Ford Explorer Recall
- Firestone Tire Requests NHTSA to Begin A Defect Investigation concerning
the Rollover Problems of the Ford Explorer, Mercury Mountaineer &
Mazda Navajo SUV Vehicles after a Tire Tread Separation of a Rear Tire
NHTSA Analysis of a
Request for a Defect Investigation Submitted by Bridgestone/Firestone,
Inc.
I. Introduction:
On May 31, 2001,
Bridgestone/Firestone, Inc. (Firestone) submitted a request to the National
Highway Traffic Safety Administration (NHTSA) to initiate a safety defect
investigation regarding the handling and control characteristics of Ford
Explorer[1] sport utility vehicles (SUVs) following a tread separation
of a rear tire. (This request includes Mercury Mountaineer and Mazda Navajo
vehicles). Although Firestone’s letter to NHTSA was not in the form
of a formal petition for defect investigation, the agency is evaluating
the request as if it were a petition for a defect investigation.
Specifically, Firestone
alleges that “…certain of the Explorer models will experience
an ‘oversteer’ condition in most circumstances following a
tread separation of a left rear tire, an event that is clearly foreseeable”
and that “…Explorers as tested are defectively designed in
that they have an inadequate margin of control (due to insufficient understeer)
to permit control by average drivers in the foreseeable events of tread
separation during normal highway driving in most load conditions and turning
circumstances.” The amount of “understeer” of a vehicle
is characterized by its understeer gradient, and a brief description of
understeer gradient appears later in this document. Firestone also contends
that this “inadequate margin of control…can make the Explorer’s
handling imprecise and unpredictable in foreseeable circumstances, such
as tread separation, where precise and predictable handling is essential
to safe vehicle control.”
With its request,
Firestone submitted a document titled, “Firestone Statement concerning
Dr. Dennis A. Guenther’s Engineering Analysis of Ford Explorer Margin
of Control in the Event of Belt/Tread Separation” (Firestone Statement),
which refers to a study conducted by Dr. Dennis A. Guenther of FTI/SEA
Consulting. The study included dynamic tests of four 4-door, rear two-wheel
drive (4X2) SUVs: two models of Ford Explorer and two peer SUVs.
In a meeting with
NHTSA staff on June 15, 2001, Dr. Guenther indicated that additional tests
of previously tested vehicles and tests of additional peer SUVs were planned
and that complete results and analyses of all testing would be made available
to NHTSA's Office of Defects Investigation (ODI) by the end of July 2001.
However, additional information was not submitted until August 22, 2001,
when Firestone submitted a document titled, “Engineering Evaluation
of Explorer Directional Control (Updated 8/16/01)” (Firestone Update),
authored by Dr. Guenther. The Firestone Update includes results of testing
of a two additional models of Ford Explorer and of additional tests of
one of the Ford Explorers tested in the earlier program, but does not
report results for any additional peer SUVs. It also included various
observations related to the circumstances under which Explorer crashes
occurred. In making these observations Dr. Guenther “examined several
crashes involving Explorers and reviewed police accident reports relating
to Explorer accidents.”
On September 24,
2001, Firestone submitted a document titled, “Supplement to Engineering
Evaluation of Explorer Directional Control” (Firestone Supplement),
which includes results of testing of two additional models of Ford Explorers
and of one additional peer SUV. The Firestone Supplement does not include
any discussion of these new test results, but states “…the
test results from one additional vehicle are being analyzed at this time
and will be included in a final report for this initial evaluation program.”
In a meeting with
NHTSA staff on December 4, 2001, Firestone submitted a final version of
Dr. Guenther’s report titled “Engineering Evaluation of Explorer
Directional Control,” dated November 30, 2001 (Firestone Final Report).
This document included the results of all of the testing previously submitted
to ODI, included the results of tests of one additional peer SUV, and
revised some of the conclusions and observations in the August 16, 2001
submission. At the meeting, Firestone also presented additional information
to NHTSA (Firestone Presentation). This presentation included a video
tape showing a demonstration of 40 mph "step steer" tests of
three SUVs (including one Explorer), a review of five fatal crashes involving
MY 1995 and later Explorers that experienced rear tire tread separations,
a list of crashes of Explorers equipped with non-Firestone tires that
experienced crashes following rear tire tread separations, a discussion
of various causes of tire tread separations, statements related to the
"undesirable nature" of vehicles that exhibit oversteer characteristics,
and a statistical analysis of data from NHTSA’s Fatality Analysis
Reporting System (FARS) comparing the fatal crash experience of various
drivetrain and body style Ford Explorers to that of other SUVs.
II. The
Firestone Statement:
The Firestone Statement
presents the "preliminary" results of dynamic testing of certain
SUVs that was conducted by Dr. Guenther at the request of Firestone’s
counsel in connection with “…the preparation of Firestone’s
defense in personal injury litigation…” related to crashes
involving Explorers. A “Statement of Dr. Dennis A. Guenther,”
which accompanied the Firestone Statement, indicates that Firestone’s
counsel requested that Dr. Guenther “…carry out an investigation
of the directional control of the Ford Explorer following a tire tread
separation” that included accident reconstruction, review of police
accident reports, and vehicle measurements. The Firestone Statement includes
information related to the dynamic testing of two MYs of 4-door, 4X2 Ford
Explorers, a MY 1996 and a MY 2000, and two peer 4-door, 4X2 SUVs, a MY
1996 Chevrolet Blazer, and a MY 2001 Jeep Cherokee. Also, the MY 1996
Ford Explorer was tested with both Firestone and Goodyear original equipment
manufacturer (OEM) tires fitted to the vehicle.
The Firestone Statement
only includes the results of a portion of the dynamic testing conducted
by Dr. Guenther, the “constant radius circle” tests to measure
the understeer-oversteer characteristics of the vehicles tested. (The
understeer gradient results from these tests, as well as those for the
other vehicles tested by Dr. Guenther, are summarized in Table 1.)
The other tests
performed, for which results were not included in the Firestone Statement,
are:
“step steer” tests to measure vehicle response times as related
to lateral acceleration and yaw velocity response and to measure the gain
of these responses in relation to steering wheel input; and
“frequency response” tests to measure lateral acceleration
gain, yaw velocity response gain and the phase angles at the frequencies
tested (up to 3 to 4 hertz (hz), cycles per second).
The following are the descriptions of these three types of tests included
in the Firestone Final Report:
"Constant Radius
Circle – The vehicle is driven around a 200-foot constant radius
circle with very slowly increasing speed beginning at 0.0 mph. The driver
adjusts the steering angle (by turning the steering wheel) as necessary
to keep the vehicle on the path of the circle. Test runs were done in
both directions, clockwise and counter-clockwise…";
"Step Steer – The vehicle is driven on the test pad area in
a straight line at a constant speed. The driver then rapidly turns the
steering wheel until it hits a mechanical stop. Steering wheel stops are
set to attain a desired lateral acceleration at the test speeds. This
steer angle is held until steady-state response is established. Tests
were run in both directions (right turn/left turn) and at two speeds (60
mph and 40 mph)…"; and
"Frequency Response – Sinusoidal sweep steering tests are frequently
used to determine the linear response of vehicles. The vehicle is driven
on a long straightaway with the driver steering with slowly increasing
frequency up to approximately 3 to 4 hz followed by decreasing frequency.
The test was run at a nominal speed of 66 mph."
The “step steer” and “constant radius circle”
tests were performed both with “four good tires” and with
“the left rear tire detreaded.” The “frequency response”
tests were only performed in the “four good tires” condition.
All of the tests
were conducted in the "lightly loaded" condition (vehicle curb
weight plus the driver and necessary instrumentation), and all but the
“frequency response” tests were also conducted in the gross
vehicle weight rating (GVWR) load condition. In the June 15, 2001 meeting
with NHTSA, Dr. Guenther stated that the GVWR load condition is one in
which the total vehicle weight is equal to the vehicle’s GVWR and
front and rear axle loads are close to being the same percentage of their
respective gross axle weight ratings. Dr. Guenther also stated that the
"detreaded" tires were supplied to him by Firestone and that
it was his understanding that the tread and "outer belt" were
removed by slicing into the shoulder area of the tire and then carefully
cutting and peeling the tread and "outer belt" from the rest
of the tire.
Based on Dr. Guenther’s
analysis of the results included in the Firestone Statement, Firestone
made the following “findings” (quoted from pages 1 and 2 of
the Firestone Statement):
the Explorer models
that he tested, as designed, have a significantly lower amount of understeer
than the other SUVs he has evaluated, less than half as much as the Jeep
Cherokee and Chevrolet Blazer;
the Explorer loses much of its small margin of understeer when it is loaded
to gross vehicle weight rating; the Cherokee and the Blazer do not;
the Explorer models tested, unlike the Cherokee and the Blazer, lose all
of their understeer and become oversteer vehicles in most circumstances
following tread separation on a left rear tire, the predominant tire position
in Explorer tread separation crashes; the only exception in Dr. Guenther’s
investigation is a light load configuration in a counter-clockwise turn,
with the separated tire mounted on the left rear, a circumstance where
the vehicle retains a very small amount of understeer; and
an oversteer vehicle is extremely difficult for most drivers to control,
particularly at interstate highway speeds where it can become directionally
unstable.
The Firestone Statement contends, based on these “findings,”
“…that the Explorer is defectively designed in that it has
an inadequate margin of control, due to insufficient understeer, in the
foreseeable circumstance of tread separation during normal highway driving
in most load and turning circumstances.”
III. The
Firestone Update:
The Firestone Update
includes the results of testing for two additional models of Explorer,
a MY 1999 2-door, 4X2 Ford Explorer, and a MY 1997 4-door, four-wheel
drive (4X4) Ford Explorer under the same test conditions used in the tests
for which results were included in the Firestone Statement. It also includes
the results of two additional tests of the MY 1996 Ford Explorer tested
in the earlier program. One of these additional tests involved equipping
the MY 1996 Ford Explorer with a non-OEM “Ford recommended replacement”
Goodyear tire, and the other test condition involved conducting tests
with the “detreaded” tire being fitted to the right rear wheel
of the vehicle rather than the left rear wheel. It also includes the results
for the “step steer” and “frequency response”
tests for the vehicles for which the understeer gradient results from
the “constant radius circle” tests were included in the Firestone
Statement.
The Firestone Update
includes additional “conclusions.” One of these is of primary
importance since it effectively revises the fourth "findings"
included in the Firestone Statement. That "finding" states that
"an oversteer vehicle is extremely difficult for most drivers to
control, particularly at interstate highway speeds where it can become
directionally unstable" and is based on a statement in the "Conclusion"
section of the Firestone Statement. That is that "[o]versteer can
make a vehicle directionally instable and subject to loss of control in
the hands of most drivers." The Firestone Update includes an additional
sentence in its "Conclusion" section that states that "[a]t
or above critical speed the Explorers are not controllable by any driver."
This is a much stronger statement regarding the safety consequences of
a rear tire tread separations on Explorers.
With respect to
the results of the “step steer” and “frequency response”
tests, the Firestone Update concludes (1) that the step steer test results
show that the Explorers exhibit longer response times and high yaw velocity
gains with a “detreaded” rear tire when compared to the other
SUVs tested; and (2) that Explorers exhibit a second “peak”
in their roll angle frequency response gain characteristics that is not
found in the other SUVs that were tested and that “[g]enerally such
large frequency responses are considered undesirable.”
IV. The Firestone Supplement:
The Firestone Supplement
includes the results of tests of three additional vehicles: two MY 1993
Ford Explorers and a MY 1996 Jeep Grand Cherokee. All of these tests were
performed in the same test conditions used in the tests for which results
were previously submitted to ODI. The information submitted with this
document on these additional test vehicles did not include information
on the body style (2-door vs. 4-door), drivetrain configuration (4X2 vs.
4X4), or the tires fitted on the vehicles during the tests. Information
related to the body style, drivetrain configuration, and tires used during
testing of these vehicles is included in the Firestone Final Report.
The Firestone Supplement
does not include any additional “findings” and “conclusions,”
or, for that matter, any discussion of the test results included in it.
V. The Firestone Final Report:
The Firestone Final
Report includes the test results of all the vehicles that Dr. Guenther
tested as part of his work for Firestone, including the results of tests
of a MY 1992 Nissan Pathfinder. This document revises several of the observations
and conclusions included in the Firestone Update. Of most significance,
the conclusions drawn by Dr. Guenther in the Firestone Update as they
relate to the particular models of Explorers that exhibit "undesirable"
performance are revised. The conclusions in the Firestone Update related
to the understeer gradients, frequency response characteristics, and gain
and response times apply to the “Explorers tested (except the 1997
4 wheel drive).” However, in the Firestone Final Report, they are
applied only to the “1995 model year and later 2WD Explorers tested.”
The Final Report states that the “pre-1994 Explorers tested and
the 4WD Explorer tested appear to have understeer characteristics closer
to the Blazer, Grand Cherokee and the Cherokee.” The report further
states that, as regards the understeer characteristics of the vehicles
tested, the “1992 Pathfinder tested appears to fall between these
groups.”
VI. The Firestone Presentation:
The entire Firestone
Presentation has been placed in the docket (public file) for this proceeding.
Some of the items presented at the December 4, 2001 meeting with Firestone
and Dr. Guenther are discussed below.
A video tape included
in the presentation shows 40 mph step steer tests being conducted with
a MY 1996 Chevrolet Blazer, a MY 1996 Ford Explorer, and a MY 2001 Jeep
Cherokee, all of which are 4X2, 4-door vehicles. In each case, the vehicles
were fitted with a "detreaded" left rear tire, and tests were
performed in a right turn. The video shows that for similar steering wheel
inputs, the Blazer and Cherokee both exhibited stable responses. However,
the Explorer's response was unstable, with the vehicle spinning out of
control.
In presenting the
list of 16 crashes of Explorers equipped with non-Firestone tires that
experienced rear tire tread separations, Firestone argued that the performance
of Explorers under circumstances of a rear tire tread separation is not
dependent upon the brand of tire involved.
In discussing the
various causes of tire tread separations, Firestone stated that beyond
issues related to the design and manufacture of tires, such as those that
resulted in the recalls of Firestone ATX and Wilderness AT tires, there
are factors related to the use of tires that can result in tread separation,
and that “tread separations will continue” to occur even in
the absence of design and manufacturing faults in tires. These factors
include punctures, improper repairs, overloading, underinflation, and
impact damage.
In its statistical
analysis of FARS data, Firestone compared the rates of fatal crashes per
100,000 registered vehicle-years for Explorers to those of the other SUVs
tested by Dr. Guenther for various crash categories and for different
combinations of drivetrain and body-style.
VII. ODI’s Analysis
of the Firestone Request:
To assist those who are not familiar with
the basic concept of dynamic systems and automotive vehicle dynamics,
ODI has prepared an appendix to this document entitled, "Discussion
of Some Basic Concepts of Dynamic Systems and Vehicle Dynamics."
A. Analysis of Test Results Submitted
in Support of Firestone’s Request:
The Firestone Statement lists four “findings”
(listed above) upon which it based its conclusion that the Explorer is
“defectively designed.” These purport to be based on the results
of the tests conducted by Dr. Guenther that were included in the Firestone
Statement. ODI believes that the vehicle performance measures that resulted
from these tests are among the appropriate measures that could be used
to examine the influence of a vehicle’s control and stability characteristics
on the likelihood of a driver’s losing control of a vehicle resulting
in a crash following a rear tire tread separation. However, to evaluate
the Firestone findings that are based on these test results, it is necessary
to examine the data reported by Dr. Guenther from two perspectives. The
first is whether the reported results accurately characterize the actual
response characteristics of the vehicles tested. The second is whether
the sample of vehicles tested is adequately diverse and representative
of the SUVs being driven on U.S. highways.
1. Accuracy of the Reported Test
Results:
ODI has examined whether the tests were
objective and performed under acceptably controlled conditions, whether
the measurements were made with an acceptable level of accuracy, and whether
the results were objectively reduced and analyzed.
ODI performed this evaluation only for
the “constant radius circle” tests from which the understeer
gradients were determined. A similar evaluation for the other tests performed
by Dr. Guenther was not performed, primarily because the results of the
“step steer” and frequency response” tests were not
primary components of any of Firestone’s findings or contentions.
To evaluate the results from the “constant
radius circle” tests, ODI compared the results of Dr. Guenther’s
tests to the results of comparable tests performed by others and independently
analyzed Dr. Guenther’s raw test data.
There were two other sources of relevant
understeer gradient data. First, recent and ongoing research at NHTSA’s
Vehicle Research and Test Center (VRTC) evaluating various dynamic rollover
test procedures included vehicle characterization tests, one of which
measured understeer characteristics. VRTC tested seven SUVs, including
a MY 1998 4-door, 4X4 Ford Explorer. VRTC used a different test procedure,
a constant speed, slowly increasing steering wheel angle test, than that
used by Dr. Guenther, and VRTC did not test any vehicles with a rear tire
tread separation.
Second, Ford Motor Company (Ford) derived
understeer gradients from tests performed on a variety of SUVs, including
various model year, drivetrain, and body style Explorers, as well as a
few other light duty vehicles. For some of these makes and models, Ford
also provided understeer gradients of the vehicles with a rear tire tread
separation.
The lightly loaded load condition used
in both VRTC’s and Ford’s tests was the same as that used
in Dr. Guenther’s testing. However, the GVWR test condition used
by Dr. Guenther was different from that used by Ford and VRTC. Ford and
VRTC used a loading procedure that results in the total vehicle weight
being equal to the vehicle’s GVWR and its rear axle load being equal
to the vehicle’s rear axle’s gross axle weight rating. Due
to the more rearward center of gravity location that would result, the
GVWR load condition used by Ford and VRTC would tend to result in slightly
lower understeer gradients than the loading used by Dr. Guenther.
Table 2 includes the understeer gradients
reported by Dr. Guenther for vehicles in the lightly loaded condition
without a rear tire tread separation along with understeer gradients for
comparable vehicles with intact tires that were tested by VRTC and Ford.
Table 3A includes the understeer gradients in the lightly loaded condition
for all vehicle models for which test results are available from more
than one source and duplicates some of the results presented in Table
2. Table 3B includes the understeer gradients in the lightly loaded condition
for vehicles for which results were only available from one source.
a. Objectivity of Procedures Used
by Dr. Guenther:
Based on the information provided by Firestone
with its request and made available to ODI during the June 15, 2001 meeting,
the tests performed by Dr. Guenther used a procedure recognized by both
U.S. and international automotive engineering standards organizations
to measure a vehicle’s understeer/oversteer characteristics and
with test conditions appropriate and adequately controlled for those procedures.
b. Accuracy of Measurements Made
by Dr. Guenther:
Only limited information is available
on the instrumentation, but an examination of the raw test data from the
“original” tests performed by Dr. Guenther (i.e., those performed
prior to May 31, 2001), which were provided to ODI, suggests that the
instrumentation used was adequate to derive accurate measurements.
c. Objectivity of Data Reduction and Analysis Used by Dr. Guenther:
No precise information on the data reduction
techniques used in Dr. Guenther’s analyses to determine the understeer
gradient from each individual test run was provided to ODI. However, since
Firestone provided ODI with the raw data from the original tests performed
by Dr. Guenther and reported in the Firestone Statement, VRTC was able
to compare his results with those that result from using the data reduction
techniques normally used by VRTC for analyzing similar test data.
Using the results of VRTC's reduction
of Dr. Guenther "raw" data, ODI separately analyzed the results
of the three clockwise tests and the three counter-clockwise tests.
There were 40 “sets” of test
runs performed in Dr. Guenther’s original test program, five test
vehicles, two load conditions (lightly loaded and GVWR), two tire conditions
(“4 normal tires” and “left rear tire detreaded”),
and two steering directions (clockwise and counter-clockwise). For those
40 sets of test runs, 34 of the understeer gradient values reported by
Dr. Guenther were lower than those that resulted from VRTC’s analysis,
and 6 of the values were higher. In 10 of the 34 cases where Dr. Guenther’s
values were lower than those calculated by VRTC, the value reported by
Dr. Guenther was outside the 95% confidence limits calculated by the VRTC
analysis. In none of the 6 cases where Dr. Guenther’s values were
higher than those calculated by VRTC were the values reported by Dr. Guenther
outside the 95% confidence limits calculated by the VRTC analysis.
Table 4A includes the understeer gradient
values reported by Dr. Guenther for all five of these vehicles for the
“4 normal tires” condition for both the clockwise and counter-clockwise
tests and for both the lightly loaded and GVWR condition. Table 4A also
includes those understeer gradients determined in VRTC’s analysis
of Dr. Guenther’s test data. A review of the data in Table 4A shows
that 18 of the 20 values reported by Dr. Guenther are lower than those
determined from VRTC’s analysis, and two are higher. In six of the
18 cases where Dr. Guenther’s values were lower than those calculated
from VRTC values, the values reported by Dr. Guenther were outside the
95% confidence limits calculated by the VRTC analysis. In the two cases
where Dr. Guenther’s values were higher than that calculated from
the VRTC values, Dr. Guenther’s values were not outside the 95%
confidence limits calculated by the VRTC analysis.
Ford also analyzed the raw test data from
Dr. Guenther’s tests for the MY 1996 Chevrolet Blazer and the MY
2001 Jeep Cherokee, and provided ODI with the average of the understeer
gradient values for the clockwise and counter-clockwise directions for
the “4 normal tires” conditions. In both cases, the average
understeer gradient values resulting from Ford’s analysis were lower
than those reported by both Dr. Guenther and by VRTC, and in the case
of the Cherokee, the Ford value fell outside the 95% confidence limits
calculated by VRTC.
ODI attempted to ascertain whether the
understeer gradient values determined from any one of the techniques would
always be higher or lower than those determined from either of the other
techniques. For both of the cases where results were available from all
three sources, the results using the VRTC method were higher than Dr.
Guenther’s and Ford’s, and Dr. Guenther’s results were
higher than Ford’s. However, based on only two examples, it is not
possible to determine with any confidence that there are any systematic
differences among the three data reduction techniques.
2. Variability of Understeer Gradient
Test Results:
In determining whether the differences
between the results obtained by Dr. Guenther, Ford and VRTC have any practical
significance, it is necessary to examine them in the context of all of
the sources of variability that exist in measuring the understeer gradients
of vehicles.
First, the tests used to determine the
understeer gradients for vehicles have an inherent level of variability
from one test run to another. VRTC’s testing included six test runs
per vehicle, three in the clockwise direction and three in the counter-clockwise
direction; this same test protocol was used by Dr. Guenther. In VRTC’s
data reduction and analysis procedure, the test data for each run were
analyzed using a linear regression to produce a straight line fit of the
data, and the slope of that line represents the understeer gradient for
that test. Then, in order to quantify the test-run-to-test-run variability,
a statistical analysis of the individual understeer gradient values for
each of the six tests runs of a vehicle was performed, and the mean (average)
and the standard deviation for the six test runs was calculated. As an
example, VRTC’s tests of the MY 1998 Ford Explorer resulted in an
average understeer gradient of 2.58°(degrees)/g and a standard deviation
of 0.31°/g for the six test runs upon which the result in Tables 1
and 2 is based. Using standard statistical techniques, the 95% confidence
interval for these test results is from 2.25°/g to 2.91°/g. This
characterizes the test-run-to-test-run variability and is a measure of
the confidence with which the results of a series of tests of one vehicle
being tested at one test site conducted at one time can be judged as representing
the “true” understeer gradient of that vehicle. Dr. Guenther
only reported the average of the understeer gradient values determined
from the six individual test runs that he performed.
Some of the variability comes from sources
other than the variations in understeer gradient values that result from
the test-run-to-test-run differences and from the differing data reduction
techniques discussed above. These other sources include variations in
testing the same vehicle at different times at the same facility, variations
in testing at different test facilities with different road surfaces,
variations in test procedures (including conducting constant radius circle
tests at different radii) and variations in testing different samples
of the same vehicle model.
Ford has indicated that it has evaluated
some of these sources of variability for the "constant radius circle"
test procedure. These evaluations involved testing the same vehicle at
different times at the same test site, at different test radii at the
same test site, and at different test sites. Based on these tests, Ford
estimates these sources of variability result in a 95% confidence interval
of +/-0.25°/g. This variability should be added to the test-run-to-test-run
variability discussed above when comparing understeer gradients that were
measured in different test series.
Moreover, variation in test results between
different samples of the “same” vehicle model may be as large
as, and possibly larger than, the other sources of variability discussed
above, due to normal production tolerances and tolerance stack-up during
the manufacturing process. Based on the above discussion, ODI believes
that unless differences of at least 1°/g are found when comparing
the measured understeer gradients of different vehicle models, one cannot
confidently state that the vehicles truly have different understeer gradients.
3. Comparison of Dr. Guenther’s
Measurements of Understeer Values to Those Measured by Others:
To further evaluate whether the results
reported by Dr. Guenther accurately characterize the responses of the
vehicles tested, those results were compared to the results of tests performed
by others. This comparison is possible since, as discussed above, understeer
gradients of vehicles that are the same as, or similar to, those tested
by Dr. Guenther are available from other sources.
As noted earlier, the test procedure used
by VRTC is different from that used by Dr. Guenther and Ford. For rear-wheel
drive vehicles, understeer gradient test results using this test procedure
will tend to be higher than that of the "constant radius circle"
test procedure. ODI is not aware of any studies that have attempted to
quantify the differences in the understeer gradients measured using the
two test procedures.
Table 3A includes results from VRTC tests
of five vehicles for which understeer gradient results are also available
from Dr. Guenther and/or Ford. In three of those five cases, the understeer
gradients reported by Dr. Guenther and/or Ford are less than those from
VRTC and are also outside the 95% confidence limits derived for the VRTC
tests, and in the other two cases, the results from Dr. Guenther and/or
Ford are close to those from VRTC and are within the 95% confidence limits
derived for the VRTC tests.
The understeer gradient results from Firestone,
Ford, and VRTC shown in Tables 2 and 3A are nearly all within the +/-1°/g
range of test variability discussed above. There are only two exceptions.
One is the difference between the reported understeer gradients for the
MY 1996 4-door, 4X2 Ford Explorer results from Dr. Guenther and that of
the MY 1997 4-door, 4X2 Explorer results from Ford, which is 1.06°/g.
This difference is so close to the 1°/g variability criterion that
ODI deems it to be insignificant. The other exception is the difference
between the understeer gradients for the MY 2001 4-door, 4X4 Ford Escape
reported by VRTC, 4.21°/g, and that reported by Ford, 2.50°/g.
Despite these two instances, ODI believes that the understeer gradient
results from Dr. Guenther, Ford, and VRTC are sufficiently consistent
such that, for the purposes of analyzing the "findings" made
by Firestone, it is legitimate to use the results reported by Dr. Guenther.
B. ODI's Review of the "Findings"
and "Conclusions" Supporting the Firestone Request:
The first three of the four "findings"
presented in the Firestone Statement are based entirely on the results
of the tests that are discussed above. In examining these findings, ODI
assumed that the reported test results upon which they are based are accurate
and valid. The last of the findings presented in the Firestone Statement
is based on Dr. Guenther’s review of technical literature, some
of which is referenced in the Firestone Statement.
Additional conclusions are presented in
the Firestone Update. The first is that the step steer test results show
that the Explorers exhibit longer response times and higher yaw velocity
gains with a “detreaded” rear tire when compared to the other
SUVs tested. Since this finding is related to the issue of the control
characteristics of an oversteer vehicle, it will be addressed below in
the “Findings Related to the Controllability of an Oversteer Vehicle”
section. The second is that the Explorers tested have “…two
distinct peak ranges in the roll angle frequency response; one of them
at a low frequency similar to the other SUVs tested and a second higher
frequency response not shared by the other SUVs tested.”
Given the additional test results and
revised conclusions included in the Firestone Final Report, ODI has assumed
that Firestone now believes that these “findings” apply only
to the “1995 model year and later 2WD Explorers tested.”
Each of these “findings” is
examined below.
1. Finding that the Explorer Has
Significantly Lower Understeer than the Other SUVs Tested:
As indicated above, understeer gradients
with “4 normal tires” for several models of Ford Explorers
and other SUVs are available from tests performed by VRTC and Ford, as
well as from Dr. Guenther’s test program. The understeer gradients
that were available from all of these sources for vehicles tested in the
lightly loaded condition are included in Tables 3A and 3B. Table 3A includes
understeer gradient values for those vehicles for which test results are
available from more than one source, and Table 3B includes those for vehicles
for which results were only available from one source.
The data in Tables 3A and 3B show that,
in general, the MY 1995 and later 4X2 Ford Explorers have lower understeer
gradients than the four peer SUVs tested by Dr. Guenther. However, they
do not have significantly lower understeer gradients than many of the
other SUVs tested by VRTC and Ford. Several recent and current model year
SUVs have understeer gradients lower than or close to those of MY 1995
and later 4X2 Explorer models.
It should be noted that all of the vehicles
included in Tables 3A and 3B whose understeer gradient is lower than or
equal to 3.0°/g were introduced to the market or were redesigned for
MY 1997 or later, and all but two of the vehicles, the MY 1994 Toyota
4Runner and the MY 1993 Ford Explorer XLT, included in the tables whose
understeer gradient is lower than 4.0°/g were introduced to the market
or were redesigned for MY 1996 or later. On the other hand, only three
of the vehicle models included in the tables whose understeer gradient
is greater than 4.0°/g were introduced or redesigned for MY 1996 or
later. This seems to indicate a significant, industry-wide trend toward
lower understeer gradients in more modern SUVs. This trend is also seen
in the Ford Explorers, where the average understeer gradient for the MY
1995 and later 4-door, 4X2 Ford Explorers is about 2.3°/g and the
understeer gradient for the MY 1993 4-door, 4X2 Ford Explorer, a model
introduced in MY 1991, is 4.1°/g. This trend has positive consequences
on vehicle response characteristics in that reduced levels of understeer
gradient will tend to increase the level of vehicle control available
to the driver. As long as the other aspects of the vehicle's response
characteristics provide an adequate level of damping and stability (characteristics
that are not directly related to or adequately quantified by the understeer
gradient), drivers will be able to safely operate a vehicle with such
lower levels of understeer gradient over the same range of driving situations
as vehicles with higher levels of understeer gradient.
In summary, although the first finding
made in the Firestone Statement is accurate when the comparison is limited
to the small sample of vehicles examined by Dr. Guenther, data reflecting
tests of a more representative sample of SUVs show that the understeer
gradients of these Explorer models are higher than or similar to those
of several other contemporaneous SUVs.
2. Finding that the Explorer Loses
Much of its Understeer when Loaded to GVWR:
When an SUV is loaded to its GVWR, its
center of gravity generally moves rearward; in fact, this is true of most
light trucks and vans (see Measured Vehicle Inertial Parameters-NHTSA’s
Data Through November 1998, Heydinger et al, SAE 1999-01-1336, page 5).
Such rearward movement of a vehicle’s center of gravity would inherently
tend to result in a reduction in the vehicle’s understeer gradient
to some extent (see Fundamentals of Vehicle Dynamics, Thomas D. Gillespie,
SAE R-114, page 226).
The understeer gradient results for the
two load conditions of the three MY 1995 and later 4X2 Explorer configurations
and the two other SUVs tested prior to the submission of Firestone’s
request in the “4 normal tires” condition by Dr. Guenther
and included in the Firestone Statement are shown in Table 4A along with
the absolute and relative (percentage) changes from the lightly loaded
condition to the GVWR condition. Also included in Table 4A are the understeer
gradient results of VRTC’s analysis of Dr. Guenther’s data
for the “4 normal tires” tests that were discussed earlier.
First, examining Dr. Guenther’s
results alone, only one of the six comparisons for the Explorer configurations
has a difference greater than the typical test-series-to-test-series variability
of +/-0.25°/g discussed earlier. That is the MY 1996 Ford Explorer
with Firestone tires tested in the clockwise direction. If the typical
industry practice of averaging the results for clockwise and counter-clockwise
tests were applied to this data, the difference of 0.40°/g between
the lightly loaded and GVWR conditions would also fall within typical
test-series-to-test-series variability. The largest absolute and relative
differences between the lightly loaded and GVWR conditions is seen for
the MY 1996 Chevrolet Blazer, where its understeer gradient changes by
1.22°/g for both the clockwise and counter-clockwise tests, but its
understeer gradient decreases in the clockwise direction and increases
in the counter-clockwise direction.
Using the VRTC’s reduction of Dr.
Guenther’s data, in all but one of the lightly loaded-to-GVWR comparisons,
the differences are smaller than those reported by Dr. Guenther. In no
case is the difference significant based on the 95% confidence limits
that were calculated. Table 5 presents an analysis of understeer gradient
data from tests performed by Ford and by Dr. Guenther. These data also
do not indicate a consistent or significant reduction of understeer with
increasing load for various MY 1995 and later 4-door, 4X2 Explorer models.
In fact, for the seven 1996 to 2000 model year Explorers included in Table
5, the average change in understeer gradient from lightly loaded to GVWR
condition is a decrease of 0.146°/g, with the 95% confidence interval
being from –0.45 to +0.16°/g. If the analysis is expanded to
include all MY 1995 and later 4X2 Explorers for which understeer gradient
results are available, the average change in understeer gradient from
lightly loaded to GVWR condition is a decrease of only 0.077°/g, with
the 95% confidence interval being from –0.33 to +0.17°/g.
The understeer gradient results for the
two load conditions of five of the Explorer configurations and of the
two peer SUVs tested in the “4 normal tires” condition by
Dr. Guenther and included in the Firestone Update, Firestone Supplement,
and Firestone Final Report are shown in Table 4B along with the absolute
and relative (percentage) changes from the lightly loaded condition to
the GVWR condition. The results for the “right rear” MY 1996
Explorer are not included in the table, since these are the same for the
MY 1996 Explorer tested with OEM Firestone tires included in Table 4A.
Only two of the ten comparisons for the
Explorer configurations have a difference greater than the typical test-series-to-test-series
variability of +/-0.25°/g discussed earlier. These are the two MY
1993 Ford Explorers tested in the clockwise direction, and the differences
are only very slightly greater than 0.50, specifically 0.51 and 0.53.
Also, if the typical industry practice of averaging the results for clockwise
and counter-clockwise tests were applied to this data, the differences
between the lightly loaded and GVWR conditions would fall within typical
test-series-to-test-series variability.
Table 6 presents the results of an analysis
of the differences in understeer gradient between lightly loaded and GVWR
conditions of 24 non-Explorer SUVs. For these vehicles, the average change
in the understeer gradient from lightly loaded to GVWR condition is –0.347
°/g, with the 95% confidence interval being from –0.59 to +0.11°/g.
This data does not support Firestone's
assertion that the MY 1995 and later 4X2 Explorer models lose much of
their understeer when loaded to GVWR. For that matter, the data indicates
that the changes in understeer gradient for the Explorer models tested
by Dr. Guenther are not as great as those for the other three SUVs tested
by Dr. Guenther, as well as those of many other SUVs tested by Ford.
3. Finding that, Unlike the Other
SUVs Tested, Explorers Become Oversteer Following a Rear Tire Tread Separation:
The understeer gradient results in the
“detreaded” tire condition for the two load conditions and
two turn directions for the nine Explorer configurations and the four
peer SUVs tested by Dr. Guenther are included in Table 7. Seven of the
nine Explorer configurations exhibited an understeer gradient less than
zero, i.e., oversteer, in at least one load/turn direction condition,
and all of these were MY 1996 or later Explorers, which are equipped with
a short-long-arm (SLA) front suspension. The understeer gradient for these
vehicles in the “detreaded” tire condition ranged from –0.04°/g
to –2.91°/g, i.e., oversteer. The understeer gradient exhibited
by the MY 1997 4-door, 4X4 Explorer in the “left rear tire detreaded,”
GVWR, clockwise turn direction test condition, 0.04°/g, is so close
to zero that the vehicle could be characterized as a neutral steer vehicle.
One of the peer SUVs, the MY 1992 Nissan Pathfinder, also exhibited an
understeer gradient of less than zero in both turn directions in the GVWR
load condition. Neither of the two MY 1993 Ford Explorers, which are equipped
with a “twin I-beam” front suspension, were found to have
understeer gradients of less than zero in any of the “detreaded”
tire test conditions. The lowest understeer gradient measured for either
of these Explorers in the "detreaded" tire test condition, 0.20°/g,
was exhibited by the MY 1993 Explorer XLT. This value is close to or greater
than the lowest understeer gradients exhibited by two of the peer SUVs,
the MY 1996 Chevrolet Blazer and the MY 1996 Jeep Grand Cherokee, when
they were tested in the “detreaded” tire condition.
Ford provided results of understeer gradient
tests for 25 SUVs, 13 with 4X4 drivetrains and 12 with 4X2 drivetrains,
two front-wheel drive compact passenger cars, and two front-wheel drive
minivans that included data for tests conducted in the GVWR condition
both with all intact tires and with a tread separation of the tire that
was on the outside rear wheel during the tests (i.e., the left rear tire
for a clockwise, right, turn or the right rear tire for a counter-clockwise,
left, turn). Based on this data, most of the SUVs (20 of the 25 in the
data), as well as one of the passenger cars and both minivans, exhibit
linear range oversteer characteristics following a rear tire tread separation.
Table 8 shows the maximum, minimum, mean, and median values of the understeer
gradient with intact tires (“tread on”) and with a tread separation
on the outside rear tire (“tread off”) and the change in the
understeer gradient from the “tread on” to the “tread
off” condition for all 29 vehicles for which results were provided
by Ford and for the 13 vehicles for which results were provided by Firestone.
For the maximum and minimum values shown in Table 8, the model year of
the particular vehicle is shown in the table. All of the vehicles that
represent these maximum and minimum values are 4-door SUVs with wheelbases
between 104 and 116 inches and curb weights between 3,500 and 4,700 lbs.
The 15 Explorer models represented in the data include six of the seven
vehicles included in Table 5 (one of the MY 1997 “Ford” Explorers
included in Table 5 was not tested with a “detreaded” rear
tire), four additional Explorer vehicles tested by Ford (a MY 1993 4-door
4X2, a MY 1997 2-door 4X2, and a MY 1996 and a MY 2000 4-door 4X4), and
five additional Explorer vehicles tested by Dr. Guenther (a MY 1993 and
a MY 1996 (tested with a right rear “detreaded” tire) 4-door
4x2, a MY 1999 2-door 4X2, and a MY 1993 and a MY 1997 4-door 4X4). The
average “tread off” understeer gradient for these Explorers
is –1.38°/g. Of the sixteen non-Explorer SUVs that became oversteer
in these circumstances, ten are more oversteer than the average for the
Explorers and four had more oversteer than the Explorer that exhibited
the most oversteer.
These data indicate that Explorers are
not unique among SUVs in exhibiting linear range oversteer characteristics
following a rear tire tread separation. The data also show that linear
range oversteer following a rear tire tread separation is not unique to
SUVs. (Although the data in the Firestone Final Report indicates that
the MY 1992 Nissan Pathfinder exhibits linear range oversteer in the GVWR
condition with a rear tire tread separation, Dr. Guenther asserts that
the results of the “constant radius circle” tests show that
“unlike the other SUVs tested, the 2WD 1995 model year and later
Explorers lose their margin of understeer when they experience a tread
separation” (emphasis added). As such, this finding is not supported
even by the results of the tests conducted by Dr. Guenther.)
4. Findings Related to the Controllability
of an Oversteer Vehicle:
The Firestone Statement asserts that “[a]n
oversteer vehicle is not safe at highway speeds in the hands of the average
driver.” It also states that an oversteer vehicle is difficult to
drive because the driver “has to deal with…a slow response
time and the following large gain.” In particular, Firestone contends
that an oversteer vehicle is extremely difficult for most drivers to control,
particularly at high, “interstate highway” speeds where it
can become directionally unstable. Firestone revised this contention in
its August 22, 2001 Update, stating that “[a]t or above critical
speed the Explorers are not controllable by any driver.”
ODI is not aware of any published results
of tests of passenger cars or light trucks that exhibit linear range oversteer
characteristics that result from operating factors such as loading or
component degradation/failure (excluding tire failures). However, based
on information provided by vehicle dynamics experts, it is likely that
station wagons and light trucks with heavy loads in the rear of their
cargo areas and light duty vehicles towing relatively large trailers (particularly,
if the trailers are improperly loaded) can exhibit linear range oversteer,
yet such vehicles are driven by average drivers without crashing. Drivers
may complain about the difficulty of driving such vehicles for extended
periods and can become fatigued driving them, since constant, minor steering
corrections are necessary to drive such a vehicle at speeds approaching,
at, or above its critical speed, but this is not relevant to the Firestone
request, since it is unlikely that drivers will continue to drive vehicles
that have experienced a rear tire tread separation for any longer than
necessary to get to the side of the road and change the tire.
The first specific statements made by
Firestone about the safety consequences of the changes in transient response
times and steady-state gains appear in the Firestone Update, and are repeated
in the Firestone Final Report. The Firestone Update states that the step
steer test results show that the Explorers exhibit longer response times
and higher yaw velocity gains with a “detreaded” rear tire
when compared to the other SUVs tested by Dr. Guenther. Given the understeer
gradients of these vehicles, this is not at all surprising since, as discussed
in the Appendix, vehicles with lower levels of understeer will inherently
exhibit longer transient response times and higher steady-state gain characteristics
at highway speeds than similar type/class vehicles with higher understeer
gradients. However, such response characteristics are not unique to Explorers
and would be exhibited by any light duty vehicle whose understeer gradient
is significantly reduced due to vehicle loading or component degradation/failure.
Although not mentioned by Firestone, ODI
notes that, in addition to the reduction in the vehicle's understeer gradient,
a tread separation has an additional effect on a vehicle’s control
and stability characteristics. That is the result of the “detreaded”
tire’s significant loss of traction, as well as a dramatic loss
of cornering stiffness. This results in not only a significant reduction
in the vehicle’s understeer gradient, but also a substantial reduction
in the lateral acceleration capability of the vehicle. Test data available
from both VRTC and Ford show that most SUVs are normally capable of controllably
achieving lateral acceleration levels of about 0.7 g, and their linear
range would typically extend to about 0.4 g on dry, paved road surfaces.
However, information provided by Ford indicates that following a rear
tire tread separation, some SUVs exhibit limits of lateral acceleration
of only 0.35 g, and linear ranges of lateral acceleration response as
low as 0.2 g. This effect likely contributed to the loss of directional
stability of the Explorer shown in the videotape presented at the December
4, 2001 meeting. Based on the 40 mph linear range yaw response gain measured
by Dr. Guenther for that vehicle, the 21° steering wheel input used
in the 40 mph step steer test of the Explorer would have resulted in a
steady turn with a lateral acceleration of 0.33 g. Based on the similarly
measured gains for the other two SUVs shown in the video, the Blazer's
and the Cherokee's responses to the 24° steering wheel inputs would
be steady turns with lateral accelerations of about 0.13 g and about 0.2
g, respectively. These lateral accelerations were well within the linear
range maneuvering capabilities of each of these vehicles with “good
tires,” and the lateral accelerations for the Blazer and Cherokee
remained in the linear range even with "detreaded" rear tires.
However, the lateral acceleration achieved by the Explorer reached the
non-linear range, causing the vehicle's response gains to increase very
rapidly, leading to eventual loss of directional stability. (During this
demonstration the driver was instructed not to make any effort to change
the steering wheel angle in an attempt to correct any undesirable vehicle
response. ODI believes that such a scenario is not representative of real
world driver behavior.)
The above discussion is not meant to minimize
the fact that the driver of a vehicle that has become oversteer as a result
of experiencing a rear tire tread separation is confronted with substantially
different vehicle response and maneuverability characteristics than those
with which the driver is familiar. Particularly if the vehicle is being
driven above its critical speed, this could lead to loss of control, if
the driver does not make appropriate corrections to the steering input
in reaction to the vehicle's response.
With regard to a driver’s ability
to control such a vehicle, a crucial factor is the control tasks with
which the driver is confronted. When the driver of a vehicle that has
become oversteer after experiencing a rear tire tread separation is confronted
with other than simple, driving tasks (i.e., straight line driving, and
possibly ”slow” lane changes and negotiating typical curves
of modern high-speed highways), the vehicle can be very difficult to control,
especially when the vehicle is being driven above its critical speed.
In such situations, there may be a loss of directional stability, e.g.,
a “spin-out” or a loss of path control, resulting in an impact
with another vehicle or roadside object. Whether or not a given driver
is able to adequately control such a vehicle depends upon the vehicle’s
speed, the precise traffic/driving situation, the vehicle’s response
characteristics, which are affected by the level of oversteer of the vehicle,
and the driver's experience, knowledge, and alertness.
Ford has made available to ODI videotapes
of tests that it conducted involving various SUVs, including various models
of Ford Explorers with a rear tire tread separation, being driven at high
speeds, in some cases significantly higher than their critical speeds.
These tests indicate that moderate lane changes and relatively hard braking
maneuvers can be performed by alert, experienced drivers (not necessarily
professional test drivers) without loss of control or stability.
Firestone claims that a driver may lose
control when “he has to deal with the unfamiliar and unpredictable
oversteer handling through a steering input/vehicle response characterized
by a slow response and a following large gain.” This is correct.
However, such loss of control situations also can occur with vehicles
that are not oversteer. For example, Dr. Guenther's test results for the
two MY 1993 Explorers suggest that MY 1991 through 1994 Explorers that
are equipped with a “twin-I beam” front suspension do not
become oversteer following a rear tire tread separation. However, a substantial
number of these vehicles that were equipped with Firestone ATX or Wilderness
AT tires were involved in loss of control crashes following rear tire
tread separations.
Based on the above, ODI believes that
the value of a vehicle's linear range understeer gradient, and in particular,
whether it is positive or negative (i.e., understeer or oversteer, respectively)
does not provide a sufficient measure for evaluating the risk of a loss
of control in the event of a rear tire tread separation.
5. Discussion of Other Issues
Raised in the Firestone Presentation:
Firestone provided a list of 16 crashes
of Explorers equipped with non-Firestone tires that experienced rear tire
tread separations. Seven of these crashes involved pre-MY 1995 Explorers,
vehicles that are not the focus of the Firestone request, and 10 of the
17 fatalities and 9 of the 35 injuries included in that list involved
these vehicles. Another of the crashes included in the list involved an
Explorer with an unknown model year. As stated earlier, ODI agrees that
a tire tread separation can lead to a crash. That is why it is important
to minimize the number of such tire failures. However, these incidents
do not indicate that an Explorer is more likely to experience a crash
under this scenario than other SUVs.
ODI believes that Firestone's analysis
of the FARS data is of limited, if any, use in understanding the issues
related to Firestone’s request. In past analyses of crash avoidance
issues, NHTSA has found that analyses based on FARS data are usually of
limited value. The likelihood of a fatality in a crash is very dependent
upon the vehicle's crashworthiness characteristics and safety belt use.
Often, analyses of fatal crashes yield results that conflict with results
of analyses based on more general state crash data files that include
information on crashes of all levels of severity.
The Firestone Presentation included a
discussion of various causes of tire tread separations, other than those
related to design and manufacturing faults in the tires. NHTSA is in the
process of several initiatives that will reduce the incidence of such
tread separations, in response to mandates established by the Transportation
Recall Enhancement, Accountability, and Documentation (TREAD) Act. These
include an upgrade of the Federal motor vehicle safety standards applicable
to tires, a requirement for tire pressure monitoring systems on vehicles,
and improved labeling of tires. NHTSA is also taking steps to encourage
the proper maintenance and inspection of tires and proper vehicle loading.
Nevertheless, ODI agrees that because various factors related to in-service
tire use and abuse can lead to tread separations, tread separations will
continue to occur. However, this has no direct relevance to the issue
of whether the Explorer contains a safety-related defect.
VI. Field Experience of Ford Explorers
and Other SUVs that Experienced a Tread Separation:
An examination of the Firestone claims
database provided to ODI during its investigation of Firestone ATX and
Wilderness tires, EA00-023, supports the conclusion that Explorers do
not respond in a significantly different manner than other SUVs after
experiencing a tire tread separation. Table 9 (Table 7 from the “Engineering
Analysis Report and Initial Decision” for EA00-023) shows that the
number of crashes per 100 “tread separation” claims for Explorer
vehicles is similar to that of other compact SUVs and all other SUVs.
The ODI consumer complaint database for the tires that were the subject
of that investigation, summarized in Table 10 (Table 8 from the “Engineering
Analysis Report and Initial Decision” for EA00-023), also does not
indicate a significant difference in the likelihood of a crash following
a tread separation between Explorer vehicles and other compact SUVs.
VII. Conclusions:
The foregoing analysis should not be construed
as suggesting that tread separations will not lead to losses of control
and crashes, particularly in SUVs. The many crashes following tread separations
of tires on these vehicles that are documented in the Firestone claims
database and that have been reported to ODI by consumers and others demonstrate
that such a tire failure can lead to loss of control, particularly when
it is a rear tire that fails and the vehicle is being driven at high speed.
However, the fact that a vehicle exhibits linear range oversteer characteristics
following a rear tire tread separation does not, in itself, indicate that
the vehicle contains a safety-related defect. Moreover, the data available
to ODI does not indicate that Explorers in general, or even MY 1995 and
later 4X2 Explorers in particular, are more likely to exhibit linear range
oversteer characteristics following a rear tire tread separation than
many of their peers.
For the foregoing reasons, Firestone’s
request for a defect investigation is denied. Appendix - Discussion of
Some Basic Concepts of Dynamic Systems and Vehicle Dynamics:
This Appendix is intended to allow those
who are unfamiliar with the basic concepts of dynamic systems in general,
and vehicle dynamics in particular, to better understand the issues raised
by Firestone's request for a defect investigation and ODI's analysis of
that request. It provides a discussion of some basic concepts with respect
to the responses of dynamic systems, including “linear range”
versus “non-linear range” dynamics, “steady-state”
versus “transient” response, “steady-state” response
gains and “transient” response times, and “open loop”
versus “closed loop” control of dynamic systems. It also includes
a discussion of some basic concepts of automotive vehicle dynamics, including
understeer gradient, the steady-state and transient stability and control
characteristics of understeer and oversteer vehicles, and the concept
of “critical speed” for an oversteer vehicle.
The concepts of “linear range”
versus “non-linear range” dynamics, “steady-state”
versus “transient” response, and “open loop” versus
“closed loop” control apply to the analyses of any dynamic
system. However, in order to simplify the following discussion, the examples
used to describe general system dynamics concepts are derived from automotive
vehicle dynamics.
A. Basic Concepts of Dynamic Systems
and their Counterparts in Vehicle Dynamics:
1. “Linear range”
versus “Non-Linear range” Dynamics:
Linear range refers to an operating range
of a dynamic system over which, in response to incremental changes in
an input to the system, the changes in the responses (motions) of the
system are proportional, with a constant ratio, to each successive input
increment. That is, it is a range of system response over which the ratio
of the amount of change in the response of the system to the amount of
change in an input to the system is equal to a constant over the entire
range.
In the case of an automobile, the linear
range usually refers to that range of vehicle operation over which incremental
increases in the steering wheel angle will result in incremental increases
in the vehicle’s lateral acceleration, such that the ratio of each
successive response increment to its corresponding input increment is
constant over the entire range of the increments. As an example, at a
given speed, assume that a vehicle responds with a lateral acceleration
of 0.2g to a 20° steering wheel input, and at that same speed, its
response to a 40° steering wheel input is a lateral acceleration of
0.4g. In both cases, the vehicle’s response to a 20° steering
increment is a 0.2g lateral acceleration increment. Since the ratio of
the response, lateral acceleration, to the input, steering wheel angle,
is 0.01g/° for the two successive increments of steering wheel angle,
the vehicle’s response up to at least 0.4g would be “linear.”
If, in response to a 50° steering wheel input, the vehicle’s
response is 0.45g, the vehicle’s response would no longer be linear,
since the ratio of the incremental response to the incremental input,
0.005g/°, has changed from that of the previous increment, 0.01g/°.
The level of lateral acceleration at which this incremental response-to-input
ratio is no longer constant is the upper bound of a vehicle’s “linear
range.” All levels of lateral acceleration above that level would
be considered to be in the “non-linear range.”
In reality, the response of a vehicle
is never perfectly linear. As a practical matter, a vehicle's response
is measured over a range of inputs and then a linear fit to the measured
data is generated. Also, while various testing techniques have been standardized,
the results always vary from one test to another, due to a variety of
factors, some of which are discussed in the main document.
2. “Steady-State”
versus “Transient” Response:
Steady-state response refers to an operating
condition of a dynamic system characterized by constant values of its
dynamic response variables (the measures of the system’s various
responses, usually in terms of measures of motion and/or position).
In the case of an automobile, steady-state
response refers to an operating condition of a vehicle in which its motion
is characterized by constant values of the vehicle’s various dynamic
response variables, forward velocity (or longitudinal acceleration, i.e.,
braking or accelerating), lateral acceleration, radius of turn, yaw velocity
(yaw rate), sideslip (yaw) angle, and roll angle. The sideslip angle of
a vehicle is the difference between the direction in which the vehicle
is “pointed” (the orientation of the vehicle’s longitudinal
centerline), and the direction in which the vehicle's center of gravity
is moving.
The steady-state response of a vehicle
to zero steering wheel angle, constant throttle position, and no brake
pedal application on a flat road surface (with no external disturbances,
such as a wind) would be straight line driving at a constant speed. This
“steady-state” condition would be characterized by constant
velocity (zero longitudinal acceleration), zero lateral acceleration,
an infinite radius of turn, zero yaw rate, zero sideslip angle, and zero
roll angle. If a vehicle is being operated in this condition and then
the steering wheel is turned a predetermined number of degrees and held
at that position, the vehicle will within a period of several seconds
achieves a “new” steady-state condition, characterized by
constant, or nearly constant, values of the motion variables above. This
new steady-state condition would be characterized by constant velocity,
a non-zero, finite level of lateral acceleration, a finite radius of turn,
a non-zero, finite level of yaw rate, a non-zero, finite sideslip angle,
and a non-zero, finite roll angle. When a vehicle is operating at a steady-state
condition, it is said to be "trimmed."
The period of time between the two steady-state
conditions discussed above constitutes the vehicle's transient response
and is characterized by values of forward velocity, lateral acceleration,
radius of turn, yaw rate, sideslip angle, and roll angle that vary with
time.
3. “Steady-State”
Response Gains and “Transient” Response Times:
One basic category of measures of the
steady-state response of dynamic systems is steady-state response gains.
This type of measure can be used to characterize the response of any of
a system’s dynamic response variables. It is a measure of the level
of steady-state response (output) of any response variable to an input
to the system and is expressed as a ratio of the level of that response
to the level of the input to the system.
In the case of an automobile, the most
common input variable used to characterize a vehicle’s steady-state
response gains is steering wheel angle. As such, these gains are expressed
as a response level per unit of steering wheel angle (SWA) expressed in
degrees (°). An example is the yaw velocity response gain, which measured
in degrees of yaw angle per second (°/sec) per degree of steering
wheel angle (SWA°, or °), or more simply, °/sec/°. This
is one of the response gains that figures prominently in the discussion
in the Firestone Update.
Another basic category of measures of
the transient response of dynamic systems is response time. This type
of measure can also be used to characterize the response of any of a system’s
dynamic response variables. It is a measure of how quickly a dynamic system
responds to an input and achieves a new steady-state condition. Not all
of a system’s dynamic response variables respond at the same rate,
and a system’s various response variables may reach new steady-state
values at different times.
For automobiles, response times are measured
during transient response testing, such as the “step steer”
tests conducted by Dr. Guenther, and are very basic measures of an automobile’s
transient response characteristics. They are measures of how quickly the
vehicle is able to respond to very rapid steering input. If the input
were a true “step input” (taking zero time) as described above,
the response time would be measured from time, t, equal zero. Since this
is not possible, the usual convention in automotive vehicle dynamics is
to turn the steering wheel as rapidly as possible and to measure the response
time from the time that the steering wheel angle reaches 50% of its final
value (this reference time is called t0). The response time of an automobile
is characterized by the time between that t0 and the time at which the
vehicle response first reaches a given percentage of its steady-state
response. The percentage most often used is 90%; however, the 50% response
time is sometimes used. The results of Dr. Guenther’s “step
steer” tests included both the 90% and 50% response times.
4. Distinction between Stability
and Control
In the broadest sense, the term “stability”
is used to describe the response of a trimmed system to disturbances external
to the system. The term “control” is used to describe the
response of a system to changes within the system that attempt to disturb
it from the trimmed condition. The following descriptions are taken from
Chapter 5 of Race Car Vehicle Dynamics by Milliken and Milliken, SAE R-146,
1995:
Basically, the term "static stability"
refers to the tendency of a system to return to a previously established
equilibrium when disturbed. One cannot talk about stability without first
having established the equilibrium or initial trim. The initial trim can
be defined by the control position, a steady-state attitude or acceleration.
In the case of the automobile the basic
directional control is provided through the steering wheel:
Static directional control is the control
moment on the vehicle as a result of steering the front wheels (emphasis
in original).
The terms stability and control are often
confused or used interchangeably, but it is essential to make the distinction
between these concepts when discussing vehicle dynamics.
5. “Open Loop” versus
“Closed Loop” Control:
The concepts of “open loop”
and “closed loop” control relate to whether the inputs to
a system are adjusted in response to the outputs (responses) to achieve
a specific desired outcome. If such an adjustment to the input to a system
that is dependent upon the output of a system exists, the system is being
operated with “closed loop” control. The output “signal”
that is used to adjust the input is known as “feedback,” and
“connection” from the output back to the input is the “control
loop.” “Open loop” control refers to the situation in
which the control loop is not “closed” and there is no “feedback,”
i.e., the control loop is “open.”
In the case of an automobile, “open
loop” refers to a situation where a predetermined input or series
of inputs are made to the vehicle, such as moving the steering wheel to
a new fixed position or moving the steering wheel in a particular manner,
such as a sine wave pattern, and the vehicle is allowed to respond without
the intervention of a driver who “closes” the control “loop”
by altering the input to achieve a predetermined vehicle output, such
as following a particular path like a lane change maneuver. A vehicle
being operated by a driver on the highway such that the vehicle will negotiate
the path desired by the driver is being operated in a “closed loop”
situation, i.e., the driver’s inputs to the system (the vehicle)
are being adjusted in response to the output (the vehicle’s motion)
to achieve a desired outcome (follow a particular path).
A special case of open loop control is
“fixed control.” As its name implies, “fixed control”
is a situation where a single “fixed” input is made to a system
in order to determine the system’s response. In the case of an automobile,
the “step steer” tests discussed earlier are an example of
a fixed control situation.
“Loss of control” can occur
for a variety of reasons. For example, in the open loop case, loss of
control occurs when a change in the steering wheel angle no longer results
in a change to the control moment—the heading and path of the vehicle
can no longer be adjusted by changes in the steering wheel position. In
the closed loop case, loss of control may be due, for example, to a change
in the vehicle's response characteristics that is beyond that to which
the driver can adapt.
B. Additional Vehicle Dynamics
Concepts:
1. Understeer Gradient:
The simplest means to describe the concept
of understeer gradient is to examine one of the tests used to measure
it, the “constant radius circle” test. In this test, a vehicle
is driven in a circular path of constant radius starting at a low speed
and slowly accelerating, usually to a speed at which the vehicle is no
longer controllable. As the vehicle’s speed increases while it negotiates
the constant radius circle, the vehicle’s lateral acceleration increases
with the square of the speed. During the test, the vehicle’s steering
wheel angle and lateral acceleration are measured (the latter can either
be measured directly or it can be calculated from the vehicle’s
speed and the known radius of the circle). When the lateral acceleration
is plotted versus the steering wheel angle, the slope (gradient) of the
line of that plot is known as the “steering wheel angle gradient.”
Dividing the value of this slope by the vehicle’s overall steering
ratio, which is the number of degrees of steering wheel angle necessary
to change the vehicle’s front wheel angle by one degree, yields
the “reference steer angle gradient.” In the special case
of the “constant radius circle” test, this “reference
steer angle gradient” is the vehicle’s understeer gradient.
The general terms, "understeer"
and "oversteer," are often used to describe a vehicle’s
control and stability characteristics. A vehicle is said to be “understeer”
if its understeer gradient is positive and is said to be “oversteer”
if its understeer gradient is negative. The distinction between these
terms is that "understeer gradient" is the metric used to quantify
this vehicle performance characteristic and "understeer" and
"oversteer" denote a category or type of vehicle performance
characteristic. A positive "understeer gradient" characterizes
an "understeer" vehicle, and a negative "understeer gradient"
characterizes an "oversteer" vehicle.
2. Steady-State and Transient
Response Characteristics of Understeer and Oversteer Vehicles:
Even within the linear range, a vehicle’s
understeer gradient alone does not provide adequate information to provide
a complete description of the control and stability characteristics of
a vehicle as it relates to the ability of the average driver to safely
control the vehicle. However, a reduction of a vehicle’s understeer
gradient sufficient for the vehicle to approach a neutral steer condition,
or in a more extreme case, for the vehicle to become oversteer, as a result
of the degradation or failure of a vehicle component (such as a tire tread
separation) could result in substantial changes in a vehicle’s control
and stability characteristics; e.g., increased transient response times
in the vehicle’s transient responses, and higher sensitivity of
steady-state response gains to vehicle speed (which could result in higher
than desirable gains at typical highway speeds). Under some circumstances,
the vehicle’s response to otherwise reasonable and appropriate steering
and/or braking inputs could result in loss of vehicle control and/or stability.
The vehicle responses for which these
changes in response times and response gains are most likely to affect
the ability of the average driver to safely control the vehicle are its
primary control responses; i.e., lateral acceleration and yaw velocity.
Of secondary influence, and much less importance, are a vehicle’s
roll, pitch and ride responses, which are only coupled responses that
result from the fact that automobiles have a “sprung” suspension
system in order to achieve a comfortable ride over irregular road surfaces.
Factors other than understeer gradient
affect a driver's ability to safely control a vehicle. For example, the
overall steering ratio is a basic vehicle characteristic measure that
has a significant effect on the vehicle control characteristics sensed
by a driver, since the vehicle’s steady-state response gains are
proportional to that ratio. Simply changing the overall steering ratio
of a vehicle will change those gains without affecting the understeer
gradient.
The following discusses the effects on
driver/vehicle performance of transient response times and the speed sensitivity
of steady-state response gains.
Transient response times that are substantially
longer than usual are considered undesirable, since it is more difficult
for the average driver to adequately predict a vehicle’s response
when the response times are relatively long. The ability of a driver to
anticipate and adjust for the delay in a vehicle’s response that
is characterized by its transient response time is referred to as “lead
equalization.” Although there are no specific guidelines for what
levels of transient response times would be best for normal, safe vehicle
operation, response times much greater than 0.5 seconds are usually considered
less desirable.
Relatively high sensitivity of a vehicle’s
steady-state response gains to changes in vehicle speed is considered
undesirable, since this would require drivers to use significantly different
steering inputs at different speeds in order to achieve the same vehicle
response. However, for the sensitivity of a typical light duty vehicle’s
steady-state response gains to changes in vehicle speed to be great enough
to result in undesirably high gain characteristics at typical highway
speeds, a vehicle’s understeer gradient would need to approach 1°/g.
Such a vehicle with an understeer gradient of 1°/g would exhibit an
increase in its yaw rate gain (the vehicle steady-state response gain
that is emphasized in the Firestone Update) of about one-third when a
vehicle’s speed increases from 40 mph to 80 mph. For comparison,
yaw rate gain of a vehicle with an understeer gradient of 2°/g would
increase by about 10% when the speed increases from 40 mph to 80 mph,
and the gain for a vehicle with an understeer gradient of 3°/g would
decrease only slightly for the same change in speed.
As discussed earlier, for typical vehicles
whose understeer gradient is reduced as a result of the degradation or
failure of a vehicle component (such as a tire tread separation), the
reduced values of understeer gradient would result in higher sensitivity
of steady-state response gains to vehicle speed and increased transient
response times of the vehicle’s transient responses. For such vehicles
whose understeer gradient is reduced to zero, i.e., neutral steer vehicles,
the vehicle’s steady-state response gains would increase along with
increases in the vehicle’s speed, and the transient response times
would approach one second or more.
In a situation where a vehicle understeer
gradient is reduced to the point that it becomes negative, i.e., it becomes
an oversteer vehicle, the vehicle’s steady-state response gains
would increase at a rate greater than increases in the vehicle’s
speed and the transient response times would usually exceed one second.
For such an oversteer vehicle being driven
at or above its “critical speed,” the vehicle’s steady-state
response gains would become so sensitive to vehicle speed that the gain
would become theoretically infinite, and the vehicle’s transient
response times would also become theoretically infinite. The concept of
“critical speed” is discussed in detail below.
3. "Critical Speed"
for an Oversteer Vehicle:
One approach to explaining the implications
on vehicle control and stability of an oversteer vehicle is to examine
the moments acting on a vehicle in a “steady-state” turn,
i.e., at constant radius and constant speed. These can be represented
using what is known as “derivative” notation. In such notation,
the forces and moments acting on a vehicle are represented by multiplying
each appropriate “derivative” coefficient by its corresponding
basic state variable. There are three basic state variables that can together
describe the operating conditions of a vehicle that affect the forces
and moments acting on the vehicle. They are the road wheel steer angle,
d; the yaw velocity, r; and the sideslip angle, b. Their corresponding
“derivatives” are the control moment derivative, Nd, the yaw
damping derivative, Nr, and the static directional stability derivative,
Nb.
Using this notation, the moments
acting on a vehicle are:
Control Moment - dNd
Yaw Damping Moment - rNr
Static Directional Stability Moment - bNb
These moments and their direction of action
for vehicles negotiating a steady-state turn are shown in Figure A-1 for
the cases of understeer, neutral steer and oversteer vehicles.
In the case of an “understeer”
vehicle (i.e., a vehicle with a positive understeer gradient), all three
of these moments act on the vehicle with the yaw damping moment and the
static directional stability moment both acting to resist the control
moment and the vehicle’s turning motion, i.e., trying to return
the vehicle to a straight path. The control moment acts in the opposite
direction to the other moments and maintains the vehicle’s curved
path.
As a vehicle’s understeer gradient
becomes smaller, the static directional stability derivative decreases,
and when the vehicle’s understeer gradient becomes zero (i.e., a
neutral steer vehicle), the derivative becomes zero. Therefore, the static
directional stability moment also becomes zero, and the only moments acting
on the “neutral steer” vehicle are the yaw damping moment
and the control moment, which in the case of a steady-state turn as shown
in Figure A-1, are equal and act in opposite directions.
In the case of an “oversteer”
vehicle (i.e., a vehicle with a negative understeer gradient), the static
directional stability derivative is negative, and its corresponding moment
will act to reduce the radius of whatever turn the vehicle is negotiating
(i.e., tighten the turn). As such, it will oppose the stabilizing influence
of the yaw damping moment. As shown in Figure A-1, in the special case
of an oversteer vehicle being driven at the “critical speed,”
only the yaw damping moment and the static directional stability moment
act on the vehicle in a steady-state turn. In this case, these moments
are equal and acting in opposite directions. While the yaw damping moment
still acts to resist the vehicle’s turning motion, the static directional
stability moment acts to resist the yaw damping moment and maintain the
curved path.
The yaw damping moment decreases as vehicle
speed increases while the static directional stability moment does not
change substantially with speed in the linear range (in fact, the simple
two degree-of-freedom model upon which Figure A-1 is based predicts that
the static directional stability moment does not change with speed in
the linear range). Therefore, there exists (only for an “oversteer”
vehicle) a “critical speed” at which the yaw damping moment
is exactly balanced by the static directional moment. At the “critical
speed,” if the steering wheel is turned, there are no stabilizing
moments available to balance the control moment. As such, any fixed steer
angle input would be unbalanced, and the vehicle’s response would
be an ever-increasing lateral acceleration and yaw velocity, and an ever
decreasing radius of turn until the vehicle experiences a “spin-out.”
At “critical speed,” an oversteer vehicle can only maintain
a steady-state response if the control moment is zero; i.e., if the steering
wheel angle is zero. In others words, an “oversteer” vehicle
traveling at the “critical speed” cannot maintain a steady-state
response with a non-zero steady steering input.
For “oversteer” vehicles traveling
in a curved path at speeds above the “critical speed,” the
static directional moment will always exceed the yaw damping moment. Therefore,
to maintain such a steady-state turn, a control moment opposite to the
direction in which the vehicle is turning is needed to balance the moments
acting on the vehicle; i.e., the driver must steer the vehicle “out
of the turn.” Without a driver’s intervention with such “reverse”
steering, the vehicle could no longer maintain a steady state response,
and the “fixed-control” response of the vehicle would become
unstable.
However, unstable does not mean uncontrollable.
The above discussion is based on the analysis of the steady-state vehicle;
i.e., one for which the steering input is fixed (does not vary with time).
During real driving, the steering input is not fixed; i.e., the driver
can and usually does turn the steering wheel to respond to an evolving
driving situation. For example, if an “oversteer” vehicle
starts to diverge from the desired course, the driver can turn the steering
wheel so as to bring the vehicle back to the desired path. This is an
example of “closed-loop” control, as discussed earlier.
4. Linear Range versus Non-Linear
Range Vehicle Characteristics:
A vehicle’s directional control
and stability characteristics in the linear range are not the sole determinant
of how the vehicle will respond following a rear tire tread separation.
Another key factor is how the vehicle behaves at its maneuvering limit.
A vehicle's response at its maneuvering limit is called its “limit
response.” A directionally unstable limit response leads to a spin
out, while a directionally stable limit response would result in a “plow”
or “push.” In either case, the vehicle would not be controllable
through steering inputs at its maneuvering limit, regardless of whether
it exhibited understeer, neutral steer, or oversteer characteristics in
the linear range.
In the linear range, higher measured understeer
gradients correlate with higher levels of vehicle stability. However,
it is not appropriate to attempt to apply the understeer-oversteer concept
to the nonlinear range because in that range one cannot assess stability
by measuring the vehicle's response to control inputs — which is
the way in which understeer gradients are determined.
The information available to ODI indicates
any vehicle that has experienced a rear tire tread separation would exhibit
an unstable limit response. Thus, even if a vehicle exhibits linear range
understeer characteristics in the linear range and is therefore stable
and controllable in that range, if a large enough steering input is made
such that the vehicle's response reaches the non-linear range; the vehicle
would exhibit a directionally unstable limit response and would spin out.
--------------------------------------------------------------------------------
[1] Throughout this document, Ford Explorer
will be used to refer not only to the Ford Explorer vehicles, but to Mercury
Mountaineer and Mazda Navajo vehicles that are the same vehicles that
are "badged" for the other two "makes."
|