We are often asked to explain why the airbags did or didn’t deploy in a particular collision. Many people have a misconception that airbags will or should “go off” if the vehicle was travelling faster than a certain speed. But travel speed is not a key factor for airbag deployment. Whether or not to deploy an airbag is an extremely complex decision that the vehicle’s “safety brain” has to make in less than a fraction of a second. This article explores some of the factors that go into that decision.

Most car-to-car collisions last for about 70 to 150 milliseconds.1,2  Even though that can be less than the duration of a human eye blink, a lot happens during that time. Crash Reconstructionists and Engineers often break down a crash into three distinct stages: initial contact when the vehicles touch, maximum engagement when the maximum deformation and intrusion occurs, separation as the vehicles depart from each other. After the third stage, the vehicles travel to their final rest positions.

Figure 1: Collision stages – 1) Initial Contact, 2) Maximum Engagement, and 3). Separation, followed by Final Rest. These video still frames come from the drone footage of one of our 2016 crash tests in which the 3 crash stages lasted 130 ms.

Figure 1: Collision stages – 1) Initial Contact, 2) Maximum Engagement, and 3). Separation, followed by Final Rest. These video still frames come from the drone footage of one of our 2016 crash tests in which the 3 crash stages lasted 130 ms.

The airbag deployment decision has to be made well before maximum engagement occurs. This is because as plastic and metal are getting broken and crushed, occupants inside will move towards the area of impact (go to article), and the appropriate airbags need to fully deploy before occupants strike any portion of the vehicle interior or reach the limit of the seat belt slack. Also, to avoid an airbag-induced injury, the airbag must be fully inflated before the occupant moves too close to it. The air bag itself also takes some time to fully deploy. So, generally, the deployment decision is made in only 15 to 50 ms for frontal collisions.3 That is less than one-third of the entire crash duration, and it is remarkably quick!

So how is this decision made so quickly? As illustrated below, the vehicle’s airbag control module (ACM; also known as the EDR or the “black box”) has built-in accelerometers and is connected to other sensors around the vehicle so that it can constantly monitor rates of change in motion (see Figure 2 below). If you want to really know how the black box is programmed to make a decision to deploy airbags, you’ll have to endure a bit of physics and calculus, and wrap your head around orders of motion. [Physics/calculus give me uncomfortable flashbacks, can I just skip this section? ]


Figure 2: A simplified illustration of the Airbag Control Module (ACM) and vehicle sensors.

The first order of motion is displacement or the distance an object moves. For example, 10 metres.

The second order of motion is the rate of change of displacement, or more simply, speed – for example, we calculate speed by looking at the distance an object moved with respect to time – 10 metres per second, or 36 km/h. Pretty straightforward so far…

The third order of motion, which we commonly refer to as acceleration (or deceleration for negative values), is the rate of change of speed. For example, if you slowed from 56 km/h (typical barrier crash test speed) to 0 km/h by hitting a wall, which happens over a time period of 80 milliseconds, the average deceleration (rate of change of speed) would be 194 m/s2 (approximately 20 g). But braking lightly to slow down from 56 km/h to 0 km/h as you reach a red traffic signal, over the course of 8 seconds, would have an average deceleration of only 2 m/s2 (only 0.2 g)

The next and 4th order of movement is called jerk. This jerk is not the driver who cut you off the other day, it is the rate of change of acceleration. It is calculated by taking the derivative of acceleration with respect to time, or the second derivative of velocity, or the third derivative of displacement.


Why did we put you through a physics/calculus lesson on rates of change of other rates of change? Because the airbag deployment decision depends upon acceleration and jerk [4]. At the moment of initial contact, as illustrated in Figure 1A above, the sensors around the vehicle and the ACM “safety brain” sense the accelerations and estimate the jerk by taking continuous samples at a high frequency in order to detect if the accelerations and jerk rates are consecutively high enough. The ACM then must make a very quick decision on whether or not to deploy the airbags and/or seatbelt pre-tensioners based on pre-programmed algorithms and thresholds.

Why not just use acceleration or speed change in the deployment decision? After all, airbag deployment has been correlated [5] to the speed change and overall acceleration. The problem is that the forces and accelerations are the highest at the maximum engagement stage of a crash, but the airbag deployment decision has to be made very quickly – before maximum engagement. To make things a little more complicated, the crash pulse shape and maximum engagement time varies from crash to crash. The crash test example we’ve been using illustrates this well – the graph below (Figure 3) compares the crash stages with the speed, acceleration and jerk experienced by the vehicle over the course of the crash.

Figure 3: Collision stages and motion data over time

By examining the illustration above, you will notice that acceleration (blue line) does not peak until the crash is nearly at maximum engagement (about 65 milliseconds after initial contact) – and as we already discussed, the deployment decision needs to be made before that. However, at the time of deployment, the speed change may not have even reached 10 km/h, but the vehicle is experiencing an acceleration of about 14 g and a jerk of almost 3000 g/s at only 20 milliseconds after initial contact. Yikes! So, the deployment decision is based on an algorithm that must predict how severe the collision may be based on factors, such as the rate of change of acceleration (jerk), sensed and recorded as the collision is developing. An algorithm is a computer code of a mathematical function, which in this case happens to be a potential injury- or death-mitigating decision of the mechanical function of safety equipment, including the seatbelt pretensioners and air bags. The deployment decision for this vehicle was made within 15 milliseconds of initial contact.

The example used here is from an offset, head-on frontal collision, but crash pulses of acceleration and jerk can look very different for different types of crashes, vehicles, and impact orientations. The deployment decision algorithm analyzes acceleration and jerk along the longitudinal and lateral axes. If the direction of the impact is coming in on another angle, the longitudinal or lateral components need to be of a high enough magnitude to warrant either a frontal or side airbag deployment (or both). If a vehicle strikes or glances a relatively soft, deformable, surface such as the side of another vehicle, the airbags may not deploy because the collision developed over a longer length of time. If a vehicle under-rides a higher vehicle, the collision again develops over a greater length of time and the air bags may not deploy. Frontal air bag deployments may also not occur during roll-over incidents, due to the time involved and varying directions of accelerations involved. In rollovers, the decision to deploy roll curtains may take slightly longer (if the vehicle is involved in sufficient accelerations to drop them). Side air bags typically deploy even more quickly than frontal air bags, because there is less crush zone and space between the occupant and the area of impact. The airbag control module also monitors several other sensors to make the deployment decision, including seatbelt use, seat position, occupant size, and weight classifications in order to optimize occupant safety.

The airbag deployment algorithm of each vehicle manufacturer is relatively complex, varied, and proprietary. Nevertheless, we do know that the factors discussed above – acceleration, jerk, and the seatbelt and state of occupant sensors – are all considered in the deployment decision. There is now a plethora of data from modern vehicles with ever increasing sources of information from sensors. With this rich data, a better understanding of the crash can be preserved to better assist the crash reconstruction expert’s analysis of the circumstances surrounding a collision in further detail, and comment on safety equipment use, function, and potential biomechanical benefit.


  1. Agaram, Venkatesh et al., “Comparison of Frontal Crashes in Terms of Average Acceleration”, Society of Automotive Engineers, Paper number 2000-01-0880.
  2. Warner, Charles Y. et al., “Pulse Shape and Duration in Frontal Crashes”, Society of Automotive Engineers, Paper number 2007-01-0724.
  3. Crash Data Specialists, “CDR Data Analyst and Applications Class”, 2015.
  4. In the special case of rollovers, angle of roll and rate of roll as reported by a rollover sensor are also considered.
  5. Wood, M., Earnhart, N., and Kennett, K., “Airbag Deployment Thresholds from Analysis of the NASS EDR Database,” SAE Int. J. Passeng.Cars – Electron.Electr. Syst. 7(1):2014, doi:10.4271/2014-01-0496.

Share this article:

Share This
Your browser is out-of-date!

Update your browser to view this website correctly. Update my browser now