2.    Roadholding Testing Offered by ATI


2.1.1  Definitions

“Control response” and “roadholding” refer to the open-loop response of the vehicle to inputs from steering, throttle, and brakes.

 “Handling” refers to the closed-loop behavior of the vehicle and driver, which involves the visual, kinesthetic and proprioceptive communication from vehicle to driver in addition to reliable control response. For example, a “couch-potato car” may have control responses that are in many respects proper, but will have poor handling. Similarly, a car may have good control response, but still “have no soul”. Obviously, “good handling” is often a subjective opinion. Objective measures of handling usually involve the driver workload requirements in a given closed-loop driving task.

“Rollover” refers to those “limit” maneuvers that challenge a vehicle’s rollover resistance.  In rollover tests the severity of the control inputs are increased until the response limit - plowout, spinout, or rollover – is reached. Rollover testing may use either closed-loop avoidance maneuvers or open-loop maneuvers: but only open-loop tests can be objective and driver-independent.

2.1.2  Control Response Testing Currently Offered

SAE J266- Steady State Directional Control Test Procedures

ISO 7401 - Lateral transient response test methods ISO 7975 - Braking in a turn - open loop test procedure

ISO 9816 - Power-off reactions of vehicle in a turn- Open-loop test procedure

ISO 9815 - Passenger car-trailer combination - lateral stability test procedure

2.1.3  Handling Tests Currently Offered

ISO 3888  Part 1 – Test procedure for a severe lane change

ISO 3888 Part 2  - Elk Test

NHTSA  ESC Test Series


2.1.4  Rollover Tests Currently Offered

NHTSA  Rollover Test Series -
     Slowly Increasing Steer
     Road Edge Recovery Maneuver

ATI  Rollover Test  Protocols (See Part 6)-
    ATI  Reversed Steer


2.1.5  Additional  Roadholding Testing Services

Split Image Videography - is performed in real time using multiple synchronized video cameras; usually at little or no extra testing costs.

Braking Tests - Simultaneous measurement of braking factors: pedal effort, booster vacuum, pressure at each brake line, speed of each wheel, vehicle speed, and brake temperatures. Measurement of vacuum booster exhaustion effects. Measurement of ABS internal electronic signals.

Aerodynamic Effects - Specially-designed, fast-response airspeed/direction system is used in conjunction with steering wheel angle and inertial sensors to study wind response, aerodynamic interaction with trucks, etc. One vehicle's response is compared to another "reference vehicle" by the methodology described in ISO/TC22/SC9/WG5 N11 (21 May 1991).

Acceleration/Braking Lift/Dive - Suspension deflections and pitch angle gradients due to acceleration and braking.

Vehicle Performance - Time to speed and distance, braking stopping distances, etc.

Accident Site Driving Task Analysis - Accident site drivethroughs using fully-instrumented exemplar vehicles, with "driver's eye view" and roadside video coverage.

Tire Deflation - Performed on test track or highway, by multiple solenoid valves mounted in tire rim, activated on switch command through a slipring assembly.

Component "Failures" - Staged failures of steering, brake, or suspension components with documentation of vehicle performance by instrumentation and real-time split-image (2 pictures-within-picture) video.

Vehicle Systems - Comprehensive testing of electric and electronic; hydraulic and pneumatic; fuel; engine management; diagnostic; servo control; active and passive safety systems.

Four-Wheel Steer, Traction Control, Stability Augmentation - Analysis and testing of electronically augmented or controlled steering, brake, throttle, and suspension systems. Experimental determination of operating algorithms.

Data Presentation - By "Mitsubishi Rhombus", "Weir Diagram", "Video G-G", and other methodologies.

Analysis of Testing by Other Facilities - Analysis of methodology, instrumentation capabilities, etc. Plotting of raw data; assessment of validity; interpretation of results.



ATI's standard test programs are carried out on the drag strip and adjacent parking area at Raceway Park in Englishtown, NJ. The drag strip is constructed of extremely smooth blacktop, 80 feet wide and 4,000 feet long. It is level in the longitudinal direction, and has a constant one-degree cross slope for drainage. Adjacent to the dragstrip is a paved (March 1996), level-surface parking area, 400 feet wide and 2,000 feet long, with an entrance road for acceleration to speed. All testing in which the steering machine is used are run in the parking area. Test programs may also be run at various proving grounds: e.g. Uvalde, Flat Rock, TRC, etc.



2.3.1  Data Channels

In vehicle handling tests the following data channels may be  recorded or plotted:

In tip-up testing the following additional data may be recorded or plotted:

2.3.2  Transducers Used in ATI Handling Tests

Speed and Heading angle:  "VBOX" GPS transducer.

Wheel speed/Tire deflection/suspension travel: A hollow-shaft encoder runs on a plate supported by lugnut extensions. The body of the encoder is in a frame that is connected through a pin/clevis joint to a vertical shaft. The vertical shaft travels through a fender-mounted eyebolt assembly which contains a string encoder for measurement of suspension travel. The encoder measures wheel speed. A downward-facing ultrasonic distance sensor measures tire deflection.

Steering wheel angle/effort: "second steering wheel" normally clamps to original steering wheel, with adjustable arm to suction cup on windshield. Optional attachment is directly to the steering column, replacing the vehicle steering wheel.  Strain-gauge paddle wheel element with built-in amplifier measures steer effort. Optical encoder angle readout. Calibration by special torque wrench. Used in lane-change and free-control studies.  

Throttle position: Tap into vehicle’s throttle position potentiometer.

Throttle position: String encoder attached to accelerator linkage.

Brake effort: LEBOW 3363-200 brake pedal load cell. 200 pounds full-scale; linearity 0.1 percent. Cal by precision 100 lb TROEMER 9082 NBS Class F weight.

Brake effort: LEBOW 3363-300 brake pedal load cell. 300 pounds full-scale; linearity 0.1 percent. Cal by precision 100 lb TROEMER 9082 NBS Class F weight.

Yaw velocity: NORTHRUP Nortronics 3-axis DC-DC rate gyro package, P/N 77025, SN17 and SN18 (2 units, one installed in Humphrey Package). 90 deg/sec FS; linearity 0.5 percent; hysteresis 0.1 deg/sec; threshold 0.01 deg/sec. Calibration on 36 degrees/sec rate table.

Yaw acceleration: SYSTRON-DONNER INERTIAL DIVISION Fluid-Rotor Angular Servo Accelerometer, Model 4591. Full-scale 570 Degrees/sec/sec; Linearity 0.2 percent. Calibration check on pendulous swing.

Linear acceleration: Force-balance servo accelerometers: SYSTRON-DONNER Models 4383 and 4310; SUNDSTRAND Models 303 and 305, SUNDSTRAND Model 2180 "MiniPal". Several of each, with full-scale 0.25 to 10 G, linearity 0.05 or 0.1 percent. Calibration on sine table with precision gauge blocks.

Stabilized Inertial Reference System: Pitch & roll Angles, X,Y,Z accelerations, pitch, roll, yaw velocities: HUMPHREY Stabilized Accelerometer System, Model SA07-0304-1. Vertical/free gyro can be used with servo erection (at 2 deg/min) or as low-drift free gyro caged until beginning of maneuver. Angles are calibrated against accelerometers by forced drift. Gimballed accelerometers held vertical by gyro are SUNDSTRAND Linear Servo Accelerometer Model 303. Full scale 1 g, linearity 0.05 percent. Calibration ±1 g by tilting. Package is modified by installing NORTRONICS rate gyro package (S/N 17) inside and incorporating power supplies in box lid.  Weight 38 pounds,

Strap-down Inertial Measurement System (IMU1): X,Y,Z accelerations, Roll, Pitch, Yaw Rates. Assembly consists of three orthogonally-mounted SUNDSTRAND Model 303 linear servo accelerometers, together with three Systron Donner GyroChip rate sensors.. Dimensions 4x4x3.5 inches, 3.2 pounds.

Strap-down  Inertial Measurement Ststem (IMU2): X,Y,Z acceleration, pitch, roll, yaw velocities. Assembly consists of  three orthogonally-mounted  Sundstrand Mini-Pal servo accelerometers and three Systron Donner GyroChip rate sensors. Dimensions 3.5x4.5x2 inches, weight 1 pound.

Sideslip angle and ground-referenced roll angle: ATI-designed BETA TROLLEY TYPE 1. Angular accuracies 0.1 degree, ground-generated noise permitting.

Wind speed and direction: Assembly consists of a lightweight rotating cup anemometer and a fast-response, fiberglass-over-styrofoam wind vane, specially designed by Princeton University Aerospace Department.


2.4  Specialized Test Equipment

2.4.1  Data Acquisition System

All data is conditioned in instrumentation amplifiers, and anti-alias-filtered in tenth-order 14.2 hz Butterworth active filters. Available conditioners include DC General Purpose; DC Strain Gauge; Shaft Encoder; Frequency Input. The filters use precision operational amplifiers, one-percent resistors, and temperature-stable capacitors matched to 0.1 percent. The filter characteristic is flat to one part in 4000 at 10 hz and is down 66 db at 30 hz. The filtered data is sampled 60 times per second, digitized to 12 bit accuracy and resolution, and stored with 12 channels on one horizontal line of the video recorder. Each data line appears as a dot-dash line across the top of the visible video picture.

The digitized data can be shown as an overlay on the video picture, either as a 12 channel horizontal analog bar display, or as a vertical array of numerics, each ranging from -1000 to +1000 full scale. Both displays can be used simultaneously, if desired. The numeric display can be updated 60 times per second, on every video field, for single-framing analysis, or at slower rates (4 times per second and 2 times per second) for real-time viewing. In running the test, these displays are shown on a video monitor usually placed on the floor of the passenger side, for a visual check of proper transducer operation.

Only the data line must be recorded, as the other displays can be regenerated during playback, or recorded on the picture during editing.

The system has a plug-in 12 channel 12 bit Digital to Analog converter module, to provide strip-chart recording capability or analog recording into other data systems.

The system also has a plug-in "RAM Buffer" module, which serves as an interface to an IBM - PC computer. Thru this module, the computer sees the data being reproduced as RAM that is updated with twelve 12-bit data samples in a 53 microsecond burst occurring sixty times per second. Various "handshake" lines and data flags are provided for easy software access.

Unique Capabilities of Data Acquisition System - In many test programs there is visual and auditory "data" which can be seen or heard but not measured. The ATI "Real-time Video Data Acquisition System" (US Patent 4,819,896) overcomes this problem by recording all three sources of information - sight, sound, and instruments - simultaneously and in synchronism, on videotape.

Display Capabilities - The real-time data display overlaid on the video picture has proved to be particularly useful. For example, in accident scene drivethroughs, in demonstration of avoidance maneuvers, and in standard lane-change tests the video picture can be single-framed, with steering wheel angle, lateral acceleration, and other variables shown in 1/30 second increments against the driver's view of lane boundaries or other obstacles.

Video G-G Diagram - The ATI "Driver-Vehicle Display Apparatus" (US Patent Number 4,716,458) shows the instantaneous cornering/braking/acceleration state of a car with respect to its various limits, overlaid on the "Driver's eye view" video picture. With acceleration/braking as the vertical axis and lateral acceleration as the horizontal axis, various diagrams bit-mapped in PROM are used to represent the driver unwillingness limit, the vehicle's control limit, the rollover limit, etc. The actual forward and lateral accelerations experienced by the test vehicle are combined into a moving dot on the screen. The instantaneous relationship between the vehicle and its limits are immediately obvious. The "Video G-G Diagram" has been utilized in demonstrating safety margins in vehicle scene drive-throughs, in showing the severity of lane change maneuvers, and in demonstrating the margins between vehicle handling limits and rollover limits.

Race Car Video G-G - In the adoption of the video G-G diagram, horizontal bar graph representations of throttle and brake are added at the top of the screen, and lap time in a numeric display at the bottom of the screen. The lap time is reset by a radio signal sent by a stopwatch operator at the side of the track.

Real-Time Split Image - The output of up to three video cameras can be combined into one composite picture, maintaining the capability for data overlay. Typical applications have a master "driver's eye" view through the windshield, with smaller picture-in-picture views to show disconnected components, wheel lift, etc. The Heitz PIP system is done in real time, using three cameras and one recorder, for economy and to enhance admissibility.

2.4.2  Programmable Steering Machine

See Heitz Section and downloadable PDF files.


2.5  Rollover Resistance Test Evaluation

2.5.1  Overview of Rollover Analysis

Objective rollover analysis includes the following areas: vehicle form and function; criteria for vehicle rollover resistance; vehicle rollover testing; vehicle handling characteristics; and the role of vehicle trajectory  and tripping mechanisms in accidents.

1.      The first rule in all industrial design is "Form Follows Function". Accordingly, a vehicle designed for off-road use at the expense of on-road performance will be relatively narrow and short for maneuverability; it will have large road clearance and measures to protect running gear; and it will have oversize tires for flotation in mud or loose sand. Small vans and short station wagons will be taller than passenger cars because passengers must "sit up straighter"; and vans will also be taller for easy entry. These characteristics, dictated by the vehicle's intended use, will generally result in a higher center of gravity and a tendency toward compromised rollover resistance in vehicles designed primarily for off-road use or for utility than in passenger cars or sports cars.

2.      A vehicle's rollover resistance should be evaluated by various objective criteria, in the light of its expected use.  The design aim must be maximum practical safety for the vehicle’s demographic target and intended use.

3.      Static rollover resistance depends primarily upon center of gravity and trackwidth; on suspension roll compliance; and on lateral elasticity of tires and suspension. Dynamic rollover resistance depends upon these factors, plus roll inertia and shock absorber characteristics, and the vehicle's handling responses.

4.      There have been five different static criteria for vehicle rollover resistance. “T/2h” (the ratio of the half-track dimension to the CG height) can best be described as the rollover resistance of the equivalent brick: it is no more than a first-order approximation because it completely disregards the static lateral elastic properties of tires and suspension, as well as all vehicle dynamic aspects. "Tilt table ratio" takes into consideration the static lateral elastic properties, but introduces a confounding, unnatural rise in the vehicle CG proportional to the departure of the cosine of the tilt angle from unity. Tilt table ratio has the advantage of simplicity and repeatability, but it is "repeatably wrong". "Side Pull Ratio" is a relatively difficult and complicated test, but when performed correctly with precision equipment it provides the best available estimate of static, as contrasted to dynamic rollover resistance. "Critical sliding velocity" (CSV) adds vehicle roll moment of inertia to track width and CG height, and provides a theoretical estimate of the minimum lateral sliding speed at which a tripped rollover will occur.  "Steady-State Rollover Safety Margin" is the margin between the maximum steady-state cornering capability of the vehicle as measured in one of the SAE J266 handling test protocols, and the sidepull ratio as measured in laboratory tests.

5.      Sidepull has the advantage of measuring separately the effects of body roll angle and the lateral compliance of suspension and tires, and the vertical movement of the sprung mass due to side load. Sidepull therefore yields true measurements of T/2 and H, under static conditions.

6.      Critical Sliding Velocity can be improved by using the values of T/2 and H measured in a sidepull test as opposed to "undeformed" values obtained from measurements without applied sideload.

7.      Because of these considerations, Side Pull Ratio is the most objective static laboratory criterion presently available for evaluation of “steady-state rollover resistance”, and CSV using sidepull data is the most complete overall criterion using only laboratory test procedures.  Although Sidepull Ratio is theoretically the most objective static criterion, it requires very sophisticated testing equipment, currently available at only a few places, such as General Motors and Heitz Chassis Lab.

8.      In the period 1971-1974 NHTSA made a concerted effort to develop a consistent dynamic test procedure for vehicle rollover. The "VHTP" procedures: braking in a turn, trapezoidal steer, sinusoidal steer, and "drastic steer & brake maneuver", all using a primitive automatic vehicle controller to remove the human element, were developed. The conclusion of this effort was that rollover testing is essentially a "can of worms" for two reasons -- tire variability and roll/yaw synergism. (1) In recent years, however, both of these difficulties have been overcome.

9.      Tire peak sideforce was found to increase with successive runs during a test due to shoulder wear, by as much as 40 percent with some square-shouldered 1970-vintage bias-ply tires. It was later found that radial tires tend to behave oppositely, but with less variability, due to their as-designed rounded shoulders. Tires with different tread patterns would change at different rates (squared-shoulder greater than rounded-shoulder), such that a consistent test procedure was impossible to develop. (Recent ATI rollover tests with modern radial tires have found stable lateral acceleration measurements in repeated test runs). (2)

10.  Roll/yaw synergism made the timing of steer reversals an important factor in rollover. It was found that a human driver, after a few practice runs to get a "feel" for the vehicle, could roll a passenger car that would not roll under the automatic controller's program.  However, a quandary arose in the driver-control testing. After “getting the feel” the driver could achieve rollover in every run; but after a tire change it would take two or three runs to “get the feel” again. The question was whether the “learning” was taking place in the driver or the tires.  This problem was solved 1n 1997 with the introduction of the Heitz Programmable Steering Machine, which provides precisely-repeatable steering inputs.

11.  In 1988 and again in 1996, Consumers Union called certain vehicles Unacceptable, in the subjective opinion of CU's experts, because they tipped up in CU's obstacle avoidance maneuver. Critics pointed out that the tests were not instrumented, so the exact maneuvers could not be duplicated for evaluation; that test results were very driver-dependent; and that other vehicles with less rollover resistance had been deemed OK by Consumers Union experts.

12.  In 1992 a NHTSA-sponsored report by Systems Technology, Inc. described  “Steady-state Rollover Propensity Margin” as a rollover criterion.(3) This margin was defined as the difference between Sidepull at tip-up and the maximum lateral acceleration (“max lat”) in a circle test. This criterion was immediately put to use by ATI as a useful steady-state criterion, and it was later adopted by General Motors as a design guide.

13.  In 1997 Toyota unveiled their "Fishhook" test, which consists of a violent swerve to miss an obstacle pylon, followed by a steer reversal to full lock in the other direction. The Fishhook test is instrumented, and the criterion obtained is the lateral acceleration at tip-up. According to the test protocol, if tip-up is not achieved, pulse braking is used to stimulate tip-up. (4)

14.  In 1998 the Swedish magazine Teknikans Verld rolled one of the new A-class Mercedes in an obstacle avoidance maneuver called the "Elk Test". Mercedes responded by stopping production to modify the vehicles, and by demonstrating rollovers of six competitive vehicles: the VW Sharon and Golf, Renault Megane Scenic, Peugeot 306, Audi A3, and Opel Astra. Opel rolled a VW and VW rolled an Opel. The methodologies for these demonstrations were not published, and may have used extreme combinations of steering, throttle, and brakes. For this reason the International Standards Organization Technical Committee 22, Subcommittee 9 (Vehicle Roadholding) responded by beginning work on a standardized Elk Test. The result was the introduction of ISO 3888 Part 2

15.  In 1997 ATI/Heitz introduced the Heitz Programmable Steering Machine, called “Sprint 1”, which could provide precisely-repeatable programmed steering inputs in an easily-installed package.(5) One of these machines was supplied to NHTSA in April 1998.

16.  In June 1998 NHTSA began their "Phase II Test Plan for Dynamic Rollover Research", using the Heitz Programmable Steering Machine. The test Plan included eight Protocols: Frequency Response; Variation of steady-state steering gain with increasing lateral acceleration at 50 mph; variation of steady-state gain with speed at constant steer angle; J-turn without pulse braking; J-turn with pulse braking; open-loop Fishhook, without pulse braking, with steer timing determined from the frequency response test; open-loop Fishhook without pulse braking, with steer compensated for the vehicle's steering ratio; and resonant steer testing. The NHTSA protocol attempted to overcome the steer timing problem in the 1972 Michigan tests, but the timing of the pulse braking was still driver-dependent. The test series included 12 vehicles: three each of SUVs, cars, trucks, and vans; in order to define a spectrum of rollover behavior. (6)

17.  In 1998 ATI/Heitz implemented “Start at preset speed” and “Roll rate feedback” in the Sprint 1 Machine. The test driver could bring the vehicle to a speed somewhat higher than the test speed, release the throttle and depress the START PROGRAM switch. When the vehicle slowed to the precise set speed the steer program would begin. In this way the uniformity of test initial conditions is enforced. With “Roll rate feedback” the steer reversal in a Fishhook-like maneuver always occurs at maximum roll angle (roll rate zero-crossing), regardless of the vehicle’s roll response time.  Roll rate feedback offered the simple solution to the “tire/driver learning problem”.

18.  In 1999 Heitz upgraded the NHTSA machine and supplied a second-generation Sprint 2 machine to several vehicle manufacturers, with the additional test automation capabilities. These features would simplify the test procedure by eliminating the need for prior frequency response testing, and would standardize the timing of steer, regardless of the vehicle dynamics.

19.  In April 1999 ATI announced three rollover test protocols: "Reversed Steer With and Without Braking" and "J-Turn With and Without Braking", and “Resonant Steer”. (7)  These were made possible by the addition of roll rate feedback to the Steering Machine capabilities. Experience with these protocols and further research led to a number of minor changes in the reversed-steer protocols and dropping the J-turn and resonant steer tests in 2002.

20.  A study of the time and energy aspects of rollover testing and the forces on the outrigger led to criteria for data filtering and an objective definition of rollover based on outrigger force. (8) Systematic tip-up testing resulted in criteria for when tires must be changed during a test. Run-to-run comparisons with different steer rates supported the selection of 600 degrees/second as an optimum balance between rollover efficiency and human capabilities, for the ATI test protocol.

21.  Reversed-steer testing in 2000 revealed that  the "Steady-State Margin" used by ATI and others since 1992 was inadequate, for two reasons. The dynamic roll angles before tip-up were much greater than those measured in the static sidepull tests; and for typical understeer vehicles the reversed steer produced much greater rear tire slip angles and therefore greater total tire forces. Reversed steer  produces higher side forces and lower "T/2h", and therefore significantly reduces the "safety margin" from that  which had been believed prior to 2001.                

22.  In October 2002 NHTSA announced its intention to use a Reversed Steer with roll rate feedback as  a rollover test for its NCAP Consumer Information Program. All tests will use the Heitz Programmable Steering Machine. 

23.  Accident investigations indicate that rollovers always involve out-of-control situations. In the most frequently reported scenario a vehicle drifts off the road on the right due to driver inattention causing, in a panic reaction, a gross overcorrection to the left followed by an even greater overcorrection to the right. Rollover occurs during the second half-cycle, or during a following portion of this "driver-induced oscillation". Careful, systematic testing which examines a vehicle's behavior in a steering reversal maneuver, with steer inputs designed to keep the trajectory within reasonable lateral "roadway" boundaries while at the same time observing human factors limits in amplitudes and rates, has relevance to safety evaluation. While the occurrence of tip-up per se may be typical of certain classes of vehicle, the broadness of the range of inputs that cause tip-ups may separate vehicles within a class. The reversed steer protocol introduced by ATI is based on these factors.

24.  Systematic dynamic rollover testing has demonstrated the inadequacy of steady-state rollover criteria. Rollover occurs when the dynamic lateral acceleration exceeds the dynamic T/2h by a sufficient amount, for a sufficient time. Dynamic rollover testing using repeatable, reproducible, objective protocols is the only sure way to evaluate rollover. In particular, it has been found that roll angles in dynamic testing are considerably greater than those in static tests such as side pull or in steady-state testing. This is a very important factor in rollover, since the resulting lateral movement of the vehicle cg causes a dramatic reduction in "T/2h". In fact, the most effective practical measure to increase rollover resistance is a reduction in dynamic roll compliance.


2.5.2  NHTSA vs. ATI Rollover Test 

The ATI Reversed Steer Protocol and the NHTSA NCAP Rollover test (variously called "NHTSA Fishhook" or "Road Edge Recovery Maneuver") are similar in that both use reversed steer with roll rate feedback. However, their basic aims are different: the ATI protocol is an engineering test intended to simulate an on-road highway rollover accident while producing data on vehicle characteristics important to rollover. The NHTSA is a Go/No-Go test intended simply to discriminate among vehicles for consumer information. "Relatability" to the real-world highway environment is not as important in the NHTSA test as the ability to discriminate between vehicles in obtaining "Star Ratings" for consumer information.

The standard ATI is run at 50 mph with steer angles incremented from 90 degrees to 270 degrees. 50 mph is a typical highway speed which is vehicle independent, and with steer angles above 270 the loaded "outside" front tire tends to suffer bead unseating. Bead unseating is almost never found in on-road highway accidents, and steer of 270 is considered to "push the limit" of panic steering in non-test situations. (Human capabilities are much greater, and in test situations extreme steer and steer rates are common (12). The ATI test is intended to simulate on-road rollovers, and too-large steer angles may also produce too-large path deviations. With incremented steer, plots can be made of lateral acceleration, yaw velocity, sideslip, roll angle, etc. to study vehicle trending toward rollover. If rollover occurs, the protocol provides small variation in test speed to determine precise thresholds. The protocol also provides optional additional testing at incremented speeds at constant steer to plot motion "partial derivatives" - variation with steer at constant speed and variation with speed at constant steer.

The NHTSA test has regulatory aspects. in which all vehicles from minicars to large vans must be tested under the same conditions For that reason the NHTSA test uses constant steer angle with speed increased incrementally from 35 mph to 50 mph. The lower test speeds are considered safer than 50 mph, especially for the larger vehicles. For fairness between vehicles, the test steer angle is set at 6.5 times the steer required to obtain 0.3g of lateral acceleration at 50 mph. This is intended to compensate for differences between vehicles in steering ratio and understeer gradient. It sometimes occurs that a vehicle will not tip up at the 6.5 ratio due to tire force saturation: in these cases a ratio of 5.5 is tried, sometimes successfully. To avoid bead unseating tubes are used in test tires: this protective measure causes no significant effect on test results.

In NHTSA testing some vehicles exhibit severe bouncing on the outriggers, which is another reason to restrict test speeds. This behavior has not been seen in ATI testing: the difference is possibly due to outrigger design. The NHTSA outriggers are relatively stiff transverse beams attached to front and rear bumper mounts. These can act as undamped springs when a vehicle tips up "hard" onto them. ATI uses one center-mounted outrigger on each side, with air cylinders fitted to soften the strike-down and absorb energy, and they may prevent the severe bouncing experienced by NHTSA. On the other hand, the ATI outriggers permit 4 or 5 degrees of additional roll angle after "outrigger down".

The ATI protocol is limited to tip-ups with the driver on the "low" side, to avoid back injury to the driver when the vehicle "falls" at the end of a tip-up run. The NHTSA test must be run in both directions, which is incentive to limit tip-up roll angles.

2.5.3  The "Hump" Problem In Reversed-Steer Testing (Reference 11)

The use of roll rate feedback, originated by ATI in 1998 was intended as an economical solution to the timing of the steer reversal, since both fast-responding and slow-responding vehicles would have their steer reversals at maximum roll angle. However, in testing during 1991 by NHTSA and ATI of the same vehicle, occasional long steering delays were observed. The problem was traced to variation in the time of zero-crossing of the roll rate signal.

After considerable research at ATI the source of the phenomenon was discovered to be in the difference in roll frequencies of the vehicle sprung mass on its suspension and the total vehicle on its tires. Long delays resulted when the total mass frequency was more than twice the sprung mass frequency. Two "countermeasures" were developed: a bandpass filter to suppress the roll rate signal and a lead network to accentuate it.

The effects of longer delay in steer reversal serves to increase sideslip angles and scrub off speed at a greater rate, and these tend to cancel each other. NHTSA decided to let the vehicle "be itself" and omit the filters. ATI uses band pass selectively: when long delay is observed the test is repeated with a filter and notes whatever difference it makes.

2.5.4  J-Turn Test

The J-Turn test, typically consisting of a 180 degree step steer, applied at 500 degrees/second and  held in for several seconds at 50 mph, has been around since the 60's at least. Its advantage is simplicity to run: the steering wheel is manually turned rapidly against a stop. The J-turn has two major disadvantages. If tip-up results, it generally occurs at deviations from the original path of 100 feet or more,  And except for an initial transient it is a steady-state test which is an inefficient rollover test. It was dropped by NHTSA because it did not offer anything in addition to theit reversed-steer "Fishhook" test.

In the J-turn with pulse braking, a hard brake pulse is applied at the peak roll angle resulting from the steer input. If all four wheels lock up, the vehicle transitions from a circular path to a straight-line tangent to the circular path, since a sliding tire has no directional sense. Releasing the brakes immediately restores the circular path. By optimizing the brake pulse, the vehicle can be "rocked" at a frequency unattainable with the steering wheel, so tip-up can sometimes be achieved. The brake input must be very precise in timing and must be sufficient to lock the wheels. Its relatability to the rear world is poor at best; and it may not work at all with anti-lockup brakes. 


2.5.6  Resonant Steer Tests

On paper the resonant steer test looks promising. However, testing of several protocols by several organizations as part of an ISO task force showed that with large steer angles the resulting trajectories can be "crazy", and repeatability is very poor. 


2.5.7  Roll Behavior Near Tip-up Thresholds (Reference 8)

The ATI Reversed Steer Protocol has enabled precise study of roll behavior at the tip-up threshold. Below the threshold the roll angle increases to an asymptotic value or (more usually) a slight overshoot, then settles to a steady-state value. Just above the threshold  the roll angle tends to follow an identical trace, but then "takes off" to a high value. At the threshold the roll can go either way, like a card standing on end. In the tip-up case the inflection point where roll angle begins to turn upward can be considered the time of rollover.

In all cases the peak roll angle before tip-up is significantly greater than that found in steady-state tests, because of the time lag between the "excess" roll angle and the take-off point.


1.      R.D. Irvin, P.S. Fancher, L. Segal, Refinement and Applications of Open-Loop Limit-Maneuver Response Methods, SAE Paper 730941.

2.      Tire Shoulder Wear In Repetitive Rollover Testing, ATI Report No. 111901, November 2001.

3.      R. Wade Allen et al, Vehicle Dynamic Stability and Rollover, Report No. DOT HS 807-956, STI Report No. TR-1268-1, June 1992.

4.   Method for Measuring Lateral Acceleration for Rollover, Toyota Engineering Standard TSA1544, revised March 1997.

5.   E.J. Heitzman and E.F. Heitzman,  A Programmable Steering Machine for Vehicle Handling Tests, SAE Paper 971057.

6.   An Experimental examination of selected Maneuvers that May Induce On-Road Untripped, Light Vehicle Rollover – Phase II of NHTSA’S 1997-1998 Vehicle Rollover Research Program, DOT HS Report, July 1999.

7.   Specifications for ATI Rollover Test Protocols,  http://www.atiheitz.com/rolltest.pdf.

8.      E.J. Heitzman and E.F. Heitzman, On Road Rollover Testing: Outrigger Height and Data Filtering, ATI Report No. 011500, January 2002.

9.      Part II, Department of Transportation, NHTSA, 49 CFR Part 575: Consumer Information Regulations; Federal Motor Vehicle Safety Standards; Rollover Resistance; Proposed Rule. Federal Register, October 7, 2002.

10.  G.J. Forkenbrock and W.R. Garrott, A Comprehensive Experimental Evaluation of Test Maneuvers That May Induce On-Road, Untripped, Light Vehicle Rollover. Phase IV of NHTSA’s Light Vehicle Research Program. DOT HS 809 513,  October 2002. 

 11. The Hump in Roll Rate Feedback: Source and Countermeasures, ATI Report No. 030112, January 2002.   

 12. G.J. Forkenbrock and D.E. Elasser, An Assessment of Human Driver  Capability.  DOT HS 809 875, June 2005.


2.6  Testing and Consulting Services:  Rates and Charges

TRAVEL TIME CHARGES: Travel time charged is based on what seems reasonable under the particular circumstances, including such factors as effective time lost from a normal office workday, work vs. recreational reading or sleep accomplished enroute, and the time "it should have taken" under optimal conditions.

PROJECT EXPENSES: No charges are made for equipment owned by ATI, including test facilities, instrumentation, video, data acquisition and computation equipment; or for normal expendables (videotapes, etc.). Project expenses that are charged as line items include travel expenses, unusual expendable materials, rented equipment, subcontracted work, special printing and reproduction, shipping charges, special fees, or supplemental insurance. Air travel is normally in first class. However, unless specifically agreed otherwise, it is charged at the published unrestricted coach (Y) fare basis, as listed in online reservations systems. Local mileage is charged in accordance with IRS guidelines.

FIXED PRICE SERVICES: When the testing services or the final work product can be sufficiently defined, ATI prefers to quote such services at an agreed-on fixed price, on a "not to exceed" basis.

PAYMENT: Invoices are rendered monthly or at project "milestones", and are due upon receipt. Outstanding balances past due over sixty (60) days are considered to be delinquent. Interest at 1.5 percent per month will be charged on delinquent accounts.


2.7  Statement of Terms and Conditions of Agreement

CONTRACTUAL RELATIONSHIP : Neither ATI nor a prospective Client is bound in any way by a quotation from ATI, except for confidentiality as stated below. ATI assumes that a contractual relationship exists only after a signed agreement equivalent to ATI’s Standard Contract Form has been received. If the start of contracted work is delayed by the Client for three months after an agreement is signed, the agreement shall be nullified.

CHARGES: Work performed on a time-and-expenses basis will be charged in accordance with the most current "Schedule of Rates and Charges". Work performed under fixed price purchase of testing services will be charged at, or less than, the agreed "not to exceed" amount.  Any unusual types of work not specifically covered by the original agreement may be subject to a revised agreement. A suitable retainer or prepayment will be required from the Client in advance. Such amounts will be held by ATI until the final invoice, at which time the Client's account will be reconciled.

PAYMENT: Invoices are rendered monthly or at project "milestones", and are due upon receipt. Outstanding balances past due over sixty (60) days are considered to be delinquent. ATI, without liability, may withhold delivery of reports and other data, and may suspend performance of its obligations to the Client, pending payment of delinquent charges. Furthermore, ATI reserves the right to decline further work with any Client delinquent in payment of charges due to ATI for previous work, until such balances are paid in full.

EXECUTION OF SCOPE OF SERVICES: ATI will perform all work in accordance with generally accepted professional engineering practice. In the event of deficiencies in the work performed, such as errors, omissions, or ambiguities, ATI's sole responsibility and liability will be to provide without additional charge, corrected, revised, amplified, or clarified data sufficient to correct the deficiency. No other warranty, express or implied, is made concerning work performed under the agreement. ATI will diligently proceed with the contracted work from the agreed-on start date and will report to the Client in a timely manner, except for delays occasioned by factors beyond ATI's control, by factors that were not reasonably foreseeable, or by factors initiated by the Client. Work under the agreement will be terminated upon receipt by ATI of written notice from the Client, except that ATI may complete such analyses, records, and reports as are reasonably necessary to protect professional reputation and to adequately document the work performed through termination.  Charges for such work will be kept to a reasonable minimum, not exceeding ten percent (10%) of total charges incurred through the date of termination. Work under the agreement may be terminated by ATI only for just cause, including but not limited to excessive delays caused by the Client.

CONFIDENTIALITY: ATI will hold in strictest confidence all proprietary or confidential information of a Client or prospective Client to which it may be given intentional or accidental access. Unless otherwise expressly agreed in writing, all test data, videotapes, reports, and other information provided to the Client under this agreement shall be the exclusive property of the Client. ATI will not divulge under any circumstances except explicit written direction by the Client any test data resulting from its work, or any aspect of that data including the existence of the test.

MISCELLANEOUS: The Client assumes full and complete responsibility for all uses and/or applications of ATI's work under this agreement, and agrees to indemnify and hold harmless ATI, its officers, directors, employees, or shareholders against any and all liability, damages, losses, claims, demands, actions, causes of action, and costs including attorneys' fees and expenses, resulting from any alleged damages resulting from the aforementioned use, application, or non-use of ATI's work under this agreement. The Client agrees that in no event shall ATI, its officers, directors, employees, or shareholders be liable for any incidental or consequential damages, direct or indirect, arising from ATI's services under this agreement.  Unless otherwise expressly agreed in writing, ATI shall retain exclusive rights to all patentable ideas developed during the performance of this agreement. In any litigation involving the Client in which ATI is compelled by subpoena or judicial order to testify at a deposition, or to produce documents regarding work performed by ATI for the Client, the Client agrees to compensate ATI, at its prevailing hourly rates, for all time spent by ATI in responding to such legal process, including all time spent in preparing for and providing such testimony.  In such cases the collection of fees billable to opposing counsel will be the responsibility of the Client.

ATI Contract Form

This page last modified on 06/28/2011