Congestion at traffic lights and emergencies
I have noted before the problems of setting the "stop"/"go" sequences at traffic lights. When the media quote the average time that a resident of X (X may be a city or country) spends in queues each year, invariably traffic queues feature as a source of delay.
So, the other day, I was stuck in a queue at one of Exeter's crossroads. It is one that I knew from my daily commute to the university, so I had measured its performance. Four roads are controlled by the lights. Let us call them North, South, East and West. The normal light sequence is:
A) Green for East and West, red otherwise
B) Green for South, red otherwise
C) Green for East and West, red otherwise
D) Green for North, red otherwise
E) All red to allow pedestrians to cross
Any of A)-D) may be omitted if the detectors show no traffic waiting and E) may be omitted if no pedestrians have pressed their signal. A)-D) are time-capped at 30 seconds, and are also capped if there is no detectable movement of cars (induction coils in the road surface). The latter appears to be a fail-safe mechanism, combining information from approaching traffic with information close to the crossroads.
What can go wrong? It looks like a good control scheme. And, generally, it is; it even detects my bicycle. But, there is one problem. The road I have called South is the main access route from the police station and ambulance station to the road East - which in turn is a major route to suburbs and out of the city. And when an emergency vehicle approaches the junction during phase B) above, the traffic stops to allow that vehicle priority and to overtake, the fail-safe mechanism detects no moving cars in a queue on the correct side of the road, concludes that there is no queue and switches to phase C). Result - longer delays for traffic on South. When I was stuck in the queue, I observed that normally about 16 cars get through on phase B) but the emergency meant only 2 did. Since the system is fairly well balanced, bringing the queue back to its normal state took several sequences of lights.
Nonetheless, the frustration did generate something - this blog!
So, the other day, I was stuck in a queue at one of Exeter's crossroads. It is one that I knew from my daily commute to the university, so I had measured its performance. Four roads are controlled by the lights. Let us call them North, South, East and West. The normal light sequence is:
A) Green for East and West, red otherwise
B) Green for South, red otherwise
C) Green for East and West, red otherwise
D) Green for North, red otherwise
E) All red to allow pedestrians to cross
Any of A)-D) may be omitted if the detectors show no traffic waiting and E) may be omitted if no pedestrians have pressed their signal. A)-D) are time-capped at 30 seconds, and are also capped if there is no detectable movement of cars (induction coils in the road surface). The latter appears to be a fail-safe mechanism, combining information from approaching traffic with information close to the crossroads.
What can go wrong? It looks like a good control scheme. And, generally, it is; it even detects my bicycle. But, there is one problem. The road I have called South is the main access route from the police station and ambulance station to the road East - which in turn is a major route to suburbs and out of the city. And when an emergency vehicle approaches the junction during phase B) above, the traffic stops to allow that vehicle priority and to overtake, the fail-safe mechanism detects no moving cars in a queue on the correct side of the road, concludes that there is no queue and switches to phase C). Result - longer delays for traffic on South. When I was stuck in the queue, I observed that normally about 16 cars get through on phase B) but the emergency meant only 2 did. Since the system is fairly well balanced, bringing the queue back to its normal state took several sequences of lights.
Nonetheless, the frustration did generate something - this blog!
Comments
Post a Comment