Traffic

A road diet isn't taking away a travel lane; it's moving the turning cars out of your way

The fear is gridlock. NYSDOT's traffic model says three lanes handle today's traffic as well as four or better. We built a simple model of our own to show how that's possible, when a road diet works better and when it works worse.

Watch it build, from light traffic to heavy

One block of Boston Post Road, 1,000 feet between two signals. Four lanes on top, the road diet below. Press Next to step up the traffic, from a quiet hour to more than the corridor ever sees. Each step replays the same three minutes of simulated traffic every time. It's a toy simulation, meant to give some intuition about the trade-offs of a road diet. It is not a comprehensive traffic simulation.

400 vehicles an hour

Moderate traffic: no real difference

This is moderate traffic, about what several intersections carry at the evening peak, including Larchmont Avenue and Beach Avenue. A driver waiting to turn left on four lanes blocks a lane, but the next lane is nearly empty, so drivers behind just go around. The two roads work about the same.

Four lanes
Road diet
Average trip through the block
23.1sec
22.8sec
The slowest trip in 20
28.0sec
25.1sec
Drivers held up behind a left turn
3%
None

Averages over twelve simulated hours at this volume, not the three minutes shown.

550 vehicles an hour

The road diet's benefits start to show

On four lanes, drivers start getting stuck behind someone waiting to turn left (shown in red). On the road diet, the turning driver waits in the center lane and nobody behind has to stop. Mamaroneck Avenue and North Barry Avenue are around this level.

Four lanes
Road diet
Average trip through the block
23.7sec
23.4sec
The slowest trip in 20
33.0sec
25.6sec
Drivers held up behind a left turn
6%
None

Averages over twelve simulated hours at this volume, not the three minutes shown.

700 vehicles an hour

The left-turn penalty on four lanes grows

With this much traffic, the next lane is full. If you're stuck behind a left-turning driver on four lanes, you're really stuck: watch for the car held up for about 20 seconds. On the road diet, nobody is. This is what the busiest intersections carry, like Old Post Road and Richbell, and Orienta Avenue.

Four lanes
Road diet
Average trip through the block
25.2sec
24.2sec
The slowest trip in 20
38.3sec
26.3sec
Drivers held up behind a left turn
11%
None

Averages over twelve simulated hours at this volume, not the three minutes shown.

850 vehicles an hour

The benefit peaks, and the road diet hits its limit

The left-turn benefit is at its largest here: on four lanes almost one driver in five spends part of the trip stuck behind somebody waiting to turn, and on the road diet nobody does. But the single through lane is close to saturated.

Four lanes
Road diet
Average trip through the block
27.2sec
24.8sec
The slowest trip in 20
44.2sec
26.7sec
Drivers held up behind a left turn
17%
None

Averages over twelve simulated hours at this volume, not the three minutes shown. The trips through the block are still quick on the road diet; the cost shows up in the line waiting to get in.

Watch the count of cars waiting at the lights: sometimes the light turns red before everyone gets in, and those drivers sit through another red. Over a full hour the road diet's line runs more than twice as long as the four-lane road's, though it still clears.

This is the situation at Delancey Avenue, the busiest spot on the corridor, which is part of why NYSDOT is proposing a separate fix there.

1,000 vehicles an hour

It doesn't work anymore

One lane can't carry this much traffic. Every signal cycle leaves more cars behind, so the line waiting to get into the block keeps growing. By the end of an hour it's about 100 cars long on the road diet and about 5 on four lanes.

No intersection on the corridor carries this much.

Play the interactive simulation

It's about reliability, not speed

At 700 cars an hour, the average trip is about the same either way. The difference is the unlucky trip.

On four lanes, about one driver in nine gets stuck behind somebody waiting to turn, and the slowest trips run far longer than the average. On three lanes, nobody gets stuck, and every trip takes about the same time.

That predictability is also the safety story. The sudden stops and lane changes that slow the unlucky driver are the same moves that cause rear-end crashes and sideswipes.

Four lanes (today)
Three lanes
Average trip through the block
25.2sec
24.2sec
The slowest trip in 20
38.3sec
26.3sec
Drivers held up behind a left-turner
1 in 9
None

Twelve simulated hours averaged, 10% of drivers turning left, 90-second signal cycle.

Every signal on the corridor

The busiest direction at each signalized intersection at the evening peak, against how much a three-lane road can carry in our simulation. That limit depends on how much green time the Post Road gets at each signal, which we don't know. So both are shown.

    Through vehicles per hour, one direction, at the evening peak. ≈900: the three-lane limit if the Post Road gets 60% of the green at each signal. ≈754: the limit at 50%. The 60% limit comes out about the same with a 60- or 90-second signal cycle; the 50% limit was simulated with a 60-second cycle. Orienta and Delancey, the intersection NYSDOT flagged, are in bold.

    If the Post Road gets 60% of the green, every intersection on the corridor sits under the limit. The closest is Delancey Avenue, the busiest movement at 831 cars an hour. At 50%, four would be over: the high school driveway, Fenimore, Alden and Delancey.

    We don't know the real signal timing. NYSDOT does, and its model found trouble only at Orienta and Delancey. Both analyses point to Delancey as the tightest spot on the corridor, and NYSDOT is already proposing a fix there.

    What our simulation doesn't show

    The simulation illustrates how a center turn lane works. It is not a traffic study, and it isn't calibrated to observed counts. It leaves out pedestrians, parking, buses, trucks and side-street traffic, and it models one idealized block, not the whole corridor. NYSDOT's model is the one to rely on for the real street.

    Some results don't favor the road diet, and we're including them:

    • At 1,000 cars an hour each way, the three-lane road breaks down and the four-lane road doesn't. Nothing on the corridor comes close to that volume.
    • With only 5% of drivers turning left instead of 10%, the four-lane road is faster on average.
    • A shorter signal cycle helps the four-lane road. It's a real partial fix, but it doesn't address the crashes.

    The detailed edition has every sensitivity test, and the write-up has the method.

    What NYSDOT found

    NYSDOT counted turning movements at every signalized intersection on the corridor and modeled the road both ways. According to NYSDOT staff, the three-lane design keeps or improves traffic flow at every intersection but one.

    The exception is Orienta and Delancey, which already has a traffic problem. That's why a roundabout is proposed there.

    That's NYSDOT's model, not ours. NYSDOT engineers are conservative, and the agency has every reason not to propose a design that jams a state highway.

    Sources
    1. NYSDOT traffic analysis, US-1 Complete Streets
      Turn counts at every signalized intersection and a corridor model, presented at the 2025 and 2026 open houses and described to us by NYSDOT staff.
      NYSDOT project documents
    2. "Stuck behind a left turn," a traffic microsimulation
      Car-following and lane-changing models, 12 simulated hours per case. Built by a Safe Routes to School volunteer; the model code is public.
      The simulation · Source code
    3. NYSDOT turning-movement counts, evening peak
      Busiest direction of through traffic at each of the 17 signalized intersections.