Aerotherm Project  7212
TECHNICAL SUPPORT  AND ASSISTANCE  FOR THE  DEVELOPMENT
      OF STANDARDS  OF PERFORMANCE  FOR STATIONARY
              INTERNAL COMBUSTION ENGINES
            S.  Youngblood, M.  MacDonald, L. Cooper
            Acurex Corporation/Aerotherm Division
                      485 Clyde  Avenue
               Mountain View,  California  94042
                        January 1977
                  AEROTHERM REPORT TR-76-234
                         Prepared  for

              EPA Project Officer - John R. Busik
              EPA Task Officer - John N. McDermon

        Industrial Environmental  Research Laboratories
             U.S. Environmental Protection Agency
                   Research Triangle Park
                    North Carolina  27711

                     Contract  68-01-3158
                          Task 14

-------
                                          TABLE OF CONTENTS


Section                                                                                        Page

   1       INTRODUCTION 	       1

           1.1  Scope of Work	       1
           1.2  Summary of Results   	       2

           1.2.1  Additions to Data  Base	       2
           1.2.2  Scatter Analysis   	       2
           1.2.3  Field Applicability  	       4

   2       ADDITIONS TO DATA BASE	       5

   3       SCATTER ANALYSIS:   SOURCES OF VARIATIONS IN NO  LEVELS FROM LARGE BORE
           ENGINES   	x	       7
           3.1  Effect of Measurement Methods
           3.1.1  Previous Studies of Exhaust Measurement Variation  	      8
           3.1.2  Variations in NOX Emissions to the NOX Instrument	     11
           3.1.3  Variations in NOX Levels Related to Sampling Procedures  	     19
           3.1.4  Comparison of Manufacturer's Measurement Error   	     29

           3.2  Effect of Ambient Conditions on NOV	     33
                                                  X

           3.2.1  Effect of Ambients on NOX Emissions from 1C Engines	     34
           3.2.2  Selection of Ambient Correction Factors for Large  Bore Engines   	     35
           3.2.3  Effect of Ambient Correction on Reduction of Data  Scatter  	     69

           '3.3  Effect of Engine Variability	     75

           3.3.1  Production Variations 	     75
           3.3.2  Model Variations  	     77
           3.3.3  Variations with Number of Cylinders	     80
           3.3.4  Variations in NOX Level Due to Other Engine Variables  	     80

           FIELD APPLICABILITY	     87

           4.1  Summary of Manufacturers' "114" Responses 	     87
           4.2  Summary of DEMA and AGA Comments on  Control Techniques	     89

           REFERENCES   	     91

           NOMENCLATURE 	     93
                                                 m

-------
                                       LIST OF ILLUSTRATIONS




Figure



   1        Nondispersive Infrared Absorption Analyzer

                                                                                                1 5

   2        Chemiluminescent  Analyzer  	

                                                                                                on

   3        Schematic  of Exhaust  Sampling  System 	


                                                                                                ?7
   4        Sampling Practices  	



   5        Measurement  Error Relative  to  EPA Procedure   	



   6        1969  Humidity Range  in Engine  Test Cells,  Dearborn,  Michigan  	      36



   7        Effect  of  Ambient Humidity  on  1C engines NO   emissions  	      37



   8        Effect  of  Ambient Temperature  on 1C Engine NO  Levels   	      38
                                                        X


   9        Effect  of  Humidity on  Emissions  Scatter for  Six HD Gasoline Engines  	      39



  10        Effect  of  Humidity and Temperature on  Emissions Scatter for Six HD Diesel

           Engines  	      40



  11        Effect  of  Humidity on  A/F Ratio  and NO  Emissions  from  HD Gasoline Engines ...      42
                                                 X


  12        Effect  of  Temperature  on  A/F Ratio and NOV Emissions for HD Gasoline Engines .  .      44
                                                   X


  13        Effect  of  Barometer  Pressure on  A/F Ratio  and NO  Emissions for HD Gasoline

           Engines	      45



  14        Exhaust Gas  Concentration Versus A/F 	      46



  15        Effect  of  Fuel Distribution on NO Emissions	      47



  16        Effect  of  Ignition Timing and  A/F on NO Exhaust Emissions, 1600 rmp, 23 bhp

           Road at mbt	      47



  17       Effect  of A/F Ratio  and Ambient  Humidity on  NO  Emissions of a LD Gasoline

           Engine	      49



  18       Comparison of SI  Ambient Humidity Correction Factors 	      50



  19       Water to Fuel Range  vs. Ratio  Air-to-Fuel   	      52



  20       Effect of Humidity and Temperature on NO Emission  for Different CI Engines  ...      54



  21       Comparison of CI  and SI Ambient  Humidity Correction Factors  	      56



  22       Correction Factors for Temperature for CI  Engines    	      57



  23       Gas Turbine Humidity Correction  Factor 	     59



  24       Observed Ambient Humidity  Influence on NO   Production  	      60
                                                    X


  25       Comparison of Existing Gas  Turbine Ambient Temperature Correction with HD Diesel

           Temperature Correction 	      62



  26       Gas Turbine Temperature  Correction Factors  	      63



  27       Relationship  of  NOX Level  and Load to f/a  Ratio for a Turbocharged Diesel Engine

           Operating at Two Ambient Temperatures  	      64
                                                 IV

-------
                                  LIST OF ILLUSTRATIONS  Concluded.





Figure                                                                                         Page



  28       Effect of Ambient Correction on Scatter in NOX Emissions  of Uncontrolled

           4-TC Engines	      72



  29       NO  Production Variation with Number of Cylinders  	      81
             X


  30       Variation in NO  Level with Number of Cylinders	      82



  31       Variation in NO  Level with Speed	      83



  32       NO  Variation with Manifold Temperature  	      84



  33       N0x Variation with Torque (BMEP)	      86

-------
                                           LIST OF TABLES


Table                                                                                          Pa9e

   1        Additions to Data Base ................      .......     .       6
   2       Accuracy of NO  Measurements .......................       10

   3       Instruments Used By Manufacturers to Measure NO  Emissions   .          ...       12

   4       Sources of NO  Instrument Error  ........................     16

   5       Comparison of CL and NDIR, NDIR/NDUV Measurements  ...............     18

   6a      Sources of Sampling Error:  The Sample Transfer  ................     21

   6b      Sources of Sampling Error:  Instrument Related ..........     ...       22

   7       Comparison of Sampling Practices ......................     24

   8       Sources of Error for DEMA, SAE, and EMD Emission  Practices ...........     26

   9       Large Bore Engine Manufacturer's Sampling Practice ..............     30

  10       Manufacturer's Sampling Error as a  Percent,  Relative to EPA Procedure  .....     31

  11       Ambient Humidity Correction Factors for SI  Engines .- ..............     48

  12       Existing 1C Engine Ambient Correction Factors for Application to Large Bore
           Engines  ..................................     70

  13       Scatter in NO  Emissions of Uncontrolled 4-TC Engines  ............     71

  14       Variations in NO  Emissions from GMC/EMD Production Engines  ..........     76

  15       Potential  Scatter in NO  Emissions  from Production Engines Due  to Ambient
           Humidity and Temperature Variations  .....................     78

  16       Variation  in NOX Due to Model Differences  .    ...     ...........     79
                                                 vi

-------
                                             SECTION 1
                                           INTRODUCTION
       In June 1974 the Aerotherm Division of Acurex Corporation was  engaged by the EPA to gather
data and information available on the effectiveness of various emission control  techniques for re-
ducing emissions of NO , CO, and HC from stationary internal  combustion engines.   These data were
documented in a draft Standards Support and Environmental Impact Statement (SSEIS) which was submit-
ted to the Industrial Studies Branch (ESED/OAQPS) in March 1976.
       The draft SSEIS has undergone review within EPA and has also been circulated among various in-
dustry groups representing manufacturers and users of stationary internal  combustion engines.   Since
the development of standards of performance for stationary internal combustion  engines  is so complex,
a number of questions and issues have been raised during this review  of the Standards Support and En-
vironmental Impact Statement.
       The objective of this current study is to answer the questions and issues  which  have arisen
as a result of the draft document review and to perform various services for the  EPA in connection
with the pending recommendation and promulgation of standards of control.   This supplemental report
documents work completed under contract 68-01-3158, Task 14.   The following section briefly outlines
the scope of work completed under this task order.

1.1    SCOPE OF WORK
       The work completed under this task order is divided into three tasks as  described in the work
plan.  These tasks are listed below by the section number of this report where  they are discussed.
       •   Section 2:  The purpose of this task was to collect and incorporate  into the existing data
           base additional emissions data supplied by the eight manufacturers of  large  bore engines.
           The responses to a June 16, 1976 Section "114" request for information were  the primary
           source of this information
       t   Section 3:  The sources of data scatter for the reported emissions data were evaluated
           according to the following breakdown:

-------
       -   Measurement Techniques:  Assess the extent to which scatter in the data is attributable
           to variations in the measurement practices among the eight manufacturers
       -   Variations in Ambient Conditions:  Determine to what extent variations in ambient inlet
           air humidity, temperature, and pressure affect exhaust emissions, and investigate the
           application of correction factors to adjust emissions to standard inlet air conditions
       —   Engine Variability:  Investigate to what extent the scatter in the data is related to;
           (a) inherent differences among engines of a common generic configuration, and (b) differ-
           ences among identical engines and models of the same manufacturers
       •   Section 4:  The final task under this contract was to reevaluate the field applicability
           of the various controls that were identified in the draft document

 1 .2    SUMMARY OF RESULTS
       The following are highlights of the major conclusions  reached during the completion  of this
 task order.

 1.2.1  Additions to Data Base
       •   The data base of uncontrolled engines containing ambient information has  increased nearly
           200 percent after incorporation of the additional  data  obtained.   The corresponding data
           for controlled engines increased 100 percent.   Thus,  substantial  emissions data  now exists
           for correction to standard ambient inlet conditions
       •   No manufacturer reported the  response of an  engine's  exhaust emissions  to  systematic
           variations of inlet air conditions.   Thus,  ambient correction  factors could  not  be derived
           directly from the data

1.2.2  Scatter Analysis

Measurement Methods

       •    Measurement practices were  shown  to  contribute  significant  uncertainty  for some  of the
           manufacturer's  data

       •    The error  in each manufacturer's  measurement method was  estimated.   It  is  recommended
           that the reported N0x levels  be  banded by the  appropriate uncertainty limits

-------
Effect of Ambient Conditions
       •   Variations in ambient conditions were shown to contribute significantly to the scatter in
           reported emission levels.  This scatter can be reduced by correcting observed levels to
           standard ambient conditions
       •   Ambient correction factors were selected from the literature based on smaller bore engines
           to be applied to large bore data, since no factors existed for large bore engines.  These
           factors are:
           —   Corrections_for humidity only in SI engines.  Factors could not be derived for changes
               in ambient pressure or temperature due to inconsistent variations in response to these
               inlet parameters
           —   Corrections for temperature and humidity for CI engines.  These factors were based on
               specific engine types  (e.g., 4-TC, 2-BS, etc.).  No factors exist for the effect of
               ambient pressure variations on emissions from CI engines
       •   Scatter in 4-TC diesel and dual fuel uncontrolled emissions was decreased slightly by
           application of an ambient  correction factor.  Scatter increased somewhat for 4-TC uncon-
           trolled natural gas emissions.  Thus, other sources of scatter, such as measurement
           methods and engine variability are believed to account for most of the data scatter
       t   Existing correction factors (based on small bore engines) were recommended for application
           to large bore data, until  manufacturers can obtain data to develop factors based on large
           bore engines.  It is recommended that standardized limits be established for those ambient
           parameters for which no factors presently exist

Effect of Engine Variability
       •   Two manufacturers reported that NO  levels vary from 3 to 8 percent for measurements of
                                             X
           identical engines (production models).  However, some of the reported variation was shown
           to possibly be related to  variations in ambient conditions
       •   No clear trend could be established for the effect of number of cylinders on emission
           levels for either diesel or natural gas engines
       •   NO  emissions for any type engine (given strokes/cycle, fuel, and air charging) vary more
             A
           from manufacturer to manufacturer than they do among models within a manufacturer's line.
           It is expected that differences among manufacturers are related to differences in speed,
           BMEP, and manifold air temperature

-------
      •   Limited  data  show that NO  emissions from 4-TC diesel  and  dual  fuel  engines  decrease some-



          what with  increasing speed




      •   NO  emissions from 4-TC, natural gas (SI) engines increase directly  with  design  manifold



          air temperature.  NO  levels from diesel and dual fuel  (CI) engines  increase  slightly with



          design manifold air temperature




      •   No trend in NO  emissions could be established for differences  in BMEP among  diesel,  dual



          fuel, and natural gas engines





.2.3   Field Applicability




      •   There is a general  consensus  among manufacturers  that 1 to 3 years is required to fully



         develop N0x control  techniques, including the evaluation of engine durability.  This  com-



         pares  to 9 to 15 months  estimated in the draft SSEIS




      •   In general, manufacturers reported potential  cost impacts of NO  controls (e.g., increased
                                                                        X


         maintenance) rather  than  technical  limitations that would prevent the application of po-



         tential  NOX control  techniques

-------
                                             SECTION 2
                                      ADDITIONS TO DATA BASE
       Following the completion of the draft SSEIS and its presentation to the NAPCTAC in March,  EPA
received comments from engine manufacturers and the AGA.   As a result of these comments EPA sent
Section 114 letters on June 16, 1976 to the relevant engine manufacturers requesting additional  in-
formation.  Seven of eight manufacturers of large bore engines responded with new or revised exhaust
emissions data, information on ambient conditions during emission tests, sales data, and field operat-
ing experience.  These new data are summarized in Table 1 and compared with the existing data base.
Entries in this table are engine models for which we have controlled and/or uncontrolled data. As
this table illustrates, the uncontrolled data base containing ambient information has more than
doubled and the controlled data with ambients has increased 150 percent.  Thus, a substantial data
base exists for which ambient variations in temperature and humidity can be considered; however,  none
of the manufacturers supplied emissions data, systematically, as a function of ambient variations.
Therefore, a correction factor for ambients cannot be developed directly.
       Two manufacturers, DeLaval and White-Alco, supplied emissions data (uncontrolled) for the  first
time.  Their submission included data for many of their present production models.   Cooper Energy
Services, White Superior, and Ingersoll-Rand also reported exhaust emissions (both  uncontrolled and
controlled) for additional, previously unmeasured engine models.  Both Colt and Waukesha reported
additional emissions data (uncontrolled and controlled) for previously measured engine models.
       These new emissions data have been incorporated into the existing data base.  The uncontrol-
led data containing ambient temperature and humidity were used to perform the scatter analysis re-
ported in Section 3.2.3.  Further evaluation of the controlled data base (with ambients) will be
completed using a computer analysis developed under contract 68-02-2530.

-------
TABLE 1.  ADDITIONS TO DATA BASE
Fuel
Strokes
Air Charging
Existing data,
uncontrol led
Additional data,
uncontrol led
Existing data,
control led
Additional data,
controlled
Diesel
2
BS
5

3

TC
5
4
4
4
4
NA
2

2

TC
7
13
6
7
Dual Fuel
2
BS




TC
2
1
2
1
4
NA




TC
3
3
3
2
Gas
2
BS
1
1
1
1
TC
1
11
1
11
4
NA
4
1
4
1
TC
8
5
5
2
Total
38
39
31
29
U_>
CO
1—
Fuel
Strokes
Air Charging
Anbient data,
existing uncontrolled
Ambient data,
new uncontrolled
Ambient data,
existing controlled
Ambient data,
new controlled
Diesel
2
BS
2

2

TC
1
4
1
4
4
NA
0

0

TC
3
13
3
6
Duel Fuel
2
BS




TC
2
1
2
1
4
NA




TC
3
3
3
2
Gas
2
BS
0
1
0
1
TC
0
11
0
11
4
NA
3
1
3
1
TC
2
4
2
1
Total
16
38
16
27

-------
                                             SECTION 3
                         SCATTER ANALYSIS:  SOURCES OF VARIATIONS IN NOX
                                  LEVELS FROM LARGE BORE ENGINES

       This task is concerned with refining the exhaust emissions data which were reported in the
draft SSEIS and in subsequent responses from the engine manufacturers.  The draft reported the ef-
fectiveness of various emission control techniques as part of an effort to develop standards  of per-
formance to limit exhaust emissions from stationary reciprocating 1C engines.   The reported data,
however, may not be directly comparable because the manufacturers used different measurement  tech-
niques and the ambient conditions varied from test-to-test.   Consequently, these measurements methods
are examined in detail in Section 3.1 to determine to what extent the reported emission  levels differ
due to differences in the procedures.  Similarly, Section 3.2 seeks relationships between  exhaust
emissions and varying ambient conditions.  Specifically that section examines:  (1)  how  variations
in ambient temperature, humidity, and pressure affect NO  emissions, (2) what correction factors
                                                        X
exist for adjusting measured exhaust emissions to standard conditions, and (3) the extent  to  which
scatter in emissions levels can be reduced by applying an appropriate ambient correction factor.
This current study will make a preliminary evaluation of ambient scatter by attempting to  correct
the uncontrolled data base only.  The controlled data will be corrected and evaluated under contract
68-02-2530.
       In addition to these two types of scatter, differences in emissions data may be attributable
to inherent differences among engines of a common generic configuration, or inherent differences
among identical engines of the same manufacturers and model.  These kinds of scatter are examined
in Section 3.3 (after correction for ambient conditions where possible).  A fourth source  of  scatter
is related to the extent to which the data represent a "cut-and-try" approach to emission  reduction,
rather than a systematic approach to obtain the greatest reduction in emissions.  This kind of scat-
ter will  be examined under contract 68-02-2530, after the controlled data base has been  corrected for
variations in measurement methods and ambient conditions.
       Thus, the following sections evaluate scatter due to:  (1) measurement methods, (2) ambient
conditions, and (3) engine variability.

-------
 3.1     EFFECT  OF  MEASUREMENT METHODS




        Previous studies have indicated that sampling instrumentation and procedures significantly



 affect  emission levels.   Furthermore, no one standard procedure has been adopted by all of the  man-



 ufacturers  reporting emissions.  That is, the eight manufacturers who reported emission data  used



 either  chemiluminescent (CL) or nondispersive infrared (NDIR) and ultraviolet (NDUV) instruments and



 one  of  four emission measurement procedures (SAE, EPA, EMD, or DEMA).   Therefore, this section  will



 present  a detailed discussion of the instrument and sample acquisition practices used by the  engine



 manufacturers.  The purpose of this discussion is to identify possible variations in exhaust  emission



 levels  attributable to particular measurement equipment and/or procedures.  Specifically, this  sec-



 tion will :




        t    Establish which instruments, sampling trains, and procedures were used by each manufac-



            turer who reported NO  emissions
                                X



        •    Summarize the potential  sources of uncertainty relating to each measurement practice




        •    Evaluate the variability among manufacturers' emissions data due to these uncertainties




 The  following  discussion will begin by illustrating typical measurement uncertainties in present



 sampling practices.  This analysis  will show that significant uncertainties in measurements can arise



 due  to  differences in both instrument and sampling procedures.  Therefore, both of these sources of



 error will  be  discussed as they apply to the instruments and procedures used by the eight large bore



 engine  manufacturers.  Then these practices will be compared and the uncertainty in each manufactur-



 er's NO  measurements estimated.





 3.1.1   Previous Studies of Exhaust  Measurement Variation




        Before  evaluating each manufacturer's sampling practice, it is of interest to examine  the



 present  "state-of-the-art" of internal  combustion engine exhaust measurements of NO .   Three  recent



 studies have been conducted to compare the measurements of NO  made by different laboratories from



 the same emission source.   In two of these studies N0x measurements were made, simultaneously,  in



 the same laboratory using identical  procedures.   In the other study, the same emission source was



 sent to each laboratory.   These studies illustrate the magnitude of emission variations attributable\



 to instruments and sampling procedures.   They also indicate the reproducibility of emissions  measure-



ments,  both  within a  laboratory and  amongst different laboratories.  The results of these studies



will  serve as  a basis  for  comparing  potential  data variations due to the measurement practices of



the eight large bore  engine manufacturers considered in the current study.

-------
       A series of cooperative emissions tests was conducted by the Coordinating Research Council
(CRC) to evaluate measurement methods used to analyze diesel exhaust emissions from truck-size
engines.  In Phase III of this program, six laboratories sent sampling teams to one location to make
simultaneous measurements (using NDIR analyzers) of a multicylinder engine (Reference 1).  The engine
used in this study was a 6 cylinder, 300 cubic inch, 4 stroke direct injection diesel.   The proce-
dures that were used during this test program to measure NO, CO, and C0« evolved into SAE Recommended
Practice J-177 (Reference 2).  In Phase IV of that program, the same engine was circulated to 15
laboratories to:  (1) verify that the generally good agreement of Phase III emission levels were the
result of improved sampling procedures, and (2) obtain NO/NO  data using CL analyzers (Reference 3).
                                                            X
       Table 2 shows the results of these two cooperative tests.  Although the Phase III variations
appear reasonable, the Phase IV results indicate poor agreement amongst the laboratories.  These
larger errors were attributed to poor calibration procedures (span gases out of spec, instruments
not calibrated) and possibly some variation in engine performance.  In addition, it was noted that
the average NO concentration measured at rated load by the CL analyzers was approximately 23-percent
lower than that measured by NDIR.
       A more recent cooperative test program conducted by the CRC evaluated EPA's revised heavy duty
diesel engine NO/NO  measurement methods and instrumentation (Reference 4).  Six participants made a
series of NO/NO  measurements on a multicylinder engine, simultaneously, and produced the range of
error shown below.  A range is shown because the results were analyzed in three different groupings:
(1) all data, (2) those which remained after eliminating questionable results from participants who
encounter sampling problems, and finally, (3) those which were left after excluding both questionable
data and results obtained from instruments with long sample transfer times.

                                 13 mode      St'd Dev.
                                 g/hp-hr      g/hp-hr      St'd Dev./Mean, %
                        NO     8.03 - 8.21    .42 - .29       5.9 - 3.5
                        NOX    8.05 - 8.16    .30 - .17       3.8 - 2.1

All but one of the participants used a CL analyzer, and the results from the one NDIR analyzer (a
new reduced interference design) were equivalent to these from the CL analyzers.  Thus, based on
simultaneous tests using the same instruments, sampling practices, and emissions source, errors in
emission measurements among laboratories ranged from 3 to 6 percent for NO, and 2 to 4 percent for
NO .  In addition, the repeatability within a laboratory ranged from 2 to 7 percent for NO and 1 to
3 percent for N0x-

-------
TABLE 2.  ACCURACY OF NO  MEASUREMENTS  (REFERENCES  1  AND  3)
                        A
Source
NO:
NDIR
CL
N02:
NDUV
CL
Phase III
6 Labs, Simultaneous Measurements
Standard Deviation
as % of Average

5
—

—
—
Spread in Data
as % of Average

10
—

—
—
Phase IV
15 Labs, Round Robin
Standard Deviation
as % of Average

8
36

21
88
Spread in Data
as % of Average

27
15

39
166

-------
       In contrast to the above program, the eight large bore engine manufacturers used one of four
test procedures (SAE, EPA, DEMA, or EMD) and either CL or NDIR/NDUV instruments.  Furthermore, the
NDIR's were not reduced interference designs.  Hence, their data are likely to vary more than the
truck engine results due to measurement practices.  The following discussion will first examine the
sources of such variations from differences in both instruments and sampling practices.  Then an
attempt will be made to suggest uncertainty bounds on the emissions data from each manufacturer.

3.1.2  Variations in NOX Emissions Related to the NOX Instrument
       Since the development of commercial chemiluminescent analyzers (1971), various studies have
been conducted to compare their operation with the already established NDIR analyzer.  All of these
studies have shown the NDIR analyzer to record consistently higher levels of NO than the CL analyzer
for a given source (References 3,5, and 6).  Three of the large bore engine manufacturers reported
N0x emissions using NDIR's and the other five used CL analyzers.
       Table 3 shows the NO  instruments used by each of the manufacturers.  Scott Research Labora-
                           X
tories made the emission measurements for Ingersoll-Rand and White-Alco, since neither manufacturer
presently owns emission measurement equipment.  Note that an electrochemical instrument was used by
Shell Oil Research in 1971 to measure NO  from one Cooper-Bessemer and one Ingersoll-Rand engine.
None of the manufacturers has used this instrument since then; therefore, no attempt will be made to
correlate those emissions with instrument differences,.  Only Waukesha and GMC/EMD continue to measure
NO  emissions with NDIR analyzers, and Waukesha has recently acquired a CL analyzer.  Thus, all but
one manufacturer will be using CL instruments in the future.  In the following paragraphs, the prin-
ciple of operation of the NDIR and CL instruments and associated sources of error will be briefly
reviewed.  Then the results of NDIR/CL comparisons will be summarized to seek a method of expressing
emissions on an ''equivalent instrument" basis.

3.1.2.1  NDIR Instrumentation
       The NDIR instrument was introducted nearly 20 years ago and has continued to be used widely
as a CO and COp detector.  In addition, it was used extensively to detect NO, HC, and SO,,, but other
methods are now supplanting it for these species.  Its principle of operation depends upon absorption
of infrared radiation by the gaseous sample.  Figure 1 illustrates a typical NDIR instrument.  Built-
in optical and gaseous filters are used to produce a narrow infrared beam bandwidth to compensate for
interference (absorption) by other constituents.
       Despite these precautions, water vapor and COp, to some extent, may cause positive interfer-
ences (high readings) even though refrigerant and chemical driers are used to remove water vapor.
                                                 11

-------
                              TABLE 3.  INSTRUMENTS USED BY MANUFACTURERS TO MEASURE NO  EMISSIONS
                                                                                       A
ro
Manufacturer
Colt
Cooper Energy
De Laval
Ingersoll-Randa
Waukesha
White Superior
(Div. Cooper Energy)
White-Alcoa
GMC/EMD
CL (with Thermal Converter)
NO/NO
A
Thermo Electron 10A
Thermo Electron 10A
Scott 125
Scott 125

Scott 125
Scott 125

NDIR/NDUV
NO, N02/N02

NO: Beckman
IR 18A


NO: Horiba
AIA-2


NO/Beckman 31 5B
N02/Beckman 255
Electrochemical
N0/N02

Envirometrics NS-200



NO: Dynasciences NX-130
N02: Envirometrics NS-200


               Measurements made by Scott Environmental Technology  Lab.

-------
                                   Infrared
                                   source
                         Reference
                            cell
OJ
                      Recorder
                      signal
                                                      Diaphragm
                                                      distended
                                                                                       Sample out
   — Component
      of interest
O —Other molecules
                                               Control unit
                                Figure 1.   Nondispersive infrared absorption  analyzer.

-------
 Desicants, however, have been found to  cause  significant  interferences  as  well  as  water vapor (see



 later discussion and Section 3.1.3).   In  addition  to  these  problems,  the  response  of the NDIR instru-



 ment to the specie of interest is  nonlinear in  some instrument  designs, necessitating a carefully



 constructed calibration curve using at  least  four, and  preferably  six to  eight, calibration gases.






 3.1.2.2  Chemiluminescent Analyzers




        Chemiluminescent analyzers,  in contrast  to  NDIR  instruments, have  been developed only recently



 (1971) for source sampling.   Nevertheless,  the  CL  instrument has gained increasing application  for



 the measurement of NO  (NO + N02)  and NO.   In this type of  instrument ozone, Og, is  reacted with



 nitric oxide, NO, to produce a chemically excited  state of  NO*, which emits light  as  it decays  to



 stable NO,,.  The intensity of this  emitted  light is proportional to the NO concentration present  in



 the sample.  The analyzer uses a photomultiplier to detect  the  light.




        Figure 2 is a schematic of  a CL  analyzer and illustrates the reaction of ozone with  NO in  the



 reaction chamber of the instrument.  The instrument is easier to calibrate than some  NDIR's due to



 its linear response to  NO, and thus requires fewer calibration  gases.  Note that a N02 -* NO con-



 verter is depicted in Figure 2.  This device consists of a  stainless  steel tube which is heated to



 ~1200°F and essentially converts all N0? in the sample to NO.   In  this way just NO or both  NO and



 N02 can be measured,  depending on whether the sample gas is fed directly to the reaction chamber  or



 passed through the converter first.  The N02 level is then  deduced by subtracting  the NO level  from



 the total  NO  reading.




        Potential  problems  encountered in this instrument include quenching of the  excited N0? by



 other species,  converter  inefficiencies, and interferences  caused by  chemiluminescence of other gases.



 In  general,  quenching (by  C02  or H20) is negligible in CL instruments; water is removed from the



 sample  before  analysis  and the  use  of low pressure (vacuum) reaction  chambers reduces quenching by



 C02.  Some  quenching  problems  have  been observed during measurement of fuel-rich automotive exhausts.



 Quenching effects,  however,  have not been observed during measurements on  truck-size  diesels, whose



 exhausts are similar  to those of large bore stationary engines  (i.e., -15  percent  oxygen).




       Converter problems are avoided by thermal conditioning of new  converters and by making regular



 checks of converter efficiency  (>90 percent).  Interferences are minimized by the  instrument manu-



 facturer through the choice of spectral  filters and photodetectors.





 3.1.2.3  Sources of Instrument Error





       Sources of error for each of the  instruments listed  in Table 3 are  summarized  in  Table 4.



Note that neither the quenching effect nor the converter problems of  the CL analyzer  should occur
                                                 14

-------
en
          Sample gas
                                 Photo-

                               multiplier

                                  tube
                                      N02 * NO


                                      Converter
                                                                                         Ozone
                                                                  1
-- L"ht
                                                                                               Principle of Operation
                                                                                                  NO + 03 - NO* + 02
                                                                                                  NO* •+ N09 + hv (Light)
                                                                                           $       c.     f-
                           o


                           <
Vent (to  vacuum source)
                         Reaction chamber
                                                 Figure 2.   Chemiluminescent analyzer.

-------
                                TABLE 4.
SOURCES OF N0x  INSTRUMENT ERROR
Chemil uminescent,
N0/N02
• Quenching of CL by
third body reactions3
(C02, H20)


• N0,/N0 converter13
L.
- NOX -* Ng at high
temperature, low
02, high CO and HC
— N02 •* NO incomplete
due to improper
converter condi-
tioning
- NH3 -* NO, N02
at high temperature
especially with
some catalytic con-
verters in rich ex-
hausts
- N02 ->• NO at low
temperatures, low 02
NDIR, NOC
• Water vapor in-
terference causes
higher NOX readings.
Also COg inter-
ference to some
extent; optical
filters and water
removal minimize
interferences


• Use of chemical
driers lead to
N02 adsorbtion
and N02 * NO
conversion
• Calibration curve
non-linear, requires
frequent checks us-
ing 4 to 6 calibra-
tion gases


NDUV, N02d
• Interference caused
by carbon particles
and other species ,
results in a drift-
ing, high reading.













Electrochemical ,e
N0/N02
• Not much appl i ca-
tion in source
sampl ing


• SOg interferences


• Nonlinearity in
30 to 300 ppm
range
• Susceptible to
error due to
temperature
changes






aNot present in properly designed instruments.   (See References 7,  8,  9.)
Condition CL converter to avoid error; perform regular converter efficiency.  Most of these errors occur
 in rich exhausts unlike the lean exhausts (~15 percent oxygen) of  large bore  engines.   (See References 4,
 10.)
cSee References 5 and 10
 See Reference 10
eSee Reference 11

-------
when measuring large bore engine exhausts with a properly operated and maintained CL analyzer.  Pres-



ent NDIR analyzers, on the other hand, are susceptible to errors due to interference despite sample



conditioning to remove water.  This interference combined with the relative ease of operation and



accuracy of CL analyzers has led to an increasing perference of CL's for NO  measurements.  As an
                                                                           X


example, EPA's proposed revisions to the Heavy Duty Diesel and Gasoline Engine Sampling Procedure



specify CL analyzers.  The following section summarizes comparisons of CL and NDIR/NDUV measurements



of N0x-  These comparisons will be the basis for expressing NDIR/NDUV measurements as equivalent CL



levels.





3.1.2.4  Correlations of CL to NDIR/DNUV




       As discussed above, NDIR analyzers are subject to water vapor interference despite sample



conditioning to remove water vapor.  These interferences cause the instrument to indicate higher



N0x levels than are actually present.  A comprehensive study conducted by TRW (with Scott Research



Labs)  using constant volume sampling (CVS) of automative exhausts determined that





                                     [NO]NDIR = 1.07 [NO]CL - 4.2                                (1)





where  [NO] denotes NO concentration in ppm (Reference 5).  Both analyzers had similar water removal



systems (condenser and chemical driers); hence, the difference was due to a gas interference only.



In addition to this finding, the study established that chemical driers promoted NO,, -*- NO conversion.



Comparison of an NDIR with water removal to a CL without water removal indicated that





                                     [NO]NDIR   1.12 [NO]CL+ 8.4                                (2)





       Furthermore, a comparison of two CL analyzers, one (CLg) preceded by a water trap and chemical



drier  and the other (CL.) without either, indicated that





                                   [NOxJ    = .873 [N0x]    '* 6.6                                (3)

                                          13               M




Thus,  the net effect of the drier and water trap is to destroy about 12 percent of the NO  in the
                                                                                         X


sample.




       The results of this study and others that corroborate these effects are summarized in Table 5.



The bias of NDIR analyzer ranges from 3 to 30-percent higher based on these studies.  As this table



indicates, the particular correction that applies to any one of the three manufacturers that used an



NDIR analyzer (and NDUV for N02) will  depend on their sampling setup and procedures (e.g., type and



location of water trap, and the NO, present).  Therefore, the appropriate correction for NDIR instru-



ment bias  will  be identified after a discussion of each manufacturer's sampling practices.
                                                 17

-------
                             TABLE 5.   COMPARISON OF CL AND NDIR, NDIR/NDUV MEASUREMENTS
Source
TRW Study
(Reference 5)
Clemens Memo
(Reference 12)
SAE 750204
Phase IV
(Reference 3)
SAE 730213
(Reference 6)
CRC Report #487
(Reference 4)
Comparison
N°xCL/N°xNDIR/NDUV = '82
NOxCL/N°xNDIR/NDUV = "75
NOxCL/NOxNDIR = '9
(NOCL/NONDIR = '88)
NOCL/NONDIR = .77
N°xCL/N°xNDIR/DUV = *94
N°xCL/N°xNDIR = "985
Comments
No water removal from CL sample.
Water trap and drier used for both CL and NDIR.
Used TRW comparison for NO (no water removal for CL)
and assumed that N02 was 2 percent of total NO in a
diesel exhaust. x
Result questionable due to other sampling errors.
None of eight large-bore manufacturers used DUVa
instrument.
NDIR instrument is "third" generation, low interfer-
ence analyzer, not comparable to NDIR's used by
large-bore manufacturers.
co
          Dispersive Ultraviolet

-------
3-1-3  Variations in NOX Levels Related to Sampling Procedures
       In addition to the choice of instrumentation, differences in sampling procedures can cause
variations in reported emission levels.  This was clearly illustrated in Table 2 for the Phase IV
cooperative tests where 15 laboratories made measurements of an identical source.  Since four basic
sampling procedures (DEMA, SAE, EMD, EPA) were used by the eight large bore engine manufacturers,
some variation between their reported emission levels can probably be attributed to differences in
their sampling practices.  The sources of these errors will be identified and the four sampling pro-
cedures compared and evaluated in this section.

3.1.3.1  Sources of Sampling Error
       The two major problems encountered in measuring engine exhausts are:  (1) chemical  changes
that occur during transfer of the sample to the analyzer, and (2) error due to improper operation of
the analyzer.  Figure 3 is a simple schematic of an engine sampling system.  In transferring the ex-
haust gas sample to the analyzer, care must be taken to ensure that all of the NO  (NO + N02) or NO
(when only an NDIR is used) originating from the engine exhaust reaches the analyzer.  For these
reasons sample lines are heated to prevent condensation or kinetic conversion of constituents.  High
sample flowrates (short sample residence times) are maintained to minimize sample degradation during
its transfer from the engine to the analyzer, and water removal devices are employed to minimize
instrument interferences from water vapor.  Similarly, it is important that the analyzer be calibra-
ted and all components, such as NO^^NO converters (on CL instruments), be functioning properly.
Therefore, adequate analyzer specifications and calibration procedures are essential  for accurate
emission measurements.
       Table 6a presents a more detailed summary of sources of sample transfer error.  As  this table
indicates, heated sampling lines coupled with low sample residence times are essential to  preserve
the initial amounts of NO and N02 contained in the sample gas.  Furthermore, leak checks of the en-
tire sampling system (particularly on the vacuum side) should always be performed to assure that the
sample gas reaches the analyzer undiluted.  Water removal is required with NDIR analyzers; therefore,
care must be taken to minimize the time available for NO to be converted to N02 which can  then be
absorbed by the condensed water from the sample gas.  Chemical driers (desiccants) are an  unacceptable
water removal device since they promote the NO,, -*• NO reaction and absorb N02 (see Section  3.1.2).

       Table 6b summarizes important analyzer related procedures required for accurate measurements.
Care must also be exercised in operating emission measurement instruments.  Frequent calibrations
should be performed using accurate, certified blends of calibration gases.  Zero and span  checks
                                                 19

-------
                                                 Sample Line
1N5
O
              Exhaust
                 Probe
                    Enqine
                                                                       Water
                                                                      Removal
NO  Analyzer
  A
                                       Figure  3.   Schematic  of  exhaust  sampling  system.

-------
            TABLE 6a.   SOURCES OF SAMPLING ERROR:   THE SAMPLE TRANSFER
                               Sampling System
        Source
          Error
                                   Correction
t  Unheated
   sampling line
•  Large sample
   residence times
•  Sample line
   connections and
   fittings
•  Water removal

   —  Refrigerant
   —  Chemical
      drier (desiccant)
NO •* N02i N02 adsorbed
during water removal:
Observed 40-percent
loss of NO when sample
residence times were
35-40 sec (Reference 13)
Heated Line: N02 ad-
sorbed during water re-
moval : 6-7% NOy loss
for system response
times9 greater than
35 sec (Reference 13)
Cold  Line: 12-percent
NOX loss  for sample
residence time of
13 sec (Reference 14)
Leaks that dilute sample
gas: Observed 25-percent
loss of NOv due to pre-
filter leak (Reference 13)
  2 adsorbed in condenser;
NO
(Reference 5)
N02 -> NO and drier "eats"
N02«  Negative or positive
errors with cold lines
(SAE procedure) depending
on how drier conditioned
(Reference 5 and 13)
                              Heat line to 375°F
                              Use short sample line
                              and/or flowrates to
                              limit system response
                              to 15 sec or less.
                              Leak check system be-
                              fore testing
CL: Locate condenser
after converter.
NDIR: Maintain high
sample flow through
condenser, remove
H20 as it forms.

Do not use chemical
driers
 System response  time  =  sample  residence  time  + instrument  response  time
                                     21

-------
           TABLE 6b.  SOURCES OF SAMPLING ERROR;  INSTRUMENT RELATED
                                Instrumentation
         Source
          Error
                                                              Correction
 •   Instrument drift,
    ozone  shortage  (CL),
    plugged  capillary
 •  Converter
   malfunction (CL)
t  Visual rather than
   stripchart readings
t  Calibration and
   span gases
 Low NO   readings
      /\
Not all N0£ converted to
NO, results in  low NOX
reading also possible for
NO to exceed NOX  levels
Analyzer meter only ac-
curate to ±3 percent.
Visual averaging less
consistent than chart
averaging (Reference 13)
Change of constituents
with time, or erroneous
certification
Calibrate instrument
frequently.  Zero and
span before and after
each measurement for
all instrument ranges.
Instrument should meet
minimum performance
criteria.
Perform converter
checks regularly.  Use
known standard to check
converter efficiency.
Use stripchart, aver-
age levels over an
interval for which
steady state condi-
tions exist.
Use certified gases of
specified blends.
Cross-check span and
calibration gases.
                                     22

-------
also serve to indicate potential instrument problems as well as necessary gain adjustments.   Span
gases should be frequently cross-checked with calibration gases or checked against a NBS standard
since even certified gases can be in error.
       Some calibration curves for NDIR analyzers are nonlinear; therefore, several calibration gases
(minimum of four) should be used to calibrate the instrument (Reference 13).  These instruments also
experience some hysterysis after sitting unused; therefore, frequent calibration is necessary.   Also,
calibration points should be curve-fit using a higher-order polynomial.  In addition, efficiencies
of N0x converters (on CL analyzers) should be checked regularly.  Finally, strip chart recordings
of data are superior to visual analyzer meter readings since a permanent record is produced, and
emission levels can be averaged over a time interval more accurately and consistently.

3.1.3.2  Sampling Procedures:  EPA, DEMA, SAE, EMD
       Having completed this brief review of sources of sampling error, we can examine each  of  the
four measurement practices presently used by the eight large bore engine manufacturers.   Table  7 is
a comparison of the EPA, DEMA, SAE, and EMD emission measurement practices (References 15,  16,  17,
and 18).  The information for the EPA procedures is based on the revised EPA Heavy Duty Diesel  Emis-
sions Regulations as proposed in the Federal Register, Volume 41, No. 101, May 24, 1976. This  re-
vised procedure requires CL analyzers for NO/NO  measurement, establishes instrument and calibration
                                               X
specifications, and defines sample transfer configurations and practices.
       This comprehensive EPA regulation was recently verified through cooperative testing  by both
manufacturers of mobile diesel engines and the EPA-Ann Arbor mobile sources laboratory (Reference 4).
The six participants in this program (see Section 3.1.1) made essentially equivalent NO  measurements
for a series of eight tests.  Two participants, however, did experience a small, but consistent NO/NO
crossover (NO levels greater than NO  levels).  Nevertheless, the standard deviation of measured NO
                                    x                                                             x
levels was generally small, ranging from 2 to 4 percent (of the mean level).  Therefore, the EPA
practice will be the basis of comparison for the other measurement practices.
       The potential sources of measurement error of the other three procedures (OEMA, SAE,  and EMD)
are summarized in Table 8.  The sampling trains for these three procedures, as well as the  EPA  setup
are illustrated in Figure 4.  The DEMA procedure specifies a CL analyzer and the SAE/EMD procedures
require an NDIR instrument.  The EMD practice also utilizes an NDUV analyzer, in series after the
NDIR, to measure NO,, in the sample.
       Table 8 clearly indicates that there are several sources of measurement error possible for the
DEMA procedure, largely as a result of its failure to define instrument and sample transfer practices.
                                                 23

-------
               TABLE  7.   COMPARISON  OF  SAMPLING  PRACTICES
Test Procedure

DEMA:  No time between modes  specified
EPA:   5 minutes
SAE:   20 minutes
EMD:   30 minutes

DEMA:  Data recorded and averaged over 10 minutes
EPA:   Data recorded over last 2 minutes
SAE:   Data recorded over last 5 minutes
EMD:   Data recorded over last 3 minutes


Instrument Calibration - DEMA and EPA are CL;  SAE  is NDIR; EMD is NDIR/NDUV

DEMA:  Analyzers calibrated semimonthly.   No details of calibration given.
EPA:   Check NOX converter once per week  (must be  at least 90-percent
       efficient); leak check system, calibrate analyzer, check sample
       line residence time, and quench check every 30 days
SAE:   Calibrate monthly
EMD:   Checked and calibrated monthly, semiannually and annually

DEMA:  Number of calibration  gases not specified
EPA:   Span analyzer with calibration gases  having nominal concentrations
       of 30, 60, and 90 percent of full  scale concentrations
SAE:   Use calibration gases  that are 25, 50,  75,  and 100 percent of
       instrument range used.
EMD:   Use 4 calibration gases

DEMA:  No specifications for  calibration  or  span gas blends and dilutents
EPA:   Blends and dilutents specified and accurate within 2 percent of
       true concentration or  traceable within  1 percent of NBS blends.
       Span gas traceable to  within 1 percent  of calibration gas.
SAE:   ±2 percent accuracy on gas analysis certification
EMD:   ±2 percent accuracy blends and dilutents defined by EMD

DEMA:  No analyzer specifications required
EPA:   Specifications for response-time,  precision, noise, zero and
       span drift, and linearity
SAE:   Accuracy should be ±2  percent full-scale deflection or better
EMD:   Same as SAE
                                  24

-------
                         TABLE 7.  Concluded
Sample Transfer

DEMA:  Stainless steel (S.S.) probe, configuration  not  specified
EPA:   Specifies number and size of holes in S.S. probe
SAE:   S.S. probe, no configuration specified
END:   S.S. probe, configuration specified

DEMA:  Sampling line material not specified
EPA:   Stainless steel, teflon, or proven inert
SAE:   Stainless steel or teflon after exhaust cooled
EMD:   Stainless steel

DEMA:  Line heated to 200°F for NOX (375°F for HCy)
EPA:   375°F + 18°F - 9°F
SAE:   Unheated
EMD:   Unheated

DEMA:  No sample line length specified
EPA:   <50 feet or sample line residence and instrument response  time  is
       less than or equal to 15 seconds
SAE:   No length specified
EMD:   No length specified

DEMA:  Chemical driers can be used
EPA:   Not allowed
SAE:   Refrigerant and chemical drier specified
EMD:   Same as SAE

DEMA:  Recommend water removal at probe or at analyzer  before  N02 -> NO converter
EPA:   Condenser at analyzer after converter
SAE:   Condenser and drier before analyzer
EMD:   Same as SAE for NDIR.  NDUV does not use water removal.
                                  25

-------
                               TABLE 8.   SOURCES  OF  ERROR  FOR DEMA, SAE, EMD EMISSION PRACTICES
                                   DEMA  (CL)
ro
CJ1
•  Unheated sampling lines permitted

•  No specification of sample residence
   time in sampling line, or system re-
   sponse time

•  Leak checks are not specified

f  Water removal device can be located
   at analyzer, but before F^-^NO
   absorption in water trap

•  Chemical driers permitted

•  No instrument specifications

•  Converter efficiency checks not
   specified

t  No calibration procedures specifed

•  Calibration and span gas specifica-
   tions not defined
                                                    EMD (NDIR/NDUV)/SAE (NDIR)
•  Unheated sample lines

0  No sample residence or system
   response time specified

•  Leak checks not specified

0  Allow chemical  drier

0  Calibration procedures not specific
   (e.g., what constitutes out-of-
   calibration, how calibration points
   are curve fit,  etc.)

0  Calibration and span gas blends and
   dilutents not specified by SAE.
   EMD has "own" specifications.

-------
                     EPA-HD Diesel and Gasoline (Proposed) Regulations
nnfififlfifiy N°;
UUUUUUU/V con
-* NO j
verter J
50' or sample residence +
instrument response time <15 seconds
EMD and SAE J-177 Recommended Practices

Sample line
(Not heated)



Li




Fi Itrr »•- . -- .





CL
00
O
O
NDIR
ro
                     DEMA SETUP
                       (Near Probe)
                       Condenser
                                    Sample line not heated
                                      No length specified
                                          Filter
               >
               " NO
          Converter
                                                                                                         CL
"  -QQQQ/
                     Heated 200"F *
                     10°F no length specified
Condenser
     - NO
Converter
>
                                                                                          CL
                                                  Figure 4.  Sampling practices,

-------
 Its most obvious shortcoming is the potential  degradation of the NO and N02 in the sample gas as it



 passes from the probe through an unheated sample line (of unspecified length) and condenser before



 entering the CL analyzer   High residence times would also promote N0x loss for this configuration.



 One large bore engine manufacturer demonstrated that the sample residence time must be less than



 3 seconds to prevent significant loss of N02 in the sample (12-percent loss for a 13-second residence



 time) as it passed through a cold sample line  and condenser (see Table 6).




       Similarly, SAE and END practices can also lead to a loss of N0x as the sample passes through



 an unheated sample line, a condenser, and a desiccant before reaching the analyzer.  As discussed



 previously, the desiccant converts NOp to NO and absorbs NOg.




       Comparisons of the SAE sampling train with the proposed EPA train indicated that SAE measured



 concentrations of NO are understated by 15 to  40 percent during the first few measurement modes of



 the federal 13-mode composite cycle, then are  equal in the seventh or eighth modes, and finally are



 overstated by about 10 percent in the last modes (Reference 13).  These results and the study con-



 ducted by TRW suggest that the desiccant requires a period of  time to equilibrate.  If a new drier



 were used before each 13-mode test (permissible under present  EPA regulations), NO levels would tend



 to be low (since levels would be understated until  the drier had equilibrated, possible not until



 the last, rated speed, low power modes were being measured).  If the drier was not replaced, or if



 it was preconditioned, the NO levels would tend to  be overstated.  This is probably the case for



 those large bore engine manufacturers who used driers, since their engines are typically operated



 about an hour before they are stabilized and measurements are  begun.




       The EMD procedure is similar to the SAE procedure with  the exception that both NO and NO- are



 measured using NDIR and NDUV analyzers.  Results from Reference 5 based on automotive exhausts indi-



 cated that NDIR/NDUV N0x measurements were about 20-percent higher than CL readings (no water removal



 for CL sample).  Apparently, the positive biases (interference and drier) more than offset potential



 N0x loss due to the cold sampling in that study.  In addition  to the positive NDIR biases, NDUV



 analyzers have a history of stability problems, usually manifested as a drifting  (high) reading.



 One study attributed this error to interference due to carbon  particle buildup (see Reference 10).





 3.1.3.3  Summary and Conclusion of DEMA, SAE,  and EMD Practices





       In summary,  it appears that the DEMA practices will generally lead to negative errors due to



N0x loss in transporting the sample from the engine to the analyzer.  At this time the absolute error



associated with this  practice is unknown relative to the EPA procedure.  Limited  results (see Table 5)
                                                28

-------
suggest this error could be as much as -10 to -15 percent depending on the amount of N0? in the



sample (estimated to be 2 to 10 percent of the total NOV) and the extent that NO is converted to NO,
                                                       X                                           c,


during the sample transfer.




       The positive bias of the SAE procedure gives a +12-percent error in the NO reading.   The EMD



procedure, which measures both NO and N02 causes a +20-percent error in the NO  reading.  Nevertheless,



both procedures could experience as much as a -40 percent error in the NO or NO  reading if the



chemical drier has not stabilized.  Thus, the uncertainties in N0x measurements for the SAE and EMD



practices appear to be significantly greater than the DEMA practice.




       Finally, the EPA procedures defined for the heavy duty diesel engines are clearly superior to



the other three procedures.  This practice appears to have minimized the various sources of measure-



ment error and should give more accurate and consistent NO  readings.  Therefore, each of the manu-



facturer's sampling practices will be evaluated in the following section by comparing it to the EPA



procedure.





3.1.4  Comparison of Manufacturer's Measurement Error




       Table 9 summarizes each manufacturer's practice in terms of procedures that could lead to



measurement error.  Note that three of the four DEMA manufacturers using cold sample lines  have lo-



cated their water traps near the analyzer instead of at the engine as is recommended in the DEMA



practice.  Therefore, this setup will promote additional loss of NO  in the exhaust sample.  In
                                                                   X


addition, Colt and Waukesha have relatively high sample residence times which will also promote



NO  loss.
  X



       The potential errors in measurement resulting from these practices are estimated in  Table 10



relative to the EPA procedure.  (Blanks appear in Table lOfor items which do not apply to a particu-



lar manufacturer.)  These errors are depicted for each manufacturer in Figure 5.  Errors due to sys-



tem leaks are not included in the figure since this error cannot be generalized.  Also, the error



bands were constructed assuming that errors of the same sign were additive.  Note that these uncer-



tainties substantially exceed the 4-percent scatter in NO  levels observed during verification tests
                                                         X


of the EPA procedure.




       NDIR and NDIR/NDUV instrument biases have been included in this figure and are based on the



TRW study (Reference 5).  Note that the potential negative 40-percent error of Table 10 is not included



in Figure 5 since the manufacturers generally operate the engine long enough before measurements are



taken to condition the drier.  As the data for Waukesha indicates, sampling errors can be as large as



instrument bias; although in the case of the Cooper and 6MC/EMD data, instrument biases predominate.
                                                 29

-------
                                TABLE  9.   LARGE BORE ENGINE MANUFACTURER'S SAMPLING PRACTICE
Practi ce/
Manufacturer
DEMA (CL)
Colt
Cooper
Delaval
White Superior
SAE (NDIR)
Waukesha
Cooper
EMD (NDIR/NDUV)
GMC/EMD
EPA (CL)
White/Alco
Ingersoll-Rand
Sample Line
Heated
No
No
No
Yes
No
No
No

Yes
Yes
Length, ft
20
20
33
30
33
18-75
40

60-80
60
Residence
Time, Sec
14
3
2
4
-15
<3
6

6-8
6
Water Removal
Refrig
Yes
Yes
Yes
Yes
Yes
Yes
Yesa

Yes
Yes
Drier




Yes
Yes
Yesa



Location
Before
Analyzer
Before
Analyzer
Before
Analyzer
Before
Analyzer
Before
Analyzer
Before
Analyzer
Before
Analyzer

After
Converter
After
Converter
Data
Recording
Stripchart
Visual
Visual
Visual
Stripchart
Visual
Stripchart

Stripchart
Stripchart
Leak
Check
Yes
No
No
Yes
No
No
Yes

Yes
Yes
Converter Eff.
Check (CL)
2/yr
No
No
?
--
--
--

Yes
Yes
aEMD measures N0? with NDUV.  No water traps are used.

-------
                  TABLE  10.   MANUFACTURER'S SAMPLING ERROR AS A PERCENT,  RELATIVE TO EPA PROCEDURE
Manufacturer' s
Practice
DEMA
Colt
Cooper
Delaval
White Superior
SAE
Waukesha
Cooper
EMD
GMC/EMD
Sampling
Cold line, large
residence time
-12a

-12a


Water trap before analyzer
Cold lineb
-5
-5
-5

-5
-5

-5
Hot line
<-5c




Leak
Checkd
-25
-25

-25
-25


Recording6
±3
±3

±3


NDIRf
and
Drier
NDUV


<-40
1+10
f-40

-40
+20
Converter
Check9
i-




aPotential loss of NOx based on Data from Reference 14.

bAssumes that five percent of NO converted to N02 and lost in  water trap.   Based  on data
 from Reference 12.

cLoss assumed to be less than five percent since heated  sample line used.

dLoss cannot be generalized.  Twenty-five percent loss based on one example reported  in
 Reference 12.

eBased on Reference 12.

 Errors based on Reference  12.   Forty percent error for unconditioned driers.

9Failure to perform converter checks could result in understated NOx values due to
 incomplete conversion of NOg to NO.

-------
OJ
ro
          100 X
+20-
+10-
NO observed \
NO correct 7
/
-10-



-20-
-30-
-40-
-50-
Mfg
Practice

55 - <
@ H @ ca H "" £3
^••J ~~~



Colt Cooper Del aval Superior Waukesha Cooper GMC/EMD
DEMA SAE EMD
                           Does not include  potential error  due to system leaks, and assumes chemical  drier  is conditioned.
                                        Figure  5.   Measurement error relative  to EPA procedure.1

-------
Conclusions
       The above analyses of instruments and measurement practices suggest the following conclusions
about the reported data:
       •   The exhaust data which were reported using the DEMA, SAE, or EMD practices should be
           banded by the appropriate uncertainty level.  In general measurement uncertainties are
           expected to range from -15 to +5 percent for DEMA data and -20 to +20 percent for the
           SAE/EMD data.
       •   Waukesha's data should not be used for setting numerical values of a standard since:  (1)
           only NO was measured, (2) other significant sampling problems could have been present,
           and (3) the amount of N0~ in the exhaust was unknown so that correction of NO to NO + N0«
           would be speculative.
       •   The Ingersoll-Rand and White-Alco exhaust emissions data appear to have the least measure-
           ment error (±5 percent) since their sampling procedure was essentially identical to the
           EPA procedure.  The previous analyses also lead one to conclude that the EPA practice is
           the best existing sampling procedure and should be used as a guideline when stipulating a
           measurement procedure for an NSPS.  Both the DEMA and SAE/EMD practices require better
           and more carefully specified sampling procedures.  In addition, the NDIR/NDUV instrument
           bias relative to CL instruments is so large that it yields unacceptable measurement uncer-
           tainty.  Cooperative testing of the revised EPA procedure indicate that measurement un-
           certainty using the best sampling procedures is about ±5 percent.

 3.2     EFFECT OF AMBIENT  CONDITIONS ON  NO
        Previous studies have recognized that variations in ambient conditions can significantly af-
 fect  observed exhaust emissions, particularly NO  levels (References 6 and 19).  As a consequence,
 EPA has adopted ambient correction factors for regulated mobile sources based on experimental studies
 These factors are used to correct observed NO  levels to values at a standard temperature (85°F) and
 humidity  (75 grains FLO/lb dry air) for heavy duty  (HD) diesel engines and to correct to standard
 humidity, only, for both heavy and light duty gasoline vehicles (References 20 and 21).  Since NO
 emissions from large bore engines will also depend  on ambient conditions, these data should also be
 corrected to standard conditions.
       Although the eight large bore engine manufacturers provided ambient conditions (barometric
 pressure and inlet air temperature and humidity) for much of their emission data, none of them re-
 ported emission data for a given engine operated under varying ambient conditions.  Thus, at this
                                                  33

-------
time it is not possible to directly derive ambient correction factors for large bore engines.  Never-
theless, the response of these large bore engines to changes in ambient conditions may be similar to
that observed in the smaller bore diesel and gasoline engines for which ambient correction factors
have been developed.  Therefore, the discussion in this section will evaluate the applicability of
existing ambient correction factors to large bore engines.  It will also examine the potential ap-
plication of gas turbine correction factors and analytically derived expressions.

       Section 3.2.1 will discuss variations in N0x levels resulting from ambient differences and the
reduction in data scatter after applying ambient correction factors developed for smaller bore diesel
and gasoline engines.  Then, in Section 3.2.2, the existing correction factors and potential ana-
lytical expressions are evaluated in detail and appropriate correction factors are selected to cor-
rect the emissions data from the eight large bore engine manufacturers of this study.  The results
obtained after applying this correction to the uncontrolled data base are summarized in Section 3.2.3
and further recommendations are made for ambient correction factors for large bore engines.

3.2.1  Effect of Ambients on NOX Emissions from 1C Engines

       Enforcement of NO  emission standards for mobile 1C engines must rely on ambient correction
factors to adjust NO  measurements to standard ambient conditions, since the observed levels may
differ considerably from levels that would be measured at standard atmospheric conditions for which
control standards are set.  Several studies have shown the effect of ambient humidity on NO  emis-
                                                                                           X
sions (References 22, 23, and 25).

       Figure 6 taken from one of these studies, illustrates the variation in specific humidity each
month during the year 1969 for Dearborn, Michigan.  For this area, ambient humidity varied from 20
to 120 grains H20/lb dry air over the year.  The effect of this variation on NO  emissions is shown
in Figure 7 using the correction factors that have been derived from experimental  work for gasoline
and diesel engines.  The N0x levels are shown to deviate as much as 25 percent from levels measured
at standard conditions.  Moreover, the correction factors used vary significantly as well, depending
on the particular study and type of emission source.

       Variations in ambient temperature have also been shown to affect NO  emissions.  This effect
                                                                          X
is illustrated in Figure 8 and ranges from 5 to 25 percent depending on the particular study.  Al-
though the effect for diesel  engines is not as large as that produced by ambient humidity variations,
the change in  brake specific emissions (g/hp-hr) can be significant.  This is particularly true for
some  large bore  spark ignition engines, which emit as much as 20 g/hp-hr.  For these sources a 5-
percent  ambient  correction means a change in the reported level of about 1 g/hp-hr.
                                                 34

-------
       Variations in barometric pressure also can be expected to affect NO  emission (References 22
and 26).  Only one study, however, has evaluated this ambient variation (Reference 22).  This in-
vestigation used carbureted gasoline engines and found NO  variations of as much as 40 percent due
                                                         X
to changes in ambient pressure.  These changes were attributed largely to variations in A/F ratio
in the carbureted gasoline engines.  However, a correction factor for this effect could not be de-
rived for the different carbureted engines because each engine's emissions response to changes in
barometric pressure was too inconsistent to generalize a correction.
       Despite the lack of a quantifiable correction for changes in ambient pressure, several studies
have shown that scatter in emissions data taken at different ambient humidities and temperatures can
be reduced, significantly, by  applying the appropriate correction factors.  For example, Figure 9 il-
lustrates the reduction in data scatter that is achieved with emissions from HD gasoline engines by
correcting for humidity only.  The average standard deviation from the six engines was reduced by
a factor of about six by the use of a correction for humidity.
       Figure 10 illustrates a similar result for ambient temperature and humidity correction of
diesel engine emissions.  The  average scatter after correction was reduced to about one third of the
scatter before correction.  Note, that in absolute terms, the reduction in scatter for diesel engines
was smaller (-0.5 g/hp-hr) compared to the reduction for gasoline engines (~1.5 g/hp-hr).  Both
studies, nevertheless, indicate that scatter in emissions data can be reduced significantly, by cor-
recting observed NO  levels to standard conditions.

3.2.2  Selection of Ambient Correction Factors for Large Bore Engines
       As stated previously, ambient correction factors have been developed for automobile  and truck
size gasoline (SI)  and diesel  (CI) engines.  The following sections will  discuss the application of
the existing SI factors to natural gas engines and the existing CI factors to large bore diesel  and
dual  fuel engines.   In addition, gas turbine ambient correction factors will  be examined for applica-
tion to large bore engine emissions.  As these sections will illustrate, no satisfactory ambJent
temperature correction factor has been developed J^_ajiy__sj^j:^rTjI_jji!£rJiaJ—CM^
only one study has considered ambient temperature corrections for CI engines.. Therefore, Section
3.2.2.4 will  discuss an analytical approach to correct emissions for variations in ambient temperature,

3.2.2.1  Ambient Correction Factors Developed for SI Engines
       A survey of the literature showed that correction factors for gasoline-fueled engines have
been  developed only for ambient humidity variations (Reference 19).  One of the studies also evalu-
ated  the effect of ambient temperature and barometric pressure variations on exhaust emissions, but
                                                 35

-------
   S-

   ro
   TD
   o
   Q.

   S-
   QJ
   CL

   OJ
   fO
   S-
   cn
-o
•i—
 E



 CD
4->
 13

 O

-Q
       140'
       120-
    100-
        80-
        60-
        40 -J
        20 -J
                i    i    i    i    i     i    I    I    I     i    i    i
              Jan Feb  Mar Apr May Jun Jul Aug  Sep Oct  Nov  Dec
Figure 6.   1969 humidity range  in  engine test cells,  Dearborn, Michigan
           (from Reference 24).
                                  36

-------
      O)
      -p
      (J
      O)
      1-
      o
      O
CO
03
  1.0  for humidity  >75  grains/lb dry air.
                   'From Reference 19.
                   'From Reference 20.
                                        Figure 7.  Effect of ambient  humidity  on  1C  engines  NOX emissions.'

-------
CO
co
1.5 .
-o
O)
o
O)
S-
S-
0
o

^-~~N
X
o
~^"


~S
i—
13
(/)
(13
E 1 0

^~~x
X
o
^^
" 	 '
II



















































^ 	
Diesel Enqi

















^•^i^
	 y
nes







I I

i i






r
Larg

















	














2 bore
N
i on
1 . L\J
ii
•1.10 ^ ^
CD O
X X


, _ 30
1.0 0) 0
&< -s
to -s
c o>
-s o
(D (-1-
• .90 °- g.

- .80

                                                                     Natural  gas  engine



                      aBased on range of ± 30°F from standard temperature of 85°F.   K >1  for temperature
                       less than 855F.

                       Reference 6

                      cBased on change in NOX emissions with manifold inlet temperature.   Assume change
                       in ambient is equal to change in manifold temperature.  From Reference 26.
                            Figure 8.  Effect of ambient temperature on 1C engine N0x levels.

-------
GO
               2.0 -
s_
_c

o.
-C

CD
            C
            o
            >  1.0
ro
-a
c
ra
-I-J
00
                                       10
                                                                	A	Average
                                                                       Before Ambient  Correction
                                                              After Ambient Correction
                                    11
12
                                                      NO  Level, g/hp-hr
                                                        X
13       14

     a
15
16
                     aAverage level (Federal. Test Cycle Composite) for four ambient humidities; 30, 45,
                      75, 120 grains/lb dry air.
                            Figure 9.  Effect of humidity on emissions scatter for six HD gasoline
                                       engines (Reference 22).

-------
s_
_c
Q.
_£Z
Dl
C
o
•I—
03
OJ
Q
i.
ro
-a
    1.0-,
     .6-
    .2
                       Before Ambient Correction
                                                                   After Ambient Correction
                 1
                5.2
                          2
                         7.3
 3
7.6
 4
7 A
 5
9.5
  6
13.3
Engine #
NOX Emission,
g/hp-hr @ 75 grains/
Ib. dry air, 85°F,
29.92 inches Hg.
          For Federal 13 mode composite cycle for HD diesel engines.
          Ambient humidity varied from 35 to 125 grains/lb dry; temperature from 70 to 115°F.
        Figure 10.   Effect of humidity and temperature on emissions scatter for six
                    HD diesel engines13 (Reference 6).

-------
found that engine-to-engine variations were too great to generalize a correction factor for either
temperature or pressure (Reference 22).  These studies will be briefly discussed in the following
paragraphs, and then the ambient humidity correction factors will be compared and evaluated for ap-
plication to the large bore natural gas-fueled engines in this study.
       The Automobile Manufacturer's Association contracted with the Ethyl Corporation to conduct a
study on the effect of ambient air humidity, temperature, and pressure on heavy duty (HD) gasoline
engines  (Reference 22).  The results of this study for the effect of humidity later became the cor-
rection  factor designated in the Federal Register for HD gasoline engines (Reference 20).
       This correction factor was derived from emission measurements conducted on seven engines in
accordance with the Gasoline Fueled Heavy-Duty Engine procedures in the 1970 Federal  Register.   The
test engines were installed in an engine dynamometer test cell where humidity, temperature, and
pressure were varied.  The engines tested were gasoline fueled, spark ignited, carbureted, heavy duty
truck engines ranging in size from 38 to 75 CID per cylinder.  The compression ratios varied from 7
to 9.4, and air-to-fuel ratios varied from 14.4 to 15.7.  The objective of study was  to develop fac-
tors to adjust composite mass emissions to a standard condition of 75 grains H^O/lb dry air, 90°F
inlet air temperature, and a barometric pressure of 29.92 inches of mercury.
       Figure 11 illustrates the effect of humidity on both A/F ratio and NO  emissions (Federal 9
mode composite cycle).  As this figure illustrates, a reasonably good correlation was established
between changes in ambient humidity and NO  emissions.  Note also that A/F ratios are essentially
constant or decrease slightly with increasing inlet air humidity.  The ambient correction factor was
of the form
                                     K = 0.634 + 0.00654(H) - 0.0000222(H)2                      (1)

where  N0x corrected =  N0x observed
       H             = specific humidity, grains H20/lb dry air
       Figures 12 and 13 illustrate the effect of temperature and pressure on A/F and N0x emissions
for these same engines.  Obviously, engine-to-engine variations were too great to generalize a cor-
rection factor for either temperature or barometric pressure.
       Both physical  reasoning and these results indicate that changes in ambient temperature and_
pressure affect the A/F ratio more than do changes in humidity.  Moreover, it is well known that the
effect of A/F ratio variations on NO  emissions depends on the engine operating point on the N0x vs.
A/F curve (see Figure 14).  For example, a decrease in A/F ratio for an engine operating at A in
                                                  41

-------
o
4->

14—
1_
s
14
13
16
15
14
16
15
14
16
15
14
16
16
	 Q 1 1 1 |
8 — —
Engine 1
till
i
_
i

1 I I 1
flD n P
UL o o
Engine 2
i i I i
I
Q
'\j
i

r\ 1 ' ' '
w O 	
Engine 3
1 i I i
i
c\
— ©-_
i

81 i " i
— 8 	 8
Engine 4
I i I i
i
— O— _

-ft— ft ' Q'
Engine 5
I 1 1 i
it;
14
P '/? ' 0 '
w CV Q 	
Engine 7
1 I 1 1
i
8 -


— _
4-/t,3o3
    20      40      60       80      100

            Humidity, grains H^O/lb dry air
120
                                                          -C
                                                           I
                                                          o.
                                                               20.0-
                                                               18.0-
                                                               16.0-
                                                               14.0-
                                                               12.0-
                                                               10.0-
                                                                8.0-
20
                                           Each number represents an
                                           engine's datum point.

                                           Curves show M02 values predicted
                                           by the R^ factor equation which
                                           is good only to 120 grains.
 I
40
                                          60
                             80
                                 Humidity, grains
                           TOO

                      dry air
Figure 11.   Effect of  humidity on  A/F ratio and N0x  emissions for HD  gasoline engines  (from  Reference 22}

-------
Figure 14 would result in a decrease in NO  emissions, whereas a decrease in A/F ratio for an engine
                                          X
at C would cause an increase in NO  emissions.  The author concluded, therefore, that engines operat-

ing at different A/F ratios with different metering characteristics should exhibit varying (possibly

contradictory) effects on exhaust emissions for similar changes in inlet air conditions.

       The author also indicated that changes in inlet air conditions could affect fuel distribution

to the engine, and this in turn affects N0v emissions.  This effect is illustrated in Figure 15.
                                          A

In addition, changes in inlet air conditions which change the engine A/F ratio also change the min-

imum spark advance for best torque  (mbt).  Figure 16  illustrates how NO  emissions vary with A/F

for the mbt setting and other settings retarded from  mbt.  Since some of the engines in that study

had no vacuum advances, they experienced, in effect,  retarded mbt settings as their A/F ratio changed.

Thus, their emissions would respond differently to A/F changes than engines with vacuum advances.

Therefore, the development of a general correction factor for ambient temperature and pressure was

not possible, although variations in these parameters affected emissions significantly.

       In two other studies ambient humidity corrections were developed from LD gasoline vehicles

(References 23 and 25).   In the first study, a correction factor was derived from emission tests  at

76°F and at four ambient  humidities on a fleet of eight passenger cars operated on a chassis dynamo-

meter set to simulate seven different road loads of the Federal 7 mode composite cycle for LD vehicles.

These passenger cars were powered by gasoline-fueled, spark  ignited, carbureted engines ranging in

size  from 42 to 59 CID per cylinder.  The compression ratios  varied from 8.5 to 10.5, and air-to-fuel

ratios from 14.6 to 16.4.

       Corrections for ambient  humidity were  derived  for  both  the  Federal Test  Cycle composite load

factor is considered more applicable to  large  bore  carbureted  SI engines which  typically operate at

a  constant load (nearly  rated  load).  These  factors are


                                     K = 0.796 + 0.175(H/100)  + 0.129(H/100)2                    (2a)
                                     composite factor


                                     K = 0.844 + 0.151(H/100)  + 0.075(H/100)2                     (2b)
                                     constant  load, 50 mph


The results of this study were  adopted by California  to  correct emissions  for  ambient  humidity for

gasoline powered vehicles under 6,000  pounds  (Reference  27).
                                                   43

-------
.i • — r 	 1 	
14 9- O~O 	 .
11 ' ' '
if
15 T- 'o~~o- 	
n ' ' '

J i i
,5 
-------
en
1
14 '
16
15
J
16
15
14
16
15
14
15
14
13
16
15
14
11
1 i < i i
o Q 	 '
fj 	 "" ' H . rnninr

. '. 	 @__J 	 8—^—
r O- 	 cr .
Enqine
1 1 1 i 1
A
1 J
-8'
2
1

| I 1 > i
p Q 	 	 | 	 
-------
CD
                                S_
                                -t->
                                c:
                                o>
                                o
                                c
                                o
                                o

                                00
                                ro
                                CD
                                         I      I
            I      I      I      I
                                   1
                                   10
12
                              NO
14
 i
16
 l
18
20
                                                          Air-Fuel Ratio
                                   Figure 14.  Exhaust gas  concentration versus A/F (from
                                               Reference  22).

-------
    2000.
    1600
    1500
g-   1000J
     725


     500-
      Most Uniform
      Fuel Distri-
      bution
Least Uniform
Fuel Distribution
                      I  1600  - 725
                      I     1600
        = 55%
                            I    1   I
            13     14   15   16   17  18 19 20

                      Air-Fuel  Ratio


         Figure 15.  Effect of fuel  distribution
                     on NO emissions (from
                     Reference 22).
2200
                                                     Q.
                                                     Q-
                                                          600
                                                                 Retard (% decrease in
                                                                         power)
                                                   Air-Fuel  Ratio


                                    Figure 16.   Effect  of ignition  timing  and A/F
                                                on  NO exhaust  emissions, 1600 rpm,
                                                23  bhp  road  at mbt  (from
                                                Reference 22).

-------
       In the second  study,  a  similar  test  program on  gasoline  vehicles was conducted to develop
an ambient humidity correction factor  that  was  adopted by  EPA to correct LD gasoline vehicles
(References 21  and  25).   This  factor  is

                                        K = I/O    0.0047(H-75))
                                                                     (3)
       In yet another study,  an  ambient  correction  factor for humidity was  developed for typical  re-
search engines where  A/F  and  mbt settings  were  held constant  (Reference 24).   The author of this
study also noted  the  effect of A/F  ratio on  N0x emissions as  illustrated in Figure 17.   Based on  this
figure, the author of Reference  22  reasoned  that composite correction  factors  would vary from constant
load factors since NO  emissions vs.  humidity curves  changed  with  A/F  ratio.   Despite the validity
of these observations, the correction factor derived in this  study is  probably less applicable to
large bore engines than those derived for  gasoline  vehicles,  since in  practice A/F ratios and spark
settings of large bore engines are  not necessarily  held constant as they were  in this study, (e.g.,
spark timing is fixed as  load changes for  some  engines, and varies with load for other engines).

Comparison of Existing Humidity  Correction Factors  for SI Engines

       Based on the preceding discussion,  three ambient humidity correction factors are potentially
applicable to large bore, natural gas fueled engines, particularly four-stroke, carbureted versions.
These factors are summarized  in  Table 11.   Figure 18 is a comparison of the three factors over a
typical range of ambient  humidities.   Note that only one of the factors plotted is a constant load
factor (Equation 2b); the other  three are  based on  composite test cycles.

                     TABLE 11.   AMBIENT  HUMIDITY CORRECTION  FACTORS FOR SI  ENGINES
              Equation No.
                  (1]
                  (2a)
                  (2b)
                  (3)
              Correction Factor
K   0.634 + 0.00654(H) - 0.0000222(H)2 Compo-
site Factor (9 Mode Federal  HD Gasoline Test
Cycle)

K   0.796 + 0.175(H/100) + 0.129(H/1002)
Composite Factor (Federal Test Cycle, LD
Gasoline Vehicles)

K   0.844 + 0.151(H/100) + 0.075(H/100)2
50 mph, Constant Load

K   1/(1   0.0047(H-75)) Composite Factor
(Federal Test Cycle, LD Gasoline Vehicles)
Reference
                                                                                  22
                                                                                  23
                                                                                  23
                                                                                  25
                                                 48

-------
CL
CL
CO
c:
o

to
CO

'i
cu

o
3500-



3000-



2500




2000



1500



1000
O  20 grains moisture

A  90 grains moisture

D  160 grains moisture

   per pound dry air
                        15
                         16
            17
18
                    Air-Fuel  Ratio
                       15.7 A/F


                       14.9 A/F


                       14.4 A/F
20              90                160
                                                          Humidity grains H 0/1 b dry air
               Figure  17.  Effect of A/F ratio and ambient humidity on NO emissions of a LD
                          gasoline engine  (from Reference 22).

-------
O)
4->
O
o>
S-
O
"O

s_
   0)
        1.25  -
        1.20  -
        1.15  -
                Q  eqn (1)

                0  eqn (2a)


                    eqn (2b)
                    20       40       60       80      100      120

                        Ambient  Humidity (grains H^O/lb dry air)


             Figure 18.  Comparison  of  SI  ambient humidity correction factors
                                         50

-------
       As Figure 18 illustrates, there is a considerable difference in correction depending on the
study.  The results from the EPA/Scott study show the greatest sensitivity to ambient humidity varia-
tion while the results for HD gasoline engines show the least.  This is not unexpected in view of the
previous discussion regarding the variation in engine responses to changes in inlet conditions de-
pending on A/F ratio, fuel metering and distribution, and ignition (distribution operation) charac-
teristics of different engines.  The correction based on constant load has been chosen as the most
suitable correction to be applied to large bore SI engines, since these engines are typically opera-
ted at constant load.  Thus, this correction factor is most applicable to carbureted (4-NA) natural
gas engines.
       Application of any of the correction factors of Figure 18 to other engine types (e.g., turbo-
charged units) is questionable due to major differences in inlet air intake systems.  For example,
data  from the draft SSEIS indicate that N0x reductions due to water induction are directly propor-
tional to water-to-fuel (w/f) rates, up to w/f ratios of one.  Moreover, w/f rates due to ambient
humidity are a function of A/F ratio, as Figure 19 illustrates.  Based on this figure, the w/f rate
of a  carbureted natural gas engine (trapped A/F -17) is about 30-percent lower for a specific humid-
ity of 100, than the w/f ratio of a turbocharged engine with a trapped A/F ratio of 25 (trapped A/F
ratios for turbocharged SI engines typically range from 20 to 25, References 28 and 29).  Note, also,
that  the curves of constant humidity diverge with increasing A/F ratio.  Therefore, it can be anti-
cipated that NO  emissions of engines with different A/F ratios will respond differently to identical
changes in ambient humidity.  Thus, the application of the constant load humidity correction factor
(based on carbureted gasoline engines) to other than carbureted large bore, SI engines is questionable
       Similar conclusions can also be reasoned regarding the application of ambient temperature and
pressure correction factors to NO  emissions from engines whose air intake as well as fuel systems
differ substantially.  As discussed earlier, however, no factors have been developed for SI engines
for either temperature or pressure.

3.2.2.2  Ambient Correction Factors Developed for CI Engines
       A survey of the literature established that two sources have reported ambient correction fac-
tors  for truck-size diesel engines (References  6  and  19).  A study by Krause, et al., was sponsored
by_ the Automobile Manufacturer's Association and the Engine Manufacturer's Association  and resulted
in correction factors for both temperature and humidity.  These corrections were  later  adopted by
j:he EPA for HD diesel engines (Reference 20)^  The other source, Reference 19, reported  corrections
for humidity only.
                                                  51

-------
                         0.80
en
ro
                OJ
                4->
                co
                CC  r—
                   OJ
                c  rs
                O  4-
                •r- E
                +-> -Q
                U i—
                13 -^
                -o  o
                C   (XI
S- J3
cu •—
                "3
         0.60-
                        0.40-
                        0.20-
                                         Figure 19.   Water to fuel  range vs.  ratio air-to-fuel.

-------
       In Krause's study, a correction factor was derived for six different engines run over a 13-
mode cycle (Reference 6).  Emission procedures used were similar to "California Procedures for Diesel
Engines in 1973 and Subsequent Model Year Vehicles Over 6001 Pounds Gross Vehicle Weight."  The
heavy duty truck engines tested were diesel fueled, compression ignition engines ranging in size
from 69 to 155 CID.  The compression ratios varied between 14.9 and 18.7.  Inlet air conditions were
controlled to one of 19 combinations of humidity and temperature.  The humidity conditions ranged
from 35 to 125 grains HgO/lb dry air, and the temperature was varied between 70°F and 115°F.  Baro-
metric pressure was controlled to 28.00 ± 0.2 inch Hg. at the air cleaner inlet.
       The engines in the study were of the following types:
       •   4-stroke turbocharged, direct injection (4-TC)
       •   4-stroke turbocharged, prechamber  (4-TC, PC)
       •   4-stroke naturally aspirated, MAN  chamber (4-NA, MAN)
       i   4-stroke naturally aspirated (4-NA)
       t   2-stroke blower scavenged (2-BS)
       •   4-stroke turbocharged, aftercooled (4-TC,AC)
       Unlike turbocharged large bore engines, which are nearly always aftercooled, only one of the
 turbocharged engines in this study  was aftercooled.  That engine as well as the 4-NA and the 2-BS
 units  were similar  in design to  large bore  engines.  A correction  factor was developed for all of the
 engines  as well as  for  individual units.  The factor was of the form

                                     K   l/[A(H-75) + B(T-85)]

where NO corrected = (K)x (NO observed) and A and B are related to A/F ratio (load).  The reference
ambient conditions are 75 grains H-0/lb dry air and 85°F.  Figure 20 is a plot of the coefficients
A and B for each engine as a function of load (or A/F ratio).
       Average values of A and B for all engines (as function of A/F) were also determined.  This
result does not appear wholly justified given the variations in response to ambient humidity and
temperature exhibited by the different engine types depicted in Figure 20.  Note that the 4-TC, AC
unit is significantly less sensitive to ambient temperature variations over the load range than the
other designs.   Therefore, separate ambient correction factors for 2-BS, 4-NA, and 4-TC, AC units
from this study were used on the corresponding large bore designs instead of just one average value
                                                  53

-------
     0.010
                                           4-TC
                                           2-BS
                                           4-NA
                                           4-TC
                                           4-TC
   -0.004
                    100
16.7
                                     A/F
Figure  20.   Effect of humidity and temperature on NO emission for
            different CI engines.
                              54

-------
of all engines.   The  coefficients  A  and  B were  determined  for  rated  load  conditions using average
rated  load A/F  ratios  that  were  reported for  the  large  bore  engines.
       The second source  (Reference  19)  which examined  ambient humidity effects reported correction
factor based  on  experimental  tests of  a  2-BS  and  a  4-NA engine.
       Ambient  humidity corrections  from both of  these  sources (assuming  inlet temperature is held
constant, hence,  the  B term of Krause's  factor  drops  out)  are  illustrated  in Figure 21.  The SI
(gasoline) factors  discussed  in  Section  3.2.2.1 are also plotted  for comparison.  Note that there is
little variation  in correction factors for  different  diesel  engine types with the exception of the
2-BS unit from  Reference  19.  Moreover,  the diesel  fueled  (CI) engines appear less sensitive to am-
bient  humidity  changes than SI units,  particularly  for  humidities less then 75 grain H20/lb dry air.
       Initially, one might expect the diesel units to  be  more responsive  to inlet humidity variations,
since  these units operate at  higher  A/F  ratios  (approximately  25  to  40), and therefore, induct more
nearly at stoichiometric  ratios  (-15 to  17, depending on the fuel).   Diesel engines, however, have a
greater  thermal  inertia than  SI  engines  according to  their higher trapped  A/F ratios.  Apparently the
higher thermal  inertia in the diesel units  more than  offsets their higher  effective water induction
rate;  thus their NO   emissions are less  sensitive to  changes in ambient humidity than SI engines.
                    X
This explanation  is corroborated by  experiments which have demonstrated that water injection, as a
control  technique,  produces significantly greater NO  reductions  in  SI units than for CI (diesel)
units  (see Section  5.3.8  of the  draft  SSEIS).
       The Krause study (Reference 6)  also  investigated the  effect of ambient temperature on NO
emissions.  Figure  22  presents the correction factors that were derived for engine types similar to
those in the present study  (humidity is  assumed constant,  hence, the A term of Krause's factor drops
out).  The reference temperature is taken as 85°F.  This figure shows that the naturally aspirated
and blower scanvenged engines NO  emissions are more  sensitive to inlet air temperature changes than
the aftercooled design.  Since Krause's  study systematically examined the effect of both inlet tem-
perature and humidity for a number of  CI engine types,  his factors were selected for application to
similar large bore engine types.

3.2.2.3  Potential Application of Gas Turbine Ambient Correction Factors to Reciprocating 1C Engines
       Since no ambient correction factors  have been  developed specifically for large bore engines,
all  existing correction factors for internal combustion  engines were examined including those for
gas  turbines.   There is a considerable amount of technical  literature on gas turbine ambient correc-
tion  factors as a result of the promulgation of emission standards for aircraft engines and the pro-
                                                 55

-------
-o
O)
•M
O
O)

i.
o
o
    1.25 -
       1.20
       1.15
       1.10
       1.05
OJ
I   1.0
        .95
        .90  •
        .85
        .80
        .75
                   O 4-TC (AC), 4-NA


                   V 2-BS

                   A 4-NA   j
                            Ref.  19
                      2-BS
                       SI  factors
                       (See  Figure  is)
                  /     Ambient Humidity  (grains  H20/lb dry air)
                    20
                         40      60      80      100     120
             Figure 21.   Comparison of CI and SI ambient humidity
                         correction factors.
                                      56

-------
                  1.10 T
in
                  1.05 -  -
             01
             QJ
O
O
                CU
                S-.
                ta
                  1.0
                   .95 -  -
                                                                     O  4-TC (AC)


                                                                     Q  4-NA


                                                                     V  2-BS
                   .90 -  -
                                   50    60    70     80    90   100    110    120

                                            Ambient  Air Temperature, °F
                                 Figure  22.   Correction  factors  for temperature for CI
                                             engines  (Reference  6).

-------
posed emission standards for stationary gas turbines (References 30, 31, 32, 33, 34, and 35).  This



section describes these various factors and their potential application to the large bore engine data



in the present study.




       Gas turbine combustion characteristics have some common features with those in diesel engines.



Both the gas turbine primary combustion zone and the diesel combustion chamber can be characterized



as a well-stirred reactor, and both use similar distillate fuels or natural gas.  On the other hand,



the gas turbine combustion is a steady-state, constant pressure process, whereas the diesel is an



unsteady, variable pressure combustion process.  Nevertheless, the similarities of the two systems



warrant an investigation of existing gas turbines correction factors.




       Many investigators have developed ambient humidity correction factors based on a model that



relates NO  formation to combustion parameters of temperature, pressure, equivalence ratio and res-



idence time (using the kinetic rate equations for NO  formation) (Reference 34).  Humidity enters



this model though its effect on reaction (flame) temperature.   Most researchers have shown that the



effect of humidity on NO  formation takes the form
                        K




                                  NO
                                    x corrected   Fxp [K ,„           H         n

                                  N0x observed  "        (Hobserved   "reference'J





where   H  ^   specific humidity at reference (standard) conditions



        K      empirical constant that ranges from 14 to 30, generally taken as 19 (Reference 30)




       Figure 23 is a plot of this correction factor and shows that it agrees reasonably well with



the HD diesel (4-TC, AC) factor discussed in Section 3.2.2.2 (the gas  turbine factor was adjusted



to a reference humidity of 75 grains H^O/lb dry air to correspond with the  reference humidity of the



HD factor).




       Other sources have shown similar agreement of humidity  effects  on NOX emissions of diesel



engines and gas turbine as illustrated in Figure 24.   The gas  turbine  humidity correction factor was



derived from empirical  data based on water injection  as a means of NO   control  in gas turbines.   The
                                                                     X


ambient humidity was converted into an effective water-to-fuel  ratio by multiplying the ambient



humidity loading by the near stoichiometric A/F ratio existing at combustion.  Then the empirically



derived water injection correction factor of Ambrose (Reference 36) was used to calculate the per-



centage reduction in NOX-   It is reasoned that the overall  A/F ratio is inappropriate since much of



the water vapor in the  inlet air never reaches the primary combustion zone because it is vented for



engine cooling or enters downstream as dilution or wall cooling air.  Based on this result one can



conclude that changes  in humidity appear to affect NOX formation in gas turbines in much the same
                                                  58

-------
en
10
                        "0
                        O)
                        
-------
CTl
o
                        1.0  -r
                                                     Predicted Curve
                  QJ
                  _
                  E
                  to
                      >» 0.9
                     -a
                     0 n
                      CM U.
un
ro
                     •P
                     (T5
   0.6 -




   0.5 -

       s
                         0
                                             Experimental
                                             Curve
                                                       Experimental Lab Data

                                                       Field Data

                                                       Diesel Engine Data
                            0.0     35      70      105     140     175     210     245

                                       Specific Humidity  grains  H?0/lbm dry air
                                                                    230
315
                                Figure 24.  Observed  ambient humidity influence on NO  production
                                            (from  Reference 36).                      x

-------
way as in diesel (1C) engines.  However, it should be remembered that a number of adequate humidity
correction factors have been developed for 1C engines; therefore this gas turbine result is of
limited value, but serves to reinforce the more global application of these corrections.
       Gas turbine temperature corrections were also examined for application to reciprocating engine
data.  Figure 25 is a comparison of the HD diesel  (4-TC, AC) correction factor with various gas
turbine factors.  Obviously, there is little agreement with the exception of one factor.  Presumably
these differences arise from differences in inlet  air compressor and combustor design.  The gas tur-
bine factors, however, have some flexibility in that they  are related to the combustor  inlet para-
meters of pressure and temperature.  Therefore, an attempt was made to relate changes in NO  forma-
tion of reciprocating engines to changes in ambient temperature by estimating the precombustion
cylinder temperature in a typical  reciprocating engine using the compressor pressure  ratio and the
assumption of isentropic compression to calculate  temperature.  This calculated  temperature was then
used with the gas turbine factors.   Figure 26 presents the results of this approach for several
different forms of gas turbine  factors.  Again, these results do not correlate well with the HD
diesel factor,  probably due to  the empirical nature of the gas turbine equations and  the large dif-
ferences  in  air intake systems  between engines  and turbines.
       On the basis of this brief  review,  gas turbine ambient correction factors do not make a use-
ful  contribution as potential ambient correction  factors  for reciprocating engines.

3.2.2.4  An  Analytical Approach  for  the Ambient Temperature Correction of NO  Emissions
       Since no temperature correction factor has  been reported in the literature  (see  Section
3.2.2.1) for SI engines, and no  systematic emissions data  exist (either SI or CI)  from  which to base
a  temperature correction for large bore engines in general, an attempt was made  to develop a semi-
analytical approach for relating changes in ambient temperature to changes in NO   level from large
bore engines.  This approach was used to estimate  how changes in ambient temperature  affect NO
levels.
       The analytical approach  is  based on the  fact that  ambient temperature changes  affect NO  emis-
sions by their direct impact on both fuel/air (f/a) ratio  and peak flame temperature.   As the ambient
temperature  rises, the inlet air becomes less dense.  Since the air intake volume  is  essentially con-
stant, the engine will inject a smaller mass of air; that  is, the f/a ratio will increase.  This
change in f/a ratio is related to  a change in NO   level for both SI and CI engines (see Figure 27).
       The temperature of the air  in a diesel engine or fuel/air mixture in a dual fuel or natural
gas engine after compression is correspondingly changed by a change in ambient  (inlet)  air tempera-
                                                   61

-------
CTi
ro
           o
           01

           s_
           o
           o
S-
OJ
to
                                                                                HD Diesel  Reference  6
                                                                                gas turbine

                                                                                Reference 37
                                                          80
                                                        100
120
140
                                                      Temperature, °F
                         Figure 25.   Comparison  of existing gas  turbine ambient temperature correction
                                      with  HD  diesel  temperature  correction.

-------
O-l
CO
-a
CD
-M
O
a>
s-
S-
o
o
                     O)
                     
-------
                                                     TREF
en
                      NO
                                                                               TREF
                                  \
                                                                               REF
                                                                                            Load
                                                              Af/a
                                                            (f/a)REF    (f/a)
                                                          f/a
                             Figure 27.  Relationship of NO  level and load to f/a ratio
                                         for a turbocharged diesel engine operating at
                                         two ambient temperatures.

-------
ture.  That is, an increase in inlet air temperature results in an even greater charge temperature



after compression.  This increase in charge temperature leads to a higher peak flame temperature and,



consequently, greater NOX levels.  The increase in N0x level due to an inlet air temperature increase



is related to specific engine design parameters such as f/a ratio, degree of aftercooling, and com-



pression ratio, as well as the fundamental nature of the combustion process (i.e., CI or SI).  There-



fore, the relationship between N0x level and inlet air temperature is anticipated to be highly de-



pendent on the particular engine design.




       The discussion above indicates that NOX level is primarily a function of both f/a ratio and



inlet air temperature.  That is,





                                             NOX = g(f, T)
                                               A



where  f = f/a, fuel/air ratio



       T 5 T,, ambient inlet temperature
            a


       N E NO , oxides of nitrogen



       The subscript R denotes a reference ambient condition




       Since we are interested in predicting changes in NO  level rather than absolute NO  level we



may mathematically represent a change in NO  to a change in ambient temperature as
                                           X
                                       dN   (§)  df * (f)  dT
 °r
                                                           (l!


dN. = 3N_  df   3|i

dT   3f  dT   3T
 Where the derivatives on the right side of Equation (1) remain to be evaluated.  The first partial


             3N
 derivative, -53-, represents the change in NO  emissions due to a change in f/a at constant ambient
             dT                             X

                                        3N
 temperature, while the second partial, •**-, represents the change in NO,, due to temperature variations
                                        d I                              X


 at constant f/a.   This mathematical formulation can be portrayed graphically as shown in Figure 27.



 The diagram shows NO  vs. fuel-to-air ratio, the downsloping curves, and load vs. fuel-to-air ratio,



 the upsloping curves, for a typical turbocharged diesel engine.  The basic problem is to find the



 change in NO  at  constant load due to some ambient temperature change.  For example, assume the am-



 bient temperature, T, is greater than the reference temperature, TR.  Starting at the uncorrected



 fuel-to-air ratio one locates point A on the NO  production curve and point B on the load curve.
                                                X


 Then  moving over, at constant load, to the reference temperature load curve, one locates point C and,



 hence,  the reference fuel-to-air ratio.   Now, knowing the reference fuel-to-air ratio,  one can move
                                                  65

-------
from point D to point E on the reference temperature curve.   Thus, the change in N0x production re-



sulting in this change of ambient temperature is indicated as ANO .
                                                                 X



       The change in NO  level with f/a for a constant ambient temperature, -^, is presumed known



for a given engine or engine type.  This relationship can be established from data that can be ob-



tained in the laboratory on a sample engine.   In addition, it is assumed that the N0x vs.  f/a plots



of Figure 27 have similar slopes for different ambient temperatures.   Therefore, only N0x  vs. f/a



data at one ambient temperature are required  to evaluate the derivative.




       The derivative df/dT, expresses the rate of change of f/a ratio with a change in ambient



temperature.  The author of Reference 38 has  derived the following expression relating f/a to ambient



temperature (at constant load) for both turbocharged and naturally aspirated engines.



                                                         B
                                                                                                 (2)

                                                               /i.   i \

and                                   B   (1  + n)    (1    ei)  n






where  k    C /Cv, rati° °^ specific heats



       n  = turbocharger adiabatic compressor efficiency



       n  = turbocharger exponent from equation T.  r    constant where .5
-------
Since B is always less than 2 and T  is 70°F = 530°R, a nominal change in ambient temperature of
                                   R
25°F makes

Since this term is small compared to unity, we can approximate the derivative by
                                            df   „ fR
                                            dt   b Tj^

Thus, the term from Equation  (1) that predicts the change in NO  due to a change in f/a is evaluated
as

                                            3N  df = /8N\/B\/!R
       An expression is now required to relate the change in NO  level to a change in ambient inlet
 air temperature,  i.e., the partial derivative vf of Equation (1).  This dependence can be evaluated
 by first relating changes in ambient temperature to changes in flame temperature using the following
 relationship from Reference  39 suggested  by the authors of Reference 40.
                                                     Y
                                     T  +    -
                                                 (Q-L)-f + Cp(Too - TB)
(5)
where  T-. = droplet diffusion flame temperature from Reference 39.
       T^  E isentropic compression temperature
           = TM(CR)k
       CR  E compression ratio of engine
       TM  E manifold air temperature, related to T. by degree of turbocharging and aftercooling
       Tn  = boiling point temperature of the fuel
       Q   E heat of combustion per unit mass of fuel
       L   s latent heat of vaporization of the fuel
       Y   E ambient oxygen mass fraction
       i   = stoichiometric oxygen to fuel ratio (by mass)
       C   E specific heat of the fuel/air mixture
                                                  67

-------
Then, the change in flame  temperature  is  related to a change in N0x production using the Arrhenius
relation

                                 dNO                  K
                                 -gt*= EXP(-K/RTn); £   123,000°R
If it is assumed that the  rate  of NO   production is time  independent, one can readily integrate the
Arrhenius equation to yield the following expression  for  the ratio of NO  produced  at a given ambient
temperature, T, and a corresponding flame  temperature, T^, to that at the reference condition.
                                   NR
The partial derivative can then be approximated  in  finite  difference  form by
                                         ».•(>    30
                                         3T-     T    TR
Now by substituting Equations (4), (6),  and (7)  into  Equation  (1) we  obtain:
Rf N
dN DTR 3N
r (T Tin
|K * f 0 f 0 P ' 1
1 cvn ^ '*• T)tK
R T x T
i\ ' f n A f f R
L ix. i x.r\ j
dT TR 8f T-TR
                                                                                                (8)
       A preliminary evaluation of this expression for a turbocharged diesel  engine  predicts  a  1-
 percent change in NO  per 1°F change in ambient inlet air temperature.   For a constant intercooler
 effectiveness and turbocharger compressor efficiency, a one degree change in ambient temperature
 corresponds to a one degree change in manifold inlet temperature.   Therefore, this analytical
 expression can be checked using N0x emissions data vs. manifold air temperature (at  constant  load)
 for a  turbocharged, diesel engine.
        Emissions data for the manifold air cooling control (see Figures 5 to 11 of the draft  SSEIS)
 indicates  that this technique produces a 0.1- to 0.3-percent change in NO  per degree Fahrenheit  for
 diesel  engines.  Emissions data reported by Ingersoll-Rand (and from Figures 5 to 11 of the draft),
 show about a  1-percent change in NO  per 1°F change  in manifold air temperature for turbocharged  SI
 engines  (Reference 26).  Since NO  emissions from SI engines are more responsive to changes in mani-
 fold air temperature, it appears that this analytical approach overestimates the effect of inlet  air
 temperature on N0x emissions from this turbocharged  CI engine.  The assumptions which were made with
 regard to  the constancy of the NO  production rate with peak flame temperature may be more valid  for
 SI  engines than CI engines owing to differences in the combustion processes of each.
                                                 68

-------
       The approach outlined here has the potential to incorporate many different parameters into a
single correction factor.  Additional terms could be added for any other operating parameters that
have a significant effect on NO  emissions, either directly or by their impact on f/a or T^.  Once
the appropriate relations are established, it may be possible to predict changes in NO  levels as a
                                                                                      X
result of changes in any one of these variables caused by ambient variations.
       These goals are somewhat ambitious; therefore, the most logical first step is to gather avail-
able data taken at different ambient conditions and apply the method to prove its validity.   At this
time, however, there is insufficient information for either SI or CI engines to substantiate this
approach.

3.2.3  Effect of Ambient Correction  on Reduction of Data Scatter
       Having selected potential ambient correction factors for large bore engines, we are now in a
position to determine whether the use of these corrections will reduce the scatter in the existing
data base.  For the purpose of this  preliminary study, ambient correction factors will be applied
only to the uncontrolled data of four-stroke turbocharged diesel, dual fuel, and natural gas engines.
Emissions data from this engine type contained the most complete ambient information (before incor-
poration of additional Colt and Cooper emissions data) and, therefore, provided a better statistical
sample for determining the effect of ambient corrections on the scatter.

3.2.3.1  Effect of Ambient Corrections on NO  Emissions Scatter
                                            X
       The ambient correction factors were selected from existing 1C engine factors discussed in Sec-
tions 3.2.2.1 and 3.2.2.2.  These factors are summarized in Table 12.  The ones for CI engines are
based on small bore diesel engine types which correspond to each of the large bore designs.   The 4-TC,
AC  factor (see Table 12), was used to correct both the 4-TC diesel and dual fuel results.
       The SI correction factor is for humidity only and is based on a typical 4-NA carbureted gaso-
lene engine operating at a constant  load.  The other humidity  factors  (see Table 11), for similar
gasoline engines all were based on composite test cycles; therefore, the factor based on a constant
load was selected for application to the large bore data.  As was indicated in Section 3.2.2.2, this
factor may not be valid for other engine types (such as the 4-TC units); nevertheless, it was selec-
ted for this preliminary evaluation  since no humidity correction factors have been developed for
other types of SI engines.
       The results of correcting the data are summarized in Table 13 and Figure 28. As the table in-
dicates, the standard deviation in the data was only reduced from 1.9  to 1.7 g/hp-hr for diesel data
and from 4.2 to 3.9 g/hp-hr for the  dual fuel data, whereas it increased from 1.4  to 1.8 g/hp-hr for
the natural gas data.
                                                   69

-------
 TABLE 12.  EXISTING 1C ENGINE AMBIENT CORRECTION FACTORS FOR APPLICATION TO LARGE BORE ENGINES
      Fuel
             Correction Factor
                                                                             Comments
Diesel  &


Dual  Fuel
                   K = l/(l+A(H-75) + B (T-85))


                    H = observed humidity, grains hLO/lb dry air

                    T = observed inlet, air temperature, °F
                                                See Section 3.2.2.2 rated

                                                load correction for humidity


                                                and temperature (from Refer-


                                                ence 6).
                       Type
                     2-BS


                     4-NA


                     4-TC, AC
                      A
                  -0.00242


                  -0.00231


                  -0.00231
                                                        B
                                                     0.00235


                                                     0.00255


                                                     0.0017
Natural Gas, SI
K = 0.844 + 0.151 (H) + 0.075 (H)
                                                                   See Section 3.2.2.1 rated


                                                                   load correction for humidity


                                                                   only (from Reference 22).
 NO  corrected = (K) NO  observed
   X                   A

-------
 TABLE  13.   SCATTER  IN  NOX  EMISSIONS OF UNCONTROLLED 4-TC ENGINES
Fuel/#
Natural Gas/4
Dual Fuel/3
Diesel/11
Engine
Type
4-TC
4-TC
4-TC
NOX Level grams/hp-hr
Uncorrected
Mean
17.5
8.1
8.1
St'd Dev.
1.4
4.2
1.9
Corrected
Mean
15.5
8.2
7.7
St'd Dev.
1.8a
3.9b
1.7b
 Humidity  correction  only
^Humidity  and  temperature  correction
                               71

-------
                   20 - -
                                             O  Uncorrected


                                             Q  corrected
                                                                                                   o
                                                                                                   .9
                   15  ..
ro
               i
               Q.
               1-

               CT>
                   10  --
                    5  ..
Natural gas     Dual Fuel
                                                               Diesel fuel
                          Figure 28.   Effect of ambient correction on scatter in NO  emissions of

                                      uncontrolled 4-TC engines.                   x

-------
       An increase in standard deviation does not necessarily imply that the correction factor is in-
valid (nor does a decrease confirm validity).  That is, it may be possible for NOX levels which are
higher than the sample average to increase when corrected for ambients (e.g., because they were mea-
sured at high humidity and/or low temperature) and, correspondingly, for the lower NO  levels to be
                                                                                     X
decreased by the ambient corrections.
       The increase in scatter of the natural gas data after correction, however,  may be the result
of (1) the application of a humidity factor to 4-TC engines that was based on 4-NA carbureted gaso-
line engines and (2) the inability to correct for ambient temperature variations which may have had
a significant effect on the reported emissions.  This increase in scatter is contrary to the results
illustrated in Figures 9 and 10  (of Section  3.2.1) where the scatter in a sample of both SI and CI
engines was reduced significantly after correction for ambient variations.  Note that the scatter re-
ported in these figures is based on an average for each engine over a range of ambients rather than
an average of the entire sample of engines and, therefore, excludes other sources  of scatter, such
as engine variability.
       A more valid test of the applicability of a correction factor would be its  ability to predict
changes in NO  production caused by changes  in ambients for an individual engine.   This was not pos-
sible since no systematic data set exists which can be used to develop and test corrections.

3.2.3.2  Conclusions Regarding Ambient Correction Factors/
       The primary finding of this study is  that the scatter in the N0y data was reduced only slightly
by the application of ambient correction factors.  Apparently other variables, such as measurement
uncertainties (discussed in Section 3.1) or  engine design parameters (see Section 3.3), account for_
most of the differences in NO  levels reported by the manufacturers.
       The search for ambient correction factors also revealed that neither humidity nor temperature
corrections exist for turbocharged natural gas engines (2 to 4 stroke).  These engine types consti-
tute a significant fraction of the existing  sources and, presumably, will continue to be produced.
Therefore new sources of this kind would be  subject to performance  standards for emissions.  Enforce-
ment of a standard based on measurements "in the field" would require application of  ambient correction
factors or, alternatively, specification that the measurements be made within given ambient  limits.
An experimental program to systematically investigate the effect of variations  in ambient  temperature
and humidity on NO  emissions from natural gas engines would be one approach to the problem  of estab-
lishing ambient correction factors.
                                                  73

-------
       In addition to the lack of humidity and temperature correction  factors,  no  factors  exist to



correct NOX emissions for variations in ambient pressure for carbureted natural  gas  engines.   Studies



of carbureted gasoline enjines_^4^^) have shown that ambient pressure variations  affect NO  emis-



sions substantially j?,y_J=lli]]^^j^4;jTe A/' Fratio (see Fielure 13).  However,  researchers  were  unable  to



generalize a correction factor for this effect.  In view of the absence of such  a  correction  factor,



barometric pressure in a given location should be specified to within certain limits,  and  qualifi-



cation measurements of large bore naturally aspirated or blower scavenged engines  should be conducted



within these limits.  This approach, however, is unsatisfactory for relating measurements  made  at



higher elevations, e.g., Colorado, to a measurement made at standard ambient conditions.





       The application of ambient correction factors to emissions data from dual fuel engines based on



the data  from diesel  engines is subject to question.   Although the combustion of the gas/air mixture in



these dual fuel  engines  is  initiated after compression by the injection of pilot oil, it proceeds  essen-



tially as a premixed flame  rather than  by droplet burning.   Hence, the combustion process has many



similarities to  that in  natural gas  engines.   Moreover, trapped A/F ratios of dual  fuel engines are



generally less than those of diesel  engines but greater than those of SI  units.   Therefore, these



engines experience different effective  ambient water-to-fuel  induction rates and thermal inertias



(see Sections 3.2.2.1  and 3.2.2.2)  than their diesel  counterparts.  Consequently, one can reason that



NO  emissions from dual  fuel  engines will  respond differently than diesel  engines to variations in



ambient humidity and temperature.  Therefore, dual  fuel engines may require different correction



factors than diesel engines.   There  is  no data at this time, however, to  prove or disprove  this hypo-



thesis.




       In summary, ambient correction factors depend on the snqjjie^t^ge^   Although correction for



large bore engines can be inferred from tests on truck size units, this approach is imperfect because



of the differences in combustion characteristics and the fact that not all large bore engine types



are represented  in the test samples.  Nevertheless, it does have enough validity that it can be used



as a basis for correcting emission data until the manufacturers develop and demonstrate their own



correction factors.  In the absence  pfii>anvcoirre>ctionii.factors for temperature or pressure,  limits



should be specified for values of ambient pressure and temperature to minimize variations  of exhaust



measurements.  These Jlimits ^wi 11  be eva1uated_aj:te£itex^imjjijjig_the expanded data  base during the



follow-on work effort.
                                                 74

-------
3.3    EFFECT OF ENGINE VARIABILITY
       The emission data supplied by the manufacturers show a great deal of scatter, that is, NO
levels from uncontrolled engines can vary by as much as a factor of 2 to 3.   Some of this scatter
can be attributed to the use of different measurement techniques or the fact that the results were
not corrected for ambient conditions.  This scatter also could be the result of:   (1) variations in
production for a particular model, (2) variations between different models of the same type (i.e.,
same strokes/cycle, air charging and fuel), or (3) different number of cylinders  for a given model.
In this section the uncontrolled data base is evaluated for these kinds of scatter.   Furthermore,
the data are examined for trends in emissions due to speed, bmep, and manifold air temperatures.
All of these analyses use data that have been correlated for ambient variations by the methods de-
scribed in Section 3.2.3.

3.3.1  Production Variations
       Unfortunately, there is little data available to determine variations in emissions of produc-
tion models.  Most of the manufacturers have concentrated on obtaining emissions  in  their laborato-
ries for different engine models.  At this time there has been little impetus for large bore station-
ary engine manufacturers to make exhaust measurements of engines leaving the production line.
Nevertheless, Colt and GMC/EMD reported variations in production models.
       Colt reported less than a 3-percent difference in two NO  measurements, one in 1972, the other
in 1975, from a two-stroke blower scavenged diesel engine (Reference 41).  Such a small difference was
unexpected, and Colt suggests (in their August 2, 1976 "114" response) that the variation would more
likely be of the order of 10 percent for production units, but they have no data  to verify this esti-
mate.  Colt has measured NO  levels of production spark ignited engines (2-TC-G)  within 3 percent of
                           X
each other under similar ambient conditions (Reference 42).
       GMC/EMD has reported average NO  levels and standard deviations for samples of their 2-TC and
                                      X
2-BS diesel models (Reference 43).  These results are summarized in Table 14.  As this table suggests,
these variations in NO  levels of production engines may have been a result of ambient variations.  In-
let air temperatures varied over a wide range for both the turbocharged and blower scavenged units, and
humidity was not recorded.  An attempt was made to determine whether these observed variations in
emissions could be due to changing ambient conditions.  First, a correction was computed for each
extreme of the reported temperature range using the methodology presented in Section 3.2.2.2.  These
two maximum possible variations were then compared to the reported data to determine if temperature
variations alone could account for the scatter.  Next a correction was computed for both the reported
                                                  75

-------
                        TABLE 14.  VARIATIONS IN NOX EMISSIONS FROM GMC/EMD PRODUCTION ENGINES (FROM
                                   REFERENCE 43)
en
Type
2-TC-D
2-BS-D
# Cylinders
20
16
16
12
# Units
11
10
13
8
Variation (standard deviation
as a % of average)
± 8
± 8
± 6
± 5
Ambient
Temperature
J 67-97°F
63-128°F

-------
temperature variation and an assumed humidity variation, ranging from 35 to 115 grains H-O/lb dry



air.  These corrections were then compared with the production variability scatter.   The comparisons



are listed in Table 15.





       Based on these corrections, temperature variations alone could not account for all  the scatter



of turbocharged units, although they would do so for blower scavenged units.   If humidity were to



vary over the range used for the calculations presented in Table 15, then ambient variations of tem-



perature and humidity could account for all the scatter reported for these engines.   It should be



noted, however, that the ambient correction factors are based on data from smaller,  truck-size



engines, and therefore, may not be applicable to these larger units.





3.3.2  Model Variations





       Variations in NOX levels may be attributed to different engine models  with an engine type for



a given manufacturer.  For example, N0x levels for a manufacturer's 4-TC-G models may vary due to



differences in bore, stroke, turbocharger, configuration (inline cylinders vs.  vee), compression



ratio, aftercooler, and other engine design parameters.  In an effort to identify the magnitude of



model-to-model variations, average N0x levels and standard deviations were evaluated for different



models from each manufacturer.




       Table 16 presents the results of this study.  GMC/EMD, White Alco, and Colt are not included



in this table since they each manufacture one engine model, which can be configured  for different



fuels.  For example Colt markets their 38D8-1/8 opposed piston engine model as  a gas, diesel, or dual



engine, either blower scavenged or turbocharged.  GMC/EMD and White Alco manufacture one basic diesel-



fueled, turbocharged design which differs primarily in number of cylinders and speed ratings.  CMC/



EMD also markets blower scavenged units.




       The other five manufacturers listed in Table 16 produce different engine models within a given



engine type.  NO  levels reported by Cooper for four 2-TC models varied by 13 percent about the aver-
                X


age.  These engines differ in bore, speed, number of cylinders, and torque (bmep), but they were all



operated at the same inlet and manifold air temperature.  Delaval's data indicate  only 4- to 5-per-



cent variation between models for both gas and diesel fueled units.  The percent variations shown for



Waukesha, Ingersoll-Rand and White Superior should not be compared to the Cooper and Delaval results



because differences due to ambient conditions could not be factored out; these manufacturers did not



report ambient data.  To the extent that one can draw any conclusions from such a small sample size,



it appears that NO  emissions for any type of engine (given  strokes/cycle, fuel, and air charging)
                  X


vary more from manufacturer to manufacturer than they do among models in a manufacturer's line.
                                                  77

-------
                         TABLE  15.   POTENTIAL  SCATTER  IN  NOX  EMISSIONS  FROM  PRODUCTION  ENGINES  DUE TO
                                    AMBIENT  HUMIDITY AND  TEMPERATURE  VARIATIONS
co
Type
2-TC-D

2-BS-D

# Cylinders
20
16
16
12
Inlet Air
Temperature
Range
) 67-97°F
)
) 63-128°F
f
Scatter,
Uncorrected
for Ambient
± 8
± 8
± 6
± 5
Correction for
Temperature
Range9
) +3 to -2
) (67 to 97°F)
) +5 to -9
) (63 to 128°F)
Correction
for Temperature
& Humidity
Range^
+16 to -11

+15 to -14

                  aUsing HD diesel  ambient correction  factor  (see  Section  3.2.2.2)
                                  = !/0  + B(T-85J)  where B - .0017  for  2-TC-D  (based  on  4-TC-D)
                                                                         2_BS_D
                                                          T  =  test  temperature
                                                              (Reference  temperature  of  85°F)
                   Using  HD  diesel  ambient  correction  factor.
                      K
                       correction
+ A(H-75) + B(T-75)) where A = -0.00231  for 2-TC-D (based on 4-TC-D)
                             = -0.00242  for 2-BS-D
                           H - test humidity ranging from 35 to
                               115 grains H20/lb dry air
                               (Reference humidity = 75 grains
                                  H20/lb dry air)

-------
                     TABLE 16.  VARIATION IN N0v DUE TO MODEL DIFFERENCES
                                               A

Manufacturer

Cooper
Del aval
Ingersoll-Rand
White Superior
Waukesha
Gasa
4-TC
#

2
3C
2C
3C
AVG

17.6
18.0
20.4
12.8
SD

0.6
1.8
2.4
0.6
%

4
10
12
4
2-TC
#
4




AVG
14.6




SD
1.8




%
13




Dieselb
4-TC
#

3


2C
AVG

9.2


7.9
SD

0.5


0.7
%

5


9
aAmbient correction for humidity only (see Table 12)
 Ambient correction for humidity and temperature (see Table 12)
cNo ambient correction, ambients not recorded
SD - Standard deviation
%  - SD/AVG

-------
Emission variations  that are due  to differences  among manufacturers  could be related to differences



in speed, bmep,  or manifold air temperature.   This  possibility will  be addressed in Section 3.3.4.





3.3.3  Variations  with Number of  Cylinders




       Several  manufacturers have suggested that NO  levels  will  vary for a basic engine design de-
                                                   X


pending on the  number of cylinders due to interaction with the manifold and turbocharger   Figure



29 is a plot of NO  level  (corrected for ambients)  vs.  number of  cylinders.  Based on this grouping,



it appears thatNO^IeveJ^fecjreasesiw^                                                for 4-TCjJ



units, and that 4-TC-D NOy emissions aj^e^




       Figure 30 presents  a different interpretation.   NO  levels  (corrected for ambients)  have been
                                                        X


plotted vs. number of cylinders for each manufacturer's units. This would tend to reduce other sources



of emission variation, such as design differences among manufacturers that may  have been reflected in the



previous figure.  Figure 29 indicates that  there is no  clear trend between NO  emissions and number



of cylinders for either diesel or gas units.   The effect of  changing the number of cylinders introduces



a scatter of 3  to  9  percent.   Note that EMD data could  not be corrected for ambients, and therefore,



this data may not  represent the effect of differences in the number  of cylinders.





3.3.4  Variations  in NOX Level Due to Other Engine  Variables




       The results of Sections 3.3.1 through  3.3.3  suggest that the  variations  observed in NOx



levels reported by all the manufacturers for  engines of a  given type may be a result of design para-



meters that differentiate  one manufacturer's  engines from  those of the others.   Consequently, uncon-



trolled NOX data (corrected for ambients) were plotted  vs. speed  (rpm), manifold air temperature, and



torque (bmep).




       Figure 31 illustrates NOX  level  variation with speed  for two  engine types.   There are too few



data points to  indicate any trend for 4-TC  units, although the 4-TC  dual  fuel and diesel NO  levels
                                                                                           X


appear to decrease with increasing speed, as  would  be expected for lower exhaust gas chamber resi-



dence times.




       NOV level variation with manifold air  temperature is  depicted in Figure  32.   The 4-TC-G NOV
         X                                                                                       X


levels appear to be  very sensitive to the design air manifold temperature, but  the diesel  and dual



fuel levels do  not.   These results confirm  expectations from physical reasoning that NO  levels from pre-



mixed vaporized fuel combustion in SI engines may be strongly influenced by the degree of aftercool-



ing.  CI engines are characterized by droplet combustion,  and NO   emissions under these conditions



would depend more  on local A/F ratio than on  overall air temperature.
                                                 80

-------
                                                     4-TC (
                                                          ( •  Natural  gas
                                                          ( O  Diesel
CXI
                        Q.

                        -£=


                        CD
                              20
15
                              10
                                                     8
                                O

                                O
                                    I   I  I   I  I  I   I  I  I   I  I   I  I  I   I  I  I   I  I
                                                                                        *
                                                                                        O
                                                                                       «*
                                           4         8        12        16        20

                                                  Number of cylinders
                              Figure 29.  NO  production variation with  number  of cylinders,
                                            y\

-------
co
ro
                      i
                     CL
                        20

                        18

                        16

                        14
?, 12
rt
 x
I 10


   8

   6
                             9.0/0.6, 1%
                             Delaval
                                                         O 2-BS-C
                                                         B 2-TC-D
                                                         V 4-TC-D
                                                         V 4-TC-G

                                                x   a   a/x
                                               17.2/1.6, 9% Delaval
                                               15.8/1.2, 8% EMD
                                                                              12.5/0.4, 3% EMD
                                                                     7.9/0.6, 7% Alco
                         4

                         2

                         0
                                                      x
                                                      a
mean
st'd. dev.
                                                        12        16
                                                         # cylinders
                                                     20
                               Figure 30.  Variation in NO  level  with  number  of  cylinders,
                                                          A

-------
CO
GO
                 25
                 20
                 15
              -C
               I
              Q.
              S-
              01
               X
              g  10
                                                  8
                                                  o
                                       • 4-TC Nat.  qas
                                              (SI)  *
                                       o 4-TC Diesel  and
                                              dual  fuel
                                              (CI)
                                       8
                                       o
                                 I  I
I  I   I  I   I  I  I   I  I   I  I  I   I  I  I   I  I   I  I  I   I  I   I
Q I  I   I  I  I   I  I  I   I  I

  0          200         400         600         800        1000         1200       1400
                                        Speed ,RPM
                                         Figure .31.   Variation  in  NO  level  with speed.
                                                                    A

-------
00
                              25
                              20
                              15
                            I
                           a.
                           .c
                           i.
                           cn
                              10
A4-TC natural  qas
0     (SI)     J

Q4-TC diesel and

       dual  fuel

       (CI)
                                                                  I  I   I  I  I   I  I   I  I  i
                                80       100        120       140       160      180      200

                                                           T  °F
                                                           I *•   I
                                  Figure 32.  NO  variation with manifold temperature.
                                                A

-------
       Finally, Figure 33 illustrates the variation of NOX level with bmep (torque) for 4-TC-G and
4-TC-D, DF units.  The data are scattered.  If one ignores the cluster of CI data around 150 psi, one
can conjecture a trend of decreasing brake specific NOX emissions with increasing bmep for these
units.  No trend is apparent for SI units, except that they are not manufactured with bmep's exceed-
ing 200 psi.
       Based on these preliminary studies, it appears that certain engin_e_jjeslgn parameters (e.g..
rpm, bmep, or manifold air temperatffif?J.,imM...^^1.SLM,,,more of the scatter in NOV levels for engines of
a given type than variations in ambient humidity or temperature.  These_factors will be investigated
further under contract 68-02-2530.
                                                   85

-------
                25
                 20
                                                                 •  4-TC   Natural  gas  (SI)


                                                                 O  4-TC   Diesel and
                                                                           dual  fuel  (CI)
s_
_c

Q-
_c

en
                 15
                                                              O
                                                              O
CD
CTi
      10
                                                                                        O

                                                                                        O
                         I  I   I  I   I  I  I   I  I   I  I   I  I  I   I  I   I  I  I   I  I   I  I
                                                                         I   I  I  I   I  I   I  I  I
                   100
                                 150
        200

BMEP, psi
250
                                        Figure  33.   NO  variation with torque  (BMEP).
                                                      /\

-------
                                              SECTION 4



                                         FIELD APPLICABILITY





       Since manufacturers have applied NO  control techniques only to engines in their emissions labora-
                                          X


tories, they have expressed concern for possible technical limitations of these controls when used



under "field" conditions.  The primary objectives of this task are (1) to evaluate manufacturers' re-



sponses regarding the adverse effects of control techniques they have tested and (2) to assess the



technical limitations that exist for field application of various control techniques.




       This section will primarily discuss the first objective.  The second objective requires addi-



tional contact with manufacturers and field users.  The second objective will be accomplished during



the Affected Facilities task (under Contract 68-02-2530).  Thus, by coupling these tasks a consider-



able amount of consultation can be consolidated.  Therefore, some conclusions regarding the field



applicability of the various control techniques will be reserved.





4.1    SUMMARY OF MANUFACTURERS' "114" RESPONSES




       In response to the June 16, 1976 Section "114" request for information, six of the eight large



bore engine manufacturers commented on potential or observed adverse effects of the emission control



devices they had tested.  Their comments are summarized and briefly discussed below.  DeLaval and



White Alco were unable to comment since neither has done any development work on the various control



techniques.




       In general, the responses produced a consensus that any control technique to reduce NOV emis-
                                                                                             A


sions must undergo a period of durability testing before being released for field application.  For



example, Inge rsolj-Rand concluded that none of the methods of NO  reduction (other than water induc-



tion) that they had demonstrated during short test runs produced immediate adverse effects such



as misfiring, detonation, or excess thermal stress.  More extensive testing would be required to



assess long term durability.,




       Colt also expressed the need for long term testing of the order of one or more years to



evaluate the effect of any N0v control on engine components and maintenance.  This time estimate



appears to be consistent with the 9 to 15 months allotted for development of existing (rather than



emerging) control techniques discussed in Section 5.3.14 of the draft SSEIS.  These existing controls
                                                  87

-------
were classified as changes to operating conditions, e.g.,  retard  or A/F,  or  additions  of existing



hardware, e.g., increased manifold aircooling, or turbocharging.




       GMC/EMD, however, stated that at least three years  of durability testing  was  a  normal  evalua-



tion period for assessing any technical changes that were  to be made  to a production engine.   Thus,



GMC/EMD's time estimate is more nearly equal to the estimates  in  the  draft SSEIS for major develop-



ment programs for controls such as modified combustion chambers or water  induction.




       Colt, on the other hand, establjsJied_t^t_alt£j^iiQ^^                 of  a 4-TC dual fuel
       •*••'"	 "    ^**aanni-«r™.i"-i-r3maa»s™B~—-                                             	   	•'•*••

engine the effect of retard (4 degrees) was to increase the maintenance period for valve  recondition-



1n9^y£S~E£T^en^'  Their estimate of the incremental  cost for this increase in  maintenance was



approximately 0.3 percent of the estimated total  annual operating cost of  this engine which appeared



in Table 7-10 of the draft SSEIS.   These economic aspects will  be given further  consideration during



Task 5 (Cost Revisions) of the follow-on work effort.   It ij_jrrteresting  to note that Colt was the



only manufacturer to quantify an adverse effect_of_a control  technique.




       Furthermore, Cooper and Waukesha noted the customary increases in  fuel consumption, CO, and



smoke levels associated with the application of NO  controls.   Both manufacturers, however, indi-
                                                  X


cated that they had insufficient technical information (e.g.,  exhaust temperature histories) to



assess the potential  adverse effects of the techniques they applied.  Waukesha believed techniques
                                                                          """*'            *"•*        Hi*


such as retard might cause iincreased maintenance dueiiit10--exc£-Ssiiv-ei_exhaust temperatures reqjri£J_ng



more frequent valve replacement.  Waukesha and Cooper indicated that a development program would be



required to determine the adverse  effects of various NOX reduction control techniques.




       In addition, to this  information, Cooper indicated that certain controlled levels  that they



reported in their new emissions data (2-TC natural gas, production engines, see  Section  2), were  un-



acceptable due to misfiring.   These NO  levels represented the greatest NO  reduction possible and
                                      X                                   X


caused the greatest increases in fuel consumption.  All of these control  approaches  which caused



excessive misfiring were combinations of retard (3 to 4 degrees), manifold air cooling (30°F), and



lean A/F ratio.




       Finally,  White Superior indicated that engines sold for compressor  applications are typically



required to meet a 10-percent overload capability during two hours of the  day.   Thus, White concluded



that some of the controls they had applied would require derating of the  engine, since,  in some cases



these engines could not meet his overload limit.   White is presently compiling this  information for



submittal to EPA.




       In conclusion, all  of the manufacturers have stated they believe some adverse effects will  re-



sult from the application of NOX controls.  Only Colt has quantified an adverse  effect of a control

-------
technique.  This effect appears to be primarily economic (i.e., an increased maintenance cost) rather
than a technical limitation.
       Other manufacturers feel that development programs are necessary to quantify adverse long
term effects of N0x controls.  Their estimates range from 1 to 3 years for all control techniques,
rather than the 9 to 15 months estimated in the draft SSEIS for retard, manifold air cooling, derate,
turbocharging and A/F changes.

4.2    SUMMARY OF DEMA AND AGA COMMENTS ON CONTROL TECHNIQUES
       Two organizations, DEMA and AGA also commented on the field applicability of NO  control  tech-
niques discussed in the March 1976 draft SSEIS (References 44 and 45).  In general, these comments re-
flect a concern for the increased cost of purchase and operation of controlled engines, rather than
technical limitations which would prevent the application of a control technique.  These cost aspects
will be given more consideration under Task 5 of the follow-on work effort.  Their comments related to
technical limitations are summarized below by the specific control technique.  Conclusions regard-
ing the field application of these techniques will be reserved until after Task 1, Affected Facilities,
is completed as discussed in the introduction of this section.
Retard (R)
       The primary concern is the possibility of reduced exhaust valve life due to higher exhaust
gas temperatures.  Increased fuel consumption is also noted with the application of this control as
well as some amount of derate with large amounts of retard in naturally aspirated or blower scavenged
engines.  Manufacturers believe 10-percent derate is a practical limit for marketing BS or NA engines.

Reduced Manifold Air Temperature (MAT)
       The amount of cooling is limited by the cooling media (i.e., radiator, cooling tower, pond).
The practical  limit is approximately the ambient air temperature plus 20 degrees.  Lower manifold
temperatures require ground water or refrigeration.  Also, increased cooling implies increased de-
scaling costs.

Air-to-Fuel  Change (A/F)
       A/F changes have limited application to blower scavenged engines due to increased parasitic
horsepower and  manifold air temperatures.   Both of these effects would tend to offset a reduction
in brake specific NO  emissions.   In addition, the durability of the blower is unknown for higher blower
speeds.   Beyond a certain  amount of A/F change in a 4-NA engine, maximum load capability will  decrease
                                                 89

-------
implying derate.  Dual fuel engines will require turbocharger development and careful turbocharger
to engine matching because combustion of dual fuel  engines is more sensitive to A/F than diesel
units.  Increases in A/F for turbocharged gas engines are limited by combustion deterioration.

Derate (D)
       Derating will cause marketing problems for 2-BS units due to increased installed cost and
fuel  consumption.  Derate is generally not effective in reducing brake specific N0x emissions on
turbocharged engines due to the relatively flat A/F curve over the power range.

Turbocharging
       Turbocharging of NA or BS engines results in a low rated turbocharged engine with higher in-
stalled costs and fuel consumption.  In addition, manufacturers indicated that a turbocharged BS or
NA engine requires the performance and development effort of any "new1' engine.  Also, turbocharged
engines incur higher maintenance costs (primarily for the turbocharger) than NA or BS units.

Internal Exhaust Gas Recirculation (IEGR)
       In essence this technique produces a marginally scavenged engine (particularly BS units),
therefore carbon buildup, higher temperatures (increased maintenance) and poorer combustion (in-
creased fuel cost) will result.

External Gas Recirculation (EEGR)
       The concern here is that only one manufacturer has limited data for this technique.  It is
believed that considerable development is required.  The requirement for cooling will mean higher
installed and maintenance costs.
                                                  90

-------
                                              REFERENCES


 1.   Coordinating Research Council (CRC) "Cooperative Evaluation of Techniques  for Measuring NO and
     CO in Diesel Exhaust:  Phase III," CRC Report, 1972.

 2.   "Measurement of C02, CO, and NO in Diesel Exhaust,"  SAE Recommended  Practice J177a. March, 1974.

 3.   J. M. Perez, L. C. Broering, and J. H. Johnson, "Cooperative Evaluation  of Techniques for Measur-
     ing NO and CO (Phase IV Tests),"  SAE Paper 750204.

 4.   Coordinating Research Council (CRC) "Cooperative Study of Heavy Duty  Diesel  Emission Measurement
     Methods," CRC Report 487, July, 1976.

 5.   "A Study of Mandatory Engine Maintenance for Reducing Vehicle Exhaust Emissions,  Volume VI:  A
     Comparison of Oxides of Nitrogen Measurements,"  TRW  Systems Group, CRC  APRAC CAPE  13-68-12,
     July 1972.

 6.   S. R. Krause, D. F. Merrion, and G. L. Green, "Effect of Inlet Air Humidity and Temperature on
     Diesel Exhaust Emissions,"  SAE Paper 730213.

 7.   H. G. Maahs, "Interference of Oxygen, Carbon Dioxide, and Water Vapor on the Analysis for Oxides
     of Nitrogen by Chemiluminescence."

 8.   R. D. Matthews, R. F. Sawyer, and R. W. Schefer, "Interference in Chemiluminescent  Measurement
     of NO and NOg Emissions from Combustion Systems," Draft of Paper Submitted to Combustion and
     Flame  Fall, 1976.

 9.   J. E. Sigsby, Jr., et al., "Chemiluminescent Method for Analysis of Nitrogen Compounds  in Mobile
     Source Emissions,"  Environmental Science and Technology. Volume 7, No.  1, January  1973.

10.   J. H. Tuttle, R. A. Shisler, and A. M. Mellor, "Nitrogen Dioxide Formation in Gas Turbine Engines:
     Measurements and Measurement Methods,"  Purdue University Report No.  PURDU-CL-73-06, December 1973.

11.   Monsanto Research Corporation, "Instrumentation for the Determination of Nitrogen Oxides Content
     of Stationary Source Emissions,"  Volume I, October 1971 and Volume  II,  January  1972.

12.   W. B. Clemens (EPA/Ann Arbor), "Impact on Existing Standards Due to Proposed Instrumentation
     Changes in the Heavy Duty Federal Test Procedure,"  Interoffice Memo, September  18, 1974.

13.   Personal Communication between S. B. Youngblood (Acurex/Aerotherm) and W.  B. Clemens (EPA/Ann
     Arbor) July 22, October 19, and November 1, 1976.

14.   F. S. Schaub and K. V. Beightol, "NOX Emission Reduction Methods for  Large Bore  Diesel  and
     Natural Gas Engines," ASME Paper 71-WA/DGP-2.

15.   "Revised Heavy Duty Engine Regulations for 1979 and Later Model Years,"   Federal  Register,
     Volume 41, No. 101, May 24, 1976.

16.   Diesel Engine Manufacturers Association (DEMA), "DEMA Exhaust Emission Measurement  Procedure for
     Low and Medium Speed Internal Combustion Engines,"  1974.

17.   SAE J177a, See Reference 2.

18.   "General Motors Corporation Statement on Draft SSEIS  for Stationary Reciporcating Internal Com-
     bustion Engines,"  May 14, 1976.

19.   Coordinating Research Council (CRC), "Effect of Humidity of Air Intake on  Nitric  Oxide  Formation
     in Diesel Exhaust,"  CRC Report 447, December 1971.

20.   "Emission Regulations for Heavy Duty Gasoline and Diesel Engines," See  Reference 15.

21.   Environmental Protection Agency (45 CFR Part 1201), "Control of Air Pollution from  New  Motor
     Vehicles and New Motor Vehicle Engine,"  (Notice of Proposed Rule Making), February 1971.

22.   S. R. Krause, "Effect of Engine Intake-Air Humidity,  Temperature, and Pressure on Exhaust Emissions,"
     SAE Paper 710835.
                                                  91

-------
23.  W. J. Brown, et al. ,  "Effect of Engine Intake-Air Moisture on Exhaust Emissions,"  SAE Paper
     700107

24.  J. A. Robinson, "Humidity Effects on Engine Nitric Oxide Emissions at Steady-State Conditions,"
     SAE Paper 700467.

25.  Scott Research Laboratories, "Effect of Laboratory Ambient Conditions on Exhaust Emissions,"
     Project No.  2846.

26.  Private Communication between R.  W.  Sheppard (Ingersol1-Rand) and D. R.  Goodwin (EPA), August 1976.

27.  State of California,  Air Resources Board, 1973,  "California Exhaust Emission Standards for
     Gasoline Powered Motor Vehicles,"  Proposed, November 18, 1970.

28.  C. R. McGowin, "Emissions Control of a Stationary Two-Stroke Spark-Gas Engine by Modification
     of Operating Conditions,"  Shell  Development Company Report, 1972.

29.  Personal Communication between  A. R. Fleischer  (Delaval  Turbine  Inc.) and D. R.  Goodwin (EPA),
     July 30, 1976.

30.  G. L. Touchton and  N.  R.  Diebeluis,  "A Correlation of Nitrogen Oxides Emissions  with  the Gas
     Turbine Operating  Parameters,"  ASME  Paper 76-GT-14.

31.  H. Shaw, "Effect of Water on Nitric  Oxide Production in  Gas Turbines Combustors,"  ASME Paper
     75-GT-70.

32.  W. S. Blazowski and D.  E. Walsh,  "Prediction of  Aircraft Gas Turbine NOX Emission Dependence on
     Engine Operating Parameters and Ambient Conditions,"  AIAA Paper 73-1275.

33.  J. M. Vaught, "The  Effect of Inlet Temperature and Pressure on an Industrial Turbine  Engine
     Exhaust Emission,"  ASME  Paper  75-WA/GT-ll.

34.  W. S. Y. Hung, "An  Experimentally Verified NOX Emission  Model  for Gas Turbine Combustors,"
     ASME Paper 75-GT-71.

35.  J. W. Marzeski and  W.  S.  Blazowski,  "Ambient Temperature and Pressure Corrections for Aircraft
     Gas Turbine  Idle Pollutant Emissions,"  ASME Paper 76-GT-130.

36.  Gas Turbine  International , May-June  1974.

37.  Private Communication  between S.  B.  Youngblood (Acurex/Aerotherm) and J. McDermon (EPA),
     October 1976.

38.  T. Wu and K. J. McAulay,  "Predicting Diesel  Engine Performance at Various  Ambient Conditions,"
     SAE Paper 730148.

39.  F. A. Williams, Combustion Theory, Addison-Wesley, Chapter 3 and 4, 1965.

40.  R. P  Wilson, Jr.,  E.  B.  Muir,  and F.  A.  Pellicciotti,  "Emissions Study  of a Single-Cylinder
     Diesel Engine,"  SAE  Paper 740123.

41.  Private Communication  between C.  L.  Newton (Colt) and D.  R.  Goodwin (EPA),  August 2, 1976.

42.  Private Communication  between C.  L.  Newton (Colt) and D.  R.  Goodwin (EPA),  April  2, 1976.

43.  Private Communication  between G.  P  Hanley (GMC)  and D.  R.  Goodwin (EPA),  January 29, 1975.

44.  Private Communication  between F  S.  Schaub (DEMA) and D. R. Goodwin (EPA), May 12, 1976.

45.  Private Communication  between G.  H.  Ewing (AGA)  and D.  R. Goodwin, September 24, 1976.
                                                  92

-------
                                            NOMENCLATURE
Abbreviation
    Fuel
     0
     DF
     G
Strokes/Cycle
     2
     4
Air Charging
     BS
     NA
     TC
  Pollutants
     N0x
     NO
     N02
     CO
     co2
     HCT
     CH4

Instruments
     CL
     NDIR
     NDUV
   Other
     A/F
     f/a
     SI
     CI
     H
     T
    Explanation


Diesel
Dual fuel
Gas (i.e., natural gas)


2-stroke/cycle engine
4-stroke/cycle engine


Blower scavenged
Naturally aspirated
Turbocharged (and intercooled)


NO + N02
Nitrogen oxide
Nitrogen dioxide
Carbon monoxide
Carbon dioxide
Total hydrocarbons
Methane
Chemiluminescent
Nondispersive infrared
Nondispersive ultraviolet
Air-to-fuel ratio
Fuel-to-air ratio
Spark ignition
Compression ignition
Humidity grains HpO/lb dry air
Ambient air intake temperature
                                                 93

-------
                                            NOMENCLATURE
Abbreviation
     Fuel
      D
      OF
      G
Strokes/Cycle
      2
      4
Air  Charging
      BS
      NA
      TC
  Pollutants
      N0x
      NO
      N02
      CO
      co2
      HCT
      CH4

Instruments
      CL
      NDIR
      NDUV
   Other
      A/F
      f/a
      SI
      CI
      H
      T
    Explanation


Diesel
Dual fuel
Gas (i .e., natural gas)


2-stroke/cycle engine
4-stroke/cycle engine
Blower scavenged
Naturally aspirated
Turbocharged (and intercooled)
NO + N02
Nitrogen oxide
Nitrogen dioxide
Carbon monoxide
Carbon dioxide
Total hydrocarbons
Methane
Chemiluminescent
Nondispersive infrared
Nondispersive ultraviolet
Air-to-fuel ratio
Fuel-to-air ratio
Spark ignition
Compression ignition
Humidity grains HUO/lb dry air
Ambient air intake temperature
                                                 93

-------