Improving LRT Signal Timing in Downtown Minneapolis

Almost every train stops at 3rd Avenue N., yet only one car per minute crosses the LRT here.

Almost every train stops at 3rd Avenue N., yet only one car per minute crosses the LRT here.

Although St. Paul’s traffic signals deserve most of the blame for slowing the Green Line, there’s room for improvement in Minneapolis as well. To be fair, Minneapolis deserves praise for its signal timing through the University of Minnesota campus, where trains are seldom delayed. They’ve also established a very good progression for westbound trains from the city limits to the Prospect Park Station, including the right turn from University Avenue onto 29th Avenue SE.

In downtown Minneapolis the results are mixed. As in downtown St. Paul, part of the problem is that intersections that previously split green time between two streets now have that time split three ways—the LRT, the cross street and auto turn movements across the LRT with the LRT stopped. With one-third less green time available, keeping the LRT moving is a greater challenge. This is the situation at 3rd Avenue N., 2nd Avenue N., Nicollet Mall, Marquette Avenue, Park Avenue and 4th Street-Chicago Avenue.

After examining the city’s own reports on traffic volumes (including intersection turn movements by hour of the day), I’ve concluded that there are opportunities to give more time to LRT in downtown without unreasonably disrupting auto traffic. The basic premise is: if auto movements don’t need all the green time allocated to them, that time should be reassigned to LRT. In some cases, the auto movements are small enough that their signal phases can even be preempted without causing undue delays.

The Green Line opening is the catalyst for making this proposal, because it has doubled the number of people traveling on 5th Street. LRT now carries many more people than the autos on most of the streets that cross the tracks. LRT also has to compete for green time with the small number of autos that share 5th Street. They currently get a third of the green time at several intersections, despite carrying a tenth of the LRT’s volume. For those reasons it’s time to take a fresh look at 5th Street signal timing.

Starting at Target Field Station, let’s take the intersections one at a time. To get an apples-to-apples comparison, each LRT passenger is given the same weight as an automobile.

3rd Avenue N.

3rd Avenue N. frequently stops the trains. The signal cycle is divided into three phases, for 5th Street and LRT, for 5th Street left turns across the tracks onto I-394 and for 3rd Avenue N. The left turn and 3rd Avenue N. movements are tiny compared to the LRT, or to any other intersection in downtown. Left turns across the tracks amount to only 5 per hour in the midday and 22 per hour in the PM peak. Traffic on 3rd Avenue N. is only 8 cars/hour in the AM peak, 31 in the midday and 69 in the PM peak. That may seem like a lot, but it’s only one car per minute. They’re holding up trains for one car per minute?

2nd Avenue N.

2nd Avenue N. is a three-way intersection. It’s busier than 3rd Avenue N. Auto and LRT volumes are about the same, but that’s only about 5400 cars and buses on 2nd Avenue compared to 3900 LRT passengers. Although the counts are about even, a review of the auto counts by time of days show only 3 cars per minute on 2nd Avenue in the AM peak, 2 per minute in the midday and 7 per minute in the PM peak. Those are spread over two traffic lanes, so it’s really 3.5 cars per lane per minute.

The other source of delay at this intersection is the 5th Street auto traffic, which crosses the tracks here. It amounts to only 2 cars per minute, except in the PM rush hour, when Garage B empties into a single lane on 5th Street. Then volumes reach 5 cars per minute.

1st Avenue N.

1st Avenue N. green time is only split two ways. 1st Avenue carries twice the traffic as 5th Street in a single lane, so there’s really no room for improvement.

Hennepin Avenue

The same goes for Hennepin, which is the busiest street the LRT crosses. Including 5200 daily bus passengers, Hennepin carries twice the volume of 5th Street. No change possible here. Thankfully, Hennepin has only two signal phases, so delays aren’t bad.

Nicollet Mall

Nicollet Mall is a 3-way signal. The Mall is bus-only and 9000 bus passengers cross 5th Street compared to about 10,700 LRT passengers. We don’t want to slow down the buses. However, there are only 2500 daily autos on 5th Street, 2-3 per minute, so that phase can definitely be squeezed.

Marquette Avenue

Marquette Avenue and 2nd Avenue S. are the paired contraflow bus lanes that carry an enormous volume of passengers on the rush hours, so they’re off limits for peak period cuts to the cross street. However, in the midday there are few buses and Marquette averages only 5 cars per minute divided between two lanes (2.5 cars/minutes/lane), so there’s room for more LRT green time. Marquette is a 3-way light, and the 1600 daily cars on 5th Street get more green time than they deserve (only 3-7 cars per minute), so there’s an opportunity.

2nd Avenue S

2nd Avenue S. is a 2-way light that seldom stops the trains. As with Marquette, rush hour buses need all the green time they currently receive, but off-peak hours offer an opportunity to help the LRT. Like Marquette, In the midday, 2nd Avenue S. sees only 5 cars per minute divided between two lanes.

3rd Avenue S.

3rd Avenue S. has 11,100 cars and no buses. That’s fairly busy (5-7 cars per minute per lane) and more LRT green time may be difficult to justify, even though 5th Street crossing 3rd Avenue S. is now up to 22,500 LRT passengers.

4th Avenue S.

The numbers for LRT versus autos improve greatly at 4th Avenue S. because the Government Center Station passengers have boarded. Now it’s 28,200 LRT passengers plus 2239 cars on 5th Street (total 30,400), against 8500 cars and 638 bus passengers on 4th Avenue for a total of 9138. These lopsided volumes argue for signal preemption.

5th Avenue S.

5th Avenue S. is somewhat busier, 9800 cars and 1455 bus passengers (total 11,255). Auto volumes vary from 7 cars/minute in the midday to 17/minute in the PM peak. Although that may seem high, there are three through traffic lanes on 5th Avenue, so each lane is handling about 6 cars/minute. It would not delay autos to reassign more green time to 5th Street, which has 28,200 LRT passengers and 2239 cars total 30,400. 5th Avenue is another 2-way signal, so there’s more green time to reallocate.

Portland Avenue

It’s even better at Portland Avenue, a 2-way signal that has only 5246 cars and no buses. That makes it almost as low as Victoria Street in St. Paul, where the city is reluctantly experimenting with full preemption (which is working well for LRT, it turns out). LRT combined with autos has almost 6 times the traffic volume.

Park Avenue

Park Avenue is a 3-way signal with 6700 cars and no buses, now that the 94 express has been rerouted. The Park Avenue phase has only 5-11 cars per minute, spread over three lanes, therefore lowering the per lane count to no more than 4 cars/minute, even in the PM rush hour. There are 4 LRT passengers for every car on Park. The right turns from 5th Street onto Park, which get their own phase, amount to only .5-1.5 cars per minute, a tiny volume that could certainly be preempted.

4th Street and Chicago Avenue

The situation at 4th Street and Chicago Avenue, where trains cross diagonally through the intersection, is somewhat more complicated. This has proven to be a real bottleneck for trains. When Blue and Green Line trains to Minneapolis converge at the junction at the same time, one has to wait for the other to clear. The first train then hits a red light at 4th/Chicago that can last two minutes. Once clear of the intersection, there’s the Downtown East Station stop, followed by the light at 5th and Park. This has turned into a cascade of delays for both lines.

Chicago has only 3300 cars (3-5 cars/minute) and will never be a busy street. Reducing Chicago’s green time, even preempting it, is probably the best way to reduce LRT waits.

4th Street has 6800 cars, along with 2083 bus passengers. 4th Street car volumes decreased dramatically when Washington Avenue was closed through the U of M campus. Bus volumes are also down sharply, because Routes 16 and 50 were replaced by the Green Line. Bus passengers on 4th Street declined from 5210 before the Green Line to 2083 now.

However, more express buses and car traffic may soon be shifted to 4th Street to access the new on-ramp to northbound I-35W under construction near Seven Corners. If so, bus passengers will rise by 1500 per day, all in the PM rush hour. If half of the car traffic to northbound I-35W is diverted from Washington Avenue, 4th Street’s car volume will increase by 5000 to 11,500 plus 3600 bus passengers, totalling 15,100. That is divided between two lanes. If my projection is correct, 4th Street will see 4 cars per lane per minute. That rate would permit more green time for LRT.

To summarize 4th & Chicago, I project that their combined auto and bus total will be 18,400 per day, assuming all possible buses are rerouted to the new on-ramp. However, LRT volume is 36,300, thanks to boardings at the Downtown East Station. That 2 to 1 ratio, plus the potential to reduce train delays, argues for more LRT green time.

Time of day variations

Most of the streets that cross the LRT are only busy in the rush hours, with the PM rush hour being noticeably heavier than the AM. Even where preemption cannot be justified in the rush hours, it should be coinsidered during the off-peak. I was recently at 5th and Marquette at 9 PM on a Saturday and watched a train wait at both 2nd Avenue S. and Marquette, even though there was no cross traffic at all. That’s unnecessary. The result will be much faster LRT travel times during the large majority of service hours, hopefully permitting a train to be removed from the schedule, a major savings.

Is priority counterproductive?

Priority can only shorten a red light or lengthen a green by a modest amount. I’m starting to suspect that priority can’t effectively move the LRT through downtown when the LRT green phases are too short. This is particularly true since the added Green Line trains are producing calls for priority every couple of minutes.

I watched the signals on 5th Street to get a better idea of how the LRT signal phases compared with those for autos at each intersection. Disclaimer–Not having access to the city’s actual timing programs, my conclusions are empirical and certainly not precise. Even so, I don’t think my conclusions are too far off.

At the 2-phase intersections (1st Avenue N., Hennepin, 2nd Avenue S., 3rd Avenue S., 4th Avenue S., 5th Avenue S. and Portland) the LRT “Go” signal was on whenever the light was green for 5th Street. That’s good as far as it goes, but out of a 110-second signal cycle, the LRT seemed to get about 30 seconds, or less than 30 percent of the green time. Giving the LRT half the 110-second cycle would double what they presently receive.

30 seconds was generous compared to the 3-phase intersections. The LRT would sit through the other two phases meant for autos, then have maybe 15 seconds of green time, less than 14 percent of the cycle. It appeared that the LRT phases were intentionally only long enough for the train to clear the intersection. If LRT simply received 1/3 of the time, it would more than double to 36 seconds.

It occurred to me that active priority with small LRT windows may be self-defeating. With so many trains moving in both directions, it’s not possible to thread that particular needle, unless the hole is made much larger. Longer fixed LRT green phases with no priority might produce fewer delays than at present.


LRT in downtown Minneapolis was never given green time proportional to the number of people it moves through an intersection.

The opening of the Green Line has greatly increased LRT’s share of the people passing through downtown intersections. At most intersections, LRT transports many more than do automobiles, yet it still receives a small minority of the green time.

At many intersections, green time is wasted on low volume auto phases that could be shortened or preempted with little or no harm to auto flow.

Signal priority with extremely short LRT green time (the current practice at 3-way intersections) is less effective at moving LRT’s than longer LRT green time without priority. Of course, both would be even better.

Aaron Isaacs

About Aaron Isaacs

Aaron retired in 2006 after 33 years as a planner and manager for Metro Transit, where he worked in route and schedule planning, operations, maintenance, transit facilities, light rail and traffic advantages for buses. He's an historian of transit, as a 40+ year volunteer with the Minnesota Streetcar Museum. He's co-author of Twin Cities by Trolley, The Streetcar Era in Minneapolis and St. Paul, and author of Twin Ports by Trolley on Duluth-Superior.

22 thoughts on “Improving LRT Signal Timing in Downtown Minneapolis

  1. Janne

    Signal timing downtown is something that I wonder about typically two or four times a day, as I bike both directions across it and spend most of my time sitting at red l ights wishing I weren’t waiting for no traffic to cross. Here’s a post I wrote last winter out of that frustration.

    This post gives me an excuse to ask some of my other questions. If doesn’t *feel* like signal timing is different off-peak times from on-peak times. QUESTION: Is timing different at different times? If so, why can’t we make it more responsive to actual traffic? If not, why not? And, couldn’t we just turn all those dang lights into flashing four-way stops every time there isn’t actually need for them?

    I’m also amazed at the astoundingly low traffic counts on many of those Downtown East streets. This summer I’ve been noticing how almost every street in the area has had one or even two lanes closed for construction, and I haven’t noticed any significant traffic delays. QUESTION: Why can’t we do something to right-size the streets down there? Portland, Park, Chicago, 4th, 3rd, they all have long-term closures (some all the way to the heart of downtown, ike 4th) and it’s working FINE.

    1. Aaron IsaacsAaron Isaacs Post author

      Way back before computers, the downtown rush hours saw all the lights turn green at once on the parallel streets throughout downtown. You could make 3, sometimes 4 blocks, before they all turned red at once. I was a cab driver in college and if you zig-zagged 3 blocks at a time, you could cross downtown without ever stopping.

      During the off-peak, all the streets had the same progression, about 21 mph. That worked well for the one-way streets, but on the 2-ways (Hennepin, 3rd Ave. S.), it worked in one direction but going the other way youhit a red light every block.

      I’m not convinced that the current system does a better job of moving traffic.

      1. Janne

        Aaron, does that mean you can’t answer even the first question? Can anyone answer? It’s not intended to be rhetorical.

        1. Students_TT

          Janne, signal timing is changed during different times of day, and can sometimes be very drastic. Aside from differences in cycle length, different plans at different times of day also have different patterns of progression and offsets for different directions (Inward at the AM, outward in the PM). I think one of the biggest issues with developing the time of day plan is that just like with turning movement counts, the results are only from one day, so the day the video was taken could be a busy day or a dead day. It’s becoming more possible to get a better shot of data by using services like the count system site, but even then results are from a sporadic time frame that could be hard to get a standard from. I saw some data as old as 2008, and traffic patterns can definitely change in that time. However, even some of the more recent volumes could be used to get a better average for the model. The big thing is that even with these volumes being available, it can be hard to try change the way that data collection is done. However I expect as technology becomes better (and a younger, potentially more innovative workforce starts to have a presence) some of these it will start to lean to the side of big data and large sample pools for the best overall result.

    2. Nick Musachio

      In answer to your question. Yes and yes. Traffic signals are controlled by pc computers in the yellow boxes you see next to every intersection. These computers are programmed, and have differing schedules depending on time of day and even day of the week. Dealers choice. They can certainly be programmed to flashing 4 way reds if desired at certain hours. It’s ridiculous to wait a minute for a red light at 4 AM when there is no traffic. But hey, Minneapolis.

      Here’s my solution to the whole thing. It would work for bicycles too. Always Green Traffic Control. Sadly, it’s hard to even talk to traffic control in Minneapolis to make suggestions. They don’t return my phone calls or emails. Have you tried?

  2. David MarkleDavid Markle

    It’s good, and potentially useful, that Aaron Isaacs devotes careful thought and professional insight to this issue. I agree that the system can probably be improved, based on continual monitoring of traffic, passenger loads, etc. Recently I waited at the West Bank Green Line station for an eastbound train for an unusually long period during which five westbound trains departed. At various times of day I’ve also seen 3-car trains carrying much less than a 1-car load.

    What I find most frustrating, however, is the fact that the tracks were laid on top of the street rather than many feet beneath. Such atrocious, short-sighted planning!

    1. Sean Hayford OlearySean Hayford Oleary

      I have been pretty amazed by the efficiency of the one-way lights in downtown. Once every week or two, I’ll drive a car into downtown, and I can drive from TH 65/35W stub to Washington Avenue without hitting a single red. Or from the same stub to Target Center maybe hitting one or two lights.

      That’s much better than even more overtly auto-oriented places. On France Avenue through the Southdale District, I seem to hit 9 of 10 red lights, whether on bike or in car.

      1. Adam MillerAdam Miller

        I lost badly from government center plaza to Target Field, but primarily because the train was able to make it out of the government center stop while I had to stop for the light at 3rd Ave.

        I wonder if the driver didn’t run a signal, though, as he pulled into the intersection with about two seconds remaining on the counter and was still in it when the signal for cars on 3rd turned green.

  3. Adam FroehligAdam Froehlig

    Aaron, I have to ask how you came up with your “cars per minute” caluclations? Were they based on actual hourly calculations, or did you go from the daily count? The danger with doing the latter is that it’s highly skewed towards peak periods and the workday, so doing a simple cars-per-day-divided-by-1440-minutes-per-day results in a potentially high error.

    1. Joseph TottenJoseph Totten

      I’d actually prefer it if you could find a peak 15 or 5 minute count and extrapolate from there vs a whole hour, unsure if this actually happened or not, but there are peaks within peaks within peaks…

    2. Aaron IsaacsAaron Isaacs Post author

      The city’s website counts the movements within each intersection in (if I recall correctly) 15 minute increments for these 2-hour periods: 7-9am, 11am-1pm and 4-6 pm. I took the total count for each 2-hour period and divided it by 120 minutes. There was probably some peaking within those periods, but I doubt that there was enough to materially change my analysis.

  4. Walker Angellwalker

    Aaron, great post. Three questions:

    – What would happen if we used shorter cycle times. Like maybe 40-70 seconds? There would theoretically be more wasted clearance time, but perhaps made up for with much less wasted unused green time?

    – What would the likely outcome be if we changed our priorities to give rail (LRT, Tram, etc.) and Pedestrians/disabled/bicycle top priority, buses second, and then individual cars?

    – Why can our signals not do a better job dynamically responding to demand instead of being locked in to rigid cycles/phases? For instance, numerous times I’ve sat in the turn lane on westbound University to turn south across the tracks on to Western. Zero trains will come through on it’s green (vertical bar), only a small handful of cars will come through eastbound while I’m waiting, same for north-south. Twice when I’ve been given a green turn, trains had to stop for me.

    1. Aaron IsaacsAaron Isaacs Post author

      One of my friends at Metro Transit has observed that Chicago uses a shorter signal cycle and he feels it moves traffic better, or at least reduces waiting times before the next green. For those intersections where crossing auto traffic doesn’t need all the current allotted green time, that would seem to make sense.

      There’s actually a potential conflict between full LRT preemption and pedestrian crossing, since the latter requires a minimum green time so peds aren’t caught in the middle of the street. That said, detecting the train early enough should be able to prevent an unsafe situation for peds.

      Not being a traffic engineer, I can’t comment on the technical flexibility of signal systems, except to say that they’re getting smarter all the time.

  5. Nick M

    Isn’t there a way they could install a sensor on the tracks to flip the light in the LRTs favor, then flip it off once it gets past a safe point? This doesn’t seem like it is advanced technology and would maximize the speed of the LRT while minimizing the impact on the rest of the light cycle.

    As someone who rides a bike to work on Nicolett ave and walks down Nicolett frequently, crossing 5th street is incredibly aggravating. I don’t have the experience with the other intersections that I do with that one but theres no way the light timing at that foot traffic on Nicolett into proper account. During the morning/evening rush you could probably write several thousand jaywalking tickets a day because there are maybe 20-25 seconds in the 3 or 4 minute light cycle where you are allowed to walk across 5th street. Part of the light cycle needs to be giving people getting off the trains walk signals. Are they supposed to wait on the platform for several minutes until they have a walk signal? I don’t understand what the thought was there. It seems like (mind you from casual observation, not scientific timing) that the light cycle for the odd car or two coming down 5th st. is significantly longer than that for the streams of pedestrians getting off the light rail. It seems like after a train enters the station there should be a period where all vehicular traffic has a red light and pedestrian traffic has walk signals in every direction.

  6. Walker AngellWalker Angell

    Also, at what point do you determine that you simply have too much of one mode of traffic and need to reduce it rather than try to manage an impossible situation? We’re not even remotely as bad as LA, NYC, WDC, or London (though seemingly worse than Amsterdam, Frankfurt, and Paris) so we’re far from in a bad situation. Still…

    When do you begin considering congestion charges? How would congestion charges compare to prioritizing preferred modes over less preferred?

  7. Benjamin Riggs

    I’d think shorter cycle times (to a minimum of pedestrian crossing time) combined with signal priority would be a small step which would improve train travel time, while impacting automobile wait time (which, let’s be honest here, is the reason stoplights exist in the first place) only mildly.

    A far better plan, to my mind, is full preemption of stoplights between rail stations for as long as it takes the train to go from one to the next (possibly with some minimum interval of preemption so that auto queues would have a chance to dissipate). This could nearly eliminate the need for a cycle in these stoplights entirely, probably leading to better traffic flow overall, despite the occasional increased traffic queue. Whether the technology in place can handle this event without too much overhead of wasted time is a problem for city engineers.

  8. Nick Musachio

    The elephant in the room is congestion and emissions. Preemption = more emissions and congestion. There is another solution to the win/lose proposition presented in the article: Always Green Traffic Control (AGTC). The fact of the matter is that there already is plenty of green light time for the Green Line. It’s just that the trains aren’t coordinated to the intersections. AGTC can coordinate the trains to the existing green lights. AGTC can also coordinate vehicles to green lights. There is plenty of green time for both to get through without waiting.

    Additionally, some very large chunks of time still won’t be addressed by the suggested solution. This, because when the Green Line crosses arterials such as Snelling or Lexington, the trains often stop when the light is still green; thus wasting up to one minute and 45 seconds. AGTC should be considered to control both the trains and the vehicles at these intersections.

    Here are two videos presenting a win/win solution to the problem:

  9. Mary G

    Thank you, thank you for mentioning 3rd Avenue North. Whether riding the train, the bus or in my car, it seems like there are huge amounts of time when no one is going anywhere – especially the train. So I guess it’s the left turners? I don’t know because I never see anyone doing that. I only see all of us – cars, buses and trains – sitting looking at each other.

Comments are closed.