EPA-BMW Correlation Program
Abstract

     Exhaust emission and gas cylinder analysis data were gathered.at
the EPA laboratory and at the new !>MU tent facility in Farmingtcm.  The
two laboratories agreed closely in the measurement of CO.  Although
there were, significant differences in the measured levels of HC, KO
and CO   it is believed that these discrepancies were caused by differ-
ences in dynamometer type and ambient conditions.  Because of .the limited
amount of testing done, it was not possible to generate ambient correla-
tion factors for the exhaust emissions.  •

-------
                        EPA-BMW Correlation Program

 1.   Introduction
      «
      BMW has  recently  completed construction and instrumentation of a
 light--duty vehicle  exhaust  emission test  cell in the Detroit area.  The
 facility is located  at Erhart  Motors at  32715 Grand River in Fannington.
 The  facility  becan;e  operational o'n  approximately Hay 1,  1975.   JiHW
 rajueyted a correlation test:  program with  tlie KPA automotive laboratory
 in order to assess  the .operation, of their new test facility.

 2.   Technic_al Digcussion

     2.1   Pro fir am Ob j cc t J.y e

     The  purpose of this study  is  to determine any differences.in exhaust
 emission tests, fuel economy tests  and gas cylinder analysis tests at
 EPA  facilities and the Ul-iW laboratory.  In addition to  comparing these.
 test  results, this study also  compares the test  equipment used  at each
 facility.                             '      .          ,

   2.2   Facilities and Equipment     .                           •        •

         2.2.1  Test Sites

     The  BMW facility  consists  of one test, cell  v;hich has temperature
but not  humidity control.  EPA  used  their  designated "master cell" which
is cell number 5.  This facility has both  temperature and humidity control

     The  equipment and instrumentation used  within each  test cell is
listed in Table I.  Perhaps the most significant  difference in  test
equipment is the dynamometers.  The EPA used  a direct-drive inertia
wheel unit, while BMW  has the belt-driven  type.

          2.2.2.   Test  Vehic3.es

     One  test vehicle was used  in this program.   This was a 1975  3.0  Si
BMW equipped with a 4-speed manual  transmission.   The engine was  an.
electronic fuel-injected 6-cylinder,- equipped with a thermal reactor.
The inertia and actual h.p. setting  for this  vehicle were 3500  Ibs, and
11.2 IT. p., respectively.  The same  fuel (commercial leaded)  was  used
for all  tests.          .     ^

    2•^  Program Design

     The  test was designed so that, after  several .vehicle tests  had .be-?.n
done at EMW,  three tests would be conducted  at the  EPA.   The day  prior
to testing at the KPA,  five tests were conducted  nt BMV'.   For purposes
of the st.-Jtiotiool analysis, the. last three  of t:lu:se. te?;t:s  were  used.
In aridition to vehicle tests, a set  of gas  cylinders was  also analysed
at each lab.

-------
           Table.I  Test". Site Equipment and Instrumentation
Analyzers

   •JIG


   CO

   NO -
     x


   C0n
Analyzer  Bench
CO Cond itionin
CVS
Dynamometer-
Driver 's Aid
Computer
                             EPA
BecbiKm  AGO,  0-30  ppm,
I12/1*2  fuel

Bendix 8501,  0-500 ppra

TECO 10A,  0-100  ppm,
0- Osone Sourer.

BccUman  3ISA,  0-3.3%

Houie.built  according to
Federal  Register

Ascaritn,  Silica Gel
in Common  Tube.

Ford .Pliilco,  CFV


Clayton  CT-50, Direct-
Drive Inertia, Au'-.o
Loading  not usexl

Varian,   5" -  60  mph,
6"/™.n.  Preprinted Trace
IBM 370, Off-line
I I'M Model R55,  0-100  ppm
100% H2  fuel

Il&B Type 2T, 0-500  ppm
TECO 10, 0--100
Air Ozone Source
ll&B Type 2T, 0-2%
          according  to
Federal- Register

Ascarite, Silica Gel
in Separate Tubes

Scott 302, 6-bag, AP  = 22r
1120

Clayton C.150, Belt-Drive
Inertia, Manual Loading
Servogor, 7.6" = 60 mph
12 cm/tain, Hand-made trace
Hand-calculated

-------
                                   -3-
     2.4  Test Procedure

          2.4.1  Vehicle Preparation

      All vehicle tests were "hot-start" so vehicle preconditioning  con-
 sisted of stabilizing the vehicle at some constant; condition  prior  to
 the start of each test.  Oil temperature was used as the control  con-
 dition.   Prior to testing, the vehicle was operated at a steady speed
 of 25 •.nph until the oil temperature rose to 72°C.  Then the vehicle was
 brought to an idle.  When the oil temperature readied 75°C  (.1.67 °F)  the
 test was begun.           .

          2.4.2  Em i ssion Test s

      The driving cycle used was the urban dynamometer driving schedule
 (LA-4).   The vehicle began the cycle at idle with an oil temperature of
 75°C.   Emissions were measured.according to the .1974 FTP.  Between  each
 test the vehicle was shut-off and allowed to cool for approximately .15
 minutes  so that  the following test could be started at an oil temperature
 of 75°C.                              .                 -

          2.4.3  Fuel Consumption Tests

      Fuel Economy was calculated for each LA-4 cycle based on .the carbon
 balance  technique as described in the Federal Register.

          2.4.4 •Cylinder Gas Check

      Five gas cylinders were analyzed at both test sites.  These cylinders
 were analyzed on the same, analyzer trains that were used for the vehicle
 emission  tests;   Three concentrations of CO  (nominally 0.8%, 1.7%, and
 3.2%)  and  two concentrations of'CO (nominally 170 ppm and 1300 ppm) were
 used.                          '

 3.   Data  Analysis

      In  addition to the calculation of means and standard deviations
 for  HC, CO,  CO-,  NOX and  fuel economy at  each test  site,  a Student's
 "t"  test  was  used to detect any  significant differences between, the
 test  sites.

 4.   Test  Resuits

     4.1   Eiiiis s ion nnd F u el Economy Da t a

      Data and statistical analysis results from the emission and fuel
 economy  tests arc contained in Table 11.   -As shown,  the two labs agreed
 in CO measurei'ient;  however,  the  10-. <-'« difference .in C0.; :;iuy..!ed high
..'• tafifatJCiil  significance.   The BMW Ir.b r.i-jar.ured ?.'.'. 3'/.: Tcv/ri: SiC ,-^nd  13.27..
 higher NO.  and these differences vere. at the. 97% and 98/j confidence.
 level, rel:jiC:cLlvely.   Repeatability of the IIC ni-asuremout \/as much
 poorer than  repeatability of tlie other emissions.;.

-------
                         -4-
Tablc II    Vehicle Emission and Fuel Economy Data
and Statistical Analysis Results (Data uncorircctecl
for any ambient or instrumentation difference)
Test
BMW
1
2
3
X
0
 A..LUU/0
XEPA
t-value
Confidence
Level -
HC, sa
' mi

0.54
0.64.
0.51
0.563
.068
12.1 %

0.71
0.90
0.78
0.796
0.096
•12.0 %
-0.233
-29.3 %
3.43

97 %
CO, £? '
' nu.

7.20
7.49
7.58 •
7.42
0.20
2.7 %

7.58
7.37
7.62
7.52
0.13
1.7 %
-0.10
-1.3%;
.72

<60 %
rn gm
LO „ , r
2 ' mi

656
647
654
65?.
4.7
0.7 %

595
585
585
588
5.77 '
0.9 %
64
10.9%
-14.86

>99..9 %
NO , £S
X Till

2.26
2,06
2.09
2.14
0.11
5.0 %

1.91
1.89
1.87
1.890
.020
1.0 %
.25
13.2 %
-3.89

98 %
Fuel Economy, — :~-
gal

13.3
13.4
13.3
13.33 	
.06
0.4 %

14.6
14.8
14.8
14,73
0.12
0.8 %
-1.40
-9.5 %
-18.78

>99.9 %

-------
                                    -5-
     Due to  the fact  that each facility  tested on  only one day,  the
ambient conditions changed very  little within each test  site.  Therefore,
it is not reasonable  to attempt  to' obtain ambient  correction  factors  for
this vehicle;.  Also,  there were  only very small differences in temperature
and humidity conditions between  the test sites.  However, Lire barometric
pressure at EPA was about 0.4 in. Hg lower than at  BMW.   In an attempt  to
adjust the data for this difference in barometric  pressure, the  correc-
tion factor obtained  from, the 1975 MVMA correlation study (-"-'was  applied
to the dai'a.  In doing this, it was assumed that barometric pressure had
the sauic percentage effect on the emissions from .vehicles used in both
studies.  So, obviou.sly, correctness of the result depends on validity
of this assumption.   In the MW1A study, a -0.4 in. Hg difference in barometric
pressure v/ould have accounted for nearly 30% of the observed difference
in NO  and about 20%  of the observed difference in HC in the KMW-EJ.'A data.
And these adjusted differences are at the 94% confidence, level, which is
a significant reduction of the 97% and 98% confidence level of the un-
ccrrected data.

     Also in the 1975 MVMA study, a difference in  dynamometer load was
observed ."between direct-drive and belt-driven inertia .wheels; -and this
resulted in a significant effect on NO .   The MVMA program yielded a
dynamometer type vs 1-:0X relationship which would account for about 35%
of the observed NO  difference in the BMW-EPA data.
                  J»

     The statistically significant difference of 10.9% in C02 values
cannot be explained by dynamometer and/or ambient  effects.  Results of
the 1975 EPA-MVMA correlation study indicate that about a 4% difference
could be a dynamometer difference and about 1% could be attributed to
the difference in barometric pressure.  But the remaining 6% difference
would stiJl be significant at greater than a 99% confidence level.

    4.2  Gas Cross-check Tests

     Results of  the five, gas cylinder analyses are shown in Table III.
As listed,  the only analysis that was not in good agreement was the
nominal 1.7%. COo.   BMW measured this concentration as being 4.2% higher
than did the EPA.   Since this analysis was conducted on the same analyzers
that were used for the vehicle test, this discrepancy accounts for part
of-the. 10.9% observed difference in C02 emissions.  Taking into account
the 4.2% analyzer difference, a 4%' dynamometer effect and a 1% barometric
pressure effect,  the  10.9% observed difference in C02 is reduced to only
about a.'2% difference.      ..                •   •         .

->•  Con c.1.11 ^ j ons

     1.  CO data showed good agreement between the two laboratories.

    .2.  Differences  in dynamometers (direct-drive vs belt-driven inertia)
and barometric pressure are believed to be responsibile for the observed
difference in NOX values.

-------
                   -6-
            Table 111
Results of Gas Cylinder Cross-Checks
Gas .
CO
CO
co2
co2
co2
-•••-" — ' 	 '"" • - '
BMW
169 ppm
1321 ppm
0.80 %
1.74 %
3.24 %
EPA
167.5 ppm
1323.8 ppm
0,80 %
1.67 %
3.21 %
A'(BMW-KPA)
1.5 ppm
—2.8 ppm
0.00 .
0.07 %
'0.03 %
A,%
0.9
-0;,2
0.00
4.2
0.9

-------
                                   -7-
     3.  The only difference in gas .cylinder analysis greater than  1% was
a 4. 2%. -difference (lli'W higher than EPA) on a C02 concentration of about  1.72.

     4.  The above mentioned 4.2% difference in CC^ analysis is responsible.
for part, of the observed CC^ emission difference of 10.9%.  The remaining
difference is believed to be caused mostly by differences in. dynamometer
type and barometric pressure.
     1.  In future correlation work, it would be desirable to have testing
performed at one location on more than one day.  This. would most likely
result in greater ambient condition variations within each laboratory
and would make possible the calculation of ambient correction factors.

     2.  During vehicle tests at BMW the te.st cell temperature rose from
about 73°!' at the beginning of a test to approximately 83°F a-t the. end
of the test.  Decreasing this temperature; rise would most likely .improve
testing repeatability.   (The temperature, range encountered during the EPA
tests was from 72°1< to 78°F.)

     3.  Dynamometer type should.be standardized or . quantitatively charac-
terized in order to improve laboratory correlation.

-------