Is There an Ideal Sensor Location?

Are there “perfect” sensor locations? For example, when we deploy a queue warning system, are there sensor locations that will get us better data? Could that data inform us of slowing traffic sooner? Or could it be a better indication of traffic conditions than data from another location would be?

For end-of-queue warning systems we submit that the ideal sensor location is just upstream of where queuing is most likely to begin and, therefore where average speeds vary most.

There are locations like that throughout the work zone. Narrowing of lanes, lane shifts, temporary concrete barrier, bridge falsework and other construction activities affect drivers sense of safety. Anything that negatively affects that feeling of comfort will reduce the 85th percentile speed.

Short on-ramps with reduced merge distance have the same affect. However, if traffic always quickly accommodates those merges and returns to the previous 85th percentile speed, then that is not a perfect location. Only when the geometry in combination with traffic volume results in dynamic queuing does that become a good sensor location for queue warning systems.

The power source is our greatest limiting factor today.  Batteries, solar systems, etc. take up space. They must be located where we can reach them easily for maintenance. For this reason, many sensors are located on message signs and arrow boards where they can draw power from them and even share communications devices.

Arrow boards are placed at the taper. Queuing begins there, of course. But we will only catch speed variance due to conflicts at that merge point. We won’t see if that variance continues upstream.

Message signs are placed in advance of the work to warn of slowing downstream. We should always place one sensor at a point that queuing would reach as a result of a worst-case scenario. And a message sign location may be able to serve both purposes. But we normally want the sensors located where queuing begins and we want the message signs located upstream to warn of that slowing – not located together. If sensors and message signs share the same locations they are likely either too close to the work zone or too far from the source of the queuing to warn traffic before they reach the problem area.

We generally space sensors out every half mile to a mile apart with the understanding that we will learn about any queuing quickly. And that is a good approach. After all, we can’t predict all causes of queuing. But couldn’t we adjust those locations a little one way or the other to catch these obvious causes of slowing a little earlier?

It would be helpful to see research into sensor location. But in the meantime, let’s evaluate our work zones and adjust our sensor locations to monitor the more obvious sources of slowing. Our systems will perform better and improve work zone safety even more than they do today.