Hiding in the Gap

The often pushed safety campaign of how much gap should be left between cars on the road seems to vary wildly just between the states of Australia, with VicRoads (the vehicle licencing and registration body of Victoria) currently suggesting “trying” to have a 2 second gap in optimum conditions and 4 seconds when any of the following conditions exist:

  • at night or in poor light
  • bad weather conditions such as rain or fog
  • when you are tired
  • when your vehicle is heavily laden and can’t stop as quickly
  • when you are towing
  • when you are unsure of the road

So it is then interesting to consider the simulated throughput under several scenarios, starting with the “default” assumptions still being pushed in design standards and being achieved in real world situations that an unobstructed road will first achieve unstable flow beyond 2000 vehicles/lane/hour.

car throughput vs gap

It was only possible to obtain such flow rates with an average vehicle to vehicle gap of 1 second. Further, when compared to the 2 or 3 second average gap scenarios the driver model required both faster responses (attention) and observation of additional characteristics of the surrounding vehicles to maintain the same low level of accident occurrence. Driving in these high throughput conditions demands additional focus from the driver, requiring more energy and increasing stress compared to leaving a longer gap. Sadly choosing to relax does not work in a conditioned population such as Melbourne, as other drivers will simply fill such a comfortable gap.

The reductions in throughput with increasing vehicle to vehicle gap should partly explain the consistently bad traffic congestion during wet periods (which is not mirrored on trains), the slight increase of vehicles on the road from people displaced from walking or cycling and the existing density of vehicles pushed down towards collapse as drivers leave longer stopping distances. Thankfully as the vehicle to vehicle gap increases the discontinuity between unstable and stop-start flow disappears and although the flow is greatly reduced the vehicles no longer have to rapidly accelerate and decelerate as stop-start behaviour is delayed or suppressed entirely.


Here the real world example of a stop-start flow developing on an unobstructed and otherwise free flowing highway. Varying the spread of the drivers target vehicle to vehicle gap showed little to no change in the overall behaviour, similarly changing the spread of the drivers target speeds did not change the average speed or throughput characteristics until stop-start conditions were reached.

For reference theoretical self driving cars (robots) were added into the above plot to illustrate this, having faster reaction times, more accurate inputs than the human drivers, tightly grouped target speed, and 2 second vehicle to vehicle gaps. They follow the 1-3 second gap group until stop-start flow causes the human model to lose throughput, while the more attentive robots are able to descend smoothly all the way to stopped. Although having a vehicle to vehicle gap of twice the contemporary driver model the self driving car is able to perform equal or better except in the region approaching the unstable flow regime. Robots with car to car communication could conceivably do even better by forming train like groups operating with reduced vehicle to vehicle gaps and still operate within current road rules, though one can only imagine the public outrage the first time any accident were to occur despite the lack of culpability human drivers appear to have.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s