A Study of Emissions from Yard Tractors Using Diesel and LNG Fuels

Transcription

1 A Study of Emissions from Yard Tractors Using Diesel and LNG Fuels Final Report August 28, 2007 Prepared for: Mr. Brad Rutledge Senior Project Manager WestStart-CALSTART 48 South Chester Ave Pasadena, CA Dr. J. Wayne Miller University of California, Riverside College of Engineering-Center for Environmental Research and Technology Riverside, CA (951)

2 Disclaimer Study of Emissions from Yard Tractors with Diesel and LNG Fuels This report was prepared as the result of work sponsored by WestStart-CALSTART and the Port of Long Beach and as such does not necessarily represent the views of WestStart-CALSTART or the Port of Long Beach, or its employees. WestStart- CALSTART and the Port of Long Beach, its employees, contractors and subcontractors make no warrant, express or implied, and assume no legal liability for the information in this report; nor does any party represent that the uses of this information will not infringe upon privately owned rights. This report has neither been approved nor disapproved by WestStart-CALSTART and the Port of Long Beach. Further neither WestStart- CALSTART nor the Port of Long Beach passed upon the accuracy or adequacy of the information in this report. ii

3 Acknowledgements The author expresses appreciation to the following associates who contributed much to the success of the project and the furtherance of knowledge of emissions from cargo handling equipment at the Long Beach Container Terminals (LBCT) with help from WestStart-CALSTART. We very much appreciated the financial support of the Port of Long Beach throughout the project and the assistance of WestStart-CALSTART in managing the selection and acquisition of the yard tractors and for conveying the contract. WestStart-CALSTART Mr. Brad Rutledge Port of Long Beach Mr. Thomas A. Jelenic Mr. Matthew Arms Long Beach Container Terminal Mr. Charles Doucette Mr. Anthony Otto University of California, Riverside Mr. Kent Johnson Mr. Don Pacocha iii

4 Table of Contents Disclaimer... ii Acknowledgements...iii Table of Contents... iv Table of Figures... vi Table of Tables... vii Executive Summary Introduction Cargo Handling Equipment (CHE) Yard Trucks -General Yard Tractors ARB Survey Emissions Standards for Yard Tractors Emission Test Methods for Yard Tractors Dynamometer Testing Experimental Work Plan Choosing the Right Yard Tractors to Test Test Driving/Operating Schedule Measurement of Gas Concentration and Flow Rates Measurement of Particulate Mass (PM 2.5 ) Quality Assurance and Quality Control Requirements Results Yard Tractors Tested and Protocol for Engine Maps Test of Mechanically Controlled Engines at Tier 1 & 2 Off-Road Standards Test of an Electronically Controlled Engine at 2004 On-Road Standards Test of Electronically Controlled Engine Using LNG fuel Measured Emissions -- Modal Data Calculating Overall Emission Factors from Modal Data Weighted Emission Factors for Engines in LBCT Yard Tractors Comparison of Emission Factors: This and Earlier Studies with Standards Deeper Evaluation of Emissions Measured for 2001 C8.3L Engines Deeper Evaluation of Values Measured for a 2003 C8.3L Engine Deeper Evaluation of Emissions Measured for ISB5.9L Engines Deeper Evaluation of Emissions Measured for CG250 Engines Added Analysis: Limited Analysis of ECM Data from the ISB Engine Findings and Recommendations Appendix A Description of UCR s Mobile Lab and Selected QA/QC Procedures Appendix B iv

5 Quality Assurance Project Plan Appendix C Raw Test Data for Various Yard Tractors Selected Test Values for C8.3L-2003 (Tier 2) & C8.3L-2001 (Tier 1) Engines Selected Test Values for ISB-5.9L-2005 & CG-250 LNG Engines Appendix D Selected ECM Data from the ISB Engine v

6 Table of Figures Figure ES-0-1 Weighted Emission Rates (grams/wheel-hp-hr) for Yard Tractors with Various Diesel Engine Technologies and a Cummins-Westport LNG Engine a) NOx & b) PM... 1 Figure 1-1 Percentage of Regional NO x & PM 10 Emissions by Source Category (ref Hahn)... 4 Figure 1-2 NOx and Diesel PM Emission Distributions for Cargo Handling Equipment at California Ports (ref ARB)... 5 Figure 1-3 Cummins Engine Breakdown by Model Year and Engine (ref. ARB 5 )... 7 Figure 1-4 Torque scales and Location of Intermediate and Rated Speed Points Figure 1-5 Schematic Shows the Most Common Water Brake, the Variable Level Type Figure 2-1 Yard Tractor Being Tested on a Chassis Dynamometer and UCR s Mobile Lab Figure 2-2 Major Systems within UCR s Mobile Emission Lab (MEL) Figure 2-3 Summary of gas-phase instrumentation in MEL Figure 3-1 (a) Matrix of Test Points for the C8.3L Engine in the Yard Tractor and (b) Data from Cummins Brochure Figure 3-2 Matrix of Test Points for a Cummins C8.3L Engine in the Yard Tractor and (b) Data from Cummins Brochure Figure 3-3 (a) Engine Map from the Chassis Dynamometer & (b) from Cummins ISB Brochure Figure 3-4 Plot of Measured Wheel hp vs. %ECM-load for ISB Engine Figure 3-5 (a) Engine Map from the Chassis Dynamometer & (b) from Cummins Brochure Figure 3-6 Plot of Measured Wheel hp vs. %ECM-load for LNG Unit Figure 3-7 Modal Data in g/whp-hr for a) C/2001 (Tier 1) and b) C/2003 (Tier 2) Figure 3-8 Modal Data in g/whp-hr for a) C/2003 (Tier 2) and b) ISB/2005 (Onroad/2004) Note difference in scale Figure 3-9 Modal Data in g/whp-hr for a) ISB/2005 (On-road) and b) CG250/2005 (Onroad) Figure 3-10 Plots of the Weighted Emission Factor (g/whp-hr) for a) NOx and b) PM.. 27 Figure 3-11 Plot of the Broadcast Fuel Rate and the Measured CO 2 Rate Figure 3-12 Comparative Data for the ISB-2004 and ISB Figure 3-13 Comparative Data for the CG250 from the POLA and POLB Tests Figure 4-1 Weighted Emission Rates (grams/wheel-hp-hr) for Yard Tractors with Various Diesel Engine Technologies and a Cummins-Westport LNG Engine a) NOx & b) PM 34 vi

7 Table of Tables Table 1-1 Regional Emissions in 2001 by Source Category for the Port of Los Angeles, tpy (ref Starcrest)... 3 Table 1-2 NOx & PM Emissions by Major Source Category, tons per year... 4 Table 1-3 Estimated Statewide Cargo Handling Equipment Populations and Emissions (ref 4 ARB)... 5 Table 1-4 Diesel Engine Manufacturer for Yard Tractors (ref. ARB 5 )... 6 Table 1-5 Selected EPA Non-road diesel engine emission standards in g/bkw hr (g/bhphr)... 7 Table 1-6 Test Modes, Torque and Weighting Factors for the ISO-8178-C1 Cycle... 8 Table 2-1 Slate of Yard Tractors and Fuels Selected for Testing Table 3-1 List of the Engines Tested and Their Properties Table 3-2 Weighted Emission Rates (g/whp-hr) for Four Engines Tested Table 3-3 Compilation of a) Measured Values in g/whp-hr & b) EPA Standards in g/bhphr Table 3-4 Comparative C1 Test Points for the POLB & POLA Tests Table 3-5 Comparative Emission Values for Tests of Two C8.3L Engines Table 3-6 Comparative Results for the C1 Cycles Used for the Two Engines Table 3-7 Comparative Results for ISB Engines from POLB & POLA Table 3-8 Comparative C1 Test Points for Two LNG Engines Table 3-9 Comparative Weighted Emission Factors for Two LNG Tests Table 3-10 Selected ECM Data from the ISB Engine Table 3-11 Emission Rates at Idle for the Various Units Table 4-1 List of Test Engines & Properties vii

8 Executive Summary Yard tractors are the workhorse of the cargo handling equipment (CHE) and there are over 1,500 in the Ports of Los Angeles and Long Beach. Because of their high numbers and frequent use, they also are the primary contributor to the emissions inventory from cargo handling equipment. Over 50% of the NOx and PM emissions associated with CHEs come from yard tractors. Based on these facts the POLB, LBCT and WestStart-CALSTART launched an investigation of the emission factors from yard tractors. There were two goals: first, measure the emission factors from a series of engines certified to ever increasingly more stringent emission standards, and second, investigate the emissions from an engine running on an alternative fuel, LNG. As direct emissions values are scarce, the report compares the results with earlier studies and EPA certification values. A number of yard tractors with various engine technologies meeting different levels of EPA emission standards were part of the planned project as shown in the table below. The test matrix represented engines meeting Tier 1 and 2 non-road (industrial) standards, on-road (automotive) standards for 2004 and an engine fueled with LNG. Table ES-1 Matrix of Yard Tractors Selected for Testing with CARB & LNG Fuels The engines were tested in their yard tractors on a heavy-duty chassis dynamometer. Testing included: first, developing an engine map to establish a C1 ISO test cycle, and second, measuring the emissions at various C1 test points in duplicate. Both modal and overall emission factors were calculated from the test data. The weighted emission factors for NOx and PM are shown in the figure below for the various engines. About 75 emissions tests were carried out during the project. Figure ES-0-1 Weighted Emission Rates (grams/wheel-hp-hr) for Yard Tractors with Various Diesel Engine Technologies and a Cummins-Westport LNG Engine a) NOx & b) PM Results for the diesel fueled engines showed the weighted emission factors on grams per wheel horsepower-hour for NO x and PM decreased with engines manufactured to meet stricter EPA certification standards. Estimates of the emission factors on a brake horsepower basis 1

9 indicate the values were within the EPA specifications for the C-2003 and ISB engines but not for the C-2001 engine. Unlike the earlier POLA testing, Donaldson control technology was used in the LBCT tests and reduced the CO, THC and PM emissions on the C-8.3L engines but unexpectedly did not reduce emissions on the ISB engine. Several issues were discussed in the analysis section. One issue was the C-2001 engine map in that the C1 load points were about 50% of Cummins engine map instead of the expected 75%. As detailed in the analysis, we believe the emissions factors are still valid. Another issue was the results comparing the ISB engines from 2004 and The 2005/LBCT engine had about 20% higher NO x emissions than the earlier POLA test unit and the reason for the difference was not clear. Several hypotheses were presented in the discussion. Finally, the LNG fueling system from the LBCT yard tractor was modified prior to testing and did not generate the fueling errors seen during the POLA test. Consequently the LBCT yard tractor had lower NOx emissions than the earlier POLA test. One frontier examined in this work was the activity data for the ISB engine. These data revealed that the engines spend considerable time in the idle mode. The use of the engine downloads appears to be useful information about how a yard tractor operates. Not examined in this study was the measurement of the in-use activity and emissions from yard tractors with portable instruments during actual use of moving containers. Obtaining those values is planned and those values should be very helpful in finalizing the most appropriate emission factors to use for yard tractors. 2

10 1 Introduction The Ports of Los Angeles and Long Beach are working with the California Air Resources Board (CARB) and the South Coast Air Quality Management District to define the total and source emissions associated with port operations. Working towards that goal, the Ports of Los Angeles and Long Beach separately contracted the Starcrest Consulting Group, LLC 1,2 to determine the inventory contribution from various sources and the total inventory. Port emissions are a significant element in the inventory in the South Coast Air Basin. As the South Coast Basin is a nonattainment area for air quality, substantial efforts 3 are underway to mitigate the effect of increased ship traffic that is anticipated for the ports. An example of the regional emissions reported in the Starcrest study for the Port of Los Angeles is shown in Table 1 below. Table 1-1 Regional Emissions in 2001 by Source Category for the Port of Los Angeles, tpy (ref Starcrest) Another perspective of the same inventory can be viewed by segmenting the emissions to either port or regional as seen from the Hahn study 3 and in Table 2 below. 1 Starcrest Consulting Group, LLC, prepared a report for the Port of Los Angeles, Port-Wide Baseline Air Emissions Inventory, June Starcrest Consulting Group, LLC, prepared a report for the Port of Long Beach, 2002 Baseline Emissions Inventory, March Townsend, C., Warren, T., Report to Major Hahn and Councilwoman Hahn by the No Net Increase Task Force, June 24,

11 Table 1-2 NOx & PM Emissions by Major Source Category, tons per year Another perspective in the Hahn report put the proportions of various sources to the regional contribution as shown in Figure 1. In any case, it is clear that the current inventories indicate the ocean going vessels are primarily responsible for NOx and PM emissions. The emphasis in this report is the cargo handling equipment (CHE) which contributes about 26% of the NOx and 36% of the PM emissions at the port where most of the cargo handling is done. Figure 1-1 Percentage of Regional NO x & PM 10 Emissions by Source Category (ref Hahn) 1.1 Cargo Handling Equipment (CHE) Cargo handling equipment is defined as the various pieces of equipment with diesel engines used to move cargo at the ports, including: cranes, excavators, forklifts, sweeper/scrubbers, tractor/loader/backhoe, container handling equipment, and yard tractors. From the ARB state survey 4 in 2004, the yard tractors represented over 60% of the units and about 66% of the NOx and PM emissions as shown in Table 1-3 and Figure 1-2 below. 4 California Environmental Protection Agency, Air Resources Board Staff Report, Stationary Source Division, Emissions Assessment Branch: Initial Statement of Reasons for Proposed Rulemaking & Adoption of the Proposed Regulation for Mobile Cargo Handling Equipment at Ports and Intermodal Rail Yards, October

12 Equipment Types Numbers Emissions, Tons/ Day NOx PM Cranes Excavators Forklifts Container Handling Equipment Other, General Industrial Equipment <0.01 Sweeper/Scrubbers <0.01 Tractor/Loader/ Backhoe Yard Trucks 2, Totals 3, Table 1-3 Estimated Statewide Cargo Handling Equipment Populations and Emissions (ref 4 ARB) Figure 1-2 NOx and Diesel PM Emission Distributions for Cargo Handling Equipment at California Ports (ref ARB) 5

13 1.2 Yard Trucks -General Yard tractors, the focus of the measured emissions for this report, are also known as terminal tractors, yard trucks yard hustlers and yard goats. In any case, these are the units that are typically powered with diesel engines and designed for the movement of containers: to/from ships/trains, on/off terminals, to/from RTG cranes or on/off stacks. The important perspective is that yard tractors are numerous, about 1,500 in the Ports of Ls Angeles and Long Beach, or 2,300 in California, and are frequently used. Given the large population and frequent use, it is not surprising that yard tractors contribute 65% of the emissions associated with cargo handling equipment (CHE). Thus yard tractors are the first unit that should be considered in any project measuring emissions from CHE units. Given that the only measured emissions for the diesel engine in the yard tractors is when the engine is certified on an engine dynamometer, it was important to measure the emissions from an engine in a yard tractor. This approach would provide emission values that could be compared with values used in figuring earlier inventories and would provide insight on potential emission reductions from proposed mitigation strategies. Some of the mitigation strategies include: switching to cleaner fuels, like LNG, or switching to cleaner engine technology, like using diesel engines certified for on-road use in yard tractors, an off-road application. The main thrust of this project was to look at relative emissions from the LNG and diesel-fueled yard hustlers. In addition this work provided an indication of the confidence limits for the current emission factors being used in various surveys. 1.3 Yard Tractors ARB Survey ARB conducted a survey to provide information on the yard tractors based on engine manufacturer and the year made. As is evident, in the Table 1-4 below, engines manufactured by Cummins were the most commonly used in the yard tractors. Furthermore, ARB partitioned the two largest Cummins engine size groups, 5.9L or B-Series and the 8.3L or C-Series, into their model years as seen in Figure 2-1 below. Table 1-4 Diesel Engine Manufacturer for Yard Tractors (ref. ARB 5 ) 6

14 Figure 1-3 Cummins Engine Breakdown by Model Year and Engine (ref. ARB 5 ) ARB s analysis of the total inventory showed the Cummins 5.9L engine was the more common and comprised 14 percent and 9 percent of the 2000 and 2002 model year inventories, respectively. Likewise the 8.3L represented 26 percent of the total inventory with the most common years, 1991 and 2001, comprising 9 percent and 5 percent of the inventory, respectively. Taken together the four models represented about 37 percent of the total inventory. In addition to population, ARB s report provides a thorough characterization of yard tractors for inventory purposes 5 and included many other dimensions such as average horsepower, activity, engine load factor, emission factor deterioration, fuel correction factors, add-on controls and emission reduction strategies. ARB s emission projections included growth factors for port growth, equipment scrappage and the inclusion of new engine standards. 1.4 Emissions Standards for Yard Tractors Emission certification for diesel engines is very different from protocols used to certify gasoline engines. First, all diesel engines are certified in a laboratory on an engine dynamometer, unlike gasoline engines that are certified in the actual vehicle. Second, the emission standards for diesel engines depend on whether the engine is used on- or off-road. The latter distinction is important. Although yard tractors look like tractors commonly seen on highways, they are regulated as off-road or non-road equipment and have a higher emission standards. Examples of the standards for a sub-set of the diesel engines used in non-road application are listed below. Table 1-5 Selected EPA Non-road diesel engine emission standards in g/bkw hr (g/bhp-hr) 5 See reference 4, Appendix B, Emission Estimation Methodology for Cargo Handling Equipment Operating at Ports and Intermodal Rail Yards in California 7

15 1.5 Emission Test Methods for Yard Tractors Selection of the driving/operating cycle for the diesel engine in the yard tractors is very important. The impact of operating cycle, while measuring emissions, is known to outweigh many other parameters. In order to meet the goals of this project, we needed to identify an operating cycle that allowed measured emission values to be compared: 1) with the reference certification values and 2) for different technologies. For example, we wanted to compare emissions from engines certified for on-road standards following the FTP heavy-duty transient cycle to engines certified for off-road standards following steady-state cycles. During this project, we considered but did move forward on the development of an actual in-use cycle as we wanted the emissions from this project to be compared with values measured in earlier studies. Two approaches to matching the emissions measured in the certification cycle surfaced in discussions and involved taking the engine out of the yard tractor and mounting it on an engine dynamometer. In one case, the engines would be operated following the EPA transient cycle used for on-road certification, and in the other case, the engines would be operated following the ARB s 8-mode test cycle that is used for certifying off-road engines. Both approaches were viewed as infeasible given the cost of taking the engine out the yard tractor and the length of time the unit would be away from the owner. In the end, a third approach was developed in which the engine was tested in the yard tractor and emissions measured on a heavy-duty chassis dynamometer while the yard tractor was operated following the ARB s 8-mode test cycle. This approach provided the needed data at considerably less cost and equally important, speeded the time for the yard tractor to be returned to the shipper. The ARB 8 Mode Cycle for certifying off-road vehicles and diesel-powered off-road industrial equipment is the same as ISO-8178-C1 6 shown in the table below. According to Reference 6, specific examples are: industrial drilling rigs, compressors, construction equipment including wheel loaders, bulldozers, crawler tractors, crawler loaders, truck-type loaders, offhighway trucks, hydraulic excavators, agricultural equipment, rotary tillers, forestry equipment, self-propelled agricultural vehicles (including tractors), material handling equipment, fork-lift trucks, road maintenance equipment (motor graders, road rollers, asphalt finishers), snow-plough equipment, airport supporting equipment, aerial lifts, and mobile cranes. Table 1-6 Test Modes, Torque and Weighting Factors for the ISO-8178-C1 Cycle 6 International Standard Organization IS Reciprocating internal combustion engines - Exhaust emission measurement -Part 4:Test cycles for different engine applications, First edition l 5 8

16 A more complete understanding of the definition of rated and intermediate speed is provided in the ISO reference 6 and captured in Figure 1-3 below. While rated speed is the governed speed, intermediate speed is less obvious and defined as: For engines designed to operate over a speed range on a full-load torque curve, the intermediate speed is the declared maximum torque speed if it occurs between 60 % and 75 % of rated speed. If the declared maximum torque speed is less than 60 % of rated speed, then the intermediate speed shall be 60 % of the rated speed. If the declared maximum torque speed is greater than 75 % of the rated speed then the intermediate speed shall be 75 % of rated speed. Figure 1-4 Torque scales and Location of Intermediate and Rated Speed Points. The torque figures given in Table 1-6 and Figure 1-4 represent, for a given test mode, the ratio of the required torque to the maximum possible torque at that speed. 1.6 Dynamometer Testing Dynamometers (dynos) are very useful tools designed to measure torque and rotational speed (rpm) from which power produced by an engine can be calculated. Dynamometers come in various configurations. A dyno that is coupled directly to an engine is known as an engine dyno. A dyno that measures torque and power delivered by the power train of a vehicle without removing the engine from the frame of the vehicle is known as a chassis dyno. With a chassis dyno the vehicle is placed on rollers that turn as the vehicle operates and the output power from the engine is measured through the wheels. While engine dynamometers provide the most accurate results of an engine operation, a chassis dynamometer is often the most practical approach as it can measure the power and torque of an engine without removing the engine. The main issue with the chassis dynamometer is the measured power and torque at the wheels is less than the values at the engine flywheel (e.g. brake horsepower) due to the various frictional and mechanical losses in the various components; for example, drivetrain transmission and gearbox, tire friction and other factors. Estimating these power losses is possible in order to estimate the brake horsepower values for the engine in a vehicle; however, the estimated values are still less accurate than those measured with an engine dynamometer. 9

17 The dynamometer is designed to operate for a range of speeds and torque and is equipped with some means of measuring the operating torque and speed. An important distinction is the type of motor/driver or absorption unit used in the design. Motoring/driving dynos are useful when testing the torque and power requirement for a pump while an absorbing dyno acts as a load and is driven by the vehicle being tested. An absorption dyno absorbs the power and dissipates it as heat by restraining the output shaft mechanically with a friction brake, hydraulically with a water brake, or electrically with an electromagnetic force. For this project we planned to use a heavy-duty chassis dynamometer with a water brake absorption unit because of the expense and time associated with removing an engine from a vehicle and then replacing it. Practically speaking, the budget did not allow for removing/replacing an engine in the vehicle and the owners of the vehicles wanted their tractors returned as soon as possible. Thus the plan was to mount a vehicle on the rollers of a chassis dyno, operate the engine at ISO load points and have the vehicle's wheels spin the rollers to enable the dynamometer to make the required measurements. The water brake operates by adding water to the absorption unit until the engine/wheels run at a steady rpm against the load. Water is kept at that level and replaced by constant draining and refilling, which is needed to carry away the heat created by absorbing the horsepower. The housing attempts to rotate in response to the torque produced but is restrained by the scale or torque metering cell that measures the torque. The concept of the design is very similar to a water pump without an outlet. Figure 1-5 Schematic Shows the Most Common Water Brake, the Variable Level Type. From Figure 1-5, since the restraining element tends to rotate with the output shaft, the force of the shaft can be determined by measuring the force required to prevent the rotation of the restraining element. Torque is then calculated by multiplying the force times the length of the lever arm, or the distance through which the force acts. Because power is the product of the torque and the rotational speed, and because the power is conserved throughout the system, the transmission increases the torque in the same proportion by which it reduces the rotational speed. With torque at the rear wheel and the overall reduction ratio you can calculate the engine torque. 10

18 At a given engine speed, the rear-wheel torque will depend on what gear is selected. A dynamometer chart providing torque must be engine torque or else it would be applicable only to a specific gear and the chart would have to specify the gear. Note, however, that if the torque is measured at the rear wheel and then the overall reduction ratio is used to calculate the engine torque, the result will not be the same as the result that you would get if you connected the dynamometer directly to the crankshaft, because the measured rear-wheel torque is subject to power train and other losses. Nevertheless, any dynamometer chart that shows torque and that does not specify the gear is most definitely the engine torque, albeit adjusted for drive train losses. 11

19 2 Experimental Work Plan The work plan called for testing tractors that represented a span of emission control technologies in the yard operated by Long Beach Container Terminal. The test vehicles included three diesel yard tractors and one yard tractor fueled with LNG. Results from this project would be compared to an earlier study carried out for the Port of Los Angeles, Tetra Tech and CARB. 2.1 Choosing the Right Yard Tractors to Test One point of differentiation between this POLB study and the earlier study was the approach used to select the yard tractors to test. The earlier project selected a few yard tractors to represent the thousands of units in the Ports of Los Angeles and Long Beach. The selection process relied on survey data and stratification of the data to select yard tractors that were representative of the fleet of yard tractors. An approach using stratification was viable as certain dimensions or characteristics of the population of yard tractors were the same for the multiple shippers. For example, Cummins was the major engine manufacturer and only two engine models were commonly used. ARB s earlier work on selection of representative units was helpful in deciding which four year tractors to select from Long Beach Container Terminal. The POLB/LBCT test matrix of engines is shown in Table 2-1 and included a C-Series from 2001 certified to Tier 1 Non-road Standards, a C-Series from 2003 certified to Tier 2 Non-road Standards, a B-Series from 2005 certified to on-road Standards and a CG250 from 2005 using LNG. The first two engines from 2001 and 2003 represented baseline non-road engines with mechanical fuel injection. The third ISB engine is a modern low-emission engine designed for on-road applications and meets lower emission certification limits than the non-road engines. The ISB included electronic controls and a cooled EGR subsystem and a turbocharger with variable geometry. On these three tractors, LBCT installed the Donaldson Diesel Oxygen Catalyst (DOC) and Spiracle Unit 7. Finally the CG250 represented a modern LNG engine with low-nox and very low PM emissions. One goal was to measure the differences in emission for the various engine technologies. Table 2-1 Slate of Yard Tractors and Fuels Selected for Testing 2.2 Test Driving/Operating Schedule For this project WestStart-CALSTART recruited yard tractors from Long Beach Container Terminal and delivered them to Johnson Machinery in Riverside for testing on their SuperFlo 8 Automotive High-Performance chassis dynamometer with a 36-inch diameter and water-brake dynamometer. At Johnson, the yard tractor was safely secured with chains such that the rear/driven wheels were on the chassis rolls. While the ISO C1 Test Cycle in Table 1-6 specifies 7 See 8 See 12

20 the speed and torque, it does not specify exactly how to reach those points for an engine in a vehicle. In order of sequence, the engine map and C1 test points were determined by: 1. First running a power curve with the engine in the vehicle starting with the rated speed or governed speed 2. Identifying the RPM nearest the intermediate speed where 100%, 75% and 50% of peak load can be run at the same RPM 3. Identifying one point at the intermediate RPM if you can not get max at peak torque 4. Last point is idle Thus before any emissions testing could take place, the first step in the emission test protocol was the measurement of the map of engine torque/power versus revolutions per minute (RPM) for the engine in the vehicle while on the chassis dynamometer. Mapping the engine in this manner allows the loads or horsepower measured at the wheel to be repeatable according to the C-1 protocol during the emissions testing. RPM is monitored both at the engine and the dyno. Having the engines in the yard tractors presented some complications in determining the engine map. One common problem was the transmission and accompanying gear ratios allowed the engine to operate at multiple torque and RPM points. In an ideal situation for mapping and testing, we simply choose the gear such that the engine and the wheels operate at the same RPM (direct drive mode) but the yard tractors had several gear ratios and limits including that the wheel speed was governed to less than 25 miles per hour to comply with off-road regulations. Both the gear ratio and 25mph limits were adjusted to develop a suitable engine map. Figure 2-1 Yard Tractor Being Tested on a Chassis Dynamometer and UCR s Mobile Lab Using a chassis heavy-duty chassis dynamometer, the rear wheels and engine were loaded to specific points established from the engine map testing and the total emissions were directed to the UCR mobile lab. The engine was first operated at full power for 30 minutes to stabilize emissions and before starting to test at Mode 1 and in sequence going to Mode 8. A duplicate cycle at each of the eight load points selected in the test matrix followed the first run, starting with Mode 1. Testing at each of the load points lasted 5 minutes or longer if needed to capture sufficient PM mass for accurate measurement. Emission testing included the measurement of the 13

21 exhaust flow rate, the concentrations of carbon monoxide (CO), carbon dioxide (CO 2 ), oxides of nitrogen (NO x ) and total hydrocarbons (THC) and the PM 2.5 mass. In addition to the concentrations of pollutants, many engine parameters were measured during the testing including: exhaust temperature, both dyno and engine RPM, wheel horsepower, ambient pressure, engine exhaust pressure, fuel rate, % torque, and others. 2.3 Measurement of Gas Concentration and Flow Rates The sampling and measurement methods of mass emission rates from heavy-duty diesel engines are specified in great detail in Code of Federal Regulations (CFR): Protection of the Environment, Section 40, Part 86. The University of California Riverside s (UCR) unique mobile, heavy-duty diesel laboratory (MEL) is designed and operated to meet those stringent specifications. MEL is a complex laboratory that was verified against ARB s heavy-duty diesel lab and routinely measures a wide range of speciated and particulate emissions from diesel engines. Design capabilities and details of MEL are described in Cocker 9 and additional information is provided in Appendix A. Thus MEL in combination with the heavy-duty chassis dynamometer at Johnson Caterpillar Company became the test platforms used to measure the emissions from the yard tractors. Diluted Exhaust: Temperature, Absolute Pressure, Throat ΔP, Flow. GPS: Pat, Long, Elevation, # Satellite Precision. CVS Turbine: SCFM, Variable Dilution. Secondary Probe. Gas Sample Probe. Secondary Dilution System* PM (size, Mass). Drivers Aid. Gas Measurements: CO 2 %, O 2 %, CO ppm, NO x ppm, THC ppm, CH 4 ppm. Other Sensor: Dew Point, Ambient Temperature, Control room temperature, Ambient Baro, Trailer Speed (rpm), CVS Inlet Temperature. Dilution Air: Temperature, Absolute Pressure, Throat ΔP, Baro (Ambient), Flow, Dew Point (Ambient). Exhaust: Temperature, ΔP (Exhaust-Ambient), Flow. Engine Broadcast: Intake Temperature, Coolant Temperature, Boost Pressure, Baro Pressure, Vehicle Speed (mph), Engine Speed (rpm), Throttle Position, Load (% of rated). Figure 2-2 Major Systems within UCR s Mobile Emission Lab (MEL) The total exhaust gases from the diesel engine entered the primary tunnel in the mobile emission lab where it was diluted with filtered ambient air. The primary dilution system is 9 Cocker III, D. R., Shah, S., Johnson, K., Miller, J. W., Norbeck, J., Development and Application of a Mobile Laboratory for Measuring Emissions from Diesel Engines. I Regulated Gaseous Emissions, Environ. Sci. Technol.,2004, 38,

22 configured as a full-flow constant volume sampling (CVS) system with a smooth approach orifice (SAO) Venturi and dynamic flow controller. The SAO Venturi has the advantage of no moving parts and repeatable accuracy at high throughput with low-pressure drop. As opposed to traditional dilution tunnels with a positive displacement pump or a critical flow orifice, the SAO system with dynamic flow control eliminates the need for a heat exchanger. Tunnel flow rate is adjustable from 1,000 to 4,000 scfm with accuracy of 0.5% of full scale. It is capable of total exhaust capture for engines up to 600kW. Colorado Engineering Experiment Station Inc. initially calibrated the flow rate through both SAOs used in the primary tunnel. The mobile laboratory contains a suite of gas-phase analyzers on shock-mounted benches. The gas-phase analytical instruments measure NO x, methane (CH 4 ), total hydrocarbons (THC), carbon monoxide (CO), and carbon dioxide (CO 2 ) at a frequency of 10 Hz and were selected based on optimum response time and on road stability. The 200-L Tedlar bags were used to collect tunnel and dilution air samples over a complete test cycle. In the design eight bags were suspended in the MEL allowing four test cycles to be performed between analyses. Filling of the bags is automated with Lab View 7.0 software (National Instruments, Austin, TX). A summary of the analytical instrumentation used, their ranges, and principles of operation is provided in the table below. Each modal analyzer is time-corrected for tunnel, sample line, and analyzer delay time. Gas Component Range Monitoring Method NO x 10/30/100/300/1000 (ppm) Chemiluminescence CO 50/200/1000/3000 (ppm) NDIR CO 2 0.5/2/8/16 (%) NDIR THC 10/30/100/300/1000 & 5000 (ppmc) Heated FID CH4 30/100/300/1000 (ppmc) FID Figure 2-3 Summary of gas-phase instrumentation in MEL Thus during the yard tractor testing we measured exhaust flow and concentrations of carbon monoxide (CO), carbon dioxide (CO 2 ), oxides of nitrogen (NO x ) and total hydrocarbons (THC). Gas phase samples were extracted and the diluted samples are analyzed second by second (modal data). Samples from the engine exhaust were also collected into sample bags over defined phases of the test cycles and analyzed later (integrated data). 2.4 Measurement of Particulate Mass (PM 2.5 ) Particle samples were extracted from the primary dilution tunnel, diluted further in a secondary dilution system and collected on Teflon filters for determining the PM mass. The temperature of the air and filter in the filter collection system is controlled as specified in the CFR. Detailed information about the design and capability of MEL for sampling and measuring PM and toxics is available in Cocker 10. The use of particle filters requires numerous sample handling and custody issues to prevent contamination, to ensure proper sample identification, and to ensure equilibration requirements 10 Cocker, D.R.; Shah, S.D.; Johnson, K.J.; Zhu, X; Miller, J.W.; Norbeck, J.M., Development and Application of a Mobile Laboratory for Measuring Emissions from Diesel Engines. 2. Sampling for Toxics and Particulate Matter, Environ. Sci. & Technol, 2004, 38,

23 of the CFR. The procedures for weighing the filters include specific detailed sequences for checking and recording the equilibration chamber conditions, zero and spanning the balance, weighing reference objects, weighing filters, performing zero and span checks, performing replicates. The filter handling and weighing procedures are described in Appendix B. 2.5 Quality Assurance and Quality Control Requirements From an overview perspective, there are numerous quality control and quality assurance procedures built into the operation of MEL, mainly due to the requirements of the CFR. A partial summary of routine calibrations performed by the MEL staff as part of the data quality assurance/quality control program follows and more detail is listed in Appendix A. The MEL uses precision gas blending to obtain required calibration gas concentrations. Calibration gas cylinders, certified to 1%, are obtained from Scott-Marrin Inc. (Riverside, CA). By using precision blending, the number of calibration gas cylinders in the lab was reduced and cylinders need to be replaced less frequently. The gas divider contains a series of mass flow controllers that are calibrated regularly with a Bios Flow Calibrator (Butler, New Jersey) and produces the required calibration gas concentrations within the required ±1.5 percent accuracy. The CFR specifies a number of quality control and quality assurance requirements in order to meet their protocol in the measurement of emissions from heavy-duty diesel engines. For example Title 40 CFR includes many performance specifications and criteria including: sampling methods instrumental methods environmental controls calibration methods and frequencies QC check methods and frequencies QC check tolerances Documentation of the program and its results include Standard Operating Procedure (SOPs), Checklists, Log Books, Data Files, Reports to Management, and Reports to Clients. Standard Operating Procedures Safety Check: Truck and Trailer Generator start-up Analytical bench start-up Pre-test set-up Test operations Post test shutdown Shutdown: to ground power Shutdown: total Balance Protocol Checklists are maintained for Start-up Calibration Shutdown Test operations Test QC 16

24 Logbooks are maintained for Environmental logbook recording environmental conditions during each test Filter sample identification logbook Equipment tested log (e.g., trucks, generators, etc) Project logs book: history of tests, checks, repairs, etc. Fuel log book: quantities and types of fuel, vehicle or generator Microbalance log book: recording all filter weight and balance QC data Filter Sampling Log: filter Ids and filter weight data associated with each test Computer Data Files. Equipment test selections, configurations, and test sequences are automated by configuration files. Each configuration file specifies a complete sequence of test and QC operations. During the sequence of operations specified in a configuration files, the data from all channels were recorded at 1Hz or faster rate in a raw test results data file. The configuration file and associated data file were identified by test ID number. The test ID numbers were assigned based on date and time of day. Each day of testing consists of a pre-test calibration and post-test calibration. When multiple tests were done in one day, the post-test calibration becomes the pre-test calibration for the next test. For example a typical triplicate test day will look like the following: pre-test calibration and QC configuration pre-test calibration and QC raw test original data Test 1: equipment test configuration with integral post test calibration Test 1: equipment test raw test original data with integral post test calibration Test 2: equipment test configuration with integral post test calibration Test 2: equipment test raw test original data with integral post test calibration Test 3: equipment test configuration with integral post test calibration Test 3: equipment test raw test original data with integral post test calibration In addition to the files generated by the data acquisition system there were a particle Filter Log File and an environmental data check log. The Filter Log file that associates filter identification numbers with sampling locations and times throughout the test, and with filter weight gains. The environmental data log verifies barometric pressure, dew point temperature, ambient temperature, and RH with lab values for at least one point during a day of testing. Raw test original data files were post processed to produce original QC summaries and original data summaries for review by the Project Manager. The post-processed files include: one test file for 1-second modal data one test file for uncorrected integrated cycle data, calibration data, MFC data, etc In addition to the individual test files, the QC data from each test and the integrated cycle data were each appended to a database containing all test results since the beginning of MEL operations. The following two database files were maintained from the raw data file post processing: database of integrated test results (emissions in grams per cycle, ppm and environmental) 17

25 database of integrated test QC results (zeros, spans, and test specific data) During data validation and review, comments and corrections are recorded by editing a COPY of the raw test original data files. The edited, corrected files are called raw test validated data files, and these validated data files are used to generate and replace entries in the post processed files. In addition to the databases of integrated results and test QC results, there were database files containing histories of QC checks not related to specific tests. These include: Propane injection mass balance CO 2 injection mass balance NOx converter efficiency Blended gas calibration checks Analyzer linearity checks In addition to the QA/QC steps required by the CFR, UCR took the data one more step in that our calculated emission factors were compared with manufacturer s value and EPA/ARB standards. 18

26 3 Results Results are reported in separate sections beginning with reporting the actual engine maps measured on the heavy-duty chassis dynamometer followed by the measurement of emissions from the diesel engines while the yard tractors operated at the selected modal points. Data used in the analyses are included in Appendix C of this report. 3.1 Yard Tractors Tested and Protocol for Engine Maps The test matrix of yard tractors planned for this project was intended to represent a cross section of the technology being used at Long Beach Container Terminal (LBCT). The actual test matrix for this project is shown in Table 3-1 below. All vehicles were assembled by Ottawa and the engine serial numbers and operating hours are listed for reference. The plan was to test a hierarchy of emissions control technologies across the various units at LBCT. Table 3-1 List of the Engines Tested and Their Properties As described earlier, several complications arose when testing an engine mounted in a vehicle rather than on an engine dynamometer. First the difficulty in establishing the test points specified in the C1 cycle, especially the test points at the intermediate speed/rpm; namely: modes 5, 6 and 7. Normally the intermediate speed is near the maximum torque point on the engine map (1500RPM for the Cummins C8.3L engine) and these points are easily achieved on an engine dynamometer setup. However, the chassis dynamometer required a minimum wheel speed to produce sufficient wheel loading. Because off-road yard trucks are governed to a maximum speed of less than 25 mph, the wheel speed was often too low for the dynamometer to generate some of the design test loads at the intermediate speed point. Therefore, the yard trucks were operated at a higher wheel speed to achieve the intermediate load points. This approach resulted in engine speeds (RPM) which were higher than those intermediate speed points specified in the C1 cycle for modes 5, 6 and 7. In general, the intermediate engine speed points were increased from specified speed points of about 1500 rpm to 1900 rpm and weighed emission factors were calculated using the modified intermediate speed points. This approach was used in the earlier work when testing with ARB s Stationary Source Division. Other issues were the automatic transmissions limited the maximum load at low RPM so we established points at 50% and 10% of maximum load at the lowest RPM. Finally, tire temperature was monitored during testing and the order was changed as needed to prevent heat build up and loss of tires or power to tires. 19

27 3.2 Test of Mechanically Controlled Engines at Tier 1 & 2 Off-Road Standards The first yard tractor tested used a 2001 Cummins C-model, 8.3L diesel engine certified to EPA s Off-Road Tier-1 emission standard. As mentioned previously, the first activity involved securing the tractor to the heavy-duty, water-brake chassis dynamometer and mapping the output of the engine with the direct drive gear (if possible) while operating at wide open throttle (WOT) in increments of 100 RPM. The resulting engine map was obtained and shown below. Note the chart shows 9-modes and this approach follows that used in the ARB testing. Basically we added an additional mode at the intermediate speed and 50% power as we could not obtain 100% power at the intermediate speed. This point allowed us to check differences between design rated speed and the higher speed (RPM) where the data were actually collected. Thus Mode 9 is idle and part of the C-1 series but the included Mode 8 is only test point and not included in the modal analysis. Figure 3-1 (a) Matrix of Test Points for the C8.3L Engine in the Yard Tractor and (b) Data from Cummins Brochure One observation with Ottawa #118 data is the unit was not driven in direct gear since the engine speed was 2200RPM and the drum speed was 22mph. Knowing the drum and tire diameters were 36 and 41inches respectively; then the 22mph would be equivalent to a drum speed of 205RPM and a tire speed of 185RPM. Clearly the gear ratio was more like 10/1 rather than 1/1. Another observation is that the maximum power achieved at the rated or governed speed of 2200RPM was 107 hp at the dynamometer wheels. Comparing this value with the published value of 215 brake horse power (bhp) at 2200RPM suggests that 50% of the power was not achieved at the wheels. This deviation is larger than expected as nominally, one might expect losses from brake-horse power of about 25% due to transmission and other losses. The relationship between brake-horsepower and wheel horse power (whp) will become more evident in later units. We noted the lower than expected relationship between the whp and bhp after the emission measurements were made on Ottawa #118 so we obtained Ottawa #117, another unit with Tier 1 technology. However, this unit provided the same power at the wheels as #118 so we did not measure emissions. We were not able to identify the reason for the deviation between the published bhp for a new engine and the dyno values that we measured. Clearly some of the 20

28 losses were due to auxiliary units on the engine (like a fan) and the transmission losses but these power losses typically total 25%. The third yard tractor tested used a 2003 Cummins C8.3L engine and control technology meeting EPA s Tier 2 Non-road standards. The figure below shows the measured and rated engine values. Note for this case that the ratio of power measured at the wheels to published brake horse power (161/205) was about 80%, a value more typical of what was observed in the past as compared with a value of 50% for the units #117 and #118. Figure 3-2 Matrix of Test Points for a Cummins C8.3L Engine in the Yard Tractor and (b) Data from Cummins Brochure 3.3 Test of an Electronically Controlled Engine at 2004 On-Road Standards The fourth yard tractor tested used a Cummins 2005 ISB engine. The on-road ISB engine/emissions control system has advanced technology as compared with the earlier C8.3L. The ISB has electronic engine controls and is using an EGR emissions control strategy to meet EPA s on-road standards that are more stringent than Tier-2 off-road standards. The figures below show the resultant C1 Cycle that was developed from the engine map data and the published values from Cummins. Note the ratio of power measured at the wheels to rated power at 2400RPM (167/237) was >70% based on the Cummins data. 21

29 Figure 3-3 (a) Engine Map from the Chassis Dynamometer & (b) from Cummins ISB Brochure A perspective of the relationship between the wheel horsepower and the power at the engine flywheel (brake horsepower) can be seen from the ISB engine data. The ISB engine included an electronic control module (ECM) that broadcasts the power at the flywheel and we recorded during the testing. For this ISB engine the figure below shows the relationship between the broadcast horsepower and the wheel horsepower. For example, using the equation and Figure 3-3b, the calculated power at the wheels is about 70% at 2400RPM, 80% at 180ORPM and ~5% at idle. Values calculated for Ottawa #126 match these values; however, the losses for Ottawa #118 significantly exceed the expected difference between the wheel and brake horsepower %load y = 0.50x R 2 = wheel hp Figure 3-4 Plot of Measured Wheel hp vs. %ECM-load for ISB Engine 3.4 Test of Electronically Controlled Engine Using LNG fuel 22

30 The fifth yard tractor tested used a Cummins Westport engine fueled on LNG and the figures below provide the data on the engine mapping exercise and cycle selected for the chassis testing. The brochure states: The powerful Cummins Westport C Gas Plus has the highest power-toweight ratio in its class. The Cummins Westport C Gas Plus engine is a six-cylinder Lean-Burn Spark-Ignited (LBSI) natural gas vehicle engine that delivers from 250 to 280 hp with ultra-low emissions. It provides excellent torque, high fuel efficiency and reliable, robust performance, with excellent ratings. Its ECM (Electronic Control Module) allows engine performance to be tailored to fit the vehicle mission with road speed governing and cruise control, as well as an engine protection system and complete self-diagnostics. Prior to testing, the fuel delivery system was modified by a Cummins field person to accommodate the dynamometer testing. From the work order, the modifications included 1) removing the air compressor intake hose from mixer inlet housing 2) plug-off mixer inlet opening to conserve boost pressure and 3) allow air compressor to be naturally aspirated for test purposes. The unit was restored to the original configuration after testing. Note the cycle time for collection of data was reduced for the LNG engine in order to conserve fuel pressure and allow the engine to hit higher load points without having to refuel. Earlier work indicated this approach was beneficial and that the emissions were stable within a very short time of hitting the targeted load points. Figure 3-5 (a) Engine Map from the Chassis Dynamometer & (b) from Cummins Brochure As can be calculated from these data, the difference between the wheel horse power and that at the fly-wheel (brake horse power) is about the same as for the ISB engine. A plot of the wheel horsepower as a function of percentage (%) of load reported from the ECM is shown in the figure below. Note for the LNG engine, the power at idle is about 12% of the rated power as compared with about 5% for the ISB engine. 23

31 %load y = 0.41x R 2 = wheel hp Figure 3-6 Plot of Measured Wheel hp vs. %ECM-load for LNG Unit 3.5 Measured Emissions -- Modal Data As described in the earlier section, three yard trucks using CARB ULSD diesel fuel and different engine/emissions control technology were tested following a modified 8-Mode ISO C1 cycle developed on the heavy-duty chassis dynamometer. Full gaseous and particulate emissions were directed to the UCR mobile lab where CO, CO 2, NO x, THC and PM 2.5 concentrations and flow rates were measured. Raw and processed data are shown in Appendix C. Results for the mode by mode emissions in grams per wheel-horsepower-hours are shown in the figure below for the 2001 engine meeting EPA Tier-1 standards and the 2003 engine meeting EPA Tier-2 standards. While Mode 9 or Idle is part of the C-1 Cycle, there is no bar chart at Mode 9 since there is no measured power. Mode 8 is not part of the C-1 cycle but was included for comparison since emissions were measured. Error bars representing the confidence limits at one standard deviation are indicated in all the figures. Note the high confidence in measured emission values since the coefficient of variation is <5% for most data sets. 30 g/whp-hr NOx & 10*PM (C/2001) 30 g/whp-hr NOx & 10*PM (C/2003) 25 NOx 25 NOx 20 10*PM 20 10*PM Figure 3-7 Modal Data in g/whp-hr for a) C/2001 (Tier 1) and b) C/2003 (Tier 2) 24

32 The values in Figure 3-7 indicate modal emissions for the C/2001 Tier 1 unit were about double (200%) the C-2003 Tier 2 values at all modes. This finding is not surprising as the Tier 1 EPA standards are about 47% higher than the Tier 2 standards. The 47% figure is reached by using the certification values in Table 1-5 and assuming the HC is ~5% of the HC+NO x standard. Thus 9.2/(6.6*95%) = 147%. One complication with interpreting the C-2001 data is the wheel horsepower was only 50% of the rated brake horsepower as compared with expectation that the wheel horsepower would be 70 to 80% of the rated brake horsepower. While clearly the modal emissions values for the C can be calculated; a question is whether the C-2001 values are comparable with the C-2003 values? Given that modal emissions in grams per horsepower-hour are relatively independent of power (or Mode) as seen in Figure 3-7, I believe the values for the two model years can be compared. Further the emissions for some points were at exactly the same percentage of brake horsepower; for example, Mode 3 of the C-2001 is at exactly the same percentage of brake horsepower as Mode 2 for the C-2003 engine. A second figure with modal data provides a side-by-side comparison for the emissions from a ISB/2005 engine meeting the on-road standards with emissions from the C/2003 engine meeting Tier 2 off-road standards. The results in Figure 3-8 clearly show the modal emissions for the ISB engine certified to the on-road standards are lower than the modal emissions for the C/2003 engine certified to off-road standards. This figure provides an interesting comparison in that the off-road and the on-road engines are certified following different operating schedules. The offroad uses steady-state cycles and the on-road testing follows a transient cycle so this comparison is rarely made and was a cornerstone of this project. Clearly the emissions from the ISB engine are half the emissions of the Tier 2 engine when compared on the same modal cycle. g/whp-hr 20 NOx & 10*PM (C/2003) g/whp-hr 8 NOx & 10*PM Emissions (ISB/2005) 15 NOx 10*PM NOx 10*PM Figure 3-8 Modal Data in g/whp-hr for a) C/2003 (Tier 2) and b) ISB/2005 (On-road/2004) Note difference in scale Normally emissions per unit of work are fairly constant, with two notable exceptions in Figure 3-8 a) & b). One is Mode #4 at 10% power where the engine may not be designed to work as efficiently at those low loads and the ratio of emissions to small power in the denominator gives a higher than expected modal rate. Another deviation is Mode 8 of the ISB engine. Given that the ISB engine is 5.9L with electronic controls designed for on-road use and the C is an 8.3L with mechanical functioning designed for off-road applications, there is likely to be significant 25

33 Study of Emissions from Yard Tractors with Diesel and LNG Fuels differences in the design for engine operation. The engine manufacturer would know the reasons for this finding. The final chart compares the modal emissions between the ISB and the Cummins CG engines. No PM are shown for the CG250 engine as earlier studies have shown that the emission levels are very low and running the cycle long enough to measure the PM emissions will necessitate that the unit be refueled. g/whp-hr 8 NOx & 10PM Emissions (ISB/2005) NOx PM g/whp-hr NOx Emissions (CG250/LNG) Figure 3-9 Modal Data in g/whp-hr for a) ISB/2005 (On-road) and b) CG250/2005 (On-road) Results comparing the modal emission profiles for the CG250 and ISB engines show significant differences. Note NOx emissions from the CG engine follow load and are higher at the higher loads (and Mode#4). If the LNG engine is running lean at higher loads, then engine temperature is higher and NOx is increased. As a consequence of the engine operation, most of the modal values for the CG250 engine are significantly higher than the same modal data for the ISB engine. Recall that the CG250 fueling system was modified for this testing from the in-use setup in order to enhance the fuel delivery during this dynamometer testing. Earlier tests with the POLA units showed that making these changes caused fewer fueling errors codes on the ECM during the dyno tests. 3.6 Calculating Overall Emission Factors from Modal Data Emission factors were calculated using the equation below with the weighting factors established in the ISO protocols. The equation is. Where: A wm = Weighted mass emission level (HC, CO, CO 2, PM, or NO X ) in g/whp-hr. g i = Mass flow in grams per hour, P i = Power measured during each mode, includes auxiliary loads, WF i = Effective weighing factor for an ISO C1 cycle. (see Table 1-6) 26

34 Note for the purposes of this calculation only the emissions and power data from our designed C-1 cycle were included. Thus data from the intermediate RPM point, identified as Mode 8 in the data tables, are not included in calculating the overall emissions factor. However, Mode 9 (Idle) data are included in the calculating the overall emission factors and the form of the equation allows idle with zero power to be included in the weighted modal emission factors. Also recognize that the overall emission factors calculated with the above equation does not give the same result as that calculated from the weighted average of the modal data. 3.7 Weighted Emission Factors for Engines in LBCT Yard Tractors Four engines with four different engine design technologies were tested during the project and the results for the weighted emission factors are shown in the Table below. Note the PM and NO x emission decreased considerably as the standards became increasingly tougher. From the table, the ISB engine has the lowest NO x emissions and the CG engine has the lowest PM emissions. With the increased emphasis on carbon consumption, it is interesting that the 2003 C8.3L was replaced with a 5.9L engine and the fuel consumption was about the same for the same amount of work. Note the CG engine has a very high THC emission factor but has a lower CO 2 profile due to the heating value contributed by the higher hydrogen content. Table 3-2 Weighted Emission Rates (g/whp-hr) for Four Engines Tested The figure below shows the relative weighted emissions factor for NO x and PM for the four engines tested in the project. 12 Weighted NOx Emission Factors (g/whp-hr) 0.30 Weighted PM Emission Factors (g/whp-hr) C-2001 C-2003 ISB-2005 CG C-2001 C-2003 ISB Figure 3-10 Plots of the Weighted Emission Factor (g/whp-hr) for a) NOx and b) PM 27

35 3.8 Comparison of Emission Factors: This and Earlier Studies with Standards One of the goals of this project was to compare the emission factors measured in this study at the POLB/LBCT with values measured in the POLA project. Second we wanted to compare the values measured on the chassis dynamometer with the certification values measured on an engine dynamometer, although there is no reason to expect engines set up in a chassis would have the same emissions as when run attached to an engine dynamometer. The table below makes that comparison. Note the values measured on the chassis dyno are for wheel horsepower and the certification values are for brake horsepower so the measured values in the table should be reduced by about 20 to 30% to be directly comparable on a brake horsepower basis. Table 3-3 Compilation of a) Measured Values in g/whp-hr & b) EPA Standards in g/bhp-hr 3.9 Deeper Evaluation of Emissions Measured for 2001 C8.3L Engines Looking at the data for the mechanically controlled C8.3L/215 we see that both the earlier POLA and the new LBCT tests used an engine rated at 215 bhp and the C1 test points were similar as is evident in the table below. This means that the POLA unit was also tested with about 50% of the engine output delivered to the wheels at 2200 RPM as in the POLB/LBCT test. Table 3-4 Comparative C1 Test Points for the POLB & POLA Tests However, the emissions data show a statistically significant difference in the measured NO x and PM values based on the measured coefficient of variation in the testing. On the surface it looks like a classic NOx/PM trade-off, perhaps by the timing of the fuel injection being advanced for the LBCT engine. However, this explanation is unlikely as the CO 2 rates and the weighted work over the C1 cycle are statistically the same as shown in the table below. 28

36 Table 3-5 Comparative Emission Values for Tests of Two C8.3L Engines We also note the POLA test unit has higher emissions of CO and THC but this finding is consistent with the fact that the POLB/LBCT yard tractors had an installed DOC after the engine. The lower PM level for the POLB test is likely due to the after control unit with an efficiency of 25% that installed on the POLB unit. No obvious explanation can account for the lower NOx in the POLA unit but it was not due to the test points since both were run at the same points. A final analysis compared the weighted emission factor is in grams/whp-hr with the EPA emission standard. We can estimate the weighted emissions factor on a brake-hp-hr if we assume the losses between the flywheel and wheels. For example, if the losses were the average of 25% over the C1 cycle, then the emissions factors on a brake hp-hr for POLB unit would be about 8.3 for NOx and 0.20 for PM and above the EPA certification standard for NOx and lower than the PM standard. By contrast, both NOx and PM were within the standard for the unit from the POLA. From the modal data in Figure 3-7, it is evident, except for Mode 4, that the modal emissions do not change much with load. Thus this approach for estimating the emission factor on a brake house power basis is reasonable. The greater emission factor for Mode 4 only slightly increases the overall weighted emission factor for the eight modes Deeper Evaluation of Values Measured for a 2003 C8.3L Engine The second technology tested was another mechanically controlled C8.3L but this one met Tier 2 specifications and was rated at 205bhp. These are no data directly comparing the power at the flywheel and wheels but from the earlier discussion the C1 points were about 80% of the rated power over the C1 cycle and much higher than the first C8.3L engine. Using those figures, then the estimated emissions factors with the flywheel power is about 4.95 and 0.10 for NO x and PM respectively. These values are in good agreement with the EPA emission standards. The PM being below standard was due in part to the Donaldson PM Control technology of about 25% Deeper Evaluation of Emissions Measured for ISB5.9L Engines The third yard tractor tested had an ISB electronically controlled engine certified to meet the EPA on-road standard tested and used cooled EGR to meet the NO x limits. One feature and benefit of engines with electronic controlled modules (ECM) is the engines have multiple sensors, the data from which is recorded and broadcast while the engine is operating. Earlier we presented data showing the relationship between ECM broadcast power and power measured at the wheels. The correlation coefficient was very high. The ECM also broadcasts the fuel consumed and this value should correlate strongly with measured emissions of CO 2. Further the inspection of the relationship between the two independent measures provides a check of the emissions data. A chart showing the measured fuel consumption and CO 2 emissions is provided below. Note the coefficient of determination indicates a strong correlation between the two independently measured but highly correlated variables. 29

37 e-fuel rate vs CO2 rate y = 0.00x R 2 = Figure 3-11 Plot of the Broadcast Fuel Rate and the Measured CO 2 Rate A comparison of the C1 cycles for the two engines and modal emissions data are shown in tables and figure below. Note the C1 test points at rated speed (Modes 1-4) have nearly equal power and emissions factors. However the intermediate RPM was 2000RPM for the 2004 unit and 1800RPM for the 2005 unit and the resulting test points were different. Table 3-6 Comparative Results for the C1 Cycles Used for the Two Engines The weighted emissions factors from the two tests of the tractors with ISB engines are compared in the table below. Table 3-7 Comparative Results for ISB Engines from POLB & POLA These data contrast with the earlier table that compared the results from the C8.3L-2001 engines in that the emissions of CO and THC are the same. This finding is surprising given LBCT s claim of having installed a DOC on all diesel units. We would have expected lower emission factors for CO and THC as observed for the C-8.3L engines from LBCT. Even the PM level is the same, again a value that was expected to be reduced by 25%. 30

38 Also interesting is the CO 2, work and NOx emissions are statically different. Recall the earlier C-engines showed emission factors for CO 2 were within 1.5%, a value within the error limits of measurement. For the ISB engines, the variation in CO 2 emission factors is 5.5% and statistically different from the measurement error. From these data it is difficult to judge whether the higher test load at the lower RPM caused the emission increase or whether the ECM was adjusted for advanced fuel injection and greater efficiency leading to the higher emissions. Note Mode 6 (2005) and Mode 5 (2004) had the same load but the emission factor was higher for the ISB g/whp-hr NOx Emissions (ISB 2004) g/whp-hr NOx Emissions (ISB/2005) Mode Figure 3-12 Comparative Data for the ISB-2004 and ISB-2005 It is also interesting that at the C1 intermediate speed (Mode 8) the emission factor is doubled for the 2005 unit. Whether the electronic controls had advanced the timing of fuel injection for the ISB-2005 is hard to discern from the data. In any case, the results show higher NOx and higher efficiency of the engine for the ISB-2005 test. Comparing the NO x and PM emissions with the certification values for the ISB-2005 unit involves multiplying the values in Table 3-7 by 75% to convert wheel horsepower to brake horsepower. Such an approach provides an estimated NO x emissions factor of 2.23 g/bhp-hr, a value within the EPA standard, although higher than the earlier ISB-2004 engine tested Deeper Evaluation of Emissions Measured for CG250 Engines The last technology tested in the sequence was the Cummins CG250 engine that is fueled by LNG. C1 test conditions for the POLA and the POLB tests were quite similar as shown in the table below. However, as pointed out earlier, the first POLA test was carried out with the yard tractor as received from the POLA while the POLB test used a yard tractor that was modified to increase fuel flow for the dynamometer test. Without modification, the POLA unit had error codes for the highest load points and these were eliminated in the POLB test. 31

39 Table 3-8 Comparative C1 Test Points for Two LNG Engines The weighted emissions factors from the two tests of the LNG fueled tractors are compared in the table below. Table 3-9 Comparative Weighted Emission Factors for Two LNG Tests For the most part the weighted work and emissions are statistically the same which is consistent with the engines following the same C1 cycle. The only difference is for NOx and the modal differences can be seen in the figure below g/whp-hr Modal Emissions for CG250 POLA POLB Figure 3-13 Comparative Data for the CG250 from the POLA and POLB Tests Although the C1 cycle modes for testing both the POLA and POLB units was similar, it is clear from the figure above that the emissions at each mode were higher from the as-received yard tractor from the POLA than for the POLB unit. Making the adjustments to the fuel system before testing the POLB unit eliminated the error codes for insufficient fueling that were found in the POLA test. Reduced modal emissions during tests of the POLB unit should not be a surprise as the engine was operating closer to the design specifications with respect to the fuel delivered to the engine. 32

40 One aspect of the LNG testing when compared to the diesel testing is difference in modal emission values for Modes 1-3 and Modes 5-7. For example, with diesel fuel the emission values per unit of work are nearly constant but with LNG, the modal values decline from about 5 to near 2.. Presumably this feature in the data is one reason for the weighted emission factor being much greater than the manufacturer s certification value of 1.8 grams per brake-horsepower-hour Added Analysis: Limited Analysis of ECM Data from the ISB Engine One of the features of the newer engines is the addition of extensive electronic controls and the capability of recording a number of operating data related to the fuel consumption, maintenance cycles and activity of the engine and gears. Appendix D reports selected data from the ECM on the LBCT yard tractor used in this project. Some of those data are listed in the table below. Table 3-10 Selected ECM Data from the ISB Engine Of particular interest in the data set is the percentage of time at idle as compared with total engine run time. For this particular yard tractor the unit spent about 50% of it time in the idle mode. Fuel consumption for the idle mode was about 20% of the total fuel consumed for this unit. In any case, this brief analysis and perspective shows the usefulness on the ECM data from a yard tractor for learning more about the activity of typical yard tractors and fuel consumption during those operations. In an earlier publication 11, this author showed the utility of ECM data for heavy-duty trucks. Another point of interest is the emission rates in grams per hour during idle and these rates can be looked up in the tables of Appendix C. The significance of these values is that with idle control, there is the opportunity to save on both fuel and emissions. A summary of the emission rates at idle are presented in the table below. Table 3-11 Emission Rates at Idle for the Various Units 11 Huai,T., Shah, S. D., Miller, J. W., Youngglove, T., Chernich, D. J., Ayala, A., Analysis of heavy-duty diesel truck activity and emissions data, Atmospheric Environment 40 (2006)

41 4 Findings and Recommendations Yard tractors are the workhorse of the cargo handling equipment (CHE) and there are over 1,500 in the Ports of Los Angeles and Long Beach. Because of their high numbers and frequent use, they also are the primary contributor to the emissions inventory from cargo handling equipment. Over 50% of the NOx and PM emissions associated with CHEs come from yard tractors. Based on these facts the POLB, LBCT and WestStart-CALSTART launched an investigation of the emission factors from yard tractors. There were two goals: first, measure the emission factors from a series of engines certified to ever increasingly more stringent emission standards, and second, investigate the emissions from an engine running on an alternative fuel, LNG. As direct emissions values are scarce, the report compares the results with earlier studies and EPA certification values. A number of yard tractors with various engine technologies meeting different levels of EPA emission standards were part of the planned project as shown in the table below. The test matrix represented engines meeting Tier 1 and 2 non-road (industrial) standards, on-road (automotive) standards for 2004 and an engine fueled with LNG. Table 4-1 List of Test Engines & Properties The engines were tested in their yard tractors on a heavy-duty chassis dynamometer. Testing included: first, developing an engine map to establish a C1 ISO test cycle, and second, measuring the emissions at various C1 test points in duplicate. Both modal and overall emission factors were calculated from the test data. The weighted emission factors for NOx and PM are shown in the figure below for the various engines. About 75 emissions tests were carried out during the project. Figure 4-1 Weighted Emission Rates (grams/wheel-hp-hr) for Yard Tractors with Various Diesel Engine Technologies and a Cummins-Westport LNG Engine a) NOx & b) PM 34

42 Results for the diesel fueled engines showed the weighted emission factors on grams per wheel horsepower-hour for NO x and PM decreased with engines manufactured to meet stricter EPA certification standards. Estimates of the emission factors on a brake horsepower basis indicate the values were within the EPA specifications for the C-2003 and ISB engines but not for the C-2001 engine. Unlike the earlier POLA testing, Donaldson control technology was used in the LBCT tests and reduced the CO, THC and PM emissions on the C-8.3L engines but unexpectedly did not reduce emissions on the ISB engine. Several issues were discussed in the analysis section. One issue was the C-2001 engine map in that the C1 load points were about 50% of Cummins engine map instead of the expected 75%. As detailed in the analysis, we believe the emissions factors are still valid. Another issue was the results comparing the ISB engines from 2004 and The 2005/LBCT engine had about 20% higher NO x emissions than the earlier POLA test unit and the reason for the difference was not clear. Several hypotheses were presented in the discussion. Finally, the LNG fueling system from the LBCT yard tractor was modified prior to testing and did not generate the fueling errors seen during the POLA test. Consequently the LBCT yard tractor had lower NOx emissions than the earlier POLA test. One frontier examined in this work was the activity data for the ISB engine. These data revealed that the engines spend considerable time at idle. The use of the engine downloads appears to be useful information about how a yard tractor operates. Not examined in this study was the measurement of the in-use activity and emissions from yard tractors with portable instruments during actual use of moving containers. Obtaining those values is planned and those values should be very helpful in finalizing the most appropriate emission factors to use for yard tractors. 35

43 Appendix A Description of UCR s Mobile Lab and Selected QA/QC Procedures While extensive detail on UCR s mobile lab is provided in Cocker 9,10 ; this section is made available for those that may not have access to that reference. Basically UCR s mobile emissions lab (MEL) consists of a number of operating systems and was designed to meet the requirements of the Code of Federal Regulations Part 40 for the testing of heavy-duty diesel engines. A schematic of MEL and its major subsystems is shown in the figure below and some description follows. Diluted Exhaust: Temperature, Absolute Pressure, Throat ΔP, Flow. GPS: Pat, Long, Elevation, # Satellite Precision. CVS Turbine: SCFM, Variable Dilution. Secondary Probe. Gas Sample Probe. Secondary Dilution System* PM (size, Mass). Drivers Aid. Gas Measurements: CO 2 %, O 2 %, CO ppm, NO x ppm, THC ppm, CH 4 ppm. Other Sensor: Dew Point, Ambient Temperature, Control room temperature, Ambient Baro, Trailer Speed (rpm), CVS Inlet Temperature. Dilution Air: Temperature, Absolute Pressure, Throat ΔP, Baro (Ambient), Flow, Dew Point (Ambient). Exhaust: Temperature, ΔP (Exhaust-Ambient), Flow. Engine Broadcast: Intake Temperature, Coolant Temperature, Boost Pressure, Baro Pressure, Vehicle Speed (mph), Engine Speed (rpm), Throttle Position, Load (% of rated). Major Systems within the Mobile Emission Lab The laboratory was designed to accept the full output from the diesel engine as input into a primary dilution tunnel. The primary dilution system is configured as a full-flow constant volume sampling (CVS) system with a smooth approach orifice (SAO) venturi and dynamic flow controller. The SAO venturi has the advantage of no moving parts and repeatable accuracy at high throughput with low-pressure drop. As opposed to traditional dilution tunnels with a positive displacement pump or a critical flow orifice, the SAO system with dynamic flow control eliminates the need for a heat exchanger. Tunnel flow rate is adjustable from 1000 to 4000 scfm with accuracy of 0.5% of full scale. It is capable of total exhaust capture for engines up to 600kW. Colorado Engineering Experiment Station Inc. initially calibrated the flow rate through both SAOs for the primary tunnel. The mobile laboratory contains a suite of gas-phase analyzers on shock-mounted benches. The gas-phase analytical instruments measure NO x, methane (CH 4 ), total hydrocarbons (THC), 36

44 CO, and CO 2 at a frequency of 10 Hz and were selected based on optimum response time and on road stability. The 200-L Tedlar bags are used to collect tunnel and dilution air samples over a complete test cycle. A total of eight bags are suspended in the MEL allowing four test cycles to be performed between analyses. Filling of the bags is automated with Lab View 7.0 software (National Instruments, Austin, TX). A summary of the analytical instrumentation used, their ranges, and principles of operation is provided in the table below. Each modal analyzer is timecorrected for tunnel, sample line, and analyzer delay time. Gas Ranges Monitoring Method Component NO x 10/30/100/300/1000 (ppm) Chemiluminescence CO 50/200/1000/3000 (ppm) NDIR CO 2 0.5/2/8/16 (%) NDIR THC 10/30/100/300/1000 & 5000 (ppmc) Heated FID CH4 30/100/300/1000 (ppmc) FID Summary of gas-phase instrumentation in MEL CARB Verification of the Heavy-duty Diesel Mobile Laboratory A cross-lab correlation check was performed between the ARB and the UCR mobile lab using a Freightliner tractor equipped with a 475 hp, MY2000 Caterpillar C-15 diesel engine. The cross lab check was carried out at CARB s heavy-duty chassis dynamometer facility located at the Metropolitan Transit Authority (MTA) facilities in Los Angeles, California. The vehicle was loaded using the chassis dynamometer, and emissions measurements were made using either the CARB laboratory or MEL. The truck was operated on the Urban Dynamometer Driving Schedule (UDDS) and two steady-speed tests. The table below shows the results of these tests. It should be noted that all MEL emissions data were submitted to CARB, which returned them with the deviation from their results. ARB s values were blind to UCR in the process. A crosslaboratory check performed by other heavy-duty diesel (HDD) laboratories reported (Traver 2002) similar deviations, as this study found. Test Cycle THC CO NO X CO 2 Hot UDDS 10.1% 13.0% 8.9% 5.2% mph 7.4% 12.3% 4.0% 4.9% 55 mph 16.4% 3.7% 4.0% 5.5% Cross-laboratory test performed at CARB s HDDT test facility (January 31, 2002) After the installation of the secondary system, a number of internal and external confirmation tests were carried out. For example, the masses of PM 2.5 collected on two parallel samples holders were compared and the results were within 5%. Also a cross-lab correlation check was performed with the same Freightliner tractor at the CARB heavy-duty chassis dynamometer facility while operating on the UDDS. Emission measurements were made using the MEL and CARB measurement benches on consecutive days. Table 11 shows the results of these tests. For these tests, the filter face temperature in the MEL was adjusted to 27 C (81 F) to match the CARB PM collection system. A retest in the MEL with the filter face temperature set to 47 C (117 F ) recovered ~11% less PM mass than the MEL test at 27 C (81 F ). Following the tests, the MEL emissions data were submitted blind to CARB who provided the percent differences 37

45 between the labs, as shown in Table 11. A cross-lab check performed by other HDD laboratories reported similar deviations as those found in an earlier CRC cross-laboratory study. Test Cycle THC CO NO x CO 2 PM Hot UDDS 11.8% 18.4% 8.0% 2.7% 0.1% Cross-laboratory test performed at CARB s HDDT test facility. (March 19, 2002) Quality Assurance and Quality Control Requirements Internal calibration and verification procedures are performed regularly in accordance with the CFR. A partial summary of routine calibrations performed by the MEL staff as part of the data quality assurance/quality control program is listed in the table below. The MEL uses precision gas blending to obtain required calibration gas concentrations. Calibration gas cylinders, certified to 1 %, are obtained from Scott-Marrin Inc. (Riverside, CA). By using precision blending, the number of calibration gas cylinders in the lab was reduced to 5 and cylinders need to be replaced less frequently. The gas divider contains a series of mass flow controllers that are calibrated regularly with a Bios Flow Calibrator (Butler, New Jersey) and produces the required calibration gas concentrations within the required ±1.5 percent accuracy. In addition to weekly propane recovery checks which yield >98% recovery, CO 2 recovery checks are also performed. A calibrated mass of CO 2 is injected into the primary dilution tunnel and is measured downstream by the CO 2 analyzer. These tests also yield >98% recovery. The results of each recovery check are all stored in an internal QA/QC graph that allows for the immediate identification of problems and/or sampling bias. An example shown below is for propane mass injected into the exhaust transfer line while sampling from raw and dilute ports (three repeats) to evaluate exhaust flow measurement on steady state basis (duration = 60 sec, Date completed January 2005). Tests Raw C3H8 g Dil C3H8 g CVS DF Raw C3H8 est Diff % % % ave % stdev COV 1.2% 0.8% 0.3% 0.9% Example of propane quality control check 38

46 EQUIPMENT CVS Study of Emissions from Yard Tractors with Diesel and LNG Fuels FREQUENCY Daily VERIFICATION PERFORMED Weekly Propane Injection CO 2 Injection Per Set-up CVS Leak Check Second by sec. Back pressure tolerance ±5 inh 2 0 Calibration System Semi-Annual Primary Standard 1% Bottle Check Analyzers Pre/Post Test Daily Zero span drifts Monthly Linearity Check Secondary System Daily Leak Check Each test CO 2 : Secondary vs. Primary Weekly Propane Injection: 6 point primary vs secondary check Semi-Annual Data Validation Each test CO 2 Balance Modal vs Integrated Bag Mass Standard Check all sensors limits/mode PM Sample Media Each test Static, tunnel & dynamic blanks Temperature Barometric Pressure & Dewpoint Sensors Daily Checks w/ ATIS; Psychrometer CALIBRATION PERFORMED Throat Pressure Absolute Pressure MFCs: Drycal Bios Meter Zero Span MFC: Drycal Bios Meter & TSI Mass Meter Performed when verification fails Sample of Verification and Calibration Quality Control Activities 39

47 Quality Assurance Project Plan Appendix B CECERT: DTQAPP QUALITY ASSURANCE PROJECT PLAN FOR MOBILE EMISSIONS RESEARCH LABRORATRY Revision: 1 January 8, 2004 Prepared for: U.S. Environmental Protection Agency Attn: Kent M. Helmer National Vehicle & Fuel Emissions 2000 Traverwood Drive Ann Arbor, MI Principal Investigator: Wayne Miller College of Engineering Center for Environmental Research and Technology University of California at Riverside Riverside, CA

48 A1 QAPP Title and Approval Title: Mobile Emissions Research Laboratory, Revision 1, January 8, 2004 Signatures indicate that this Quality Assurance Project Plan (QAPP) is approved and will be fully implemented in conducting the research project described in this document. Wayne Miller Principal Investigator CE-CERT University of California at Riverside Signature Date John Collins Quality Assurance Officer CE-CERT University of California at Riverside Signatr Date Program Officer U.S. Environmental Protection Agency Signature Date Quality Assurance Officer U.S. Environmental Protection Agency Signature Date xli

49 A2 Table of Contents A PROJECT MANAGEMENT A4 Project Organization A5 Project Background A6 Project Description and Schedule A7 Data Quality Objectives and Criteria A7.1 Quality Objectives...Error! Bookmark not defined. A7.2 Measurement Performance Criteria...Error! Bookmark not defined. A8 Special Training Requirements A9 Documentation and Records B MEASUREMENTS AND DATA ACQUISITION B1 Sampling Design B2 Sampling Methods...Error! Bookmark not defined. B3 Sample Handling and Custody...Error! Bookmark not defined. B4 Analytical Methods Requirements...Error! Bookmark not defined. B5 Quality Control Requirements B6 Instrument Testing, Inspection, and Maintenance B7 Instrument Calibration Frequency B8 Acceptance of Supplies and Consumables B9 Non-direct Measurements B10 Data Management C MANAGEMENT ASSESSMENTS C1 Assessment and Response Actions C1.1 Technical Systems Audits C1.2 Performance Evaluation Audits C2 Reports to Management D DATA VALIDATION D1 Data Validation Criteria D2 Data Validation Methods D3 Reconciliation with Data Quality Objectives A3 Distribution List Program Manager, EPA QA Manager, EPA Wayne Miller, CE-CERT Kent Johnson, CE-CERT John Collins, CE-CERT Dave Gemmill, CE-CERT Appendix A References Appendix B Minimum Detection Limits Appendix C Balance Protocol A1 R1.1 Appendix D MEL SOP s xlii

50 A4 Project Organization A PROJECT MANAGEMENT The organization of the project is shown in Figure A1. The Principal Investigator, Dr. Wayne Miller, handles coordination with the EPA Program Manager and establishes contract relationships with clients who need tests conducted using the facility. Mr. Kent Johnson coordinates test schedules with clients, handles day-to-day operations of the facility, supervises operations staff, conducts data reporting, implements quality control (QC) systems, and compiles QC data. Dr. John Collins reviews QC data and compliance with this QAPP. EPA Program Manager Wayne Miller Principal Investigator Kent Johnson Project Manager John Collins QA Officer Driver Don Pacocha A5 Project Background Kathy Cocker Balance Room Supervisor Figure A1. Project Organization MERL Technicians The Mobile Emissions Laboratory (MEL) was developed under sponsorship of the EPA. In order to build a system capable of measuring exhaust emissions from the fleet of on-road vehicles under real-world, on-road operating conditions. The approach taken during development of this system was to build a full flow CVS emission measurement system mounted in a trailer. The system mounted in the trailer is used to measure the exhaust emissions of the heavy-duty diesel tuck towing the trailer while operating on public roads. To the extent possible, the laboratory complies with methods and specifications for regular fixed laboratory test cell emission measurements of heavy-duty diesel engines. Development and testing of the laboratory is described in Cocker et al [1]. A6 Project Description and Schedule The purpose of this project is to operate MEL in support of various test programs. The measurement results are mass emission rates of CO 2 and criteria pollutants, which include CO, NOx, THC, NMHC, and PM. The equipment to be tested and the load activity cycles depend on the specific project. The types of equipment/projects planned for testing include both Class A heavy-duty diesel trucks, stationary back-up generators, heavy-duty diesel engines with control devices, portable on-road instruments, switch type locomotives, and aircraft. A list of planned projects this year is as follows: Switch Locomotives at LA Harbor; Back Up Generators Steady State 5 Mode Cycle; Portable instrument evaluation; On-road, steady state, and chassis dyno; 43

51 On road Heavy Duty driving cycles and real world driving. The level of quality control and quality assurance reporting required comparable to that for engine certification testing. These projects require special personnel such as an experienced class A driver, qualified engineering, and technicians capable of working in an analytical environment. A7 Data Quality Objectives and Criteria A7.1 Quality Objectives The quality objective of this project is to produce emission measurement data that satisfies the requirements of the Title 40, Part 86 of the Code of Federal Regulations (CFR) as it apples to measurement of emissions from heavy-duty diesel engines. A7.2 Measurement Performance Criteria Title 40 CFR includes many performance specifications and criteria including: sampling methods instrumental methods environmental controls calibration methods and frequencies QC check methods and frequencies QC check tolerances The requirements and the methods used to satisfy the requirements are described in section B. A8 Special Training Requirements There are three categories of staff operations requiring special training. These are Driver MEL technician Balance room technician Driver. The driver of the tractor-trailer rig must have Class A license. The driver must also be able to communicate with laboratory personnel in the chase vehicle during an emission test sequence. And the driver must be able to follow the speed/time trace displayed on the driver s aid during the course of an emission test sequence. The ability to follow a test sequence and driver s aid while driving on public highways takes practice. The driver s performance is measured in terms of absolute meters of deviation in real time. Drivers are evaluated and certified as satisfactory by the Project Manager. At this time the project has only one driver. The driver is highly experienced with MEL. 44

52 MEL Technician. The MEL technicians include both staff and students. These technicians in conjunction with the Project Manager are responsible for warming up equipment, selecting and overseeing automated QC and emission test sequences, filling out data logs, data keypunching, changing gases, and handling filter samples. The MEL technicians receive on the job training from the Project Manager. Performance is evaluated by the Project Manager through direct observation, and through evaluation of cross checks and redundancies built into technical operations. Incorrect performance leads to discrepancies in cross checks of failures in QC criteria. The Project Manager sends each technician a report evaluating his performance on a regular basis. Balance Room Technicians. Balance room technicians are generally students or other part time workers. The technicians must be able to weigh substrates with quantifiable accuracy and precision. To accomplish this, balance operations are conducted following very specific protocols that produce substantial amounts of QC data. It takes practice and experience to follow these procedures correctly. New technicians are given practice sets of substrates for weighing, and practice following the balance protocols. The technicians to the Project Manager report results for the practice sets. The Project Manager evaluates the practice results and identifies any problems. Problems are corrected and the new technicians try again with a new practice set, until Project Manager is satisfied that they are able to follow the protocols correctly. A9 Documentation and Records Documentation of the program and its results include Standard Operating Procedure (SOPs), Checklists, Log Books, Data Files, Reports to Management, and Reports to Clients. Paper records and computer records will be maintained by CE-CERT for a period of at least 5 years. Standard Operating Procedures MEL Operating SOPs. Safety Check: Truck and Trailer Generator start-up Analytical bench start-up Pre-test set-up Test operations Post test shutdown Shutdown: to ground power Shutdown: total MicroBalance Protocol Log Books. Paper records are kept in binders holding checklist forms and in bound logbooks. Checklists are maintained for Start-up Calibration Shutdown Test operations Test QC 45

53 Logbooks are maintained for Environmental logbook recording environmental conditions during each test Filter sample identification logbook Equipment tested log (e.g., trucks, generators, etc) Project logs book: history of tests, checks, repairs, etc. Fuel log book: quantities and types of fuel, vehicle or generator Microbalance log book: recording all filter weight and balance QC data Filter Sampling Log:filter Ids and filter weight data associated with each test Computer Files. Equipment test selections, configurations, and test sequences are automated by configuration files. Each configuration file specifies a complete sequence of test and QC operations. During the sequence of operations specified in a configuration files, the data from all channels is recorded at 1Hz or faster rate in a raw test results data file. The configuration file and associated data file are identified by test ID number. The test ID numbers are assigned based on date and time of day. Each day of testing consists of a pre-test calibration and post-test calibration. If multiple tests are done in one data the post-test calibration becomes the pre-test calibration for the next test. For example a typical triplicate test day will look like the following: pre-test calibration and QC configuration pre-test calibration and QC raw test original data Test 1: equipment test configuration with integral post test calibration Test 1: equipment test raw test original data with integral post test calibration Test 2: equipment test configuration with integral post test calibration Test 2: equipment test raw test original data with integral post test calibration Test 3: equipment test configuration with integral post test calibration Test 3: equipment test raw test original data with integral post test calibration In addition to these files generated by the data acquisition system there is a particle Filter Log File and an environmental data check log that must be key punched. The Filter Log file that associates filter identification numbers with sampling locations and times throughout the test, and with filter weight gains. The environmental data log verifies barometric pressure, dew point temperature, ambient temperature, and RH with lab values for at least one point during a day of testing. These raw test original data files are post processed to produce original QC summaries and original data summaries for review by the Project Manager. The postprocessed files include: one test file for 1-second modal data one test file for uncorrected integrated cycle data, calibration data, MFC data, etc 46

54 In addition to the individual test files, the QC data from each test and the integrated cycle data are each appended to databases containing all test results since the beginning of MEL operations. The following two database files are maintained from the raw data file post processing: database of integrated test results (emissions u grams per cycle, ppm and environmental) database of integrated test QC results (zeros, spans, and test specific data) During data validation and review, comments and corrections are recorded by editing a COPY of the raw test original data files. The edited, corrected files are called raw test validated data files, and these validated data files are used to generate and replace entries in the post processed files. In addition to the databases of integrated results and test QC results, there are database files containing histories of QC checks not related to specific tests. These include: Propane injection mass balance CO 2 injection mass balance NOx converter efficiency Blended gas calibration checks Analyzer linearity checks All of the files above, including the raw data and the validated data are maintained for a period of at least five years. Reports to Management. The Project Manager reports project status every two weeks to the Principal Investigator, but the contents of these meetings are not recorded. Test result displays and QC result displays are generated by the Project Manager every two weeks. These displays are available for review continuously by the Principal Investigator on file servers over the computer network. A report of projects completed, data summaries, and QC summaries are reported annually to the EPA Program Manager. Reports to Clients. Reports to clients consist of delivery of validated data including: integrated cycle emissions data 1-second modal emission data (if requested) depending on the request of the client. Supporting documentation and QC data summary documentation is available upon the request of the client. 47

55 B1 Sampling Design B MEASUREMENTS AND DATA ACQUISITION MEL is used to conduct test programs for various clients. The selection of equipment to be tested, manufacturer, model year, and the selection of emission tests depend on the aim of the client. The engine activities to be tested are combinations of steady state loads and engine speeds, specific speed and load versus time cycles, or real-world driving. Activities cycles generally consist of standard activity cycles specified by ISO, EPA, or ARB. The specific cycles for testing are selected by the client. For some on-road transient cycle emission tests, the measurements are designed to match a speed/time driving trace originally designed for implementation on a dynamometer. Due to the constraints imposed by available public highways, MEL is not always able to directly implement established dynamometer traces as originally derived. However, MEL is able to re-order segments of the dynamometer driving traces and to sample selected portions of those traces, so that emissions equivalent to the originally dynamometer trace are collected. MEL emission measurements collected over the UDDS cycle while operating on a dynamometer at the Los Angeles MTA, have been compared with MEL measurement collected while operating over the rearranged UDDS cycle on public roads. The data comparisons verify the validity of the cycle re-arrangement method Cocker et. al. [1]. B2 Sampling Methods The sampling methods, data calculation equations, quality assurance specifications, and quality control criteria required for measurement of mass emission rates from heavyduty diesel engines is specified in great detail in Title 40, Part 86 of the Code of Federal Regulations. These methods are briefly discussed in the sections below. B3 Sample Handling and Custody Gas phase samples are drawn directly from the vehicle exhaust into the CVS sampling system where they are diluted with filtered ambient air. The diluted samples are analyzed second by second (modal data) and also collected into sample bags over defined phases of the test cycles. The sample bags are then analyzed (integrated data). Sample handling and sample custody by the operator are not applicable. Particle samples are extracted from the CVS dilution tunnel, diluted further in a secondary dilution system, and collected on Teflon filters. The temperature of the air and filter in the filter collection system is controlled as specified in the CFR. The use of particle filters requires numerous sample handling and custody issues to prevent contamination, to ensure proper sample identification, and to ensure equilibration requirements of the CFR. The filter handling procedures are described in MEL SOP s Appendix D. Throughout the procedures filters are handled only using forceps. The filters never leave the custody of CE-CERT. The steps involved are as follows : unused filter numbers are assigned to new plastic Petri dishes 48

56 new filters are placed into the numbered plastic Petri dishes the Petri dishes with unexposed filters are placed into a temperature and humidity controlled equilibration chamber after 12 hours the filters are weighed filters are reweighed daily until two filter weights within 3 ug of each other filters with consistent weights are transported to the trailer for testing prior to the test, filters are transferred from Petri dishes to filter holders in the trailer, the filter ID numbers are recorded on the particle Filter Data log sheet at the end of the test filters are returned to their original Petri dishes the Petri dishes of the exposed filters are marked with an X the filters are returned to the balance room and placed in the equilibration chamber and equilibrated 12 to 72 hours the exposed filters are weighed once daily until two weights with 3 ug are obtained The procedures for weighing the filters are detailed in Balance Protocol A1 R1.1 Appendix C. These procedures define specific detailed sequences for checking and recording the equilibration chamber conditions, zero and spanning the balance, weighing reference objects, weighing filters, performing zero and span checks, performing replicates. In summary, filter weights are only considered valid if environmental chamber conditions are within tolerance, and the filter weight was preceded and followed by a valid zero and valid span. B4 Analytical Methods Requirements The specific instruments chosen to implement the analytical requirements of 40 CFR are described in Cocker et al [1, 2]. A general discussion of the equipment is included in this section. Gas phase The regulated emissions measured are dry CO, CO 2 and CH 4 and wet THC and NOx. Particles secondary dilution designed to achieve 47C +/- 5C equilibration at 22 C and 45% RH for 12 to 72 hours before sample weighing single weight measurement resolution 10 ug. or better Vehicle Speed For transient driving cycles, vehicle speed is critical measurement required to demonstrate compliance with the driving cycle. MEL uses a doppler sensor to record speed over the ground. Position Position measurements are not required by the CFR. However, when collecting onroad emissions data, geographic position and elevation including height collected by a GPS unit is useful auxiliary data. The GPS also provides speed over the ground that is useful for verification of speed data collected from the Doppler sensor. 49

57 B5 Quality Control Requirements The primary QC requirements imposed by 40 CFR are as follows: Span gas bottles: certified to within 1% tolerance CVS checks (propane mass balance) performed weekly and performing within 2% tolerance Linearity checks performed monthly on all analyzers performing within 2% of point from full scale down to 15% of full scale for each instrument range scale in use Zero checks and span checks on appropriate range before and after each test In addition to those checks, MEL requires the following QC activities: Once during all test/calibration days CVS leak check Comparison of bag data with integrated modal data, agreement to with-in 5% for CO 2 and NOx, and 15% for CO (tolerances depend on species). THC are typically different by 30% due to wet modal sample and dry bag sample. The percent difference is checked, but there are no tolerance requirements. Collections of blanks with each day s filter sampling: o trip blanks: carried from lab to trailer, but never leaves Petri dish o static blanks: loaded into sampler and unloaded immediately, o dynamic blanks: loaded and left in unused (non-flowing) sampler channels during sampling, o tunnel blanks: samples dilution air during full flow test with no exhaust flow. B6 Instrument Testing, Inspection, and Maintenance Generator: follow schedule of maintenance specified by manufacturer. Every 250 hours, oil filter, fuel filter, oil analysis Class A truck: follow schedule of maintenance specified by manufacturer based on accumulated mileage or months of operation. Heated filters: check monthly, replace if needed HEPA filter: checked every 6 months HEPA pre-filter: replace monthly Pump diaphragms: replace annually B7 Instrument Calibration Frequency Criteria gas analyzers: instruments are zeroed and spanned before every test and checked after every test. Analyzer linearity is checked monthly. Constant Volume Sampler Flow Rate: The venturi was calibrated at purchase. The flow is verified weekly in combination with the analytical system weekly by the propane injection mass balance and by the CO 2 injection mass balance. Mass flow controllers: calibrated every six months against a BIOS dry cell meter. 50

58 Temperature probes: checked annually against an ice bath, boiling water and elevated temperature calibrator. Barometric pressure sensors: checked every test against an aneroid barometer, and against nearest airport barometric pressure. CVS Differential pressure sensors: electronic self-calibration every day. Other differential pressure sensors calibrated annually against U-tube manometers. Humidity and dew point temperature: once per test day against a motorized psychrometer and nearest airport ATIS recordings. B8 Acceptance of Supplies and Consumables Gas supplies. The fuel gas used for flame ionization detectors is Praxair FID FUEL, a blend of 40% hydrogen and 60% helium. Zero Air is Praxair THC FREE and certified to have less than 0.1 ppm CO, CO 2, and THC. Zero Nitrogen is Praxair Vehicle Emissions Zero or 4.8 Grade nitrogen and is certified to have less than 0.1 ppm NOx and THC, and less than 0.5ppm CO. Single component calibration gases are Praxair Primary Master and are certified by Praxair to within ± 1%. Multi component calibration gases are Praxair Primary Master and are certified by Praxair to within ± 1% for each component. Liquid CO 2 cylinder of 100% CO 2 are Praxair and are certified by Praxair to within ± 1%. Independent audit gases and blended gases are obtained from Scott-Marin. Each component in the blend is certified to within ± 1%. When calibration gases or audit gases are replaced, the new bottle is compared with the old bottle. If the measured response of the new bottle is not with 1.5% of the named concentration on the bottle, it is returned to the supplier for re-analysis. Filter Media Sample filters are 47-mm ringed Pall Teflo 47mm with a 2.0um pore size. The filters are inspected for holes and obvious defects during weighing, but are not otherwise tested. B9 Non-direct Measurements Airport measurements. Ambient temperature, dew point temperature, humidity, and barometric pressure data are obtained daily from the nearest airport at least once during each test day. These data are recorded in logbooks, and keypunched into the data files associated with each test. These data serve as a crosscheck of comparable data measured using MEL equipment. Engine broadcast data. For engines equipped with broadcast data streams, engine data such as percent load fuel rate vehicle speed 51

59 engine speed throttle position coolant temperature boost pressure are recorded at a minimum rate of 1 Hz. B10 Data Management MEL data. The system of paper records and computer data files is described in section A9. Descriptions of validation criteria are given in section D. The current section describes the mechanics of generating, transferring, and archiving those data files. The configuration files that specify the test to be performed are edited on the consoles in the emissions trailer before a test is to be conducted. The raw test original data are generated by the data acquisition computers during the test. Operator comments can be entered during the test and, at the end, comments are incorporated into the raw test original data file, which is finalized at that point. Data remains on the MEL data acquisition computers until the trailer returns to the CE-CERT parking lot. Once the trailer returns to the parking lot, a copy of the computer data files are transferred to Project Managers computer for validation. Once validated the files are moved to a validated directory. Every night the raw and validated data is automatically copied to CE- CERT s server over a private network. At the end of a test, particle Filter Log File has been filled out with the data to associate individual filter identification numbers with sampling locations and times throughout the test. After the final weights have been obtained by the balance room personnel, the final weights are entered onto the particle Filter Log sheet. The Filter Logs are then keypunched to create Filter Log files. The Filter Logs are each keypunched twice, by two separate individuals, and the two files compared in order to catch keypunch transcription errors. The hard copy of the particle Filter Log is returned to the trailer and stored in binders along with log data sheets for the test. Post processing programs are run within one week to perform emission calculations, generate test summaries and displays, and generate QC summaries and displays. Copies of the original data files are created so that they can be edited. The edited copies after review and correction are referred to as raw test validated data. The post processing programs are re-run on the validated files to regenerate emission calculations, test summaries, QC displays, etc., in order to continue the review process. The edited files are flagged to indicate validation status. The validated files are used to generate the final reportable results: integrated cycle emissions and 1-sec modal emissions with validation status included, within one month of data collection. All of these original, intermediate, and edited files are maintained on the network server. To provide redundancy, the files on the network server are automatically mirrored to a second hard drive daily. The contents of the mirror drive are also automatically backed up to tape daily. 52

60 Handwritten logs including the particle Filter Logs remain in binders in the trailer. The binders are archived approximately annually. C MANAGEMENT ASSESSMENTS C1 Assessment and Response Actions C1.1 Technical Systems Audits EPA personnel performed a technical systems audit on 10/22-23/2002. The auditors reported no major problems, were very satisfied with the operations, the quality control procedures and results, and the validation procedures being applied to the data. The primary recommendations were to document the operating and QC procedures in more detail. There are no current plans for additional audits. C1.2 Performance Evaluation Audits The MEL has not received a performance audit for the various individual components of its operations, i.e., the MEL has not received an audit of the type where auditors supply independent audit test gases mixtures, simulated balance samples, independently verify flow rates, etc. However, MEL has participated in five correlation studies where overall mass emission rate measurements of a vehicle or generator are made by two independent systems and the results compared. The correlation studies performed to date include: CE-CERT Stationary Source group, six comparisons at steady state, June 2002 through September 2003, EPA Method 5: report being prepared. Los Angeles Metropolitan Transportation Authority (MTA), February 2003, full flow CVS system, heavy-duty truck, transients and steady state: criteria gasses agreed well; found discrepancies for particulate mass emissions. Los Angeles Metropolitan Transportation Authority (MTA), March 2003, full flow CVS system, heavy-duty truck, transients: criteria gasses agreed well; particulate mass emissions agreed well. Engine Fuels and Emissions Engineering Inc, (EFEE), March 2003, steady state, RAVEM mini-dilution system: MEL identified problems with RAVEM probe tip selection. GE Environmental Energy Research Group, April 2003, two different mini-diluter systems, steady state: data not yet received from GE EER. Engine Fuels and Emissions Engineering Inc, (EFEE), October 2003, steady state, RAVEM mini-dilution system: experiment failed due to RAVEM probe location problems. Sierra Instruments, April-May 2003, Sierra BG-2 proportional sampling system, particulate mass only, steady state: particulate mass agreed within 10%. 53

61 C2 Reports to Management Project Manager reviews test and performance status after every test. The Project Manager updates QC charts every week. Results out of tolerance receive immediate attention. The Project Manager and MEL Staff work to eliminate out of tolerance responses results before further testing of client equipment. The Project Manager meets with the Principal Investigator every two weeks to review project status. The contents of these meetings are not recorded. Status reports to the EPA Program Manager will be prepared quarterly. These reports will include: inventory of tests completed and validation status summary of QC checks development of improvements or additions to operations D DATA VALIDATION Before data is reported it passes through several validation levels and is marked with appropriate validation status. Data Validation Level indicates the level or extent of data screening that has been applied. It does not indicate whether or not the data has met the screening criteria. Validation Status indicates how well the data meets or fails to meet the screening criteria. Data Validation Levels are as follows: Level 0: data is converted to engineering units but has not been subject to any review Level 1: data passes a specific list of quality control criteria. Level 2: data reflects review and comments by CE-CERT and client data users Level 3: data incorporates review and comment resulting from peer review Data Validation Status Codes are currently applied to entire test records, i.e. the collection of all species and parameters making up an emissions test. The Validation Status Codes used in this program are as follows: Valid (V): test data meets all validation criteria at the validation level indicated Suspect (S): test data initially fails to meet some aspect of validation criteria and are under investigation. These data will eventually become designated as Valid, Partially Valid, or Invalid. Partially Valid (P): The majority of the data meet all screening criteria, but one or more species fail to meet all criteria (e.g. THC analyzer did not function but all other data are valid). Or; all data meet the majority of screening criteria, but one or more criteria failed (e.g., time since last linearity check exceeds 30 days). A description of the reason for partially valid is recorded in the comment section of the database. Invalid (I): Substantial failures to meet validation criteria 54

62 The specific reasons for marking a test Suspect, Partial, or Invalid are stored in the comments section of the data record for that test. D1 Data Validation Criteria Validation Level 0 Lowest level of data processing and manipulation. Data corrected from measured value to engineering units and process calculations are made to get mass and corrected values at process conditions. Validation Level 1A (through 9/30/03) In order for data to be Valid at Level 1, it must meet the following criteria: Data inspected for good calibrations (zero and span drift during test days, weekly linearity checks), Bag data compared with modal data when bag sampling used (expected tolerance is less than 5% for NOx, CO 2, and 10% CO, but non for CH 4 or THC). Subject to change depending on test type, CVS verification performed and with-in 2% for both CO 2 gravimetric injection and propane CFO injection, Steady state mode standard deviations less than 5% on CO 2, NOx and CO and 15% for THC and CH 4. Secondary filter flow standard deviations less than 0.5% of span Secondary propane CFO injection with-in 5% NOx converter checks greater than 90% Linearity checks with-in 2% of point Dew point temperature and barometric pressure checked with local ATIS system Temperatures and standard deviations with-in expectations PM mass tare and filter weights checked for stability Repeated tests compared for test-to-test variation. Steady state test-to-test variation was determined to be valid if less than 5% for CO 2 and NOx, 15% for CO and 25% for THC assuming the total mass was greater than 10 times the Minimum Detection Limits specified in Appendix B. Transient test-to-test variation is still under investigation for transient tests and new vehicle types. Validation Level 1B (through 10/1/03 through present) Beginning on 10/1/03 additional checks and validation criteria were incorporated into the system. In order to be considered valid at Level 1 after this date, the data must pass all of the validation criteria described in Level 1A above, and must also meet the following criteria: Secondary Media Blanks (PM 2.5u, DNPH, Quartz, and C1-C10) started. Blanks include dynamic, static, trip, and tunnel. Blank data used to evaluate problem with handling and/or running operations of the secondary system. Tolerance still under investigation. Dynamic Titration Calibrator checks with a multi blended primary standard bottle 55

63 for CO, CO 2, NOx, and CH 4 and a single blend for THC. All species (THC, CO, NOx and CO 2 ) fall with-in 2% of audit bottle. PM filter weighing chamber RH and temperature conditions are saved in 5 min and 1 hr averaged data with min, max, and stdev for archived records. This data is published on the web at [ select Chamber Room Status to display history. Validation Level 2 After Level 1 validation has been applied, data is delivered to users within CE-CERT and to client users. After approximately 6 months, the users are questioned about whether they have found any problems or suspect any problems in the emissions data, based on their experience using the data. Any problems are investigated, comments are incorporated into database if appropriate, and the data are marked with Validation Level 2. Validation Level 3 In some cases data is distributed beyond the immediate clients to the general scientific community either through peer-reviewed publications or as a result of use by the regulatory community. Such data sometimes becomes subject to substantial scrutiny such as comparisons with data from related measurement studies or theoretical studies. It is not uncommon for peer-review to identify issues that were not apparent from within the original study in isolation. Data is released to peer review will be marked with Validation Level 3 approximately one year after its release to the public. Reviewer comments will be investigated and incorporated into the database as necessary. D2 Data Validation Methods The Project Manager on a weekly basis conducts data validation for routine calibrations and on a monthly basis performs data validation for tests. The validation process is broken down into two main steps, post process calibration data (includes CFO propane injection, linearity checks, NOx converter checks, CO 2 gravimetric injection, and calibrator audit bottle check) and post process test data (includes analyzer zero/span drift, bag vs. integrated modal, modal standard deviations, bag sample standard deviations, ambient reference checks, test to test variability, PM stability and PM RH with-in tolerance). The validation process involves the following steps: Routine calibration data validation process: Each calibration test is set up as a routine automated test cycle that has a data file and a configuration file that runs the test. Operators run the tests and data is stored to the data acquisition hard drive for later post processing. With-in one week the calibration files are copied to the Project Manager s computer for post processing. The post processing involves running a program that extracts the important information from the raw data file using the configuration file. All test comments are permanently archived in the 56

64 configuration files for future post processing. The results from the post processing are appended to the main database. The database maintains a chronological order of all the data calibrations, tests, experiments etc. This makes it easy to see if before and after calibrations were valid or invalid. Once all the calibrations are processed, the Project Manager views the main database file for the results. One calibration cycle will increase the database size by one row. Each calibration row is appended to another control charting file to track and document its performance. The Project Manager evaluates the tolerance requirement for each test and updates the main database validation level to a 1 after reviewing each file. If the specific calibration test meets the tolerance limits the data validation specification is updated to valid otherwise invalid. An invalid does not imply the data is unusable, but that there was some reason the calibration did not meet the tolerance specifications. If an invalid is recorded in the main data base file, a comment will be provided in the main database to understand its severity. The raw and post processed data files are copied to the server automatically at the end of each day. Tests and experiment data validation process: Each test and experiment is set up with an automated test cycle that has a data file and a configuration file that runs the test. Operators run the tests and data is stored to the hard drive for later post processing. With-in one month the test files are copied to the project manager s computer for post processing. The post processing involves running a program that extracts the important information from the raw data file using the configuration file. All test comments are permanently archived in the configuration files for future post processing. The results from the post processing are appended to the main database. The database maintains a chronological order of all the data calibrations, tests, experiments etc. This makes it easy to see if before and after calibrations were valid or invalid. Once all the tests are processed, the Project Manager views the main database file for the results. By inspection one can see that before and after calibrations were valid and invalid. If invalid the test cycle data validation field is updated with invalid and a comment added as to what tolerance test failed. Next the test data is checked for zero/span drift and stability. This is done by taking all the calibration data (pre test and post each test for the day), and verifying that zero and span drift is with-in tolerance using the calibration database. The calibration database maintains a record of every calibration performed on each analyzer and a comment section. If the zero/span drift or stability fails, the main database data validation field is updated with an invalid and a comment is added to explain the type of error and severity. Next the modal and integrated bag data is compared. A history file contains statistical calculations to view quickly the variability between integrated bag and modal data. Some tests do not require bags thus there will be no comparison made. If the results are invalid the test cycle is updated with invalid and a comment added as to what tolerance test failed. Ambient bag results are compared to past and present data. If the ambient bag 57

65 appears to be an outlier (still being identified) the data is considered invalid and the main database is updated. Test to test variability check. The main database file is imported into Access where mass specific emission data is calculated by category type. Records of average, count, and standard deviation are output for many database fields. These data are then further analyzed in excel to find test-to-test variability quickly. If the test-to-test variability is established for the test type and is outside the tolerance specified the data validation field is updated with invalid and a comment as to the failure mode is recorded in the comment fields. Operator comment saying some aspect of the test was not performed correctly will invalidate the data and be assigned to the data validation field with invalid and a comment added explaining the nature of the problem. In this case the data validation process is not required unless the problem only affect a part of the data. All issues that arise during any test are immediately investigated under the Project Manger s direction and with the help of his staff. A second database is maintained that has tests identified as invalid. In this file resolution comments are provided as to what may be wrong. If a test can be repaired the solution is added in a separate column. This maintains the trial an error and effort of the MEL staff to maintain and keep up with the operations of the lab. Data that is fixed and becomes valid is updated in the main test database and a comment about the solution is added. The comment explains what was wrong and what was done to repair the data. Also a comment is added if there is a level a seriousness for the type of test to see if it should be used for the project. Data that cannot be fixed remains in the data set with an invalid label in the data validation record. All data that is either identified as valid or invalid is available to the user in the main database. D3 Reconciliation with Data Quality Objectives The process for informing the user of problems that may affect utility of data is available in the main database. There is enough information in the database to provide an explanation and a conversation between the end user and the project manager in order to understand the limitations of its use. 58

66 Appendix C Raw Test Data for Various Yard Tractors Selected Test Values for C8.3L-2003 (Tier 2) & C8.3L-2001 (Tier 1) Engines 59

67 Selected Test Values for ISB-5.9L-2005 & CG-250 LNG Engines 60

68 Appendix D Selected ECM Data from the ISB Engine 61