Data Collection of Heavy-Duty Drayage
Trucks in Houston-Galveston Port Area
£% United States
Environmental Protect
Agency
-------
Data Collection of Heavy-Duty Drayage
Trucks in Houston-Galveston Port Area
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. EP-C-06-080 and EP-C-12-017
Work Assignment 2-02
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-23-012
May 2023
-------
Table of Contents
Acronyms vii
Acknowledgments ix
Executive Summary ES-1
I.0 Introducti on 1-1
2.0 Background 2-1
3.0 Study Design 3-1
3.1 Workplan and Quality Assurance Project Plan (QAPP) Development 3-1
3.2 RSD Testing and Vehicle Sampling Plan Development 3-2
3.3 Develop Recruiting Plan 3-33
4.0 Study Preparation 4-1
4.1 Acquiring and Preparing PEMS and PAMS Equipment 4-1
4.2 PEMS and PAMS Mockups in Ann Arbor 4-9
4.3 Conduct Recruiting 4-9
4.4 Preliminary Visits to Candidate Establishments 4-11
5.0 PEMS and PAMS Testing 5-1
5.1 PEMS Testing 5-1
5.2 PAMS Installations 5-25
5.3 PEMS and PAMS Testing Summary 5-30
6.0 Data Analysis and QC 6-1
6.1 PEMS Data Processing and QC 6-1
6.2 PAMS Data Processing and QC 6-12
7.0 Study Results and Conclusions 7-1
7.1 Emissions results 7-1
7.2 PAMS Data 7-50
8.0 Lessons Learned and Program Observations 8-1
8.1 General Lessons and Recommendations 8-1
8.2 PEMS Lessons Learned 8-2
8.3 PAMS Lessons Learned 8-6
9.0 MSOD Data Conversion and Data Delivery 9-1
9.1 MySQL Database Delivery 9-1
10.0 References 10-1
II.0 Index of Appendices 11-1
The Appendices cited in this document have been provided separately in electronic format. An
index of these appendix files is provided at the end of this report.
l
-------
List of Figures
Figure 2.0-1 Terminals of the Port of Houston Analyzed in this Study 2-1
Figure 3.2-1 Location of RSD Testing Equipment at Port Entrance 3-3
Figure 3.2-2 RSD Test Equipment 3-4
Figure 3.2-3 Distribution of the Number of RSD Beam Blocks Per Vehicle in the Sample ..
3-6
Figure 3.2-4 Distribution of Drayage Vehicle Model Years in the Sample 3-8
Figure 3.2-5 Distribution of RSD Measured NOx Values in the Sample 3-9
Figure 3.2-6 Trend of Standard Deviation of NOx with Mean of NOx for Vehicles with
2 RSD Replicates 3-12
Figure 3.2-7 Trend of Standard Deviation of NOx with Mean of NOx for Vehicles with
3 RSD Replicates 3-13
Figure 3.2-8 Trend of Standard Deviation of NOx with Mean of NOx for Vehicles with
4 RSD Replicates 3-14
Figure 3.2-9 Trend of Standard Deviation of NOx with Mean of NOx for Vehicles with
5 RSD Replicates 3-15
Figure 3.2-10 Trend of Standard Deviation of NOx with Mean of NOx for Vehicles with
6 RSD Replicates 3-16
Figure 3.2-11 Trend of Standard Deviation of NOx with Mean of NOx for Vehicles with
7 RSD Replicates 3-17
Figure 3.2-12 Trend of Standard Deviation of NOx02 with Mean of NOx02 for Vehicles with
2 Replicates 3-18
Figure 3.2-13 Trend of Standard Deviation of NOx02 with Mean of NOx02 for Vehicles with
3 Replicates 3-19
Figure 3.2-14 Trend of Standard Deviation of NOx02 with Mean of NOx02 for Vehicles with
4 Replicates 3-20
Figure 3.2-15 Trend of Standard Deviation of NOx02 with Mean of NOx02 for Vehicles with
5 Replicates 3-21
Figure 3.2-16 Trend of Standard Deviation of NOx02 with Mean of NOx02 for Vehicles with
6 Replicates 3-22
Figure 3.2-17 Trend of Standard Deviation of NOx02 with Mean of NOx02 for Vehicles with
7 Replicates 3-23
Figure 3.2-18 Distribution of RSD NOx02 Measured Values in the Sample 3-25
Figure 3.2-19 Distribution of Z-values for Vehicles in the Sample 3-28
Figure 3.2-20 NOx Bin Assignment of Vehicles with Differing Numbers of RSD Replicates
3-29
Figure 4.1-1 SEMTECH-DS (left) and 3-Chamber Gravimetric PM Sampler (right) 4-3
Figure 4.1-2 Gaseous and PM System Used in Drayage Study 4-5
Figure 4.1-3 Gaseous-only Flowmeter Mounting on Drayage Truck 4-6
Figure 4.1-4 Isaac datalogger used Drayage study 4-8
Table 4.3-1 Summary of Drayage Recruiting, by Companies 4-10
Figure 5.0-1 Drayage Fieldwork Schedule, Dec 2009 - March 2010 5-1
Figure 5.1-1 Gaseous-only Installation on Truck with Bulkhead 5-6
Figure 5.1-2 PM/Gaseous Test Components with Generators on Roof 5-7
Figure 5.1-3 PM/Gaseous System on 2004 Kenworth 5-8
li
-------
Figure 5.1-4 Grav and MPS Box Outside of a 1993 Freightliner 5-8
Figure 5.1-5 Grav and MPS Box Outside 1993 Freightliner Day Cab 5-9
Figure 5.1-6 Gaseous-only Installation on 2003 Freightliner Fleet Truck 5-10
Figure 5.1-7 Gaseous-only Installation on 1998 International Sleeper Cab 5-10
Figure 5.1-8 PEMS and Gravimetric Sampler in Large Sleeper Cab 5-11
Figure 5.1-9 PEMS and Gravimetric Sampler in 1980 Kenworth Cabover 5-12
Figure 5.1-10 PEMS and Gravimetric Sampler in Cab with Window Removed 5-12
Figure 5.1-11 Sleeper Cab with full PM System and Window Removed 5-13
Figure 5.1-12 PEMS installed on Top Bunk of Sleeper Cab 5-14
Figure 5.1-13 PEMS On Crate in Sleeper Cab 5-14
Figure 5.1-14 Sample Lines Routed through Sleeper Cab Window 5-16
Figure 5.1-15 Sample Lines Routed through Sleeper Cab Vent 5-16
Figure 5.1-16 Sample Lines Routed through Toolbox 5-17
Figure 5.1-17 Sample Lines Routed through Passenger Window 5-17
Figure 5.1-18 PEMS Between Seats in Day Cab 5-18
Figure 5.1-19 PEMS Mounting Platform in Day Cab 5-19
Figure 5.1-20 Typical Under-Dash VI Connector Location 5-20
Figure 5.1-21 Typical Dash Front VI Connector Location 5-20
Figure 5.1-22 Optical Sensor with Cardboard Alignment Tool 5-21
Figure 5.1-23 Access to Filters within the Gravimetric Sampler Assembly 5-24
Figure 5.2-1 Switched Power Terminal used for Non-ECU PAMS Installations 5-27
Figure 5.2-2 PAMS Installed on a Seat Base 5-28
Figure 5.2-3 Typical PAMS GPS Mounting Location 5-29
Figure 5.2-4 PAMS wiring from a Deutsch Connector 5-29
Figure 5.3-1 Trucks PAMS Tested vs. RSD-Tested Fleet 5-33
Figure 6.1-1 Process for Calculating Brake Specific Power Output with ECU Data 6-8
Figure 6.1-2 Extrapolated lug curve for the Cat 3406E Engine Rated at 355 HP 6-10
Figure 6.2-1 ECU Engine Percent Load vs. Speed for a J1708 protocol truck 6-14
Figure 7.1-1 HC Emissions for 1978-1993 Trucks by VSP BIN, Port 7-3
Figure 7.1-2 HC Emissions for 1978-1993 Trucks by VSP BIN, Non-port 7-3
Figure 7.1-3 HC Emissions for 1994-1997 Trucks by VSP BIN, Port 7-4
Figure 7.1-4 HC Emissions for 1994-1997 Trucks by VSP BIN, Non-port 7-4
Figure 7.1-5 HC Emissions for 1998-2003 Trucks by VSP BIN, Port 7-5
Figure 7.1-6 HC Emissions for 1998-2003 Trucks by VSP BIN, Non-port 7-5
Figure 7.1-7 HC Emissions for 2004-2006 Trucks by VSP BIN, Port 7-6
Figure 7.1-8 HC Emissions for 2004-2006 Trucks by VSP BIN, Non-port 7-6
Figure 7.1-9 CO Emissions for 1978-1993 Trucks by VSP BIN, Port 7-7
Figure 7.1-10 CO Emissions for 1978-1993 Trucks by VSP BIN, Non-port 7-7
Figure 7.1-11 CO Emissions for 1994-1997 Trucks by VSP BIN, Port 7-8
Figure 7.1-12 CO Emissions for 1994-1997 Trucks by VSP BIN, Non-port 7-8
Figure 7.1-13 CO Emissions for 1998-2003 Trucks by VSP BIN, Port 7-9
Figure 7.1-14 CO Emissions for 1998-2003 Trucks by VSP BIN, Non-port 7-9
Figure 7.1-15 CO Emissions for 2004-2006 Trucks by VSP BIN, Port 7-10
Figure 7.1-16 CO Emissions for 2004-2006 Trucks by VSP BIN, Non-port 7-10
Figure 7.1-17 CO2 Emissions for 1978-1993 Trucks by VSP BIN, Port 7-11
Figure 7.1-18 CO2 Emissions for 1978-1993 Trucks by VSP BIN, Non-port 7-12
iii
-------
Figure 7.1-19 CO2 Emissions for 1994-1997 Trucks by VSP BIN, Port 7-12
Figure 7.1-20 CO2 Emissions for 1994-1997 Trucks by VSP BIN, Non-port 7-13
Figure 7.1-21 CO2 Emissions for 1998-2003 Trucks by VSP BIN, Port 7-13
Figure 7.1-22 CO2 Emissions for 1998-2003 Trucks by VSP BIN, Non-port 7-14
Figure 7.1-23 CO2 Emissions for 2004-2006 Trucks by VSP BIN, Port 7-14
Figure 7.1-24 CO2 Emissions for 2004-2006 Trucks by VSP BIN, Non-port 7-15
Figure7.1-25 NOx Emissions for 1978-1993 Trucks by VSP BIN, Port 7-15
Figure 7.1-26 NOx Emissions for 1978-1993 Trucks by VSP BIN, Non-port 7-16
Figure 7.1-27 NOx Emissions for 1994-1997 Trucks by VSP BIN, Port 7-16
Figure 7.1-28 NOx Emissions for 1994-1997 Trucks by VSP BIN, Non-port 7-17
Figure 7.1-29 NOx Emissions for 1998-2003 Trucks by VSP BIN, Port 7-17
Figure7.1-30 NOx Emissions for 1998-2003 Trucks by VSP BIN, Non-port 7-18
Figure 7.1-31 NOx Emissions for 2004-2006 Trucks by VSP BIN, Port 7-18
Figure 7.1-32 NOx Emissions for 2004-2006 Trucks by VSP BIN, Non-port 7-19
Figure 7.1-33 VSP-Based Activity And NOx Emissions for 0160699-A 7-20
Figure 7.1-34 VSP-Based Activity And NOx Emissions for 0160910-A 7-20
Figure 7.1-35 VSP-Based Activity And NOx Emissions for 0178515-B 7-21
Figure 7.1-36 VSP-Based Activity And NOx Emissions for 0180990-A 7-21
Figure 7.1-37 VSP-Based Activity And NOx Emissions for 0181096-A 7-22
Figure 7.1-38 VSP-Based Activity And NOx Emissions for 0181157-A 7-22
Figure 7.1-39 VSP-Based Activity And NOx Emissions for 0182003-A 7-23
Figure 7.1-40 VSP-Based Activity And NOx Emissions for 0182022-A 7-23
Figure 7.1-41 VSP-Based Activity And NOx Emissions for 0182291-A 7-24
Figure 7.1-42 VSP-Based Activity And NOx Emissions for 0183558-A 7-24
Figure 7.1-43 VSP-Based Activity And NOx Emissions for 0183710-A 7-25
Figure 7.1-44 VSP-Based Activity And NOx Emissions for 0183713-A 7-25
Figure 7.1-45 VSP-Based Activity And NOx Emissions for 0183716-A 7-26
Figure 7.1-46 VSP-Based Activity And NOx Emissions for 0183718-A 7-26
Figure 7.1-47 VSP-Based Activity And NOx Emissions for 0184781-A 7-27
Figure 7.1-48 VSP-Based Activity And NOx Emissions for 0185067-A 7-27
Figure 7.1-49 VSP-Based Activity And NOx Emissions for 0185576-A 7-28
Figure 7.1-50 VSP-Based Activity And NOx Emissions for 0185728-A 7-28
Figure 7.1-51 VSP-Based Activity And NOx Emissions for 0185774-B 7-29
Figure 7.1-52 VSP-Based Activity And NOx Emissions for 0186029-A 7-29
Figure 7.1-53 VSP-Based Activity And NOx Emissions for 0188529-B 7-30
Figure 7.1-54 VSP-Based Activity And NOx Emissions for 0189106-A 7-30
Figure 7.1-55 VSP-Based Activity And NOx Emissions for 0190306-B 7-31
Figure 7.1-56 VSP-Based Activity And NOx Emissions for 0190786-A 7-31
Figure 7.1-57 VSP-Based Activity And NOx Emissions for 0190992-A 7-32
Figure 7.1-58 VSP-Based Activity And NOx Emissions for 0218126-A 7-32
Figure 7.1-59 VSP-Based Activity And NOx Emissions for 0267938-A 7-33
Figure 7.1-60 VSP-Based Activity And NOx Emissions for 1095729-B 7-33
Figure 7.1-61 VSP-Based Activity And NOx Emissions for 1095730-A 7-34
Figure 7.1-62 VSP-Based Activity And NOx Emissions for 1144157-A 7-34
Figure 7.1-63 VSP-Based Activity And NOx Emissions for 1157629-A 7-35
Figure 7.1-64 VSP-Based Activity And NOx Emissions for 1190901-A 7-35
iv
-------
Figure 7.1-65 VSP-Based Activity And NOx Emissions for 1191018-A 7-36
Figure 7.1-66 VSP-Based Activity And NOx Emissions for 1191083-A 7-36
Figure 7.1-67 PEMS vs. RSD HC Emissions Comparison, by VSP Bin 7-37
Figure 7.1-68 PEMS vs. RSD CO Emissions Comparison, by VSP Bin 7-38
Figure 7.1-69 PEMS vs. RSD NOx Emissions Comparison, by VSP Bin 7-38
Figure 7.1-70 NOx Emissions (g/s) for 0178515-B During aDay's Route 7-40
Figure 7.2-1 Drayage Truck Activity by Hour of Day 7-51
Figure 7.2-2 Drayage Truck Activity by Day of Week 7-52
Figure 7.2-3 Drayage Truck Speeds for In-Port Activity 7-53
Figure 7.2-4 Drayage Truck Speeds for Non-Port Activity 7-54
Figure 7.2-5 Drayage Truck Percent Engine Load for In-Port Activity 7-55
Figure 7.2-6 Drayage Truck Percent Engine Load for Non-Port Activity 7-56
Figure 7.2-7 Drayage Truck Operation by MOVES Operating Mode Bin, In-Port 7-57
Figure 7.2-8 Drayage Truck Operation by MOVES Operating Mode Bin, Non-Port 7-58
Figure 8.2-1 Exhaust System Repairs Performed During Study 8-4
Figure 8.2-2 Side Stack Trucks Not Tested During Drayage Study 8-5
Figure 9.1-1 Entity-Relationship Diagram for MySQL Database 9-2
v
-------
List of Tables
Table 3.2-1 Count of Sample Vehicles in Each Model Year Bin 3-7
Table 3.2-2 Allocation of the 1,877 Sampled Vehicles Among 25 Strata 3-30
Table 3.2-3 Distribution of 32 Proportionally Selected Vehicles Among 25 Strata. 3-30
Table 5.3-1 PEMS Testing by NOx Bin and Model Year Group 5-31
Table 5.3-2 PEMS and PAMS Test Counts, By Truck 5-31
Table 5.3-3 PAMS Tests by Model Year Group 5-33
Table 6.1-1 MOVES HD Operating Mode Definitions (update) 6-12
Table 7.1-1 Dynamic and Field Blank Measurement Results 7-2
Table 7.1-2 Drayage Average Work-Based Emissions, In-Port 7-41
Table 7.1-3 Drayage Average Work-Based Emissions, Non-Port 7-43
Table 7.1-4 Drayage Average Distance-Based Emissions, In-Port 7-45
Table 7.1-5 Drayage Average Distance-Based Emissions, Non-Port 7-47
Table 7.1-6 Drayage Idle PM Emission Rates 7-49
Table 7.2-1 Activity Measurement Result Summary, In-Port 7-59
Table 7.2-2 Activity Measurement Result Summary, Non-Port 7-60
vi
-------
Acronyms
A/C
Alternating Current
ASD
Assessment and Standards Division
BSFC
Brake-Specific Fuel Consumption
CO
Carbon Monoxide Emissions
CO2
Carbon Dioxide Emissions
CSV
Comma-Separated Variable file
DLC
Diagnostic Link Connector
EAM
Emissions and Activity Measurements
ECU
Electronic Control Unit
EFM
Sensors' Exhaust Flow Meter
EPA
United States Environmental Protection Agency
ERG
Eastern Research Group
FEAT
Fuel Efficiency Automobile Test
FID
Flame Ionization Detector
g
grams
GPS
Global Positioning System
H-GAC
The Houston-Galveston Area Council
HC
Hydrocarbon
HP
Horsepower
IOO
Independent Owner / Operator
hr
hour
kW
Kilowatt
LOD
Laboratory Operations Division
MPS
Micro-Proportional Sampling System
MSOD
Mobile Source Observation Database
MOVES
Motor Vehicle Emission Simulator Model
NDIR
Non-dispersive infrared
NH3
Ammonia
NOx
Oxides of Nitrogen
OTAQ
Office of Transportation and Air Quality
OTR
Over-the-road (long-haul truck)
PAMS
Portable Activity Measurement System (aka datalogger)
PEMS
Portable Emissions Measurement System
PHA
The Port of Houston Authority
PM
Particulate Matter
QA
Quality Assurance
QAPP
Quality Assurance Project Plan
QC
Quality Control
QMP
Quality Management Plan
RSD
Remote Sensing Device
RPM
Revolutions Per Minute
SAE
Society of Automotive Engineers
SAS
Statistical Analysis Software
SCFM
Standard Cubic Feet per Minute
vii
-------
SIP
State Implementation Plan
so2
Sulfur Dioxide
SOP
Standard Operating Procedure
TCEQ
The Texas Commission on Environmental Quality
THC
Total Hydrocarbons
TxDOT
Texas Department of Transportation
VSP
Vehicle Specific Power
viii
-------
Acknowledgments
The Port of Houston Drayage Study was a multi-year effort resulting from collaboration
between the U.S. Environmental Protection Agency (EPA), the Texas Commission on
Environmental Quality (TCEQ), the Houston-Galveston Area Council (H-GAC), and the Port of
Houston Authority (PHA). Many people were responsible for the success of this project. ERG
would like to recognize the commitment and contributions from Shelley Whitworth and Patricia
Franco Lawhorn of the H-GAC, Mary Mcgarry-Barber of the TCEQ, and Ken Gathright of the
PHA. From EPA's ASD division, Carl Fulper, Robert Giannelli, Connie Hart, Cheryl Caffrey,
David Hawkins, John Koupal, Michael Christianson, Prashanth Gururaja and Jim Warila all
provided significant contributions, both in project design and analysis and also with fieldwork
deployment. Carl Fulper's dedication and overall direction was instrumental, as was Bob
Giannelli's support in diagnosing, maintaining and repairing the complex equipment used in the
study. A number of EPA LOD personnel also provided significant contributions, including
Robert Caldwell, Charles Schenk, Ethan Schauer, Dale Van Earp, Craig Swan and Brian Ratkos.
The efficiency and expertise of the LOD team was evident and critical, in particular with the
planning, complex fabrication, cutting, welding, equipment setup and other work necessary for
each unique installation. ERG's subcontractors on this project, Sensors, Inc. and the University
of Denver, also provided mission-critical support during the study. In particular, Louis Moret
and Chris Darby of Sensors both proved (once again) invaluable during the field deployment,
and Chris Darby's detailed analysis of PEMS data after fieldwork was complete helped ensure
the highest quality data was available from this study. Gary Bishop, Don Stedman and Brent
Schuchmann from the University of Denver demonstrated their expertise as they efficiently
conducted the RSD sampling effort which preceded the PEMS deployment. Recruiting support
by Patricia Lawhorn, Shelley Whitworth, Ken Gathright, and Diane Preusse (ERG) provided an
abundance of trucking companies to support the study, and field support from ERG staff
including Alan Stanard, Scott Fincher, Gopi Manne, Sandeep Kishan and Mike Sabisch all
helped keep things running smoothly on the ground and during subsequent analysis, validation
and reporting.
IX
-------
Executive Summary
This work was conducted by EPA, ERG and Sensors, in coordination with the Texas
Commission on Environmental Quality (TCEQ), the Houston-Galveston Area Council (H-GAC),
and the Port of Houston Authority (PHA). The objective of this work assignment was to collect
emissions and activity data on drayage trucks for incorporation of this source category into the
EPA Motor Vehicle Emission Simulator (MOVES) emission model used for state
implementation plan modeling,
University of Denver personnel performed one of the initial steps in this study by using a
remote sensing device (RSD) to make NOx emission measurements of drayage trucks at the Port
of Houston. This part of the study was conducted for EPA in order to both begin characterizing
drayage truck emissions and to provide a group of trucks to sample from for further exhaust
testing. Staff recorded these RSD measurements along with each truck's model year so that ERG
could then develop a stratified sampling plan for selecting drayage trucks to be tested. Trucks
were selected from a sample pool of 1877 trucks which had received an RSD test, and these
trucks (and their associated companies) were identified using port terminal gate data merged
with TxDOT registration data. Both fleet and independently owned and operated trucks were
recruited to participate in the program, and company and driver incentives were offered to
encourage participation.
Twelve companies that operate trucks that routinely service terminals in the Port of
Houston ultimately contacted ERG to inquire about the program, and of those, six companies
participated in the program. Portable emission measurement system (PEMS) and portable
activity measurement system (PAMS) testing was conducted in three phases between December
6, 2009 and March 17, 2010. Twenty-three PAMS instrumentations were performed during the
study and were approximately one week in duration and forty-six one-day PEMS tests were
performed on 37 trucks (some trucks were tested multiple times). PAMS tests consisted of
collecting 1 Hz date/time, truck speed, location, engine RPM, and engine datastream data (when
available). Two types of PEMS tests were performed, gaseous-only tests (of which 24 were
performed) and gaseous / PM tests (of which 22 were performed). Gaseous testing involved 1
Hz collection of THC, CO, CO2 and NOx emissions and exhaust mass flow rate, and gaseous /
PM testing involved collecting all gaseous and exhaust data along with micro-proportionally-
sampled PM mass emissions collected on gravimetric filters. Activity data and other system and
environmental parameters were also collected during the PEMS tests.
ES-1
-------
Once fieldwork was complete, study data was processed, analyzed, validated (problems
identified, flagged, and corrected as appropriate) and eventually processed for input into the EPA
Office of Transportation and Air Quality's (OTAQ's) Mobile Source Observation Database
(MSOD). Gaseous and PM mass-based emissions have been calculated on a fuel basis (grams
per gallon), time basis (grams per second), distance basis (grams/mile) and work basis (grams
per brake-horsepower hour). Work based emissions have been calculated using engine control
unit load data (and lug curves, as needed). Similarly, activity data has been further characterized
in terms of brake-specific load. Emissions and activity results have been calculated and are
presented both within and outside of port terminals, and have been further broken down on a
vehicle-specific-power (VSP) basis as is used in MOVES modeling. These results are available
in Section 7 of the report. As can be seen in Section 7, higher idle operation was seen within
port boundaries than outside of port boundaries, both for PEMS and PAMS testing. Similarly,
average speeds were lower within ports than outside of ports. Both work-based and distance-
based emissions were higher within ports than outside of ports. Analysis of PAMS results
similarly showed significantly lower speeds and loads for in-port activity vs. non-port activity.
Hence, dominance in lower MOVES operating mode bins is seen for port activity when
compared to non-port activity. It should be noted that non-port activity only includes local
operations and does not include over-the-road (interstate) operations. Inclusion of interstate
operation in these results would likely increase the differences seen in MOVES operating mode
bins between port and non-port activities.
This study was a significant undertaking and involved a high level of planning, staffing
and resources. This report describes methods and procedures for preparing for and performing
the field study, and also processing and validating the data. A summary of the data collected is
provided, but additional work needs to be done to better understand the true emission values and
uncertainty limits of this data and to further characterize port emission and activity profiles using
the data collected in this study. In addition, future studies to better understand emissions from
other port activities such as emissions from marine vessels, emissions from cranes and gantries,
emissions from dedicated port vehicles (aka yard mules), emissions from railcars and other
miscellaneous port emission sources are necessary in order to understand the overall emissions
contribution from all port activities.
ES-2
-------
1.0 Introduction
The objective of this work assignment was to collect data on drayage trucks for
incorporation of this source category into the EPA Motor Vehicle Emission Simulator (MOVES)
emission model used for state implementation plan modeling. Drayage trucks are heavy-duty
trucks that are used to transport containers, bulk and break-bulk goods to and from port terminals
and intermodal rail yards to other locations. This work only involved diesel-powered trucks,
which is the fuel type powering almost all drayage trucks in the study area. This work was a
continuation of Work Assignment 2-7 of Contract EP-C-06-080, during which remote sensing
device (RSD) testing was performed to gather drayage truck emissions data for use in developing
sampling criteria for this study. After development of the truck sampling criteria, recruiting and
equipment preparation took place and portable emission measurement systems (PEMS) and
portable activity measurement systems (PAMS) were then used to gather emissions and activity
data from trucks selected for participation in this study.
This work was conducted by ERG and EPA in coordination with the Texas Commission
on Environmental Quality (TCEQ), the Houston-Galveston Area Council (H-GAC), and the Port
of Houston Authority (PHA). TCEQ and H-GAC developed a partnership with the EPA to work
together on improving mobile source emission inventory estimates used in the SIP process.
Emphasis was placed on the Houston area, with particular focus on drayage trucks due to their
possibly significant emission contribution resulting from port activities.
The emissions measurements that were performed during this study were 40 CFR 1065-
compliant onboard PEMS measurements of gaseous and aggregate particulate matter (PM)
exhaust emissions. Each test was conducted over a typical workday. For the activity
measurements, Isaac dataloggers were used to collect date, time, location (GPS), engine speed
and ECU datastream data (for SAE J1708 and J1939-controlled engines) over a period of
approximately one week.
Once fieldwork was complete, study data was processed, analyzed, quality assured
(problems identified, flagged, and corrected as appropriate) and eventually processed for input
into the EPA Office of Transportation and Air Quality's (OTAQ's) Mobile Source Observation
Database (MSOD).
Information regarding the sample design, study preparation and execution and data
processing and QC are presented in this report, along with complete study results. Analysis,
interpretation of the results and implications are not presented here but will be the focus of future
work.
1-1
-------
2.0 Background
Drayage trucks operate in large numbers, with activity focused at a limited
number of locations in an urban area, such as port terminals and intermodal facilities.
These trucks are heavy duty trucks that are usually diesel-powered and are used to
transport containers, bulk, and break-bulk goods to and from ports and intermodal rail
yards to other locations.
These trucks conduct the majority of their travel on short-haul runs, repeatedly
moving containers across fixed urban routes. Drayage truck activity in the Houston area
is largely focused at the Port of Houston, particularly at the Barbours Cut Container
Terminal, but also with activity at the Bayport Container Terminal, the Jacintoport
General Cargo Terminal and Greens Port Industrial Park Terminal. A map of the
locations of these four port terminals is shown in Figure 2.0-1.
Figure 2.0-1 Terminals of the Port of Houston Analyzed in this Study
hts
We/,
ark
Nontiside
Village
C3 Greater
Fifth Ward
Green
Houston
Denver Harbor/
Port Houston
Northshore
Cloverleaf
Channelview
Greater
Third Ward
OST/ South
Galena Park
SLoo9
Union
Pant F*v
Pasadena
' ¦ • ' A
Jacintoport
Baytown
Airport'
P sodeoa Fwy
Greens Port
Deer Park
Greenwood o jr
Place
Kingsdale
Houston
Barbours Cut
F
r
+
William P
n Edgebrook
\
South AcresJ
Crestmont Park
Hobby Airport
Mmnetex
Southbelf
Ellington
Brookside
&
Village
T
x La Porte
a Municipal
Airport
La Porte Bay Ridge
Park
Oakhurst
Ellington
Airport
+
Armand Bayou —
Nature Center
Clear Lake
Shoreacres
&
+
Skyway
Manor Airport
Pearfand
BaioyAve
if
\
CwrtyRd10"> \
Fnftnriswonri
\ CD
Heritage Webster
Park
Seabro-ok
Clear Lake
Shores
Bayport
Th
Baybrook
Place
Baytown
Schilling @
& Massey
Vinson
^aAsstead
Map data ©2012 Google
Emissions from on-road drayage truck operations may be estimated using the
default short-haul combination truck source type within the MOVES model. However,
the activity and emissions occurring off-network within the port itself, which consists of
2-1
-------
significantly more idle and low speed operation than on road (non-port) activity is not
included in the MOVES model used in the SIP process.
To address this, EPA conducted this study to collect information required in order
to create a new source category within the MOVES model for drayage activities. Toward
this goal, EPA developed a partnership with TCEQ and H-GAC for collecting the
information needed to improve the inventory in the SIP process. The focus of this
partnership was placed on the Houston area because of the large emission inventory that
exists due to all of the drayage truck operations that take place in and around the port
terminals. In this study, references are made to "in-port" and "non-port" driving modes.
In-port operation refers to vehicle operation within the secured and gated boundaries of
the port terminals. All port terminals studied in this project were a part of the Port of
Houston.
For this study, ERG and EPA, in partnership with TCEQ and H-GAC, collected
information necessary to add drayage trucks as a source category in the MOVES model.
The PEMS and PAMS devices were used to gather emissions and activity data. A
secondary objective was for this study to serve as a template for local areas to expand
default MOVES source categories to better reflect local conditions. This report describes
the preparation for and execution of the data collection study and presents results from
and lessons learned during the study.
2
-------
3.0 Study Design
This section describes the key elements of the study design including the preliminary
work performed in preparation for conducting this study. A comprehensive workplan and
quality assurance project plan (QAPP) were developed outlining all steps in the study and a
remote-sensing study was conducted under a prior work assignment in order to gather emissions
information to be used in developing the truck sampling plan used in this study. Details of the
sampling plan which is based on the remote sensing data are presented in Section 3.2. Section
3.3 contains a description of the strategy used for recruiting candidate trucks specified in the
sampling plan.
3.1 Workplan and Quality Assurance Project Plan (QAPP) Development
Several iterations of the workplan and QAPP were developed during this study. SOPs,
included by reference as appendices to the QAPP, were also created for all facets of fieldwork
and were revised as needed throughout the study.
3.1.1 Development of the Workplan
With the goal of characterizing emissions and activity from drayage trucks operating in
and around the Port of Houston, EPA and the ERG team developed a workplan that outlined the
study design to collect emissions data on approximately 32 trucks and activity data on
approximately 24 trucks during the study. This workplan was revised as needed throughout the
study as work requirements changed.
3.1.2 Development of the QAPP
As part of the proposal effort for this contract, ERG tailored the corporate Quality
Management Plan (QMP) into a guidance document which provides corporate quality guidelines
for work under this contract. In addition, a QAPP was developed as part of this work assignment
to provide project-specific guidelines covering all facets of the study. This QAPP was drafted
prior to the commencement of project activities, and was revised during the project. As specified
in the work assignment, the QAPP was based on the following two guidance documents:
• Requirements for Quality Assurance Project Plans. EPA QA/R-5. EPA/240/B-01/003.
USEPA Office of Environmental Information. Washington, D.C. (Available at
http://www.epa.gov/quality/qs-docs/r5-final.pdf).
• Guidance for Quality Assurance Project Plans. EPA QA/G-5. EPA/240/R-02/009. USEPA
Office of Environmental Information. Washington, D.C. (Available at
http://www.epa.gov/quality/qs-docs/g5-final.pdf).
3-1
-------
The QAPP conforms to requirements specified in Task 2 of the performance work statement, and
describes the following measures:
• standard procedures for calibration of all portable measurement instruments
• standard schedules for regular calibration of portable measurement instruments, and the
maintenance of permanent and retrievable records of all calibrations
• procedures or decision rules for verifying proper operation of a portable measurement
system when reviewing records of calibrations, spans, or zeroes
• maintenance of operating logs for all portable measurement systems
• standard operating procedures for equipment used to perform calibrations
• standard operating procedures for portable measurement instruments (PEMS/PAMS)
• procedures for sampling and recruitment of respondents
• procedures for data transfer, entry and management
• procedures for regular transfer of all data generated within this project to the Work
Assignment Manager for review and audit
• procedures for the protection of respondent confidentiality and,
• data tracking and chain of custody procedures
A copy of the QAPP is provided in Appendix A of this report.
3.2 RSD Testing and Vehicle Sampling Plan Development
Remote sensing device (RSD) testing was performed by University of Denver personnel
operating with ERG under a previous work assignment (WA 2-7 of this contract) in order to
gather preliminary emissions information on the fleet which operates at the Barbours Cut
Terminal. This RSD emissions information was used in developing the PEMS sampling plan for
the study. Descriptions of RSD testing performed and the sampling plan which was developed
follow.
3.2.1 RSD Testing and Development of the PEMS Sampling Plan
Under Work Assignment 2-7 of this contract, RSD testing was conducted From July 21
through July 31, 2009 at the Port of Houston's Barbours Cut Terminal to measure emissions of
diesel drayage trucks as they entered the gate at the port. Gary Bishop from the University of
3-2
-------
Denver set up the Fuel Efficiency Automobile Test (FEAT) RSD instrument to take
measurements at the top of the exhaust stacks of each vehicle. Barriers were used to funnel
trucks through a narrow chute so that the RSD instrument would take measurements on one
vehicle at a time. Figure 3.2-1 shows the location of the RSD instrumentation just beyond the
entrance of the Barbours Cut Terminal. Approximately 4500 HC, CO, NOx1, SO2, NH3 and
opacity RSD measurements were collected for approximately 2000 trucks with Texas license
plates over the 2-week test period. The instantaneous speed and acceleration of each vehicle
were measured at the same time as the RSD beam block. As is customary, a video photo was
taken of each truck's license plate.
Figure 3.2-1 Location of RSD Testing Equipment at Port Entrance
The following RSD instruments were used to measure diesel exhaust from heavy duty
diesel drayage trucks as they entered the Barbours Cut terminal.
FEAT 3002 developed by University of Denver
1 In relation to RSD measurements described in this section, we make no distinction between NOx and NO. We refer
to these generically as NOx.
3-3
-------
- NDIR Component - Measures CO, C02, HC, and percent opacity
2 UV Spectrometers - Measure NO, N02, S02, and NH3
ESP 4600 by ESP, Inc
- NDIR Component — Measures CO, C02, HC,
Dispersive UV Spectrometer - Measures NO, and Smoke factor
The measurement system required the use of scaffolding to position the RSD instrument
and detector at a suitable height so that the system could "see" the drayage truck's exhaust plume
with its light detector. Figure 3.2-2 shows the RSD equipment immediately before a truck passes
through the beam.
Figure 3.2-2 RSD Test Equipment
Emissions data was collected over a 9 day period for as many vehicles that entered or
exited the port as possible in order to obtain an understanding of the gaseous and particulate
emissions for trucks operating in and at the port. License plate information for each truck, which
was photographically captured during the RSD test, was then transcribed and used to merge the
3-4
-------
RSD data with the Texas Department of Transportation's (TxDOT's) heavy-duty vehicle
registration data, which included information such as each vehicle's weight, model year,
manufacturer, etc. In addition, data from the Barbours Cut Terminal entry database was also
procured and merged (via license plate), providing information on each truck's trailer chassis and
container's ownership and weight as the truck was driven into the port.
Using this merged dataset, ERG developed a sampling strategy including sampling strata
and participation targets, based on the anticipation that there would be six weeks of PEMS and
four weeks of PAMS testing. The RSD emissions data within the dataset were grouped into
emission level "bins" for the measured vehicle population. These RSD emission level bins were
used as a stratification variable for selecting test vehicles, accounting for anticipated vehicle and
operator variability and possible participation bias. The database was also analyzed to identify
companies (also known as establishments in project terms) with the highest volume of traffic at
the port in order to maximize vehicle selection opportunities within the those companies. These
companies with high volumes of port traffic in many different emission level bins were known as
"targeted establishments".
3.2.2 PEMS Sampling Plan Development
As previously described, data from each RSD beam block (measurement episode) was
merged by transcribed license plate with Texas registration records, yielding the model year of
each truck that was registered in Texas. After matching RSD emissions measurements with
Texas registration data, there were successful registration matches for 4,032 of the RSD beam
blocks on 1,877 vehicles. Vehicles were selected from these 1,877 vehicles for PEMS
installations. Some vehicles had more than one RSD beam block. The distribution of the number
of beam blocks per vehicle is shown in Figure 3.2-3.
3-5
-------
Figure 3.2-3 Distribution of the Number of RSD Beam Blocks Per Vehicle in the Sample
2
3
4
5
(I)
6
o
7
o
CD
8
E
9
cc
CD
10
CD
11
Q
CO
12
cr
H—
13
o
1—
14
0
-Q
15
E
13
16
17
I O
19
20
21
22
FREQ.
CUM.
FREQ.
PCT.
CUM.
PCT.
869
869
46.30
46.30
470
1339
25.04
71.34
260
1599
13.85
85.19
132
1731
7.03
92.22
71
1802
3.78
96.00
40
1842
2.13
98.14
16
1858
0.85
98.99
5
1863
0.27
99.25
4
1867
0.21
99.47
4
1871
0.21
99.68
0
1871
0.00
99.68
1
1872
0.05
99.73
1
1873
0.05
99.79
1
1874
0.05
99.84
1
1875
0.05
99.89
1
1876
0.05
99.95
0
1876
0.00
99.95
0
1876
0.00
99.95
0
1876
0.00
99.95
0
1876
0.00
99.95
0
1876
0.00
99.95
1
1877
0.05
100.00
100
200
300
400 500
FREQUENCY
600
700
800
900
3-6
-------
3.2.2.1 Use of Model Years as a Stratification Variable
The distribution of model years of the RSD beam blocks for the 1,877 vehicles in the
sample is shown in Figure 3.2-4. EPA chose five groups to stratify the model years. The model
year strata and the counts of sample vehicles in each stratum are shown in Table 3.2-1.
Table 3.2-1 Count of Sample Vehicles in Each Model Year Bin
Model Year Bin
Vehicle Count
1978-1993
122
1994-1997
494
1998-2003
1065
2004-2006
131
2007-2010
65
3.2.2.2 Use of RSD Emissions as a Stratification Variable
The two most important emissions for diesel engines are NOx and particulate matter
(PM). The RSD instrument measures instantaneous (0.5s) NOx (as grams NOx per gram fuel)
and opacity for each beam block. Initial regressions of the effect of vehicle identity on opacity
showed no significant difference among the 1,877 vehicles. On the other hand, vehicle identity
had a strong influence on NOx. Therefore, since NOx is an important diesel emission and since
the initial regressions indicated strong differences among the vehicles, we chose to stratify the
sample set on RSD NOx emissions, as well as model year group.
Figure 3.2-5 shows the distribution of measured RSD NOx values for the dataset. Clearly,
the distribution is positively skewed. However, before dividing the NOx emissions into strata, it
needed to be determined if, in addition to vehicle identity, NOx was significantly influenced by
vehicle speed, vehicle acceleration, vehicle specific power (VSP), or vehicle weight. If NOx was
significantly influenced by one or more of these independent variables, then a correction would
be needed for those effects so that the NOx emissions of the vehicles in the sample would all be
on the same measurement condition basis. The intent of this correction would be to make sure
that the vehicles were only stratified based on their NOx emissions, not by the other factors such
as speed or VSP, which could possibly affect the measured NOx results and, therefore,
stratification.
3-7
-------
Figure 3.2-4 Distribution of Drayage Vehicle Model Years in the Sample
3-8
-------
Figure 3.2-5 Distribution of RSD Measured NOx Values in the Sample
3-9
-------
Ordinary least squares regression of the measured NOx values against vehicle identity
(using license plate as a categorical variable), speed, acceleration, VSP, and weight can be used
to determine which independent variables are important to NOx. However, for the regression
results to be reliable, the errors of the NOx measurements need to be approximately
homogeneous and approximately normal.2
The homogeneity of the NOx variance can be evaluated by looking at the variances (or
standard deviations) for individual vehicles in the sample that received more than a single beam
block. Figures 3.2-6 through 3.2-11 show plots of the standard deviation of the NOx readings vs.
the mean of the NOx readings for vehicles with 2, 3, 4, 5, 6, and 7 replicate RSD readings3,
respectively. All six plots use the same horizontal and vertical scales. Each point on a plot
represents the results for a single drayage vehicle in the dataset.
Figures 3.2-6 through 3.2-11 indicate that the standard deviation of the NOx readings
tends to be proportional to the mean NOx readings. This means that the NOx readings have an
inhomogeneous variance. For the purposes of performing regressions, we would like the NOx
variance to be more homogeneous, that is, more nearly constant at all levels of mean NOx
values.
Now, consider Figures 3.2-12 through 3.2-17, which are for the same observations as
Figures 3.2-6 through 3.2-11. For these new plots we have raised all NOx readings to the 0.2
power and then calculated the standard deviation and mean of the power-transformed values for
2 Ordinary least squares is based on the assumption that the error variance is homogeneous, that is, the same under
different conditions. Certain probability calculations discussed later are facilitated if, in addition, the response
variable (NOx or a transformation thereof) is approximately normal. These probability calculations pertain to the
use of the NOx values to identify strata to use as the basis of the stratified experimental design. Below, we discuss
the determination of a transformation that achieved these objectives (homogeneous variance, normality of the
transformed NOx values).
Another standard assumption of ordinary least squares is that the errors in the response variable be normally
distributed. Again, the transformed NOx value is the response variable here. This assumption affects hypothesis tests
to determine which of the variables listed above, vehicle identity, speed, acceleration, VSP, and weight, are
significantly related to NOx. Briefly, certain effects of using a very large sample size (4,032) mitigate the concern
about this normal assumption. While we do not want to digress into the full mathematical details, the effects of the
large sample size to which we refer are loosely related to the central limit theorem. Further, the hypothesis tests are
just a mechanism for identifying important variables and do not constitute the final results of the study. Thus, the
normality of the errors in the y-variable is not a critical issue that merits a lot of attention in this particular study.
3 Plots for more than 7 replicates are not shown because so few vehicles had more than 7 replicates that a trend is not
clear.
3-10
-------
each vehicle4. Again, all six plots use the same horizontal and vertical scales, and each point on a
plot represents the results for a single drayage vehicle in the dataset.
The plots in Figures 3.2-12 through 3.2-17 show that the standard deviations of the
transformed NOx values are relatively independent of the mean of the transformed NOx values.
The vertical scatter of the standard deviation in these plots is a consequence of random statistical
fluctuations in the standard deviation of the NOx readings. The vertical scatter in each plot can
be described by a transformation of the chi-square distribution.5
These results indicate that the 0.2-power transformation is appropriate for homogenizing
the variance of the RSD NOx values in this dataset. Accordingly, regressions to determine the
influences on NOx were performed on NOx0 2. The regression of NOx0 2 against just vehicle
identifier (as a categorical variable) had an r2 of 0.795 and a root mean square error6 of 0.1033.
For the regression of NOx0 2 against vehicle identifier plus speed, acceleration, VSP, and weight,
only speed had a significant regression coefficient. The regression of NOx0 2 against vehicle
identifier plus speed had an r2 of 0.830 and a root mean square error of 0.1002. Since the size of
the speed coefficient was quite small, since adding speed to the regression caused the residual
error to go down only a small amount (from 0.1033 to 0.1002), and since classifying a vehicle's
RSD emissions characteristic is easier using just RSD emissions information (and not also using
vehicle speed), we decided to drop speed from further consideration. In this case the influence of
the independent variable (speed) is not very important even though it was statistically significant.
4 We chose the 0.2-power transformation because it caused the variances to be closest to homogeneous across the
range of NOx values. Powers less than 0.2 were too strong, and powers greater than 0.2 were too weak.
Nevertheless, even with the 0.2-power transformation the variance is probably not exactly homogeneous.
5 While the full details are given in basic statistics books, the estimate of a standard deviation behaves like the
square root of a chi-square-distributed variable, after accounting for certain constant factors that do not change the
basic shape of the distribution.
6 The root mean square error is the standard deviation of the residuals left over from the regression.
3-11
-------
Figure 3.2-6 Trend of Standard Deviation of NOx with Mean of NOx for Vehicles with 2 RSD Replicates
+ T- '
+"%4ti"H-H~+
+ ++++ +
++
+
+
+
+ -
+ + +4i%+T_|+
+ + + 4-++^ ,T + ++
+, + . ++ +++ +!*+'+Hff T + + #
+i+ +
i "H" i. + ' "~J~
+ + .r^^' ~+
+ +
'+ f ^
^ ++ + +
+"4+"^+ +*
+ ++ +
„j_ + 4.
+ +
f 4.
+1-
+
10
20
30
40
50
60
70
80
90
100
110
120
Mean of NOx
3-12
-------
Figure 3.2-7 Trend of Standard Deviation of NOx with Mean of NOx for Vehicles with 3 RSD Replicates
120
Mean of NOx
3-13
-------
Figure 3.2-8 Trend of Standard Deviation of NOx with Mean of NOx for Vehicles with 4 RSD Replicates
120
Mean of NOx
3-14
-------
Figure 3.2-9 Trend of Standard Deviation of NOx with Mean of NOx for Vehicles with 5 RSD Replicates
0 10 20 30 40 50 60 70 80 90 100 110 120
Mean of NOx
3-15
-------
Figure 3.2-10
Trend of Standard Deviation of NOx with Mean of NOx for Vehicles with 6 RSD Replicates
120
Mean of NOx
3-16
-------
Figure 3.2-11
Trend of Standard Deviation of NOx with Mean of NOx for Vehicles with 7 RSD Replicates
120
Mean of NOx
3-17
-------
Figure 3.2-12 Trend of Standard Deviation of NOx°2 with Mean of NOx°2 for Vehicles with 2 Replicates
+
+
+
%
+
+ *
+
+++
+
+
+ + ++++^ + I
+ + ++ + +_+++! + +
j I ' -f-
+-h- +
+ ++ + ^
^ +
+
* J
+ ¦ ijtf+++4, ++
1 "t. — : +
+ + +
i Tj. •
+ .^4-
"bL4- "t"
^ - + + +'
+- -tfc ¦
-*r' +
+ ¦ - -1- + +' +' -t
+ + ±++#
*++ ++ + I * J+++-fc(=+ ++H+ +-
+++ ++ +I4f+ * + + 4J+
+ ++ ++ ++-|J + ,'$++*+ +
T+ -|j_ + + + 4iF!j-+ '+4f+T# +T
+ ++-H- -~ + 4,+ itf +1j++t+
! u V. + -f+' f -*++i* + + ++"i +
^ "Hf- | |-r- -if -f- ^
+ , -i-
• x+
. _j_ -f
++ + +
+
+
+
12
13
14
1.5
16
17
19
2.0
2.1
2.2
2.3
2,4
2.5
2.6
2.7
Mean of N0x^0.2
3-18
-------
Figure 3.2-13 Trend of Standard Deviation of NOx°2 with Mean of NOx°2 for Vehicles with 3 Replicates
++~
-ft-
"+L +
+-f~
-+- -i-
_t" "^"l+
+_!_ ^ -f- _j_
^ + _j_-r
-f- l rff-
++ , ++
~f" t~j"
+ +~^ I ~r "+ H"
++i-44- -f- + , . +
.J. > . -r 'i '
+- ' T jjf"
+t++i1f+ ++^+
i
-i~ +
+ 4-
+
+
1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1
Mean of N 0^0.2
2.2 2.3
2.4
2.5 2.6
2.7
3-19
-------
Figure 3.2-14 Trend of Standard Deviation of NOx°2 with Mean of NOx°2 for Vehicles with 4 Replicates
0.43
0.42
0.41 -
0.40
0.39 '
0.38
0.37
0.36
0.35
0.34
OJ 0.33'
0.32
^ 0.31
^ w 0.30
-O 0.29
CJ 0.28
Z 0.27 •
0.26'
O 0.25 -
0.24
o a23
0.22 "
co 0 21
¦ST 0.20'
© °'19
0.18
L-l 0.17'
"D 016
Jrt a15'
iS 0 14
^ 0.13
£ 0.12
™ 0.11
CO °-10
0.09
0.08
0.07
0.06 '
0.05 '
0.04 '
0.03
0.02
0.01
0.00
+ -f
+
+ +
4-
+
+ +++ ++# + +
+ + + +
¦#- ++" +4J" ^ ,
++ a++ *t + + ++++
1.2
1.3
1.4
1.5
1.6
1.7
1.9
2.0
2.1
2.2
2.3
2.4
2.5
2.6
2.7
Mean of N0x^0.2
3-20
-------
Figure 3.2-15 Trend of Standard Deviation of NOx°2 with Mean of NOx°2 for Vehicles with 5 Replicates
+
++# f +
1.2
1.3
1.4
1.5
1.7 1.8 1.9 2.0 2.1
Mean of NO/^
2.2
2.3
2.4
2.5
2.6
2.7
3-21
-------
Figure 3.2-16 Trend of Standard Deviation of NOx°2 with Mean of NOx°2 for Vehicles with 6 Replicates
0.43
0.42
0.41
0.40
0.39
0.38
0.37
0.36
0.35
0.34
0.33
0.32
0.31
0.30
0.29
0.28 -
0.27
0.26
0.25
0.24
0.23
0.22
0.21
0.20
0.19
0.18
0.17
0.16
0.15
0.14
0.13
0.12
0.11
0.10
0.09
0.08
0.07
0.06
0.05
0.04
0.03 -
0.02
0.01
0.00
+
1.2
1.3
1.4
1.5
1.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.5 2.6 2.7
Mean of NOX^0.2
3-22
-------
Figure 3.2-17 Trend of Standard Deviation of NOx°2 with Mean of NOx°2 for Vehicles with 7 Replicates
0.43 -
0.42 -
0.41 "
0.40 -
0.39 -
0.38 "
0.37 -
0.36 "
0.35 -
0.34 -
C\| 0.33"
O °-32"
^ 0.31 -
< 0,30"
>< 0.29 -
O 0.28
2 0-27 "
0.26 -
O 0.25 -
0.24-
J- 0.23 -
O 0.22 -
m 021"
-ir 0.20 -
> 0.19-
-------
The distribution of the 4,032 NOx0-2 values is shown in Figure 3.2-18. Comparison of
Figure 3.2-18 with Figure 3.2-5 shows that the 0.2-power transformation also converts the
skewed distribution of measured NOx values into a distribution that is more nearly normal.
Each vehicle in the dataset has a "true" NOx value that is characteristic of that vehicle.
This "true" value is the actual average NOx concentration that would typically be emitted by the
vehicle while in the operating mode that the RSD was set up to measure (ie. low speed
acceleration). Some vehicles have characteristically low "true" NOx values; other vehicles have
characteristically high "true" NOx values. The measured NOx values are not only a function of
the "true" value, but also have contributions due to each truck's day-to-day emissions variability
and the RSD unit's inherent measurement error. The intent of the stratification process was to
stratify the vehicles according to their "true" NOx values rather than their NOx measurements.
Another complicating factor is that some vehicles received only a single RSD measurement
while others received up to 22 RSD measurements, as shown in Figure 3.2-3. Because of the
differences in the number of RSDs received, a simple stratification by the mean RSD NOx would
cause individual vehicles to be stratified in an inconsistent manner. Stratifying vehicles by the
mean of the measurements would be reliable for the vehicles with many RSDs and unreliable for
vehicles with a single RSD. ERG needed to find a method that ccould stratify the vehicles
according to their "true" NOx value in a way that treated all vehicles similarly whether they
received a single RSD or many RSDs.
Fortunately, because the 0.2-power transformation homogenizes (approximately) the
variance of replicate RSD NOx measurements (see Figures 3.2-12 - 3.2-17) and normalizes
(approximately) the distribution of the RSD NOx measurements (see Figure 3.2-18), analysis of
the NOx data for the purposes of NOx stratification can overcome these two complications.
Each of the 4,032 measured NOx02 values shown in Figure 3.2-18 can be thought of in
terms of two sources of variation. The first source of variation characterizes the distribution of
the "true" NOx values of the 1,877 vehicles in the dataset about the mean "true" NOx value for
the dataset. The variance of this first source of variability is called varveh. The second source of
variation causes the RSD NOx measurements for an individual vehicle to scatter around the
"true" NOx value for a given vehicle. The variance of this second source of variability is called
varMeas. The standard deviation of 0.1033 as calculated above is used to calculate the varMeas
value of 0.0107 (=0.1033*0.1033). The contributions to this second source of variability are the
emissions variability of the vehicle and the variability of the RSD measurement. In order to
perform the stratification, it is not necessary to separate these two contributions to the second
source of variability.
3-24
-------
Figure 3.2-18 Distribution of RSD NOx0 2 Measured Values in the Sample
FREQ.
CUM.
FREQ.
PCT.
CUM.
PCT.
0
0
0.00
0.00
1
1
0.02
0.02
0
1
0.00
0.02
2
3
0.05
0.07
2
5
0.05
0.12
5
10
0.12
0.25
13
23
0.32
0.57
22
45
0.55
1.12
36
81
0.89
2,01
61
142
1.51
3.52
124
266
3.08
6.60
191
457
4.74
11.33
324
781
8.04
19.37
393
1174
9.75
29.12
472
1646
11.71
40.82
489
2135
12.13
52.95
491
2626
12.18
65.13
445
3071
11.04
76.17
337
3408
8.36
84.52
235
3643
5.83
90.35
150
3793
3.72
94.07
77
3870
1.91
95.98
58
3928
1.44
97.42
55
3983
1,36
98,78
28
4011
0.69
99.48
9
4020
0.22
99.70
8
4028
0.20
99.90
2
4030
0.05
99.95
0
4030
0.00
99.95
2
4032
0.05
100.00
0
4032
0.00
100.00
500
FREQUENCY
3-25
-------
Since the distribution of measured NOx0-2 values is influenced by the two sources of variation, its
variance varveh+Meas can be described in terms of varveh and varMeas:
varveh+Meas = varveh + varMeas Equation 1
The equation describes how the distribution of the "true" NOx0-2 values of the 1,877 vehicles is
broadened by the variability in the NOx measurements for each vehicle to produce the slightly
wider distribution of the measured NOx values for the dataset.
Thus, varveh+Meas can be estimated from the distribution of measured NOx0 2 values. But we
need to be sure that the set of data on which varveh+Meas is calculated has the same number of
RSDs for each of the 1,877 vehicles. Otherwise, the calculated value of varveh+Meas will be
incorrect. Again, some vehicles have one RSD and others have up to 22 RSDs as shown in
Figure 3.2-3. To include all of the 1,877 vehicles in the calculation, we used all of the RSDs for
vehicles that had one RSD and a randomly selected one RSD for each vehicle that had more than
one RSD. This subset of 1,877 RSD NOx0 2 values had a mean of 1.965, a standard deviation of
0.1644, and a variance of 0.0270.7
Since the value of varveh+Meas is 0.0270 and the value of varMeas is 0.0107, then using Equation 1
the value of varveh is 0.0164 by difference. This is the variance of the "true" NOx emissions of
the 1,877 vehicles in the dataset.8 If each of the 1,877 vehicles had a very large number of RSD
measurements, the distribution of the vehicle mean NOx0 2 values would have a standard
deviation of SQRT(0.0164). However, in our case of 4,032 measurements on the 1,877 vehicles,
almost all vehicles have one or a few measurements. Because of this, the distribution of the mean
vehicle values is broadened. Quantitatively, the standard deviation of the mean NOx0 2 value for
the distribution of mean vehicle NOx0 2 values each having N replicates for each vehicle can be
expressed in terms of varveh and varMeas:
Standard Deviation of the Mean NOx0 2 = SQRT [varveh + (varMeas /N)] Equation 2
where varveh = 0.0164
varMeas = 0.0107
N = the number of RSD NOx replicates for the vehicle.
7 Calculation in /proj l/EPA_Drayage/StratificationPlan/rsd_variability2.sas
8 As an aside, the 0.2-power transformation of NOx values which approximately normalizes the variances, the
determination of varveh of 0.0164, and the calculation of the sample mean NOx0 2 value of 1.965, allows us to
estimate the distribution of the "true" RSD NOx values for this set of 1,877 vehicles. Note that this distribution does
not contain the influences of the RSD measurement error and vehicle NOx emissions variability that are presentO in
the RSD measurements. The only exception to this statement is that the estimates of the parameters of the normal
distribution are not absolutely free of the influence of measurement errors or random variations in vehicle emissions;
because of the large sample size (1,877), this influence is small. The cumulative distribution is described by the
Excel function NORMDIST (NOxA0.2, 1.965, SQRT(0.0164), TRUE), where NOx is the RSD-measured NOx in
grams NOx per kilogram of fuel.
3-26
-------
Equation 2 indicates that, as the number of replicates (N) increases, the standard deviation of the
mean approaches SQRT(0.0164).
The mean NOx0-2 value and the standard deviation of the mean for a given vehicle can be used
along with the mean NOx0-2 value for the 1,877 vehicle sample (1.965) to calculate a z-value for
the vehicle:
Z = Vehicle's Mean NOx0,2 - Sample Mean NOx0,2 Equation 3
Standard Deviation of the Mean NOx0 2
The z-value places the vehicle in the distribution of NOx0 2 mean values for the 1,877 vehicles
whether the vehicle in question received one RSD or many RSDs. Equation 3 properly calculates
a z-value for each vehicle's mean NOx0 2 emissions while taking into account the number of RSD
replicates. Figure 3.2-19 shows the distribution of the z-values for the 1,877 vehicles in the
dataset.
We want to assign each of the 1,877 vehicles to one of a few bins that are related to the general
NOx level of the vehicle. Figure 3.2-19 shows that some vehicles tend to be low NOx emitters,
some tend to be high NOx emitters, and most tend to be medium NOx emitters. For the purposes
of characterizing the full range of NOx emitters we need to select vehicles across the range of
NOx emissions. We also want to stratify the number of vehicles selected for second-by-second
PEMS installations so that that data can be used to characterize low, medium, and high NOx
emitters equally well. To do this stratification, we have arbitrarily chosen to use five NOx strata
defined by the 2.5, 22.5, 77.5, and 97.5 percentile points in the z-value distribution. These points
occur at z-values of-1.992, -0.726, +0.651, and +2.148.9 We name the NOx bins -2, -1, 0, +1,
and +2. NOx Bin = 0 has the middle 55% of the z-values. NOx Bin = -1 has the vehicles with the
next lower 20 % of the z-values and NOx Bin = +1 has the vehicles with the next higher 20 % of
the z-values. NOx Bin = -2 has the vehicles with the lowest 2.5% of the z-values and NOx Bin =
+2 has the vehicles with the highest 2.5% of the z-values.
The effect of the number of RSD replicates on the NOx bin assignments is shown in Figure 3.2-
20. Each point in the plot represents one vehicle's number of RSD replicates and its mean NOx0 2
value. As the number of replicates decreases, the lines separating the NOx bins get farther apart
as the influence of varMeas increases according to Equations 2 and 3. If the effect of the number
of replicates had not been included in making bin assignments, many vehicles would have been
incorrectly binned.
9 The z-values given were determined empirically from the distribution of points and differ somewhat from the
theoretical, tabulated values (-1.96, -0.755, +0.755, and +1.96) for the z-distribution for the specified percentile
points.
3-27
-------
Figure 3.2-19
Distribution of Z-values for Vehicles in the Sample
FREQ.
CUM.
FREQ.
PCT.
CUM.
PCT.
-4.5
0
0
0.00
0.00
-4.3
|
1
1
0.05
0.05
-4.1
0
1
0.00
0.05
-3.9
0
1
0.00
0.05
-3.7
0
1
0.00
0.05
-3.5
|
1
2
0.05
0.11
-3.3
I
2
4
0.11
0.21
-3.1
I
2
6
0.11
0.32
-2.9
¦
3
9
0.16
0.48
-2.7
¦
5
14
0.27
0.75
-2.5
7
21
0.37
1.12
-2.3
6
27
0.32
1.44
-2.1
16
43
0.85
2.29
-1.9
23
66
1.23
3.52
-1.7
42
108
2.24
5.75
-1.5
54
162
2.88
8.63
-1.3
57
219
3.04
11.67
-1.1
72
291
3.84
15.50
-0.9
94
385
5.01
20.51
-0.7
130
515
6.93
27.44
CD
-0.5
147
662
7.83
35.27
-0.3
134
796
7.14
42.41
cc
>
I
-0.1
157
953
8.36
50.77
0.1
166
1119
8.84
59.62
0.3
151
1270
8.04
67.66
N
0.5
142
1412
7.57
75.23
0.7
125
1537
6.66
81.89
0.9
85
1622
4.53
86.41
1.1
71
1693
3.78
90.20
1.3
57
1750
3.04
93.23
1.5
26
1776
1.39
94.62
1.7
25
1801
1.33
95.95
1.9
18
1819
0.96
96.91
2.1
14
1833
0.75
97.66
2.3
19
1852
1.01
98.67
2.5
11
1863
0.59
99.25
2.7
6
1869
0.32
99.57
2.9
¦
3
1872
0.16
99.73
3.1
¦
3
1875
0.16
99.89
3.3
0
1875
0.00
99.89
3.5
0
1875
0.00
99.89
3.7
0
1875
0.00
99.89
3.9
0
1875
0.00
99.89
4.1
0
1875
0.00
99.89
4.3
¦
2
1877
0.11
100.00
4.5
0
1877
0.00
100.00
10
20
30
40
50
60
70
80
90
100 110 120 130 140 150 160 170
FREQUENCY
3-28
-------
Figure 3.2-20 NOx Bin Assignment of Vehicles with Differing Numbers of RSD Replicates
t
• m
• • •
• • •
• mm
' i !
1.5
s i
1.6
I
1.7
2.7
1.2
1.3
1.4
1.8
1.9
2.0
2.1
2.2
2.3
2.4
2.5 2.6
NOx Bin:
• • • 2
Mean of N0^0.2
• • • 4-1
+2
3-29
-------
3.2.2.3 Vehicle Allocation among 25 Strata Defined by Model Year Group and RSD NOx
Bin
The discussions above defined five model year group strata and five RSD NOx strata.
Together these make 25 overall strata. Allocation of the 1,877 vehicles in the dataset among
these 25 strata is shown in Table 3.2-2. The allocation among the 5 model year group strata is
shown by the totals in the right column. Allocation among the 5 NOx Bin strata is shown by the
totals in the bottom row. Both of these totals show that many vehicles are assigned to the center
strata and few vehicles assigned to the extreme strata.
Table 3.2-2 Allocation of the 1,877 Sampled Vehicles Among 25 Strata
NOx
Bin
-2
-1
0
1
2
1978-1993
8
23
69
20
2
122
1994-1997
1
34
259
175
25
494
1998-2003
11
234
636
168
16
1065
2004-2006
11
65
43
8
4
131
2007-2010
15
20
26
4
0
65
46
376
1033
375
47
1877
EPA anticipated that PEMS needed to be installed on about 32 of the 1,877 vehicles. If
32 vehicles were selected randomly or proportionally from the 1,877 vehicles, a distribution of
vehicles similar to that shown in Table 3.2-3 would result. The table shows that the selected
vehicles would also be distributed among the model year strata and the NOx bin strata with large
counts near the center and low counts near the extremes. Such an allocation does not provide a
distribution of selected vehicles evenly across both model year group and NOx bin strata that
would serve to well define the characteristics of vehicles in the extreme strata.
Table 3.2-3 Distribution of 32 Proportionally Selected Vehicles Among 25 Strata
NOx
Bin
-2
-1
0
1
2
1978-1993
0.1
0.4
1.2
0.3
0.0
2.1
1994-1997
0.0
0.6
4.4
3.0
0.4
8.4
1998-2003
0.2
4.0
10.8
2.9
0.3
18.2
2004-2006
0.2
1.1
0.7
0.1
0.1
2.2
2007-2010
0.3
0.3
0.4
0.1
0.0
1.1
0.8
6.4
17.6
6.4
0.8
32
3-30
-------
A more desirable allocation of 32 vehicles is shown in Table 3.2-4. In this allocation,
most of the 25 strata have at least one vehicle, and a few of the strata have more than one
vehicle. The total row and total columns show that, while center strata still contain more vehicles
than the extreme strata, vehicles are more evenly distributed among each of the 5 levels of model
year group and NOx bin strata. The result is that all of the model year group and NOx bin strata
have at least 4 vehicles in each stratum level. Analysis of second-by-second data from this
allocation of vehicles will be better able to quantify the effects of model year group and NOx bin
effects, as well as interactions between them.
Table 3.2-4 A Stratified Distribution of 32 Selected Vehicles Among 25 Strata
NOx
Bin
-2
-1
0
1
2
1978-1993
1
1
1
1
1
5
1994-1997
0
1
2
2
2
7
1998-2003
1
2
3
2
2
10
2004-2006
1
2
1
1
1
6
2007-2010
1
1
1
1
0
4
4
7
8
7
6
32
3.2.2.4 Vehicle Selection and Ranking Within Each of the 25 Strata
The number of vehicles shown in Table 3.2-4 in each of the 25 strata need to be selected
from each of the vehicles assigned to the 25 strata shown in Table 3.2-2. For example, Table 3.2-
4 shows that 3 participants are needed from the stratum defined by Model Year Group =1998-
2003 and NOx Bin = 0. Table 3.2-2 shows that that stratum contains 636 vehicles. Which of
those 636 vehicles should be chosen? This question needs to be answered for each stratum.
Clearly, practical considerations are important in choosing the vehicles to solicit for
PEMS installations. For example, the vehicle's owner must be willing to participate in the study
by allowing access to the vehicle. Beyond the practical considerations, two statistical issues are
important for the 32 vehicles in the instrumented vehicle set:
1) Probability that a vehicle is actually in the assigned NOx Bin - Proper NOx bin
assignment is important because we want to be reasonably certain that all 32 vehicles are in the
assigned bins so that we can be certain that all levels of NOx are covered by the 32 vehicles.
Because a vehicle's RSD measurements were used to assign a vehicle to a NOx bin, if a vehicle
was RSDed many times, it is more likely to actually be in the NOx bin to which it was assigned.
The probability that a vehicle's characteristic NOx measurement is actually in the assigned NOx
3-31
-------
bin depends on the mean NOx0-2 value, the number of RSDs that were used to calculate that
mean value, and the lower and upper NOx0 2 values that define the bin to which the vehicle was
assigned.10 The probabilities range from about 50% to 100%.11
2) Frequency that the vehicle will drive in the Port of Houston during the period of
instrumentation - After going through the expense of instrumenting a vehicle, we want to be
reasonably certain that the vehicle will drive in one of the port terminals so that the second-by-
second emissions and activity data that is collected will be representative of driving in the port.
We want to give higher selection priority to vehicles that were observed by RSD to drive many
times in the port. The logic is that vehicles that were seen many times in the port during the RSD
data collection phase will be more likely to be seen in the port during the vehicle instrumentation
phase of the study. We need to be aware, however, that the methods that trucking companies use
to assign vehicles to jobs may spoil the connection between past and future port activity levels of
individual vehicles. Also, because selection priority was given to vehicles that appeared to
service the port more frequently, care needs to be taken if the emissions results of this study are
to be applied to the heavy-duty truck fleet as a whole. It is possible that duty cycle differences
for drayage trucks could bias emissions results as compared to the entire on-road fleet.
To assist in soliciting trucking companies and vehicles for participation in the project, we
need a prioritized list of vehicles within each of the 25 strata. Companies that have vehicles at
the top of each of the 25 lists could be solicited first. Ideally, vehicles that have both a high
probability of being in the NOx bin and that had received a large number of RSDs would be
given high priority. Because a ranking cannot be made simultaneously on two variables, we
arbitrarily ranked vehicles within each stratum by the product of the fractional probability of
being in the assigned NOx bin and the square root of the number of RSDs received. While many
ranking variables could be devised, this ranking variable produced rankings that appeared to
have acceptable trade-offs between the probability of being in the assigned NOx bin and the
number of RSDs received.
Prioritized lists for each of the 25 strata are provided in Appendix B, RSD Stratification
Results. Vehicles were targeted for recruitment based on these prioritized rankings.
10 The probability that a vehicle is actually in the assigned bin is calculated in
/projl/EPA_Drayage/StratificationPlan/rsd_variability2.sas.
11 For example, if a vehicle' s mean value is very near a bin border, then the probability of being in the assigned bin
will be about 50% since the vehicle might actually be in the lower bin or the upper bin. If a vehicle received many
RSDs, then the probability of being in the assigned bin will be near 100% since the uncertainty in the mean value
will be low.
3-32
-------
3.2.3 PAMS Sampling Plan
The same pool of vehicles used for PEMS testing was intended for use in selecting
trucks for PAMS installations (each vehicle was to be one of the 1877 RSD-screened trucks).
However, since activity was not generally expected to be correlated to emissions, RSD emission
measurements were not used in determining PAMS selection bins.
In order to efficiently obtain a stratified random sample of activity measurements which
would yield a representative sample of fleet activity under our current sampling structure, the
same model year groupings used for PEMS selections were used for PAMS selections. In order
to roughly match the model year distribution of the 1877 RSD-screened vehicles, the team
attempted to recruit vehicles in such a way as to yield PAMS model year bin sizes which were
proportional to the percentage of trucks in each of the five model year bins shown in Table 3.2-1.
An effort was also made to conduct PEMS and PAMS testing on some of the same vehicles,
providing a "matched" set of emissions and activity data.
3.3 Develop Recruiting Plan
ERG worked with EPA, TCEQ, H-GAC and the Port of Houston Authority in developing
the strategy and materials for recruiting for this study. Informational brochures and "frequently
asked questions" sheets were developed by EPA, along with TCEQ and the H-GAC, which
provided basic information about the project to potential candidates. These brochures, which
were produced in English and Spanish, are shown in Appendix C.
The recruitment approach was tailored to maximize the yield from the recruited
establishments. Companies which operated at the port were selected for recruitment based on
their count of "high value" trucks from the list of 1877 trucks which had received RSD
measurements. Value in this context pertained to the probability that a truck was within its
assigned NOx bin and also the likelihood that the truck would be driven to Barbours Cut
Terminal over the instrumentation period, as described in Section 3.2.2.4. As shown in
Appendix B, each truck was assigned a bin ranking based on these probabilities. Companies
with the largest counts of candidate trucks were generally recruited by H-GAC, TCEQ and PHA
personnel through telephone calls and emailing of the Appendix C brochures and FAQs. During
the initial phone calls, a brief description of the study and incentive program was provided, and
interested participants were asked to contact ERG after review of the literature and consideration
3-33
-------
of the program. In order to increase the "callback" response rate, an additional incentive was
offered to the contact in each company who initiated a phone call to ERG.
The overall study's incentive program was as follows:
• $100 to each individual who responded to ERG as a result of initial contact by PHA, H-
GAC or TCEQ personnel
• $1000 (one-time incentive) to each company whose fleet trucks or independent
owner/operator trucks contracted to that company received one or more PEMS and/or
PAMS tests
• $500 for each truck that was successfully PEMS tested, paid to the independent
owner/operator (if independently owned) or the company that owned the fleet (if it was a
fleet truck)
• $250 for each truck that was successfully PAMS tested, paid to the independent
owner/operator (if independently owned) or the company that owned the fleet (if it was a
fleet truck)
3-34
-------
4.0 Study Preparation
4.1 Acquiring and Preparing PEMS and PAMS Equipment
EPA provided all equipment used in this study. Sections 4.1.1 and 4.1.2 describe
preparation of the PEMS and PAMS equipment prior to the study. Section 4.2 describes
mockups which were conducted prior to fieldwork in order to prepare the equipment and team
for PEMS and PAMS installations. After equipment preparations and installation and operation
mockups were completed, recruiting began, followed by preliminary onsite visits, as described in
Sections 4.3 and 4.4.
4.1.1 PEMS Preparation
For emissions measurements, the PEMS team used SEMTECH-DS PEMS manufactured
by Sensors, Inc. and provided by the EPA for use during this work assignment.
The SEMTECH-DS system collected the following information in one-second intervals, as
specified in the work assignment:
engine speed (revolutions per minute, rpm),
oxygen concentration in the exhaust stream ([O2], percent by weight, wt%),
carbon-dioxide concentration in the exhaust stream ([C02], percent by weight, wt%),
oxides of nitrogen concentration in the exhaust stream ([NOx], parts per million, ppm),
carbon monoxide concentration in the exhaust stream ([CO], percent by weight, wt%)
total hydrocarbon concentration in the exhaust stream, ([THC] parts per million, ppm)
aggregate particulate matter by gravimetric methods (g),
ambient temperature (EC),
exhaust temperature (EC),
exhaust mass flow rate (via the Sensors EFM)
relative humidity (%), and
barometric pressure (kilo-Pascals, kPa).
date/time stamp.
The following derived measurements were also provided for all emissions measurements:
exhaust flow volume (adjusted to standard temperature and pressure, cu. ft/min (scfm)),
fuel flow volume (g/sec, gal/sec),
carbon dioxide emission rate (g/sec, g/kg fuel),
pollutant emission rates for NOx, CO, THC, and PM, (g/sec, g/gal).
In addition to the above measured and derived values, information from each truck's
electronic network (either SAE J1708 or SAE J1939 compliant) was collected, when available.
The SEMTECH DS' default parameter lists for each protocol were used. Information was
collected on a second-by-second basis within the same datafile as the emissions measurements.
4-1
-------
Exhaust flow rate measurements were made using Sensors' five-inch diameter exhaust
flowmeters (EFMs). Although flowmeters of different diameters were available, all trucks tested
during the study had five inch diameter exhaust systems and sufficient exhaust flow volume for
use of five-inch flowmeters.
Gravimetric filter sampling was accomplished using Sensors' micro-proportional
sampling system (MPS) and their 3-chamber gravimetric filter sampler provided by EPA.
Gravimetric PM samples were collected on 47mm Teflon filters housed in the gravimetric filter
sampling unit which was heated to 1065 specifications. For gravimetric sampling, an air
compressor and flow control unit was used to provide filtered dilution air to the MPS, and a
rotary vane vacuum pump was used for drawing the filter sample. Compressed nitrogen (from a
compressed gas cylinder) was used for activating gravimetric filter solenoids. A SEMTECH-DS
adjacent to a 3-chamber gravimetric PM sampler is shown Figure 4.1-1.
4-2
-------
Figure 4.1-1 SEMTECH-DS (left) and 3-Chamber Gravimetric PM Sampler (right
A small air compressor and filtration unit was used to operate the MPS and to
automatically back-purge the EFM pressure lines at specified intervals. This air compressor,
mounted inside the MPS chamber, operated on A/C power provided by a Honda portable
generator. Ambient air scrubbed with carbon and particulate filters was supplied by the system
to allow zero-calibrations to be performed during periods of non-sampling.
Heated sample lines were used to transport exhaust sample from the exhaust flow meter
to the SEMTECH-DS, and also from the MPS chamber to the gravimetric filter sampler.
Rigid and flexible stainless steel tubing was custom fabricated for each installation to
allow transport of the complete exhaust flow from the truck's exhaust pipe to the EFM mounted
4-3
-------
on the top of the MPS rack. Tubing was mounted to the outlet of the muffler, in place of the
tailpipe stack, after the tailpipe stack had been removed. Joints were secured with welds, ring
clamps and band clamps, as appropriate. This entire system is shown installed on a custom rack
on the back of a city bus in Figure 4.1-2. Although components were mounted differently in the
drayage study (the SEMTECH-DS and the 3-chamber gravimetric sampler were generally
housed inside the cab in the drayage study), Figure 4.1-2 shows the layout of all components
relative to one another. In this figure, the SEMTECH-DS is mounted in an environmental
chamber, which was not necessary in the drayage study since the SEMTECH-DS was placed
inside each truck's cab.
For trucks for which only gaseous samples were taken (no PM measurements), the
exhaust flowmeter was typically mounted directly to the outlet of the exhaust muffler, and a
heated sample line was then used to transport the sample from the exhaust flowmeter to the
SEMTECH-DS, as shown in Figure 4.1-3. Also shown in this figure is the FID fuel cylinder
(mounted on the frame platform next to the generator) and the CL-size calibration and zero gas
cylinders (on the concrete) used for calibrating the SEMTECH-DS prior to (and after) each test.
Filter sampling could be controlled automatically by way of an integral timer, through
user-input conditions such as elapsed times at certain operating conditions, or manually through
laptop-interface control. For this study, all gravimetric filter switching was performed manually
by field personnel riding along with the driver and test equipment, and was primarily based on
operations (whether operations were within or outside of the port and idle vs. non-idle
operation).
EPA provided multiple backups for all critical components, including the SEMTECH-
DS, the 3-chamber gravimetric filter sampler, the MPS sampling system, exhaust flowmeters,
heated sample lines and generators. Raw materials and hardware required for multiple daily
installations were also maintained in overstock in EPA's onsite trailer. Spare parts for repairing
and maintaining all equipment were also maintained on-site (in EPA's trailer) throughout the
study.
Equipment was inventoried, acquired as needed and prepared for several months prior to
the start of this study. A verification of performance of gaseous measurement equipment,
exhaust flow measurement equipment and proportional sampling equipment with standards listed
in 40 CFR 1065 was performed by Sensors and EPA staff on all equipment to be used in the
study.
4-4
-------
Figure 4.1-2 Gaseous and PM System Used in Drayage Study
MPS
Chamber
FID fuel and solenoid
valve cylinders (N2)
(behind grav sampler)
Gaseous (gray) and PM (blue)
heated sample lines
MPS interface
box (blue box)
SEMTECH-DS
in environmental
box
Custom Fabricated
Exhaust Tubing
Exhaust Flowmeter
Gravimetric
Sampler
AC to DC Power converters
iWeather station
GPS
GPS Antenna
b
Generators
4-5
-------
Figure 4.1-3 Gaseous-only Flowmeter Mounting on Drayage Truck
Due to space constraints, equipment for this study would have to be mounted in various
locations in trucks, including inside the cab (SEMTECH, power supply and gravimetric sampler)
and outside of the cab (MPS chamber, exhaust flowmeter, pressurized cylinders and generator).
Consequently, equipment and materials beyond that which had been required for prior studies
were needed to allow installations on drayage trucks. The space constraints of drayage truck
installations also prevented the PEMS rack, which was used to house the MPS, gravimetric
sampler, power supply and SEMTECH-DS as a single unit during the recently completed
nonroad diesel equipment field study, from being used during this program.
EPA acquired and equipped a 24-foot Haulmark trailer for PEMS field operations
support. All materials and tools necessary for PEMS and PAMS testing and equipment repairs
were maintained in this trailer. Specially-designed racks and workbenches stored all equipment,
tools, materials and consumables and provided working space for field personnel. A specialized
rack with dedicated ventilation in the front upper deck of the trailer provided storage of various
compressed gas cylinders to be used in the study.
4-6
-------
EPA fabricated mounting hardware, including platforms for the SEMTECH-DS and
gravimetric sampler and retention cylinders for the nitrogen and FID-fuel gas bottles, prior to the
study. Electrical harnesses with appropriate 6-pin and 9-pin Deutsch connectors and associated
communication boxes for collection of SAE J1708 and SAE J1939 data from truck ECUs were
acquired. Transport tubing, elbows, adapters and clamps were acquired to accommodate all
expected exhaust configurations. Hand tools, power tools such as air wrenches, an onboard air
compressor, band saw, welder and other equipment were also acquired for use in the study and
stored in EPA's trailer.
All consumables and hardware necessary for conducting the study, such as gravimetric
filters and shipping materials, PEMS system filters, exhaust tubing and clamps, quick-release
power cables and other electrical harnesses and auxiliary sensors with their associated harnesses,
fuel sample bottles, calibration and audit gases and FID fuel and nitrogen gas cylinders and
associated high-pressure gas transport lines were also prepared as needed.
A preliminary inventory of PEMS and PAMS equipment which was used during study
preparation is provided in Appendix D.
PEMS SOPs, developed during the prior nonroad field study, were revised for the new
equipment hardware and software requirements for this work assignment. A copy of the PEMS
SOPs used in this study is provided in Appendix E.
4.1.2 PAMS Preparation
Prior to the start of fieldwork, ERG evaluated the PAMS available from the previous
nonroad study to determine which candidates would be suitable for use in this study. The six
Isaac V8 sealed dataloggers were selected for use in this study. An Isaac datalogger is shown in
Figure 4.1-4.
Isaac harnesses with 6-pin and 9-pin Deutsch connectors were acquired in order to allow
collection of SAE J1708 and SAE J1939 data from truck ECUs. Garmin GPS antennas were
purchased both from Isaac and from Garmin to allow incorporation of GPS in the PAMS data.
EPA technologists installed Isaac harness connectors on the GPS antennas purchased from
Garmin to allow them to be used with the Isaac dataloggers. Additional wiring harnesses for
operating the Isaac dataloggers from switched ignition power and optical RPM collection sensors
were also acquired, and brackets were fabricated to allow Isaac optical sensors to be used with
the Caterpillar magnetic-base RPM sensor mounts used in the prior nonroad field study.
4-7
-------
Figure 4.1-4 Isaac datalogger used Drayage study
The functionality of all six Isaac units was verified prior to field deployment for this
study. Repairs and software reconfigurations were performed as needed to ensure all units had
maximum memoiy capacity, 3-dimensional accelerometer data collection capabilities and
complete SAE J1708 and SAE J1939 ECU data collection capabilities. Configuration files were
created for the three anticipated types of installations (RPM collection via optical pickup, SAE
J1708 data collection and SAE J1939 data collection). Desired data parameters for both J1708
and J1939 communication protocols were defined in each of the configuration files to ensure
consistent and complete data collection for each PAMS installation. These units were configured
to record engine on and off status, associated date and time stamps, engine speed, location and
speed (via GPS) and ECU data on a 1-Hz basis. Datalogger electrical current usage rates in
active and standby mode under different installation configurations were measured in order to
4-8
-------
ensure the drayage truck batteries would not be drained after a period of extended inactivity with
the dataloggers installed.
PAMS SOPS were also developed based on the current equipment and installation
procedures. A copy of these SOPs is provided in Appendix F.
4.2 PEMS and PAMS Mockups in Ann Arbor
After preliminary PEMS and PAMS equipment inventory and preparation had taken
place, the ERG team assisted EPA in performing "mockup" testing of the PEMS and Isaac
PAMS equipment on Class 8 trucks similar to those to be tested during the study. Three mockup
sessions were performed at EPA's National Vehicle and Fuel Emissions Laboratory in Ann
Arbor, Michigan. These mockups served several purposes, including:
• Installation team members were able to increase their familiarity with the equipment
configuration and software operation for this study
• Team members practiced installations on vehicles similar to those being targeted for
this study, which allowed the team to determine the best procedures and locations for
installations of various equipment to help optimize efficiency and safety,
• Team members were given an opportunity to confirm the PEMS equipment was of a
size and configuration which would allow it to be installed in the desired locations on
targeted test vehicles, and make necessary hardware changes to reduce the chances of
installation problems
• The mockups allowed identification and acquisition or development of any equipment,
tools, materials, or procedures which would be needed for field activities
• The mockups allowed identification of previously unforeseen technical or logistical
challenges, with sufficient time for resolving these issues prior to field activities
• Team members were able to ensure all equipment was operating as expected and
ensure any systematic data quality issues were resolved, and
• The mockups allowed the need for all consumables necessary for field testing to be
identified in time for the consumables to be sourced and acquired
Sensors', Inc. personnel assisted with PEMS equipment support and provided training for
operation of the PEMS and PM sampling equipment to EPA and ERG staff during the mockups.
Many of the PEMS and PAMS preparation tasks described in Section 4.1 were performed during
the mockups.
4.3 Conduct Recruiting
Prior to the start of recruiting, ERG attended and presented an overview of the drayage
study at H-GAC's September 22, 2009 Federal Diesel Grant Program presentation held at the
Port of Houston. This provided ERG and H-GAC an opportunity to inform candidates about the
4-9
-------
upcoming study and to gauge interest and obtain feedback to be used to tailor recruiting for the
study.
Several weeks prior to the beginning of fieldwork, ERG worked with the EPA and
project partners to begin recruiting candidate drayage trucks. As described in Section 3.3, ERG
identified the "target" trucking companies with the highest number of "high value" trucks from
the sample of 1877 RSD-tested trucks. Preliminary contact information for these companies was
obtained through Internet research and these company lists (with the preliminary contact
information) was provided to PHA, H-GAC and TCEQ staff, who attempted to contact each
company, using the preliminary contact information provided by ERG or their own contact
information (i.e., acquaintances and the H-GAC Air Quality Database records). Once a
successful contact was made with each company, the PHA, H-GAC or TCEQ staff member
briefly explained the study and incentive program and offered to provide additional information
(the recruiting materials shown in Appendix C). The company contact was asked to consider the
study and contact ERG if they had interest in participating. Recruiting began several weeks prior
to the start of fieldwork and was performed throughout the majority of the study. Table 4.3-1
provides a summary of recruiting performed.
Table 4.3-1 Summary of Drayage Recruiting, by Companies
Recruiting Parameter
Count of
Companies
Total number of companies from pool of 1877 RSD'd trucks
178
Companies for which recruitment attempts were made (by PHA, H-GAC,
and TCEQ personnel)
86
Companies for which no recruitment attempts were made (companies with
lowest number of "high value" trucks in the RSD sample pool)
92
Companies who called ERG to inquire about study participation
12
Of companies who called ERG, those who agreed to testing, but didn't have
any eligible trucks available
2
Of companies who called ERG, those who agreed to testing, but were not
ultimately involved in the study because all their trucks were in full bins or
because the study concluded before they were needed
2
Of companies who called ERG, those who ultimately refused testing
2
Of companies who called ERG, those who received a preliminary "onsite
visit" to discuss study logistics and evaluate trucks for eligibility /
testability, and ultimately participated in PEMS and PAMS testing
6
As indicated above, 12 companies interested in participating (or even interested in only
receiving the "initial call incentive" as described in Section 3.3) contacted ERG personnel for
4-10
-------
additional study information. ERG staff described the study, answered any questions the
participant had, and attempted to schedule a visit to meet with the respondent, discuss testing
logistics and assess the company's eligible trucks for "testability". As can be seen in Table 4.3-
1, six companies received onsite visits, and field testing was ultimately performed at all six of
these companies. An additional four of the 12 companies expressed interest in participating in
the program, but did not ultimately participate either because their eligible trucks were
unavailable (out of service or over-the-road long-haul trucks), were in sampling bins that were
already full, or were just recruited too late in the study for testing (we completed fieldwork
without the need for testing these company's trucks). Only two companies who contacted ERG
ultimately refused to participate. All companies who agreed to an onsite visit did participate in
testing.
4.4 Preliminary Visits to Candidate Establishments
The preliminary onsite visits were performed several days to several weeks before testing
began at an establishment. ERG attempted to minimize the delay between the preliminary onsite
visit and the start of testing at each company, in order to prevent participation attrition. By this
point in the recruitment process, the company contact was generally familiar with the study and
incentives, but usually was interested in learning more about exactly what participation entailed.
ERG staff discussed PEMS and PAMS testing, including how installations were performed
around working schedules, where equipment would be placed and how it was installed and
secured, temporary modifications that would be made to the truck (and later reversed) and driver
/ company commitments. Also during this preliminary visit, logistics of the study were
discussed with the company contact in further detail, addressing issues such as identification of
the day-to-day contact for recruiting and testing, location to be used for installing test equipment
on trucks, operation and installation schedules, a suitable location for the onsite PEMS trailer,
availability of shore power (a/c power distributed from the facility where installations were
taking place) for onsite operations, off-hours facility access, method of incentive disbursements
(to the company or individually to each driver), insurance requirements for ride-alongs, tax
issues associated with the incentives, liability issues, how best to recruit independent drivers and
other issues.
After describing the technical and logistic aspects of the study and answering any
questions the contact had, ERG personnel attempted to survey as many of the eligible trucks
(from the pool of 1877 trucks which had been RSD screened) as possible. This typically
required personnel to wait onsite until the trucks arrived to pick up or drop off a load or until
they arrived to stop for the night. Since many of the trucks were operated by independent
4-11
-------
owners / operators (IOOs), ERG personnel would typically attempt to recruit these drivers at this
time. The onsite contact typically assisted with this recruiting to assure the driver that the study
(and incentives) was legitimate. ERG found that although some drivers were reluctant at first,
ultimately no drivers refused the study, once their questions were answered and their concerns
were addressed.
As many eligible trucks were surveyed as possible during this time. The primary goal of
this survey was to assess each truck for testability. Some issues evaluated were the condition of
the truck, the number and location of the exhaust stacks on the truck, whether the truck had
visible exhaust leaks, how much room was available for equipment outside and inside the cab,
availability of access points for routing sample and gas hoses and electrical lines into the cab,
ECU connector type, etc. A copy of the complete evaluation form is provided in Appendix G,
and PDF copies of all completed forms are provided in Appendix Y. Multiple photographs were
also taken of each truck to help PEMS and PAMS installation team members prepare equipment
for each individual installation.
4-12
-------
5.0 PEMS and PAMS Testing
PEMS and PAMS installations were performed at trucking companies operating at the
Port of Houston. These companies were located in the cities of Houston, Texas and La Porte,
Texas. The fieldwork for this study was conducted in three phases, over the period of December
8, 2009 through March 17, 2010. Four waves of PAMS installations, monitoring, and removals
were performed over this time, concurrent with three waves of PEMS testing. The overall
fieldwork schedule is shown in Figure 5.0-1.
Figure 5.0-1 Drayage Fieldwork Schedule, Dec 2009 - March 2010
Date Range
1st PAMS install Wave:
Dec 8 through 10
2nd PAMS install Wave:
Jan 20 through 22
3rd PAMS install Wave:
Feb2&3
4th PAMS install Wave:
March 5 & 6
The work assignment for this study specified PEMS testing be conducted over a six-week
period, and PAMS testing over a four-week period. Approximately two to five PEMS tests per
week were anticipated, for approximately 32 total PEMS tests over the six-week period.
Approximately 6 PAMS installations were also anticipated per week over the four week period,
based on an estimate of approximately 2 PAMS installs per day, installations being performed
over the weekend as much as possible. This would yield roughly 24 week-long PAMS
installations.
As described in Section 5.1, the PEMS test team performed both PM and gaseous tests as
well as gaseous-only tests. A total of 46 tests were conducted, 22 of these were full PM and
gaseous tests, and 24 were gaseous-only tests. A total of 23 week-long activity measurements
were also collected, as summarized in Section 5.3.
5.1 PEMS Testing
5.1.1 Process Overview
As described in Section 4.4, ERG staff performed preliminary site visits to all companies
prior to the arrival of the field testing team. Tentative plans were developed for PEMS and
5-1
-------
PAMS testing at each company, and candidate trucks were evaluated for testability. ERG
personnel then coordinated onsite testing once the field team was ready to deploy at each
company's site.
PEMS installation teams consisted of an ERG PEMS/PAMS field manager, two EPA
technologists from EPA's Laboratory Operations Division (LOD), two (or more) EPA staff
members from EPA's Assessment and Standards Division (ASD) and one technician from
Sensors, Inc. Additional staff, such as ERG PAMS installation team members or additional ASD
staff members, also periodically assisted with PEMS support operations, as needed.
The ERG field manager served as a liaison between the participating establishment and
the installation team (as well as the PAMS installation team), performed scheduling and
coordination of deployment, provided assistance with PEMS installations, setup and testing,
assisted with collection of truck and instrumentation data on instrumentation data collection
forms, assisted with PAMS installations and removals and reviewed and posted PEMS data onto
the project-secure FTP site. The field manager also tracked incoming and outgoing PM filters
and coordinated PM filter shipments between the field and the EPA laboratory. The EPA LOD
technologists and ASD staff members performed PEMS installation and operation duties,
including equipment installation, setup and gaseous and PM testing. Sensors technicians assisted
with PEMS installation, setup and operation, assisted with resolving in-use testing issues,
performed in-field review of the PEMS data and performed service and maintenance of the
PEMS gaseous and PM sampling equipment, along with EPA ASD personnel. EPA ASD
personnel also performed on-going testing support, in which one person would "ride-along" with
each truck during testing.
As described in Section 3.2.2, trucks were selected for PEMS testing from the pool of
1877 trucks which had previously received one or more successful RSD measurements, and
truck selection was stratified using truck model year and RSD NOx emission categories. Table
3.2-4 lists the distribution of the target sample of trucks within the 25 sample strata. Following
this weighting scheme, prioritized lists of trucks to be sampled were created for each trucking
company in an effort to fill all 23 strata (two bins had zero recruitment goals). In order to
maximize testing across as many bins as possible, priority was given to "uncommon" bins (bins
with few trucks) with consideration of eligible trucks at companies where testing was scheduled
but had not yet taken place. Also, as described in Section 3.2.2.4, vehicle selection within each
strata was based on the probability that the vehicle's true NOx emissions were within the NOx
emissions strata and also the number of times RSD readings were successfully captured on each
5-2
-------
truck. The "probability within a bin" and "RSD count" criteria were used to create the
prioritized stratum rankings shown in Appendix B.
Prioritized lists of desired trucks were provided to each company contact, and
information was collected regarding availability of each truck. Trucks could be excluded from
testing for several reasons, as listed below:
• The truck's usage shift did not allow the team adequate time to install and prepare the
PEMS equipment (i.e., the truck would leave too early in the morning for the team to
complete the installation, warm-up and calibration of the equipment)
• The truck was not scheduled to operate within a port
• The truck was not testable (dual exhaust, side stacks or insufficient available space
behind or in the cab prevented PEMS equipment installation from being possible)
• The truck was out of service or no longer worked at the company
• The truck was domiciled elsewhere and could not be left at the installation site overnight
It should be noted that no drivers refused to participate in the program, suggesting the
PEMS incentive was sufficient, and in fact a smaller incentive could suffice in future studies
with a similar participation commitment.
After equipment selection for PEMS testing, the ERG onsite installation manager
scheduled instrumentations with the appropriate establishment contact(s), sought permission to
instrument the specific pieces of equipment, and asked the driver about each truck's expected
operation during the measurement period (likelihood of port operations, local driving, over the
road, etc.). If the driver or dispatch contact indicated the truck was not to be used for port
operations or would not be available for testing during the anticipated measurement period, an
alternate truck was selected for instrumentation from the list of candidate trucks. Dispatch
personnel were generally accommodating and willing to schedule drivers for port work during
days when PEMS testing was to occur.
PEMS installations began with thorough collection of truck-specific data, which was
hand written onto "Instrumentation Data Collection Forms" as shown in Appendix H. Digital
photographs were taken of plates, VINS, and equipment serial numbers and specification tags
whenever possible, to help clarify or correct any ambiguous or inaccurate information which
may have been recorded during data collection. Photographs were also taken of engine tags,
when possible, in order to help confirm power and torque ratings, engine family and serial
numbers and other identifying information. Vehicle odometer readings were also recorded,
though neither odometer or engine family data was used as a part of any analysis discussed in
this report. A compilation of all information collected for all PEMS and PAMS instrumentations
5-3
-------
throughout the study is provided in Appendix I, with PDFs of the original hardcopy forms
provided in Appendix X. Emissions and instrumentation data is also being provided as part of
the MSOD data submission for this project. In Appendix I (and elsewhere), the primary key
field for each test is the test ID, which provides a unique identifier for each PEMS or PAMS test
conducted. The naming convention is as follows: PEMS; plate date E , PAMS; plate A. For
example, "ERG983_20100202_E" would represent a PEMS test performed on February 2, 2010
on a truck with license plate "ERG983". "ERG983 A" would represent a PAMS installation on
a truck with license plate "ERG983". No dates were assigned for PAMS instrumentations since
PAMS testing spanned multiple days and only one PAMS file (containing all test data) is
provided for each instrumentation.
Drayage trucks tested during the study are expected to have been operated on Texas Low
Emission Diesel (TxLED), in accordance with the Texas Commission on Environmental
Quality's TxLED program(3). Diesel fuel samples were collected for each PEMS-tested truck.
An electric automotive fuel pump along with a small rechargeable 12V battery was used to
collect fuel samples, which were labeled with sample ID tags and stored in fuel sample jars
provided by the EPA. All fuel samples were collected and handled according to guidelines listed
in Appendix N, Drayage Fuel Sampling SOPs, and stored and transported to the EPA laboratory
in Ann Arbor in the EPA PEMS trailer. After fieldwork was complete, fuel analysis was
performed by both the EPA in-house laboratory and by Paragon Laboratories. Results from all
fuel sample analysis are presented in Appendix O. Specific gravity and molar ratio (hydrogen to
carbon ratio) obtained from this sample analysis was used as input values when processing the
PEMS data. However, analysis was only performed on a subset of the samples collected, and
some samples were damaged (the sample containers were broken) during transport, so complete
sample analysis results were not available for all samples. Average values were used when
sample analysis results were not available for specific tests. For IOO trucks with unknown
refueling sources, these averages were based on the overall averages calculated over the entire
study. For fleet trucks refueled from an establishment's common fuel source, averages were
based on the fleet average for that establishment.
PEMS installation, operation and maintenance was scheduled and performed in such a
way as to minimize interruption of operations. PEMS instrumentation teams generally
performed installations during each site's non-working hours (after the trucks were no longer
needed for that working day). Hence, PEMS installations usually took place the evening prior to
the day of testing, and the instrumentation team would then arrive the next morning at least two
hours prior to the truck's scheduled departure time to warm-up, calibrate and verify the
functionality of the PEMS equipment prior to emissions testing. Generally, two trucks were
5-4
-------
instrumented each night (installations were generally performed Sundays through Thursdays),
one for gaseous only and one for gaseous and PM. Although equipment was available to
perform two concurrent PM tests, approximately six hours of cleaning and calibration of an MPS
system was required after each use, so performing two concurrent PM tests would prevent the
team from being able to perform PM testing the following day. Therefore, the optimum test
schedule incorporated one PM test each day.
5.1.2 PEMS Equipment Installation
After team members evaluated each truck for testability and test equipment layout,
various support activities associated with each installation were initiated. The primary
equipment installation tasks are described in the following subsections.
5.1.2.1 Determine Layout of Major Equipment Components
Prior to each installation, PEMS installation team members evaluated the truck to be
tested to identify the best configuration for the test system and to resolve any potential
installation or operation issues. Assessments were made of the space available behind the cab
for mounting the generator, MPS box, nitrogen and FID fuel gas bottles and other equipment,
and the exhaust system was evaluated to determine how to mount the exhaust flowmeter and
how to route the heated sample lines into the cab. As discussed in Section 4.1.1, gaseous-only
installations were somewhat less complex than full PM and gaseous installations since an MPS
chamber wasn't required and therefore the exhaust flowmeter could be mounted directly on the
muffler (after removal of the truck's exhaust stack). In addition, no gravimetric sampling
chamber, PM sampling line or nitrogen cylinder were needed for gaseous-only testing. Figure
4.1-2 shows the equipment used for full PM and gaseous installations (on a bus), and figure 4.1.3
shows a gaseous-only instrumentation on a truck the team tested during this study. As shown in
Figure 5.1-1, a gaseous-only installation could be performed on trucks with bulkheads, but full
PM installations were not possible without removal of the bulkhead, as insufficient room was
available between the bulkhead and the fifth wheel to mount the additional equipment.
5-5
-------
Figure 5.1-1 Gaseous-only Installation on Truck with Bulkhead
5.1.2.2 Install Deck Plates and Exterior Test Components
Many trucks were already equipped with deck plates, which were grated aluminum plates
clamped to the truck's frame rails directly behind the cab. If not already equipped, these were
installed on the trucks to be tested and served as platforms on which to mount equipment. Deck
plates were purchased at local truck supply outlets located near our various testing sites. After
the deck plates were installed, the exterior test system components were placed in their
respective locations. Figure 5.1-2 shows the placement of a MPS/flowmeter assembly on a deck
plate with generators mounted on a rack over the cab of a 1980 Kenworth cabover truck. In this
figure, the sample lines are routed along different paths into the sleeper cab though the
passenger's window. Figure 5.1-3 shows a full PM/gaseous system on a 2004 Kenworth sleeper
cab. Note that in Figure 5.1-3, due to space constraints, the nitrogen cylinder used for
gravimetric solenoid activation is mounted on the frame steps near the deck plate. Figure 5.1-4
shows installation of both the MPS chamber and the gravimetric sampler outside the cab of a
1993 Freightliner sleeper cab. Although the gravimetric sampler was usually located inside the
cab, insufficient room was available in this particular cab for the sampler, so this was mounted
on the deck plate adjacent to the MPS. The gravimetric sampler was positioned with the access
door facing outward to allow gravimetric filter changes when the truck was stopped. Figure 5.1-
5 shows installation of a MPS and gravimetric sampler outside a day cab on a 2003 Freightliner,
again, due to insufficient room inside the cab.
5-6
-------
When installing equipment on the deck plate behind the cab, it was critical to ensure
adequate distance was available between the tractor's fifth wheel coupling device and the
sampling / measurement equipment. Even through adequate distance may be available when the
truck is pulling a trailer in a straight path, once a turn is made, the corner of the trailer (or cargo
container) arcs toward the cab and significantly reduces the available distance from the rear of
the cab as it travels through the arc path. During installation, care was taken to ensure adequate
clearance was maintained in order to prevent trailer or cargo container interference, and in
"tight" installations (where limited room was available), before starting the workday (but after a
trailer was installed) the driver of the truck was asked to slowly make a sharp turn as installation
personnel watched the arc of the trailer corner to ensure sufficient clearance was available.
Adjustments were made to mounting configurations and locations as needed.
Figure 5.1-2 PM/Gaseous Test Components with Generators on Roof
5-7
-------
Figure 5.1-3 PM/Gaseous System on 2004 Kenworth
Figure 5.1-4 Grav and MPS Box Outside of a 1993 Freightliner
5-8
-------
5.1.2.3 Install Exhaust Flowmeter
As previously described, procedures for installing the exhaust flowmeter varied
depending on whether the test was a gaseous-only test or a PM/gaseous test. For PM/gaseous
testing, LOD personnel removed the truck's existing exhaust stack and fabricated (using welding
and band clamps) a new exhaust system to route the exhaust through the exhaust flowmeter
mounted on the MPS, as shown in Figures 5.1-3 through 5.1-5. For gaseous-only testing, the
exhaust flowmeter was mounted directly on the truck's muffler (after removal of the stack), as
shown in Figures 5.1-6 and 5.1-7. In Figures 5.1-6 and 5.1-7, it can be seen that no MPS was
required, so only the generator and the FID fuel cylinder were needed, and were secured to the
deck plates which had been installed by LOD team members.
5-9
-------
Figure 5.1-6 Gaseous-only Installation on 2003 Freightliner Fleet Truck
Figure 5.1-7 Gaseous-only Installation on 1998 International Sleeper Cab
5.1.2.4 Install In-cab Test Equipment and Route Heated Sample Lines
Concurrent with installation of the deck plate, external test systems and exhaust
flowmeter equipment, the interior of the truck was prepared for installation of the test equipment.
5-10
-------
For sleeper cabs, generally objects on the bed were cleared, the mattress was removed and
mounting platforms were installed to secure the test equipment. Some installations were
hampered due to the excessive distance from the MPS (outside) to the gravimetric sampling
chamber (inside). This was limited by the length of the system's heated PM sampling lines, so it
was critical that the proper location and orientation of all equipment be determined before
installations began, in order to avoid having to remove and reinstall equipment in new locations
Figures 5.1-8 through 5.1-10 show various sleeper cabs with PEMS and gravimetric sampling
equipment installed. Figure 5.1-8 shows a sleeper cab in which the PEMS and gravimetric
sampling unit were mounted on the lower sleeper platform. Sample lines were routed into the
cab through a sleeper window. Routing lines through the sleeper window generally provided the
most flexibility in equipment locations due to the relatively short distance required for routing
the sample lines. On the other hand, the sleeper cab shown in Figure 5.1-9 had no convenient
access ports through which the sample lines could be routed, so these lines were routed into the
cab through the passenger window (this is the 1980 Kenworth cabover shown in Figure 5.1-2).
In Figure 5.1-10, the mattress of this sleeper cab was removed and the rear side window was
removed in order to mount the equipment and route the sample lines into the cab. Figure 5.1-11
shows the outside of this same cab with the rear side window removed. Cardboard was placed
over the window opening in order to protect the cab from rain while the equipment was installed.
Figure 5.1-8 PEMS and Gravimetric Sampler in Large Sleeper Cab
5-11
-------
Figure
Figure 5.1
5.1-9 PEMS and Gravimetric Sampler in 1980 Kenworth Cabover
PEMS and Gravimetric Sampler in Cab with Window Removed
5-12
-------
Figure 5.1-11 Sleeper Cab with full PM System and Window Removed
> i
'i.iiJ.j , r
_ • "I
L Iff I
81 -i »
ll '3 Mr-
f
T> 1 iff >1
i
i Is
^ m '¦\
12/14/2009
For sleeper cabs with dual over/under bunks, frequently the best location for mounting
the PEMS was on the top bunk, as shown in Figure 5.1-12. Mounting the PEMS on the top bunk
minimized the distance the gaseous sample line had to travel from the exhaust flowmeter, and it
also made more room available on the floor of the cab for the gravimetric box and other
equipment. Figure 5.1-13 shows equipment in a sleeper cab where placement on a top bunk was
not possible and the sample line was not long enough to reach the PEMS. For this installation,
the team members installed the PEMS on a shipping crate in order to position the PEMS close
enough to the exhaust flow meter to allow the sample line to reach the PEMS. Figure 5.1-13 also
shows the blue PM sample line attached to the gravimetric filter box, shown in the lower portion
of the picture.
5-13
-------
Figure 5,1-12 PEMS installed on Top Bunk of Sleeper Cab
Figure 5.1-13 PEMS On Crate in Sleeper Cab
5-14
-------
Generally, team members attempted to route the sample lines into the sleeper cab through
a vent or window, as shown in Figures 5.1-7, 5.1-11, 5.1-14 and 5.1-15. Another option was to
route the sample lines through an access port within the cab's external tool box, as shown in
Figure 5.1-16. Team members also attempted to route sample lines though the deck's work light
located on the back of the cab, but this was not generally possible due to the small size of the
hole and obstructions such as interior panels. Often, it was necessary to route the lines through
the passenger front window, as shown in Figure 5.1-17 (this can also be seen in Figures 5.1-1,
5.1-2 and 5.1-5). Due to the weight of the heated sample lines, perches were sought on which to
support the lines. In the absence of these, duct tape was used to secure the lines to the side of the
cab, as shown in Figure 5.1-17. No residue remained as the tape was removed the next day.
Duct tape was also used to seal passenger window openings through which heated sample lines
were routed.
Other installation parameters of concern included access to the front of the PEMS unit to
monitor controls, reset the unit if necessary and change system filters, access to the gravimetric
sampling access door (to allow the operator to change gravimetric filters throughout the day),
access to the generator to allow refueling and routing of the PEMS exhaust and FID drain tubes
from the PEMS to outside the cab. Figure 5.1-9 shows an installation with easy access to both
the gravimetric sampling unit access door and the PEMS control panel. The red cooler between
the PEMS and the gravimetric unit contains the gravimetric filters to be used during the day's
testing, and the small white box on top of the PEMS unit contains extra PEMS system filters.
Figure 5.1-12 shows the clear PEMS FID and exhaust drain tubes routed outside the cab.
5-15
-------
Figure
02/04/2010
Sample Lines Routed through Sleeper Cab Window
5-16
-------
Figure 5.1-16
Sample Lines Routed through Toolbox
Figure 5.1-17 Sample Lines Routed through Passenger Window
5-17
-------
Installation in day cabs was more challenging, due to space constraints, as shown in
Figures 5.1-18 and 5.1-19. As shown in these images, the PEMS unit was installed between the
driver's and passenger's seats. Figure 5.1-5 shows the associated PM sampling equipment
located outside of the cab for the in-cab PEMS shown in Figure 5.1-18. Figure 5.1-19 shows a
platform that was constructed to support the PEMS rack, which minimized seat interference with
the PEMS' sample line, control lines, and heated filter.
Limited by the length of the heated sample lines, it was imperative that careful planning
occur prior to installing sampling equipment, especially for PM tests (primarily the SEMTECF1-
DS, MPS chamber and gravimetric sampler). Even small changes such as switching locations
between the SEMTECFI-DS and gravimetric sampler within the cab could be a significant
undertaking because of the cab's space constraints and the size and weight of the sampling
equipment, so minimizing changes after equipment installed was critical. This was especially
true when testing vehicles for which the sampling lines could not be routed into the cab except
through the passenger side window (as shown in Figure 5.1-17).
5-18
-------
5.1.2.5 Install Test System Accessories
Other test system components that were installed included the weather station, GPS
antenna and VI connector (or optical sensor for RPM measurements if no VI connector was
available). Weather stations can be seen mounted to the cab hand rails in Figures 5.1-14 through
5.1-16, and on the rear view mirror in Figure 5.1-17. Figures 5.1-20 and 5.1-21 show two
common locations for VI connectors. Another common location was at the outside (door-facing
side) base of the driver's seat. Figure 5.1-22 shows a mounted optical sensor. The cardboard
strip at the base of the sensor was used to reduce the strength of the high-powered magnet while
positioning the optical sensor, and this cardboard was removed once the optical sensor was
aligned. GPS antennas were frequently located on the dash below the windshield, although
occasionally they were routed outside the cab and mounted on cab exterior platforms.
Due to the heat of the generator exhaust, the generator had to be mounted in such a way
as to direct the exhaust away from the pneumatic hoses attaching the cab to the trailer (glad
hands). In addition, cable tie-wraps were used to tether the pneumatic hoses away from the
generator exhaust or other hot or pinch points on the sampling equipment.
5-19
-------
5.1-20 Typical Under-Dash VI Connector Location
Typical Dash Front VI Connector Location
5-20
-------
5.1.3 Prepare Equipment for Testing
5.1.3.1 Preparations the Night of Installation
Once all equipment was installed and connected, the system was turned on, warmed up,
calibrated and prepared for the following days' testing. System checks were performed which
included verification of the proper exhaust flow meter and input settings, review of measured
exhaust flow rates, verification of dilution ratios, verification of MPS block pressures and
verification of MPS proportionality. The SEMTECH heated filters were replaced, and a system
leak check was performed. Gas audits and calibrations were then performed, as necessary, and
all system settings were established and verified. Acquisition of external data, including weather
station data, GPS data and engine computer datastream data or external RPM data was also
verified at this time. If any systems were found to be malfunctioning or operating out of range,
in-vehicle repairs were made or the system was replaced with a backup unit. An overview
checklist of the tasks associated with preparing the system for testing is provided in Appendix P,
"Drayage PEMS Checklist_20100129'\
After all systems were prepared for the following day's testing, the vehicle was secured
for overnight storage. Generally, the team attempted to leave all systems "hot", operating on
shore power, eliminating the need for lengthy warm-ups the following morning. If rain was
5-21
-------
forecast, the system was covered with a large tarp to prevent water incursion, as shown in Figure
5.1-14.
5.1.3.2 Test Day Preparations
The next morning, on the day of testing, field crew members would arrive approximately
two hours before the truck's scheduled departure time to prepare the truck for testing. If shore
power had not been available to keep the system "hot" overnight, some field crew members
would arrive approximately additional hour earlier (before the calibration/setup crew) in order to
start the generators and equipment to allow the pre-test warm-up using generator power.
The functionality of all systems would again be verified, including acquisition of external
parameters. Gravimetric filters were inventoried and recorded on gravimetric filter data
collection sheets, which were provided to the person scheduled to do the ride-along (along with
tools necessary to switch gravimetric filters). The ride-along also carried spare SEMTECH
heated filters, in case they were needed during the day.
All systems were verified and recalibrated as needed and described in Appendix P. This
would include performing a zero on the MPS transducers, checking for proportionality and
appropriate exhaust flow rate ranges, and performing zeros and calibrations on the PEMS
gaseous measurement systems.
Immediately before testing was to commence, the ride-along would install the
gravimetric filters in their holders, recording all necessary information on gravimetric filter data
collection forms as shown in Appendix Q. Gravimetric filters were handled according to the
"Gravimetric Filter Handling SOPs" provided in Appendix R. A PEMS test session and a PEMS
test were initiated immediately before the truck was placed into service.
5.1.4 PEMS Testing
When the truck driver arrived, typically the truck was started and allowed to warm up for
10-15 minutes (or more) while the driver received the day's orders from dispatch. During this
time, any last minute system checks or calibrations were performed, along with collection of an
"idle" gravimetric filter (if sufficient time was available after the truck had been warmed up but
prior to departure). Immediately before departure, the generator fuel was topped off, spare
gasoline containers (for generator refills) were secured on the truck behind the cab, all tie-down
cables were tightened, and hoses and lines were routed away from the generator exhaust or
interference points.
5-22
-------
PEMS measurements were gathered over a complete working day, usually between six
and ten hours. No specific drive cycles were used during testing, and all PEMS testing was
based on actual in-use loads and speeds. Using a laptop computer, the ride-along monitored all
system parameters and also refilled the generator and collected and recorded data on gravimetric
filters. If questionable readings or problems were encountered during the day, the ride-along
would contact ASD or Sensors personnel back at the on-site support trailer. Remote diagnostic
support was provided, as needed, and if necessary the team would meet the truck during the day
(usually on a return visit to the installation facility) in an effort to diagnose or correct any
problems that could not be remotely corrected.
Gravimetric filters were collected and switched throughout the day, as listed in Appendix
T, the Drayage Filter Log. As shown in this appendix, certain filters were dedicated to "in port"
activities. The in-port (vs. non-port) emissions are segregated in Section 7, Study Results.
Appendix U identifies "port" activity for all testing (including gaseous-only testing), as
identified through review of ride-along activity logs presented in Appendix V. PDF scans of all
hardcopy filter data collection forms are provided in Appendix W.
Throughout the day, the ride-along would reload gravimetric filters, following guidelines
in Appendix R, the Gravimetric Filter Handling SOPS, and Appendix Z, the Filter System User
Manual. Figure 5.1-23 shows the gravimetric filter access door open, allowing access to the
URG gravimetric filter holders within.
5-23
-------
MEAf ED lltlC
Figure 5.1-23
Access to Filters within the Gravimetric Sampler Assembly
S iTlTECH-DS
s sors. inc. mM» *
mOBILE Emipsic
When not in use, gravimetric filters were transported and handled in plastic holders
sealed in Ziploc plastic bags and carried/stored in small coolers. However, since these filters
were loaded in the field, they were briefly exposed to ambient contamination as they were
removed from their bags and holders and placed in the sampling equipment. Filter loading was
performed inside the cab of the truck to minimize exposure to ambient contamination. Dynamic
and field gravimetric filter blanks were collected during the study in order to identify and
quantify the extent of ambient PM contamination on PM data resulting from gravimetric filter
transport and handing activities. Field blanks were treated as actual samples, including all
shipping, handling and transport to the field during testing, although they were not placed in the
gravimetric sample system holder. Dynamic blanks were also treated as actual samples but in
addition to field handling they were placed in the gravimetric sample system holder rack during
emissions testing. However, no exhaust sample (or air) was routed through the dynamic blanks
(the flow-control solenoid on the gravimetric sampler which held the dynamic blank remained
closed during testing). Comparison of the measurement results for dynamic blanks and field
blanks can help provide information regarding potential system contamination resulting from
either the filter holder or leaks in the flow-control solenoids used to i solate that specific filter
holder in which the dynamic blank was placed. Dynamic and field blank results are presented in
5-24
-------
Table 7.1-1. Appendix T (Drayage Filter Log) lists results for each sampled filter and also lists
sample type such as port or idle operation. Appendix M (Drayage Emissions Measurement
Results) lists by-filter emissions for all sampled filters.
Other ride-along support activities included refueling the generator during breaks in
activity, monitoring data and system parameters and ensuring all systems remained safely
secured. Upon return to the facility at the end of the day's activities, the PEMS test would be
ended, and a post-test calibration and zero would be performed to quantify the measurement
instrument's drift throughout the day. After the post-test calibrations, the PEMS data file session
was ended, and the data was processed, plotted and reviewed to identify any problems. All
gravimetric filters were retrieved and transferred to the onsite manager for logging and
recording. A complete list of ride-along responsibilities are provided in Appendix S, the
Drayage Ridealong Checklist.
5.1.5 Post-Test Activities
As mentioned in the preceding section, immediately after retrieval from the PEMS, all
data was processed using the SEMTECH-DS post-processor for an in-field data review. System
settings, flow rates, proportionality and pollutant concentrations were evaluated to assess test
validity. If any suspicious parameters or suspect data was encountered, an equipment evaluation
was performed and onsite managers would discuss whether an additional day of testing would be
performed on that truck.
PEMS data (along with all other relevant project data and information) was archived in
multiple locations for later transfer to the project-specific secure FTP site.
Every day after testing, the MPS was removed from service for a complete
disassembly, cleaning and recalibration to be performed the following day (while the next day's
test was underway). In this manner, two MPS units were alternated for use/service every other
day, allowing daily PM testing (as weather permitted). Service on the MPS was performed
according to guidelines presented in Appendix AA, the MPS User Manual (rev 1.01, revision
still underway).
5.2 PAMS Installations
This task involved the instrumentation of selected drayage trucks with Isaac PAMS for
acquiring activity data, including date/time, engine speed (via ECU data when possible) and
position (GPS). As described later in Section 5.3, fourteen of the twenty-three trucks which were
used in the PAMS study also received a PEMS test. The remainder of trucks generally were also
5-25
-------
trucks which had received an RSD test, but a few of the trucks were not part of the 1877-truck
RSD-tested fleet. During the course of the study, a mutual decision was made between onsite
ERG and EPA personnel that trucks not included in the fleet of 1877 RSD-sampled trucks could
receive PAMS instrumentations, as long as the trucks would be doing port work and fell within
the model year profile of the 1877 truck fleet which had received RSD tests. For those trucks
which received both a PEMS test and a PAMS test, the PAMS unit was disconnected from the
truck during the day of the PEMS test (to allow the ECU datastream or optical RPM signal to be
used for the PEMS test). For these trucks, the activity data from the PEMS test was later merged
into the PAMS data for a complete activity datafile (in essence adding an extra day of activity
data).
Prior to each PAMS instrumentation, vehicle information was written onto data collection
forms (the same form as used for PEMS testing) shown in Appendix H. All information
collected on these forms is included with the PEMS instrumentation data provided in Appendix
I. As with PEMS instrumentation, all PAMS data collection was enhanced with digital
photographs.
As shown in Figure 5.0-1, PAMS installations occurred in four waves coinciding with the
PEMS fieldwork activities. PAMS data was collected for approximately one week for each
installation. This relatively short activity collection cycle was believed to be appropriate for
several reasons:
• Because drayage drive-cycle variability was expected to be lower than that for
many other source categories, a one-week data collection duration would allow
collection of adequate activity information for most vehicles
• One-week data collection periods reduced the need for staff to revisit
instrumented vehicles in order to download data, verify functionality and correct
problems
• Because of a lower chance of instrumented vehicle reassignment, shorter data
collection duration was felt to reduce the risk of loss of equipment, and
• Short data collection duration allowed a greater number of vehicles to be tested
over a given timeframe
Six PAMS units were used throughout the study, as shown in Figure 5.0-1. PAMS
installations generally began near the start (or prior to) each PEMS testing phase, and continued
until all six dataloggers were installed, calibrated and functioning properly. Two to three ERG
personnel performed PAMS installations, and each phase of six truck installations was
completed in two to three days. The ERG PEMS/PAMS field manager served as a liaison
between the participating establishment and the installation team (as well as the PEMS
5-26
-------
installation team) to ensure PAMS instrumentations did not disrupt drayage activities or impede
PEMS testing.
Two general types of PAMS instrumentations were performed during this study. For
vehicles which were not computer controlled (or ECU data was not available for any reason),
RPM was captured using an optical RPM sensor (with reflective tape affixed to a rotating object,
such as the engine's harmonic balancer), and DC power for the datalogger was taken from the
truck's switched power source (generally the ignition switch), as shown in Figure 5.2-1. For
computer-controlled vehicles with an available ECU port, power was taken from the ECU port
(either a 6-pin or 9-pin Deutsch connector). ECU-port installations collected RPM (along with a
number of other engine parameters) broadcast in the ECU datastream.
Figure 5.2-1 Switched Power Terminal used for Non-ECU PAMS Installations
Once the type of installation was determined, a suitable location was identified for the
PAMS box and an appropriate harness was selected. As the Isaac units are equipped with 3-D
accelerometers, an attempt was made to mount each PAMS vertically, so its "z" axis (as
indicated on the top of the unit) pointed along the length of the truck, and the "y" axis pointed
directly upwards. However, this was not always possible, depending on the type of truck and
available mounting options. Datalogger orientations were noted on the instrumentation forms
(and static values of the 3-accelerometers while the truck is at rest can also be used to determine
datalogger orientations). Figure 5.2-2 shows a PAMS installed against a seat base. For this
5-27
-------
particular installation, the "z" axis is properly aligned, but the "x" and "y" axes are not in their
ideal orientations (although this can be corrected during data analysis, as needed).
Figure 5.2-2 RAMS Installed on a Seat Base
Once a suitable mounting location for the PAMS unit was identified, the appropriate
wiring harness was routed from the optical sensor or the 6-pin or 9-pin connector (whichever
method was used). The installation type (optical, 6-pin or 9-pin) determined the type of harness
to use. The GPS antenna was also connected to the Isaac harness and mounted somewhere on
the dash, as shown in Figure 5.2-3. Wires were routed from the GPS antenna and the power
switch (or Deutsch connector) carefully to be hidden from the truck driver and occupants (as
much as possible). Figure 5.2-4 shows wires leading from a Deutsch connector to the Isaac
datalogger prior to them being covered by the truck's sill plate (which had been removed for the
installation).
5-28
-------
Figure 5.2-3 Typical PAMS GPS Mounting Location
Figure 5.2-4 PAMS wiring from a Deutsch Connector
Once the hardware was installed and the wiring was connected, a laptop was connected to
the PAMS and configured using an appropriate configuration file developed during the mockups
5-29
-------
described in Section 4.2. One of three configuration files was used for each installation, either
external RPM/GPS (for trucks with an optical RPM sensor), SAE J1708 data (for trucks with 6-
pin Deutsch connectors), and either the SAE J1708 configuration file or the SAE J1939
configuration file (depending on model year, for trucks with 9-pin Deutsch connectors). After
download of the configuration file, the truck was started, successful data acquisition of all
parameters was verified, and the protocol was changed as necessary. All parameters were
reviewed, and the a final check was performed of the datalogger and harness/accessory
installations. At this time, the truck was ready for release. Detailed installation guidelines are
provided in Appendix F, Drayage PAMS SOPS.
After installation was complete, a tentative appointment for removal of the unit was made
with each truck driver and the truck was released. After the week of instrumentation had passed,
ERG contacted each driver and confirmed the removal appointment. During removal, data from
each PAMS unit was downloaded onto a laptop, processed and evaluated (while the PAMS was
still installed on the truck). The PAMS and accessories were then removed, and the truck was
restored to its original condition.
5.3 PEMS and PAMS Testing Summary
PEMS and PAMS testing totals are shown in Tables 5.3-1 through 5.3-3. Table 5.3-1
summarizes PEMS testing, by NOx bin and model year group, as described in Section 3.2.2. In
Table 5.3-1, bins which were completed (i.e., the target number of trucks received full
gaseous/PM tests) are shaded gray. Green numbers (shown italicized in the top of each cell) in
Table 5.3-1 represent bin targets, and red numbers (bottom of each cell) represent the total
number of trucks which received gaseous/PM tests. Black numbers (which are shown in
parenthesis) represent trucks which received gaseous only (oversample) tests. Some trucks
received multiple PEMS tests, as shown in Table 5.3-2. In total, 37 trucks received PEMS tests,
and 9 of these 37 trucks were tested twice (resulting in 46 PEMS tests performed throughout the
study). Overall, 22 gaseous/PM tests were conducted, and 24 gaseous only tests were conducted.
In Table 5.3-1, only one test is shown for trucks which received more than one test, i.e., a PM
test represents either two gaseous/PM tests or a gaseous and a PM test for trucks 2JC711,
1190901-A, 0182003-A and 0182022-A. Consequently, the totals obtained by summing the cells
in Table 5.3-1 will not yield 37 (trucks) or 46 (PEMS tests) shown in Table 5.3-2.
5-30
-------
Table 5.3-1 PEMS Testing by NOx Bin and Model Year Group
(0
Q.
3
o
L.
0
>-
"53
T3
o
Table 5.3-2 PEMS and PAMS Test Counts, By Truck
Recruiting Bin
Info
Truck Information
PEMS Test Types
PAMS
MY NOx
Group Bin
Model Truck
Yr Plate ID Make
Gaseous Gaseous
Only & PM
04-06
2d
2004
1191083-A
3300
Kenworth
2
1
04-06
-1d
2004
0185067-A
787
Kenworth
1
1
04-06
-1d
2004
1191018-A
647
Freightliner
1
98-03
-1c
2003
0190306-B
321
Volvo
1
98-03
-1c
2000
0218126-A
232
Freightliner
2
98-03
-1c
1999
0180990-A
829
Freightliner
1
98-03
-1c
2000
0186029-A
455
Freightliner
1
94-97
-1b
1997
0160699-A
559
Freightliner
1
78-93
-1a
1980
0181276-B
668
Kenworth
2
1
04-06
Od
2004
1191084-A
3400
Kenworth
1
98-03
0c
2003
0183710-A
C105
Freightliner
1
1
98-03
0c
2000
0185774-B
260
Kenworth
1
98-03
0c
2003
0183713-A
C108
Freightliner
1
1
98-03
0c
2003
0183716-A
c111
Freightliner
1
1
98-03
0c
2001
1095729-B
1800
Freightliner
1
1
98-03
0c
1998
0185576-A
778
International
1
1
98-03
0c
1999
1144157-A
None
International
1
98-03
0c
1999
0183558-A
356
International
1
NOX Emission BINS
-2
-1
0
1
2
No
RSD
2007 -2010
0
0
0
0
0
0
2004 - 2006
0
2
1
0
1
(2)
1998 -2003
0
2(2)
2(8)
2
0
(4)
1994-1997
0
0
0(1)
1 (1)
3
0
(2)
1978 -1993
0
1
1
1
0
5-31
-------
Recruiting Bin
Info
Truck Information
PEMS Test Types
PAMS
MY NOx
Group Bin
Model Truck
Yr Plate ID Make
Gaseous Gaseous
Only & PM
98-03
0c
2001
0185728-A
812
Freightliner
1
98-03
0c
2001
1095730-A
2000
Freightliner
2
1
94-97
0b
1994
0178515-B
414
Freightliner
2
1
94-97
0b
1997
0190992-A
834
Freightliner
1
78-93
0a
1993
0181096-A
435
Freightliner
1
1
98-03
1c
2003
1190901-A
771
International
1
1
1
98-03
1c
2003
0183718-A
C113
Freightliner
1
1
94-97
1b
1996
0182003-A
9210
Volvo
1
1
1
94-97
1b
1996
0181157-A
398
Kenworth
1
94-97
1b
1997
0184781-A
776
Freightliner
2
78-93
1a
1993
0181021-A
None
International
1
04-06
NONE
2004
0160910-A
T796
Freightliner
1
04-06
NONE
2006
1157629-A
851
Freightliner
1
94-97
NONE
1997
0190786-A
T639
Freightliner
1
94-97
NONE
1997
0267938-A
20
Freightliner
1
98-03
NONE
1999
0188529-B
830
International
1
98-03
NONE
1998
0189106-A
T657
Freightliner
1
98-03
NONE
2000
0182291-A
794
Kenworth
1
98-03
NONE
1998
0182022-A
1161
International
1
1
98-03
N/A
2002
0183438-A
L-
0247
International
1
98-03
N/A
2003
0183439-A
L-
0306
Freightliner
1
98-03
N/A
1999
0186809-A
459
Freightliner
1
98-03
N/A
2001
0186844-A
785
Freightliner
1
94-97
N/A
1997
0190758-A
L-
9754
Freightliner
1
98-03
N/A
2000
1075087-A
001
Peterbilt
1
98-03
N/A
2000
1190370-A
3000
International
1
98-03
N/A
2003
1191097-A
C121
Freightliner
1
98-03
N/A
2000
Temp
821
Freightliner
1
Occasionally some of the gaseous-only oversampled trucks were not part of the 1877
RSD-tested fleet (some gaseous-only trucks did not received a pre-study RSD test). The
decision to test non-RSD sampled trucks was made between ERG and EPA onsite staff in order
to maximize the number of trucks tested, since an adequate number of RSD-sampled trucks was
insufficient to meet gaseous/PM and gaseous-only sampling needs at some facilities. All trucks
which received gaseous/PM tests were within the fleet of 1877 RSD-tested trucks.
5-32
-------
Table 5.3-2 also lists the 23 PAMS instrumentations which were performed during the
study. Of the 23 PAMS instrumentations, 14 of those were on trucks which also received PEMS
testing. As previously described, during the course of the study, a mutual decision was made
between ERG and EPA that trucks not included in the fleet of 1877 RSD-sampled trucks could
receive PAMS instrumentations, as long as the trucks would be doing port work and fell within
the model year profile of the 1877 truck fleet which had received RSD tests. Table 5.3-3 lists
general statistics regarding the number of PAMS-instrumented trucks within each model year
group. This is shown graphically in Figure 5.3-1. Due to truck availability and fieldwork
logistics, the 1998-2003 model year group was oversampled, while the 1994-1997 group was
undersampled. However, this is not expected to influence the nature of activity data collected
during this study.
Table 5.3-3 PAMS Tests by Model Year Group
MY Group
78-93
94-97
98-03
04-06
07-10
# in Sample
122
494
1065
131
65
Sample %
7%
26%
57%
7%
3%
PAMS Count
1
3
17
2
0
PAMS %
6%
6%
76%
12%
0%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Figure 5.3-1 Trucks PAMS Tested vs. RSD-Tested Fleet
PAMS Testing Summary
L
i Sample Percentage
1 PAMS Percentage
78-93 94-97 98-03 04-06 07- 10
MY Group
Additional information pertaining to each PEMS or PAMS test, including details
instrumented trucks PEMS/PAMS operating and setup parameters for each test is provided in
5-33
-------
Appendix I. A log summarizing all tests performed throughout the study and participation
incentive payments issued for each test is provided in Appendix L.
5-34
-------
6.0 Data Analysis and QC
As much as possible, all PEMS and PAMS systems (and datastreams) were monitored
during collection followed by extensive QC and processing performed after data collection was
completed. Data processing, QC and analysis steps are described for PEMS and PAMS data in
the following sections.
6.1 PEMS Data Processing and QC
As mentioned in Section 5, the operation of PEMS test equipment was continuously
monitored by personnel throughout each test. Real-time monitoring of test parameters was
performed by the ride-along team member via a laptop communicating with the PEMS (and
MPS) in order to identify and correct any data or equipment issues. In addition, test files were
extracted during and immediately after each test, processed and reviewed for data quality issues
or problems. This real-time "in-field" QC was performed in order to identify issues with exhaust
and MPS flows and gravimetric filter system flowrates, system temperatures and pressures,
pollutant concentrations and other measured and recorded parameters.
After fieldwork was complete, all PEMS data underwent processing to create time-
aligned, comma-delimited data files incorporating test-specific input settings for each test. Initial
data review and QC was also performed at this time. This initial data processing and QC was
performed by Sensors, Inc. using their SEMTECH post-processing software, Excel-based
plotting macros, and also through manual review of data files. Once Sensors processed, QC'd
and corrected all PEMS files, they transferred these files along with all notes pertaining to data
review and corrections to ERG. All data correction and review notes provided by Sensors can be
found in Appendices J and K. Appendix J contains general PEMS data QC guidelines used
during data review (App J-l), notes compiled by Sensors during data processing and corrections
(App J-2), and information pertaining to data issues, processing and corrections which were
made to the PEMS data during Sensors' processing and QC (App J-3). Appendix K contains a
by-test/by filter review of all PEMS second-by-second data performed by Sensors. This
appendix contains details of the QC described in Appendix J.
Upon receipt of the processed PEMS data and notes from Sensors, ERG performed
additional data analysis and validation in order to resolve all outstanding data issues. Results of
this second data review are provided in Appendix J-4. Once the second review was complete,
ERG imported the data into Statistical Analysis Software (SAS) and processed results for
analysis, reporting and data delivery. In addition to applying data analysis filters within SAS, all
6-1
-------
previously identified issues were flagged or corrected, as appropriate. A complete list of filters,
corrections and flags applied in SAS is provided in Appendix J-5.
The following sections describe the specific areas of focus during processing and analysis
of the drayage PEMS data. Test-specific details of analysis can be found in Appendices J and K.
6.1.1 Initial Drayage Processing and QC
The primary areas of focus during Sensors' data processing and QC included the following:
Exhaust transport delays: Sensors time-aligned all data, accounting for exhaust transport and
gas bench measurement delays, ECU data acquisition delays and GPS signal acquisition delays.
Alignments were performed using known system rates and were confirmed visually by
comparing relevant time-series plots (i.e., CO2 concentration vs. exhaust mass flow rate, CO2
concentrations vs. ECU derived torque, exhaust mass flow rate vs. ECU RPM, and GPS speed
vs. ECU speed and PEMS fuel rate vs. ECU fuel rate). Data was generally aligned within one
second during Sensors' post-processing.
Mass emission calculations: Mass emission calculations were based on measured carbon
concentrations and exhaust flowrates, excluding tests in which the flowmeter data was invalid or
missing. In those cases, mass emission calculations were based on the ECU fuel rate. Mass-
based emissions from the following tests were based on ECU fuel rate: 0182003-
A_20100309_E, 0183710-A_20091215_E (only part of this test) and 1190901-A_20100208_E.
Vehicle speed: Vehicle speed was primarily based on GPS data. If GPS-based speed was found
to be invalid, vehicle speed from the ECU datastream was used for that particular test.
Engine speed: Engine speed (RPM) was primarily taken from the ECU datastream, if an ECU
datastream was available. For a few tests, the truck was either mechanically controlled (hence
no ECU data) or the ECU port was inaccessible, i.e., the available diagnostic link connector
(DLC) had been removed. In these instances, an optical RPM probe was used and calibrated
with either a handheld optical tachometer or the truck's onboard tachometer. The following tests
were performed using an optical sensor: 0181021-A_20091215_E, 0181276-B_20091208_E and
1191084-A_20080127_E. For these tests, Sensors verified the PEMS software's RPM multiplier
was correct during post-processing.
Kh Calc method: 40 CFR 1065.670 methodology was used for the NOx humidity correction
factor.
Weather data: Sensors post-processed PEMS results using data from the weather probe.
Fuel settings: #2 diesel, specific gravity and hydrogen to carbon molar ratios used during post
processing were based on fuel testing results EPA performed on samples collected during the
study. Specific values based on fuel testing results are provided in Appendix AC, Drayage
Processing Parameters. When data was not available for an individual truck, averages were used
6-2
-------
from that establishment (when fleet vehicles were refueled from a common source), or overall
study averages, as appropriate.
Detection Limits: All zero (default value) detection limits were applied
Calculation limits: Default values (1000 rpm/s, 21 mph/s, 0.05 gal/s, 0.5 % C for fuel specific
dropout, 0.005 bhp-hr for brake specific dropout, 4 samples for THC FID auto range) were
applied.
Drift corrections. For tests in which valid pre and post-zeros and spans had been collected, drift
corrections were applied during data processing. Frequently, this required Sensors merge
separate XML files containing the various pre and post-test calibrations. Once these were
merged into the test data file, the SEMTECH post-processor applied drift corrections (on a
percentage basis) based on zeros and spans (linear interpolation correction of drift throughout the
test). The post-processor also applied corrections based on any autozeros which had been
performed during testing. The test files ERG received from Sensors had drift corrections applied
(on a percentage basis) whenever possible. Details of which files were drift corrected are
provided in Appendices J-4 and K.
O2 sensor corrections /adjustments. O2 ranges were evaluated, but O2 sensor corrections were
not applied. Exhaust is assumed to have no dilution.
Autozeros: Sensors output the sample path field to identify when autozeros were performed
during sampling (an "ambient" sample stream suggests an autozero is being performed). In SAS,
ERG has flagged and excluded autozeros. Field guidelines specified autozeros only be
performed during times when the truck was not in active use, so autozeros generally did not
affect active data collection.
Other data review & corrections . Through review of time-series plots of all primary measured
and system parameters, Sensors identified and flagged improper temps, flows, dilution, pollutant
concentration, proportionality, etc., as listed in Appendices J and K. In addition, any data
corrections which were applied during post-processing are described in these appendices.
Correct certain test files collected when flowmeter 33341 was plumbed incorrectly: At some
point during the study (prior to February 9th, 2010), the tubing on flowmeter 33341 was switched
(labeled wrong and tubing consequently plumbed incorrectly). This was identified (and
corrected) by a Sensors technician during a field installation on March 14, 2010. Four tests were
performed with this "misplumbed" flowmeter. This resulted in an associated bias in the exhaust
mass flow rate (measured flowrate was approximately 40% underreported). To correct this,
Sensors performed a lab correlation comparing measurements from the flowmeter in the mis-
plumbed configuration with results from their laminar flow element across varying flow ranges.
This comparison was used to develop a correction factor (of 1.62) to be applied to the exhaust
flow data during post-processing, as described in Appendix J-3. This correction should be
applied if the PEMS data is reprocessed in the future. This affects the following four tests:
0160910-A_20100211_E, 1144157-A_20100308_E, 0267938-A_20100310_E, and 0185728-
A 20100312 E.
6-3
-------
6.1.2 Subsequent Drayage QC, Analysis and Reporting Summaries
Building on the work already performed by Sensors, ERG continued analysis and QC of
the processed PEMS data using Sensors' data review notes. The following steps were
performed.
Perform final data QC and validation: As described in Section 6.1, ERG performed additional
data engineering review and analysis on the drayage PEMS data in order to identify and resolve
all outstanding data issues. Second-by-second review involved evaluation of all gaseous
pollutants, review of sampling system pressures such as the MPS inlet pressure and SEMTECH
pressures, evaluation of all system flows including the exhaust mass flow rates, calculation of
fuel flow rate, all MPS sampling flowrates and gravimetric filter flowrates, and evaluation of all
system and sampling temperatures such as exhaust temperatures, external heated line, chiller,
cyclone, manifold and gravimetric filter temperatures, and ambient and internal PEMS
temperatures. Results of this secondary data review are provided in Appendix J-4. This review
was primarily performed by analyzing the time-series plots of all measured pollutants and system
parameters that had been created by Sensors during the initial data processing process.
Perform date and time assignments and corrections - All dates and times were reviewed and
corrected, as necessary. This primarily corrected all timestamps to be consistent with the central
time zone (either Central Standard Time or Central Daylight Savings Time). Adjustments were
made based on GPS times recorded in the datafile, which represent Greenwich Mean Time.
Perform gravimetric filter reconciliation: As previously described, information pertaining to
gravimetric filters collected during a test was recorded by ride-along personnel on filter data
collection forms. Information collected in order to identify the filter during data processing
included filter ID (unique laboratory serial number for each filter) along with holder location
(sample holder 1, 2 or 3). In this way, the overall sequence of holder IDs activated could be used
to identify each gravimetric filter. However, for several reasons, including multiple sampling of
filters (with bypass in-between) and switching back and forth between filters multiple times,
some discrepancies were identified between the gravimetric filter sequence recorded in filter data
collection forms and the gravimetric sampling sequence recorded in the data file for each test.
Using information contained in the ride-along notes collected during sampling and filter pressure
differential data recorded in the data file, ERG reconciled information in the filter log to
information recorded in each data file. The final sampling sequence is shown in Appendix T,
Drayage Filter Log.
SAS readin and QC: All data provided by Sensors was read into SAS. Test files were compiled
into a single file for each day of testing. In SAS, ERG flagged data collected during autozeros,
data in which one or more pollutant or RPM values are missing and data identified during ERG's
or Sensors' second-by-second engineering review as suspect or invalid. In general, suspect data
is excluded from emission summaries included in this report, and is also identified in the final
emissions database provided for this project. The filter ID sampling sequence previously
determined was assigned in SAS, and filters affected by excluded data are identified in emission
summaries included in this report and in the filter log in Appendix T. GPS speed
("iGPS Ground Speed, mph") was used as the primary vehicle speed. Vehicle speed from the
6-4
-------
ECU "iVEHSPEED, mph" was used if the GPS speed was invalid, missing, or suspect. When
GPS speed was used, this variable was advanced by 2 seconds (2 seconds earlier relative to the
rest of the data file) in order to refine the GPS alignment to other system parameters. All
corrections/adjustments made to variables are listed in Appendix J-5.
MPS proportionality to flow review: In SAS, ERG plotted the MPS average sample flow rate
(iMPS_Average_Q, SCCM) against the exhaust mass flow rate (icMASS_FLOW, kg/hr) and
calculated the best fit line slope, intercept, coefficient of determination (r2) and root mean square
error (RMSE) in order to assess the quality of proportionality. Results are provided in Appendix
AF, MPS to Exhaust Flow Proportionality Plots. Proportionality plots are provided both for the
overall test and also for only the time periods when filter sampling was being performed. Filter
proportionality is also used in assessing gravimetric filter result validity listed in Appendix T
(Drayage Filter Log).
Calculate brake-specific power for each observation: In SAS, ERG calculated brake-specific
emissions using several different methods. Most vehicles were electronically controlled and
some indication of load was available via the ECU datastream. For these vehicles the ECU load
was used, but needed to be manipulated in order to calculate the net 'brake' torque at the
flywheel for VSP bin assignments.
For this discussion of the calculations that were performed in order to calculate brake torque, the
following definitions apply (much of this information is listed in Section 12.6.2 of the Oct 2008
SEMTECH-DS User Manual):
Brake torque - Engine torque value (N-m or ft-lb), which is the total torque (cylinder
torque) minus the engine's frictional torque loss. Brake torque is zero when engine is
running with no accessory or powertrain loads. The brake torque value is used for VSP
calculations.
Frictional torque - Frictional torque, representing the frictional losses within the engine,
is calculated by multiplying the two J1939 ECU datastream parameters "percent friction
torque" and "reference engine torque". Frictional torque may then be subtracted from the
calculated cylinder torque to calculate brake-torque (i.e., net torque). For trucks using the
J1708 protocol, the frictional torque is accounted for by using the recorded cylinder
torque values at idle as described below.
Cylinder torque - Total engine torque value (N-m or ft-lb) before frictional torque is
deducted. Cylinder torque is non-zero during unloaded engine operation.
Derived torque - ECU datastream torque value (N-m or ft-lb) representing the engine's
torque output at any point in time - ECU derived torque may be brake or cylinder torque,
depending on the engine's operating load at that point in time and the way in which the
different engine manufacturers program ECUs.
Lug curve - Total maximum brake torque value achievable for any specific RPM, as
defined by the engine manufacturer. Lug curves are included in J193 9 datastream as
6-5
-------
"reference engine torque" but needed to be obtained from engine manufacturers for many
of the J1708 tests. Most of the lug curves used for these calculations were taken from
advertising brochures printed by the engine manufacturers.
Percent load- ECU datastream value (%) which is the derived torque divided by the lug
curve torque at the current RPM (ie for a given RPM, it is the current torque output
divided by the maximum torque output). Note that percent load may be brake or
cylinder. Ideally, the "unloaded" percent load would be determined by operating the
engine at various speeds with no load (no accessories, a/c, power take off, or vehicle
motion). However, the status of air conditioner or other accessories was not known for
this study's tests. Therefore, for this analysis, we assume that the "unloaded" percent
load is just the average of the values seen with the vehicle at idle with no motion.
Idle Offset - As defined by ERG, this is the value of either ECU torque or percent load
that is read from the ECU while the engine is at idle with zero net torque output. ERG
calculated this value based on the average torque or percent load during the first extended
idle recorded for each truck in which the engine was warmed up to or near its typical
coolant temperature. It can also be thought of as the friction torque at idle.
For the data from each truck, one of the following three methods was used for correcting
the ECU derived torque to brake torque:
Method 2 (J1708) - This is Method 2 listed in Section 12.6.2.2 of the Oct 2008
SEMTECH-DS User Manual (Method 1, the 30 second window, is not applicable for this
study). Method 2 uses a manufacturer-provided lug curve to convert percent load to
derived torque. Specifically:
Derived torquebrake = percent loadbrake x lug curve
In Method 2, if percent load is given as cylinder load, it may be converted to brake load
using the following WVU equation which uses the percent load idle offset as defined
above:
% loadbrake = (% loadcyi - % loadcyi @ idle) / (100 - % loadcyi @ idle)
This method was used for the trucks that used the 1708 protocol but did not
output a derived torque, only a percent load.
Modified Method 2 (J1708) - This method was created by ERG to adjust derived torque
values from J1708 protocol engines for frictional losses. It was used for all J1708 trucks
that output an absolute value for derived torque such that an outside lug curve was not
necessary. The method uses a similar correction to Method 2, in which the torque is
adjusted downward at low load to account for frictional losses, but not adjusted at the
maximum load. The adjustment is linear across the range of engine loads. The equation is
as follows:
6-6
-------
Net Torquebrake = (1 + Torqueidie/(Torquemax - Torqueidie) x Torque -
(Torqueidie+ Torqueidie2 / (Torque max Torqueidie))
Where
Torqueidie = The average derived torque reading when the engine is idling (idle
offset)
Torquemax= The maximum recorded derived torque reading over the entire test
Torque = The derived torque reading to be corrected to net torque
Modified Method 2 was created because the maximum derived torque values
observed in the ECU data were very close to the maximum torque values given in the
torque curves found based on engine tag data. If Torqueidie had been subtracted directly
from the maximum torque, the resulting data would not have matched the maximum
torque curve values for many of the trucks. For this reason, the equation was developed
to adjust for frictional losses at idle, but not to adjust the torque at points where the torque
was close to the correct net torque as given by the manufacturer.
All J1708 protocol trucks had a single assigned idle offset value, but the idle torque (or
load) value was not constant during operation and each truck had variability in this value.
Because of this, the recorded data would still include many positive and negative values
for torque at idle. Because net torque cannot be negative at idle and because negative
torque does not have any significance from an emissions perspective, all negative values
for torque were set to zero. In order to avoid the upward bias in the total engine work
output that this would cause, all positive idle torque values were then set to zero based on
other parameters. At times when the vehicle was stationary, the engine was on, and the
accelerator pedal was not depressed, the net torque was set to zero.
Method 3 (J1939) - This is the method listed in Section 12.6.2.3 of the Oct 2008
SEMTECH-DS User Manual in which the derived torque is calculated as in Method 2,
except the "reference engine torque" (max engine torque provided in the ECU
datastream) is used instead of a lug curve. The derived torque is then converted to a
brake torque by subtracting out the friction torque (also calculated from ECU
parameters). This procedure was used for all J1939 protocol trucks. Specifically
Net Torquebrake = (Percent Torque - Friction Percent Torque) x Reference Torque
The method used to calculate Net Torquebrake from the ECU derived values for each truck was
chosen according to the Figure 6.1-1.
6-7
-------
Figure 6.1-1 Process for Calculating Brake Specific Power Output with ECU Data
Sourcing torque andBSFC curves: As mentioned above, engine manufacturer advertising
brochures were the primary source of torque and BSFC curves. ERG conducted Internet searches
to find brochures to match the different engines tested during the program. EPA assisted by
providing a number of brochures and torque curves from its own internal records. Torque and
BSFC curves were matched to the engines tested based on the engine tag information from each
engine. It is possible that an engine could have been reprogrammed to have a different rating
than it had originally, but there was no practical way to check for this so all curves were chosen
based on the observed engine tags. ERG also contacted both Caterpillar and Cummins in search
of lug curves that were more difficult to find, but there were three trucks for which curves could
6-8
-------
not be found. For these trucks, ERG has supplied the logged engine and exhaust emissions data
without any torque or VSP data.
The published torque curves of most engines consist of linear segments depending on
RPM. Most engines have flat, constant torque values at the middle of the RPM range, with a
linear decrease in torque at higher engine speeds. For this reason, ERG used torque tables for
each engine based on linear, piece-wise functions that were fit to the advertised curves.
Many of the published advertising curves do not plot the engine torque at low engine
speeds. Most of the available curves begin between 1000 and 1200 RPM and extend up to
maximum engine speed. Because the PEMS and PAMS data included points throughout the
entire engine speed range including these lower engine speeds, ERG made an assumption for the
shape of all curves at low engine speeds. Two torque curves were found in research literature
that included full sweeps of maximum torque over the complete range of engine speeds(4'5). For
both of these curves, it appeared that the maximum torque at idle speed was about half of the
maximum torque value for the engine, and increased continuously up to the maximum torque
value. For this reason, ERG applied a similar shape to the other curves which had published
curves that began at elevated engine speeds. The maximum idle-speed torque was taken to be
half of the maximum engine torque, with a linear increase up to the first point on the published
advertising curves. An example of this assumption, made for the Caterpillar 3406E engine rated
at 355 hp, is provided in the Figure 6.1-2. The solid blue line represents the torque as given by
the advertised torque curve, which does not contain information below 1200 RPM. The dashed
red line includes the extrapolation down to idle speed as performed by ERG at lower engine
speeds.
For the four PEMS tests conducted on vehicles with no ECU datastream (0181021-
A_20091216_E, 0181276-B_20091209_E, 0181276-B_20100125_E and 1191084-
A 20100128 E), BSFC emissions could be calculated by dividing the SEMTECH-DS' fuel
consumption rate and optically-measured RPM by the maximum fuel consumption rate obtained
from each engine's brake-specific fuel consumption (BSFC) vs. RPM curves. Multiplying this
ratio of "measured" fuel consumption (via PEMS) to maximum fuel consumption (from the lug
curve) by the maximum engine power output at that RPM for each observation would provide an
estimate of power for each second of data, which could be summed over the test to provide
overall brake-specific emission estimates. This was the methodology employed in the Nonroad
work assignment under this contract(6). However, no brake-specific fuel consumption vs. RPM
curves were identified which corresponded to the known characteristics of engines in the above
tests (these are the three trucks mentioned above for which no curves were available). Therefore,
for the above four tests, emissions are provided on a fuel and mileage basis, but not on a work
basis.
A list of which methodology was used for estimating BSFC emissions for each truck is provided
in Appendix AC, Drayage Processing Parameters. Appendix AD provides the torque curves
used in this study.
6-9
-------
Figure 6.1-2 Extrapolated lug curve for the Cat 3406E Engine Rated at 355 HP
Engine Speed {RPM)
Calculate Emissions. Using the brake-specific power estimates previously calculated, ERG
calculated mass-based brake-specific gaseous and PM emission estimates for all tests for which
lug curves were available (lug curves and fuel maps were not available for several tests, so
brake-specific emissions were not calculated for these tests). In addition, ERG also calculated
fuel-based and mileage-based gaseous and PM emission estimates. Estimates are presented for
the overall test, by filter, and also segregated by port/non-port and idle/non-idle operation, as
described below.
Designate Port and Idle Operation: In SAS, gaseous and PM emission estimates are provided
both within and outside of port terminals (based on GPS / ride-along notes). The terminals in
this study were initially identified by review of the filter log (Appendix T) and ride-along notes
(Appendix U). This review revealed four terminals in which operations were conducted. These
were Barbour's Cut Container Terminal, Bayport Container Terminal, Jacintoport General Cargo
Terminal, and Greens Port Industrial Park Terminal. Boundaries for these four areas were then
established as shown in Appendix AB, and all PEMS and PAMS data with GPS coordinates
falling within these boundaries was identified as "port" activity. In addition to port
identification, gaseous and PM emission estimates are distinguished as idle vs. non-idle
operation. Idle operation was determined by evaluating vehicle speed (under 1 mph), throttle
position (under 3%), and engine status (on). Filters collected during idle periods were identified
using filter data collection notes (Appendix T).
6-10
-------
Assign Emissions to MOVES Operational Bins: Using the brake-specific power estimates
previously calculated, ERG categorized emissions into MOVES operational bins according to
methodology provided in the Heavy-Duty MOVES Emission Rate Development
Documentation^.
The equation used to assign MOVES operational bins was:
VSP = T|driveline X ((Pbs)x0.746 - Pi oss,acc) /niavg,regclass
For our analysis, the following values were used:
r|driveiine, driveline efficiency, was estimated to be 90% for all trucks tested, based on
information in the guidance document.
Pbs, brake specific horsepower, was calculated as described above. Note that the factor of
0.746 is used to convert HP to kW.
Pioss,acc was set at 8 kW to account for total accessory loads, and is roughly equivalent to the
low HDT value provided in Table 6 of the reference document. This value was selected
based on conversations between ERG and EPA(2)
mavg,regciass, the average running weight value, was set to 17.1 metric tons, to fall within the
MHD to HHD range listed in Table 8 of the reference document. This value was selected
based on conversations between ERG and EPA(2)
Once a VSP value was calculated for each second of operation, an operating mode bin was
assigned according to the definitions in Table 6.1-1.
RSI) vs. PEMS emissions comparison - Using results in MOVES operational mode bins, ERG
compared truck emissions as measured during the July 2009 RSD study to PEMS results
collected for the same trucks.
6-11
-------
Table 6.1-1 MOVES HD Operating Mode Definitions (update)
Operating Mode
Operating Mode
Description
Vehicle-Specific
Power (VSPf)
(kW/ metric tonne)
Vehicle Speed
(v,mi/hr)
Vehicle
Acceleration
(gs, i.e., acc/gravity)
1
Idle
-1.0 < v < 1.0
t
11
Coast
VSP<0
t
1
-------
Perform date and time assignments and corrections - All date and times were reviewed and
corrected, as necessary. For example, the PAMS units would not correctly log the time at
exactly midnight and therefore corrections for all midnight times were applied. There were no
indications of issues related to setting the logger time during installation. There were, however,
six trucks that had loggers recording during the spring shift to Daylight Saving Time. To
compensate for this, ERG increased the time by one hour for all observations that were recorded
after 2:00 AM on March 14, 2010. This compensation was performed for the trucks with these
license plates: 0182003-A, 0182022-A, 0183438-A, 0183439-A, 0190758-A, and 1075087-A.
Time-align data streams. ERG performed time alignment of the data where necessary. For the
PAMS units, the only signal requiring alignment was the GPS speed. The time lag was
calculated by plotting the GPS speed and the ECU speed over time and finding the average lag
between the maxima and minima of the two plots. This value was found to typically be two
seconds, so all of the GPS speed data was aligned two seconds earlier than originally recorded.
Correct torque output spikes. ERG identified and modified all tests with ECU output torque
"spikes" identified during PAMS analysis. These spikes refer to unreasonable torque values
which were most often identified from the Detroit Diesel and Mercedes Benz engine datastreams
while in fuel shut-off mode. These spikes appear to be associated with the way the Isaac
datalogger processed the ECU datastream rather than the actual datastream from the ECU.
Because these spikes occurred during fuel cutoff or at idle, the (net) output torque during these
episodes was therefore set to zero in SAS. For the Mercedes-Benz trucks using the J1939
protocol, there were also spikes observed in the value for reference torque, which is the ECU
value for the maximum cylinder torque of the engine. These spikes resulted in unreasonably high
values for the engines' maximum torque (approximately 65,000 newton-meters, or 48,000 ft-
lbs), but would usually last only a few seconds. During these observed spikes, ERG set the
reference torque to the previous reasonable value. It is likely that these observations were
artifacts of the operation of the Isaac dataloggers as well.
Correct other ECU output data as required. ERG also performed all other necessary
modifications of ECU torque or load values as necessary in order to calculate brake-specific
power output values for MOVES classifications. Many of these corrections were necessary
because of a scaling factor that is built into the PAMS units' J1708 protocol setup. The PAMS
data required a scaling factor of 50 to calculate the true torque values, and a scaling factor of
1.25 to calculate the true percent load values for all J1708 protocol trucks. ERG determined these
factors by comparing the observed resolution in the data to the resolutions given in the J1708
protocol specification. Also, some trucks had brief RPM spikes that took place when the engine
otherwise appeared to be off. These isolated spikes were not considered trips or engine-on time if
the engine never reached its typical idle speed. Whenever a datafield was corrected, a new field
was created (and the original data was retained).
An example of the percent engine load data from a J1708 protocol truck is provided below. The
figure is representative of typical data recorded by the PAMS units after scaling to correct for the
PAMS unit's J1708 setup.
6-13
-------
Figure 6.2-1 ECU Engine Percent Load vs. Speed for a J1708 protocol truck
ECU Percent Engine Load vs Engine Speed - 2CS498
Three areas of interest are highlighted in the above figure. The dark spikes shown in Area 1
occur during cold start idle operations. When the engine is first started, the ECU holds an
elevated RPM so that the engine can run smoothly. The engine percent load as given by the ECU
is initially at an elevated value as well, and it decreases as the engine warms up. This is because
the friction torque of the engine is higher when cold. Area 2 is the region representing typical
idle and near-idle operation. As with Area 1, the engine experiences higher percent load values
during cold operation, even at idle. This results in overestimation of engine torque when the
engine is cold, because the ECU percent load is elevated even though the net output torque is
zero at idle. Area 3 indicates percent load values that exceed 100% at high engine speeds. This
was observed for a number of the trucks. Typically, only a very few values at high RPM
exceeded 100% load such as in the figure, though for one truck a large number of values
exceeded 100%. The method of accounting for that vehicle's engine load is described in
Appendix AE. For all other trucks, no correction was applied and there were a few points that
exceeded the maximum torque as given by the lug curves. It is not known why many of the
ECUs represented load in this way.
Merge relevant PEMS Data: For trucks on which the PAMS was disconnected to perform
PEMS testing, ERG merged the PEMS "activity" data into the PAMS data file during the PEMS
test day when the PAMS was disconnected. The following table presents the trucks for which
PEMS data was merged, along with the corresponding dates that the PEMS test took place.
6-14
-------
Truck
Date(s)
0178515-B
12/10/09,12/11/09
0183710-A
12/15/09
0185067-A
2/5/10
0185576-A
2/4/10
1095729-B
1/28/10
1095730-A
2/1/10, 2/2/10
1190901-A
2/8/10, 2/9/10
1191083-A
2/1/10, 2/2/10
0182003-A
3/9/10, 3/10/10
0182022-A
3/8/10
0183716-A
1/25/10
0183718-A
12/14/09
Removal of engine-off data: The PAMS units were set to record all data continuously,
irrespective of whether the engine was on or off. During analysis, only observations that
occurred when the key was in the 'on' position were kept for analysis. Depending on the
protocol, certain ECU values were used to indicate key on/off, and engine RPM was used to flag
whether the engine was on or off. As a QC check, if GPS data indicated that a truck was moving
even though the ECU or engine appeared to be off, this data was kept and flagged as having lost
ECU data.
Assign Activity to MOVES Operational Bins: ERG categorized each second of operation into
MOVES operational bins, using ECU data for power determination (and external lug curves
when needed), according to the methodology described in Section 6.1.2. When ECU data was
not available, no surrogates to power were available, so MOVES classifications were not
performed.
Optical RPM Only one of the 23 trucks which received a PAMS instrumentation collected
engine RPM optically (this was truck 0181276-B, a 1980 Kenworth). For this truck, activity data
is not presented on a VSP basis, as no indicator of load was available.
Designate Port and Idle Operation. As with the PEMS data analysis described in Section 6.1.2,
ERG distinguished activity both within and outside of the four primary port terminals in the
study (Barbour's Cut Container Terminal, Bayport Container Terminal, Jacintoport General
Cargo Terminal and Greens Port Industrial Park Terminal). In addition, idle activity is
distinguished from non-idle activity in the PAMS data, with summaries provided in this report.
An engine was flagged to be at idle when the GPS speed was zero, the pedal position was at its
unpressed base value, and the engine was on. Net torque values were set to zero if the idle flag
was set regardless of whether they were logged as positive or negative values. This was
performed to reduce the bias in total work output that would be caused by only setting negative
torque values to zero at idle.
Assignment of trips and trip counts: In SAS, counts of trips (and number of observations for
each trip) were assigned for each test file. Trips were defined as episodes of engine operation.
6-15
-------
Trip counts were the total number of "trips", separated by engine off periods. In some cases,
mostly during very long trips, the PAMS unit would not record at the intended one hertz
frequency and some data appeared missing. If there was no log of engine off time, the trip count
was not increased even if there was a large amount of missing data. The first observation after a
period of missing data (longer than 5 seconds) was flagged to indicate data was lost. Most engine
start events upset the voltage to the PAMS loggers such that recording was briefly interrupted.
These interruptions that occurred for all trucks for a few seconds at engine start were not flagged
as missing data. Trip assignments are included in the data submission which will accompany
this report.
Other QC Checks: After all of the above corrections were applied to the data, checks were
performed to ensure the final data was reasonable. ERG compared the maximum torque and
horsepower values from the data to those given on the engine tags where available. Also, ranges
of values for each recorded variable were applied, and SAS was used to flag any data that fell
outside these ranges. Based on observations in the final QC checks, ERG applied further
corrections to the data recorded for a small number of the trucks. These truck-specific
corrections are described in Appendix AE.
6-16
-------
7.0 Study Results and Conclusions
7.1 Emissions results
The following subsections provide results from PEMS particulate matter and gaseous
emissions measurements collected throughout the study. Emission results are provided in units
of emissions per work performed and emissions per mile traveled. Idle emission rates are
presented on a time basis (grams/second). Results of gravimetric measurements of dynamic and
field blanks collected during the field study are also presented. A more thorough compilation of
emission results is provided in Appendix M, Drayage Emissions Measurement Results.
7.1.1 PM Filter Weights
EPA supplied pre-weighed 47 mm Teflon filters in pre-loaded URG-2000-30FL filter
cassettes. After PM collection, filter samples were transported to the EPA laboratory in the URG
filter cassettes for post-test gravimetric measurements. Resultant data was provided to ERG on a
total mass per filter basis (i.e., mg/filter).
Gravimetric sample particulate measurements were collected in accordance with
guidelines provided various field SOPs described in Section 5.1. Weights of all gravimetric
filters collected throughout the study are provided in Appendix T, the Drayage Filter Log.
7.1.1.1 Dynamic and Field Blanks
As described in Section 5.1, dynamic and field blanks were collected throughout the
study to quantify the effect of handling and system contamination on the gravimetric filters
collected during the study. Table 7.1-1 provides the laboratory measurement results from
dynamic and field blanks collected during the study. These results (and additional information
pertaining to collected filters) are also included in the Drayage Filter Log, Appendix T.
7-1
-------
Table 7.1-1 Dynamic and Field Blank Measurement Results
Date used
Filter ID
Test ID
Sample Type
Net Wt (mg)
0181276-
12/9/09
Q-8
B E 1
Dyn Blnk
0.0422
12/15/09
9047499
N/A
field blank
-0.154
1/28/10
9051848
N/A
field blank
0.0131
1/28/10
9051849
N/A
field blank
0.0133
1/28/10
9051850
N/A
field blank
0.0061
0184781-
2/3/10
9051921
A E 1
Dyn Blnk
0.0199
0184781-
2/4/10
9048079
A E 2
Dyn Blnk
0.0427
2/9/10
9051880
1190901-A-E-2
Dyn Blnk
0.0023
0190306-
3/12/10
9055038
B E 1
Dyn Blnk
0.0207
0181157-
3/17/10
9054999
A E 1
Dyn Blnk
0.0251
3/17/10
9052421
N/A
field blank
0.0174
7.1.2 Summary of Gaseous and PM Emission Results
Figures 7.1-1 through 7.1-32 present PEMS emissions (on a time-basis) within MOVES
operating mode bins (aka VSP bins). Emissions are presented by the model-year bins that were
used for sample stratification. Emissions are further segregated as port and non-port. As
explained in Section 6.1.2, work-based emissions were not calculated for those trucks for which
lug curves were not available. These results are not presented in Figures 7.1-1 through 7.1-31.
"Suspect" data has been excluded from the values shown in Figures 7.1-1 through 7.1-31.
7-2
-------
Figure 7.1-1 HC Emissions for 1978-1993 Trucks by VSP BIN, Port
o
X
0
CD
2
0
>
<
0025
0.024
0.023
0022
0.021
0.020
0.019
0.018
0.017
0.016
0 015
0 014
0.013
0.012
0 011
0.010
0.009
0008
0.007
0.006
0.005 "
0.004 -
\
VSP Bin frequency and emissions - HC
modelyrgrp=1978-1993 N=1 corrPortflag=PORT
<25 mph --
< 25 - 50 mphi
!
V">
>50 mph >
? ? r ? f # ? ? ? ? p f f ;§> <§>¦ i? p f $
< 25 - 50 mph —
>
•
< — >50 mph — >
90
0.014
0.013
?
3
!5
1
i
i
p
1
•
•
•
80
y 0.012-
0)
w
^ 0011"
1
1
-70
•
•
60
Q 0.010-
X
•
"50
(1) 0.009
•
•
cn
2 0.008
<1)
^ 0.007
A
/ \
•
•
"40
30
0.006
0.005
/
j
i
>
•
20
0.004
0.003
1
i
*
•
\
V-—
\
•
^ \
\
V
10
0
j9 $ ? ? ? ? p $ ? /* /• ? ? f f ? # ? f f $ * $
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
PLOT • • • hcmean
7-3
-------
Figure 7.1-3 HC Emissions for 1994-1997 Trucks by VSP BIN, Port
VSP Bin frequency and emissions - HC
modelyrgrp-1994-1997 N=9 corrPortflag=PORT
c
o
CD
Q_
O
a;
a
sz
cd
>
CD
2
a>
Cl
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
Figure 7.1-4 HC Emissions for 1994-1997 Trucks by VSP BIN, Non-port
VSP Bin frequency and emissions - HC
modelyrgrp=1994-199/ N=9 corrPortflag=OTHER
0.052
0.050
0.048
0.046
0.044
0.042
0.040
O 0.038
CD
< 25 - 50 mph > <— >50 mph —>
\
i-.
<>) 03 Q>
«S '
>s> fg- <§> <§>
c
o
CO
i_
CD
Q_
O
c
CD
O
i_
CD
CL
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
PLOT • • • hcmean
7-4
-------
Figure 7.1-5 HC Emissions for 1998-2003 Trucks by VSP BIN, Port
VSP Bin frequency and emissions - HC
mo del yrgrp=1998-2003 N=28 corrPortfiag=PORT
o
0
CD
O
X
0
O)
CO
s_
0
>
<
0.028
0.027
0.026
0.025
0.024
0.023
0.022
0.021
0.020
0.019
0.018
0.017
0.016
0.015
0.014
0.013
0.012
0.011
0.010
0.009
0.008
0.007
0.006
0.005
0.004
0.003
l
< <25 mph >
\ •
I
25 - 50 mph >
< — >50 mph — >
# #¦ i?
.Or
/ -1 ^
»! «5 »
/ / /
O- <>>• <6"
°>' &
f <§> <9'
&>¦ nV
f <§
¥ 4
o
"S
s_
CD
Q.
O
0
O
'sz
0
>
0
O
0
0_
VSP (kW/tonne) Bin
PL0T2 Perce nt of Tota I F req uen cy
PLOT • • • hcmean
Figure 7.1-6 HC Emissions for 1998-2003 Trucks by VSP BIN, Non-port
VSP Bin frequency and emissions - HC
modelyrgrp= 1998-2003 N=28 corrPortflag=OTHER
0.016-
0.015
0.014
0.013
0.012
o
0
•
< 25 — 50 mph
>
< — >50 mph — >
I I
1
1
•
•
•
A
1 \ •
/ \
l \
•
•
•
•
•
• •
I r
•
•
/
/
/
•
•
,
c
o
0
Q_
O
0
O
JZ
0
>
c
30 0
o
0
Q.
# £ ? ? ? f * i?' ? T ? f f f
O <0 1 O- o>-
-------
Figure 7.1-7 HC Emissions for 2004-2006 Trucks by VSP BIN, Port
VSP Bin frequency and emissions - HC
modelyrgrp-2004-2006 N=5 corrPortflag=PORT
0.014
0.013-
0.012 -
0.011 -
O 0.010
CD
< 25-50 mph >
\ •
- &
< — >50 mph — >
^
c
o
CD
Q.
O
0
O
JZ
(D
>
C
0
o
CD
Q_
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
PLOT • • • hcmean
Figure 7.1-8 HC Emissions for 2004-2006 Trucks by VSP BIN, Non-port
0.006 - fj
O
CD 0.005
CO
O
X 0.004
CD
O)
2
CD
> 0.003
<
VSP Bin frequency and emissions - HC
modelyrgrp=2004-2006 N=5 corrPortflag=OTHER
<25 mph
>50 mph
/'" ? ? f * p P T ? f> f ft f * ? f ? p f i
i ° ^ ^ «• * J " •' #' 4
C
O
CD
Q_
O
0
O
JZ
0
>
£Z
CD
O
i_
0
CL
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
PLOT • • • hcmean
7-6
-------
Figure 7.1-9 CO Emissions for 1978-1993 Trucks by VSR BIN, Port
VSP Bin frequency and emissions - CO
modelyrgrp=1978-1993 N=1 c<
O
CD
co
O
o
CD
O) 0.3
CO
i_
CD
>
50 mph
O
"S
CD
Q_
O
_CD
O
H
CD
>
CD
O
i_
CD
CL
<»>
o, .g, & * «, <» # # ^ $ $ p f> f <
" b v *' • J J -1 ® ~ «
t?" t§> <§>
/ / -1
Figure 7.1-10
O
0
(/)
O
O
CD
O) 0.3
2
©
>
25-50 mph >
< >50 mph >
C
o
CD
i_
CD
Q.
O
0
O
iZ
CD
>
c
(D
O
CD
D_
# i? ? r ? p f ? ? r ? f> f * f <§>¦ ? f f $ #
/ ° • 4 at »¦ v " * 4 tf ^ -1
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
PLOT • • • comean
7-7
-------
Figure 7.1-11
CO Emissions for 1994-1997 Trucks by VSR BIN, Port
VSP Bin frequency and emissions - CO
modelyrgrp=1994-1997 N=9 corrPortfIag=PORT
O
CD
CO
O) 1
0.17
0.16
0.15
0.14
0.13
0.12
0.11
o a10
O 0.09
a> 0 08
03
4— 0.07
0
5 0.06
<
0.05
0.04
0.03
0.02
A
D •
<25 mph
I •
25-50 mph
>50 mph
4f
J 1 ^
& ®
Sf #
/ /
f f f f ft ;§> f
C
o
CO
i_
CD
Q.
O
0
a
CD
>
CD
O
CD
Q_
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
Figure 7.1-12
0.31:
0.30
0.29
0.28
0.27
0.26
0.25
0.24
0.23
O 0.22
0.21
0.20
019
" ' 0.18
O 0.17
Q 0.16
0 015
O) 014
03 0 1 3
0 012
> 0.11
I
CO Emissions for 1994-1997 Trucks by VSP BIN, Non-port
VSP Bin frequency and emissions - CO
modelyrgrp=1994-1997 N=9 corrPortflag=OTHER
/ *
<25 mph
O 50 mph
& c>-
J A v
? ? ?
o-
c
0
o
0
0_
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
PLOT • • • comean
7-8
-------
Figure 7.1-13
CO Emissions for 1998-2003 Trucks by VSP BIN, Port
VSP Bin frequency and emissions - CO
modelyrgrp=1998-2003 N=28 corrPortflag=PORT
c
o
CD
Q.
O
0)
O
CD
>
c
CD
o
1—
CD
CL
c> ?> ,«¦ <§> ,&& s> & <$¦ ,
- " " - - - I J " - J J
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
PLOT • • • comean
Figure 7.1-14
CO Emissions for 1998-2003 Trucks by VSP BIN, Non-port
0.19
0.18
0.17
0.16
0.15
o-014
g> o.i3
3 a12
o 011
O o.io
0.09
O)
2 0.08
CD
> 0.07
^ 0.06
0.05
0.04
0.03
0.02
0.01
i\
l s
U •
£
VSP Bin frequency and emissions - CO
modelyrgrp= 1998-2003 N=28 corrPortflag=OTHER
<— <25 mph —>
25-50 mph >
< — >50 mph — >
i? ? T ? p f $ f <§>'
CN- Oc f>v ' ' / / A
% c§>
C
o
CO
1—
CD
Q_
O
0)
O
SZ
CD
>
c
CD
o
u
CD
~_
VSP (kW/tonne) Bin
f2 Percent of Total Frequency
7-9
-------
Figure 7.1-15
CO Emissions for 2004-2006 Trucks by VSR BIN, Port
o
o
CD
CT)
CD
i—
CD
>
<
0.28"
0.27"
0.26"
0.25"
O)
0.24-
0.23"
1?
0.22"
n
0.21
8
0.20-
i
0.19-
i
0.18-
0.17-
0.16
0.15
0.14
0.13
0.12
0.11
0.10
0.08-
0.07
i
0.05
1
l
0.04
0.02-
»
0 01 -
A
i \
jt
VSP Bin frequency and emissions - CO
modelyrgrp=2004-2006 N=5 corrPortflag=PGRT
<25 mph
25-50 mph
oj -
9?
>50 mph
/ ? #
* / / si
O
i
CD
s_
CD
Q_
O
C
30 CD
CD
Q_
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
PLOT • • • cornea n
Figure 7.1-16
CO Emissions for 2004-2006 Trucks by VSP BIN, Non-port
O
CD
CO
o
o
CD
CD
i—
CD
>
<
0.13
0.12
1
1
CJJ |
0.11
¦Z 1
m ®
8 2
a |
i
i
0.08"
0.07"
A
0.06-
i
0.05
i
0.04
1
1
0.03
1
I •
i
0.02
I
1
0.01
I
VSP Bin frequency and emissions - CO
modelyrgrp=2004-2G06 N=5 corrPortflag=OTHER
< — <25 mph — >
• •
< 25 - 50 mph > <— >50 mph —>
cb
c
o
CD
Q_
O
jD
O
in
CD
>
C
CD
y
CD
CL
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
PLOT • • • comean
7-10
-------
Figure 7.1-17 CO2 Emissions for 1978-1993 Trucks by VSP BIN, Port
VSP Bin frequency and emissions - C02
modelyrgrp=1978-1993 N=1 corrPortflag=PORT
O
<
/\
/ \
! \
< — <25 mph
< 25-50 mph
1 / \
9\
<>> C&
c/ <4 50 mph >
100
90
C
O
03
70
i—
CD
Q_
O
60
CD
O
50
'sz
CD
>
40
4—
O
c
30
CD
O
CD
20
CL
10
& &
/ y\ V
.«>¦ s> «> p .sy
f f f j j
• <&•
» «*•
*
¦5' ^
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
PLOT • • • co2mean
7-11
-------
Figure 7,1-18
O
0
C/> 40
C\l
830
CD
O)
03
0 20
>
<
» •
CO2 Emissions for 1978-1993 Trucks by VSR BIN, Non-port
VSP Bin frequency and emissions - C02
modelyrgrp=1978-1993 N=1 corrPortflag=OTHER
<25 mph — i
< 25-50 ri^h > < >50 mph >
O
CO
CD
CD
Q_
4? P ? J° ? f f ? ? ? * * f <§>•
0 * * +' " « * * or' */ ^ «
*> ~ *>'
c
0
o
I—
0
CL
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
PLOT • • • co2mean
7-12
-------
Figure 7.1-20
CO2 Emissions for 1994-1997 Trucks by VSP BIN, Non-port
O
0
CO 40
CN
o 30
0
CO
CD
i_
CD 20
>
<
» •
VSP Bin frequency and emissions - C02
madelyrgrp=1994-1997 N=9 corrPortflag=OTHER
<— <25 mph —>
25 - 50 mph >
<— >50 mph —>
1? r ? f> * ? ? ? ? f $ ¦§> <§>• ? f f <
0 * * "> *>'
40
0
+J
c
30
0
O
l—
0
20
CL
-10
-0
/ / f
® eg"
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
PLOT • • • co2mean
Figure 7.1-21
CO2 Emissions for 1998-2003 Trucks by VSP BIN, Port
VSP Bin frequency and emissions - C02
modelyrgrp=1998-2003 N=28 corrPortflag=PORT
O
CD
if)
CM
o
o
<
£ 1
CQ £
< — <25 mph —>
< 25-50 mph > < —. >50 mph >
c
O
2?
0
Q.
o
0
o
JZ
0
>
c
0
O
0
Cl
•I? I- / / /
O ^ a, c,. <5 © ® ^ ^ £ a # ^ ,§> #
4 ~ - - ' ' " »' V "
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
PLOT • • • co2mean
7-13
-------
Figure 7,1-22 CO2 Emissions for 1998-2003 Trucks by VSR BIN, Non-port
VSP Bin frequency and emissions - C02
modelyrgrp= 1998-2003 N=28 corrPortflag=OTHER
O
CD
C/) 40
C\J
8 ®
CD
O)
CD
0 20
>
<
]
<— <25 mph —>
< 25 - 50 mph >
< — >50 mph — >
£ 1
jZ 1
•
1 .
•
8 f
•
•
3 ;
•
1
•
•
A
•
/
•
A
•
A
•
•
/
•
/
/
\
•
1
t •
1
\
• \
•
1
D •
X
C
o
CD
Q_
O
JD
O
JZ
CD
>
C
CD
O
0
0_
# # j? ?> /> * f * j* ? ? ? f # $ <§> <§>'
f 0 * * ^ ^ ^ 1
£ * * <§> <§>
•' j "
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
Figure 7.1-23
CO2 Emissions for 2004-2006 Trucks by VSP BIN, Port
VSP Bin frequency and emissions - C02
modelyrgrp=2004-2006 N=5 corrPortflag=PORT
O
<
A
f\
I \
<— <25 mph —>
25 - 50 mph >
< — >50 mph — >
Q t>3 <0 CQ
k J J J
O * 41 *• ay »¦ # *
C
O
CO
i_
CD
Q_
O
0)
O
H
CD
>
c
30 CD
O
CD
CL
^ $
-
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
PLOT • • • co2mean
7-14
-------
Figure 7,1-24
O 50
< s
A
/ \
CO2 Emissions for 2004-2006 Trucks by VSP BIN, Non-port
VSP Bin frequency and emissions - C02
modelyrgrp=2004-2006 N=5 corrPortflag=OTHER
<25 mph
25-50 mph
>50 mph
$ ? ? f T f * F ? T ? f f $ f «F
o
C
30 0)
o
CD
20 Q_
!? * * # ^ «
" » S><*
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
PLOT • • • co2mean
Figure 7.1-25
NOx Emissions for 1978-1993 Trucks by VSP BIN, Port
VSP Bin frequency and emissions - NOx
modeIyrgrp=1978-1993 N=1 corrPortflag=PORT
X
O
<
k
i \
<25 mph —>
/ \
< 25-50 mph
< — >50 mph — >
<* & Of
f f " / / J
4
VSP (kW/tonne) Bin
PLOT2 Percent of Total Frequency
PLOT • • • noxmean
7-15
-------
Figure 7,1-26
NOx Emissions for 1978-1993 Trucks by VSP BIN, Non-port
X
O
a>
CD
CD
5_
CD
>
<
0.08
0.06
0.04
0.02
A
VSP Bin frequency and emissions - NOx
modelyrgrp=1978-1993 N=1 corrPortflag=OTHER
<25 mph —>
25-50 mph
"3 03 Q)
O fe
/ / /
O- -
>50 mph
$ # ¦§>•
/ / -i
^ |