (replacement rate), to generate the
third phase for the vehicle ranking. The BWI2 allows the adjustment of the BWh index to
reflect the relative wear rate of different vehicles having a similar registration count, and
a similar wearable mass. Low replacement rates will demote the total ranking for the
vehicle as it wears (releases debris and PM) at a lower rate, compared to another
vehicle with similar registration count and wearable mass but with a higher wear rate
(reflected indirectly by the replacement rate).

Vehicle models were grouped by all series and model years within a model that had the
same FMSI codes. Table 3 lists an excerpt of the top vehicles of various make, series,
model year, and models. Shaded rows indicate vehicles that are 10 years old or newer
from the publication date of registration statistics. This age range is considered to avoid
uncertainties of working condition, availability and procurement of specific vehicles for
track testing with older vehicles. The last column in Table 3 'FMSI FRONT AXLE' shows
the friction material identifier for brake lining on front axle. The first four digits represent
the friction formulation and the digits after the letter represent the geometrical features
and dimensions of backing plate. As seen, the same FMSI identifiers apply for various
entries (vehicle make, series, model year, and model).

Table 3. Series, Model, Model Year, and FMSI Code for Most Common Vehicles

MAKE

SERIES

MY

MODEL

COUNT

FMSI FRONT AXLE

TOYOTA

COROLLA

2016

LX

58637

8969-D1210

TOYOTA

COROLLA

2015

LX

55315

8969-D1210

TOYOTA

COROLLA

2014

LX

45202

8969-D1210

TOYOTA

COROLLA

2013

BASE

45180

8330-D1210

TOYOTA

COROLLA

2012

BASE

27327

8330-D1210

TOYOTA

COROLLA

2011

BASE

26460

8330-D1210

TOYOTA

COROLLA

2010

BASE

54362

8330-D1210

TOYOTA

COROLLA

2009

BASE

38783

8330-D1210

TOYOTA

COROLLA

2007

CE

36541

7824-D923

TOYOTA

COROLLA

2006

CE

38106

7824-D923

TOYOTA

COROLLA

2005

CE

39963

7824-D923

TOYOTA

COROLLA

2004

CE

30255

7824-D923

TOYOTA

COROLLA

2003

CE

34015

7824-D923

TOYOTA

COROLLA

2001

CE

20481

7611-D741

TOYOTA

COROLLA

1999

VE

20106

7611-D741

TOYOTA

CAMRY

2016

LE

45855

8331-D1293

TOYOTA

CAMRY

2015

LE

46855

8331-D1293

TOYOTA

CAMRY

2014

L

53570

8331-D1293

TOYOTA

CAMRY

2013

L

30383

8331-D1293

8


-------
Table 4 shows the revised list that shows the combination of all rows in Table 3 with a
same FMSI # into a single entry that covers a model year range and all model
identifiers.

Table 4. Consolidating Top Series by Model Year and FMSI

MAKE

SERIES

MY

COUNT

AXLE

FMSI FA

TOYOTA

COROLLA

2014-2016

159154

FA

8330-D1210

TOYOTA

COROLLA

2014-2016

159154

RA

1635-S945

TOYOTA

COROLLA

2009-2013

192112

FA

8330-D1210

TOYOTA

COROLLA

2009-2013

192112

RA

1635-S945

TOYOTA

COROLLA

2003-2007

178880

FA

7824-D923

TOYOTA

COROLLA

2003-2007

178880

RA

1515-S801

TOYOTA

COROLLA

1999-2001

40587

FA

7611-D741

TOYOTA

COROLLA

1999-2001

40587

RA

1515-S750

TOYOTA

CAMRY

2009-2016

342992

FA

8331-D1293

TOYOTA

CAMRY

2009-2016

342992

RA

8332-D1212

TOYOTA

CAMRY

2007

50693

FA

8331 -D1222

TOYOTA

CAMRY

2007

50693

RA

8332-D1212

TOYOTA

CAMRY

2002-2006

185766

FA

7787-D908

TOYOTA

CAMRY

2005-2006

67612

RA

1617-S911

TOYOTA

CAMRY

2002-2004

118154

FA

7787-D908

TOYOTA

CAMRY

1998-2004

198316

RA

1447-587

TOYOTA

CAMRY

2001

25651

FA

7357-D697

Table 5 lists the top 25 light-duty vehicles registered in the state of California, as
grouped according to the above process. It should be noted that the column labeled
'weight class' is defined according to CFR 45, Part 565 VIN (Class A = GVWR less than
or equal to 3000 lbs., B for (3001 to 4000) lbs., C for (4001-5000 lbs., etc., up to GVWR
of 10 000 lbs.). The top 25 vehicles list includes a mix of a variety of vehicle weight
classes spanning from class B thru F. The goal in vehicle selection was to select
representative models from a range of vehicle types including compacts, sedans, SUVs,
minivans, full-size trucks, as well as at least one vehicle with regenerative braking.
These corresponded in some cases to the weight classes, but the priority in diversifying
vehicle selection was on vehicle type, not weight. Based on the registration count
ranking only, the vehicles selected would have been: Camry, Prius, Corolla, Altima,
Civic, and Sentra. These vehicles fall under two vehicle weight classes and one vehicle
type (4-door sedan), and alone do not provide enough representation of the vehicle
population in California to suit the needs of this program. The top 25 vehicles were
selected to ensure multiple models would remain for each of the following desired
vehicle groups: compacts, sedans, pickups, minivans, SUVs, and hybrids.

9


-------
Table 5. Counts of the Top 25 FMSI-Grouped Vehicles in California

MAKE

SERIES/MODEL

MY

WEIGHT
CLASS

GVWR

Reg #

RANK

BY

Reg

/ A-to-H

/ kg

#

TOYOTA

CAMRY (BASE, L, LE)

2009-2016

C

2073

342992

1

TOYOTA

PRIUS REGULAR

2010-2016

B

1800

241055

2

TOYOTA

COROLLA L

2014-2016

B

1732

159154

3

NISSAN

ALTIMA (BASE, 2.5)

2012-2016

C

1910

149096

4

HONDA

CIVIC LX

2012-2015

B

1595

140733

5

NISSAN

SENTRA S

2013-2016

B

1687

110629

6

HONDA

ACCORD LX

2014-2016

C

1934

52193

7

TOYOTA

SIENNA LE

2011-2015

D

2715

44921

8

LEXUS

RX 350

2014-2015

D

2527

43306

9

NISSAN

ROGUE S

2014-2016

C

1968

41213

9

HYUNDAI

SONATA (GLS, SE, SPORT)

2013-2015

C

2074

40117

11

HONDA

ACCORD EX

2014-2016

C

1904

39344

12

HONDA

ACCORD SPORT

2014-2015

C

2107

37332

13

TOYOTA

RAV4 XLE

2014-2016

C

2035

36803

14

TOYOTA

TACOMA DOUBLE CAB

2015-2016

D

2540

36052

15

FORD

F150 SUPERCREW

2013-2014

F

3239

33721

16

FORD

F150 SUPERCREW

2015-2016

E

3000

32921

17

HYUNDAI

ELANTRA GLS

2013

B

1720

30566

18

CHEVROLET

SILVERADO 1500

2014-2015

E

3085

27578

19

HONDA

CIVIC LX

2016

B

1695

25782

20

HONDA

ACCORD SPORT

2016

C

1964

22978

21

DODGE

RAM 1500 ST

2004

E

2989

19739

22

CHEVROLET

TAHOE C1500

2007

F

3266

19517

23

LEXUS

RX 350

2016

D

2562

12540

24

HYUNDAI

SONATA SE

2016

C

2074

11363

25

Table 6 shows the revised and re-ordered ranking based on the brake wear index BWh.
The index BWh is a product of registration count and the total wearable mass of
pads/shoes and discs/drums at all four brake corners of a given vehicle. On comparing
the registration counts with the BWh rankings, the influence of wearable mass on BWh
ranking was more evident for SUVs and trucks such as the Nissan Rogue, Toyota
Tacoma, and F150 trucks. Based on the BWh ranking, the vehicles selected would have
been: Camry, Corolla, Prius, Civic, Altima, and Tacoma. These vehicles fall under three
vehicle weight classes and two vehicle types (4-door sedan and pick-up truck). The
selection of the Tacoma would overemphasize this vehicle due to its wearable mass
(over two times the average of the other five vehicles selected with the BWh ranking).
In addition, this selection limits the selection of vehicle types, makes, and models
present in California.

10


-------
Table 6. Top FMSI-Grouped Series, with BWh

MAKE

SERIES/MODEL

MY

Reg
#

RANK

BY

Reg

TOTAL
WEARABLE
MASS
/gm

BWh
/ton

RANK

BY

BWh

TOYOTA

CAMRY (BASE, L, LE)

2009-2016

342992

1

2133

732

1

TOYOTA

COROLLA L

2014-2016

159154

3

3028

482

2

TOYOTA

PRIUS REGULAR

2010-2016

241055

2

1749

422

3

HONDA

CIVIC LX

2012-2015

140733

5

2322

327

4

NISSAN

ALTIMA (BASE, 2.5)

2012-2016

149096

4

1510

225

5

TOYOTA

TACOMA DOUBLE CAB

2015-2016

36052

15

5256

189

6

NISSAN

SENTRA S

2013-2016

110629

6

1436

159

7

TOYOTA

SIENNA LE

2011-2015

44921

8

2717

122

8

LEXUS

RX 350

2014-2015

43306

9

2707

117

9

FORD

F150 SUPERCREW

2013-2014

33721

16

2878

97

10

FORD

F150 SUPERCREW

2015-2016

32921

17

2895

95

11

TOYOTA

RAV4 XLE

2014-2016

36803

14

2462

91

12

HONDA

ACCORD LX

2014-2016

52193

7

1598

83

13

NISSAN

ROGUE S

2014-2016

41213

10

1845

76

14

HYUNDAI

SONATA (GLS, SE, SPORT)

2013-2015

40117

11

1678

67

15

CHEVROLET

SILVERADO 1500

2014-2015

27578

19

2431

67

16

HYUNDAI

ELANTRA GLS

2013

30566

18

1649

50

17

CHEVROLET

TAHOE C1500

2007

19517

23

2521

49

18

DODGE

RAM 1500 ST

2004

19739

22

2180

43

19

HONDA

CIVIC LX

2016

25782

20

1666

43

20

HONDA

ACCORD EX

2014-2016

39344

12

993

39

21

LEXUS

RX 350

2016

12540

24

2668

33

22

HONDA

ACCORD SPORT

2014-2015

37332

13

803

30

23

HYUNDAI

SONATA SE

2016

11363

25

1803

20

24

HONDA

ACCORD SPORT

2016

22978

21

803

18

25

The total wearable mass may or may not indicate the actual amount of wear debris
seen during a typical year of driving activity because different models' brake
components do not all wear out after the same amount of driving. The annual estimates
of the brake replacement rates were acquired through a business intelligence survey.
Table 7 presents the continued analysis by including the replacement rates by vehicle
age. A newer vehicle is expected to have low value of replacement rate and vice versa.
A value of 16% in the first row of Table 7 implies that 16% of all vehicle population with
year of manufacture between 2009 and 2016 are estimated to have had a brake
service. The prevailing brake service job at the time of this survey was to replace friction
couple i.e. pad and rotor or shoes and drum at any given service. A second wear index
is formulated which assumes that a brake service would include replacement of brake
parts. The wear index BWI2 is a product of BWh and the brake replacement rate. Table
7 includes vehicles in the order determined by BWI2. This ranking still leaves the
Sentra, Altima, and Corolla near the top of the list, though they are in the same vehicle
class.

11


-------
Table 7. Top FSMI-Grouped Series, with BWh

MAKE

SERIES/MODEL

MY

RANK

BY

BWh

REPLACEMENT
RATE PER
VEHICLE AGE

BWh

RANK

BY

BWh









/%

/ ton



TOYOTA

CAMRY (BASE, L, LE)

2009-2016

1

16

117

1

TOYOTA

COROLLA L

2014-2016

2

11

53

2

HONDA

CIVIC LX

2012-2015

4

14

46

3

NISSAN

ALTIMA (BASE, 2.5)

2012-2016

5

14

32

4

NISSAN

SENTRA S

2013-2016

7

11

17

5

TOYOTA

SIENNA LE

2011-2015

8

14

17

6

FORD

F150 SUPERCREW

2013-2014

10

16

16

7

LEXUS

RX 350

2014-2015

9

11

13

8

CHEVROLET

TAHOE C1500

2007

18

23

11

9

FORD

F150 SUPERCREW

2015-2016

11

11

10

10

TOYOTA

RAV4 XLE

2014-2016

12

11

10

11

TOYOTA

TACOMA DOUBLE CAB

2015-2016

6

5

9

12

HONDA

ACCORD LX

2014-2016

13

11

9

13

DODGE

RAM 1500 ST

2004

19

21

9

14

TOYOTA

PRIUS REGULAR

2010-2016

3

2

8

15

NISSAN

ROGUE S

2014-2016

14

11

8

16

HYUNDAI

ELANTRA GLS

2013

17

16

8

17

HYUNDAI

SONATA (GLS, SE, SPORT)

2013-2015

15

11

7

18

CHEVROLET

SILVERADO 1500

2014-2015

16

11

7

19

HONDA

ACCORD EX

2014-2016

21

11

4

20

HONDA

ACCORD SPORT

2014-2015

23

11

3

21

HONDA

CIVIC LX

2016

20

5

2

22

LEXUS

RX 350

2016

22

5

2

23

HYUNDAI

SONATA SE

2016

24

5

1

24

HONDA

ACCORD SPORT

2016

25

5

1

25

Table 8 indicates the final vehicle selection after a joint engineering review with CARB,
ERG, and LINK. The first update was to include an SUV, and the team selected the
Nissan Rogue. Note that the Lexus RX 350 was higher ranked among SUV's than the
Nissan Rogue, however the RX 350 was excluded due to it being a luxury vehicle and
being likely to have a significantly higher cost for brake components than the Rogue
which was not accounted for in the project budget estimate. The second update was to
include a pick-up truck, with the F-150 being selected for being the most common pick-
up truck in California (and the United States). The model year range for the F-150 was
selected to be 2015-2016 (instead of the previous FMSI model year range of 2013-
2014) because interviews with industry experts suggested that this year range was a
very common benchmarking and development candidate. After confirming the
availability for rental and availability of brake parts, the project moved on to its next
phase, to prepare the logistics and technical documentation for track testing.

12


-------
Table 8. The 6 Specific Makes, Series, and Model Years Selected for Testing

Selected Vehicle Model

Reg

BWh

BWI2

COMMENTS



RANK

RANK

RANK



2009-2016 Toyota Camry

1

1

1

Top rank by all three metrics

2012-2015 Honda Civic

5

4

3

Rear drum brakes

2011 -2015 Toyota Sienna

8

8

6

Top in the list of class 'D' Reg
#, Minivan

2015-2016 Ford F-150

17

11

10

Top in the list of class 'E' Reg
#, Large Pickup,

Very common vehicle for
friction material formulation
evaluations

2010-2016 Toyota Prius

2

3

15

Regenerative braking

2014-2016 Nissan Rogue

10

14

16

Top in the list of non-luxury
SUVs

Medium level ranking based
on BWh and BWI2

The processes followed during this phase of the project allowed the evaluation of a
significant list of vehicles in terms of make, model, and trim level, with the goal of
selecting a subset for track and dynamometer testing. The final list of six vehicles given
in Table 8 provides a good cross section of vehicle weight class/type and powertrain
systems while representing common light-duty vehicles used for brake development
and brake testing. The range of vehicle weights and disc brake dimensions allow the
proper characterization of the thermal regimes of a wide variety of light-duty vehicles
during the planned measurements of brake emissions on the brake inertia
dynamometer. This task also introduced the concept of BWI as a predictor for the total
potential contribution of a given vehicle to PM (fallout, airborne, and resuspension). The
inclusion of BWI2 as a metric for evaluation did not significantly alter the vehicle
selection as compared to selecting based on registration counts when also accounting
for selecting from a variety of vehicle types, however.

At the conclusion of this task, LINK procured one of each of the following vehicle
models for track testing. These vehicle models were also used as the basis for acquiring
brake assemblies for testing on the brake dynamometer. The selected FMSI numbers
are also included.

•	2011 Toyota Camry LE

o Front axle FMSI# 8331-D1293
o Rear axle FMSI# 8332-D1212

•	2013 Honda Civic LX

o Front axle FMSI# 8791-D1578
o Rear axle FMSI# 1618-S913

•	2013 Toyota Sienna LE

o Front axle FMSI# 8436-D1324

13


-------
o Rear axle FMSI# 8500-D1391

•	2015 Ford F-150 Supercrew

o Front axle FMSI# 8528-D1770
o Rear axle FMSI# 9018-D1790

•	2016 Toyota Prius Two Eco

o Front axle FMSI# 8538-D1184
o Rear axle FMSI# 8463-D1423

•	2016 Nissan Rogue S

o Front axle FMSI# 8449-D1737
o Rear axle FMSI# 8501-D1393

In addition to representing the vehicles registered in California, the listing provides the
appropriate representation of vehicle weight classes from 1700 kg to 3200 kg of gross
vehicle weight; vehicle types with sedans, pickup trucks, and sport utility vehicles; brake
systems with disc brakes all around and with drum brakes on the rear axle; and vehicles
with conventional gasoline powertrains and with regenerative braking systems. Note
that specific model years are given from the ranges of model years with the same FMSI
values provided in previous tables. The model years given above represent the actual
model years of test vehicles sourced by LINK for track testing.

Track Testing and the Brake Temperature Model

Existing literature regarding PM emissions from braking indicates that brake
temperature has a significant effect on emission rates.34 To account for this, ERG
selected brake temperature as a parameter for use in cycle development. Because
brake temperature was not measured during the Caltrans survey, ERG developed a
model of brake temperature based on track testing data. This temperature model was
used to determine the distribution of in-use brake temperatures associated with vehicle
speed data from the Caltrans household travel study and to estimate the brake
temperature profile during operation on the dynamometer for the new test cycle.

ERG and LINK conducted track testing to gather data about operational brake
temperatures. LINK acquired the test vehicles and replaced their wearable brake
components with new components. Test vehicles were instrumented for temperature
measurement of various brake system components and the vehicles were subject to
controlled driving in three different phases. A photograph of the type of temperature
measurement equipment installed by LINK is presented in Figure 4, which depicts a
brake mounted thermocouple installed on the Camry test vehicle. The thermocouple

3	Garg, Bhagwan D. et. al. "Brake Wear Particulate Matter Emissions." Environmental
Science & Technology 34.21 (2000): 4463-4469

4	Sanders, Paul G. et. al. "Airborne Brake Wear Debris: Size Distributions, Composition,
and a Comparison of Dynamometer and Vehicle Tests." Environmental Science &
Technology 37 (2003): 4060-4069

14


-------
wires are routed through the wheels to a wireless transmitter that rotates with the
vehicle's wheel and transmits measured data to a receiver mounted inside the vehicle.

Figure 4. Brake Thermocouples and wireless hub transmitter installed on the Camry test

vehicle

After installing the instrumentation, LINK then burnished the new friction materials of
each vehicle on-track over five repeats of Trips 2 and 4 of the WLTP-Brake cycle. Then,
brake temperatures were logged while driving the complete WLTP-Brake cycle followed
by an ERG-defined Heating and Cooling Matrix. The Heating and Cooling Matrix
consisted of a series of standardized stops and steady-speed cruises to help separately
analyze brake heating and cooling patterns. The events making up the Matrix are
tabulated in Appendix B. The WLTP-Brake and Heating and Cooling Matrix temperature
measurements were used in the development of ERG's brake temperature model.
Historgrams of rotor/drum temperatures measured over the WLTP-Brake cycle on the
test track are presented in Figure 5. In the figure, the temperature range is presented as
the temperature difference above the ambient temperature in the wheel well (which was
approximately 25°C during most of the testing.

15


-------


0.16





0.14
0.12

	 Civic • FA Disc

	 Prius - FA Disc

		Camry - FA Disc

	 Rogue - FA Disc

	Sienna - FA Disc

		 F150 - FA Disc



0.10



Density

0.08
0.06





0.04
0.02

	

>s3niiiiP" 	 _ _





0 30 60 90 120 150 180
Temperature above wheel well value / deg-C



0.16





0.14
0.12

?	Civic - RA Drum

	 Prius - RA Disc

\	Camry - RA Disc

... - Rogue - RA Disc

		Sienna - RA Disc

F150 - RA Disc



0.10



£
c

3

0.08
0.06
0.04

0.02





0.00

0 30 60 90 120 150 180
Temperature above wheel well value / deg-C

Figure 5. Histograms of temperatures of front rotors (top) and rear rotors/drums (bottom)
over the WLTP-Brake cycle operated on the test track.

ERG developed a generalized form of an equation to describe brake heating and
cooling rates based on an energy balance of the energy flow into the brakes during
deceleration and the energy flow from the brakes due to convective cooling. The
equation for change in temperature, AT, is of the form:

AT = (A + B Vo + OVo2) • (To - Tamb) ¦ Atime + D (V02 - V12) • Atime

In the equation, A, B, C, and D are heating/cooling coefficients specific to a given
vehicle. Vo is the vehicle speed (kph) at the prior instant and Vi is the speed (kph) at the
next instant. To is the brake temperature at the given instant (°C), and Tamb is the air
temperature around the brake system. Track testing was conducted for six vehicles to
determine brake temperature trends over the two measured driving cycles, the WLTP-
Brake cycle (adapted for track driving) and the Heating and Cooling Matrix consisting of
standardized stops of various intensities to determine brake heating rates as well as
steady-speed cruises to determine cooling rates. The temperature measurements
during track testing were used to determine the coefficients A, B, C, and D for the
different vehicles. Note that D is set to zero if braking is not taking place (i.e. the vehicle
is not decelerating more rapidly than would occur during a coastdown).

Track data excerpts containing only the cooling periods from the Heating/Cooling matrix
were initially extracted to determine cooling coefficients A, B, and C. The cooling
periods were extracted first because the cooling is a relatively slow process that follows

16


-------
a readily-modeled exponential decay, and any time delays in the data for this
operational regime would not have much effect on measured temperature and thereby
confound the modeling. Cooling coefficients were modeled by first using the nonlinear
regression procedure (Proc NLIN) in SAS on each of the 13 steady-speed cooling
segments of the Heating and Cooling Matrix to fit eA(A+Bv). Then, a polynomial fit to
speed was applied to the results of the NLIN procedure for the 13 different segments
using Excel. The coefficients of this polynomial model became the cooling coefficients.
Using these, ERG then determined the best single heating coefficient by determining
the best least-squares fit between the entire modeled and measured temperatures by
means of iteration using the already-determined cooling coefficients. For the units given
in the previous paragraph, the coefficients for the Toyota Camry are given in Table
9Table 9. The Toyota Camry was selected for use in modeling the Caltrans data (and
for cycle development) due to its representativeness of the vehicle fleet as well as
having a good fit between the modeled and measured temperatures of the test vehicles.
ERG determined that the decision to use the results of a single vehicle was acceptable
because the temperature model used for the complete Caltrans dataset was the same
as that used for cycle-building. The form of the equation is the key aspect, not the
specific coefficients; the temperature model was used only as a bridge between the
relative amounts of heating energy and cooling time in the in-use dataset and in the new
cycle.

Table 9. Temperature Model Coefficients (based on the Toyota Camry)

A, 1/s

B, 1/(skph)

C, 1/(skph2)

D, °C/(s kph2)

-0.001264

-0.000053926

0.0000001431

0.0088

The level of agreement between the model and the measured data is presented in
Figure 6. Note that there were 10 instances in the following graph in which the test
vehicle's braking system was allowed to return to at or near ambient (and in some
cases datalogging was stopped during those intervals to allow for driver rest). At these
times, the model was also set back to match the measured temperature (which was
near ambient at the end of these intervals). The plot presents approximately 8 hours of
operation.

17


-------
Track-Measured Right Front Rotor
Modeled Right Front Rotor
Vehicle Speed

Observation NumBer (s)

	tn to oo

fOfommfififOfOfito

500
450
400
350
300 ^

Q.

250 -a
aj
aj

200 a-

CO

150
100
50
0

Figure 6. The modeled right front outer rotor temperature and the corresponding track-

test measured temperature.

18


-------
Test Cycle Development

After completing the temperature model, ERG proceeded to the selection of a test cycle
to be followed on a brake dynamometer during PM emissions testing. The goal for the
project was to utilize a cycle that is representative of California driving and practical for
brake dynamometer operation. In this project, ERG developed a completely new cycle
as a candidate for use during testing under this project. Consideration was also given to
two existing test cycles, the EMFAC UC and its associated SCCs, as well as the World-
Harmonized WLTP-Brake cycle, developed in Europe specifically for use on brake
dynamometers. The UC was designed for exhaust emissions testing on a chassis
dynamometer, so this report will describe its potential application as a braking test. The
WLTP-Brake cycle is an "engineered" cycle, meaning it is not intended to be directly
based on actual driving traces, but rather consists of engineered braking events at
various deceleration thresholds.5 The engineered aspects were constructed from
vehicle activity data from Europe, the US, India, Japan, and South Korea. ERG
evaluated all three cycles with various measures of their representativeness of real-
world California driving. ERG presented the results of the evaluation to CARB and
collaborated to select the cycle that would be used during the PM testing during this
program. The driving cycle was developed prior to EPA involvement in the program.

Data Sources

This section describes the data sources used in this work including the two existing
candidate test cycles that were evaluated. The data sources included in-use on-road
vehicle survey data, temperature measurements performed on test vehicles operating
on the track, and data from EPA's new vehicle emissions certification results. The
existing cycles were either sourced from public information (for the WLTP-Brake cycle
and the UC) or provided by CARB (for the EMFAC SCCs). This section also describes
the methods that ERG followed to prepare and use these data sources in the selection
of the test cycle for use on the brake dynamometer in this project.

The key material gathered for cycle selection was the Caltrans 2010-2012 California
Household Survey data. This data includes actual in-use second-by-second operational
data (vehicle speed over time is of primary interest for this work) from a variety of
vehicle types operating across the state. ERG analyzed logged data from the operation
of over 2,000 vehicles including over 14,000 hours of operation. This data served as the
basis for evaluating the representativeness of both existing cycles (WLTP-Brake cycle
and the EMFAC UC/SCCs) as well as for the creation of a new ERG-developed brake
dynamometer test cycle.

5 Marcel Mathissen, Jaroslaw Grochowicz, Christian Schmidt, Rainer Vogt, Ferdinand
H. Farwick zum Hagen, Tomasz Grabiec, Heinz Steven and Theodoros Grigoratos, A
novel real-world braking cycle for studying brake wear particle emissions, Wear,
https://doi.Org/10.1016/j.wear.2018.07.020e

19


-------
Caltrans Survey Data

The Caltrans Survey included instrumentation of vehicles using either On-board
Diagnostic (OBD) dataloggers, GPS dataloggers, or both OBD and GPS dataloggers.
The survey was designed to be as random as possible and included participants from
each county in California making it an excellent data source to represent typical
California driving. The quantity of vehicles and logged data for each logger type is
presented in Table 10. It can be seen that the vast majority of the data was collected
using OBD-only dataloggers.

Table 10. Statistics on data logged with the 3 different logger types in the Caltrans

Survey Data



OBD-Only

OBD+GPS

GPS Only

Number of Vehicles

2130

365

677

Hours of Data

14,001

1,819

3,162

Time gaps of 2s in data

1.6%

0.1%

0.1%

The survey data included over 60 million seconds of data. Because of the large (and,
therefore computationally-intensive) amount of data, ERG elected to use only the
cleanest subset of data. After reviewing samples of each data type, ERG determined
that the OBD-only dataloggers appeared to generate the highest quality data. This
decision was based on reviews of the amount of clipping of trip starts and ends, the
data resolution, and the steadiness of the zero measurement when vehicles were at
idle. ERG also looked for time lags in the data (which would be of particular concern for
braking analysis). Calculated mean and median deceleration rates during braking
events were higher for the OBD-measured data than for the GPS data. This was
consistent with ERG's previous experience that GPS data tended to have speed lag in
time during acceleration or deceleration (which would likely result in a lower reported
deceleration rate for a given braking event). For all reasons given, ERG selected only
the OBD-only survey data for use in this work.

ERG applied further adjustments/corrections to the relatively clean OBD-only Caltrans
data. Most notably, OBD speed data is reported to the nearest whole kph value, with no
decimal places given. As a result, the speed traces were digitized, which challenged the
ability to discern braking events from cruising events in the speed-trace data on a 1-
second basis. This is because, due to digitization, an actual gradual reduction in speed
by a vehicle would be represented as a constant value with a single 1 kph/s jump,
followed by further constant values. Using a formula to assign braking would assign
braking to that one second even though the vehicle was actually coasting for the whole
period of time. To address this, ERG numerically smoothed all OBD data using the local
weighted regression procedure (Proc LOESS) in Statistical Analysis Software (SAS).
This procedure included a feature in which the optimum smoothing parameter can be
automatically detected. ERG found the average optimum smoothing parameter for each
vehicle and applied the average to all vehicles. The goal in the regression was to
address the digitization without over-smoothing and reducing the measured
deceleration rates of all braking events. Finally, some vehicles that had various speed

20


-------
discrepancies were dropped from analysis. Less than 5% of the OBD data was dropped
for this reason.

When reviewing a vehicle's smoothed speed trace data, the coastdown rate of the
vehicle is an important input for use in determining whether and when that vehicle's
brakes were applied. Because the specific make and model of each participating vehicle
in the Caltrans survey data was not available, ERG used a Generalized Coastdown
Curve sourced from EPA emissions certification result report data. The road loads and
inertia for different vehicles can be found in the EPA Certified Vehicle Test Results
Reports published for each model year.6 These reports include the road load curves
used during emissions certification on the chassis dynamometer; these road loads are
also relevant to the setup of a brake dynamometer. To determine a general coastdown
curve for use in this work, ERG averaged the EPA-published target coastdown curves
for the 6 vehicles chosen for testing on the test track. These 6 vehicles covered the
range of light-duty vehicle types available and can therefore be used to represent an
average or reasonably representative overall vehicle coastdown rate. The derivation of
the generalized coastdown curve is described further in Appendix C, which also
includes the target road load coefficients for the 6 test vehicles. Note also that specific
vehicle coastdown coefficients are an input to brake dynamometer testing. The
coastdown coefficients for the vehicle being simulated are entered into the
dynamometer control at the start of a test, and these coefficients govern the system's
application of brakes over the test cycle using the same type of calculation as was used
to determine braking events in the Caltrans set.

ERG then assigned driving modes to the smoothed Caltrans survey data. There were
only two important driving modes for this analysis, braking and non-braking. The braking
driving mode was assigned any time the vehicle's deceleration rate exceeded that
which would be experienced while following the generalized coastdown curve found
previously. All remaining times in which the vehicle was not decelerating more rapidly
than that coastdown curve were assigned as non-braking. Once the vehicle came to a
stop, it was no longer considered to be braking (even through the brakes may have still
been applied) because there was no further sliding at the brake friction interface and
therefore no appreciable brake heating or PM emission taking place.

Further classifications of the survey data were applied for use in the overall cycle
building process. The survey data was divided into Microtrips, defined as the period
starting the moment a vehicle comes to a stop, through an idle period, acceleration,
cruise, deceleration and ending at the next moment that vehicle comes to a stop. Each
microtrip contains one or more Braking Events, defined as the moment the brakes are
applied to the moment that either the brakes are released or the vehicle comes to a
stop, whichever occurs first. The ERG new dynamometer cycle was constructed based
on combining a selected series of braking events that actually occurred in the Caltrans
survey data. Information about the microtrip from which each braking event was
extracted was retained to associate the braking event with microtrip average speed and

6 https://www.epa.gov/compliance-and-fuel-economy-data/annual-certification-data-
vehicles-engines-and-equipment

21


-------
distance traveled. This is done because, while some distance is traveled during braking,
emission rate results from braking in EMFAC are given on a per-mile basis and these
must also appropriately consider vehicle distance traveled between braking events for
inventory purposes. An excerpt of a speed trace is presented in Figure 7, with graphical
depictions of the start and end of a microtrip and its braking events.

50
45
40
35

IE" 30

Q.

-o 25


-------
manner until the built-up cycle reached a target overall duration or distance. The key
variables used in this work were:

•	Vehicle speed (distribution)

•	Modeled brake temperature (distribution)

•	Deceleration (distribution)

•	Braking event duration (single value for each event)

Three of the variables allow for a distribution of second-by-second values within any
braking event, but the event duration is just a single value for each event. Each of the
four parameters were weighted equally as there was no reason to justify prioritizing one
of the parameters over the others. The resulting distributions of the parameters show
that all matched the target reasonably well, so it wasn't necessary to weight the
matching of one at the expense of any other (i.e. weighting them differently would not
have significantly affected the outcome).

The vector collinearity method was applied to create a series of braking events whose
distributions of speed, modeled temperature, deceleration, and event durations best
matched those in the entire Caltrans set. It should be noted that this selection method
results in selecting the group of braking events that best match the target, but the
procedure does not result in any particular order for the events (meaning they need to
be ordered later).

The evaluation of the distributions of the four different parameters of braking events on
the millions of braking events present in the Caltrans survey data was extremely
computationally intensive. To keep computation time reasonable, ERG randomly
selected a pool of 1000 Caltrans survey braking events to choose from during the vector
method. The target vector was created from the entire Caltrans dataset; however, the
new cycle could be built up from only selections from the 1000 randomly-selected
microtrips.

Matching the braking events' distribution for temperature created a new challenge for
the cycle development process. During brake dynamometer testing, the test brake
system cools according to how quickly it is rotating over time. However, applying the
brakes is the only way to heat the brake system. Because of this, the method could
select different braking events whose associated temperatures could result in
temperature gaps between the different selected braking events, meaning one or more
of the higher temperature events could not be reached by the temperature increases
available from any of the other chosen events. As a result, ERG developed rules
regarding temperature for selected events to allow for the testing to be as realistic in
temperature as possible while also reaching all specified modeled temperatures.
To eliminate temperature gaps, in which one or more selected braking events had an
initial temperature that couldn't be reached based on the heating of any of the others,
ERG implemented rules regarding temperature selection. To be eligible for selection by
the vector collinearity method, brake events had to have an initial temperature less than
the maximum ending temperature of any of the events that had been previously
selected. For the initial selection, any event with a starting temperature of less than 2°C

23


-------
above ambient was eligible. While necessary to eliminate temperature gaps in which the
highest temperatures could not be reached, this rule created a downward bias in
temperature, as throughout the early stages of event selections only relatively low-
temperature events were eligible for selection. To compensate for this, the selection
pool of 1000 microtrips was adjusted to include 800 events selected at random, and 200
events selected at random from only those events with initial temperatures between
140°C and 190°C. As a result, ERG was able to develop a test cycle that could have
completely reachable temperatures on the dynamometer as well as be representative of
the distribution of modeled temperatures reached during on-road driving. The
representativeness of the actual temperatures can be verified in histograms of operating
temperature of the cycle and the target (included in the following sections).

After the microtrips were selected, they next needed to be ordered. Because of the rule
that microtrips could only be selected if their initial temperature was less than the
maximum temperature previously reached, they could, by definition, be ordered to
increase without gaps (but there may be only one order that could reach the maximum
without gaps). Braking events were ordered to achieve the maximum temperature
relatively early in the cycle; those events that were needed to reach high temperatures
were used to move rapidly up to the maximum. Otherwise, they could end up being
"wasted" or lost by being followed by a cooler brake event leaving a temperature gap
later in the cycle. So, the resulting trend of this process is a prompt rise to the maximum
temperature followed by an oscillating and slowly decreasing temperature trend for the
rest of the cycle. During cooling, brake events were ordered by always prioritizing the
use of the highest starting temperature event that was accessible from the previous
ending temperature. Sometimes, the time required by the dynamometer for speed
changes after the point of highest temperature did result in a small number of
temperature gaps. After the maximum-selected temperature had been reached, any
further brake events with gaps were dropped from consideration. Dropping these events
(after the maximum selected temperature had been reached) did not affect the resulting
temperature distribution significantly. This is because there were generally few gaps at
this point of ordering and they generally occurred across the entire temperature range
such that their removal did not skew the resulting temperature distribution appreciably.

After each braking event was selected and ordered, the different events needed to be
connected by a continuous speed trace that could be followed on the dynamometer.
ERG added "engineered" segments between each braking event to allow the
dynamometer speed and brake temperature to arrive at the initial speed and
temperature of the next braking event. They included ramping to the next event's speed
as well as allowing time to pass at constant speed if further cooling was necessary to
match the starting temperature associated with the next selected event. Cooling was
assumed to take place as a function of speed (simulated by dyno RPM) according to the
temperature model. The dynamometer positive acceleration ramp rate was limited to a
maximum of 8 kph/s to stay within typical brake dynamometer capability. Some brake
events had initial speeds less than the ending speed of the previous brake event. The
negative deceleration level was specified at -3 kph/s. If a large amount of cooling was
needed to get to the next event's temperature, the speed was kept at the higher of
either the previous event's end speed or the next event's initial speed to maximize the

24


-------
cooling rate. For those segments in which the speed was held at the previous event end
speed, deceleration took place at -3 kph/s near the end of the segment in order to arrive
at the next event speed at the correct target temperature.

After the cycle was completely selected, ordered, and the engineered segments added,
it was taken from that point as only a speed trace. The modeled temperatures are no
longer a part of the trace, they were used in its creation only. The speed trace is the
cycle and is independent of temperature now that it has been created. This means that
testing of different brake assemblies will follow the same speed trace but will be allowed
to run at completely different temperatures depending on the vehicle/brake assembly
characteristics and modeled vehicle mass.

While some vehicle distance is traveled during the braking events selected in the cycle,
each braking event actually represents a much greater distance traveled in the Caltrans
set. Because the braking events were extracted from microtrips, the distance of the
source microtrip must be accounted for to generate a representative on-road emission
rate in g/mi. To do this, information on the microtrip source of each braking event was
retained to associate braking events with a total distance traveled (which is greater than
the distance traveled just during braking). For microtrips that contained multiple braking
events, the braking energy for each event was used to proportionally assign the total
microtrip distance traveled to that represented by each braking event. Dynamometer
distance is also traveled during the engineered segments of the cycle; however, these
segments only exist to set the dynamometer speed and the amount of brake cooling.
The dynamometer operation during the engineered segments has no actual basis in
vehicle distance traveled, which also contributes to why the dynamometer cycle
distance traveled is not useful for g/mi calculations. The ERG cycle will specifically
advise a represented distance to be used in all g/mi calculations and it will not be the
same as the integrated distance traveled by the rotational assembly on the
dynamometer. Figure 8 illustrates an example microtrip from which one braking event
was selected and inserted into the test cycle. The speed trace of the braking event is
identical, however the rest of the microtrip is excluded.

25


-------
80

Q.

¦a 40

(u

QJ

a.

00

50

70

20

60

30

10

0

0

20

40

60

80

100

120

Next
microtrip
begins

Time (s)

	Caltrans Speed Trace 	Test Cycle

Figure 8. Illustration of a braking event, extracted from the original microtrip speed trace
(blue), and inserted into the test cycle between engineered segments (red)

In addition to representativeness of California driving, the test cycle also needed to have
the ability to resolve emission factors by speed. The ERG cycle is divided into three
speed segments representing different average speed ranges. ERG selected three
speed ranges based on trends in the average speed of all microtrips in the Caltrans
data. Figure 9 depicts a histogram of the number of microtrips with each average speed.
The histogram represents all microtrips in the Caltrans set, weighted by the duration of
each microtrip (i.e. longer microtrips have higher weighting to reflect the greater
duration and distance of driving that they represent). The average speed ranges for the
segments of the new cycle were selected as 0-21 kph, 21 -69 kph, and 69 kph and
above. ERG used the cycle building approach to develop a brake dynamometer cycle
for each average speed range. Each phase would contain braking events taken only
from microtrips falling within its respective microtrip average speed range.

26


-------
1.6%

0 10 20 30 40 50 60 70 80 90 100 110 120 130

Speed Range, kph

Figure 9. The percentage of the number of microtrips within each bin of average speed

ERG used the vector method described previously to construct each of the three speed
segment. For a given speed segment, the selection pool of 1000 brake events were
chosen from only those microtrips with an average speed within the segment's speed
range. The target vector from Caltrans used for each speed segment was taken to be
that made up of the braking events from all microtrips with an average speed in that
range. The overall cycle is made up of the three speed segments run in succession.
The representative distances of each was also taken to represent the relative distances
traveled in the Caltrans survey by all microtrips falling within each speed range. This
results in the overall per-distance emission rate being representative of overall brake
emission rates in California. Because the speed segments are specified to run in
succession, the speed-based emission factors could be based on either continuous or
batch (i.e. filter-based) PM measurements.

ERG used a specific method to determine the overall duration/represented distance of
the speed segments as well as how the distances were apportioned within each of the
three speed segments. The following procedure was followed:

1.	Use the vector collinearity method to build up 200 braking events for each of the
three microtrip speed ranges

2.	Determine the total distance represented by each range's 200 events

3.	Determine the total distance traveled in the Caltrans dataset by microtrips in
each of the 3 speed ranges and find the percentage for each

4.	Correct the two speed segments that over-represent distance by removing trips
at random, but do not remove trips that would cause a temperature gap in the
cycle

27


-------
After determining the relative distances of each cycle, ERG inserted engineered
portions between each speed cycle to allow the brakes time to cool to a near-ambient
starting temperature. The percentages of distance traveled in the overall Caltrans
dataset of microtrips in each of the three speed ranges are presented in Table 11 along
with the resulting represented distances and times for the new cycle.

Table 11. The percentage of Caltrans-survey and new test cycle total distance traveled by

microtrips within each average speed range

Microtrip
Avg. Speed
Range

Percent of Total
Caltrans Distance
traveled

New Cycle
Represented
Distance (km)

New Cycle
Represented
Distance %

New Cycle
Duration (s)

0-21 kph

3.96 %

6.163

4.7 %

2,741

21 -69 kph

38.34 %

47.309

36.0 %

8,339

69+ kph

57.7 %

77.775

59.3 %

3,853

Total

100 %

131.247

100 %

14,933

Figure 10 presents a speed trace of the new ERG-developed braking cycle for this
program.

140

120
100

J- 80

"O


-------
of the 10 trips to reflect the cooling soaks to 30°C between trips as specified in the
WLTP-Brake test procedure.

In this report, findings and comparisons for the EMFAC UC/SCCs are included,
however these cycles may not be able to be repeatably run on the brake dynamometer.
These cycles have a range of brake event durations that includes events down to 1 s
and 2s in duration. The LINK brake dynamometer cannot repeatably test braking events
of this short of duration. This cycle could not be chosen for this project for this reason;
however, the EMFAC cycles are left in the analyses in this section. Correspondingly, the
ERG cycle development algorithm described previously was modified to only select
braking events of 3s duration or longer to meet the dynamometer requirements for
repeatable operation. Speed traces of the WLTP-Brake and the concatenated UC/SCCs
are presented for reference in Figure 11.

150

WLTP-Brake

0	2000 4000 6000 8000 10000 12000 14000 16000 18000

Time (s)

Figure 11. The Speed traces for the WTLP-Brake and the concatenated UC/SCC

Comparisons across the three cycles are presented in two ways. First, Table 12
describes some of the relevant properties of the three cycles. Later in this section,
distributions of various parameters of interest are presented for the three cycles.

29


-------
Table 12. Relevant parameters of the three candidate brake test cycles



Duration

(s)

Number of
Braking
Events

Distance
(km)

Brake Events/
Distance (#/km)

Average
Speed
(kph)

Max.
Speed
(kph)

ERG
Vector
Method
(Overall)

14,933

347

131

2.65

50.3V30.32

123.9

0-21 kph
speed
segment

2,741

65

6.16

10.55

21.7V8.02

70.9

21-69
kph speed
segments

8,339

198

47.31

4.19

45.7V19.42

121.8

69+ kph
speed
segment

3,853

84

77.76

1.08

80.6V69.82

123.9

WLTP-
Brake

15,826

303

192

1.58

43.7

132.5

EMFAC
UC/SCCs

16,952

1,015

272

3.73

57.7

129.8

1	- Calculated based on the actual number of dynamometer revolutions

2	- Indicates the represented distance for inventory (accounting for brake cooling and
elimination of unnecessary cruises - this is the value relevant for EMFAC)

It is important to note some specifics regarding the values in the table as follows:

•	Duration: Duration is a count of the number of cycle-specified seconds only. The
WLTP also includes cooling between most of the 10 Trips and so will take longer
to complete. The time listed for the ERG method includes all required cooling.

•	Distance: As described previously, the overall ERG cycle spins the
dynamometer further than 131 km due to the engineered segments. The 131 km
listed specifically describes the distance represented by those events for the
purposes of g/mi calculations. The distance listed for the other two cycles
describes the distance traveled by the dynamometer. The dynamometer spins a
longer distance over the ERG cycle to allow the necessary amount of cooling.

•	Average Speed: Because the ERG cycle spins the dynamometer farther than
the on-road distance represented, the average speed is presented two different
ways. For the ERG cycle and its constituent cycles, average speeds denoted
with a "1" indicate the average speed of the rotation on the dynamometer
including the cooling intervals. The average speeds denoted with a "2" indicate
the average speed based on the distance represented by the cycle for the
purposes of inventory modeling.

The following section includes distributions of various parameters of interest for the 3
cycles. The Caltrans survey data is shown as the target for representation of real
California driving. In all plots, the distributions reflect the parameters from only the time

30


-------
during brake events; the values during periods of time with accelerations and cruises
are not included in the distributions. Braking events are defined by decelerations that
exceed the generalized coastdown curve used in this work. Temperature distributions
are estimated using the temperature model developed for the Camry test vehicle.
Additional detail is presented in Appendix E, which contains similar distributions further
broken down by each of the three speed segments that make up the new cycle.

The distributions of brake event durations are presented in Figure 12. Note that the
ERG New Cycle does not have any brake events shorter than 3 seconds. This is
intentional due to a limitation of the brake dynamometer used for testing and would
otherwise be likely to result in the UC/SCCs not being repeatably testable as they have
many 1s and 2s events. The number of events in the 3-second bin of the ERG cycle are
higher as a result because the vector collinearity method was targeting the Caltrans
distribution which contains a large number of 1s and 2s events.

Figure 12. Distribution of brake event durations for the candidate cycles and the Caltrans

data

The distributions of speeds encountered during braking events are presented in Figure
13. The distribution of (negative) acceleration rates during braking is presented in
Figure 14.

31


-------
Speed Bin (kph)

Figure 13. Distribution of vehicle speeds for the candidate cycles and the Caltrans data

45

40

35

- 30

T3
£

O
u
CD
00
CuO
c

25

20

oo 15

£
CD
U
&_
CD
Q.

10

ERG New Cycle
UC/SCCs
WLTP-Brake
Caltrans Survey

-9 -10 -11 -12

-1 -2 -3 -4 -5 -6 -7 -8

Acceleration (kph/s)

Figure 14. Distribution of braking event (negative) acceleration rates for the candidate

cycles and the Caltrans data

The distributions of modeled brake temperatures for each cycle are presented in Figure
15. The same temperature model was used to estimate temperatures for the three
candidate cycles as well as the Caltrans survey data.

32


-------
30

_c

CO

_c

tn
~o
c
o

(J
(D
00
W
_C

ro

i_

CO

25

20

15

o 10
+-»
c

(D
(J

QJ
Q_



A ERG New Cycle



^^UC/SCCs



WLTP-Brake



M Caltrans Survey

	1 9 « ¦	¦

20 40 60 80 100 120 140 160 180 200 220 240 260
Temperature Above Ambient/Wheel Well (C)

Figure 15. Distribution of modeled brake temperatures for the candidate cycles and the

Caltrans data

The distribution of relative power per brake event is presented in Figure 16. In this plot,
relative power is defined as the change in speed squared (accounting for the
coastdown-rate of energy loss) divided by the duration for the entire brake event in
seconds, with units of kph2/s. Relative power was not one of the parameters used
during cycle building, but is presented here for completeness.

33


-------
20

18
16

a 14

c

+-> -Lz-
c

(L)

w 10

M—

O

E 8

(D
U

£ 6
4
2
0

0

Figure 16. Distribution of brake event relative power for the candidate cycles and the

Caltrans data

Each candidate cycle has advantages and disadvantages. The following summaries for
each cycle describe the key considerations for the evaluation and final selection of the
cycle used during dynamometer testing.

ERG Vector Method Cycle. The distributions on the parameters of interest for this
cycle match very well with the Caltrans survey results because it was designed to result
in a match on the four main parameters. It allows directly for determination of speed-
based emission factors using its three constituent speed segments. However, it is an
unproven cycle with no reputation across the research community. Some members of
the community may not agree with the approach to have a represented distance for g/mi
calculations (for the purposes of inventory modeling) that is separate from the actual
dynamometer distance traveled. ERG planned to have a separately-specified
represented distance from the start of this project as it ensures that g/mi calculations
are appropriate for representing California driving.

WLTP-Brake. The WLTP-Brake is an engineered cycle designed for use on a brake
dynamometer. As a result, it contains brake event durations appropriate for
dynamometer testing. However, of the three candidate cycles, its distributions on the
parameters of interest have the least similarity to the Caltrans survey results. Also, it
was not designed to have the distance traveled over the cycle be on the same basis as
PM emissions from California driving, resulting in potential error in g/mi calculations.

2	4	6	8 10 12 14 16 18 20

Brake Event Specific Power (kW/ton)

34


-------
The WLTP-Brake cycle also does not necessarily lend itself to the measurement of
speed-based emission factors. With this cycle, speed-based factors would only have
been able to be generated from extracting different brake event segments from the
continuous PM measurement traces. An advantage of the WLTP-Brake cycle is
commonality and comparability with European brake testing which will heavily utilize the
WLTP-Brake.

UC/SCCs. The analysis of braking events showed that the EMFAC cycles matched the
Caltrans survey results reasonably well for the parameters of interest. The key limiting
factor of these cycles will be that they contain a significant number of short duration (1 -
2s) brake events that cannot be repeatably simulated on the brake dynamometer. It is
not possible to remove these short duration events without significantly reworking these
cycles. For this reason, it would have been inadvisable to utilize these cycles for this
program despite EMFAC being designed around them such that test results would be
readily adapted into EMFAC emission factors including those that are speed based.

Cycle Selection. The three cycles were evaluated for how well they represented the
driving logged during the Caltrans survey. The EMFAC UC and SCCs represented the
Caltrans data fairly well and could be readily adapted to speed-based emission factors
because the speed correction cycles already exist. However, the EMFAC cycle
consisted of many 1s and 2s duration brake events, which cannot be repeatably
simulated on the brake dynamometer. For this reason, the UC/SCCs could not be used
in this work. The WLTP-Brake cycle is designed specifically for brake dynamometer
testing but was not specifically designed to represent California (or US) driving as it was
developed from data from multiple nations. The WLTP-Brake is also not designed to
directly determine speed-based emission factors. The ERG vector-based method
generated a cycle that both represented different speed ranges of operation and was
very similar to the Caltrans travel survey data on the four important parameters of
interest, so CARB and ERG staff agreed that ERG's newly-developed cycle would be
used during the dynamometer testing in this work. From this point forward, it will be
known as the California Brake Dynamometer Cycle (CBDC) for light duty vehicles.

Brake Burnish Cycle. Newly installed brake friction materials go through a process of
"bedding in," in which the friction couple equilibrates and a layer of pad material
becomes adhered to the disc or drum. Particulate emission rates may not be stable
during this time. Also, brand new materials may have a protective coating to prevent
oxidation prior to installation. After installation, this coating is worn off in the early stages
of use but may result in particulate emissions that are not representative of emissions
during the remaining life of the components. For this reason, a burnish procedure was
performed after the installation of new components but prior to testing.

ERG developed a new brake burnish cycle with the goal of being as short as possible
(to allow for a 24 hour test turnaround) while still resulting in a stable friction couple at
completion. The PMP was developing a standardized burnish cycle concurrently with
this project, and industry experts participating in the process indicated that a minimum
of 5 repeats of the WLTP-Brake would be necessary for a stable burnish. ERG used this

35


-------
as the source of the development of a new burnish cycle for this work. ERG developed
a new burnishing cycle by using the following method:

•	Calculate the total braking energy in 5 WLTP-Brake cycles

•	Select a relatively high energy segment of the newly developed CBDC brake
cycle that has a similar start and end temperature (starting at 707s, proceeding to
1740s)

•	The burnish cycle starts from the beginning of the CBDC and runs through the
end of the selected high-energy segment. Then, that segment is appended
repeatedly until the total braking energy of 5-WLTP-Brake cycles is reached

•	An engineered high-speed cruise is added to cool the brake assembly to near
ambient temperature

•	Finally, to cool down and equilibrate the friction couple so that it doesn't end
during high intensity operation, a single, low speed segment is appended to the
end of the burnish (the complete 0-21 kph speed segment).

The resulting burnish cycle has a duration of approximately 11 hrs 30 mins (as
compared to approximately 30 hrs for 5 WLTP-Brake cycles when including the
specified cooling between trips). A speed and temperature trace of the resulting burnish
cycle (used for all tests in this program) is presented in Figure 17. The repetitive nature
of the cycle facilitates the determination of whether PM emissions reach steady state.



200



180

u

o

160

(D





140

"to



cD

Q.

120

E



(D



I—

100

~o



c



CU

80





Q.





60

~o




-------
Test Matrix

After the test cycle was developed, ERG and LINK then began the development of a
test matrix that would describe the various tests that would be conducted during the
dynamometer testing phase of the project. The first step in the development of a test
matrix was to consider the parameters of interest to be tested. This section lists those
parameters, the different options or values to test within each parameter, and their
relative importance. The parameters that were considered are:

Vehicle make and model. The initial parameter that was considered was the vehicle
make and model of each brake assembly. The selection of the test vehicles was
described previously in the Representative Test Vehicle and Friction Material Selection
section, in which the selection of the following 6 vehicles was documented:

•	2011 Toyota Camry LE

•	2013 Honda Civic LX

•	2013 Toyota Sienna LE

•	2015 Ford F-150 Supercrew

•	2016 Toyota Prius Two Eco

•	2016 Nissan Rogue S

These vehicles are all equipped with front and rear disc brake systems except for the
Honda Civic, which utilizes front brake discs and rear brake drums. The Toyota Prius is
a hybrid vehicle and is equipped with regenerative braking. The operation of the
regenerative braking system was simulated by the brake dynamometer for this vehicle's
assemblies only. These are the same vehicle make, model, and model year as those
that were track tested during the development of the brake temperature model.

Vehicle front/rear brake assembly. The brake dynamometer tests the brake system
components of a single vehicle wheel/hub assembly at a time. Because of weight
distribution and weight transfer during braking, front and rear brake assemblies are
designed differently. Front brakes typically receive a larger proportion of braking energy
than do rear systems, and for this reason, their design is different from rear assemblies
and the components typically have a greater mass and greater surface area for heat
rejection. However, because they are generally lighter and less vented, rear brakes tend
to operate within approximately the same temperature as front brakes. To estimate total
PM emissions from braking, this project involved testing both front and rear assemblies
to better understand the relative and total emission rates from these components.

Brake pad material. Modern brake pad materials fall into the categories of low metallic
(LM) or NAO. A given vehicle may be fitted with different brake pad materials at different
points in its life as aftermarket options may differ from original equipment components.
Not all brake assemblies have the same brake lining materials available. Thus, brake
lining materials were selected based on the individual vehicle assemblies tested. At a
minimum, for each test vehicle the OES friction material was chosen for one of the pad
material options (which was NAO for all models). Testing of each vehicle then included

37


-------
one or two aftermarket options, either one or both of a common NAO aftermarket pad
and/or a common metallic aftermarket pad depending on availability.

The test plan included matching the rotor selection (or drums in the case of the Civic) to
the selected pad type where possible. OES pads were tested with OES rotors. For
aftermarket pads, LINK used business intelligence and existing brake supplier
relationships to determine the most likely/representative aftermarket rotor to be matched
with each pad material during a real-world aftermarket purchase. This allowed for the
friction couple between the pad and rotor to be more likely to be representative of real-
world operation.

The selection of brake friction materials also included consideration of the copper
content in each material formulation. California legislation in SB 346 specifies a phase-
out of copper (as well as selected other metals7) in commercial brake pad formulations
due to environmental harm associated with these compounds being carried into
waterways by roadway runoff. Other states have also adopted the legislation and it is
likely that, to simplify supply chains, eventually most or all brake manufacturers will
produce only friction materials that meet SB 346 requirements nationwide. Under SB
346, copper must be reduced to 5% or less of total material content by weight by 2021,
and to 0.5% or less by 2025. Friction material manufactured prior to the January 1, 2021
with <5.0% copper by weight may be sold until January 1, 2031. The Brake
Manufacturer's Council has developed the LeafMark letter labeling design to indicate to
consumers which of the above thresholds that a given pad meets. The LeafMark letter
labels and their respective thresholds are defined as follows:

•	A - formulation contains more than 5% copper by weight

•	B - formulation contains between 0.5% and 5% copper by weight

•	N - formulation contains less than 0.5% copper by weight.

In this work, LINK had limited control over which formulations were associated with the
different brake assemblies that were tested. OES materials were tested in whichever
formulation was used by the OES component. For aftermarket, in which a project goal is
to test high-selling, representative components, one of two options existed:

•	Only one LeafMark is associated with the best or second-best selling component.
In this case, that LeafMark was selected.

•	The LeafMark is not yet specified for a given aftermarket component. This
situation exists for those components for which current inventory exists under
multiple LeafMarks. In this case, the desired component was ordered and
whichever LeafMark was delivered was tested (as there is no way to order these
components based on the LeafMark).

Based on LINK business intelligence, there were no instances in which the top two
selling components each had a different, specified LeafMark. So, there was no

7 SB 346 also limits the presence of cadmium, chromium salts, lead, mercury and
asbestiform fibers in brake friction materials sold in California during or after 2014.
(https://leginfo.legislature.ca. gov/faces/billNavClient.xhtml?bill_id=200920100SB346)

38


-------
assembly for which ERG and LINK had to select one LeafMark over another. The two
vehicles for which metallic pads will be tested are most likely to carry the A label. The
test matrix contains the LeafMark that was planned for testing if it was known for a given
friction material.

Simulated vehicle weight/load. To decelerate at a given rate, more braking energy is
required by a more heavily laden vehicle. To better understand the effect of this,
additional tests of some vehicles will be performed at a higher weight than the normal
test weight for each vehicle. The normal test weight simulated for each vehicle will be
calculated as given for passenger cars in 40 CFR 86.129-00. The nominal equivalent
test weight (ETW) for each vehicle will be the curb weight plus 300 lbs.

ERG selected the three vehicles with the largest cargo-carrying capacity for additional
testing representing higher-laden weight operation. For these vehicles a heavily-laden
weight (HLW) was defined as an additional two thirds of the capacity added between
the curb weight and the gross vehicle weight (GVW). HLW was calculated as:

2

HLW = Curb Weight + — ¦ (GVW — Curb Weight)

Test cycle. For the majority of tests, ERG used the cycle that was developed and
selected for use in this program. This cycle was developed based on actual in-use
vehicle survey data from California. The new CBDC test cycle has a duration of
approximately 4.3 hours and represents a total of about 81.55 miles of driving. It
contains 347 braking events. The cycle consists of three different segments
representing slow, medium, and high average speed driving. The three segments are
proportioned in distance similarly to the actual in-use California distances driven in
those average speed ranges based on the Caltrans survey data.

For comparison with outside brake PM research, a small number of tests with a subset
of the test vehicles was conducted using the World Harmonized Brake Dynamometer
Cycle (WLTP-Brake). The WLTP-Brake cycle has a duration of approximately 4.4 hours
(plus cooling intervals) and represents approximately 119.3 miles of driving. It is divided
into 10 trips containing a total of 303 braking events.

Test replicates and references. Conducting replicates helps determine the
repeatability of the testing procedure. The selection of the quantity of replicates was
optimized in terms of the number of replicates given that the total number of tests was
limited. The number of replicates was decided prior to testing and was not adjusted
during testing. Given that testing had not yet been conducted at the time of test matrix
development, ERG had limited data from which to estimate the number of replicate tests
required to determine statistical significance. Based on the total number of different
brake materials and vehicles in the study, ERG decided to do two replicate tests for
almost all test matrix combinations. Replicate tests were conducted using new
components of the same specification (or part number) for each test. So, test
repeatability could be influenced by both the test setup and the differences from using
separate sets of friction materials manufactured to the same specification.

39


-------
Additionally, one vehicle/assembly/material combination was selected to serve as a
reference test. The single reference combination was repeated in the test matrix at
regular intervals throughout the project. The reference test can help to better
understand and track any measurement drift in the laboratory setup over time. ERG
selected the F-150 front brake with OES pads (which are NAO) to serve as the
reference. This was selected due to this vehicle being the most popular of the six test
vehicles for brake component benchmarking based on LINK market research and
business intelligence.

Additional considerations. In addition to the parameters that define the dimensions of
the matrix, there were some additional considerations and line items in the matrix that
were added to enhance the test plan. These include the types of filter blank
measurements to be conducted, a planned testing pause, and additional notes
regarding expected EPA and CARB chemical analyses to be conducted after the post-
test weighing processes.

LINK conducted two types of filter blanks during the program, zero blanks and
tunnel/background blanks. Zero blanks are intended to identify the level of
contamination that may occur during the handling of the filter between weighings but
outside of the actual test cycle. Zero blanks are not specifically listed in the matrix but
were performed approximately once every two weeks during the testing program.

Zero blanks were performed as follows:

1.	Pre-weigh filter

2.	Transport filter from weigh room to the test site and install normally

3.	Pause and do not turn on sample pump or expose filter to any sample flow

4.	Remove filter, transport back to weigh room and allow to stabilize

5.	Post-weigh filter

Tunnel/background blanks required a more rigorous procedure and attempt to quantify
not only the handling effects but also any contamination that exists within the complete
sampling tunnel. Tunnel/background blanks are listed in the test matrix and were
conducted as follows:

1.	Pre-weigh filter

2.	Transport filter from weigh room to the test site and install normally. Operate
and/or log data with all measurement equipment.

3.	Install the F-150 rotor, fixture, and caliper, but do not install brake pads. Run the
tunnel dilution air pump as well as the relevant sample pump and cooling airflow
for the test duration, however do not open the hydraulic brake line valve. Allow
the installed rotor to rotate and follow the speed trace (as closely as possible
given that no braking will take place). Run the cooling airflow at the flow rate
used for the front assembly of the F-150 during tunnel blank measurements. The
brake pads will not be present to eliminate any PM that could be generated from
them lightly rubbing on the brake rotor.

4.	Remove filter, transport back to weigh room and allow to stabilize

5.	Post-weigh filter

40


-------
As described in the original proposal, ERG and LINK planned a testing pause early in
the program. This allowed time for an initial data review to determine if any problems
exist with sampling plan or execution of testing. The pause allowed for specifically
delineated time for any necessary changes to be made prior to conducting the bulk of
the testing plan. The test matrix included the expected point for the testing pause.

In addition to the work conducted by LINK and ERG for this project, both CARB and
EPA offered to contribute work in performing chemical speciation of some samples. For
planning purposes, the labs that planned to perform different filter analyses are also
presented in the matrix. Primarily this included the two different workflows to take place
for the Teflon filters. One subset of them originated at CARB, were shipped to LINK for
weighing and testing, and then were shipped back to CARB for XRF and/or ICP-MS
analysis. A different subset originated at EPA's National Vehicle, Fuel, and Emissions
Laboratory (NVFEL) and were used in a filter-weighing "round-robin" in which the pre-
test and post-test weighings took place both at LINK and at EPA NVFEL in series. The
test matrix includes the tests that were assigned for each of those two workflows.

The different parameters to be tested for each vehicle are presented in the following
table. The total number of planned tests was 85.

Table 13. Brake Dynamometer Test Matrix Parameter Summary

Test
Vehicle

Front/
Rear

Pad
Material

Wheel
load

#

Replicates

Reference
repeats

Test
Cycle

Total
Tests

Camry

Front
Rear

OES
After-Met.
After-NAO

ETW

2 each

NA

CBDC,
WLTP-
Brake

14

Civic

Front
Rear

OES
After-NAO

ETW

2 each

NA

CBDC

8

F-150

Front
Rear

OES
After-Met.
After-NAO

ETW
HLW

2 each

5 of a
single
condition

CBDC
WLTP-
Brake

25

Sienna

Front
Rear

OES
After-NAO

ETW
HLW

2 each

NA

CBDC

16

Prius

Front
Rear

OES
After-NAO

ETW

2 each

NA

CBDC

8

Rogue

Front
Rear

OES
After-NAO

ETW
HLW

2 each

NA

CBDC

12

Tunnel B

lanks

2 total



The list of parameter options presented previously must then be ordered based on the
relative quantities to be tested for each parameter. It was preferable to test as many
different assemblies as possible prior to the pause, but it was also preferable to test at
least one replicate prior to the pause to get an initial indication of the level of variability
between two tests.

41


-------
Where possible, the order was then randomized and mixed to reduce the likelihood of
external factors biasing the measurements. So, most of the different assemblies were
tested prior to the pause, but after the pause the testing order was assigned in groups in
random order. Tests were conducted in blocks for each vehicle's front or rear brake
assembly to reduce the turnaround time between testing where possible. The first
replicates for each pad material were grouped together, but the order of these groups
was then selected at random. For example, the "A" replicates of a given vehicle's front
or rear assemblies for the OES, aftermarket NAO, and the aftermarket metallic friction
materials make up a group to all be tested consecutively. These groups were then
ordered, generally at random, to minimize the effects of any time-based biases that
could be encountered during the test program. One exception to the complete
randomization was that a few changes in order were made in order to reduce the
turnaround time from switching between different assemblies. The LINK dynamometer
system uses combinations of large and small inertia discs to simulate vehicle inertia.
Where possible, assemblies using the same number of large inertia discs were grouped
together to reduce the longer amount of time required to change these large discs.
However, "A" and "B" replicates for a given vehicle were still kept separate in the matrix,
and the vehicle order was not "sorted" by inertia (i.e. vehicles may be grouped together
with a common number of required large discs, but the vehicles weren't ordered by
ascending or descending number of discs).

The reference tests (of the F-150) were interspersed regularly throughout the testing
program. One tunnel blank was conducted at the start of the program, and the other
was conducted around two thirds of the way through the program. The complete test
matrix, along with the dates of each test, is presented in Appendix F.

Procurement of Test Components

LINK began procuring brake parts once the test matrix was finalized. LINK procured
OES components from local dealerships. Aftermarket components were prioritized
based on sales levels and availability for non-asbestos organic (NAO) and low-metallic
(LM) friction materials. The aftermarket components included several products acquired
from Bosch, Wagner, Autozone, and RockAuto.

LINK Test Laboratory Setup

The LINK test laboratory that was used for this project is built around a constant volume
(i.e. air velocity) sampling system that operates in a closed-airflow circuit. The cooling
and airflow rates are fixed for a given test. The airflow through the test enclosure
provides the sampling medium for emitted PM as well as the cooling flow for the brake
assembly. The sampling airflow is filtered using HEPA filters after passing through the
climate control unit and before entering the brake system enclosure. Figure 18 contains
a schematic of the LINK laboratory layout that was used for this test program.

42


-------
air temp and

humidity

sensor

HE PA 13 filters

*	Isokinetic & electropolished

CVS tunnel

TSI Instrument cluster 6 nm - 18 urn

Figure 18. Schematic of LINK Laboratory Setup

Climate control unit
20 °C & 50% RH

air speed sensor

air temperature sensor

Incoming air

Electropolished
brake test enclosure

The cooling air was controlled to stable temperature (within ± 5°C of the setpoint by
vehicle) and held to a relative humidity of 50 ± 10%; this helps ensure a stable set of
conditions when the particles enter the sampling train (between the aspiration position
and the point of sampling). The LINK system allows for adjusting the airflow rate prior to
each test to reflect the cooling rates established during the project for each brake
assembly.

The airflow circulating layout involves the use of round ducting in stainless steel, with
internal electropolished finish, with minimal constrictions and at least 8 diameters
without disturbances between the brake assembly enclosure and the point of sampling.
The sample duct is oriented horizontally, and samples are taken at the point of entry of
flow into the 90° elbow downstream of the brake enclosure. Sampling is performed
using four separate sampling lines, each originating from an isokinetic sample nozzle
arranged in parallel at the upstream end of the sampling elbow. The sampling lines and
instrumentation are described later in this document.

From the brake assembly enclosure to the sampling instrumentation, the layout is
designed to minimize aerodynamic losses with minimal bends and constrictions. These
design characteristics are intended to minimize turbophoretic losses, gravitational
deposition, diffusion, and aspiration at the nozzle. Sampling is performed isokinetically
(0.95 to 1.15 isokinetic according to ISO 9096) to avoid skewing the particle size
distribution data. A transport time of less than 5 seconds from brake assembly to
instrument is specified (with a target transport time of 2 seconds) to minimize potential
changes in size distribution due to coagulation. In addition, the short transport time will
allow the particle size distribution to be closer to the actual distribution as-generated by

43


-------
the friction surface. A photograph of a brake rotor installed in the test enclosure is
presented in Figure 19.

Figure 19. A brake rotor installed in the LINK test enclosure.

CVS Loop Cooling Airflow Setting

The air flow rate through the sampling chamber was set at different rates for the front
and rear assemblies of each different test vehicle. The air flow rate was set in an
attempt to best match the cooling that takes place in real-world operation, given that the
flow rate must stay constant during a test to allow for constant volume PM sampling.
During track testing, LINK logged temperature data for the front and rear assemblies of
each of the six test vehicles when operating over the WLTP-Brake cycle driven on a test
track. This data was used to determine the dyno cooling air flowrate for testing over the
new CBDC.

Because all test vehicles were operated over the WLTP-Brake cycle on the test track
with temperatures logged, this data was used as a source for flowrate setting. ERG and
LINK selected a subset of the 10 WLTP-Brake trips that was most representative of the
characteristics of the CBDC, and these were used to set the flowrate to best match the
track temperature over the same subset when operating on the dynamometer. To select
the WLTP trip or combination of trips to use for setting flow rate, ERG analyzed each
WLTP trip in terms of distributions of the same parameters used during test cycle
selection: deceleration rates, speed, brake event duration, temperature, and braking
energy. From this, ERG determined that WLTP trips 1,2,5, and 10 were most similar in
the above five parameters to the CBDC test cycle. As an example of the type of
distribution that ERG reviewed, Figure 20 presents a cumulative distribution of modeled
brake temperature for the Camry over the braking events of the CBDC and the 10
different trips of the WLTP-Brake.

44


-------
0

0	50	100	150	200

Temperature (over ambient) Bin (C)

Figure 20. Cumulative distribution of brake temperature over ERG's New CBDC
(Modeled) and the 10 trips of the WLTP-Brake (on track)

ERG then investigated the same distributions for various combinations of the four best-
matching trips (1,2,5, and 10). This was done by comparing the sum of squares
differences between the different trip combinations and the CBDC for the five
parameters of interest. ERG determined that trip 10 was the best fit, and the addition of
any other trip did not necessarily improve the representativeness. For this reason, LINK
used only WLTP-Brake trip 10 for airflow rate setting for each different assembly.

LINK ran Trip 10 on the dynamometer for each brake assembly at three equally spaced
flow rate settings. LINK then fit a curve to the average braking temperature at each flow
rate and calculated the flow rate that best matched the track test temperatures for each
brake assembly. Table 14 presents the flow rate used for all tests of each given
assembly over the CBDC test cycle. These values represent the airflow speeds through
the sampling duct that LINK found would allow the assembly temperatures to best
match the temperatures measured on the test track. The cooling airstream speed was
measured 8 diameters downstream of the sampling elbow, complying with the
requirement defined in EPA Method 1A (the ducting has the same diameter over this
entire length). Temperature plots comparing the measured dynamometer temperatures
to the test track temperatures are presented for the front and rear assemblies of each
vehicle in Appendix G.

45


-------
Table 14. The CVS cooling/sample flow settings for each vehicle/axle combination



Front Axle Flow
Speed (kph)

Rear Axle Flow
Speed (kph)

Camry

7.5

7.5

Civic

7.5

7.5

Sienna

7.5

7.5

F-150

51

7.5

Prius

28

28

Rogue

7.5

7.5

The different sample flow settings do affect the PM residence time between the brake
enclosure and points of mass measurement (either the TSI 100S4 or 47mm filter).
Overall nominal residence time varies from approximately 0.7s to 1,2s depending on the
CVS flow rate.

Dynamometer Operation

The LINK brake dynamometer simulates the rotation and braking functions for a single
brake assembly. The unit can simulate front or rear brakes depending on the brake-
fixture assembly and the inertia. The system uses both an electric servo motor and the
line pressure of the hydraulic brakes to follow a set speed trace over time. The
dynamometer controller balances these two sets of torques based on the programmed
vehicle road load and inertia. The inertia is the simulated wheel load at the tested brake
corner, and the road load describes the force curve representing the drag on the vehicle
across a range of speeds when traveling along a level road. During this work, the EPA-
published coefficients from the annual certification reports for each model and model
year were used to simulate road load for each test vehicle during brake dynamometer
testing.

One significant difference in the operation of the brake dynamometer and a chassis
dynamometer is the need to split the braking force between the front and rear brakes.
Because vehicle weight is transferred forward during braking, the front brakes are
designed to absorb and convert a larger amount of energy than rear brakes. The
Society of Automotive Engineers (SAE) standard J2789, Inertia Calculation for Single-
Ended Inertia-Dynamometer Testing, specifies how this energy split can be simulated.8
Table 15 depicts the standard percentage torque splits for various vehicle categories for
two levels of deceleration and two levels of vehicle loading, gross vehicle weight (i.e.
fully laden) and lightly loaded vehicle weight (LLVW). Where possible, LINK followed
J2789 for the proportioning of brake torque between front and rear assemblies in this
work. All braking in the braking test cycles evaluated in this work consist of braking
events falling within Low deceleration (< 0.65 g-force) as referenced in the table. One
exception to the use of J2789 was made for vehicle types in which, during cooling flow

8 https://www.sae.org/standards/content/j2789_201008/

46


-------
setting, a temperature mismatch was observed between front and rear assemblies for a
given vehicle compared to track test data. For example, if using J2789 resulted in the
front assembly for a given vehicle running hotter than the track data and the rear
assembly running cooler than track data, LINK adjusted the inertia split until the brake
temperatures matched the track trends (this was the case for the F-150).

Table 15. Brake Torque Split Percentages Based on SAE Standard J2789

Percent of brake force done by each axle (X and Y values)



Fixed Proportioning

Vehicle type

Low deceleration < 0.65 g

High deceleration > 0.65 g

GVWR

LLVW

GVWR

LLVW



Front

Rear

Front

Rear

Front

Rear

Front

Rear



X

Y

X

Y

X

Y

X

Y

Passenger car - FWD

78

28

78

28

80

25

80

25

Passenger car - RWD

78

28

78

28

75

30

75

30

Minivan and crossover

78

28

78

28

75

30

75

30

Pick-up trucks

68

38

63

45

80

25

80

25

SUV-RWD

73

33

73

33

75

30

75

30

Appendix C includes the road load coefficients for each model as well as the by-axle
inertia settings programmed into the dynamometer for each vehicle, axle, and test
weight combination.

Simulation of Regenerative Braking

The Toyota Prius test vehicle is equipped with regenerative braking, in which some
amount of braking energy is converted to charge the vehicle's powertrain batteries
instead of converting all energy to waste heat as is the case for the other test vehicles
equipped with only the hydraulic brakes. LINK conducted the brake emissions testing
for the Toyota Prius using their 'DutyCycleRegen' control program. This control program
is a combination of two operating principles: Duty cycle simulation to simulate a given
drive cycle (using the standard ProLINK 'DutyCycle' Program), and the addition of
regenerative braking functions.

LINK first analyzed the operational accuracy of the 'DutyCycle' (non-regen) program on
a ProLINK-controlled inertia dynamometer using the SAE J2951 procedure, Drive
Quality Evaluation for Chassis Dynamometer Testing9 Table 16 presents the results of
this analysis for the three segments of the CBDC for two vehicle assemblies.

The table presents the percent of time the parameter exceeded the control limits as well
as the root mean squared speed error (RMSSE) for each segment of the test cycle and
each assembly. RMSSE is the actual error as a percent of the maximum allowable
error. At the initiation of this program, the PMP inter-laboratory round robin had begun
the process of defining thresholds of acceptability for these two error metrics. At that
time, PMP members indicated that approximately 10% violation time is acceptable for

9 https://www.sae.org/standards/content/j2951_201111/

47


-------
maximum violation time. The threshold for RMSSE was not yet completely defined, but
was expected to be at or below 200%.

The overall error levels for both metrics were acceptable based on the threshold of the
PMP inter-laboratory round robin (which assigned a threshold of 100% on RMSSE).

Table 16. Results of SAE J2951 Analysis of the Sienna Front Assembly Operating over

the CBDC Test Cycle





F150 Front

Sienna Front

CBDC

Duration

Violation

%

RMSSE

Violation

%

RMSSE

Segment

(s)

Time (s)

Violation

Time (s)

Violation

1

8,767

308

4%

77%

578

7%

92%

2

4,011

109

3%

67%

207

5%

85%

3

2,784

77

3%

70%

290

10%

118%

Overall

15,562

494

3%

74%

1,075

7%

96%

Using the successful 'Dutycycle' program as a starting point, LINK created the
'DutyCycleRegen' program by adding the following primary features:

•	A dedicated section for burnish, the actual test cycle, and any intermediary
cooldown phases

•	Import of vehicle speed profiles into the control program using '.csv' file format

•	User input window to enter the number of repeats of the test cycle (e.g. number
of repeats of the ERG-CARB mini-trips for burnish or the WLTP-brake cycle)

•	User input window to enter the vehicle coastdown coefficients to account for
vehicle running resistance

•	User-interface to enter the regenerative brake system specifications

The regenerative braking activity of the selected Toyota Prius vehicle was simulated
using mainly four regenerative brake parameters of the electric motors:

•	Regensim_power: This is the maximum power that the vehicle electric motors
can convert to electrical energy

•	Regensim_Trq_Limit: This is the maximum torque that the regenerative system
can compensate for without any pressure applied tothe friction brakes.

•	Regensim_On_Above: This is the minimum speed above which the
regenerative system can operate at full capacity

•	Regensim_Off_Below: This is the maximum speed below which the
regenerative system cannot provide any braking support and all braking energy
is handled by friction brakes.

The following scenario is an example of the function of the regenerative braking control
feature. Consider a vehicle with regenerative system parameters shown in the screen
capture presented in Figure 21. In addition to the four parameters specific for a vehicle,
the program includes a generic parameter Regen_Brk_Trq_Min for additional control on
brake actuation times. If set to 0, the friction brake torque may take some time to build
as the brake is filled. This causes a delay and / or torque overshoot when the brake
finally clamps down onto the rotor.

48


-------
RegenSim Safeties Script Service Brake ServoTuning

Van

Desc (5 items)

Unit

Value

10080

Regensim_Power

kWatt

i

5 :

	i

10081

Regensim_T rqJJmit

lb-ft

368.7816

10082

Re g e n si m_0 n_Ab o ve

mph

6.21402

10083

Re g e n si e 1 ow

mph

1.2428

10084

Regensim_B rk_T rq_M i n

lb-ft

0

Figure 21. ProLINK screen capture of regenerative braking parameters

If a brake application from 50 km/h to 0 km/h requires 200 Nm (368.78 lb-ft) of
retarding torque, the graph in Figure 22 shows how the torque would be split between
the regen system and the friction brakes at different stages of braking.

•	Stage A: Power is speed- torque and thus for the first 5 seconds, the
regenerative brake torque gradually ramps up as the speed comes down with a
regenerative peak power of 5 kW.

•	Stage B: Regenerative system provides all braking needed from 25 km/h (=5000
W/200 N m) to 10 km/h (6.2 mph).

•	Stage C: Once the vehicle speed is below the Regensim_On_Above value of 10
km/h (6.2 mph), the friction brake kicks in and ramps up. At the same time, the
regenerative torque ramps down at the same rate. This is called "brake
blending".

•	Below the Regensim_Off_Below speed, friction brakes absorb all torque.

49


-------
200- 200- 200- 80-

§ 150)-? 150(— £ 150(— 
O

tr

50-

cn
::
cc

Figure 22. Plot of retarding torque (blue), friction brake torque (red), regenerative torque
(green), wheel speed (black), and regenerative power (purple) during a regenerative-

equipped braking event.

The original specifications for the Prius regenerative brake system were not available
due to non-disclosure policies of the manufacturer so LINK developed preliminary
estimates of the regenerative system parameters for the 2016 Toyota Prius Two-Eco
vehicle based on previous experience of conducting dynamometer tests for a
customer's Prius brake evaluation.

Figure 23 presents vehicle-to-dynamometer comparison plots of brake pressure and
total retardation torque for two brake events. These brake events were run on the test
track for a range of 0.1g-0.4g deceleration levels. Retardation torque is the combined
torque of resistance provided by the friction brake as well as the regenerative motor.
Torque was not measured during the track testing as torque wheels were not installed
at the time. Instead, the vehicle torque 'Veh Trq' shown here correspond to the values
calculated using the deceleration, tire rolling radius, and the wheel load front-to-rear
split percent (according to SAE J2789).

50


-------
20 to 0 km/h, 0.1 g

-Dyno V	Veh P	Dyno P	Veh Trq •

Dyno Trq
200

E
z

b 150

100

50





D

150

O"

i_



£



C

100

o



4—'



CT3



~o

50

i—

fD



%



Cd

- 0



10

12

Figure 23. Dynamometer-to-track comparison of velocity, brake pressure, and wheel
torque for example regenerative braking events

Torque measured on the dynamometer for a 0.1g deceleration level was within 10% of
track-tested calculation. Actuation times of the friction brakes during the dynamometer
test match well with the vehicle brake pressure. The Dyno brake pressure (P) was
slightly lower than the Vehicle P. A brake event with the duty cycle program starts in a
pressure-based control mode (more common in regular brake testing) and then
switches to torque mode. This is the reason behind a small spike in brake pressure at
the start of brake application on the dynamometer. This spike currently offsets the dyno
speed profile with respect to the vehicle speed profile.

LINK refined their regeneration simulation parameters for the dynamometer simulation
in the weeks leading up to testing based on continued review of the track test data. This

51


-------
included creating a separate value for one parameter for tests of the front and rear axle
assemblies of the Prius. Even though the Prius' regenerative system only acts on the
front wheels of the vehicle, the system does affect the demand on the foundation brakes
of both the front and rear axles. LINK used the parameters and values in Table 17 for
the simulation of the Prius regeneration capacities and speed ranges during all tests of
that model's components.

Table 17. Updated parameters used for the simulation of the Prius regenerative braking

system on the dynamometer

Parameter

Front Axle Simulation

Rear Axle Simulation

Maximum Power

2.5 kW

2.5 kW

Maximum Torque

90 Nm

60 Nm

Speed above which 100%
regeneration is available

8 kph

8 kph

Speed below which no
regeneration is available

3 kph

3 kph

Measurement Instruments

LINK equipped its test laboratory with a variety of TSI instruments for the measurement
of PM mass, count, and size distribution. This section describes the capabilities of the
various instruments that were used during this program.

TSI 100S4. The TSI 100S4 is the central instrumentation for this project. It has 4
different particle size classifications for the measurement of PM mass. The 100-S4 has
an 18 pm inlet stage (i.e. sampling is 18 pm and smaller), which is followed by cut-point
stages of 10, 2.5, and 1 pm. The instrument has Micro-Orifice Uniform Deposition
Impactors (MOUDI) for the collection of mass at each of these cutpoints. The impactors
are followed by a final filter to collect particles smaller than 1 pm. In the 100S4, LINK
used coated aluminum impactors, with a glass fiber final filter.

Figure 24. TSI 100S4 MOUDI

52


-------
TSI QCM MOUDI 140. The Model 140 Quartz Crystal Microbalance (QCM) MOUDI is
designed to perform continuous, real-time size-segregated mass concentration
measurements of particles smaller than 2.5 pm. The system uses six cutpoint stages at
960, 510, 305, 156, 74 and 45 nm and operates at a 10 L/min inlet flow rate.

i	t

Figure 25. TSI QCM MOUDI

TSI CPC. The 3790A Condensation Particle Counter (CPC) is a full-flow design PM
particle counter that has a particle size lower detection limit of 23 nm. The unit is
designed to linearly respond to particle concentrations from 1 to 10,000 particles/cm3
and can operate continuously taking 10Hz measurements. TSI indicates a counting
accuracy of ± 10%. The PMP has specified the use of this unit as the baseline for brake
particle counting without the use of a catalytic stripper or other volatile particle removal
(VPR) device. No VPR device was used in this program.

Figure 26. TSI CPC

TSI APS. The 3321 Aerodynamic Particle Sizer (APS) measures the aerodynamic size
of particles between 0.5 - 20 pm. The system operates using time-of-flight aerodynamic
sizing to determine the particle's behavior while airborne and is unaffected by index of
refraction or Mie scattering. The unit also measures light-scattering intensity in the
equivalent optical size range of 0.37 to 20 pm. The system offers continuous sampling
at 1 Hz.

53


-------
J.



Figure 27. TSI APS

TSI EEPS. The 3090 Engine Exhaust Particle Sizer (EEPS) is a spectrometer that
measures the size distribution of particle emissions from 5.6 to 560 nm continuously at
up to 10 Hz. The EEPS provides outputs of size distribution in the above range as well
as particle number concentrations down to 200 particles/cm3.

A •

Figure 28. TSI EEPS

Additional sampling was performed on behalf of EPA through their participation in this
project. EPA directed ERG and LINK to perform gravimetric sampling of brake PM
during all tests planned in the test matrix in a manner consistent with 40 CFR 1065.
Under this direction, LINK conducted parallel sampling of PM captured on 47 mm filters
during the tests that were already planned for this work.

The other specific addition to the test plan funded by EPA was the collection of PM
sample on Transmission Electron Microscope (TEM) grids using a Partector device
during a large number of tests. NVFEL staff loaned this Partector system to LINK for the
duration of the brake dynamometer testing. During use, this device only sampled for a
part of each test cycle (automatically stopping when loaded completely) at a relatively
low flow rate compared to the other instruments. These TEM grids were provided to
EPA for analysis at the end of the program.

The sampling lines and instruments are arranged as shown in Figure 29. Flow splitters
were used to separate the samples in Lines 2-4 into multiple instruments or
components. Lines 1-3 provided sample to the TSI equipment described for use for
CARB. Sample Line 4 provided sample to the filter equipment installed on behalf of

54


-------
EPA. The four probes were arranged in an equally spaced fashion in a single plane of
the sampling elbow.

Figure 29. Sample line schematic for this program

Sample line 4 included the measurements added on behalf of EPA. This sample ran
through a splitter into two PM10 cyclones. One leg of the splitter fed a 47 mm Teflon
filter followed by a 47 mm Quartz fiber filter (QFF). The other leg of the splitter fed only
a single 47 mm QFF. To equalize the pressures and flowrates, LINK installed a 47 mm
Teflon filter after the lone QFF; however, this filter was not used for any analyses. A
schematic of the layout of Sample Line 4 is presented in Figure 30.

55


-------
Teflon + Quartz
filterholderPM,,

Quartz-only
filterholderPM,,

16.7 Ipm

Flow direction
(vertical)

Flowmeter
+

Regulating
valve

Flowmeter
+

Regulating
valve

IVacuum Pump

33.41pm

SAMPLING
EXIT

Flowrate data
logged in
Pro LINK

Teflon + Quartz,
dual-stage
filter holder
PMln

Flow direction
(vertical)

Quartz-only
dual-stage
filter holder
PMln

Cyclone
10 um
Cutpoint

16.7 Ipm

Flow direction
(horizontal)

33.41pm

Flowsplitter

4th sampling line

SAMPLING
ENTRY

Cyclone
10 um Cutpoint

Figure 30. Detail schematic of Sample Line 4

The purpose of each of the three specified filters, along with potential analyses options
that could be performed on the TSI equipment media or 47mm filters, is described in
Table 18.

Table 18. The various filter media types and the respective analyses for each

Filter Material

Analysis Options

Teflon

•	Gravimetric Mass

•	XRF elemental

•	ICP-MS

Quartz Fiber (QFF) -
Following Teflon

• Volatile Organics that pass through Teflon (artifact
collection)

Quartz Fiber (QFF) -
Lone

• Particle Phase Organic Molecular Weight

Distribution (on initial filters to inform further test
types)

Coated AL Impactor

• Gravimetric mass

Glass Fiber

•	Gravimetric Mass

•	Possibly ICP, TBD

•	No further chemical analysis recommended

In preparation for testing, LINK acquired the Teflon filters for use in gravimetric testing
from two sources. Approximately half of the filters were provided by CARB, and half
were sourced from EPA NVFEL in Ann Arbor, Ml. EPA also provided all quartz fiber
filters (QFFs) for use in capturing sample for later speciation analysis. Details of each
filter type are presented in Table 19.

56


-------
Table 19. The particulate filter types used during testing

Brand/Model

Diameter

Pore Size

Source

Material

Whatman 7592-104

47mm

2pm

CARB

Teflon

MTL PT47DMCAN

47mm

2[jm

EPA

Teflon

PALL 2500QAO-UP

47mm

-

EPA

Quartz Fiber

For the TSI 100S4. LINK sourced impactor part number 0100-47-AF, and used silicone
spray from MSP, Part #07041. LINK sourced two dual-stage stainless steel filter holders
from URG, model URG 2000-30FVT, to collect PM samples on PTFE and QFF 47mm
filters. A picture of this filter holder type is shown in Figure 31.

Figure 31. Dual-stage stainless steel filter holder for PM10 sampling

Arizona Dust Experiment

Pr ior to the commencement of testing, LINK evaluated the PM sampling system using
Arizona dust as the particulate medium. Evaluations included the particle transport
efficiency and system responses to different airflows, brake speed, brake rotation
direction, and particle sampling setup. Arizona dust (per ISO 12103-1:2016) was used
for these evaluations. The dust was injected at the brake enclosure, travelled through
the sampling system, and collected on the 100S4 gravimetric filter. These experiments
were conducted prior to the installation of the 47mm filter system, so only the 100S4
was used. Multiple experiments were conducted to determine the recovery efficiency,
the repeatability, and the level of detection of the system.

Dust was emitted using a TSI 3410U dust aerosol generator. The generator's feed rate
(i.e. dosing speed) was set to 5% and injector pressure was 1.2 bar. Different airflows

57


-------
were evaluated to represent the possibility of covering various axles and vehicle
combinations. A brake rotational speed of 45 km/h was chosen to match closely to
average speed of WLTP cycle (and later for the CBDC), and a brake speed of 115 km/h
was selected as the upper range value as it is equivalent to a typical highway speed.
Brake rotation direction is designated viewing from the perspective opposite the
dynamometer drive shaft, and both clockwise (CW, airflow opposed to particle exit
direction from caliper) and counterclockwise (CCW, airflow parallel to particle exit from
caliper)) rotation directions were evaluated. The dust injection duration was 3 minutes
for each experiment. Table 20 presents the values that were tested during the Arizona
dust evaluation.

Table 20. LINK Arizona Dust Test Parameters

Airflow (m3/h)

500, 900

Brake speed (km/h)

45, 115

Brake rotation

CW, CCW

Sampling elbow

3-nozzle, 4-nozzle

Figure 32 shows the test setup for the dust injection experiments. Brake pads were not
installed inside the caliper to avoid possible emission of particles from brake drag (rotor
and pad surface interaction).

Figure 32. Brake assembly for PM system evaluation with Arizona dust

LINK utilized the 100S4 and APS to evaluate the test results. Figure 33 presents the
mass collection results of two replicate tests (Test #67 and #68) for 500 m3/h airflow, 45
kph brake speed, and CCW brake rotation. The PM system exhibits a collection
efficiency (i.e. mass recovered / mass injected) of about 90% with Arizona dust.

58


-------
67

68

Test/#

Figure 33. Mass collection efficiency results of two tests of the Arizona dust experiment

as measured by the 100S4

In addition, LINK supervised a computational fluid dynamics (CFD) simulation of
particle-laden airflow inside the brake enclosure and the sampling system. Arizona dust
was used for the CFD study as well. Details of this study can be found in the technical
paper SAE 2019-01 -2139, Design of Experiments for Effects and Interactions during
Brake Emissions Testing Using High-Fidelity Computational Fluid Dynamics'10. Figure
34 presents the flow velocity contour inside the brake enclosure with brake assembly
mounted, the stabilization duct, and into the sampling nozzles located at an 8 diameter
distance from the enclosure exit.

Velocity [m/s)

Nozzles 1, 2, 3

Figure 34. Velocity contour of cooling air resulting from CFD simulation

directs

Figure 35 compares the particle count history of the Arizona dust test replicates,
measured using APS, with the values calculated from the CFD simulations for various
particle sizes. Probed locations for particle count in the CFD study are at the enclosure
exit (inception), nozzle inlet, and inside the nozzles. Predicted particle counts (based on

10 Agudelo et. al., "Design of Experiments for Effects and Interactions during Brake Emissions Testing
Using High-Fidelity Computational Fluid Dynamics," SAE Technical Paper 2019-01-2139, 2019,
https://doi.Org/10.4271/2019-01-2139.

59


-------
the Arizona dust size distribution) match well with the measured profiles. Also, particle
count profile is seen to be very similar in nozzles 1, 2, and 3. This latter result indicates
that the PN measurements are independent of sampling nozzle location radially along
the duct cross section.

100%

90%

g 80%

§ 70%
o

v 60%

U

% 50%

CD
Q.

£ 40%

"I 30%

E

5 20%

10%

0%

0.5	5.0	50.0

Particle size / urn

Figure 35. Simulated and measured cumulative particle count at different locations of PM

sampling system

Figure 36 presents a graphical representation of a main effects analysis for measured
PM mass emission rate after correcting for duct airflow (duct-to-sampling flow
correction) in the Arizona dust experiment. Brake rotation direction showed significant
variation in PM mass rate based on the steepness of its slope, with higher values (i.e.
less mass loss) measured during tests with counterclockwise (CCW) direction of
rotation. Sampling setup, airflow, and brake speed did not have noticeable effect on the
measured PM mass rate. LINK conducted all brake emissions tests using the CCW
direction for brake rotation as default because that direction resulted in higher measured
PM mass (i.e. a higher collection efficiency) during the Arizona dust evaluation.

/

A

	Incep.



•

Run 1-Nozz. 1

• Run 1-Nozz. 2

•

Run 1-Nozz. 3

• Run 1-Duct 8D

•

Test 67

60


-------
Main Effects Plot for PM.Corr.

Fitted Means



Elbow

Airflow

Brake speed

Brake Rotation

2.00-









1.75







/

1.S0







/

125





















3-nozzle 4-nozzle

500 900

45 km/h 115 km/h

cw ccw

Figure 36. Graphical plot of the PM response to the parameters of the number of
sampling nozzles, airflow rate, brake rotational speed, and brake rotational direction

during the Arizona dust evaluation

Teflon Filter Weighing interiaboratory Evaluation

At the completion of LINK'S setup of the laboratory, EPA NVFEL staff participated in a
filter "Round Robin" comparison with LINK. EPA NVFEL provided approximately half of
the Teflon filters used during this work (CARB provided all others). After EPA provided
all of the test filters, LINK staff selected 30 at random and weighted each of them three
different times in their on-site weighing room. Then, the filters were returned to NVFEL,
and allowed to stabilize. Then, EPA staff weighed the same 30 filters three times. This
allowed for direct comparison to weights measured at NVFEL and at LINK. During
transport between laboratories, the filters were kept cool and subject to minimal
handling.

Results are presented as LINK measurements plotted against NVFEL measurements
for each filter in Figure 37. Note that the plot includes all three measurements made by
each lab on each filter (though they are superimposed over one another). The equation
for a linear fit to the data is presented on the plot; the R2 value of this fit is greater than
0.99999. The values in the plot are buoyancy corrected per 40 CFR 1065.690 by both
laboratories. EPA data did include a further drift correction, however that data was not
used in this discussion as, for the purposes of this evaluation, drift correction was not
necessary.

61


-------
LINK LAB [mg] = 1.0000824 x NVFEL [mg] - 0.0348252

405

"
-------
•	Record pre-test background air data and then start the CBDC (California
Light-Duty vehicle) burnish cycle to bed the friction materials together
(approximately 11.5 hrs)

•	Monitor the dynamometer and emission data recordings. PM mass was not
measured during burnish. To protect from excessive amounts of brake dust
noticed in some cases, CPC was turned off during burnish for one of the two
replicates of the test matrix

•	Record post-burnish background air data. Review burnish data after the cycle
is complete.

•	Clean the inlet cyclones of EEPS and CPC (whenever used during burnish)

•	Take the QFF from freezer to the weighing room for 1 hr stabilization

•	Weigh the pre-conditioned coated aluminum, glass fiber, and PTFE filters. PM
Weighing is done only if the room temperature is within (22±1) °C and the dew
point is within (9.5±1) °C.

•	Turn on the QCM unit and install the sampling filters/media in the 100S4
stacks and 47mm particulate mass sampler (PMS) filter holders

•	Connect 100S4 and PMS to the dynamometer system

•	Run the CBDC brake emissions cycle (4.5 hrs)

•	Upon test completion, perform data quality assurance test to check for any
defects as defined in the Quality Assurance Project Plan (QAPP for EPA
Work Assignment 1-04)

•	Remove filters and media and return to weighing room for stabilization

•	Remove brake assembly after the data passed the quality test (15 mins)

•	Take the dimensions, weights, and pictures of the brake components after the
test (15 mins)

•	Weigh the post-tested coated aluminums, glass fiber, and PTFE filters.

•	Store the QFFs and PTFEs at -20 °C in the freezer

The above schedule required approximately 18-19 hours if no issues were encountered.
This allowed a reasonable margin of time to address any problems and with LINK staff
working in shifts, the project was generally able to stay on a 1 test per day schedule.

Quality Assurance (QA) of Test Data

Both LINK and ERG staff performed a QA review of the results of each test. LINK
completed their internal review first, then data was provided to ERG for further external
review.

LINK developed a methodology to check the quality of data for various variables at the
completion of each test. These variables include shaft speed (km/h), rotor temperature,
cooling air settings, digital emission instruments, and the PM10 mass sampler. LINK
staff:

- used the GTR15 drive quality regulation to assess the accuracy of the control

program in simulating the CBDC test cycle on the dynamometer;

63


-------
-	checked the rotor temperature to ensure there were no loose-wires or defective
thermocouples;

-	checked that the cooling airflow settings i.e. temperature, relative humidity and
air speed, were within the PMP (Particulate Measurement Programme)
recommended limits;

-	applied certain thresholds for specific channels of APS, CPC, and EEPS, to
validate that the data recorded was well above background levels;

-	evaluated the PMS flowmeter stability using certain criteria specified based on
preliminary runs.

Figure 38 illustrates an example of the software tracking of LINK'S quality assurance
review for a selected test. It indicates the tolerance range for each parameter and
provides a color-coded output indicating the status of the actual value.

Speed Conformity

Condition

%Violation

RMSSE

PASS/FAIL

Speed Profile

% Violation < 10% and RMSSE < 100%

0%

25%

Pass

Temperature Metrics

Condition

Min Temp

Max Temp

PASS/FAIL

Rotor Temperature

Must be > 15 °C and < 400°C

31 °C

273 °C

Pass

Cooling Air Check

Condition

Trip Avg.

%Violation

PASS/FAIL

Cooling Air Temperature

Trip Average within (20 ± 2) °C % Violation < 15%

21.0

14%

Pass

Cooling Air Speed

% Violation < 15%

7.5

0%

Pass

Cooling Air Humidity

Trip Average within (50 ± 5) %RH % Violation < 15%

50.1

0%

Pass

Instrumentation Conditions

Condition

Engineer Check

PASS/FAIL

CPC

Blob 3 — Col 3 from 40,000 to 50,000 sec average must be > 50

P

Pass

APS

Blob 2 from 40,000 to 50,000 sec awrages — Col 3 > 20 / Col 25 > 1

P

Pass

QCM

Blob 4 — Col 7, 9, 11, 13, 15, 17 from 35,000 sec positiw trend

P

Pass

PMS

Condition

Min

Max AvgiineerCh

PASS/FAIL

PMS Flowmeter 1 (QFF)

Must be less than 15% fluctuation (14.195 
-------
Results

This section presents the results of the brake dynamometer testing that took place
during this program. Brake assemblies were tested using a brake dynamometer and
CVS system for the measurement of PM. The air flow through the system provided
cooling to the tested brake assembly during the test and also served as the medium to
carry particulate to the point of sampling. PM mass was measured gravimetrically in
batch and continuously. Particulate size and count were also measured continuously
throughout each test. Assemblies from six light-duty vehicles equipped with various
OES and aftermarket friction materials were tested between September 30, 2019, and
January 29, 2020. Eighty-five valid tests (including 2 tunnel blanks) were conducted
during this time, including a one-week pause to review the preliminary data after the first
two weeks of testing. Appendix F includes the dates that each test in the matrix was
conducted.

Where possible, testing was conducted according to the order in the test matrix. In
some cases, certain external factors forced LINK to re-order the testing. For example,
some of the aftermarket parts that were ordered did not arrive in time to be ready when
their test was scheduled to take place. In these cases, the test order was re-arranged,
and these tests were inserted into the test schedule once the parts arrived.

The following sections present various test results from this work in graphical and
tabular form. Appendix H contains a table of numerical results of all direct
measurements for all tests. The table includes the test parameters, gravimetric mass
results, condensation particle counter (CPC) results, and some selected operational
measurements such as temperatures and brake line pressures. Appendix I presents the
test reports generated by LINK for each individual test. For brevity, in this section some
plots or analyses refer broadly to the pads and rotor as the only components; in
analyses that include the Civic rear axle these terms are intended to also include that
vehicle's brake shoes and drum, respectively.

Operational Parameter Results

This section presents a brief overview of relevant parameters measured over each test
day. Test-level averages of three main parameters will be presented, average brake
rotor temperature, maximum rotor temperature, and the average brake fixture torque.
These values are averaged for each model, axle, and test weight in each of the
following figures. Figure 39 presents the average rotor temperature during each test for
each model, axle, and test weight combinations. Error bars present the 95% Confidence
interval of the mean of all tests of each combination. Similarly, Figure 40 presents the
peak rotor temperatures in each test, averaged by model, axle, and test weight.

65


-------
_ 180
u

— 160

0)

| 140

ro

S 120

Q.

100

O
cc

a»

CUD
ro
&_
0)

<
+-»

to

80
60
40
20

 ^ ^ ^



v v o>jy

 s *• J? ^ J J 
-------
Figure 41 presents the average brake torque measured over each test, and then
averaged across all combinations of model, axle, and test weight. The figures in this
section are presented to provide some context to the subsequent PM analyses. For
example, in terms of brake torque, the F-150 front axle tests exhibited the highest level
of braking torque, and the Prius test resulted in the lowest braking torque due to the
simulation of that vehicle's regeneration function.

350

300

— 250

CD
D
O"

200

0)

ro
&_
CO

0)
CUD
ro
&_
0)

<
-M

l/)

150

100

50

li

i

f/vv c?'////^ v

Figure 41. Average brake torque measured during the CBDC, averaged over all tests of
each model, axle, and test weight combination.

Batch Gravimetric Results

The highest priority measurements during this project were the various gravimetric PM
mass emissions, collected in batch for each test. This section presents those mass
measurements, on a by-distance basis, for all tests. Results are shown in bar charts by
test vehicle for single-wheel emission rates from front and rear axle assemblies as
indicated. The 100S4 stages are shown in the blue bars (the left bar of each test pair)
consisting of the three size cutpoints up to PM10. The Teflon filter system sampled
PM10 mass and is shown in the green bars (the right bar of each pair). Tests are
labeled with the vehicle model and the pad material. All tests can be assumed to be at

67


-------
vehicle mass of ETW tests unless specifically labeled as an HLW test. Each figure
depicts front axle results at left, and rear axle results at right.

Figure 42 presents the mass emission measurements for each test of the Camry. Three
different pad materials were tested both front and rear. It can be seen that the low
metallic pad materials resulted in the highest measured emissions. Figure 43 presents
the mass emissions results for the Civic. Both the OES and aftermarket materials tested
for the Civic are NAO formulations. Note that the Civic rear brake assembly is a drum;
the drum brake geometry is likely the cause of the particularly low emission rate for that
assembly, especially for the OES material.

147mm PTFE PM10 (mg/mi)

100S4Sg3 (PM2.5-10) Emission mg/mi
1100S4 Sg4 (PM1-2.5) Emission mg/mi
1100S4 Aft Filter (PM < 1) Emission mg/mi

REAR

J*







c*





&











c?

4?

Figure 42. Single-wheel PM Mass Emission Rates for Camry as measured by 100S4 (Blue)

and 47mm Teflon filter (Green)

68


-------
2.5

CD J
+-» £-
ro
cc
c
o

"S 1-5
E

~J •cr

° -E1

"> ClO -1-

£ £

ID

(D

^ 0.5

_oj

CtO

c

0

..O^ .,0^	.,0^ .,0^	$*¦'

Figure 43. Single-wheel PM Mass Emission Rates for Civic as measured by 100S4 (Blue)
and 47mm Teflon filter (Green). Note the Civic rear brake is a drum system

Figure 44 presents the mass emissions results for the F-150. This vehicle was
represented in the greatest number of tests because it served as the reference vehicle,
was tested with three different friction materials, and was tested at both test weight
levels. The PM emissions of the low metallic material was measured at many factors
higher than the NAO materials for the front axle assemblies. The various test matrix
parameters tested for this vehicle model resulted in the widest range of resulting PM
emissions of any model in this program.

FRONT

147mm PTFE PM10 (mg/mi)

100S4Sg3 (PM2.5-10) Emission mg/mi
1100S4 Sg4 (PM1-2.5) Emission mg/mi
1100S4 Aft Filter (PM < 1) Emission mg/mi

REAR

II ll

¦JT









¦ST





69


-------
16

j=

14

CUD









12

(D



+->



ro



cc
c

10

o



"oo



CO

8

E

LU



O

6

T—\







Q_





4

ID



(D



_C



5

2

(D



CuO



C

0

on



FRONT

II li II

II ii ll

h II

147mm PTFE PMIO (mg/mi)

100S4Sg3 (PM2.5-10) Emission mg/mi
1100S4 Sg4 (PM1-2.5) Emission mg/mi
1100S4 Aft Filter (PM < 1) Emission mg/mi

REAR

II II ll

\?	v . y > . y	< \y < \?	^jv







Figure 44. Single-wheel PM Mass Emission Rates for F-150 as measured by 100S4 (Blue)

and 47mm Teflon filter (Green).

70


-------
Figure 45 presents the mass emission rates for the Prius, which was tested with both
OES and aftermarket NAO materials. While the Prius has a similar mass to the Camry,
the mass emission rate tended to be approximately 50% lower for both the Prius front
and rear axles. This is likely to be due to the reduction in demand on the hydraulic
foundation brakes caused by the regenerative braking system function.

147mm PTFE PM10 (mg/mi)

100S4 Sg3 (PM2.5-10) Emission mg/mi
1100S4 Sg4 (PM1-2.5) Emission mg/mi
1100S4 Aft Filter (PM < 1) Emission mg/mi

REAR

II II ll ll





9*







•> ,„£>>







Figure 45. Single-wheel PM Mass Emission Rates for Prius as measured by 100S4 (Blue)

and 47mm Teflon filter (Green).

Figure 46 displays the mass emission results for the Rogue, and Figure 47 presents the
results for the Sienna. Both vehicles show elevated front-axle emission masses for the
HLW tests. For the rear axle, the HLW tests did not result in appreciably higher
emission masses, however the Sienna HLW test did result in a higher emission rate
than the ETW tests.

71


-------
60

E

tc
QC
C

o

¦F 3

(L)
(L)

0

FRONT

147mm PTFE PM10 (mg/mi)

100S4Sg3 (PM2.5-10) Emission mg/mi
1100S4 Sg4 (PM1-2.5) Emission mg/mi
1100S4 Aft Filter (PM < 1) Emission mg/mi

REAR

II II ii i. ¦¦ II

I ^	^

.J? .J*.Vs* " "

>»" >?•

,o«- ^ 
-------
As presented in Table 19, two different Teflon filter types were used during testing
(generally alternating across each test matrix replicate pair). Appendix J presents the
initial and final weights of each filter along with the buoyancy-corrected weight gain.
LINK performed buoyancy correction based on the laboratory ambient conditions as per
the procedure outlined in 40 CFR 1065.690.

ERG also reviewed the test results against the as-delivered BMC Leafmarks for each
pad material. The delivered LM pads all were assigned the letter "A" (the highest copper
level). The overall material type appeared to have a larger effect on PM emission mass
than the BMC Leafmark; after removing the LM pad tests with elevated emission levels,
there was no apparent trend between emissions and BMC Leafmark across the OES
and Aftermarket NAO tests.

Vehicle-Level Mass Results

Each test measured the PM emissions from a single wheel from either the front or rear
axle. This section aggregates the test results to estimate vehicle-level emissions. For a
given vehicle model, friction material, and test weight, the average front and rear
emissions are added together and doubled to estimate 4-wheel emissions rates for
each combination. Results presented in this section use only the PM mass
measurements made by the 100S4 system.

Figure 48 presents the vehicle-level PM10 mass emission rates for each vehicle and
test weight combination. Values are calculated as applicable for those friction materials
that were tested on both front and rear assemblies within each model. Values range
from approximately 3 mg/mi for the Prius OES material to around 30 mg/mi for the F-
150 when heavily loaded with aftermarket low metallic pads.

73


-------
35

¦ OES-NAO ¦ Afte

-—- Ofl

r-NAO ¦ After-LM

j u

j=

CuO



00

¦- 70





oo ZU

i15





1 J

m 1 n



.Ml


-------
Table 21. Estimated balance of friction materials by model for vehicles models at 3 and

11 years old.

Age

Model

OES NAO%

AM NAO%

AM LM%



Camry

29.8

57.8

12.5



Civic

29.8

70.2

N/A

(/)
i_

F-150

78.8

8.8

12.4

CO
0

Prius

29.8

70.3

N/A

CO

Rogue

29.8

70.3

N/A



Sienna

29.8

70.3

N/A



Camry

22.4

43.6

34.0

TD
O

Civic

22.4

77.6

N/A

W

!	

TO
0

F-150

59.4

6.6

34.0

Prius

22.4

77.6

N/A



Rogue

22.4

77.6

N/A



Sienna

22.4

77.6

N/A

These values were then used to calculate overall by-model PM emission rates for these
models in the in-use fleet. The values presented in Table 22 are multiplied by the in-use
fleet balance to calculate an estimate of in-use emissions for each model and test
weight combination at an average age of 7 years. This report revision corrects a
previous error in the table's reported vehicle-level PM2.5 mass emission rates.

Table 22. Measured in-use brake emission rates by model, estimated for 7 year old

vehicles.

Model

Estimated In-Use PM2.5

Estimated In-Use PM10



Emission Rate (mg/mi)

Emission Rate (mg/mi)

Camry

3.4

9.3

Civic

2.2

5.7

F-150

4.1

9.8

F-150 HLW

5.1

13.9

Prius

1.9

3.3

Rogue

3.9

9.2

Sienna

4.2

9.9

Sienna HLW

5.4

13.9

Results by Speed Segment

The CBDC test cycle was designed such that the relative distance traveled in each of
the speed segment approximates the relative distances traveled by all microtrips within
each speed range in the Caltrans dataset so that the overall cycle emissions represent
real-world driving. However, the three speed segments were created to allow further
refinement of the emissions factors based on trip average speeds as in EMFAC.

75


-------
The EMFAC model uses speed correction curves (SCCs) to adjust pollutant emission
factors for trips of varying average speeds. The test cycle in this work was comprised of
three different cycles representing microtrips falling in average speed ranges of 0-21
kph, 21-69 kph and above 69 kph. ERG used these different speed ranges to develop
trends for speed correction of the brake PM emission rates.

The QCM measures continuous mass (PM2.5) and is the measurement that can resolve
differences across the 3 speed segments. (The 100S4 and 47mm Teflon systems were
used for cumulative measurements over the cycle, not by speed segment). In this study,
the QCM was considered a relative measurement; while it reports in units of mass,
there was a large amount of noise (R2 ~ 0.35) when comparing cumulative QCM mass
for a test to the 100S4 PM2.5 results. To use it as a relative measurement only, ERG
used the QCM results to apportion the 100S4 PM2.5 total mass to each speed segment
based on the percentage of QCM mass recorded in each.

The QCM measures only PM2.5. However, the APS was measuring the size
distributions throughout the tests as well. ERG used the APS size distribution to also
estimate the PM10 mass in each speed segment from the QCM results. ERG took the
following steps to perform this estimation:

•	Gather APS data that indicates the particle size distribution (by counts) in the
range from 0.5-20 |jm.

•	Assume that particles are spherical, then calculate ratio of PM10 particle volume
to PM2.5 volume by summing the total particle volume measured at each APS
cutpoint (counts of particles multiplied by the volume of a sphere at that
cutpoint's diameter)

•	Assume particles have a homogenous density such that mass is proportional to
volume. Calculate the ratio of PM10 mass to PM2.5 mass based on the total
volume.

•	Multiply the apportioned PM2.5 by this ratio to estimate the percentage of the
total PM10 emitted in each speed segment

•	Use the calculated percentages of total PM10 for each cycle as weighting factors
to apportion the cycle PM10 mass measured by 100S4 to each speed segment

This method required significant assumptions but did allow for the QCM to be used to
estimate the PM2.5 and PM10 masses emitted within each speed segment. Figure 49
presents the overall mass emission trend by speed segment observed when averaging
all tests. It can be seen that the 21-69 kph (13-42 mph) speed segment has the highest
per-mile emissions, followed by the low speed segment with moderate per-mile
emissions, and the high speed segment has the lowest per-mile emission rate. The
error bars in the plot present the 95% confidence intervals based on the variability
across all tests only (they do not necessarily indicate any error caused by the QCM or
the assumptions used in this approach). Figure 50 presents, for the same data, the
trend in the PM2.5 mass percentage of PM10 averaged across all tests. At higher
speeds, the data indicate that PM2.5 makes up an increasing share of PM10 mass
emissions.

76


-------
7


-------
3

£ 2.5

0-21	21-69	69+

Speed Range

Figure 51. Speed correction factors for the three speed bins. These factors can be
multiplied by the overall cycle emission rate to estimate the emissions of operation

within each speed bin.

Because the speed-based emission results are not monotonic with respect to the speed
cycle ranges, ERG further investigated the overall trend presented in the speed
segment analysis. There are some potential factors that confound the agreement
between speed segment and emissions, notably the distance represented by each
cycle, the stops/distance, the braking energy/distance, and the actual duration of
braking in each speed segment. Figure 52 presents a bar chart of total braking energy
within each speed segment along with the total PM2.5 emission average of all 5
reference tests of the F-150 (axes are scaled differently to determine proportionality). In
the high and medium speed segments, the energy and emissions are proportional. In
the low speed segment, the relationship trends together (and neither are monotonic) but
the energy is high relative to the emissions.

78


-------
140000

^120000

>100000
cm

Cl)

c

80000

§ 60000

ro

™ 40000

ro

£ 20000

Total Braking Energy ¦ F-150 Ref. Test Avg

69+

35
30
25
20

DO

(N

15 o-

<1)

10

5

0

to

+-»
,o

0-21	21-69

Average Speed Range (kph)

Figure 52. Total braking energy (proportional to kJ) within each speed segment and the
F-150 Reference Test Average Total-Cycle PM2.5 emission mass for each speed segment

Figure 53 presents the total PM2.5 emission divided by the total seconds of braking
time within each speed segment, averaged over the F-150 reference tests. Presented in
this way, the emission measurements are monotonic. A key factor in the interpretation
of the SCFs is the cycle-represented distance. Emissions per braking second trend with
speed, however accounting for the distance traveled appears to confound that trend.

Figure 53. Total PM2.5 emission divided by total braking time in each speed segment,

averaged for the F-150 reference tests

79


-------
Emission Mass by Vehicle Weight

Figure 54 presents the vehicle-level test emission masses versus the simulated vehicle
test weight, categorized by pad material. Labels of vehicle model are shown at the
location of each model on the x-axis. The trends within each pad material do appear to
be linear, so linear fits are shown by pad material. The fit to the aftermarket LM
materials has the highest slope and the OES-NAO has the lowest slope. In this
analysis, the fits are not forced through the origin (i.e. intercepts set to zero). The slopes
and intercepts for these fits are presented in Table 23. While there may be engineering
justification to set the intercepts to zero, there is little justification or utility in doing so as
the Civic was the lightest vehicle tested (meaning no data is available near the origin)
and is likely to be near the lower limit of weights of any light-duty vehicle to be modeled
(meaning any modeling error at a zero vehicle weight will not affect results).

35

30

25

M 20

15

£ 10

a>
>

"Hi
+-»

o







¦













F-150

¦



Camry



Sienna



1.
•

Sienna

HLW^

	.ft-

~



	 p

togue





r

Civic ^
Prius





1200

1700	2200

Vehicle Tested Weight (kg)
OES-NAO	~ AfterMkt-NAO

2700

3200

AfterMkt-LM

Figure 54. Total vehicle test cycle PM10 mass emissions vs simulated vehicle test

weight, categorized by pad material.

Table 23. Slopes and intercepts for linear fits by material for PM mass emission rate

(mg/mi) versus vehicle tested weight (kg)

Material

Slope
(mg/kgmi)

Intercept (mg/mi)

Coefficient of
Determination (R2)

P Values,
Slope, Intercept

OES NAO

0.003

1.55

0.91

0.23, 0.75

After NAO

0.006

-2.21

0.64

0.03, 0.58

After LM

0.013

-8.37

0.23

0.19, 0.55

80


-------
Emission Mass and Component Mass Loss During Testing

LINK staff weighed the friction materials before and after each test. Rotors (or drums for
the Civic) were weighed separately from pads. This allowed for the calculation of the
total mass lost from the components over the course of the burnish and test cycles.
Components could not be weighed after the burnish and before the test cycle because
removal would have potentially upset the friction couple and nullified the burnishing
process. ERG investigated the extent to which the total mass loss was proportional to
measured PM mass emissions. This section references the components as pads and
rotors generally, but for the Civic rear, the corresponding parts were shoes and drums
and they are included.

Figure 55 presents the total PM mass emissions (y-axis) measured during the test cycle
plotted against the total mass loss of pads and rotor for each burnish and test. The plot
shows PM10 as measured by the 100S4 and has somewhat less noise than that of
PM2.5 (though the trends are otherwise similar). Fits are not presented by material as
the specific slope of the relationship between the test PM and the pad and mass loss is
arbitrary due to the burnish cycle (for which emissions were not quantified by the
100S4). The relationship is presented to show the general level of proportionality.

1400

ao
£

1200

o 1000

£
LU

O

800
600

Q_

a! 400

(O
+J

o

200
0







¦





¦

¦

• ~ ~

~

... T.

•

f

¦







10	15

Total Pad and Rotor Loss (g)

20

• OES-NAO ~ After-NAO

After-LM

Figure 55. Total test-cycle PM10 emissions plotted against the total mass loss of the pad
and rotor during the burnish and test cycle. Each point represents one test.

ERG also investigated whether the PM10 emission rates were more correlated to the
mass of pad lost, the mass of rotor loss, or the sum of pad and rotor. Figure 56 presents
the PM10 emissions totals for each test against the brake component mass loss for
three categories: pad + rotor (same data as Figure 55), pads only, and rotor only. The
plot includes R-squared values calculated for a linear fit made to each category. The
best linear fit is for the sum of pad and rotor, and the noisiest fit is to the rotor only. This
finding indicates that mass losses of both pads and rotors contribute to and are

81


-------
correlated with emitted PM. However, it does appear as though the emitted PM mass is
more responsive to the mass of pad lost. This could be because, in most tests, the pad
mass loss exceeded that of the rotors. The pad mass loss averaged 3.2 times that of
the rotor mass loss across all tests. However, this ratio varied widely across tests did
not appear to be sensitive to the friction material type.

1400

1200

c 1000
o

800
600
400
200
0

CL

+-»

(/)
0)

tc
+-»
o



•

~

¦









	A.

• ~

~ ¦

¦

R2

R2

= 0.3145.

= 0.5941..•••'"'

		 = 0.6205

¦ •

•

•

P -

			

¦



*	 #



:Jp *

jA .





5	10	15

Component Mass Loss During Burnish and Test (g)

20

Pad and Rotor

Pad Only

Rotor Only

Figure 56. Total cycle PM10 emissions relative to mass lost for pads, rotors, or the sum
of pads and rotors. R-squared values are presented for linear fits to each.

Particle Counts

The CPC was used to measure particle counts per unit of sample volume throughout
the duration of each test. Particle number totals for each test are included in Appendix
H. Table 24 presents the range, median, and average number of cumulative particles
for all CBDC tests on a per-mile basis. In general, there was a weak trend in vehicle
weight and particle count; larger vehicles tended to have larger particle counts.

Table 24. Selected statistics for single-wheel particle number counts for all tests (#/mi)

Test Statistic

Particle Number (#/mi)

Average

1.681 x 109

Median

1.258 x 109

Maximum

8.429 x 109

Minimum

3.913 x 108

82


-------
ERG calculated vehicle-level particle number emission rates by doubling the single
wheel emission averages for each model/material/test weight/axle combination. These
axle-level values were summed within each model to calculate vehicle-level values.
Figure 57 presents these estimates by vehicle model and categorized by pad material
type. Compared to the mass results, the spanned range of values across models and
pad materials appears lower for particle count results, meaning all vehicles appear to
have more similar particle count emissions than they do mass. Generally, the trends
otherwise follow those observed for PM mass with a notable exception of the Prius
aftermarket NAO material. The elevated values for this combination were caused by
high emissions measured in both tests of the front axle; the rear axle emission rates
were in line with the other sedans.

1.4E+10

¦ OES-NAO ¦ After-NAO ¦ After-LM

~ 1.2E+10

Camry Civic F-150 F-150 HLW Prius Rogue Sienna Sienna

HLW

Figure 57. Vehicle-level particle number emission rates for each vehicle and friction

material combination

Figure 58 presents the overall trends in single-wheel particle emission rates across the
three speed segments. The figure presents the emission rates averaged across all
tests. Error bars represent the 95% confidence interval for the mean of the results of all
tests. Appendix K presents the corresponding vehicle-level results by individual model.

83


-------
3.0E+09

E

^ 2.5E+09

(D
+-»

* 2.0E+09

c
o

to

| 1.5E+09

LU

U

u 1.0E+09

Q)

CD
_C

^ 5.0E+08

Q)

ClO
_c

10 0.0E+00

Figure 58. Overall average single-wheel particle number emission rate across the three

speed ranges

Particle Size Distributions

Particle size was measured in two size ranges; the EEPS measured from 5.6 - 560 nm,
and the APS measured from 0.5 - 20 |jm. The APS results tended to show noticeable
trends and responses to the test matrix parameters, and their results follow in this
section. The EEPS results were more similar across all tests with a lower degree of
apparent responsiveness to the test parameters, so EEPS results are presented only in
Appendix L. Size distributions are presented in graphs by vehicle and axle, and each
distribution is color coded by brake pad material. Note that all distributions in the size
distribution figures are normalized to sum to 1. This is to enhance the comparability of
particle size without any bias from particle number differences at the test level. Particle
number is measured specifically by the CPC so reporting absolute count is not a
necessary function of the particle sizers.

Figure 59 presents the size distribution of particles from the front brake tests of the
Camry. It can be seen the test replicate pairs of the three different pad materials follow
similar patterns. The low metallic material tends to emit larger particles, and the two
NAO materials are relatively similar in particle size distribution.

0-21	21-69	69+

Speed Range

84


-------
0.04

c n rvac

'I 0.01
o 0.005

¦ OES-NAO
•OES-NAO
¦After-NAO
•After-NAO
After-LM
After-LM

h-

m cd oo o
o o o

O) (D CO to in O) N
c\| ^ O) ^ O N ^
CN CO CO

m oo

co r-

Particle Diameter Bin (|jm)

Figure 59. Size distribution of Camry front brake PM as measured by APS

The PM size distribution for the rear brake tests of the Camry is presented in Figure 60.
The low metallic material has the largest particle sizes, and the aftermarket NAO tends
to have the smallest.



\- 
-------
Figure 61 and Figure 62 present the size distributions for the Civic front and rear,
respectively. The OES particles from the front tend to be larger than the aftermarket, but
that trend appears to be reversed for the rear drum brake emitted particle size.

^N^\fO)(Doo(DiflfflNnnoo
tf}COOOOCMT_CT>-*J-Oh-,^:OOC\JCT>

o o o -t-	¦ osi co co Lor^cb

^ w m
T °P T
^ (
-------
The size distributions for F-150 front and rear are presented in Figure 63 and Figure 64,
respectively. For the front brakes, the low metallic pads have the largest particle size,
but the trend is more inconclusive on particle size emitted from the various rear pads.

0.05
0.045

-»—<

i 0.04

o

« 0.035

^¦N^"to)oco(Dino)NnnooTtMifl
miDOOOCM^ffl^ON^OOCMOlT-OOi-

CN CO CO

m n to

Particle Diameter Bin (|jm)

• OES-NAO
¦OES-NAO
¦OES-NAO
¦OES-NAO
¦ OES-NAO
¦OES-NAO HLW
¦OES-NAO HLW
¦After-NAO
¦After-NAO
After-LM
After-LM
¦After-LM HLW
¦After-LM HLW

Figure 63. Size distribution of F-150 front brake PM as measured by APS

0.045
1 0.04
O 0.035

_CD

O

tS 0.03

ra 0.025

CL

t, 0.02

CD

n 0.015

0.01
0.005

OOOCM'CD- -t-
lO Is- 00 r-

Particle Diameter Bin (|jm)

•OES-NAO
¦OES-NAO
¦OES-NAO HLW
¦OES-NAO HLW
¦After-NAO
¦After-NAO
After-LM
After-LM
¦After-LM HLW
¦After-LM HLW

Figure 64. Size distribution of F-150 rear brake PM as measured by APS

87


-------
Figure 65 and Figure 66 present size distributions for the Prius. The two NAO pad
materials are largely overlapped in emissions from both the front and rear assemblies.

0.045
§ 0.04
O 0.035
75 0.03
ro 0.025

CL

T3 0.02

- oo ¦>-

•OES-NAO
• OES-NAO
•After-NAO
¦After-NAO

Particle Diameter Bin (pm)

Figure 66. Size distribution of Prius rear brake PM as measured by APS

88


-------
Figure 67 presents the size distribution of particles from the front brakes of the Rogue.
The OES material tends to have a larger particle size than the aftermarket material.
Figure 68 presents the findings for the Rogue rear, in which the distributions are largely
overlapped.

0.045
§ 0.04
O 0.035
0.03
ro 0.025

CL

T3 0.02


-------
Figure 69 and Figure 70 present the size distributions for the Sienna front and rear,
respectively. Both display significant overlap in the size distributions of the different
friction materials and test weight. Both the front and rear distributions tend to have a
minor bimodal effect, with a major peak at 1.98 pm and a minor peak at 1.29 pm.

infflOOOCM-ffl^ON^OONOlT-OOi-

¦OES-NAO
¦OES-NAO
¦OES-NAO HLW
•OES-NAO HLW
¦After-NAO
¦After-NAO
¦After-NAO HLW
•After-NAO HLW

o o o

co

— t-cncoco m s co

Particle Diameter Bin (|jm)

Figure 69. Size distribution of Sienna front brake PM as measured by APS

¦OES-NAO
¦ OES-NAO
¦OES-NAO HLW
¦OES-NAO HLW
¦After-NAO
¦After-NAO
¦After-NAO HLW
¦After-NAO HLW

Particle Diameter Bin (|jm)

Figure 70. Size distribution of Sienna rear brake PM as measured by APS

90


-------
Tunnel Blanks

LINK conducted two tunnel blank experiments, one early in the test program and the
other late in the test program. Tunnel blanks involved running a complete CBDC test
cycle without brake pads mounted or the brake hydraulic pressure active to help
determine the level of background PM present during a test. Table 25 presents the
results of the tunnel blank experiments as compared to all other brake assembly tests
for particle counts and mass measurements. Mass measurements of PM10 were
conducted with both the 100S4 as well as the two parallel 47 mm PTFE filter holder
systems (note that in normal tests only 1 of the filters was directly measuring PM using
a PTFE filter). The particle counts during blanks averaged approximately 0.2% of the
test average. The 100S4 blank measurement was about 1 % of the average test mass,
and the PTFE filters averaged approximately 2.8% of the average value when including
a single outlying high value during Tunnel Blank 2. No corrections were applied to the
test results based on the tunnel blank findings. The main concern was that one of the
mass measurements was an outlier in the 100S4 vs PTFE results by test. ERG and
LINK were concerned that performing corrections based on an outlying point would
have a detrimental effect on all results.

Table 25. PM counts and mass measurements for the two tunnel blank procedures as
well as the averages for all brake emissions tests



Test

CPC#

100S4 PM10

47mm PM10 1

47 PM10 mm



Day



(mg)

(mg)

2(mg)

Tunnel Blank 1

17

9.5x107

1.4

2.7

2.3

Tunnel Blank 2

66

2.1x108

2.7

3.7

12.5

Overall Test

All

1.3x101

208.3

192.31

192.31

Average

Others

1







1-Only 1 filter holder was used during non-blank dynamometer tests of brake
components, so the average of those values is presented for both holders.

Figure 71 presents the particle size distributions measured by the APS during the two
tunnel blank experiments. The measured size distributions are generally similar to the
brake component tests with a peak around 1.6-2 |jm. The similarity in shape between
the tunnel blanks and actual tests may be due to the exchange of particles deposited
within the tunnel and the sample airflow; the particles from previous tests that adhere to
the surfaces during "seasoning" may be entrained into the sample flow during the blank,
leaving a similar size distribution even though the total counts are far less. Appendix L
includes the tunnel blank measurements from the EEPS. The EEPS measurements
during tunnel blanks showed approximately 50% of the counts measured during CBDC
tests, and the normalized size distributions were similar between tests and blanks. For
this reason, ERG and LINK believe that the EEPS size distribution is strongly affected
by the distribution of ambient dust that can pass through the CVS intake filter.

91


-------
0.035

Tunnel Blank 1
Tunnel Blank 2

^¦N^^oitDooiDmoNcooooo^-cMLn

mtDcoocvj^ffl^oN^mNfflT-mT-
o o d t-	-^cvicoco irir^oo-^cor^

Particle Diameter Bin (um)

Figure 71. Size distribution of Tunnel Blanks as measured by APS

LINK also performed 6 zero blank experiments, in which filters were weighed, loaded
into their respective sampling systems, then removed and re-weighted to determine any
trends in weight gain or loss from handling. The weight gain of the PTFE filters during
the 6 tunnel blanks averaged approximately 0.5% of the average weight gain measured
in the PTFE filters during the CBDC tests. No corrections were applied based on the
zero blank findings (as some filters gained weight and some lost weight during the zero
blank process). The complete listing of zero blank results is presented in Appendix M.

Trends in Individual Brake Events

In addition to test level and phase level analysis, ERG also reviewed emissions for
individual brake events. The QCM measured instantaneous mass and the CPC
measured instantaneous particle count and findings of each are presented in this
section.

For these analyses, event emissions for each measurement type are calculated as the
sum of each deceleration event plus the emissions captured in the cruise or
acceleration event following the braking. This segment is added to capture any
emissions that might be released in the moments after the brake pressure is released
and the friction materials move apart. The QCM data was processed similarly to the
method presented in the Results by Speed Segment section; the QCM was used to
estimate the emissions percentage of the 100S4-measured total PM2.5 across each
brake event. These analyses present only findings for PM2.5 so no factor was used to
estimate PM10. The CPC particle data was used directly as measured by the
instrument.

92


-------
ERG reviewed plots of the by-event emissions from both instruments as compared to
parameters of the event, including the 4 parameters used in test cycle development.
ERG reviewed emissions trends as a function of:

•	Average vehicle speed during braking event

•	Average event deceleration rate

•	Brake rotor/drum temperature

•	Brake event duration

•	Total event braking energy

•	Average event braking power

ERG reviewed plots of QCM and CPC data against each of the above parameters for
every combination of model, axle, and pad material. No trends were apparent for
deceleration rate, and a very weak trend was observed with temperature. Stronger
trends were observed for speed, duration, energy, and power. This section presents the
results overall for both measurement types across the five parameters in cases where
trends were observed. In general, the QCM data appeared to have more noise than the
CPC. The QCM appeared to have more noise in both measurement and time alignment.
The time alignment noise is largely due to the QCM having a 1 -minute time resolution
during operation. This resulted in emissions responses that would appear to sometimes
lead and sometimes lag the corresponding brake event, making complete time
alignment impossible on a by-event basis (though brake events were typically on the
order of 1 minute apart). In contrast, the CPC data responded favorably to time
alignment. So, in the plots in this section, the CPC data has less overall observed noise.
The following by-event plots include all braking events from all CBDC tests in this
program. Note that all plots reflect single wheel emission rates.

Figure 72 presents the total brake event particle count against the braking event
average speed. Power equation fits were performed by each friction material and are
presented in the plot; the LM pads show the highest slope of the three materials. Figure
73 presents the corresponding plot for total event QCM-measured mass by brake event
average speed.

93


-------
l.E+11

l.E+10

l.E+09

l.E+08

S l.E+07

a.

u

l.E+06

l.E+05

l.E+04

l.E+03

x. .

. ++ *)

y = 10911x2-2512

y = 11094X2088
y = 13 5 9 2x2 0735

•	OES-NAO

~	After-NAO
+ After-LM

= Power (OES-NAO)
¦ — Power (After-NAO)
Power (After-LM)

1000

1	10 Event Avg Spd (kph) 100

Figure 72. Trends in CPC count against braking event average speed, categorized by

friction material

100

10

£

0.1

y 0.01

0.001

0.0001

0.00001

y = 0.0258x°-597
y = 0.0144X0-557
V = 0.0122x° 5861

* ~

•	OES-NAO

~	After-NAO
+ After-LM

=>Power (OES-NAO)
» —Power (After-NAO)
Power (After-LM)

1000

1	10 Event Avg Spd (kph) 100

Figure 73. Trends in QCM brake event PM mass against braking event average speed,

categorized by friction material

94


-------
Figure 74 presents the total brake-event CPC count vs brake rotor temperature (note
the linear scale of x-axis). Linear fits are applied to the data for reference and to indicate
the gradual upward trend. The equations are not given, however, as the R2 fits were all
less than 0.05. Figure 75 presents the QCM-measured total brake event PM2.5 mass
against average event rotor temperature. As with the CPC, linear trendlines are
presented to show the very shallow increase in emission rate with temperature, but the
fits have very low R2 values. The temperature data in general did not show the level of
responsiveness in emission rates as cited in literature. This is likely due to the relatively
low temperatures used in this study (that is primarily intended to be representative of
normal on-road use). The gradual uptrends in these plots may also be subjected to the
confounding factor that the higher-temperature events are likely to be of higher intensity
(since the rotor temperature is driven primarily by braking energy). So, it is difficult to
draw meaningful conclusions of the effect of brake temperatures on on-road PM
emission rates.

l.E+ll

l.E+10

l.E+09

l.E+08

¦*->

C

o l.E+07

u

u

Cl
<_)

l.E+06

l.E+05

l.E+04

l.E+03

Figure 74. Trends in total braking event CPC count against average rotor temperature,

categorized by friction material

WiT.-r. .y- •.

• •

~ » ~

•	OES-NAO

~	After-NAO
+ After-LM
=»Linear (OES-NAO)

— Linear (After-NAO)
• Li ne a r (Afte r- LM)

50

100	150	200

Event Avg Temperature (C)

250

300

95


-------
100.00

QJD
£

c5

10.00

1.00

0.10

0.01

0.00

CL>

> o.oo

0.00



50

100	150	200

Event Avg Temperature (C)

•	OES-NAO

~	After-NAO
+ After-LM

=> Linear (OES-NAO)
» — Linear (After-NAO)
Linear (After-LM)

Figure 75. Trends in QCM brake event PM mass against braking event average rotor
temperature, categorized by friction material

96


-------
Figure 76 presents the by-event total particle count plotted against braking event
duration. In this plot, it appears that there is less scatter in the data at the longer
durations; however, it is likely that this is only because the events with durations over
approximately 10 seconds are extremely rare in the tests performed over the CBDC.
Analysis of the QCM -by-event PM mass data did not resolve an apparent upward or
downward trend in emissions over duration, so that plot is not presented.

Figure 76. Trends in CPC count against braking event duration, categorized by friction

material

l.E+ll

l.E+10

l.E+09

l.E+08

l.E+07

l.E+06

l.E+05

l.E+04

l.E+03

1

10

Event Duration (s)

y = z.esE+oex1-371^

= ^yE+oex1-301^0

= ZJBE+Oex1-46^

•	OES-NAO

~	After-NAO
+ After-LM

<	'Power (OES-NAO)

— —Power (After-NAO)
Power (After-LM)

97


-------
Figure 77 presents the by-event total particle count against event total braking energy.
The energy calculated in these plots does account for the differences in vehicle mass (it
is not calculated purely on the speed change).

l.E+ll

l.E+10

l.E+09

l.E+08

C

o l.E+07

u

u

Q.

(_)

l.E+06

l.E+05

l.E+04

l.E+03

0.01

y = 616429X1-4108
&++. *+ ~ y = 483402X1-4817
^++ . ++

= 409392x1-:

•	OES-NAO

~	After-NAO
+ After-LM

< >Power (OES-NAO)
— —Power (After-NAO)
Power (After-LM)

0.1

100

1000

1	10

Event Total Energy (kJ)

Figure 77. Trends in CPC count against braking event total braking energy, categorized

by friction material

98


-------
Figure 78 presents the by-event PM mass against the braking event total energy. As
with the presentation of QCM mass against vehicle speed, it appears that the QCM was
less responsive to totai braking energy; this could be due to the noise in the instrument,
especially in the time domain.

100

00

E

10

+ * y = 0.0738x°-3832

^ = 0.0405x°-3829

0.01. .~*

~	• •	* **~ p*

* * • *.~ V

; v*	\ . *•

~ ~ • * * ~ ~ ~ +

0.0001

0.00001

y =*0.0372x0-3566

•	OES-NAO

~	After-NAO
+ After-LM

~ Power (OES-NAO)
" — Power (After-NAO)
— Power (After-LM)

0.1

10

Event Total Energy (kJ)

100

1000

Figure 78. Trends in QCM braking event PM mass against braking event total energy,

categorized by friction material

99


-------
Figure 79 presents the by-event total particle number plotted against the average
braking power for the event. It is interesting to see that the event duration, energy, and
power all show identifiable responses in emissions. Given that the average power is the
energy divided by the duration, it is not completely expected that power would also
show such a clear trend.

l.E+ll

l.E+10

l.E+09

l.E+08

o l.E+07
u

(_>

Q.

U

l.E+06

l.E+05

l.E+04

l.E+03

•	OES-NAO

~	After-NAO
+ After-LM

< >Power (OES-NAO)
™ « Power (After-NAO)
Power (After-LM)

0.01

y = 3E+06X1-8048

"= 3E+06X1-3482



• v?

~

•?*

~

• «

* M

y = 2E+06X1-7085

	

0.1

10

100

Event Avg Power (kW)

Figure 79. Trends in CPC count against braking event average braking power,

categorized by friction material

100


-------
Figure 80 presents the QCM-measured event PM mass against average braking event
power. As with previous plots, the responsiveness appears reduced as compared to the
CPC, but there are definite upward trends as shown in the power trendiines.

100

10

i

tso
E

V 0-1

00
03

2

y 0.01

a

4—>

c



^ 0.001

0.0001

0.00001

0.

Figure 80, Trends in QCM brake event PM mass against braking event average power,

categorized by friction material

It is not immediately clear what differences may have caused the relatively low
correlation between emission rates and temperature in this work. Earlier studies, in
which brake events were measured using matrix-style cycles instead of representative
driving cycles, may have had a confounding factor that the type of braking events
necessary to achieve high temperatures would necessarily be high energy events, and
these two aspects were correlated in the previous works. There could also possibly be
trends in temperature at brake temperatures greater than those encountered in this
study, but the Caltrans data suggested that this would be encountered very rarely in on-
road use. It is apparent that, of the parameters evaluated in this study, the energy and
power are the most correlated to emission rates.

WTLP-Brake Tests

The test matrix included four tests over the WLTP-Brake cycle for the possibility of
future benchmarking to other test programs that use that test cycle. Results are
presented here primarily to allow for comparison between the WLTP-Brake and the
CBDC results. Figure 81 presents a comparison of average emission mass results from

y = 0.0922x°-:

y = 0.0538X0-5972
%

* * ' v \ <* i~> ,vx."	+ •

*. •• y:l•

• • • «v r&n +C •

% • ~ ~ + ~ j

01

0.1

10

Event Avg Power (kW)

100

101


-------
the CBDC to two replicate tests over the WLTP-Brake for each of the front OES-NAO
friction materials for the Camry and F-150. The WLTP-Brake emissions for the Camry
are approximately one third of that over the CBDC; while for the F-150, the results from
the two cycles are much more similar.

2.5
E

CD

1.1

E
o

015

147mm PTFE PM10 (mg/mi)

100S4 Sg3 (PM2.5-10) Emission mg/mi
1100S4 Sg4 (PM1-2.5) Emission mg/mi
1100S4 Aft Filter (PM < 1) Emission mg/mi

II II

,G^

G*







*-e







v<£



G*



G®1





y



Figure 81. Comparison between the CBDC average results to replicate tests of the WLTP-
Brake for the OES-NAO materials for the Camry and F-150 front axles

TEM Grid Loading

LINK attempted to collect sample on TEM grids using the Partector during almost all
brake dynamometer tests. However, the system did not always successfully collect a
fully loaded sample. Figure 82 presents results for each of the 75 TEM grid loading
attempts; the x axis presents the time until loading was either manually or automatically
stopped, and the y axis shows the indicated saturation level at the time sampling was
stopped. Sixty six percent of the grids were loaded to more than 90 percent saturation,
and half of the filters were loaded to 100 percent.

102


-------
• MM • .

* • •J

A M a A ¦ _





w

WW

•

• •

















•



•





• m-\









»







t

*1 •















• •

•





*









100%

-	90%

« 80%
g 70%

£ 60%

I 50%

£ 40%
i2 30%

| 20%

-	10%

0%

0	100	200	300	400	500

TEM Grid Loading Duration (mins)

Figure 82. Partector-indicated saturation level plotted against the duration of loading

time during the CBDC

Discussion

Brake dynamometer testing for drive-cycle based representative particulate emissions is
a relatively new concept and few standard procedures existed at the time of the
initiation of this project. During this project, ERG and LINK worked to modify existing
exhaust emissions and brake durability testing procedures to create new appropriate
and effective brake PM emissions test procedures. Some of these new procedures have
already been adapted for use by the PMP within the Ell's JRC as they develop brake
PM related procedures in parallel with this work.

One early project goal was to be able to perform component installation, burnishing,
emissions testing, and assembly removal within a 24 hour period. This goal was created
to maximize the number of tests that could be completed within the limits of the project
budget. This goal was achieved as LINK was able to conduct almost all tests within this
amount of time, and LINK staff could keep a relatively consistent daily schedule
throughout the work. In general, testing was reliable and no significant changes to the
project plan were needed once testing began.

The results in the previous section show that the test setup was sensitive to the various
parameters varied in the test matrix. The different vehicle platforms, brake pad
materials, and test weights resulted in appreciable differences in PM mass emissions.
This section aims to evaluate and discuss the experimental design, test methods
employed, and test results.

103


-------
Instrument Agreement in Mass Measurements

Generally, there was good agreement between the 100S4 and the 47mm Teflon filter
mass measurement. Figure 83 presents a plot of the 100S4 vs the Teflon filter mass
measurements, categorized by all emissions tests and the tunnel blanks. Note that two
tunnel blanks were performed. In each, a single measurement of the 100S4 took place,
and PTFE filters were placed in both legs of the filter system for parallel measurement.
Both results for each blank are included as separate points with the same 100S4 value.

• Emissions Tests ~ Tunnel Blanks

M 1000

T3
0)

100

tc
a>

<3-

i/i

o
o

10



••

.

•

•





~

~~

~





1	10	100	1000

47mm PTFE-Measured Test Mass (mg)

Figure 83. Agreement between 100S4 mass measurements (Y-axis) and 47mm PTFE
mass measurements (X-axis) for all tests and tunnel blanks

Evaluation of the Burnish Procedure

The burnish procedure and dynamometer cycle were developed by ERG specifically for
this work. The development of the cycle involved applying engineering judgement to
early PMP burnish guidelines. This section is presented to evaluate the burnish
procedure to determine the effectiveness of the cycle and inform any changes that may
improve future testing. During burnish cycles, LINK operated the APS, EEPS, and
logged brake temperature and hydraulic pressure data. LINK also ran the CPC during
some burnish cycles but did not run that unit during all test days to reduce the amount
of cleaning and maintenance required and prevent any delays to the test schedule.
LINK did not run the QCM during the burnish cycle as that would have risked the unit
becoming overloaded and losing data during the subsequent test cycle. So, plots of
particle size, particle count, and the logged parameters of hydraulic pressure, speed,
and brake temperature are available to evaluate the burnish cycle. For brevity, not all
combinations of plots are shown; a representative plot is presented in this section for
each vehicle model. The burnish cycle can be evaluated by determining where in the

104


-------
cycle the measured values become stable (given that the cycle is made up of many
repeats of the same speed trace)

Camry. Figure 84 presents the cumulative particle count measured during the burnish
cycle on Test Day 59. This test day was the Camry rear assembly, equipped with
aftermarket low metallic pads. The CPC trend stabilizes around 18,000 seconds into the
test cycle (approximately 40% through the cycle).

0	10000	20000	30000	40000

Cycle Time (s)

Figure 84. Cumulative CPC Particle Count Measured during a burnish of the Camry rear

Aftermarket LM pads

105


-------
Civic. Figure 85 presents the cumulative particle count measured during the burnish
cycle on Test Day 84. This was a burnish of the Civic front assembly with OES-NAO
friction materials installed. The particle count trend appears to stabilize at approximately
13,000 seconds (approximately 32% through the cycle).

10000

20000	30000

Cycle Time (s)

c

3

o
u
u

Q_

u



£

u

5E+12

4.5E+12

4E+12

3.5E+12

3E+12

2.5E+12

2E+12

1.5E+12

1E+12

5E+11

0
0

40000

Figure 85. Cumulative CPC Particle Count Measured during a burnish of the Civic front

OES-NAO pads

F-150. Figure 86 presents the calculated brake effectiveness over the burnish cycle of
the F-150 front assembly during Test Day 63. Brake effectiveness is calculated by the
deceleration torque divided by the brake hydraulic pressure and is proportional to the
coefficient of friction between the pads and rotors. Each point on the plot represents the
average effectiveness during a single braking event. This burnish took place with
aftermarket metallic pads installed. The trend in effectiveness appears to stabilize at
around 22,000 seconds (approximately 53% through the cycle). Note that the values are
only an approximation of coefficient of friction and so they have a repetitive scattering
across the different braking events. This is likely because there is some internal
resistance in the hydraulic system that must be overcome such that the effectiveness is
disproportionately lower in low intensity braking events (i.e. a larger proportion of the
hydraulic pressure is used only to move the caliper pistons and pads prior to making
significant contact with the rotors).

106


-------
0.65	140

0	10000	20000	30000	40000

Cycle Time (s)

Figure 86. Calculated brake effectiveness (proportional to coefficient of friction) during a
burnish of the F-150 front aftermarket metallic pads

Prius. Figure 87 presents the calculated brake effectiveness during the burnish of Test
Day 49. This was a test of the front assembly with aftermarket-NAO friction materials
installed. The effectiveness trend appears to stabilize at approximately 10,000 seconds
into the cycle (24% through the cycle). The burnish of the Prius was a point of concern
given that the burnish cycle was developed based on an assumed total amount of
braking energy required. Given the simulation of the Prius' regenerative braking system,
the friction materials for this vehicle did not receive the same amount of relative burnish
energy as did the materials of the other vehicles. To further investigate this concern,
Figure 88 presents another plot for the Prius, the cumulative particle count during the
burnish for Test Day 27. This test day was a test of the rear assembly with OES-NAO
materials. The figure indicates that the stabilization took place later in the cycle than that
shown for the previous vehicles. It appears to stabilize at around 35,000 seconds (85%
though the cycle. In future tests, to mitigate this potential concern, it may be beneficial
to turn off the regenerative braking simulation during the some or all of the burnish cycle
of regenerative braking-equipped vehicles.

107


-------
0.5

140

0	10000	20000	30000	40000

Cycle Time (s)

Figure 87. Calculated brake effectiveness (proportional to coefficient of friction) during a
burnish of the Prius front aftermarket NAO pads

7E+11

6E+11

5E+11

4E+11

3E+11

2E+11

1E+11

100

Figure 88. Cumulative CPC Particle Count Measured during a burnish of the Prius rear

OES-NAO pads

10000	20000	30000

Cycle Time (s)

0

40000

108


-------
Rogue. Figure 89 presents particle size data for the burnish cycle of Test Day 38. This
test was a front-assembly test of the aftermarket-NAO pads for this vehicle. The APS
size distribution (color coded by particle count) is the upper plot, and the EEPS side
distribution (also color coded by count) is the lower plot. APS data appears to stabilize
around 25,000 seconds (60% of the cycle), and the EEPS appears to stabilize around
30,000 seconds (73%).

20

APS

1.E+04

£

ZL

"tr 5

Ol
+-»

CD

I "

H3

5

V
u

t 1
as

£L

0.5 r—r
0

! (

5000

10000

15000

20000 25000
Seconds

EEPS

i i—i—i—i—i—r—t—J—i—r—t—i—|—r

30000 35000 40000

10000

15000

20000 25000
Seconds

30000

35000

40000

1.E+03

1.E+02 *
z

Q.

1.E+01

1.E+00

1.E+08

1.E+07

1.E+06

1.E+05 „
E

1.E+04 |
z

1.E+03 °"
1.E+02
1.E+01
1.E+00

Figure 89. Particle size data during the burnish cycle of the Rogue front aftermarket-NAO
pads. The upper plot presents the APS result and the lower presents the EEPS.

109


-------
Sienna. Figure 90 presents particle size data for the burnish cycle of Test Day 13. This
test was a rear-assembly test of the OES-NAO pads for this vehicle. The APS size
distribution is the upper plot, and the EEPS side distribution is the lower plot. The APS
data appears to stabilize around 20,000 seconds (49% of the cycle), and the EEPS
appears to stabilize around 27,000 seconds (66%).

20

APS

1.E+04

I 10

01

«5 5
E

TO

Q 2.5
_0J
u

*+-'

L_

CO

1.E+03

1.E+02 *
z
a

1.E+01

0.5

0.005

0

Figure

5000

10000

15000

20000 25000
Seconds

30000

35000

40000

EEPS

5000

10000

15000

20000 25000
Seconds

30000

35000

40000

1.E+00

1 E+08

1.E+07

1.E+06

1.E+05 „
E

1.E+04 |

Z

1.E+03 °-
1.E+02
1.E+01
1.E+00

90. Particle size data during the burnish cycle of the Sienna rear OES-NAO pads.
The upper plot presents the APS result and the lower presents the EEPS.

The burnish cycle appears to have included a large enough amount of braking energy
for all friction materials to reach a stable condition from the various metrics presented in
this section. The Prius rear assembly was a point of concern because it appeared to
reach a stable level just as the burnish cycle was ending. This concern could be
mitigated by disabling regenerative braking simulation during burnish of this type of
vehicle in future testing. For all other vehicles, stabilization appeared to be reached at or
before 30,000s into testing. So, the burnish cycle could potentially be reduced in length
by up to approximately 3 hours and still achieve the desired level of stability for
conventional-braking vehicles.

The CPC plots indicate that particle count emissions are elevated during the early
stages of burnish after installation. Given that the duration of time that this elevated
emission rate was observed (-5-6 hours typically), it is not likely that the burnishing
process in-use has an appreciable effect on the overall emissions inventory given that

110


-------
brake pads typically last for thousands of hours of operation, therefore this portion of
brake service will not appreciably affect inventory values.

Evaluation of the Prius Regen Simulation

LINK used their internal 'DutyCycleRegen' program for the simulation of the Prius on the
dynamometer. Given that brake dynamometer testing for emissions is a relatively new
field, and the Prius (and most other regenerative braking equipped vehicles) operate
using a proprietary algorithm, there is not a common or recognized procedure or
approach for simulation of the regenerative system on a brake dynamometer. This
section aims to evaluate the accuracy of the regenerative system used in this work. The
Prius ETW was 1,606 kg in this work, and the Camry was the closest vehicle in weight
at 1,668 kg ETW and so it is presented as the non-regenerative benchmark for
comparison.

The first comparison is between the overall average brake torque measured during all
dynamometer tests. Table 26 presents the overall average brake torque for Camry and
Prius for front and rear assemblies. For direct comparison, only OES-NAO and
Aftermarket-NAO pads are included (as the Prius was not tested with metallic pads). It
can be seen that the Prius braking torque averages less than half of that of the Camry
due to the simulation of the regenerative braking system.

Table 26. Comparison of Overall Average Measured Brake Torque for Camry and Prius

Test Vehicle

Avg. Front Brake

Avg. Rear Brake



Torque (nm)

Torque (nm)

Camry

138.4

48.3

Prius

55.9

20.5

Prius % of Camry Torque

40.4%

42.5%

Figure 91 presents an example instance of the source of the difference in torque trends
between the Camry and Prius. The figure depicts a selected interval of the CBDC test
cycle and presents traces of the speed and front brake hydraulic pressure for the two
vehicles. The figure shows an example of how the hydraulic brake pressure differs
during the same deceleration event, even approaching zero during moments that are
most favorable to the regeneration system operation based on the regeneration
parameters presented previously.

111


-------
Prius Speed	Camry Speed = = — Prius Brake Press ----Camry Brake Press

60

1400

50

Q.

— 40

"O
CD
<3J
Q.

00

o5 30

4—'

<1>

E

0

1	20

c.

-

Q

10

0 —
33900

Figure 91. Speed and Brake Pressure traces for a selected point of the dynamometer test
cycle for a test of the Camry and a test of the Prius

Another method to evaluate the accuracy and effectiveness of the regenerative system
simulation is to make a comparison between the test track experiments and the
dynamometer tests. On the test track, all models were driven over the WLTP-Brake
cycle to inform the dynamometer cooling airflow setting process. During the cooling air
flow setting, the WLTP-Brake Trip 10 was used as the test cycle. Comparing the brake
heating trends between the dynamometer and the test track over Trip 10 can give an
indication of the accuracy of the regenerative braking simulation. Figure 92 presents
temperature traces for the front brake assemblies of the Camry and Prius operating
over the WLTP-Brake Trip 10 on the test track and brake dynamometer. It can be seen
that the rate of brake heating necessary to follow the same trace is much lower for the
Prius than for the Camry. It can also be seen that the heating rates for the Prius are
very similar between the test track and brake dynamometer, indicating that the
regenerative simulation results in similar brake usage on the dynamometer as on the
test track. Figure 93 presents similar data for the rear brake assemblies of both
vehicles, and similar results are observed.

1200

1000

800

600

400

200

- 0

33920 33940 33960 33980 34000 34020 34040 34060 34080 34100

Cycle Time (s x 10)

112


-------
Camry Front Track Temp
• Prius Front Track Temp
¦Camry Front Dyn Temp
¦ Prius Front Dyn Temp
Speed



1000

4000

5000

2000	3000

Cycle Time (s)

Figure 92. Temperature Traces of the Front Brakes of the Camry and Prius overWLTP-
Brake Trip 10, operating on the test track and brake dynamometer

80

60

40

20

Camry Rear Track Temp

•	Prius RearTrackTemp

•	Camry Rear Dyn Temp

•	Prius Rear Dyn Temp
Speed

1000

2000

„ , 3000 ,
Cycle Time (s)

4000

5000

Figure 93. Temperature Traces of the rear Brakes of the Camry and Prius over WLTP-
Brake Trip 10, operating on the test track and brake dynamometer

Reference Tests

The test matrix included additional replicate tests of one configuration to be run
periodically throughout the program to potentially identify any sources of drift in the
testing system. The reference test was chosen to be the front axle assembly of the F-
150, equipped with OES-NAO friction materials and operating at ETW test weight.
Figure 94 presents mass emission and particle count measurements by date. The
original intent was to distribute the reference tests approximately equally throughout the

113


-------
program, however the 4th test was swapped into the middle of the testing program due
to a test that was scheduled as a WLTP-Brake being inadvertently run as a CBDC test
and therefore matching the reference test parameters.

1.8
1.6

E 1.4

"ao
P

1.2

a>
+-»

tc
QC

C

o

1.0
0.8
0.6
0.4
0.2

0.0
9/22/2019

¦









~



~





•	





~



# t

	





¦ i

		

¦























• 47mm PTFE PM10









~ 100S4 PM10









¦ CPC Particle #



















1.4E+09

1.2E+09

1.0E+09

8.0E+08 £

o
u

6.0E+08 -5?

u

4.0E+08

2.0E+08

10/22/2019

11/21/2019
Test Date

12/21/2019

tc

CL

0.0E+00
1/20/2020

Figure 94. Reference trends by date for the 47mm PTFE and 100S4 PM10 mass
measurements (left axis); and CPC particle count (right axis)

Trendlines fit to the test date are included for each of the three sets of values. Both
mass measurement types have a minor downward trend over time. This appears to be
driven primarily by the first test, which has the highest values for all three measurement
types. From the time of the second test date, there is no discernable down trend in
either mass measurement. The CPC trend has a steeper apparent down trend; the first
value is the highest and the final value is the lowest. However, for both the mass and
particle count plots, the noise in each variable appears to be equal or greater in
magnitude than the strength of the trends. For this reason, ERG did not apply any
corrections to the data based on test date.

Issues encountered

In early November, the humidity controller of CVS airflow loop failed, causing the
humidity to eventually exceed the test limits (Limits are 50 +/-10 %RH). LINK ordered a
new control board to solve the problem. In the meantime, LINK continued running only
the burnish cycle with the existing humidity controller for multiple upcoming tests without
running the test cycle. This interim schedule lasted for approximately 1 week up until the
humidity controller was repaired. After this, LINK re-mounted the previously burnished
brakes, ran a partial burnish cycle (1 hour) to stabilize the contact surface of friction

114


-------
couple, and measured emissions during the actual test cycle. This issue resulted in
slight downtimes to re-assemble tested parts and run the test cycle separate from the
burnish cycle.

LINK encountered another issue with the dynamometer control computer that aborted a
test while conducting the burnish cycle. After troubleshooting this computer with limited
staff during the holiday schedule, it was replaced with a new computer. This issue
caused approximately 5 days of delay in the test schedule.

Comparison of Results to Literature

Studies in literature cite a variety of brake PM emissions test procedures and test
cycles. As the concept of PM emissions testing for environmental protection is relatively
new, there are few standard procedures and a wide variety of estimates of braking PM
emissions factors. ERG reviewed a study that identified a range of published light-duty
brake emissions rates.11 This range of results, converted to mg/veh*mi for comparison
to this study, is presented in Table 27. The results for this study tend to be somewhat
higher than literature, though there is significant overlap in both sets of ranges.

Table 27. Ranges of vehicle-level PM emission rates (mg/mi), summarized in literature

and for the vehicle models in this study



Literature Range for
Overall LD Emissions
Factor (mg/mi)

Range of Emission
Rates by Model in this
Study (mg/mi)

PM2.5

00

1

CO

1.9-5.4

PM10

1.9-5.0

3.3 -13.9

Figure 95 presents a different review of past studies of brake wear emission rates of
PM2.5 and PM10 for light- and heavy-duty vehicles.12 The overlaid rectangles represent
the range of emission rates by model in this study. In the figure, squares represent
heavy-duty vehicles and circles represent light-duty vehicles. The figure shows a
greater level of agreement with literature than the study in the previous table. Notably,
the findings from this study are lower in both PM2.5 and PM10 than values currently in
EMFAC. The values for PM2.5 overlap with MOVES2014 but are lower than MOVES for
PM10.

11	Grigoratos, Theodoras & Martini, Giorgio. (2014). Brake wear particle emissions: a review.
Environmental science and pollution research international. 22. 10.1007/s11356-014-3696-8.

12	Sonntag et. al. Modeling Brake and Tire Wear Emissions in Regulatory Models in the United
States, 2018 ISES-ISEE Joint Annual Meeting

115


-------
Brake wear emission factors reported in the literature

100 r

JZ

o
>

cc

(/)
C.

o

t/1

S

tu

10r

lr

Type of Study
0 US Reg Model

#	Brake dynamometer

#	Emission Inventory

#	Receptor modeling

Vehicle Class
¦ Heavy Duty

#	Light Duty

EMFAC2017

\bu-Allaban (200})*

| MOVES2014.il

EMFAC2017

EMFAC2017

|-UK

•t

Abu-Allaban (200})*

|MOVES20i4J^ Abu Allaban (2003)

P /

EMFAC2017 |

IMOVES2014-4
Boulter (2006)

^ >Luhana (2004)

Sanders (2003)-

li|lma (2008)

JL	

Carg(2000J* Abu-Allaban (200;

lljima (2008)
Garg (2000)'

, •flu

r

P Bfeult

.' _ \

ukowiecki (2009)

NAEI (2012)

	P

[MOVES2014J |

ulter (2006)
Barlow (2007)*

L0ekewllle(2OOt)* \

\.

*

Bukowierki (200$)
Rautcrberg Wulff (1999)

Hagino (2016)* "

PM2.5	PM10

Figure 95. Various literature values for brake emissions, with the ranges from this study

overlaid for comparison

PM emission rates cited in literature do appear to be trending down with time. Older
emissions models may have been reliant on by-event type brake measurement with
much higher temperatures (up to 400 °C), higher deceleration rates, and higher speeds.
The prevalence of metallic friction materials is decreasing in the fleet with time and are
therefore likely to be less prevalent in recent research. Also, friction material and rotor
life has extended significantly in recent years, with current (OES and good quality
aftermarket) formulations averaging 50-90k miles of life; based on the presented
component mass loss vs PM emission plots, the longer life is indicative of lower PM
emission rates.

Potential MOVES Emissions Factors

The Results section of this report presented emissions rates on the basis of grams per
mile of represented distance traveled. For comparison to existing emissions factors in

116


-------
MOVES, however, it is useful to present the emissions measured over the test cycle on
a time basis. The Test Cycle Development section introduced the concept of the
CBDC's represented distance, which is the distance of actual on-road travel that is
represented by the braking events in the cycle when accounting for cooling and
extended on-road cruises that were abridged to minimize the time required to run the
cycle on the dynamometer. Just as the events that make up the CBDC have a
represented distance, they also equivalent^ have a represented time, both for the
overall cycle and for each speed segment. These times, along with the represented
distances, are presented in Table 28 (in U.S. Customary units) for the Overall CBDC
and the speed segments as well as the resulting calculated average speed. These
values can be used to convert the previously given emissions per distance traveled to a
time basis representing an overall trip.

Table 28. Represented time and represented distances for the overall CBDC and each

speed segment

Cycle/Segment

Represented
Distance (mi)

Represented
Time (s)

Represented Average
Speed (mi/hr)

CBDC Overall

81.56

11,564

25.39

0-21 kph

3.83

2,911

4.73

21-69 kph

47.31

5,438

19.46

69+ kph

48.33

3,215

54.12

For comparison with MOVES factors, however, it is of more interest to isolate the
analysis from the overall test cycle (i.e. driving trip) down to time spent in actual braking.
MOVES has various operational modes (Opmodes) for the purpose of binning
emissions rates depending on the activity of a vehicle during a given second of
operation. Some of the main opmodes are those for cruise/acceleration, which bin
emissions factors based on the instantaneous level of engine power output on a
vehicle-specific power (VSP) basis. MOVES also has a single opmode for braking,
opmode 0, In general, the braking opmode is active when modeling deceleration rates
either greater than 2 mph/s over 1 second or greater than 1 mph/s over 2 continuous
seconds (the braking opmode also includes provision for braking over road grades
equivalent to those levels of deceleration, however this project did not include any
simulation of road grade). For comparison to MOVES, it should be noted that the CBDC
was limited to not include any 1s or 2s braking events.

For comparison with the MOVES Braking opmode, braking emissions in this work must
be presented as the total cycle emission rate divided by the total number of seconds of
braking. For the CBDC, the conversion from mg/mi to mg/s (of braking) can be
performed by multiplying by the factor of 0.052 [mi/shaking], which is the cycle
represented distance divided by the 1564 total seconds of braking in the CBDC. For this
approach, it is not necessary to consider the cycle represented time, only the seconds
of braking as the CBDC contains actual in-use braking events representative of real-
world operation. However, the cycle does include some low-deceleration braking that
would not fall into the requirements of opmode 0. The in-use data indicated that very
light braking was performed frequently (even if accounting for vehicle coastdown
deceleration rate). For this reason, approximately 60% of the braking events in the

117


-------
CBDC falls in the category of opmode 0 and the remaining braking takes place at a
deceleration rate less than 1 mph/s (Figure 14 presented the distribution of deceleration
rates for the CBDC, note the units of that distribution are in kph).

Table 29 presents the vehicle-level emission rates on the basis of mass per second of
braking. These values are based on the emission rates given previously in Table 22,
corrected based on the factor of represented distance divided by seconds of braking.
Values are presented based on measurements made by the 100S4 system so that
factors can be generated for both PM2.5 and PM10. The values in the tables are based
on batch gravimetric measurements made over the entire CBDC. Given that the test
cycle is built specifically to reflect actual in-use light-duty braking, this is an appropriate
approach for comparison to MOVES emissions factors. Appendix H, referenced earlier,
includes overall tabulated test results including the average PM emissions rates per
second of braking for each individual test.

Table 29. Estimated in-use brake emission rates per unit time of braking by model

Model

Estimated PM2.5

Estimated PM10



Emission Rate per

Emission Rate per



Second of Braking (mg/s)

Second of Braking (mg/s)

Camry

0.177

0.486

Civic

0.114

0.296

F-150

0.212

0.512

F-150 HLW

0.266

0.723

Prius

0.099

0.174

Rogue

0.203

0.478

Sienna

0.218

0.514

Sienna HLW

0.281

0.723

The average of the vehicle models in the table correspond to approximately 0.71 g/hr,
which is somewhat higher than the MOVES2014 average PM2.5 braking emission rate
of 0.558 g/hr.13 However, a simple unweighted average of the vehicle models used in
this program is not necessarily representative of all vehicles represented by MOVES.
Also, MOVES braking emissions are also modeled, in part, in the coasting bins so the
modeled total may be closer to the average for this work.

Given that the test cycle represents in-use driving weighted appropriately across trip
average speeds in-use, it is preferable to use the above cycle-level emissions rates for
comparison to existing MOVES factors as compared to the continuous measurements
of the QCM system. This also has the co-benefit of minimizing the effect of the greater
error level of the QCM system as compared to the 100S4.

13 US EPA. Brake and Tire Wear Emissions from On-road Vehicles in MOVES2014. EPA-420-
R-15-018, November 2015.

118


-------
Informing Braking Emissions as a Function of Negative VSR

The Trends in Individual Brake Events section presented total event emission rate
against various parameters such as deceleration, temperature, and power. To further
inform brake modeling on a basis comparable to MOVES, it was also of interest to
present the continuous emissions measurement data on the basis of emission rate per
second plotted against VSP. Figure 96 presents the CPC-measured particle number
emission rate (#/s) plotted against brake event averaged VSP (kW/ton) for all brake
events in the test program. Points are colored by friction material and power fits are
presented for each material.

l.E+ll

l.E+10

l.E+09

y = 134296X20154

108405X1-9532
y = 94577X1-9234

l.E+04

l.E+03

OES-NAO

After-NAO

After-LM

Power (OES-NAO)
Power (After-NAO)
Power (After-LM)

1	10	100

Event Avg VSP (kW/ton)

Figure 96. Particle number emission rates vs brake event average VSP for all braking
events in the test program. Power fits are shown by friction material type

Similarly, Figure 97 presents the equivalent plot for the QCM measurements; PM2.5
mass emission rate (mg/s) is presented against event-average VSP. Power fits are
presented for each material type. This type of analysis may be useful in informing the
potential to expand the MOVES braking opmode into multiple negative VSP bins in a
manner that is similar to the current cruise/acceleration VSP bins.

119


-------
10.0000

£

to
oe



1.0000

0.1000

0.0100

0.0010

0.0001

0.0000

++• +*

f)t

y = 0.0266x°-3489
y = 0.0129X03623

v j„.»v % ' >	rts-jpx:

~ * * • .» ~ + . ,

# •	~	~	a	•

'*	+ — \

~	+• ~

• 	

i'

Event Avg VSP (kW/ton)

• OES-NAO
After-NAO
+ After-LM

Power (OES-NAO)
Power (After-NAO)
Power (After-LM)

100

Figure 97. QCM-measured PM2.5 mass emission rate vs brake event average VSP for all
braking events in the test program. Power fits are shown by friction material type

Modeling Emissions Deterioration

This study involved testing friction materials that were relatively new; they were
purchased new, operated over an aggressive burnish cycle, and then were tested over
the CBDC. As a result, the project did not generate emissions measurements indicating
the deterioration of friction materials over time. Even if that effect had been quantified,
vehicles periodically have their friction materials replaced with completely new
components as the vehicle ages, potentially "resetting" the brake PM emission rates as
compared to the age of the vehicle. However, the trend in change of installed friction
materials was presented in the Representative Test Vehicle and Friction Material
Selection section. Trends in the change in installed friction materials overtime can
potentially be used to estimate deterioration rates in braking PM emissions as vehicles
age. LINK business intelligence was used to estimate the balance of materials at
vehicle ages of 3 and 11 years, which were presented previously in Table 21. The
business intelligence was based on vehicle age in years, and fitting a linear trend based
on emission rates by material yields yearly deterioration rates as given in Table 30.
These rates can be multiplied by the modeled vehicle age in years to adjust the
previously presented emission rates (which were estimates for 7-year-old vehicles).

Note these adjustment factors are based only on the estimated changes in installed
friction materials as vehicles age. To estimate new-vehicle levels, these factors could be
multiplied by 7 (years) and subtracted from the Table 29 emissions rates. Or, for older

120


-------
vehicles, they could be multiplied by the number of years older than 7 and added to the
estimates.

Table 30. Estimated Deterioration Rates Based on Friction Material Trend with Vehicle

Age

Vehicle Type

PM2.5 Deterioration

PM10 Deterioration



per year (mg/mi/yr)

per year (mg/mi/yr)

Conventional Passenger Car

0.008

0.056

Light Truck

0.074

0.209

Regenerative-Equipped

0.006

0.005

Summary and Conclusions

This report presented the results from PM emissions testing of light-duty brake
assemblies operating on a brake dynamometer. A new brake dynamometer test cycle
was developed to represent the operation of real-world vehicles in California based on
the Caltrans dataset. ERG adapted a previously developed algorithm to develop a test
cycle that would be as similar as possible to the speeds, deceleration rates,
temperatures, and braking durations encountered by real vehicles in the Caltrans
survey. This new cycle was divided into three speed bins such that emission rates could
be differentiated across different ranges of trip average speeds.

Six test vehicles (with common cross-platform brake components) were selected to
represent the range of vehicle types in the light-duty fleet. Up to three different friction
formulation types were tested for the front and rear assemblies of each model.

Replicate tests were conducted for each test matrix combination.

The LINK lab site included a constant volume sampling (CVS) system with an integrated
brake dynamometer. Measurements were made in batch and continuously by a variety
of instruments including gravimetric sampling in parallel on coated aluminum impactors
(TSI 100S4) as well as on 47mm Teflon filters. Instrumentation was also installed to
measure particle size distributions, particle counts, and continuous particle mass.

Results were presented for emitted PM mass, particle count, and size distribution from
the front and rear assemblies of six test vehicles. The test procedure developed for this
work was able to resolve effects on particle mass emissions from the various
parameters varied in the test matrix. Particulate mass emissions on a per-wheel-mile
basis ranged from approximately 0.5 mg/wheel*mi for the Prius rear brakes operating at
ETWwith NAO pads up to 15 mg/wheehmi for the front brake of an F-150 operating
with a simulated cargo load and equipped with a low metallic friction material.

In most cases, low metallic friction materials resulted in higher mass emissions than
NAO materials. Vehicles tested at higher simulated weights also tended to have higher
mass emissions. For a given friction material type, PM emission masses trended
linearly with vehicle test weight. Figure 98 presents the average by-vehicle emission
mass vs test weight for each vehicle and test weight combination. The F-150 reference

121


-------
vehicle tended to have relatively low PM emissions for its vehicle weight when equipped
with OES friction materials (as compared to the other vehicles).

35

30

§ 25

^ 20

P 15

T5 10

a>
>

"Hi
+-»

o







¦













F-150

¦



Camry



Sienna



W
•

Sienna

HLW^

	.»-•

~



		 p

togue





(^/ic ^
Prius





1200

1700
• OES-NAO

2200

Vehicle Tested Weight (kg)
~ AfterMkt-NAO

2700	3200

AfterMkt-LM

Figure 98. Total vehicle test cycle PM mass emissions vs simulated vehicle test weight,

categorized by pad material

ERG used LINK business intelligence to estimate the in-use friction material fractions
within each model. Using these fractions, ERG estimated the average in-use emission
rates for each model. Table 31 presents these estimated emission rates of PM2.5 and
PM10; they represent an inventory-level estimate for in-use emissions assuming an
average vehicle age of 7 years.

Table 31. Estimated in-use brake PM emission rates by model (7 years old)

Model

Estimated In-Use PM2.5

Estimated In-Use PM10



Emission Rate (mg/mi)

Emission Rate (mg/mi)

Camry

3.4

9.3

Civic

2.2

5.7

F-150

4.1

9.8

F-150 HLW

5.1

13.9

Prius

1.9

3.3

Rogue

3.9

9.2

Sienna

4.2

9.9

Sienna HLW

5.4

13.9

122


-------
For the purposes of comparison to the emission factors used in MOVES for opmode 0,
the above values were also presented on the basis of mass emissions per second of
braking. Table 32 presents these values by model on the basis of mg per second of
braking.

Table 32. Estimated in-use brake emission rates per unit time of braking by model

Model

Estimated PM2.5

Estimated PM10



Emission Rate per

Emission Rate per



Second of Braking (mg/s)

Second of Braking (mg/s)

Camry

0.177

0.486

Civic

0.114

0.296

F-150

0.212

0.512

F-150 HLW

0.266

0.723

Prius

0.099

0.174

Rogue

0.203

0.478

Sienna

0.218

0.514

Sienna HLW

0.281

0.723

This report also presented emissions results for particle size distribution and particle
counts. Particle count emissions generally trended with overall mass, however the
observed range from the lowest to highest emitter was reduced as compared to
measured mass. In general, low metallic pads resulted in somewhat larger particle sizes
than did NAO pads.

Recommendations

The original project scope included a task in which ERG and LINK would use the
experience in this work to develop recommendations for future work in the development
of measurement techniques for non-exhaust PM emissions from vehicles.

Realistic Emissions Factors

The brake dynamometer system intentionally represents idealized braking particulate
emission circumstances as compared to braking systems mounted on vehicles. They
are idealized in that there is not a wheel/tire mounted, there is only the minimum
hardware necessary for support of the components in the airflow, and there are not
bodywork or vehicle frame elements in the airflow. This is intentional to maximize the
repeatability of testing and to ensure that the maximum percentage of generated
particles reach the point of sampling. The goal is to understand the particles generated
at the friction couple, so it is important that the test setup maximize the sensitivity to
particles originating at that point. In actual on-road use, however, some amount of
brake-generated particulate settles on or adheres to vehicle components and surfaces.
As a result, brake dynamometer testing may result in measurements that overestimate
real in-use braking emissions.

123


-------
ERG did not find any recent literature regarding the level of PM losses from settling. The
Sanders (2003) study indicated estimates of the fraction of PM mass remaining airborne
on a real vehicle being in the range of 50% - 70% of the emitted mass from the friction
couple.14 However, there was also some level of loss present in the dynamometer
sampling system. The Arizona dust experiment indicated recovery percentages around
90%. Accounting for this, a reasonable estimate for in-use emission rates would
be 66% of the emission factors determined in this work.

Further testing may help to inform a better understanding of the particle losses from
settling, transient behavior like coagulation (and potential increased settling as a result),
other dispersion considerations for brake particles, and temperature or meteorological
effects on vehicle emissions in-use.

LINK and ERG recommend the following testing ideas for continued investigation into
brake emissions:

Long-term effects. In this work, the tested friction materials were relatively new as they
were tested after approximately 11 hours of burnishing. While the friction materials used
in a given pad or shoe are likely to be homogenous throughout in terms of chemistry, it
may be beneficial to determine whether there is a PM emissions trend as the pad or
shoe proceeds through its life toward end-of-life. Also, the friction materials in this
program were free of corrosion or any other wear outside of that caused by the burnish
cycle. One potentially valuable subject to research in future work would be to determine
any effect of corrosion (especially for metallic materials), other environmental effects, or
mal-maintenance and mechanical problems on PM emissions. This could include stuck
or constantly wearing brakes that could be emitting at disproportionate levels as
compared to other vehicles. This could potentially be done by working with vehicle fleets
to make an agreement to acquire and/or evaluate their used brake components for
testing. A more costly option would be to conduct a full aging program in which the
aging, corrosion, wear, and/or mechanical defects would be conducted as a part of the
experimental design.

Meteorological effects. Weather can affect both driving habits as well as the
morphology of particles as they are released into the atmosphere. The Caltrans travel
survey that formed the basis for the development of the CBDC test cycle was a wide-
ranging project that would have captured various weather events across the state,
however the weather was not logged or noted so no analysis is possible on that basis.
Additionally, rain may have a large effect on mitigating PM emissions (though this would
be likely to shift air quality effects of PM to watershed effects). Understanding these
effects, taken with climate data for the state, may allow for a correction to be applied to
the brake PM emissions factors based on weather.

14 Sanders, Paul G. et. al. "Airborne Brake Wear Debris: Size Distributions, Composition, and a
Comparison of Dynamometer and Vehicle Tests." Environmental Science & Technology 37 (2003): 4060-
4069

124


-------
(Urban) Geographic location. There may also be reason to prioritize the analysis of
urban environments (and their associated driving styles) in which vehicle and
pedestrians would be grouped together in large numbers and close proximity. The test
cycle in this work represented vehicle use across the state, but PM emissions may be a
higher priority in urban areas due to population density. The low-speed portion of the
CBDC reflects the driving in low-speed trips from across the state, not necessarily those
from urban environments.

By event analysis. To further inform the understanding of the mechanisms and
physical causes of varying levels of braking events, it may be of interest to conduct
testing over a prescribed or gridded set of stops instead of the representative driving
cycle used in this program. It could resemble the Heating and Cooling Matrix used in
this work to develop the brake temperature model. A series of the same engineered
stops could be performed repeatedly and characterized as a group. This could reduce
the noise observed in the Individual Brake Event analysis in this work.

Alternative brake dynamometer setup. For dynamometer testing in which the intent is
not necessarily to attempt to capture all of the PM from the brake friction interface,
minor changes from the design used in this work may be beneficial. For example, with a
slightly larger brake enclosure, it may be feasible to mount a "dummy wheel" to the
tested brake rotor to determine whether it would collect a measureable amount of PM
(i.e. measurably less PM would reach the point of sampling). The dummy wheel could
be a typical vehicle wheel with just the spokes and inner drum, with some of the outer
drum machined off so that it would fit in the enclosure. Depending on the design, it may
also need to have a hole drilled out at the center to accommodate the dynamometer's
brake hub support. Likewise, it may be possible to add an inner fender shape to the
enclosure to disturb the airflow to determine if there is a measurable amount of particle
loss. These are speculative ideas but potentially worth investigating.

Empirical/numerical models for transport losses. LINK developed a model of the
transport losses of the CVS and sampling train setup used in this project. This was a
relatively simple model assuming little turbulence and gradual curves in sample ducting.
It may be of benefit to pursue modeling of transport losses in greater detail to create
estimates of particulate settling on in-use vehicles, though this would require a much
more complicated model than that used by LINK in analyzing the smooth surfaces of the
setup used in this work.

Heavy Duty Vehicles

This work focused on light duty vehicles ranging in size from a Honda Civic up to a Ford
F-150. For on-road inventory modeling, however, heavy-duty vehicles may be the
source of a significant amount of PM emissions from the on-road vehicle fleet. The
increased weight of these vehicles means that braking energy to stop from a given
speed is much higher than that for a light-duty vehicle. As a result, even though the
population of these vehicles is lower than that of light-duty, the total emissions may be
equivalent or even higher. One confounding factor is that heavy duty vehicles are more

125


-------
often equipped with drum brakes than their light-duty counterparts. In this study, the
Civic rear axle was drum brake equipped and emitted at lower rates than the other
vehicles, likely due to particles being trapped within the drum assembly. So, it is
possible that heavy duty brake PM emissions do not completely scale up with weight at
the same slope as the light-duty vehicles in this study.

Since the time of the project initiation, Caltrans has initiated a parallel project to
measure the PM emissions associated with heavy duty vehicle braking. One of the key
challenges to this work involves the wide variety of vehicle types and vocations in the
heavy-duty fleet. Additionally, heavy duty vehicles tend to have multiple axles (including
trailer axles) that all may have different braking characteristics and associated emission
rates. The Caltrans project will include 36 dynamometer tests from different axles of
Class 8 tractor trailers, refuse trucks, beverage haul trucks, and urban buses. The
results of this project will better inform the need for and direction for potential further
testing.

Tire Wear

The other potentially significant source of non-exhaust PM from vehicles is that of tire
wear. The tire slip between the tire and roadway during operation results in abrasion of
both sides of the friction couple. Rolling tires can also lift roadway debris and cause it to
become airborne, and it is not always possible to determine whether the source of
particulate is the tire, roadway material, or settled debris without chemical analysis. For
the sake of measurement of all PM emission caused at the tire/road interface, the
source is not important as all three directly contribute. However, if reductions are to be
made possible, awareness of the source (either the roadway material or tire material) is
critical to informing possibilities for reduction. Tire formulations generally consist
primarily of styrene butadiene rubber, natural rubber, and polybutadiene. To enhance
the engineering properties of the tire, zinc oxide is often present at levels of
approximately one to two percent, and this zinc can comprise some percentage of the
emitted particulate. Roadway surfaces vary more in material formulation, including
stone/mineral matter, bituminous binder, sand, and various binders.15

The most common research publications discussing tire and roadway PM emissions are
roadside or tunnel studies of the ambient air around vehicles traveling the roadway.
These types of experiments reflect real in-use values but do not have the same level of
control over vehicle types, tire materials, and driving styles as a dynamometer study in
which a complete test matrix can be prescribed. Because many literature references
describe roadside studies, many of the available emissions factors relate to ambient air

15 Ntziachristos and Boulter. (2013). EMEP/EEA air pollutant emission inventory guidebook -
2013: 1.A.3.b.vi Road vehicle tyre and brake wear.

126


-------
on a by-volume basis, not to a per-mile or per-second basis as needed for informing
emissions models.1617

Some studies have used a test vehicle with an onboard portable emissions
measurement (PEMS) system mounted in such a way that it draws in tire and road or
brake PM. These types of measurements generally involve a funnel-shaped capturing
device with a sample pump that draws the air and PM into the system for measurement.
Depending on the size and shape of the funnel, it can sample either tire/road interface
PM or brake PM (with some risk of cross-contamination). These types of measurements
allow for control of the test vehicle operational conditions and the tire and roadway
materials. This type of testing is also subject to environmental conditions and
contaminations, unlike the controllable conditions in a lab.

There have been studies involving various laboratory dyno systems that simulate the
tire/road interface within controlled laboratory conditions. Like the brake dynamometer,
these systems were generally developed for testing other tire parameters (such as tread
life or noise) and were then adapted to PM testing. It is challenging to develop a
representative laboratory simulation of the tire/road interface, however. Two existing
designs are the system used by the German Federal Highway Research Institute
(BASt), in which a large cylinder is fabricated, and the inside surface is made of
roadway material. A wheel and tire combination is driven around the inside of this loop
in a manner similar to a planetary gear in a gearset. Turning loads, tire scrubbing,
acceleration, and deceleration can all be simulated with this design, and the volume
within the confines of the cylindrical enclosure can be sampled for emitted PM.

Another method is a flat circular roadway surface with a centrally mounted arm that
supports a wheel and tire traveling around the roadway surface. Acceleration,
deceleration, and scrubbing can also be simulated with this design; however, the
constant turning may emit a greater number of particles due to shear than would be
emitted while traveling along a straight path. This type of laboratory setup has been
used by the Swedish National Road and Transport Research Institute (VTI).

There is also the design that operates more like a light-duty chassis dynamometer in
which a wheel and tire rides on the outside of a large textured drum, such as a 48"
dynamometer roll. This interface can be mounted within an enclosure and the generated
particles drawn and sampled from that enclosure. Because the tire is riding against a
round surface, the tire stresses, deformation, and likely PM emissions would be higher
than that from driving along a flat surface.18 Some adjustment factor to account for this
would be necessary. Of the various laboratory designs, this design appears to be the

16	Panko, Julie & Hitchcock, Kristen & Fuller, Gary & Green, David. (2019). Evaluation of Tire
Wear Contribution to PM2.5 in Urban Environments. Atmosphere. 10. 99.

10.3390/atmosl 0020099.

17	Sommer et. Al. (2018). Tire Abrasion as a Major Source of Microplastics in the Environment.
Aerosol and Air Quality Research. 18. 10.4209/aaqr.2018.03.0099.

18	Dalmau, Eugenia & Augsburg, Klaus & Wenzel, Felix & Ivanov, Valentin. (2017). Tire particle
emissions: Demand on reliable characterization.

127


-------
most cost-effective to fabricate, especially in a lab already containing a chassis
dynamometer and PM sampling equipment.

The literature reveals a wide range of particle emissions estimates and wear factors.
One aspect that adds complexity is that tires can sometimes separate larger particles
up into the ~1mm range. These particles settle immediately so do not become airborne
particulate, but they are a part of the mass represented by reported "wear factors."19 So,
literature review must take care to specify the type of emission of interest. This
difference contributes to the wide range of observed values for tire wear in literature.
ERG has attempted to compile some estimates of tire and road wear and they are
presented in Table 33. The table presents two emission sources, tires only or the tire
and road interface. For comparison, value ranges are given for PM2.5, PM10, and all
wear. Some values are given in literature on a per-tire basis and some are given at the
vehicle distance traveled level, indicated by vkm.

Table 33. Selected light-duty tire and roadway PM emission factors from literature

Emission Source

PM2.5

PM10

All Mass

Tire Only





0.02-0.11 g/vkm20;

Tire + Roadway

0.31-0.5 ug/tire*km

0.54-0.95 ug/tire*km;
3.5-9.0 mg/vkm21

3.5-6.4 mg/km

(>PM10)22

Based on literature, there appears to be a wide variety of research into measuring
particulate emissions from tires, however there does not appear to be a consensus or
any standardization of test processes. One potential avenue of work would be to
conduct a survey or poll from various researchers to start developing a consensus on
what can be agreed upon and standardized to start moving toward a test method that
can be widely accepted.

19	Sommer et. Al. (2018). Tire Abrasion as a Major Source of Microplastics in the Environment.
Aerosol and Air Quality Research. 18. 10.4209/aaqr.2018.03.0099.

20	Ntziachristos and Boulter. (2013). EMEP/EEA air pollutant emission inventory guidebook -
2013: 1.A.3.b.vi Road vehicle tyre and brake wear.

21	PMP- Particle Measurement Program UNECE Informal Group, Non-exhaust traffic related
particle emissions. Non-exhaust traffic related particle emissions (brake and tyre/road wear).
Informal document GRPE-73-14 73rd GRPE, 6-10 June 2016.
https://www.unece.org/fileadmin/DAM/trans/doc/2016/wp29qrpe/GRPE-73-14.pdf

22	Aatmeeyata, D.S. Kaul, Mukesh Sharma, Traffic generated non-exhaust particulate emissions
from concrete pavement: A mass and particle size study for two-wheelers and small cars,
Atmospheric Environment, Volume 43, Issue 35, 2009, Pages 5691-5697

128


-------
References

Mathissen, M. et. al., A novel real-world braking cycle for studying brake wear particle
emissions, Wear, Volumes 414-415, 2018

Garg, Bhagwan D. et. al. "Brake Wear Particulate Matter Emissions." Environmental
Science & Technology 34.21 (2000): 4463-4469

Sanders, Paul G. et. al. "Airborne Brake Wear Debris: Size Distributions, Composition,
and a Comparison of Dynamometer and Vehicle Tests." Environmental Science &
Technology 37 (2003): 4060-4069

Marcel Mathissen, Jaroslaw Grochowicz, Christian Schmidt, Rainer Vogt, Ferdinand H.
Farwick zum Hagen, Tomasz Grabiec, Heinz Steven and Theodoras Grigoratos, A
novel real-world braking cycle for studying brake wear particle emissions, Wear,
https://doi.Org/10.1016/j.wear.2018.07.020

EPA Annual Certification Data for Vehicles, Engines, and Equipment

https://www.epa.gov/compliance-and-fuel-economy-data/annual-certification-data-

vehicles-engines-and-equipment

Society of Automotive Engineers (SAE) Standard J2789, Inertia Calculation for Single-
Ended Inertia-Dynamometer Testing
https://www.sae.org/standards/content/j2789 201008/

Grigoratos, Theodoras & Martini, Giorgio. (2014). Brake wear particle emissions: a
review. Environmental science and pollution research international. 22.

10.1007/sl 1356-014-3696-8.

US EPA. Brake and Tire Wear Emissions from On-road Vehicles in MOVES2014.
EPA-420-R-15-018, November 2015

Ntziachristos and Boulter. (2013). EMEP/EEA air pollutant emission inventory
guidebook - 2013: 1 ,A.3.b.vi Road vehicle tyre and brake wear.

Panko, Julie & Hitchcock, Kristen & Fuller, Gary & Green, David. (2019). Evaluation of
Tire Wear Contribution to PM2.5 in Urban Environments. Atmosphere. 10. 99. 10.3390/
atmosl 0020099.

Sommer et. Al. (2018). Tire Abrasion as a Major Source of Microplastics in the
Environment. Aerosol and Air Quality Research. 18. 10.4209/aaqr.2018.03.0099.

Dalmau, Eugenia & Augsburg, Klaus & Wenzel, Felix & Ivanov, Valentin. (2017). Tire
particle emissions: Demand on reliable characterization.

PMP - Particle Measurement Program UNECE Informal Group, Non-exhaust traffic
related particle emissions. Non-exhaust traffic related particle emissions (brake and

129


-------
tyre/road wear). Informal document GRPE-73-14 73rd GRPE, 6-10 June 2016.
https://www.unece.org/fileadmin/DAM/trans/doc/2016/wp29grpe/GRPE-73-14.pdf

Aatmeeyata, D.S. Kaul, Mukesh Sharma,Traffic generated non-exhaust particulate
emissions from concrete pavement: A mass and particle size study for two-wheelers
and small cars, Atmospheric Environment, Volume 43, Issue 35, 2009, Pages 5691-
5697

130


-------
Glossary of Terms, Abbreviations, and Symbols

100S4 - The TSI instrument used in this work to measure gravimetric PM mass at
various size classifications using coated aluminum impactors

APS - Aerodynamic Particle Sizer, the TSI instrument used to measure particle size
distributions in the range from 0.5 - 20 |jm

Brake Event - In a driving trace, the period of time from the initial application of the
brakes until the brakes are either released or the vehicle comes to a stop

BWI - Brake Wear Index (or Indices), a measure of the amount of brake emissions
potential from a given vehicle or brake component, weighted by friction material mass
and the counts in-use in the fleet

California Brake Dynamometer Cycle (CBDC) - The new brake dynamometer test
cycle developed and used during this test program

EEPS - Engine Exhaust Particle Sizer, the TSI instrument used to measure particle size
distributions in the range from 5.6 to 560 nm

ETW - Equivalent test weight, the simulated test weight used for most tests in this
program; corresponds to curb weight + 300 lbs.

Generalized Coastdown Curve - The estimated coastdown curve, derived from the
average of the road load coastdowns of the 6 test vehicles in this program, used to
identify braking events in the Caltrans dataset (any decelerations greater than this curve
were flagged as braking events)

Heating and Cooling Matrix - The test matrix of standardized braking events and
cruise intervals followed by each vehicle on the test track while recording brake
temperatures in order to provide inputs to the temperature model.

HLW - Heavily laden weight, the test weight used in some tests of the cargo-carrying
vehicles amounting to two thirds of the difference between the curb weight and gross
vehicle weight

LM - Low metallic friction materials used in some tests of the Camry and F-150
assemblies. Indicates the materials have a larger amount of metal than NAO

Microtrip - An identified portion of the Caltrans driving trace consisting of the time that
a vehicle starts moving, through the cruise and lasting until the next stop.

NAO - Non asbestos organic friction materials, the most common friction material and
that comprising all OES pads

OES - Original Equipment Service, the friction material specified in dealer-sourced
replacement parts

PM - Particulate Matter, often classified by PM2.5, indicating particulate up to 2.5|jm in
diameter, and PM10, particulate up to 10|jm in diameter

QCM - Quartz Crystal Microbalance, the TSI instrument used to measure continuous
PM2.5 mass

131


-------
Represented distance - the distance represented by the CBDC for inventory
purposes; the cycle consists of braking events extracted from real trips as well as
engineered cruises added for cooling, neither of which has distance basis in real use;
represented distiance also includes the actual distance traveled between each braking
event

SCC - Speed Correction Cycles, the EMFAC speed cycles used to factor UC emission
rates into the different EMFAC speed bins

SCF - Speed Correction Factors, the values provided in this work that may be used to
to factor the overall vehicle-level emission rates found in this work into groups of
EMFAC speed bins

UC - The EMFAC Unified Cycle, the standard basis for exhaust emissions factors in
EMFAC

WLTP-Brake - The World-Harmonized Brake cycle, developed recently in Europe in
cooperation with the JRC; a new cycle designed specifically for measuring brake PM
emissions.

132


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix A

Appendix A

Vehicle and Friction Material Selection Supporting Data


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix A

This Appendix includes supporting data for the Representative Test Vehicle and Friction Material Selection section of the report.
This table presents the details of the top-25 vehicle analysis including the calculated values for the two brake wear indices (BWI).
In the table, VIO indicates vehicles in operation. BWI1 is VIO x total wearable mass, and BWI2 is VIO x total wearable mass x
replacement rate.

MAKE

MODEL

MY

Comm
on for
bench-
markin

g

Curb
wt

(kg)

Brake
system

VIO

VIO
Rnk

Total
wear
mass
per

vehicle

/g

BWI1 =
(VIO x
total

wearable
mass) /
ton

BWI1
Rnk

Repl.
Rate
(%)

OE/OES
mkt
share
(%)

BWI 2
(tons)

BWI2
Rnk

TOYOTA

CAMRY
(BASE, L,
LE)

2009-
2016

yes-
FA

1460

Disc/
Disc

342992

1

2133

732

1

16%

34.0%

117

1

HONDA

CIVIC LX

2012-
2015



1221

Disc/
Drum

140733

5

2322

327

4

14%

34.0%

46

3

NISSAN

ROGUE S

2014-
2016



1550

Disc/
Disc

41213

10

1845

76

14

11%

34.0%

8

16

TOYOTA

SIENNA LE

2011-
2015



1940

Disc/
Disc

44921

8

2717

122

8

14%

34.0%

17

6

FORD

F150

SPRCREW

2015-
2016

yes-

FA&

RA

2206

Disc/
Disc

32921

17

2895

95

11

11%

90.0%

10

10

TOYOTA

PRIUS
REGULAR

2010-
2016



1382

Disc/
Disc

241055

2

1749

422

3

2%

34.0%

8

15

TOYOTA

COROLLA L

2014-
2016



1265

Disc/
Drum

159154

3

3028

482

2

11%

53.5%

53

2

NISSAN

ALTIMA
(BASE, 2.5)

2012-
2016

yes-
FA

1429

Disc/
Disc

149096

4

1510

225

5

14%

40.7%

32

4

NISSAN

SENTRA S

2013-
2016



1277

Disc/
Disc

110629

6

1436

159

7

11%

53.5%

17

5

FORD

F150

SPRCREW

2013-
2014

yes-
FA

2549

Disc/
Disc

33721

16

2878

97

10

16%

17.0%

16

7

LEXUS

RX 350

2014-
2015



1900

Disc/
Disc

43306

9

2707

117

9

11%

34.0%

13

8

CHEVROLET

TAHOE
C1500

2007

yes-
FA

2462

Disc/
Disc

19517

23

2521

49

18

23%

14.0%

11

9


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix A

MAKE

MODEL

MY

Comm
on for
bench-
markin

g

Curb
wt

(kg)

Brake
system

VIO

VIO
Rnk

Total
wear
mass
per

vehicle

/g

BWI1 =
(VIO x
total

wearable
mass) /
ton

BWI1
Rnk

Repl.
Rate
(%)

OE/OES
mkt
share
(%)

BWI2
(tons)

BWI2
Rnk

TOYOTA

RAV4 XLE

2014-
2016



1560

Disc/
Disc

36803

14

2462

91

12

11%

34.0%

10

11

TOYOTA

TACOMA
DOUBLE
CAB

2015-
2016



1975

Disc/
Drum

36052

15

5256

189

6

5%

90.0%

9

12

HONDA

ACCORD
LX

2014-
2016

yes-
FA

1465

Disc/
Disc

52193

7

1598

83

13

11%

53.5%

9

13

DODGE

RAM 1500
ST

2004



2260

Disc/
Disc

19739

22

2180

43

19

21%

5.0%

9

14

HYUNDAI

ELANTRA
GLS

2013



1207

Disc/
Disc

30566

18

1649

50

17

16%

17.0%

8

17

HYUNDAI

SONATA
(GLS, SE,
SPORT)

2013-
2015



1486

Disc/
Disc

40117

11

1678

67

15

11%

90.0%

7

18

CHEVROLET

SILVERADO
1500

2014-
2015

yes-
FA

2240

Disc/
Disc

27578

19

2431

67

16

11%

53.5%

7

19

HONDA

ACCORD
EX

2014-
2016



1492

Disc/
Disc

39344

12

993

39

21

11%

53.5%

4

20

HONDA

ACCORD
SPORT

2014-
2015



1468

Disc/
Disc

37332

13

803

30

23

11%

34.0%

3

21

HONDA

CIVIC LX

2016



1276

Disc/
Disc

25782

20

1666

43

20

5%

90.0%

2

22

LEXUS

RX 350

2016



1970

Disc/
Disc

12540

24

2668

33

22

5%

90.0%

2

23

HYUNDAI

SONATA SE

2016



1486

Disc/
Disc

11363

25

1803

20

24

5%

90.0%

1

24

HONDA

ACCORD
SPORT

2016



1507

Disc/
Disc

22978

21

803

18

25

5%

90.0%

1

25


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix B

Appendix B

Heating and Cooling Matrix for Track Testing


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix B

The following table depicts the braking events that made up the ERG heating and
cooling matrix conducted by LINK at the test track. Each vehicle was subject to a series
of braking snubs to achieve the desired initial temperature. The initial and final speeds,
along with the deceleration rate is given for each braking event. The cooling speed,
where applicable, refers to the steady-state speed that should be held after the braking
event to allow the brakes to cool down below 50°C.

Event #

Initial Front Axle

Initial

Final

Cooling

Deceleration,



disc temperature,
°C

speed,
km/h

speed,
km/h

speed, km/h

g

1

60

55

0

0

0.10

2

60

55

< 5

55

0.25

3

160

55

< 5

55

0.25

4

160

55

< 5

NA

0.35

Warm-up

open

120

60

55

0.40

5

300-350

55

55

55

-

6

60

55

< 5

NA

0.25

Warm-up

open

120

60

55

0.40

7

300-350

55

55

55

-

8

60

55

< 5

55

0.35

9

60

95

0

0

0.10

10

60

95

< 5

55

0.10

11

60

95

< 5

95

0.25

12

160

95

< 5

95

0.25

13

160

95

< 5

NA

0.35

Warm-up

open

120

60

95

0.40

14

300-350

95

95

95

-

15

60

95

< 5

NA

0.35

Warm-up

open

120

60

95

0.40

16

300-350

95

95

95

-

17

60

95

< 5

95

0.35

18

60

130

0

0

0.10

19

60

130

< 5

55

0.10

20

60

130

< 5

130

0.25

21

160

130

< 5

130

0.10

22

160

130

< 5

NA

0.25

Warm-up

open

120

60

130

0.40

23

300-350

130

130

130

-

24

60

130

< 5

NA

0.35

Warm-up

open

120

60

130

0.40

25

300-350

130

130

130

-

26

60

130

< 5

95

0.35

Warm-up

open

120

60

NA

0.40

post Warm-up

300-350

60

0

0

0.25

27

300-350

0

0

0

-

Warm-up

open

120

60

NA

0.40

Warm-up

300-350

60

0

0

0.25

28

300-350

0

0

0

-


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix C

Appendix C

Derivation of the Generalized Coastdown Curve and Road Load Coefficients


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix C

The generalized coastdown curve was determined using the EPA-published road load
coefficients for the six test vehicles used in this project. EPA publishes these values,
including test weight, yearly as a part of new-vehicle exhaust emissions certification. 1
The coefficients are used to allow determination of the simulated dynamometer drag
force on the vehicle as a function of speed. The published coefficients for the test
vehicles used in this work are shown in the following table.



Target Coef. A
(Ibf)

Target Coef. B
(Ibf/mph)

Target Coef. C
(lbf/mph**2)

Weight, lbs

Toyota Camry

35.941

-0.01201

0.020084

3670

Honda Civic

21.290

0.11890

0.018670

2970

Toyota Sienna

37.384

0.03816

0.029553

4810

Ford F-150

46.83

0.7658

0.03132

5770

Toyota Prius

31.145

0.35285

0.013956

3510

Nissan Rogue

35.59

-0.1577

0.028

3640

The EPA target coefficients are the target that the vehicle will experience as a function
of speed. For this project, the force was not the parameter of interest as the main focus
was on the vehicle deceleration rate. Using the weight and the road load force curve,
ERG determined the expected deceleration rate as a function of speed for the six
vehicles. ERG performed a polynomial curve fit to the average (by speed) of the
deceleration rates for the six vehicles. The calculated coastdown deceleration rates as a
function of speed are presented for the six test vehicles in the following figure. The
larger black curve represents a fit to the average of the six vehicles and forms the basis
of the generalized coastdown curve. The generalized coastdown curve used in this work
was:

AV = -7.931x10"5«V2 - 8.558x10"4«V- 0.3023
where AV has units of kph/s and V has units of kph.

1 https://www.epa.gov/compliance-and-fuel-economy-data/annual-certification-data-
vehicles-engines-and-equipment


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix C

The following table presents the actual inertia values used by LINK when programming
tests of each vehicle/axle combination into the dynamometer control software:

Vehicle

Front Inertia (kg*m2)

Rear Inertia (kg*m2)

Camry ETW

79.8

28.6

Civic ETW

55.1

19.8

F-150 ETW

161.0

58.6

F-150 HLW

184.3

67.0

Prius ETW

53.0

26.5

Rogue ETW

85.9

30.8

Rogue HLW

95.9

34.4

Sienna ETW

98.0

47.5

Sienna HLW

113.2

54.9


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix D

Appendix D

The ERG Vector Collinearity Cycle Building Approach


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix D

This section references two past reports in which ERG described the use of the vector
collinearity method to build up a cycle from a larger in-use dataset. The process
required further refinement for use in 17RD016 because of the temperature
requirements and that temperature could not be directly controlled on the brake
dynamometer (as the temperature is a function of only speed, braking intensity, and
time).

The first reference was included as a footnote in ERG's proposal for this project. This
reference is Roadway-Specific Driving Schedules for Heavy-Duty Vehicles, an ERG
report to EPA, August 15, 2003. Section 5 of that document includes the description of
the implementation of the vector method. This report is available at:
http://nepis.epa.gov/Exe/ZvPURL.cgi?Dockev=P100LWCT.TXT

The next reference is presented as an excerpt, and is from Development of NONROAD
Load Factors, Emission Factors, Duty Cycles, and Activity Estimates, an ERG report to
EPA, February 12, 2013. In this reference, the term microtrip is the equivalent to the
brake event in Project 17RD016 (meaning it is the building block taken from the larger
dataset and used to build up the new cycle). The section describing the technique reads
as follows:

In 1995 we developed a technique for creating duty cycles based on the microtrip
concept. Since then, we have built engine dynamometer cycles (load and RPM vs. time)
for wheeled loaders2 and telescoping boom excavators,4 and chassis dynamometer
cycles (speed vs. time) fordrayage trucks, heavy-duty trucks3, dump trucks, and
Bangkok cars and motorcycles. In most cases we also collected the data used to build
those duty cycles. In addition, many of the cycles built were actually used to make
dynamometer measurements of the emissions of engines and POEs.

The idea of a duty cycle is that it contains the essence of actual operating
behavior. To make a representative cycle practical, it should be no longer than needed.
A key challenge for the cycle builder is to compress the dataset collected on each type
of construction equipment to produce a reasonably short cycle while maintaining the
essence of the engine operating behavior. Such a short cycle can then be used to
characterize engine operation to estimate engine emissions for a type of construction
equipment during typical operation for that type of equipment.

2T.H. DeFries, G.F. Baker, B. Limsakul, M.A. Sabisch, P. Henson, S. Kishan, "ERG
Contributions to the Texas Department of Transportation Evaluation of PuriNOx Diesel
Fuel," prepared for R.D. Matthews, University of Texas at Austin, prepared by Eastern
Research Group, TxDOT-030218, February 18, 2003.

3T.H. DeFries, S. Kishan, B. Limsakul, M.J. Hebets, "Roadway-Specific Driving
Schedules for Heavy-Duty Vehicles," prepared for U.S. Environmental Protection
Agency, prepared by Eastern Research Group, EPA-030815, August 15, 2003.


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix D

Representative cycles can be built using different methodologies. The
methodology we have chosen for this study is to use pieces of real engine operation,
called microtrips, from the data collected on construction equipment, which when
connected together can be expected to have similar emissions behavior to the same
type of equipment in normal operation. The cycle is built around parameters of
equipment operation and usage that are known to be important to exhaust emissions.
By building up a duty cycle from snippets (microtrips) of actual engine operation such
that the characteristics of the cycle in some way matches the characteristics of a
database of typical engine operation, it can be inferred that the emissions behavior of
the engine over the cycle will be similar to the emissions behavior of the engine in a
particular type of construction equipment.

The cycle is created by selecting and combining microtrips taken from the
dataset of actual engine operation for each type of construction equipment. Two or
more variables are used to define and select microtrips for the cycle. This cycle-building
introduction uses relative mass fuel rate and relative RPM as the two variables used to
define engine operation. To identify specific segments of equipment operation for
inclusion in the cycle, the entire activity dataset is converted to a set of microtrips.
Typically, a microtrip is defined as a contiguous time trace of engine operation that is an
all-non-idle period or that is an all-idle period.

To use the microtrip cycle development approach, all of the microtrips in the
dataset need to have all of their second-by-second observations binned in terms of
relative mass fuel rate and relative RPM. While the size of the bins is arbitrary, bins in
general need to be narrow enough to resolve important emissions effects. On the other
hand, from a practical perspective, the number of bins needs to be small so that the
program that selects microtrips can run in a reasonable amount of time.

Selecting microtrips for the cycle is based on a strategy of minimizing the
difference between a cycle vector C representing operation in the candidate cycle and a
target vector T representing operation in the activity database. As microtrips are added
to the kernel of the candidate cycle, the difference between the two vectors C and T
tends to become smaller and smaller. The build-up process ends when the cycle
developer decides that the two vectors are substantially the same and the duration of
the cycle that has been built up is acceptably short. The multi-dimensional space that
these vectors are in will be described shortly, but first let us consider how the build-up
process works for developing a cycle.


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix D

The goal of building the cycle is to select microtrips such that when their vectors
Mj are added together, the vector C of the resulting cycle is as similar as possible to the
target vector T of the activity database. Figure A3-1 shows the hypothetical situation of
the vectors after two microtrips have been used to create a cycle. In this hypothetical
example, the first microtrip was selected from the activity database for the case as the
one whose vector Mi was closest to the target vector T for the database. Then, a
second microtrip is searched for such that when its vector M2 is added to Mi to create
the resultant vector C shown in Figure A3-1, the distance between the tips of C and T is
minimized. This distance is the length of the vector T-C as denoted in the figure by the
dashed vector. As microtrips are added to create the built-up cycle represented by C,
the length of T-C is calculated after each additional microtrip is added to the cycle to
follow the progress of the build-up process. It should be noted that the order of the
microtrips in the final cycle is unimportant from the point of view of the selection of the
microtrips. The reason for this is that the resultant C is independent of the order in
which the microtrip vectors Mj are added together.

Figure A3-1. Vector Description of Comparing Target and Cycle Activity

T-C

iv±!

It should also be noted that we are forcing microtrips to be added to the
candidate cycle. This is done even if the addition of the best incremental microtrip
causes the length of T-C to increase in some instances. Generally, as the cycle is built
up there will be a decrease in the length of T-C. After several microtrips have been
added, the length of T-C may increase slightly. Later, with the addition of more


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix D

microtrips, a "discovery" will be made that will produce a relatively abrupt decrease in
the length of T-C so that the accumulated cycle will be substantially better than the
cycle was much earlier in the build-up process.

All of the vectors used above to describe the build-up process are based on
representations of the cumulative frequency distributions of observations in relative
mass fuel rate / relative RPM space. This statement requires some explanation. A
segment of operation, whether it is a microtrip, a piece of a duty cycle, or the entire
activity database can be described as a frequency distribution. The distribution consists
of combinations of the two variables: relative mass fuel rate and relative RPM. The
continuous values for these variables were converted into frequency distributions
through the use of bins. Each one-second observation in the database was placed in a
particular relative mass fuel rate / relative RPM bin. The cumulative frequency
distribution is made up of the number of observations that fall "below" the current bin for
each of the two-binned variables. The binning criteria for the variables will be described
in Section 5.2. To help the reader understand the process, we will present a numerical
example in one dimension and another example in two dimensions to demonstrate how
the comparison of the vectors T and C works.

Suppose we wanted to compare a candidate cycle with the database using a
single POE operation variable that was monitored second-by-second in the collection of
data for the activity database. The single variable might be engine load. In this
hypothetical example, we have 35,900 one-second observations of engine load in the
target activity database and 68 one-second observations in the cycle. The first step in
comparing T and C is to bin the observations of load in the target data and in the cycle
data. TableTable A3-1 shows the binning of the hypothetical data in Columns 2 and 3.
Note that the number of observations in the target data in Column 2 is much higher than
the number of observations in the cycle data in Column 3. This is a consequence of the
activity database containing all of the observations for all microtrips and the cycle
having just one microtrip. The frequency counts in Columns 2 and 3 are then converted
to cumulative frequency counts in Columns 4 and 5. This is done to provide proximity
information for the microtrip searching algorithm. In other words, we wanted the
algorithm to be able to select a microtrip even if the observations for a given microtrip
were not in exactly the same bins as the target but did have observations at least in a
nearby bin. The use of the cumulative distributions helps ensure that proximity
information is available.


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix D

Table A3-1. Comparison of Cycle and Target Vectors
for a Hypothetical One-Dimensional Example



Counts

Cumulative
Counts

Vector
(Normalized
Cumulative
Counts)

Square of
Difference

Bin

Target

Cycle

Target

Cycle

T

C

T-C

1

1000

0

1000

0

0.028

0.000

0.001

2

11000

30

12000

30

0.334

0.441

0.011

3

7000

10

19000

40

0.529

0.588

0.003

4

6000

7

25000

47

0.696

0.691

0.000

5

4500

5

29500

52

0.822

0.765

0.003

6

2800

1

32300

53

0.900

0.779

0.014

7

1500

4

33800

57

0.942

0.838

0.011

8

800

6

34600

63

0.964

0.926

0.001

9

600

1

35200

64

0.981

0.941

0.002

10

700

4

35900

68

1.000

1.000

0.000



Sum of
Squares

6.139

5.657

0.047

Vector
Length

2.478

2.379

0.217

A comparison of the cumulative counts for the target and cycle information in
Columns 4 and 5 shows that if we used these counts to create the T and C vectors, the
lengths of the vectors would be greatly different simply because the target vector, which
is made up of the 10 elements in Column 4, would be a much longer vector then the
cycle vector, which is made up of the 10 elements in Column 5. Accordingly, we
normalize the target and cycle cumulative counts in 4 and 5 to produce the target vector
elements and the cycle vector elements as the fractional values between 0 and 1 shown
in Columns 6 and 7.

The values in Columns 6 and 7 become the elements of the T and C vectors,
which are in 10-dimensional space. A visualization of the elements of these vectors is
provided in Figure A3-2. This figure shows the normalized cumulative counts of the
target and cycle from Columns 6 and 7 as a function of the bin number. What we want
to do in developing the cycle is select microtrips so that the curve for the cycle is as
close as possible to the curve for the target in this figure. The way we do this is to
minimize the sums of the squares of the differences between the value for the
corresponding elements of the target and cycle vectors. This corresponds to the square
of the length of T-C. TableA3-1 shows the calculated length of T, C, and T-C. These
lengths can be determined from the values of the elements for T and C in Columns 6


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix D

and 7 using the standard relationship for determining the length of a vector when its
elements are known.

Figure A3-2. Visual Comparison of Vector Elements

Target
-Q— Cycle

1 23456789 10

Bin

Extension of the one-dimensional example shown in TableA3-1 and Figure A3-1
to multiple dimensions is demonstrated by the spreadsheet calculations shown in Table
A3-2. In this example, 100 matrix elements are used. The table shows 10 rows which
might be relative mass fuel rate and 10 columns which might be relative RPM. The left
side of Table A3-2 shows the calculations for the target matrix and the right side shows
the calculations for the cycle matrix. In Tables a) and b), the second-by-second
observations of the target and cycle data are binned. The numbers in each bin
represent the frequency of observations that meet the criteria for those bins. In Tables
c) and d), the counts in the Tables a) and b) are accumulated across each row. Then, in
Tables e) and f), the accumulated frequencies in Tables c) and d) are accumulated
down each column. This produces a field of frequencies on a cumulative basis that run
from a low value in the upper left corner of each matrix to a high number in the lower
right corner of each matrix. The value in the lower right hand corner of Tables e) and f)
is equal to the total number of observations in the target or cycle matrix. These total
observation numbers in the lower right hand corner of e) and f) are used to normalize all
of the frequencies in Tables e) and f) to arrive at the normalized cumulative matrices in
g) and h). The values in g) and h) are then used to calculate the square of the
differences in each corresponding matrix element to produce the values in Table i). The
value in Table j) is just the summation of all of the elements of Table i) and represents
the square of the length of the T-C vector. This is the value that we attempt to minimize
when selecting microtrips for the cycle. Note that the counts in a) and b) did not need to


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix D

be in corresponding bins for this comparison process to work. The use of cumulative
distributions permitted the two matrices to be compared successfully.

Extension of the technique to more than two dimensions can be made by
analogy.


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix D

Table A3-2. Comparison of Cycle and Target Matrices for a Hypothetical Two-

Dimensional Example

Target Activity Matrix

Cycle Activity Matrix

a) Count the second-by-second obseivations in each bin.
ABCDEFGH I

b) Count the second-by-second obseivations in each bin.
A BCDEFGH I

c) Accumulate the above frequencies across each row

2

2

2

2

2

2

2

2

2

2

0

1

1

1

1

1

1

1

1

1

0

2

2

7

7

7

7

7

7

7

0

0

5

5

8

8

10

11

11

11

0

5

5

14

15

15

15

17

26

29

0

0

2

2

2

6

7

7

7

7

0

0

0

0

0

0

0

0

0

0

0

0

6

6

6

7

7

7

7

7

0

1

1

1

1

1

1

1

1

1

0

0

0

0

0

0

0

0

0

0

d) Accumulate the above frequencies across each row

0

0

0

0

0

0

0

0

0

0

0

1

1

1

1

1

1

1

1

1

0

0

0

4

4

4

4

4

4

4

0

4

4

4

4

7

7

8

8

8

0

0

0

0

0

0

0

4

5

5

0

0

0

8

8

8

8

8

8

10

0

0

0

0

3

3

3

3

3

3

0

0

0

0

0

0

1

1

1

1

1

6

6

6

6

6

6

6

6

6

0

0

0

0

0

0

0

0

0

0

e) Accumulate the above frequencies down each column.

2

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

3

3

3

2

5

5

10

10

10

10

10

10

10

2

5

10

15

18

18

20

21

21

21

2

10

15

29

33

33

35

38

47

50

2

10

17

31

35

39

42

45

54

57

2

10

17

31

35

39

42

45

54

57

2

10

23

37

41

46

49

52

61

64

2

11

24

38

42

47

50

53

62

65

2

11

24

38

42

47

50

53

62

65

f) Accumulate the above frequencies down each column.

0

0

0

0

0

0

0

0

0

0

0

1

1

1

1

1

1

1

1

1

0

1

1

5

5

5

5

5

5

5

0

5

5

9

9

12

12

13

13

13

0

5

5

9

9

12

12

17

18

18

0

5

5

17

17

20

20

25

26

28

0

5

5

17

20

23

23

28

29

31

0

5

5

17

20

23

24

29

30

32

1

11

11

23

26

29

30

35

36

38

1

11

11

23

26

29

30

35

36

38

g) Normalize the elements in the above matrix.

1

0.031

0.031

0.031

0.031

0.031

0.031

0.031

0.031

0.031

0.031

2

0.031

0.046

0.046

0.046

0.046

0.046

0.046

0.046

0.046

0.046

3

0.031

0.077

0.077

0.154

0.154

0.154

0.154

0.154

0.154

0.154

4

0.031

0.077

0.154

0.231

0.277

0.277

0.308

0.323

0.323

0.323

5

0.031

0.154

0.231

0.446

0.508

0.508

0.538

0.585

0.723

0.769

6

0.031

0.154

0.262

0.477

0.538

0.600

0.646

0.692

0.831

0.877

7

0.031

0.154

0.262

0.477

0.538

0.600

0.646

0.692

0.831

0.877

8

0.031

0.154

0.354

0.569

0.631

0.708

0.754

0.800

0.938

0.985

9

0.031

0.169

0.369

0.585

0.646

0.723

0.769

0.815

0.954

1.000

10

0.031

0.169

0.369

0.585

0.646

0.723

0.769

0.815

0.954

1.000

h) Normalize the elements in the above matrix.

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.026

0.026

0.026

0.026

0.026

0.026

0.026

0.026

0.026

0.000

0.026

0.026

0.132

0.132

0.132

0.132

0.132

0.132

0.132

0.000

0.132

0.132

0.237

0.237

0.316

0.316

0.342

0.342

0.342

0.000

0.132

0.132

0.237

0.237

0.316

0.316

0.447

0.474

0.474

0.000

0.132

0.132

0.447

0.447

0.526

0.526

0.658

0.684

0.737

0.000

0.132

0.132

0.447

0.526

0.605

0.605

0.737

0.763

0.816

0.000

0.132

0.132

0.447

0.526

0.605

0.632

0.763

0.789

0.842

0.026

0.289

0.289

0.605

0.684

0.763

0.789

0.921

0.947

1.000

0.026

0.289

0.289

0.605

0.684

0.763

0.789

0.921

0.947

1.000

i) Calculate the squares of the differences in corresponding elements of the above two matrices.



A

B

C

D

E

F

G

H



J

1

0.001

0.001

0.001

0.001

0.001

0.001

0.001

0.001

0.001

0.001

2

0.001

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

3

0.001

0.003

0.003

0.000

0.000

0.000

0.000

0.000

0.000

0.000

4

0.001

0.003

0.000

0.000

0.002

0.002

0.000

0.000

0.000

0.000

5

0.001

0.000

0.010

0.044

0.073

0.037

0.050

0.019

0.062

0.087

6

0.001

0.000

0.017

0.001

0.008

0.005

0.014

0.001

0.021

0.020

7

0.001

0.000

0.017

0.001

0.000

0.000

0.002

0.002

0.005

0.004

8

0.001

0.000

0.049

0.015

0.011

0.010

0.015

0.001

0.022

0.020

9

0.000

0.014

0.006

0.000

0.001

0.002

0.000

0.011

0.000

0.000

10

0.000

0.014

0.006

0.000

0.001

0.002

0.000

0.011

0.000

0.000

j) Sum the squares of the differences.


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix D

An example of the discovery process as microtrips are built up is shown in Figure
A3-2. The figure shows a plot of the length of the T-C vector as microtrips were added
to a cycle for wheeled backhoe loaders. The figure shows that as microtrips were
added, the length of the vector first dropped to a local minimum after 12 microtrips were
added. The next minimum vector length was encountered after 14 microtrips were
added. Subsequent lower minima were achieved when 20, 23, 28, 30, ... microtrips
were added. Major decreases in the length of the T-C vector occurred after 14, 28, and
59 microtrips were added. This alternate drop/plateau/increase/drop behavior is
commonly seen when using this method of cycle building. At this point in cycle
development, the duration of the cycle becomes important. Depending on the
acceptability of the duration of the cycle being built up, a cycle with any of the specific
14, 28, or 59 microtrips could be used. These three candidate cycles would have
durations of 73, 124, and 437 seconds, respectively.

Figure A3-2. Square of the Length of T-C as MicroTrips are Added:
Example for Wheeled Backhoe Loaders

o
o
0

> 3

o

£ 2
D)
C
0

600

500

400

300

100

15 20 25 30 35 40 45 50 55 60 65 70

Number of Micro-Trips

/pfoj1/EPA_MOVES_NONROAD/DutyCycle/stats_backhoeloader_wtieeled.sas 12DEC12 13:13

co
"O
c
o
o
0)
C/)

Ui
sz
CD
_l

0)

o
O

200 a)
>

TO
3

E
O


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix E

Appendix E

Distributions of Parameters of Interest for the Vector Method's (New CBDC) 3 Speed

Segments


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix E

Further distributions of the four parameters of interest, broken down by the three speed
segments created by the vector method, are presented in this Appendix. The
distribution for the overall vector cycle (previously shown in the distributions in the
Results section) is presented for comparison. Note again that the distributions are for
the periods during brake events only; acceleration and cooling/cruise periods are not
included.

The distributions of brake event durations are presented in the following figure. As
described previously, the ERG cycle intentionally does not have any brake events
shorter than 3 seconds.

Brake Event Duration (s)

The distributions of speeds encountered during braking events are presented in the
following figure. It can clearly be seen that the slow speed segment has more time in
the slow speed bins, and the high speed segment has more time in the high speed bins.
The distribution of (negative) acceleration rates during braking is presented in the
subsequent figure.


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix E

25

S 20

~o
c

o

a15

on
Ctf)
c

CD

10

C

(D
(J
i—


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix E

0 20 40 60 80 100 120 140 160 180 200 220 240 260
Temperature Above Ambient/Wheel Well (C)


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix F

Appendix F

Test Matrix and Test Dates


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix F

The complete test matrix is presented in this Appendix. Note under Teflon Filter
analysis, all planned tests in which CARB was planned for analysis of the Teflon filters
are indicated. Tests in which this column is left blank may be either analyzed by EPA or
not subject to any further speciation analyses depending on EPA preferences. There
are four tests in the matrix that were conducted over the WLTP-Brake cycle. These are
labeled in the Replicates column as WLTP-Brake A or B; there were 2 replicates of
each planned WLTP-Brake test. The BMC LeafMark is listed if known and listed as
Indeterminate (Ind.) if it was not known until ordered components are delivered. The
17RD016 proposal included provision for 90 days of available dynamometer time. LINK
estimated that the cooling airflow setting experimentation for all assemblies would
require a minimum of 5 days of dynamometer time. So, in the matrix the first 5 test days
were reserved for cooling air flow rate setting, 85 tests were prescribed (including the 2
tunnel blanks), adding to a total of 90. This matrix presents the planned tests and order
at the onset of the program. During the program, minor changes to the order were made
for the following reasons:

•	Aftermarket components were not received in time for their planned tests after
ordering

•	Tests were voided due to equipment malfunctions or other issues that invalidated
a test that had been initiated. In this case, this test was postponed if new
components needed to be ordered and the matrix was continued

•	One test that was scheduled to run over the WLTP-Brake cycle was inadvertently
run over the CBDC; this test was kept and the planned WLTP-Brake test was
swapped into the place of an equivalent upcoming CBDC test.

As mentioned in the report, there were various reasons why the originally planned
testing order was not maintained throughout the program. Delays in sourcing brake
components, voided and repeat tests and other minor reasons resulted in the actual test
order being changed; however LINK kept to the original plan where possible The table
in this appendix also includes the actual date of all valid tests.


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix F

Test Matrix Including Planned Test Order and Actual Test Dates

Test
Day
#

Test
Vehicle

Front/Rear

Pad

Material

BMC
Leaf
Mark

Wheel
load

# Replicates

Ref.
repeat
#

Teflon

Filter

Analysis

Test Date

1-5

Air/Sample flow setting days for each assembly



6

F-150

Front

OES-NAO

N

ETW

N/A

1

ARB

9/30/2019

7

F-150

Front

OES-NAO

N

HLW

A



EPA

10/13/2019

8

Camry

Front

OES-NAO

A

ETW

A



ARB

10/2/2019

9

Civic

Rear (Drum)

OES-NAO

N

ETW

A



EPA

10/3/2019

10

Camry

Rear

OES-NAO

A

ETW

A



ARB

10/5/2019

11

Sienna

Front

OES-NAO

A

ETW

A



EPA

10/6/2019

12

Prius

Front

OES-NAO

A

ETW

A



EPA

10/7/2019

13

Sienna

Rear

OES-NAO

B

ETW

A



EPA

10/7/2019

14

Rogue

Front

OES-NAO

A

ETW

A





10/12/2019

15

Camry

Front

OES-NAO

A

ETW

B



EPA

10/9/2019

16

F-150

Rear

OES-NAO

A

ETW

A



EPA

10/10/2019

Testing PAUSE

10/14-20/2019

17

Tunnel Blank 1

A





10/29/2019

18

Rogue

Rear

After-NAO

Ind.

HLW

A





10/28/2019

19

Rogue

Rear

OES-NAO

A

ETW

A



ARB

10/21/2019

20

Rogue

Rear

After-NAO

Ind.

ETW

A



ARB

10/29/2019

21

Camry

Rear

OES-NAO

A

ETW

B





10/22/2019

22

Camry

Rear

After-NAO

Ind.

ETW

B





10/24/2019

23

Camry

Rear

After-LM

A?

ETW

B





10/23/2019

24

F-150

Front

OES-NAO

N

ETW

N/A

2



10/25/2019

25

Civic

Rear (Drum)

OES-NAO

N

ETW

B



ARB

10/30/2019

26

Civic

Rear (Drum)

After-NAO

Ind.

ETW

B





12/13/2019

27

Prius

Rear

OES-NAO

A

ETW

A



EPA

10/27/2019

28

Prius

Rear

After-NAO

Ind.

ETW

A



ARB

10/27/2019

29

F-150

Rear

After-NAO

N

ETW

A



ARB

11/1/2019


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix F

Test
Day
#

Test
Vehicle

Front/Rear

Pad

Material

BMC
Leaf
Mark

Wheel
load

# Replicates

Ref.
repeat
#

Teflon

Filter

Analysis

Test Date

30

F-150

Rear

After-LM

A

ETW

A



ARB

11/2/2019

31

F-150

Rear

OES-NAO

A

HLW

A



ARB

11/2/2019

32

F-150

Rear

After-LM

A

HLW

A



ARB

11/3/2019

33

Sienna

Front

OES-NAO

A

ETW

B



ARB

11/4/2019

34

Sienna

Front

OES-NAO

A

HLW

B





11/19/2019

35

Sienna

Front

After-NAO

N

ETW

B





12/26/2019

36

Sienna

Front

After-NAO

N

HLW

B





1/26/2020

37

Rogue

Front

OES-NAO

A

HLW

A



ARB

1/26/2020

38

Rogue

Front

After-NAO

Ind.

ETW

A



ARB

11/20/2019

39

F-150

Front

OES-NAO

N

HLW

B





11/9/2019

40

F-150

Front

After-LM

A

ETW

B





11/10/2019

41

F-150

Front

After-LM

A

HLW

B





11/21/2019

42

F-150

Front

After-NAO

N

ETW

B





11/10/2019

43

F-150

Front

OES-NAO

N

ETW

N/A

3

ARB

11/11/2019

44

F-150

Front

OES

N

ETW

WLTP A





11/12/2019

45

Camry

Front

After-NAO

Ind.

ETW

A



ARB

11/22/2019

46

Camry

Front

After-LM

A?

ETW

A



ARB

11/23/2019

47

Camry

Front

OES

A

ETW

WLTP A





1/18/2020

48

Prius

Front

OES-NAO

A

ETW

B



ARB

11/26/2019

49

Prius

Front

After-NAO

Ind.

ETW

A



ARB

11/26/2019

50

Prius

Front

After-NAO

Ind.

ETW

B





1/27/2020

51

Civic

Front

OES-NAO

A

ETW

A



ARB

12/6/2019

52

Civic

Front

After-NAO

Ind.

ETW

A



ARB

12/7/2019

53

Sienna

Rear

OES-NAO

B

ETW

B



ARB

12/8/2019

54

Sienna

Rear

OES-NAO

B

HLW

B





12/9/2019

55

Sienna

Rear

After-NAO

N

ETW

B





12/10/2019

56

Sienna

Rear

After-NAO

N

HLW

B





12/11/2019

57

Civic

Rear (Drum)

After-NAO

Ind.

ETW

A



ARB

12/12/2019


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix F

Test
Day
#

Test
Vehicle

Front/Rear

Pad

Material

BMC
Leaf
Mark

Wheel
load

# Replicates

Ref.
repeat
#

Teflon

Filter

Analysis

Test Date

58

Camry

Rear

After-NAO

Ind.

ETW

A



ARB

12/19/2019

59

Camry

Rear

After-LM

A?

ETW

A



ARB

12/20/2019

60

Rogue

Front

OES-NAO

A

ETW

B



ARB

12/16/2019

61

Rogue

Front

OES-NAO

A

HLW

B





12/17/2019

62

Rogue

Front

After-NAO

Ind.

ETW

B





12/18/2019

63

F-150

Front

After-LM

A

ETW

A



ARB

12/20/2019

64

F-150

Front

After-LM

A

HLW

A



ARB

12/21/2019

65

F-150

Front

After-NAO

N

ETW

A



ARB

12/22/2019

66

Tunnel Blank 2

B





1/5/2020

67

F-150

Front

OES-NAO

N

ETW

N/A

4



1/16/2020

68

Sienna

Front

OES-NAO

A

HLW

A



ARB

12/23/2019

69

Sienna

Front

After-NAO

N

ETW

A



ARB

12/24/2019

70

Sienna

Front

After-NAO

N

HLW

A



ARB

12/25/2019

71

Camry

Front

After-NAO

Ind.

ETW

B





12/27/2019

72

Camry

Front

After-LM

A?

ETW

B





12/28/2019

73

Camry

Front

OES

A

ETW

WLTP B





1/21/2020

74

F-150

Rear

OES-NAO

A

ETW

B



ARB

12/29/2019

75

F-150

Rear

After-NAO

N

ETW

B





12/30/2019

76

F-150

Rear

After-LM

A

ETW

B





12/31/2019

77

F-150

Rear

OES-NAO

A

HLW

B





1/19/2020

78

F-150

Rear

After-LM

A

HLW

B





1/25/2020

79

Prius

Rear

OES-NAO

A

ETW

B



ARB

1/12/2020

80

Prius

Rear

After-NAO

Ind.

ETW

B





1/28/2020

81

Rogue

Rear

OES-NAO

A

ETW

B





1/4/2020

82

Rogue

Rear

After-NAO

Ind.

ETW

B





1/4/2020

83

Rogue

Rear

After-NAO

Ind.

HLW

B





1/5/2020

84

Civic

Front

OES-NAO

A

ETW

B





1/10/2020

85

Civic

Front

After-NAO

Ind.

ETW

B





1/11/2020


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix F

Test
Day
#

Test
Vehicle

Front/Rear

Pad

Material

BMC
Leaf
Mark

Wheel
load

# Replicates

Ref.
repeat
#

Teflon

Filter

Analysis

Test Date

86

Sienna

Rear

OES-NAO

B

HLW

A



ARB

1/13/2020

87

Sienna

Rear

After-NAO

N

ETW

A



ARB

1/29/2020

88

Sienna

Rear

After-NAO

N

HLW

A



ARB

1/14/2020

89

F-150

Front

OES

N

ETW

WLTP B





1/17/2020

90

F-150

Front

OES-NAO

N

ETW

N/A

5

ARB

1/17/2020


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix G

Appendix G

CVS Flow Setting Results


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix G

This appendix presents the temperatures measured to validate the CVS air flowrate
setting. Brake rotor temperatures as measured on the proving ground track over the
WLTP-Brake Trip 10 are presented as the target for matching. The measured brake
rotor temperatures from the same speed trace run on the dynamometer are presented
for comparison. Results are shown for the final selected flowrate for each brake
assembly (shown in Table 14 of the main report). Note that the selection method biased
the temperature match to be best around the peak temperature for each assembly;
ERG's literature search indicated that high temperatures represent the driving mode in
which brake emissions could be expected to be highest4, and matching at elevated
temperatures also allowed a similar temperature range to be covered during each test
as the track given that tests begin at room temperature. Temperature traces are
presented for the front and rear assemblies of each vehicle.

Temperature and speed traces for each assembly are presented; traces labeled as
"PG" represent the proving ground track testing, and those labeled "D" represent the
dynamometer test. At the right of each trace are corresponding box plots indicating
various statistics on the proving ground and dynamometer temperature traces. The
central line of each box is the median temperature, and the top and bottom of each box
are the 75th and 25th percentile values, respectively. The ends of each bar display the
maximum and minimum values for each trace.

4 B.D. Garg, S.H. Cadle, P.A. Mulawa, P.J. Groblicki, C. Laroo, G.A. Parr, "Brake Wear
Particulate Matter Emissions," Environmental Science and Technology, 2000, Volume
34, Number 21, pages 4463-4469, DOI: 10.1021/es001108h.


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix G

Camry Temperature Traces for CVS Flow Rate Setting

Front

¦ PG/km/h

D/km/h	PG/oC 	 D / oC


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix G

Civic Temperature Traces for CVS Flow Rate Setting

¦ PG/km/h

D/km/h — PG/oC 	 D / oC


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix G

F-150 Temperature Traces for CVS Flow Rate Setting

Front

¦ PG/km/h

D / km/h

PG/oC

D/oC

T3

CD

Q.

00



200



180



160

dJ

140

3

120

CD



&_

(1)

100

Q.



£

80

oj



i-

60



40



20



0

1000 2000 3000 4000
Time / s

5000 6000

Rear

¦ PG/km/h

D / km/h

PG/oC

D/oC

2000 3000 4000
Time / s

5000 6000

200
180
160
Pl40
E! 120

3

re 100
&_

o. 80
a; 60
40
20
0

200
180
160
p" 140
£! 120

3

H ioo


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix G

Prius Temperature Traces for CVS Flow Rate Setting

¦ PG/km/h

D / km/h

PG/oC

D/oC

1000 2000 3000
Time / s

4000

5000 6000

¦PG/km/h

D / km/h

PG/oC

D/oC

CD

^ I

(D QJ
Q- Q_

oo £
CD

200
180
160
140
120
100
80
60
40
20
0

1000 2000 3000 4000
Time / s

5000 6000



200



180



160

u

o

140


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix G

Rogue Temperature Traces for CVS Flow Rate Setting

Front

• PG/km/h

D/km/h — PG/oC 	 D / oC

2000 3000 4000
Time / s

5000 6000

Rear

• PG/km/h

D/km/h — PG/oC 	 D / oC



ro

100

(D



Q.

HO

£



CD
1-

60



40



20



0

PG Dyn

PG

Dyn


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix G

Sienna Temperature Traces for CVS Flow Rate Setting

Front

¦ PG/km/h

D / km/h

PG/oC

D/oC

1000 2000 3000 4000
Time / s

5000 6000

Rear

¦ PG/km/h

D / km/h

PG/oC

D/oC

5000 6000



200



180



160

u

o

140



100



ro

100


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix H

Appendix H

Tabulated Test Result Summary


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix H

Test results and measured operational parameters are presented in this appendix. "Test day" refers to the test day number
given in the Task 2 test matrix, not necessarily the actual order in which tests were conducted. "F/R" indicates the front or rear
axle assembly being tested. "Pad" indicates the friction material used during the test; OES indicates the OES material, all of
which were NAO, NAO indicates Aftermarket NAO, and LM indicates the aftermarket low metallic. "Avg. Torq" refers to the
average torque applied to the dynamometer drive shaft by the hydraulic brakes during the test, and "Avg. Press" refers to the
average brake-circuit hydraulic pressure during the test. "PN" refers to particle number count as measured by the condensation
particle counter (CPC), which measures between the range of 23 nm to 2.1 |jm. The three size cutpoints are given for the
gravimetric mass measurement of the 100S4, and the PM10, as measured by the 47mm Teflon PMS, is given in next column.
The rightmost columns indicate the 100S4 PM2.5 and PM10 values presented on the basis of mass per second of braking to
be on the same basis as MOVES opmode 0.























Total

100S4

100S4





100S4

100S4



















PN 23



Pad

Sg3

Sg4

100S4



PM2.5

PM10

Test













Avg.

Peak

nm-2.1



and

(PM2.5-

(PM1-

Aft Filter

Brake

per s

per s



1-





Avg.

Avg.

Rotor

Rotor

Hm

Avg. CPC

Rotor

10)

2.5)

(PM < 1)

PMio per

braking

braking

Day

Test

/



Tst

Torq

Press

Temp

Temp

CPC

PN 23 nm -

Wear

Emission

Emission

Emission

PMS

time

time

#

Veh.

R

Pad

Wt

N-m

kPa

°C

°C

#/cc

2.1 nm #

(g)a

mg/mi

mg/mi

mg/mi

(mg/mi)

(mg/s)

(mg/s)

6

F-150

F

OES

ETW

259

594

103.2

234.9

29.0

1.072E+11

3.7

0.7512

0.4933

0.3315

1.3432

0.0430

0.0822

7

F-150

F

OES

HLW

316

674

119.2

251.2

33.9

1.252E+11

3.3

1.1711

0.6792

0.5587

1.5741

0.0645

0.1256

8

Camry

F

OES

ETW

146

570

110.4

246.3

83.6

4.532E+10

6.2

1.2773

0.4987

0.1436

1.6942

0.0335

0.1001

9

Civic

R

OES

ETW

34

832

125.2

231.7

91.9

4.977E+10

1.5

0.1156

0.1472

0.1120

0.3493

0.0135

0.0195

10

Camry

R

OES

ETW

51

497

114.3

181.9

118.7

6.430E+10

2.7

0.6329

0.4848

0.1752

1.0943

0.0344

0.0674

11

Sienna

F

OES

ETW

183

493

117.6

243.4

255.3

1.341E+11

9.9

2.4060

1.0530

0.3115

3.2632

0.0711

0.1966

12

Prius

F

OES

ETW

58

302

73.1

186.0

41.1

8.311E+10

2.9

0.5794

0.4008

0.1483

1.3922

0.0286

0.0588

13

Sienna

R

OES

ETW

88

792

134.1

237.5

149.6

8.110E+10

3.7

0.9403

0.5772

0.0831

1.3773

0.0344

0.0835

14

Rogue

F

OES

ETW

152

646

118.1

236.5

577.1

3.130E+11

6.2

1.6787

1.0086

0.3467

2.7309

0.0707

0.1582

15

Camry

F

OES

ETW

133

526

110.4

245.3

138.0

7.522E+10

6.2

1.5849

0.5995

0.1228

2.0689

0.0377

0.1203

16

F-150

R

OES

ETW

95

430

73.5

119.1

184.4

1.000E+11

6.5

0.7476

0.5345

0.1941

1.3135

0.0380

0.0770

17

Tunnel Blank 1











0.2











0.0291





18

Rogue

R

NAO

HLW

58

599

76.8

162.7

80.4

4.405E+10

3.3

0.4307

0.3169

0.1102

0.7834

0.0223

0.0447

19

Rogue

R

OES

ETW

56

661

70.2

142.3

151.8

8.223E+10

2.6

0.7191

0.4743

0.1219

1.2108

0.0311

0.0686

20

Rogue

R

NAO

ETW

50

564

75.3

154.2

132.0

7.227E+10

3.3

0.4856

0.3565

0.1291

0.9789

0.0253

0.0506

21

Camry

R

OES

ETW

51

552

116.2

186.2

107.4

5.819E+10

2.8

0.6533

0.4907

0.1183

1.1458

0.0318

0.0658


-------
Te:

Da

#

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

Contract 68HE0C18C0001, WA 1 -04

Appendix H

Test
Veh.

F
/
R

Pad

Tst
Wt

Avg.
Torq
N-m

Avg.
Press
kPa

Avg.
Rotor
Temp
°C

Peak
Rotor
Temp
°C

PN 23
nm-2.1
Urn
CPC
#/cc

Avg. CPC
PN 23 nm -
2.1 urn #

Total
Pad
and
Rotor
Wear

(g)a

100S4
Sg3
(PM2.5-

10)
Emission
mg/mi

100S4
Sg4
(PM1-
2.5)
Emission
mg/mi

100S4
Aft Filter
(PM < 1)
Emission
mg/mi

Brake
PMio per

PMS
(mg/mi)

100S4
PM2.5
per s
braking
time
(mg/s)

Camry

NAO

ETW

45

413

127.7

206.6

128.1

7.014E+10

0.7179

0.5572

0.1571

1.2446

0.0372

Camry

LM

ETW

51

552

111.1

185.6

186.6

1.022E+11

3.1

1.7370

0.6974

0.2095

2.2983

0.0473

F-150

OES

ETW

249

549

103.5

237.1

16.4

6.123E+10

5.8

0.6689

0.3496

0.3622

1.0859

0.0371

Civic

OES

ETW

32

788

96.9

153.9

58.1

3.179E+10

5.2

0.0731

0.0804

0.0713

0.3296

0.0079

Civic

NAO

ETW

33

739

92.8

175.6

104.2

5.707E+10

3.2

0.7868

0.3270

0.1264

1.3536

0.0236

Prius

OES

ETW

20

246

62.3

156.1

17.7

3.630E+10

1.1

0.2391

0.2019

0.0944

0.5442

0.0154

Prius

NAO

ETW

21

274

63.7

160.9

25.6

5.229E+10

1.3

0.1784

0.1582

0.1449

0.3637

0.0158

F-150

NAO

ETW

84

329

70.9

117.8

269.0

1.475E+11

3.4

1.4611

0.9500

0.2989

2.5550

0.0651

F-150

LM

ETW

88

579

67.4

117.6

144.5

7.918E+10

1.4611

0.9500

0.2989

2.6306

0.0651

F-150

OES

HLW

105

453

86.5

146.2

164.0

9.176E+10

5.7

0.8994

0.4939

0.1011

1.1918

0.0310

F-150

LM

HLW

111

631

72.8

125.4

139.0

7.619E+10

3.2

1.2050

0.5050

0.0903

1.5381

0.0310

Sienna

OES

ETW

172

504

113.8

248.3

240.0

1.316E+11

7.3

1.6431

0.8754

0.1959

2.6295

0.0559

Sienna

OES

HLW

197

523

121.8

268.3

266.6

1.462E+11

7.1

2.3284

1.1091

0.2546

3.3531

0.0711

Sienna

NAO

ETW

170

520

120.9

250.8

663.3

3.641E+11

6.4

1.8041

1.2351

0.3557

3.4047

0.0829

Sienna

NAO

HLW

189

530

129.3

260.6

424.5

2.331E+11

1.7

3.0364

1.7150

0.5246

5.3524

0.1168

Rogue

OES

HLW

159

622

139.4

264.6

438.4

2.403E+11

0.5

3.3931

1.6843

0.5282

5.5660

0.1154

Rogue

NAO

ETW

146

530

122.4

255.4

312.3

1.712E+11

6.8

2.1538

1.1387

0.3025

2.5123

0.0751

F-150

OES

HLW

257

559

104.1

253.6

36.8

1.372E+11

4.5

0.7348

0.4031

0.3193

1.0776

0.0377

F-150

LM

ETW

226

557

91.6

213.1

62.2

2.322E+11

16.6

3.9896

1.9996

0.4236

5.5549

0.1264

F-150

LM

HLW

276

635

99.9

244.0

88.4

3.301E+11

18.3

7.5629

2.6190

0.2579

9.6947

0.1500

F-150

NAO

ETW

226

651

98.5

214.4

31.8

1.188E+11

5.3

0.7222

0.4763

0.0614

1.3003

0.0280

F-150

OES

ETW

241

456

106.4

218.9

20.4

7.612E+10

3.8

0.8705

0.5081

0.0860

1.2329

0.0310

F-150

OES

ETW

228

487

96.9

211.4

16.3

6.070E+10

3.9

0.6250

0.3792

0.0061

1.0556

0.0201

Camry

NAO

ETW

135

505

103.1

234.5

188.9

1.035E+11

4.5

1.7274

0.7565

0.2221

2.5411

0.0510

Camry

LM

ETW

144

523

107.0

243.2

383.0

2.084E+11

7.9

3.2827

0.8459

0.2772

4.1068

0.0586

Camry

OES

ETW

149

613

76.9

165.6

17.0

1.664E+10

6.4

0.3780

0.1267

0.0383

0.5591

0.0126


-------
Te:

Da

#

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

73

Contract 68HE0C18C0001, WA 1 -04

Appendix H

Test
Veh.

Pad

Tst
Wt

Avg.
Torq
N-m

Avg.
Press
kPa

Avg.
Rotor
Temp
°C

Peak
Rotor
Temp
°C

PN 23
nm-2.1
Urn
CPC
#/cc

Avg. CPC
PN 23 nm -
2.1 urn #

Total
Pad
and
Rotor
Wear

(g)a

100S4
Sg3
(PM2.5-

10)
Emission
mg/mi

100S4
Sg4
(PM1-
2.5)
Emission
mg/mi

100S4
Aft Filter
(PM < 1)
Emission
mg/mi

Brake
PMio per

PMS
(mg/mi)

100S4
PM2.5
per s
braking
time
(mg/s)

Prius

OES

ETW

56

300

64.0

173.2

20.1

4.113E+10

2.6

0.4632

0.2710

0.0539

0.8795

0.0169

Prius

NAO

ETW

56

304

67.8

190.3

66.8

1.368E+11

2.6

0.4466

0.3859

0.1820

0.9305

0.0296

Prius

NAO

ETW

54

257

77.0

183.5

334.9

6.860E+11

0.8

0.6171

0.5630

0.3135

1.7420

0.0457

Civic

OES

ETW

106

542

104.7

235.4

263.3

1.446E+11

5.4

1.0051

0.4334

0.0668

1.4537

0.0261

Civic

NAO

ETW

99

531

107.1

239.9

364.0

1.999E+11

1.1477

0.6194

0.1725

1.7565

0.0413

Sienna

OES

ETW

74

657

130.5

239.1

144.3

7.919E+10

3.4

0.6065

0.6103

0.0849

1.6315

0.0362

Sienna

OES

HLW

84

765

149.1

274.5

184.3

1.012E+11

3.6

1.5234

0.8009

0.1372

2.3943

0.0489

Sienna

NAO

ETW

79

715

138.9

250.7

79.2

4.349E+10

4.7

0.6038

0.3699

0.0488

1.0051

0.0218

Sienna

NAO

HLW

92

822

156.2

274.0

71.6

3.930E+10

4.2

0.8197

0.4015

0.0542

0.8939

0.0238

Civic

NAO

ETW

36

755

113.9

218.0

65.9

3.609E+10

2.6

0.6065

0.2886

0.0135

0.9999

0.0158

Camry

NAO

ETW

46

387

145.6

214.9

143.5

7.865E+10

3.5

0.6286

0.5292

0.1481

1.2192

0.0353

Camry

LM

ETW

48

511

122.1

192.1

342.0

1.874E+11

2.1

2.2237

1.0269

0.2528

3.2695

0.0667

Rogue

OES

ETW

127

548

122.6

231.7

412.5

2.269E+11

9.5

3.0699

1.5073

0.5092

4.9036

0.1051

Rogue

OES

HLW

148

608

136.7

253.5

314.4

1.723E+11

13.8

3.7672

1.5381

0.5390

5.7154

0.1083

Rogue

NAO

ETW

133

472

131.5

247.3

268.9

1.474E+11

5.9

1.7445

1.0346

0.2664

2.7544

0.0678

F-150

LM

ETW

261

637

88.4

210.8

105.1

3.925E+11

14.5

6.5346

2.7330

0.7429

8.3941

0.1812

F-150

LM

HLW

291

626

110.2

264.1

115.7

4.277E+11

18.9

10.8371

3.8848

0.7797

13.199

0.2432

F-150

F NAO

ETW

242

610

102.1

227.3

53.6

2.004E+11

2.9

1.8582

0.9182

0.3070

2.6385

0.0639

Tunnel Blank 2

0.4

0.0962

F-150

OES

ETW

261

630

69.1

159.3

14.5

8.013E+10

4.2

0.5668

0.2393

0.2140

1.3995

0.0346

Sienna

OES

HLW

198

577

124.5

259.6

275.8

1.514E+11

10.1

2.6269

1.2261

0.2320

3.9597

0.0760

Sienna

NAO

ETW

156

482

115.1

248.2

803.5

4.448E+11

3.8

2.6351

1.3625

0.5110

4.0958

0.0977

Sienna

NAO

HLW

184

573

126.7

274.6

484.0

2.656E+11

8.4

4.0160

1.7572

0.6438

6.2149

0.1252

Camry

NAO

ETW

141

514

104.8

254.2

189.0

1.037E+11

7.2

1.9406

0.8306

0.2826

3.1981

0.0580

Camry

LM

ETW

143

505

101.1

223.5

146.0

8.011E+10

4.8

2.6269

0.6129

0.1381

3.1771

0.0392

Camry

OES

ETW

143

591

79.3

178.4

24.7

2.273E+10

5.2

0.5023

0.1566

0.0475

0.6961

0.0156


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix H























Total

100S4

100S4





100S4

100S4



















PN 23



Pad

Sg3

Sg4

100S4



PM2.5

PM10

Test













Avg.

Peak

nm-2.1



and

(PM2.5-

(PM1-

Aft Filter

Brake

per s

per s



h





Avg.

Avg.

Rotor

Rotor

Urn

Avg. CPC

Rotor

10)

2.5)

(PM < 1)

PMio per

braking

braking

Day

Test

/



Tst

Torq

Press

Temp

Temp

CPC

PN 23 nm -

Wear

Emission

Emission

Emission

PMS

time

time

#

Veh.

R

Pad

Wt

N-m

kPa

°C

°C

#/cc

2.1 nm #

(g)a

mg/mi

mg/mi

mg/mi

(mg/mi)

(mg/s)

(mg/s)

74

F-150

R

OES

ETW

92

414

73.9

144.6

203.4

1.117E+11

5.9

0.9210

0.5715

0.1228

1.6133

0.0362

0.0842

75

F-150

R

NAO

ETW

79

323

72.2

119.7

224.8

1.234E+11

6

2.0653

1.0736

0.3268

3.4768

0.0730

0.1807

76

F-150

R

LM

ETW

79

431

68.6

120.2

306.3

1.681E+11

4.4

1.4341

0.7892

0.2266

2.4870

0.0530

0.1277

77

F-150

R

OES

HLW

99

349

91.4

150.5

229.9

1.263E+11

4.8

2.0013

0.9708

0.2411

3.2885

0.0632

0.1675

78

F-150

R

LM

HLW

97

422

83.6

131.7

383.3

2.106E+11

2.4

2.2069

1.0535

0.3422

3.6268

0.0728

0.1879

79

Prius

R

OES

ETW

20

261

60.0

151.3

20.3

4.170E+10

0.9

0.2083

0.1915

0.0472

0.4232

0.0124

0.0233

80

Prius

R

NAO

ETW

21

242

66.6

168.4

20.0

4.092E+10

0.7

0.1649

0.1818

0.1551

0.3686

0.0176

0.0262

81

Rogue

R

OES

ETW

51

631

68.2

145.3

163.8

8.976E+10

2.8

0.7546

0.4059

0.1156

1.3110

0.0272

0.0665

82

Rogue

R

NAO

ETW

46

477

72.2

141.3

104.9

5.747E+10

2.3

0.5387

0.3816

0.0831

0.4514

0.0242

0.0523

83

Rogue

R

NAO

HLW

51

510

79.7

168.3

127.4

6.987E+10

3.6

0.6408

0.4449

0.0822

1.1996

0.0275

0.0609

84

Civic

F

OES

ETW

112

560

107.4

233.0

157.5

8.669E+10

3.3

0.9255

0.4161

0.0930

1.1391

0.0265

0.0748

85

Civic

F

NAO

ETW

103

532

105.9

245.6

409.3

2.254E+11

4.2

1.4231

0.7178

0.1923

2.2485

0.0475

0.1217

86

Sienna

R

OES

HLW

92

811

155.6

273.5

193.9

1.068E+11

4.2

1.5052

0.7826

0.1535

2.4259

0.0488

0.1273

87

Sienna

R

NAO

ETW

84

738

143.6

248.5

84.2

4.623E+10

1.2

0.6640

0.3804

0.0650

1.1487

0.0232

0.0578

88

Sienna

R

NAO

HLW

96

836

167.3

298.7

93.1

5.105E+10

8.5

0.7798

0.4349

0.1129

1.1979

0.0286

0.0692

89

F-150

F

OES

ETW

251

613

72.9

159.5

14.2

7.996E+10

3.5

0.5392

0.2495

0.2728

0.7214

0.0398

0.0810

90

F-150

F

OES

ETW

259

557

100.2

221.6

14.9

5.553E+10

2.2

0.7425

0.3616

0.2149

1.1561

0.0301

0.0688

a - Note that the total pad and rotor wear is measured as the difference between the components' weight before installation into the
dynamometer and the weight after removal from the dynaomomter. As such, it includes mass lost during both the burnish cycle and the
test cycle.


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix I

Appendix I

LINK Test Result Reports


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix I

This appendix references the attached LINK test result reports for each test day. Each
filename references the test day from the matrix as well as abbreviations of the model,
axle, friction material, LeafMark, testweight, and replicate count. The reports are in
Excel format with multiple tabs containing a variety of test parameters, result tables,
plots, and before and after photographs and measurements of the test components.

The final tab, "Test Data", contains top-down data from the continuous measurements
made during the test day. To reduce file space, each row represents a single event (ie
either a deceleration (braking), a cruise, or an acceleration. Values are given as
average or max/mins of the measurements taken during each event. The first column,
CSV File, indicates the segment of the test day. In general, segment 2 is the burnish
and 4 is the test cycle (this can be verified by cross-referencing the times given in the
Cumulative Schedule Duration column). Note that, in these automated test reports,
values given on a per-distance basis are based on the actual dyno distance traveled,
not the CBDC represented distance as in the main report. For analysis of the data, any
analysis is encouraged to use the total emissions and divide by the CBDC represented
distance as described in the report.

•	E.Test-6-F150-FA-OES-NAO-N-ETW-Refn.xlsx

•	E.Test-7-F150-FA-OES-NAO-N-HLW-rA.xlsx

•	E.Test 8-CAM-FA-OES-NAO-A-ETW-rA.xlsx

•	E.Test 9-CIV-RA-OES-NAO-N-ETW-rA.xlsx

•	E.Test 10-CAM-RA-OES-NAO-A-ETW-rA.xIsx

•	E.Test-11-SIE-FA-OES-NAO-A-ETW-rA.xlsx

•	E.Test-12-PRI-FA-OES-NAO-A-ETW-rA.xlsx

•	E.Test-13-SIE-RA-OES-NAO-B-ETW-rA.xlsx

•	E.Test-14-ROG-FA-OES-NAO-A-ETW-rA.xlsx

•	E.Test-15-CAM-FA-OES-NAO-A-ETW-rB.xlsx

•	E.Test 16-F150-RA-OES-NAO-A-ETW-rA.xlsx

•	E.Test-18-ROG-RA-AM1 -N AO-N-H LW-rA.xIsx

•	E.Test-19-ROG-RA-OES-NAO-A-ETW-rA.xlsx

•	E.Test-20-ROG-RA-AM1-NAO-N-ETW-rA.xlsx

•	E.Test 21 -CAM-RA-OES-NAO-A-ETW-rB.xIsx

•	E.Test-22-CAM-RA-AM1 -NAO-N-ETW-rB.xIsx

•	E.Test-23-CAM-RA-AM2-LM-A-ETW-rB.xlsx

•	E.Test-24-F150-FA-OES-NAO-N-ETW-rNA.xlsx

•	E.Test-25-CIV-RA-OES-NAO-N-ETW-rB.xlsx

•	E.Test-26-CIV-RA-AM1-NAO-N-ETW-rB.xIsx

•	E.Test-27-PRI-RA-OES-NAO-A-ETW-rA.xlsx

•	E.Test-28-PRI-RA-AM1-NAO-N-ETW-rA.xlsx

•	E.Test-29-F150-RA-AM1-NAO-N-ETW-rA.xlsx

•	E.Test-30-F150-RA-AM2-LM-A-ETW-rA.xlsx

•	E.Test-31-F150-RA-OES-NAO-A-H LW-rA.xIsx

•	E.Test-32-F150-RA-AM2-LM-A-HLW-rA.xlsx

•	E.Test-33-SIE-FA-OES-NAO-A-ETW-rB.xlsx

•	E.Test-34-SIE-FA-OES-NAO-A-HLW-rB.xlsx

•	E.Test-35-SIE-FA-AM1-NAO-N-ETW-rB.xIsx


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix I

•	E.Test-36-SIE-FA-AM1-NAO-N-HLW-rB.xlsx

•	E.Test-37-ROG-FA-OES-NAO-A-HLW-rA.xlsx

•	E.Test-38-ROG-FA-AM1 -NAO-N-ETW-rA.xIsx

•	E.Test-39-F150-FA-OES-NAO-N-HLW-rB.xlsx

•	E.Test-40-F150-FA-AM2-LM-A-ETW-rB.xlsx

•	E.Test-41-F150-FA-AM2-LM-A-HLW-rB.xlsx

•	E.Test-42-F150-FA-AM1 -NAO-N-ETW-rB.xIsx

•	E.Test-43-F150-FA-OES-NAO-N-ETW-rNA.xlsx

•	E.Test-44-F150-FA-OES-NAO-N-ETW-rNA.xlsx

•	E.Test-45-CAM-FA-AM1-NAO-N-ETW-rA.xIsx

•	E.Test-46-CAM-FA-AM2-LM-A-ETW-rA.xlsx

•	E.Test-47-CAM-FA-OES-A-ETW-rWLTP A.xlsx

•	E.Test-48-PRI-FA-OES-NAO-A-ETW-rB.xlsx

•	E.Test-49-PRI-FA-AM1-NAO-N-ETW-rA.xIsx

•	E.Test-50-PRI-FA-AM1-NAO-N-ETW-rB.xIsx

•	E.Test-51 -CIV-FA-OES-NAO-A-ETW-rA.xIsx

•	E.Test-52-CIV-FA-AM1-NAO-N-ETW-rA.xIsx

•	E.Test-53-SIE-RA-OES-NAO-B-ETW-rB.xlsx

•	E.Test-54-SIE-RA-OES-NAO-B-HLW-rB.xlsx

•	E.Test-55-SIE-RA-AM1-NAO-N-ETW-rB.xIsx

•	E.Test-56-SIE-RA-AM1-NAO-N-HLW-rB.xlsx

•	E.Test-57-CIV-RA-AM1-NAO-N-ETW-rA.xIsx

•	E.Test-58-CAM-RA-AM1-NAO-N-ETW-rA.xIsx

•	E.Test-59-CAM-RA-AM2-LM-A-ETW-rA.xlsx

•	E.Test-60-ROG-FA-OES-NAO-A-ETW-rB.xlsx

•	E.Test-61-ROG-FA-OES-NAO-A-HLW-rB.xlsx

•	E.Test-62-ROG-FA-AM1-NAO-N-ETW-rB.xIsx

•	E.Test-63-F150-FA-AM2-LM-A-ETW-rA.xlsx

•	E.Test-64-F150-FA-AM2-LM-A-HLW-rA.xlsx

•	E.Test-65-F150-FA-AM1 -NAO-N-ETW-rA.xIsx

•	E.Test-67-F150-FA-OES-NAO-N-ETW-WLTP rA.xIsx

•	E.Test-68-SIE-FA-OES-NAO-A-HLW-rA.xlsx

•	E. T est-69-S IE-FA-AM1 -NAO-N-ETW-rA.xIsx

•	E. Test-70-S I E-FA-AM 1 -N AO-N-H LW-rA.xIsx

•	E.Test-71-CAM-FA-AM1 -NAO-N-ETW-rB.xIsx

•	E.Test-72-CAM-FA-AM2-LM-A-ETW-rB.xlsx

•	E.Test-73-CAM-FA-OES-A-ETW-WLTP rB.xIsx

•	E.Test-74-F150-RA-OES-NAO-A-ETW-rB.xlsx

•	E.Test-75-F150-RA-AM1-NAO-N-ETW-rB.xIsx

•	E.Test-76-F150-RA-AM2-LM-A-ETW-rB.xlsx

•	E.Test-77-F150-RA-OES-NAO-A-HLW-rB.xlsx

•	E.Test-78-F150-RA-AM2-LM-A-HLW-rB.xlsx

•	E.Test-79-PRI-RA-OES-NAO-A-ETW-rB.xlsx

•	E.Test-80-PRI-RA-AM1-NAO-N-ETW-rB.xIsx

•	E.Test-81 -ROG-RA-OES-NAO-A-ETW-rB.xIsx

•	E.Test-82-ROG-RA-AM1-NAO-N-ETW-rB.xIsx


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix I

•	E.Test-83-ROG-RA-AM1-NAO-N-HLW-rB.xlsx

•	E.Test-84-CIV-FA-OES-NAO-A-ETW-rB.xlsx

•	E.Test-85-CIV-FA-AM1-NAO-N-ETW-rB.xlsx

•	E.Test-86-SIE-RA-OES-B-HLW-rA.xlsx

•	E.Test 87-SIE-RA-AM1-NAO-N-ETW-rA.xlsx

•	E.Test 88-SIE-RA-AM1-NAO-N-HLW-rA.xlsx

•	E.Test-89-F150-FA-OES-N-ETW-WLTP rB.xIsx

•	E.Test-90-F150-FA-OES-N-ETW-rNA.xlsx

The two tunnel blank tests are:

•	TBIank.Test 2_Test day 17_ROG-RA-rA.xlsx

•	TBIank.Test 3_Test day 66_ROG-RA-rB.xlsx


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix J

Appendix J

Teflon Filter Masses and Weight Gains


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix J

The Appendix 3 table includes the Teflon filter weights for each test. The table presents
the test information such as matrix test day, vehicle and friction material, then initial and
final filter masses along with the filter weight gain. Measured masses were corrected by
LINK according to 40 CFR 1065.690.

Test
Day

Vehicle

Axle

FM

Initial Mass,
mg

Final mass,
mg

Gain, mg

6

2015 Ford F-150

Front

OES-NAO

397.300

397.420

0.120

7

2015 Ford F-150

Front

OES-NAO

398.036

398.186

0.150

8

2011 Toyota Camry

Front

OES-NAO

140.149

141.174

1.026

9

2013 Honda Civic

Rear

OES-NAO

393.764

393.971

0.206

10

2011 Toyota Camry

Rear

OES-NAO

137.179

137.878

0.699

11

2013 Toyota Sienna

Front

OES-NAO

393.233

395.281

2.048

12

2016 Toyota Prius

Front

OES-NAO

397.415

397.649

0.234

13

2013 Toyota Sienna

Rear

OES-NAO

392.106

392.965

0.859

14

2016 Nissan Rogue

Front

OES-NAO

136.503

138.207

1.704

15

2011 Toyota Camry

Front

OES-NAO

393.155

394.453

1.298

16

2015 Ford F-150

Rear

OES-NAO

390.599

391.409

0.810

17

Tunnel Blank 1

Holder 1

N/A

391.932

391.952

0.020

17

Tunnel Blank 1

Holder 2

N/A

392.407

392.424

0.017

18

2016 Nissan Rogue

Rear

AM1-NAO

393.805

394.271

0.466

19

2016 Nissan Rogue

Rear

OES-NAO

134.383

135.112

0.729

20

2016 Nissan Rogue

Rear

AM1-NAO

139.931

140.506

0.575

21

2011 Toyota Camry

Rear

OES-NAO

393.603

394.331

0.728

22

2011 Toyota Camry

Rear

AM1-NAO

392.703

393.475

0.772

23

2011 Toyota Camry

Rear

AM2-LM

393.149

394.566

1.417

24

2015 Ford F-150

Front

OES-NAO

389.913

390.016

0.103

25

2013 Honda Civic

Rear

OES-NAO

139.579

139.781

0.202

26

2013 Honda Civic

Rear

AM1-NAO

394.218

395.017

0.800

27

2016 Toyota Prius

Rear

OES-NAO

393.438

393.525

0.087

28

2016 Toyota Prius

Rear

AM1-NAO

140.551

140.608

0.057

29

2015 Ford F-150

Rear

AM1-NAO

138.622

140.263

1.641

30

2015 Ford F-150

Rear

AM2-LM

138.622

140.263

1.641

31

2015 Ford F-150

Rear

OES-NAO

136.124

136.880

0.757

32

2015 Ford F-150

Rear

AM2-LM

140.479

141.382

0.903

33

2013 Toyota Sienna

Front

OES-NAO

140.272

141.922

1.650

34

2013 Toyota Sienna

Front

OES-NAO

395.212

397.267

2.055

35

2013 Toyota Sienna

Front

AM1-NAO

385.961

388.048

2.087

36

2013 Toyota Sienna

Front

AM1-NAO

390.25509

393.5353

3.280213

37

2016 Nissan Rogue

Front

OES-NAO

137.16998

140.54

3.369999

38

2016 Nissan Rogue

Front

AM1-NAO

139.26646

140.8154

1.548947

39

2015 Ford F-150

Front

OES-NAO

398.76226

398.8559

0.093603

40

2015 Ford F-150

Front

AM2-LM

386.14182

386.6455

0.503642

41

2015 Ford F-150

Front

AM2-LM

397.83607

398.7045

0.868463

42

2015 Ford F-150

Front

AM1-NAO

394.15255

394.2648

0.112251

43

2015 Ford F-150

Front

OES-NAO

140.57629

140.6887

0.112449

44

2015 Ford F-150

Front

OES-NAO

391.06811

391.1615

0.093414

45

2011 Toyota Camry

Front

AM1-NAO

138.90356

140.5453

1.641713

46

2011 Toyota Camry

Front

AM2-LM

136.4385

138.9856

2.547125

47

2011 Toyota Camry

Front

OES-NAO

397.24408

397.7454

0.501278


-------
Te:

Da

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

66

67

68

69

70

71

72

73

74

75

76

77

78

79

80

81

82

83

84

85

86

87

88

89

90

Contract 68HE0C18C0001, WA 1 -04

Appendix J

Vehicle

Axle

FM

Initial Mass,
mg

Final mass,
mg

Gain, mg

2016 Toyota Prius

Front

OES-NAO

138.26622

138.4028

0.136554

2016 Toyota Prius

Front

AM1-NAO

139.52668

139.6785

0.151833

2016 Toyota Prius

Front

AM1-NAO

394.66786

394.9486

0.28079

2013 Honda Civic

Front

OES-NAO

138.64294

139.4748

0.831865

2013 Honda Civic

Front

AM1-NAO

138.72689

139.8098

1.082931

2013 Toyota Sienna

Rear

OES-NAO

139.70822

140.7141

1.005873

2013 Toyota Sienna

Rear

OES-NAO

397.6197

399.0959

1.476172

2013 Toyota Sienna

Rear

AM1-NAO

398.91986

399.5321

0.612234

2013 Toyota Sienna

Rear

AM1-NAO

393.55956

394.1107

0.551122

2013 Honda Civic

Rear

AM1-NAO

138.67537

139.2955

0.620158

2011 Toyota Camry

Rear

AM1-NAO

139.12557

139.8727

0.747152

2011 Toyota Camry

Rear

AM2-LM

140.84926

142.8771

2.027863

2016 Nissan Rogue

Front

OES-NAO

137.08673

140.11

3.023235

2016 Nissan Rogue

Front

OES-NAO

397.17737

400.5745

3.397165

2016 Nissan Rogue

Front

AM1-NAO

398.49649

400.1743

1.677852

2015 Ford F-150

Front

AM2-LM

137.55389

138.3013

0.747395

2015 Ford F-150

Front

AM2-LM

138.04894

139.2027

1.153721

2015 Ford F-150

Front

AM1-NAO

134.84165

135.078

0.236363

Tunnel Blank 2

Holder 1

N/A

137.5176

137.5451

0.027471

Tunnel Blank 2

Holder 2

N/A

399.15441

399.2404

0.085962

2015 Ford F-150

Rear

OES-NAO

395.69223

395.8723

0.180073

2013 Toyota Sienna

Front

OES-NAO

140.67803

143.1193

2.441265

2013 Toyota Sienna

Front

AM1-NAO

140.3819

142.9374

2.555467

2013 Toyota Sienna

Front

AM1-NAO

138.91632

142.7021

3.785822

2011 Toyota Camry

Front

AM1-NAO

391.42171

393.358

1.936318

2011 Toyota Camry

Front

AM2-LM

402.79979

404.6765

1.876715

2011 Toyota Camry

Front

OES-NAO

391.02543

391.6269

0.601518

2015 Ford F-150

Rear

OES-NAO

133.68903

134.6658

0.976815

2015 Ford F-150

Rear

AM1-NAO

401.15934

403.3029

2.143558

2015 Ford F-150

Rear

AM2-LM

400.94102

402.4835

1.542485

2015 Ford F-150

Rear

OES-NAO

394.76378

396.767

2.003203

2015 Ford F-150

Rear

AM2-LM

393.19452

395.4306

2.236035

2016 Toyota Prius

Rear

OES-NAO

137.48165

137.5553

0.073652

2016 Toyota Prius

REAR

AM1-NAO

138.53058

138.5905

0.059953

2016 Nissan Rogue

Rear

OES-NAO

395.58496

396.3594

0.774408

2016 Nissan Rogue

Rear

AM1-NAO

401.93869

402.2087

0.269963

2016 Nissan Rogue

Rear

AM1-NAO

394.0117

394.738

0.726318

2013 Honda Civic

Front

OES-NAO

393.52493

394.2314

0.70647

2013 Honda Civic

Front

AM1-NAO

393.7694

395.1889

1.41946

2013 Toyota Sienna

Rear

OES-NAO

138.36691

139.8178

1.450881

2013 Toyota Sienna

REAR

AM1-NAO

140.40394

141.1037

0.699745

2013 Toyota Sienna

Rear

AM1-NAO

136.94338

137.6687

0.725299

2015 Ford F-150

Front

OES-NAO

398.31912

398.4171

0.097973

2015 Ford F-150

Front

OES-NAO

139.29281

139.3964

0.103568


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix K

Appendix K

Vehicle-Level Particle Number Emission Rates by Speed Segment


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix K

This Appendix presents the CPC-measured vehicle-level particle number emission rates
for each vehicle model. Bar charts are given for each model, categorized by pad
material and grouped by average speed range. The values presented indicate the
vehicle-level particle count emission rate on a per-distance basis.

1.2E+10

I OES-NAO

Camry

I After-NAO

After-LM

1E+10

tc
QC
c
o

tc

CL

0)
>

8E+09

6E+09

^ 4E+09

y 2E+09

0-21

21-69
Speed Range (kph)

69+

1.2E+10

Civic

I OES-NAO ¦ After-NAO

1E+10

tc
QC
c
o

tc
CL

0)
>

8E+09

6E+09

^ 4E+09

y 2E+09

0-21

21-69
Speed Range (kph)

69+

F-150


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix K

1.8E+10
-E1.6E+10
$ 1.4E+10

cc

o 1.2E+10

"l/5
—

E 1E+10
8E+09
6E+09
4E+09
2E+09
0

_
CD
	1

0)
U

j=

CD
>

I OES-NAO

After-NAO ¦ After-LM

0-21

l

69+	0-21

Speed Range (kph)

69+

2.5E+10

Prius

I OES-NAO ¦ After-NAO

2.0E+10

ro
cc
c
o

£ 1.5E+10

_cy
u

ro
Q_

"a;
>
CD

1.0E+10

-g 5.0E+09

CD
>

0.0E+00

0-21

21-69
Speed Range (kph)

69+


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix K

1.6E+10
1 1.4E+10

(L)

1.2E+10

Ui 1E+10

8E+09

g> 6E+09

4E+09

a>
>
0)

	I

_aj

u

aj 2E+09
>

Rogue

I OES-NAO ¦ After-NAO

0-21

21-69
Speed Range (kph)

69+

6E+10

^ 5E+10

QC

£ 4E+10

3E+10

13 2E+10
>
a>

1E+10

Sienna (outlying value presented as measured)
¦ OES-NAO ¦ After-NAO

0-21	21-69	69+	0-21	21-69	69+

ETW	Speed Range (kph)	HLW


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix L

Appendix L

EEPS Particle Size Distributions


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix L

This appendix includes the particle size distributions measured by the EEPS ((5.6 - 560
nm)) during each test. Results are presented for the front and rear assemblies of each
vehicle, and are color coded by the friction material and, where applicable, the test
weight.

Camry Particle Size Distributions Measured by EEPS

Front

Rear

O

O

_ K ¦> Q?

0)- ,p-	ffi'

Particle Diameter Bin, nm

After-LM
¦After-NAO
¦OES-NAO

After-LM
¦After-NAO
¦OES-NAO

c

o
O

.32
o
t

CD
CL

T3
CD
N

TO

E

2.5E-01

2.0E-01

1.5E-01

1.0E-01

5.0E-02

0.0E+00

After-LM

After-LM

After-NAO

After-NAO

OES-NAO

OES-NAO

& <§> sfa c£>	t*5 #

^ £ £ £ £ &

Particle Diameter Bin, nm


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix L

Civic Particle Size Distributions Measured by EEPS

Front

Rear

1.8E-01

1.6E-01

g 1.4E-01
O

a; 1.2E-01
o

1.0E-01

t
CO
CL

"O

CD
N

TO

E

8.0E-02
6.0E-02
4.0E-02
2.0E-02
0.0E+00













—^ After-NAO After-NAO







OES-NAO OES-NAO























•	^	A & rvV

O' * (£>*	\*

 VP	^ Nv

Particle Diameter Bin, nm

o

o

2.0E-01

1.8E-01

1.6E-01

1.4E-01
a;

1.2E-01
£ 1.0E-01
"S 8.0E-02

N

I 6.0E-02
o 4.0E-02
2.0E-02
0.0E+00







After-NAO ^^After-NAO









OES-NAO OES-NAO

























r& & J° jS> NN , £>	J>  . J&

to* ^	(fo r£b & a& ,<\V k$* >& <<& r& <£>

 W N
Particle Diameter Bin, nm

* & <&J

F-150 Particle Size Distributions Measured by EEPS

Front

2.5E-01

g 2.0E-01

O

.22
o
t
CO
CL

"S 1.0E-01

2 1.5E-01

TO

E

5.0E-02

0.0E+00

After-LM

After-LM

After-LM HLW

After-LM HLW

After-NAO

After-NAO

OES-NAO

OES-NAO

OES-NAO

OES-NAO

OES-NAO

OES-NAO HLW

OES-NAO HLW

• ^ J* J** ^	A ^ n"V K&

O- • <§>• <§>• A- Nfc <3^- rS>'

1/ -3 V- -U -O NVJ •"*	rp tip $p

Particle Diameter Bin, nm


-------
Rear

Contract 68HE0C18C0001, WA 1 -04

Appendix L

2.0E-01

O 1.5E-01

_(L>

O

CL 1.0E-01

~C5
(D
N

| 5.0E-02

0.0E+00

LM
LM

LM HLW

LM HLW

NAO

NAO

NAO

NAO

NAOHLW
NAOHLW

& Jp JO & nN A	rQ,	<$> &  A &

9? Particle Diameter Bin, nm Prius Particle Size Distributions Measured by EEPS Front Rear O O Q) O CO CL T3 0) N "co E 2.0E-01 1.8E-01 1.6E-01 1.4E-01 1.2E-01 1.0E-01 8.0E-02 6.0E-02 4.0E-02 2.0E-02 0.0E+00 ^^After-NAO ^^After-NAO ^^OES-NAO ^^OES-NAO r& \N <£> •& <8>, & <£> ^s° ^ Particle Diameter Bin, nm o o _ <§> <& & <8> . & n* K> <£> ¦ $>¦ <§>• 9^'^' »v> $>' Ky N ^


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix L

Rogue Particle Size Distributions Measured by EEPS

Front

Rear

1.8E-01

1.6E-01

g 1.4E-01
O

oj 1.2E-01
o

t 1.0E-01
CL

¦c 8.0E-02
a>

= 6.0E-02

03

5 4.0E-02
Z

2.0E-02
0.0E+00







After-NAO —^After-NAO
OES-NAO —^OES-NAO
^^OES-NAOHLW —^OES-NAO HLW



























• ^ J? S1 n>N	J* A & ^ a*3

0-	(& (& {A	>&	rS> <£>

V &> <&> \v
Particle Diameter Bin, nm

*> ^

1.8E-01
1.6E-01

c

g 1.4E-01
O

oj 1.2E-01
o

£ 1.0E-01
CL

¦a 8.0E-02
a>

= 6.0E-02

05

| 4.0E-02
Z

2.0E-02
0.0E+00

¦After-NAO
•After-NAO HLW
•OES-NAO

After-NAO
After-NAO HLW
OES-NAO

^	& \N .£> <£>	^ K> 

%° >$° \*° $>' <§>' &' <§>' ^

 t) "O K{J

Particle Diameter Bin, nm


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix L

Sienna Particle Size Distributions Measured by EEPS

Front

Rear

After-NAO
After-NAO
After-NAO HLW
After-NAO HLW
OES-NAO
OES-NAO
OES-NAO HLW
OES-NAO HLW

• ^ a* J*	J* JP /V &	of

O' "b*	kN" Sb' rf?' r&' !&' <&' oP°	^ J&' .<$'

f£J ri,3 jjp	

2.5E-01

2.0E-01

3
O

O

Q)

.2 1.5E-01

-i—>

CO
CL

ft 1.0E-01

ro
£

Z 5.0E-02

0.0E+00

After-NAO

After-NAO

After-NAO HLW

—^After-NAO HLW

OES-NAO

OES-NAO

^^OES-NAOHLW

OES-NAO HLW

cJ3 ^	^ yP ^

o* <6*	Kb< A rfo f£b «&	c& (A iJ*	<^" r& <£>

 


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix L

Tunnel Blank Particle Size Distributions Measured by EEPS

2.0E-01

Particle Diameter Bin, nm


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix M

Appendix M

Zero Blank Results


-------
Contract 68HE0C18C0001, WA 1 -04

Appendix M

The table in this Appendix presents the PTFE filter and 100S4 impactor weights and
weight gains during the zero blank experiments.

Experiment

Measurement
Location

Initial (mg)

Final (mg)

Weight Gain (mg)

Zero Blank 1

100S4 Stg1

78.648

78.638

-0.010

100S4 Stg2

78.224

78.222

-0.002

100S4 Stg3

78.251

78.253

0.002

100S4 Stg4

78.303

78.304

0.001

100S4 AF

120.988

120.979

-0.009

PMS PM10

391.103

391.114

0.011

Zero Blank 2

100S4 Stg1

78.152

78.149

-0.003

100S4 Stg2

78.149

78.146

-0.003

100S4 Stg3

78.142

78.140

-0.002

100S4 Stg4

78.361

78.360

-0.001

100S4 AF

123.260

123.240

-0.020

PMS PM10

136.924

136.943

0.019

Zero Blank 3

100S4 Stg1

77.825

77.827

0.002

100S4 Stg2

77.863

77.862

-0.001

100S4 Stg3

78.048

78.048

0.000

100S4 Stg4

77.840

77.838

-0.002

100S4 AF

122.583

122.581

-0.002

PMS PM10

398.319

398.320

0.001

Zero Blank 4

100S4 Stg1

76.757

76.758

0.001

100S4 Stg2

77.828

77.823

-0.004

100S4 Stg3

78.090

78.088

-0.002

100S4 Stg4

77.822

77.819

-0.002

100S4 AF

121.674

121.667

-0.007

PMS PM10

140.422

140.489

0.067

Zero Blank 5

100S4 Stg1

77.837

77.836

0.000

100S4 Stg2

77.354

77.354

0.000

100S4 Stg3

78.256

78.255

-0.001

100S4 Stg4

77.661

77.660

-0.001

100S4 AF

121.697

121.693

-0.004

PMS PM10

383.604

383.621

0.017

Zero Blank 6

100S4 Stg1

76.758

76.761

0.003

100S4 Stg2

77.823

77.830

0.007

100S4 Stg3

78.088

78.089

0.001

100S4 Stg4

77.819

77.823

0.004

100S4 AF

121.667

121.673

0.006

PMS PM10

140.489

140.404

-0.085


-------