EPA-AA-EOD-80-14
EPA Technical Report
Comparative Analysis of GM
Inertia Weight Distributions
MY75-MY81
by
Don Paulsell
August, 1980
Engineering Staff
Engineering Operations Division
Office of Mobile Sorce Air Pollution Control
Environmental Protection Agency
Ann Arbor, Michigan 48105
-------
Comparative Analysis of Gh
Inertia Weight Distribution for
MY75-MY81
Introduction:
General Motors has claimed that the change from 250 pound dyno inertia
increments to 125 pound intervals has penalized their fuel economy
results relative to the 1975 test procedure. EPA contends that this
would only happen if a manufacturer's fleet were biased in such a way
that the new test weights were higher than the old test weights. GM's
July 27, 1979 petition presented data on a fleet ot eleven cars to sup-
port their claim of .6mpg penalty due to "all" test procedure changes
since 1975. The one noteworthy difference in the test conditions for
this program was that nine of the eleven moved to the higher inertia
weight. The other two remained the same. None moved to a lower inertia
as would have occurred for an unbiased fleet.
EPA's analysis indicates that GM's weight reduction program resulted in
a fleet of vehicles that were more often tested at inertia increments
below the actual vehicle weight. However, these reductions in vehicle
weight do not translate to real world fuel economy improvements of
equal magnitude.
This paper summarizes the analysis of these inertia weight versus
loaded test weight differences tor GM and other major manufacturers.
The data were derived from the certification data in which over 12,500
data entries were considered from 1974 to 1980.
Discussion of Analysis:
Two essential parameters were available for 5421 of the 12,500 certifi-
cation entries. Curb weight and inertia class for the EPA test were
used to calculate the loaded test weight and weight difference, DEL_WT.
The DEL_WT is the value of curb weight plus 300 pounds minus the value
of the inertia test setting. Hence, a positive value for DEL_WT means
that the vehicles' actual loaded weight exceeds the inertia simulated
for the test. Therefore, the fuel economy measured during the test
will likely overstate the real world fuel economy.
The statistical summary of the DEL WT data are shown on Tables A
through E and on Figures 1 to 4. Individual plots were made of DEL_WT
versus inertia weight for each manufacturer and for each year. The one
observation that was readily seen from these plots was the tendency for
GM's DEL WT values to be biased positive at inertia values of 3500
pounds or less. Their data for inertia values over 4000 pounds are more
equally distributed.
-------
Tables A through E show how many data points fell on the positive and
negative sides of the inertia values for each test year. These, data
were stratified above and below the inertia of 3875 pounds and plotted
on Figures 1, 2, and 3 to illustrate how the smaller inertia classes
are biased to the positive side. All the domestics have one or more
occurrences of this positive bias for some specific model years such as
Chrysler in 1975. However, GM's bias is consistently positive and is
the largest both in numbers and average DEL WT, as shown on Figure 4.
Chrysler and AMC data indicate DEL WT averages that are generally
negative. The use of smaller inertia increments appears to decrease
this average. This reduction should therefore minimize the potential
for understating the CAFE data for these manufacturers.
These figures show several other noteworthy characteristics. First,
when all inertia weights are used (Figure 3), the overall distributions
appear equally balanced. However, the bias becomes more pronounced at
the lighter inertia values, where the advantage of a change from a
higher inertia to a lower inertia is maximized. When the 125 pound
inertia increments were implemented for 1979 and 1980 test years, the
data for each strata assumes a balanced pattern.
Figure 4 shows that the overall average DEL_WT generally got closer to
zero, except for the "OTHER" category, which encompasses all non-
domestic manufacturers. In fact, the overall averages for 1979 and
1980 exceeded -70 pounds! This should be theoretically impossible since
no individual DEL_WT should have exceeded 62.5 pounds. If it did, the
vehicle would be tested at the next inertia setting.
This anomally was investigated with Certification engineers and the
explanation was found (see attached memo). In 1979, Certification
issued a policy that allowed any manufacturer whose estimated loaded
vehicle weight was close to being in the next higher test weight to
elect to test at the higher weight. This policy avoided the problems
of having to retest vehicles whose actual production weight exceeded
the limits on the estimated weight class. Hence, a significant number
of manufacturers have elected to do this. The impact of this option is
to offset the DEL__WT data by 62 to 125 pounds, thus causing the ob-
served negative bias. However, this is a self imposed bias rather than
the result of the new inertia increments.
(Conclusions/Recommendations:
Vehicle inertia and curb weight data indicate that the smaller inertia
classes for GM were biased to the positive side (actual weight greater
than test weight).
The use of smaller inertia increments tends to recenter the distribu-
tions for all inertias to eliminate the biases, while at the same time
-2-
-------
provides incentives for even a small weight reduction which can be re-
flected in the test conditions.
GM's test fleet would likely receive a lower (but more representative)
overall CAFE under the smaller inertia test increments, but only be-
cause of the biased conditions under the broader inertia increments.
-------
APPENDIX
Table A - AMC DEL_WT Data
Table B - Chrysler DELJWT Data
Table C - Ford DEL_W1 Data
Table D - Gh DELJWT Data
Table E - Average DEL_fc/T by MFR/YR
Figure 1 - DELJWT (+/-) Data for Inertia Values Less than 3875
Figure 2 - DEL_WT (+/-) Data for Inertia Values Greater than 4000
Figure 3 - DELJWT (+/-) Data for All Inertia Values Combined
Figure 4 - Average DEL_WT Values Plotted by MFR and YR
Attachment A - Memo from R.E. Harrington, dated 4/24/79, entitled
"Policy for Handling Test Weight Intervals in
Certification"
-------
15' ET-RCltiNCY LINE NO. C-.u i:-3
TEST Y£
/r\l££T/Al
CMss2
3
tiSo *
25oo '
2.625" b
275o 9
2275" "
2eoo "
3125 l2
?Z5o rj
33V5 l4
35oo 15
3425 16
275b 17
3875 >8
V25o 20
^5bo 2l
475*0 "
$000 »
27
/o >•
TOT A L. 2a
A\l6. Awn0
31
1 n*?4-
-t- -
i 0
0 }
i . 0
0 £
3L -7
^
-^1
i
2 IT7S
•f- -
10 A
^8* 7
•/ o
3 ^
^5" -i/
i~4
33
3 (^^^
-h ~
i 0
H 7
*^ 1 I
"B -M
0 /o
^ -MC
^5
~ £"/
4l*m
-f -
^ 0
£ 4
S l€
5" //
! b
3L» -W
4>^l
-67
5 /^7S
-h -
3 0
CJ ^
( Q
b l
i 7
H' -T
0 1
53 OS"
4^
— i (j
6
^¥-7?
-h -
7 0
"H /^
t7 «
'lr 3o
2 31
0 i
rjo -/jD
3 GO
— —
a
9/T?*7
V -
} 0
"*\ /'"\
c/v .^
o *•(
i 0
0 <*
1 -0
! 5"
1 /
0 J
0 2
0 2
2 !7
"Pv S'
-70
lo/??d>
-f- —
/ 0
1 -!
i 0
o i
/"/ j
0 j
^ h1
7
-'A'^
11
TT-So
-f -
^ 0
? 0
i r
/ 0
i ^
! 1
L ^
1 i
') i
0 i
o - a
0 i
0 3
;• ~^i
32.
i-nA / /
ft L~-\~m
-\~
^ '(
7 ,,
3
35" 3
j
l ;
j
b% l>
i l
^
N
0
S "
0
o
/'-// Sf>
1 c( J
-------
FOi:i\JCY LIME NU.
TABLE.
OF
TEST Yfc
'2S2
ZZSo 5
2.375 6
25o O 7
2625 "
275o B
2875 "»
3000 u
3/2S '2
325o "
35oo i5
3625 «•
3750 "
3*75 iB
V£5c? 20
m?oo 21
^AT^O 22
5&OO "
5250 24
5500 "
6000 >B
27
M 28
~p>TA t- 2a
/tys.AtOT
31
1 I6!!1*
t-
2 2.
o z.
O 2
O /
2. V
GJ
-J2.&
2 Hl^
-H -
7 0
-31 0
3^ /fe>
37 /9
7 T S"
vS !l?
1^ n
!C?S~
-i- :; :r
1}°1~L6L
+• -
0 i
S i
a '7 L/y
'.-•:> iX5
/s- id
I 7
'ft 27
/S"2
-*t>
4 (977
•*" '
., 0
1 7 ^
0 s'
c] 0
^a 3^
lo asi
6 /?
"2. /^"
S7 103
1 9 a
-V7
MITS"
-h -
£" cu
0 ^
o f}
u1/ 17
£" ^
? 7
0 1
^^ L,L/
\ j,^
^y~7
t>
74-7^
+ -
/ ^
22^ //
0 6
7 /
57 /
1/3 /If
22L 101
CZ 1 i/-2
1 1 it "3
m 320
&&f-
— -
a
H75
•f- —
/ 0
2- Z_
Z. 1
/ o
/ I
^Z f
/ s
& V-
5 S
2. 4
O /
Z o
/ 0
26 17
53
-£>
•mo
/ 2.
2- 6
O 3
/ 2_
0 /
/ 3
£ f
2. //
/ 3
I 2.
O 2-
o /
15 V-d
55
-35"
11
7*? -SO
-f ~
/ O
3
3 7
0 3
2- -2.
C> /
/ o
// ^7
10^
"^L^.
/ ^
/ C
2*» IS
^ 7
O *?
/ 2.
~7 /
53- 1
/ 2,
3 7
1 S*
/2S /2Z
7 /6
?5 /0g
0 3
S'S 45
^ /
12. ^/ 3
5?S 5? 7
772.
.
-------
IS' EiTOt-NCY LINE NO. tiGCJIJ
TA8L.E. C -
TEST
Zooo
2575
25co
3J2S
3?75-
3625"
375o
3*75-
rlAl
3
? *
f"5 5
5,
0 7
9 •
5 n
3 '"
«•*
2 15
M^
D "
r »
? 19
? «>
5 21
^ 22
J 23
> "
\ 25
27
28
(__29
WT°
31
i /cpif
+ -
•V
•'
•~\ '"'
!oO
+ t>EL
— DEL
2I97^
•»- -
'•: '
-,•
5*f
43=i
*7 (J •> 1 C?/
^•f? lofo
vo
wf" "
WT -
1*176
-t -
;-
15 0
V, ;\
^3
IS'I
"7
/VC-TC
• \
4/977
•h -
;j ,
1 >
".-'
-;
' • '/
,4 - ••';
27
A«~ CO
5)^7^
•f- -
<'. ••
', ; , ,
'
1
/ •-:•':
•" ^
ll
C)
G>/€£/l
L-ES
7
'i
•:
,-
"7 (;
.
V.
j i ."
.- . • -."'•'
— ...
eSr""Ci
1 1
'/9^
-h —
1
,.
• ;
II -';
- '-/O
)£. t &/T
''
,i
~M~8d
^ -
/ 0
i /
/ 0
/ /
J 0
1 1 12-
-> o
^ o
2- 4~
0 f
3 5
1 7
/ ^
4 3
*? V-
_—
., ji / /
fi* *
i .
12 o
1 /
1 0
5Z 1
/ 0
/rr 76
3^r
o
-Z- H-
o i
6^ /of
/ 7
/ S
£/- 3
l%7 111
1 /S
/^•/ £V
2- 0
^ 7 ^7
/ Z-
S3 ^7
I32.1L
7W $7%
5*250
-------
TABLE D ~ P15TR.IBOT/OM OF
FoK.
(l«n4-
TEST y/e
JhJE/ZT/A '
^ / ^ ^r^x ~
\^-Cp*^7«J *J
3
2ooo *
ZZ5o *
"2375 b
2500 7
275* J
2#7S 10
3ooo "
3IZ5 "
3zso '3
3375- •«
3Soo 15
3^2.5 16
375o l7
3#7S ia
40oo 19
*fzso 20
45*00 21
lfj5p "
Sooo "
ST25"O 24
55oo 2S
6000 2C
27
N 2B
TOTAL. as
A\fe.AU)T°
31
*
1 l^7*f
-h -
0 1
l 0
0 1
/¥ 3
(o 3
3 3
•?r
3tf //
*73>
P ETT. u
' l«175
•t —
3V 2
0 /
J 1976
•h -
9 I
3 0
23 10
23 /#
68 3/
y& y6
?4 3.4
~^0 5"
380
iV9 131
t3y.
3T =~
OT~ =
2& 3
/O S
3S 36
?3 39
13 27
if g
35*0
131 /l*7
-37.
teTUftl
J 1^77
H- -
7 0
0 6
/3 0
vy s
V5" 2^
VI II
3 }0
2 7
23^"
155 W
*4?.
. u;ei<
b i97_?
•f- -
2 0
0 9
/3 ^
V/ ^
33 ^^
37 2.1
y a
j g^
? 0
201
/ay 77
+31.
JHT G
L
-
_£SS
7y--7£
•»- —
V/ 3
0 /4
^ 0
&) /t
//
b
J TES'
1979
-»- —
3 0
0 S1
/ 0
(o 0
3 7
0 d>
1 0
0 2
3 0
w i
3 3
2 /O
/6 13
0 (o
5" S
o y
o /
0 2
lay
5-3 7/
~i*i.
r W£/
nso.
+ -
/ <5
3 /
/ /
0 /
/ V
^ 6
/ 6
5" /i
3 7
y v
/ 3
i 2
0 I
1 0
71
*i V7
-^o.
SHT
^
75-HCL
*rf 0
0 S"
/ 0
9 /
y ^
O k
\ 0
0 3
3 /
J/ S
? 9
? /(.
a/ M
3 J?
9 /^
/ 7
' 3
0 3
1 O
/O ^34
-------
• I F 1-lC.L IMCY L!Nt£ Nl.J. Lii-ifjij
YEAR.
MFfc
flfWC
18
19
21
22
23
25
2o
27
28
23
JO
31
x
SD
N)
x
5D
10
X
N
x
N
X
M
_X_
PEL.
1°!!^
148
/fS
36
35
73
12-1
33
/o?
137
52.
33
13
35
5IO
/O
It6
ZZ
I35
1*5?
/6/
-37
178
-23
1(0
1^77
62-
'67
237
235T
I2o
-3
//Z
16*
2.01
31
-7
in
7*0
AAJL
-70
S3
Tfi*W»
-------
-------
»»:»»-»:-::"::::-:
^^
:i^
-------
-------
Au)T BY TE-ST VEflf^. Ffc*.
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
DATE: April 24, 1979
SUBJECT: Policy for Handling Test Weight Intervals in Certification
FROM-. R> E. Harrington,
Certification Division
T0: All Certification Branch Members
With the modification of the light duty vehicle and light-duty truck test
procedures to incorporate equivalent test weights within inertia weight
classes, several operating and policy questions have arisen which need
clarification and resolution in order to promote consistent interpreta-
tion and inplementation among the certification teams and the Fuel
Economy Group. This memorandum will serve to formally state those policy
decisions and clarify the operational implementation of this regulatory
change. At some time in the future, we intend to formally communicate
these policy decisions and operating guidelines to the industry via an
advisory circular. Until such time, you should feel free to informally
share this guidance with your respective manufacturers.
As in the past, manufacturers will provide their best estimate for
vehicle weights. In selecting test vehicles, EPA will specify equivalent
test weights consistent with the declared loaded vehicle weight of the
vehicles.
The regulations [40 CFR 86.080-26(a)(2)] explicitly allow a manufacturer
to test a vehicle at the next higher equivalent test weight if the esti-
mated loaded vehicle weight is within 100 pounds of being included in the
next higher inertia weight class. However, under the authority of 40 CFR
86.079-27 ("Special Test Procedures"), manufacturers will be allowed to
similarly test any vehicle at the next higher equivalent test weight even
if not within 100 pounds of being included in the next higher inertia
weight class.
If, due to a change in design intent, a loaded vehicle weight changes,
the manufacturer must correct his application either via an update
(prior to certification) or a running change (after certification). If
the loaded vehicle weight increases to the extent that a test vehicle
selection would change or a correctly selected but already tested vehicle
would be appropriately tested at a higher equivalent test weight, manu-
facturers will be required to emission test to determine certification
compliance. Unless the certification team has identified some unusual
circumstance which would suggest the need for a new emission-data ve-
hicle, for a properly selected vehicle only a back-to-back test would be
necessary to appropriately redetermine compliance. Therefore, even
before initial certification, back-to-back testing will be allowed to
account for a change in design intent. Of course, if a new vehicle
configuration would be selected as a result of a design change (e.g., a
different "B" vehicle), then a new 4,000-mile emission-data (before
initial certification) or running change (after initial certification)
vehicle would be required.
EPA Form 1320 6 (Rev. 3-76)
-------
-2-
If the manufacturer determines that due to slippage from design intent,
production vehicle weights have changed to the extent that the appro-
priate equivalent test weight has changed, the manufacturer must correct
his application via a running change. If the corrected equivalent test
weight is lower, no running change test will be required. If the equiva-
lent test weight increases by only one category, running change testing
would only be considered in those cases where the original emission-data
vehicle marginally passed. In these instances, the need for testing
should be considered on a case-by-case basis by evaluating the expected
impact of the weight change (that is, the vehicle's emission sensitivity
to weight change and consequently the likelihood that it would fail
standards if retested). If the production slippage results in an in-
crease of more than one equivalent test weight category, running change
testing would be more frequently expected, but again evaluated on a
case-by-case basis as to the likelihood that the vehicle would fail if
retested at the higher equivalent test weight. In cases where the
production weight increased such that the vehicle was appropriately
tested in a higher inertia weight class, testing will usually be required
unless the manufacturer has previously opted to voluntarily test at an
equivalent test weight in the higher inertia weight class.
With respect to carryover consideration, if the vehicle has been tested
at a higher test weight in a previous year and the request for carryover
is to represent a vehicle at a lower test weight, carryover will not be
jeopardized. If the converse is true (request to carryover from a lower
to a higher test weight), carryover will be jeopardized. Prior to EPA
granting carryover, the manufacturer will have to present evidence demon-
strating that the increase in weight will not significantly affect
emission levels.
As is our current practice, requests for carryover, running change
approvals, etc. must be coordinated with the Fuel Economy Group so that
they may assure compliance with the fuel economy regulations.
For your information, it is our current understanding that SEA testing
will be conducted according to the vehicle weight estimates provided in
the manufacturer's certification application. Additionally, for the time
being (probably at least through the 1980 model year), HSED will only
consider inertia weight classes when determining if production vehicles
are covered by the certificate (that is, weight changes from one equiva-
lent test weight to another will not be cause for enforcement action as
long as the intertia weight class has not changed).
------- |