AIR BAG DEPLOYMENT CRITERIA

By:

Jesse Kendall, P.E., Forensic Scientist at the Institute of Risk & Safety Analyses

([email protected], (818) 226-9974 x8)

Kenneth Alvin Solomon, Ph.D., P.E., Post Ph.D., Chief Scientist at the Institute of Risk & Safety

Analyses ([email protected], (818) 348-1133)

ABSTRACT

Air bag control modules utilize complex algorithms to make air bag deployment decisions based

on crash severity related to the change in vehicle speed or deceleration over time. Due to the

proprietary nature of air bag deployment algorithms, the velocity, acceleration, or displacement

thresholds for air bag deployment during a collision are not easily obtained; however, a range of

frontal barrier impact speeds and corresponding deceleration and displacement threshold values

for air bag deployment can be approximated using known vehicle stiffness-to-weight ratios.

KEY WORDS

air bag deployment, decision, event, data, sensor, trigger, threshold, algorithm, acceleration,

accelerometer, crash pulse, wakeup, enable

LEARNING OBJECTIVES

1. Gain an understanding of the air bag system and components.

2. Gain an understanding of when an air bag should or should not deploy.

SCOPE OF PAPER

An introduction to air bag systems and inflation processes will be followed by a brief history of

crash sensors. Variables used in air bag deployment algorithms will be described, and examples

of several patented systems compared. A method to estimate the range of speed, deceleration, or

displacement thresholds for air bag deployment will also be provided.

DEFINITION OF TERMS - See Appendix:Glossary

INTRODUCTION

The purpose of the air bag is to provide a cushion between the occupants and the vehicle’s

interior. For air bags to be effective they have to be fully inflated in a short amount of time,

before the occupants make contact with them; however, this rapid inflation can potentially cause

fatal injuries to certain people if they are in contact with the air bag during its inflation.

Therefore, air bags must have a control system that can recognize a crash correctly, and early

enough for the air bags to inflate safely.

THE INFLATION PROCESS

Air bags inflate after an electric current from the air bag control module is sent to a detonator.

This ignition starts a chemical reaction producing nitrogen gas which rapidly inflates the nylon

fabric air bag. The deployment and inflation process takes approximately 0.04 seconds [1].

After full deployment, as the occupant impacts and compresses the air bag, the nitrogen gas is

released through small vent holes. The holes are specifically sized and spaced to reduce the

volume of the bag at different rates, depending on the type of vehicle. The gas is released along

with dust particles from material used to lubricate the bag (typically talcum powder and

cornstarch).

CRASH SENSOR HISTORY

Early air bag deployment systems in older vehicles utilized mechanical sensors for crash

detection, which were phased out of the US market around 1994 [10]. Early mechanical sensors,

such as the “rolamite” by Sandia National Laboratories, relied on a metallic sphere that was

stabilized at a standby position by a spring or a magnet (see Figure 1).

 Figure 1. Mechanical Air Bag Sensor [10]

When the sensor was subjected to a force beyond a designed threshold, the spring or magnet

could no longer keep the metallic mass in place. The mass moved and made contact with an

electrode, sending an electrical signal to the air bag control module, which then sent a signal to

the air bag control module. Systems with mechanical sensors were generally inaccurate at

interpreting minor collisions. Movement within mechanical sensors can be underrepresented

with frontal collisions, and the acceleration the sensor experiences are sometimes slightly

delayed. As an improvement, modern air bag deployment now relies on microelectromechanical

system (MEMS) components.

NEW CRASH SENSING SYSTEMS

New MEMS crash sensors measure acceleration with an accelerometer that sends a continuous

stream of data to the air bag control module. Accelerometers are typically piezoelectric or

variable capacitance sensors. The most common MEMS accelerometer in use today is the

ADXL-50 by Analog Devices (see Figure 2 below).

Figure 2. Microelectromechanical Sensor (MEMS) [Source: Analog Devices, Inc.]

As an anchored mass moves relative to the sensor’s body due to acceleration, a plate attached to

the anchored mass moves closer to a stationary plate. The change in distance between the plates

affects the capacitance of the sensor, or the ability to hold an electrical charge. This change in

capacitance is easily measured and is then converted to a change in voltage. The voltage change

is directly correlated to force due to acceleration, and the readings are interpreted as acceleration

by the air bag control module. Using an algorithm, the control module can determine if air bag

deployment is necessary based on the pattern of the acceleration pulses over time.
THE DECISION PROCESS

The air bag control module (ACM) receives a continuous signal from each MEMS sensor and

records the data for a fixed period after a specific event. With a central processing unit (CPU), it

performs algorithmic computations and controls the “fire” or “no-fire” command for air bag

deployment. The triggering algorithms determine crash severity by evaluating one or more of

the kinematic variables (derivatives/integrals of acceleration) shown in Table 1. Examples of

algorithm decision flow charts are shown in the following Figures 3, 4, and 5.

Variable Expression Units

Acceleration ft/s2

Velocity (v) ft/s

Displacement ft

Jerk ft/s3

1
Energy Density 2 (ft/s)2

1
, :
Energy 2 ft-lb

Power ̅ ft-lb/s

̅
Power Density ft2/s3

Power Rate Density (ft/s2)2

Table 1. Kinematic variables used in air bag triggering algorithms.

Figure 3. Algorithm Flow Chart for US Patent 5948032 [Huang]
∆

s(t) = vehicle speed at time (t)

vth = 5 m/s  11 mph

Figure 4. Algorithm Flow Chart from [13] Hussain.

AVG1 = average acceleration over 4 ms,

AVG2 = average acceleration over 8 ms,

AVG3MAX = average acceleration over 24 ms,

Figure 5. Algorithm Flow Chart from US Patent 6236921 [McConnell]

ALGORITHM VARIATIONS

Crash sensing schemes vary greatly between patents. A majority of systems patented after 1995

utilize delta-V, acceleration, or jerk, as variables in the system wakeup command, and for

triggering the air bag. Recent systems also include occupant sensing and analysis of the distance

from the occupant [13]. Table 2 outlines the approaches used between 1995 and 2008 by several

inventors. The differences are considerable and widely varied; however, the basis for

deployment relies on one or more of the basic kinematic expressions previously described.

US Patent No. Year Inventor Assignee Title Description

Air bag deployment control
5394326 1995 Liu Delco Electronic Corporation V + acceleration
system and method
SIR deployment method based on
V + jerk + displacement +
5430649 1995 Cashler Delco Electronic Corporation occupant displacement and crash
acceleration
severity
Method and apparatus for sensing
a vehicle crash condition using
5587906 1996 McIver TRW Inc. V + acceleration
velocity enhanced acceleration
crash metrics
5668720 1997 Takahashi Toyoda Gosei Co., Ltd. Air bag controlling apparatus V + jerk + acceleration
V + jerk + displacement +
5777225 1998 Sada Sensor Technology Co. Crash sensor
acceleration
Method and apparatus for crash
5835007 1998 Kosiak Delco Electronic Corporation sensing using anticipatory sensor V + acceleration
inputs
Polynomial windowing algorithm V + jerk + displacement +
5948032 1999 Huang Ford Global Technologies
for impact responsive activation energy
Control method for variable level
5999871 1999 Liu Delphi Technologies V + jerk
airbag inflation
Three Speed Algorithm for Airbag
6236921 2001 McConnell Visteon Global Technologies V + jerk + displacement
Sensor Activation
Supplemental restraint deployment
7424354 2008 Shen Delphi Technologies method using dynamic crash V + jerk + displacement
classification

Table 2. Algorithm patents and criteria.

WHEN AIR BAGS DEPLOY

According to the National Highway Traffic Safety Administration [6], “Air bags are typically

designed to deploy in frontal and near-frontal collisions, which are comparable to hitting a solid

barrier at approximately 8 to 14 mph.” Specific thresholds are calibrated by each manufacturer

according to vehicle size and stiffness. In frontal collisions, the system ‘wake-up’ or ‘algorithm-

enable’ command is used to distinguish between events such as hitting a pothole and a collision

with an automobile. It is generally initiated when two consecutive acceleration pulses less than

(approximately ) -1.0 g for smaller vehicles or less than (approximately ) -2.0 g for larger

vehicles, occur within a short period of time (10 milliseconds) [3]. After system wake-up from a

pulse exceeding the deceleration threshold (stand-by mode), a decision is made to either fire the

air bag or return to normal state.

Due to the proprietary nature of air bag deployment algorithms, the velocity, acceleration, or

displacement thresholds for air bag deployment during a collision are not easily obtained;

however, using the NHTSA guideline for an air bag to deploy in frontal barrier collisions within

impact speeds of 8 to 14 mph, a range of threshold values can be estimated using known vehicle

stiffness-to-weight ratios.
THRESHOLD ESTIMATES

In a collision, the amount of crush (C, in inches) at a given impact speed (V, in mph) is related to

the ratio of the stiffness of a vehicle (k, in lb./in) and the vehicle weight (w, in lb.) by the

following equation [14]:

0.9

The time from the beginning of the impact to the time of the maximum crash pulse is:

56.8
0.64

By substituting for C/V, the time (tm) can be calculated using the weight-to-stiffness ratio as

follows:

0.9
56.8
0.64

Vehicle stiffness, (k) can be determined from collision test results, which report mass (m) , crush

(c), and impact velocity (v) for vehicles subjected to frontal rigid barrier collision testing.

Vehicle stiffness is calculated as:

Table 3 shows the corresponding range of decelerations and displacements in frontal barrier

collisions, at which air bags are designed to deploy, given the calculated time to maximum crash

pulse and different vehicle stiffness-to-weight ratios.

 Vehicle Vehicle Ratio Time to max. v a x
Vehicle Vehicle Vehicle Weight, w Stiffness, k k/w pulse, tm Delta‐v range Deceleration Range Displacement Range
Year Model Class (lbs) (lbs/in) (lb/in/lb) (ms) (mph) (mph) (g's) (g's) (in) (in)
2010 Ford Fusion midsize 3640 3839 1.055 78 8.0 14.0 4.7 8.2 7.0 12.3
2010 Lexus RX350 SUV 4747 5690 1.199 73 8.0 14.0 5.0 8.7 6.6 11.5
2009 Ford Escape SUV 4198 5070 1.208 73 8.0 14.0 5.0 8.8 6.6 11.5
2010 Honda Insight compact 3115 3917 1.257 71 8.0 14.0 5.1 9.0 6.4 11.2
2010 Toyota Prius midsize 3499 5672 1.621 63 8.0 14.0 5.8 10.2 5.7 9.9
2007 Chevrolet Corvette sportscar 3132 5526 1.764 60 8.0 14.0 6.1 10.6 5.4 9.5
2007 Chevrolet Equinox SUV 3172 8647 2.726 48 8.0 14.0 7.5 13.2 4.4 7.6

Table 3. Air bag deployment ranges (based on vehicle front impact with a rigid barrier).

There is no significant correlation between vehicle weight and stiffness. Two vehicles of similar

weight may have very different stiffness values, as seen when comparing the 2010 Ford Fusion

to a 2010 Toyota Prius. Both vehicles have approximately the same vehicle weight, yet the

front-end stiffness of the Toyota Prius is substantially greater than the Ford Fusion. Since both

the amount of displacement and the duration of impact for a Ford Fusion are greater, an air bag

would need to deploy in the Ford Fusion within a range of deceleration values lower than those

required for the Toyota Prius.

Comparing values

Real-world crashes are often not identical to solid barrier crashes, and care should be taken when

comparing ranges of tested and calculated values. Impact duration does not vary significantly

with impact velocity, but varies greatly with the type of collision. Air bags might not deploy

where there are extreme deformations, such as a collision with a telephone pole where only one

portion of the car is deformed. Air bags sometimes do not deploy when the impact is gradual,

over a longer period time as when a vehicle rides under or over another object. Air bags might

not deploy in collisions in which the relative stiffness are vastly different, such as the front of a

vehicle impacting the side of another vehicle. Additionally, collisions that occur at oblique
angles do not always result in air bag deployment, when significant deceleration does not occur

in a direction concurrent with the sensing device.

Case study: Air bag non-deployment

An example of an impact where there was a significant change in velocity, but the driver or front

passenger air bags did not deploy, involved a 2007 Chevrolet Equinox colliding with an

oncoming Harley-Davidson motorcycle. The air bag control module in the Equinox reported a

maximum V of 9.27 mph. This value was within the 8.0 to 14.0 mph range where deployment

(for frontal impact with a solid barrier) is expected; however, the maximum deceleration

recorded for the subject collision was only 3.27 g’s. The deceleration of the Equinox in this

accident was well below the estimated deployment range (7.5 to 13.2 g’s) as shown in Table 3.

Therefore, the driver and passenger air bags rightfully did not deploy.

Figure 6. Collision damage resulting in non-deployment of front passenger/driver air bags.

Case study: Air bag deployment

An example of an impact that did not result in a significant change in velocity, but yet the air bag

deployed, involved a 2007 Chevrolet Corvette that struck several small signs, trees, and a utility

pole off road at a very high rate of speed. As the vehicle struck the first object at over 60 mph,

the air bag control module reported a maximum V of 4.96 mph, which is well below the 8.0 to

14.0 mph range where deployment (for frontal impact with a solid barrier) is expected.

Fortunately, the maximum deceleration recorded at the same time during impact was 11.3 g’s,

which is above the estimated threshold range (6.1 to 10.6 g’s) shown in Table 3. As a result, the

air bags deployed and saved the lives of both the passenger and driver.

Figure 7. Collision damage resulting in air bag deployment.

SUMMARY

Air bags must have a control system that can recognize a crash correctly, and early enough for

air bags to inflate safely. Deployment systems generally use electronic sensors that continuously

report a vehicle’s acceleration to an air bag control module. The modules utilize complex

algorithms to make air bag deployment decisions based on one or more kinematic variables. Due

to the proprietary nature of air bag deployment algorithms, the velocity, acceleration, or

displacement thresholds for air bag deployment during a collision are not easily obtained.

Instead, a range of impact velocity, deceleration, or displacement threshold values can be

calculated (based on vehicle stiffness-to-weight ratios) and used to estimate when an air bag

should or should not deploy in a collision.

APPENDIX GLOSSARY OF TERMS

Acceleration: A vector quantity specifying the rate of change of velocity.

Accelerometer: A device which converts mechanical motion into an electrical signal

proportional to the acceleration value of the motion; it measures inertial
acceleration or gravitational force.

ACM: Air bag Control Module – the control module for air bags and related
restraint systems.

Algorithm: A series of steps designed to accomplish a specific task.

Algorithm enable: “AE” – A programmed threshold for a specific ACM at which the ACM
begins the deployment decision making algorithm.

Control Module: An electronic device that makes decisions and controls other devices.

Crash Pulse: The period of time defined by the moment when two vehicles come into
contact until the point where they separate at the centroid of damage and
the exchange of momentum between the vehicles ends. As “crash
duration,” it is defined by time.

Delta-V: The vector change of speed of a vehicle involved in an “event” described

by a magnitude and direction.

Deployment Acceleration observed along one of the car’s axes sufficient to cause the
(Event): control module’s crash sensing algorithm to “enable” or “wake up” and
which is sufficient to warrant a commanded deployment.

Enabled: When a threshold has been met satisfying one of the criteria necessary to
begin a process or deploy a device.

Event: The occurrence of some level of acceleration which causes an ACM to

evaluate available data and decide whether or not to deploy restraint
devices(s). A crash or other physical occurrence that causes a module’s
trigger threshold to be met or exceeded.

Frontal Air Bag: The primary inflatable occupant restraint device that is designed to
deploy in a frontal crash to protect the front seat occupants. It requires
no action by vehicle occupants and is used to meet the applicable frontal
crash protection requirements of FMVSS No. 208.
Jerk: The rate of change of acceleration.

Millisecond: A millisecond is 0.001 seconds.

Non Deployment Acceleration observed along one of the car’s axes sufficient to cause the
(event): module’s crash sensing algorithm to “enable” or “wake up” but which
does not warrant a commanded deployment.

Wake-up: A programmed threshold for a specific ACM at which the ACM begins
the deployment decision making algorithm. See also algorithm enable.
 REFERENCES

1. National Highway Traffic Safety Administration. What You Need to Know About Air Bags -

DOT HS 809 575. US Department of Transportation, 2003.

2. Chan, C. Fundamentals of Crash Sensing in Automotive Air Bag Systems. SAE

International, 2000.

3. Collision Safety Institute. Bosch Crash Data Retrieval System – Crash Data Retrieval Data

Analyst Course Manual. CSI, 2011.

4. Federal Motor Vehicle Safety Standards and Regulations. Occupant Crash Protection –

Standard No. 208. US Department of Transportation, 1998.

5. Hyde, Alvin S. Crash Injuries: How and Why They Happen. Hyde Associates, Inc., 1992.

6. National Highway Traffic Safety Administration. What You Need to Know About Air Bags -

DOT HS 809 575. US Department of Transportation, 2003.

7. National Highway Traffic Safety Administration. Effectiveness of Occupant Protection

Systems and Their Use - DOT HS 809 442. US Department of Transportation, 2001.

8. Nusholtz, GS. Famili, F. Di Domenico, L. Shi, Y. Aoun, ZB. Hongsakaphadana, Y. Air Bag

Effectiveness as a Function of Impact Speed. Traffic Injury Prevention, June, 2003; 4(2):

128-35.

9. Manko, Susan. Air bag and No Seatbelt a Bad Combination in Accidents. University of

Pittsburgh. Pitt Chronicle, July, 2004.

10. Rowell, John. Electronic Crash Sensing In Air Bag Litigation. Cheong, Denove, Rowell, &

Bennett. Web. June 2012.

11. Chan, Ching-Yao. Fundamentals of Crash Sensing in Automotive Air Bag Systems. Society

of Automotive Engineers, 2000.

12. Huang, Mathew. Vehicle Crash Mechanics. CRC Press, 2002.

13. Hussain, Aini et al. Decision Algorithm for Smart Air bag. International Journal of

Electrical and Computer Engineering 1:5 2006.

(Definitions taken from: Collision Safety Institute)

ABOUT THE AUTHORS

Mr. Jesse Kendall, P.E., obtained a Bachelor of Science degree in Civil Engineering from the

University of Vermont in Burlington, Vermont. He completed his engineering internship in

Denver, Colorado, working for civil engineering consulting firms before becoming a licensed

professional engineer in six states. With over fifteen years of civil engineering experience, Mr.

Kendall now lives and works in California for the Institute of Risk and Safety Analysis,

specializing in forensic engineering and accident reconstruction.

Dr. Solomon obtained a Bachelor of Science, Master of Science and Doctorate in Engineering, as

well as a Post-doctorate in Risk Benefit Assessment from UCLA. Dr. Solomon also holds a

Professional Engineering License. Dr. Solomon's studies are limited primarily to accident

reconstruction, biomechanics, and risk-benefit assessment as demonstrated by his 39 years of

independent research; his more than 200 internationally distributed publications, reports, and

presentations; his thirteen book co-authorship; and his journal guest editorships. In December of

1998 and after over 22 years of service, he retired as Senior Scientist with the RAND

Corporation. He was on the faculty at the RAND Graduate School for eighteen years, and has

taught as an Adjunct Faculty at UCLA, USC, Naval Post-Graduate School, and George Mason

University. Dr. Solomon has published studies in Transportation Accidents (automotive, trucks,

motorcycles, bicycles); Industrial & Recreational Accidents (pressure vessels, rotating

machinery, forklifts and cranes, exercise, gym, & recreational equipment, swimming pools,

manufacturing and punch presses); Slip- or Trip-and-Fall Accidents; and Adequacy of Warnings.