Bus Stop Location: Near-side or Far-side

[Richard Eade]

Originally Posted by bradnixon

. . . I'm not sure why far-side stops are "anti passenger" but they have been proven to reduce transit trip times.

Studies, like the one that the referenced article is based on, really need to be understood for what they are. This one appears to be an exercise in statistical analysis. It does not seem to be a meaningful study of where to, generally, put bus stops – although in its Conclusion, there are wild statements about fantastic savings for transit companies.

The study, “The Far Side Story” (Download the PDF), involves calculating the length of time buses spent stopped at Near-side, Far-side, and Mid-block bus stops in Montreal and Portland. Various factors were taken into consideration, including the number of doors on the bus, whether there was a right-turn lane before the intersection, the bus load, etc. However, the fundamental point of the study was to compare the length of time that a bus stopped at a near-side verses a far-side bus stop at a stop level. That is, records were used that told how long a bus was stopped at each type of stop, and the statistics provided the results.

According to the study, a near-side stop will delay a bus by an average of 4.2 seconds in Portland, and 5 seconds in Montreal, more than a far-side stop in those locations. Part of the extra time in Montreal could be attributed to the fact that right turns are not allowed on red lights there. This means that buses in Montreal can get stuck behind vehicle waiting longer to turn right. In these cases, Montreal bus operators will often open the doors before reaching the near-side stop and customers walk down to where the bus is to board it; that walk takes some extra time, lengthening the boarding time, which is included in the statistic.

It all sounds great for a stunning conclusion: If the bus company can save 5 seconds for each of 23 stops (the average number of near-side stops on the bus routes examined), then over the entire day, a particular route could have up to 9 extra runs! Sensational.

Except, nowhere in the study did they take into account the most fundamental reason for a stop lasting longer before the bus crosses a signal-controlled intersection – that a red light was involved. Remember, the measurement was from the time the bus stopped at a stop until it left that stop. It makes perfect sense that at a far-side stop the bus arrives, stops, and can leave immediately, with no extra delay. At a near-side stop, a bus arrives, stops and then has to wait for a green-light to leave. According to the study, apparently at a Stop-sign-controlled intersection, there is much less of a delay than at one controlled by lights. Well, imagine that. It sounds as if the only thing delaying a bus from leaving a near-side stop, in that case, is any vehicle that needs to clear the intersection.

And this leads us to the flaw in the conclusion stated by this study; that there can be time saved by moving the stops to after the intersections. The flaw arises from the comparison of the times at a stop level, based solely on the records of when the bus arrived at the stop and when it left that stop. The time that a bus waits at the intersection BEFORE it stops at a far-side stop is not included in the time for that stop. That delay can never over-lap the time for the bus stop; as it can at a near-side stop.

To be a fair, apples-to-apples, comparison to determine which is better, a study should look at the following time periods:

• For a near-side stop – the time from when the bus arrives at the stop (or near-enough for people to quickly walk to it and board/alight while the bus sits in traffic) until the bus crosses the far-side of the intersection;
• For a far-side stop – the time from when the bus stops for the last time before crossing the intersection (in case it takes multiple light cycles to cross), or from when the bus crosses the near-side edge of the intersection (if it does not stop), until the bus leaves the far-side bus stop.

This study, which is often reference as proving that far-side stops save time over near-side stops, does not prove anything of the sort. All it does is show that, if you include the random timing of a signal light into the time of stops at one location, then it will, generally, take longer than if no such additional time is included for stops at another position.

[lrt's friend]

Originally Posted by Richard Eade

Studies, like the one that the referenced article is based on, really need to be understood for what they are. This one appears to be an exercise in statistical analysis. It does not seem to be a meaningful study of where to, generally, put bus stops – although in its Conclusion, there are wild statements about fantastic savings for transit companies.

The study, “The Far Side Story” (Download the PDF), involves calculating the length of time buses spent stopped at Near-side, Far-side, and Mid-block bus stops in Montreal and Portland. Various factors were taken into consideration, including the number of doors on the bus, whether there was a right-turn lane before the intersection, the bus load, etc. However, the fundamental point of the study was to compare the length of time that a bus stopped at a near-side verses a far-side bus stop at a stop level. That is, records were used that told how long a bus was stopped at each type of stop, and the statistics provided the results.

According to the study, a near-side stop will delay a bus by an average of 4.2 seconds in Portland, and 5 seconds in Montreal, more than a far-side stop in those locations. Part of the extra time in Montreal could be attributed to the fact that right turns are not allowed on red lights there. This means that buses in Montreal can get stuck behind vehicle waiting longer to turn right. In these cases, Montreal bus operators will often open the doors before reaching the near-side stop and customers walk down to where the bus is to board it; that walk takes some extra time, lengthening the boarding time, which is included in the statistic.

It all sounds great for a stunning conclusion: If the bus company can save 5 seconds for each of 23 stops (the average number of near-side stops on the bus routes examined), then over the entire day, a particular route could have up to 9 extra runs! Sensational.

Except, nowhere in the study did they take into account the most fundamental reason for a stop lasting longer before the bus crosses a signal-controlled intersection – that a red light was involved. Remember, the measurement was from the time the bus stopped at a stop until it left that stop. It makes perfect sense that at a far-side stop the bus arrives, stops, and can leave immediately, with no extra delay. At a near-side stop, a bus arrives, stops and then has to wait for a green-light to leave. According to the study, apparently at a Stop-sign-controlled intersection, there is much less of a delay than at one controlled by lights. Well, imagine that. It sounds as if the only thing delaying a bus from leaving a near-side stop, in that case, is any vehicle that needs to clear the intersection.

And this leads us to the flaw in the conclusion stated by this study; that there can be time saved by moving the stops to after the intersections. The flaw arises from the comparison of the times at a stop level, based solely on the records of when the bus arrived at the stop and when it left that stop. The time that a bus waits at the intersection BEFORE it stops at a far-side stop is not included in the time for that stop. That delay can never over-lap the time for the bus stop; as it can at a near-side stop.

To be a fair, apples-to-apples, comparison to determine which is better, a study should look at the following time periods:

• For a near-side stop – the time from when the bus arrives at the stop (or near-enough for people to quickly walk to it and board/alight while the bus sits in traffic) until the bus crosses the far-side of the intersection;
• For a far-side stop – the time from when the bus stops for the last time before crossing the intersection (in case it takes multiple light cycles to cross), or from when the bus crosses the near-side edge of the intersection (if it does not stop), until the bus leaves the far-side bus stop.

This study, which is often reference as proving that far-side stops save time over near-side stops, does not prove anything of the sort. All it does is show that, if you include the random timing of a signal light into the time of stops at one location, then it will, generally, take longer than if no such additional time is included for stops at another position.

Good grief! Of course, we should consider the entire time to cross the intersection including the time at the bus stop, whichever side it may be on. Analyzing the stop time in isolation is useless information. Of course it will be longer on the near side if there is a red light. This is perfect example of statistics that can be used and misused to prove anything. Thank you Richard for sharing this.

[HighwayStar]

Originally Posted by lrt's friend

Good grief! Of course, we should consider the entire time to cross the intersection including the time at the bus stop, whichever side it may be on. Analyzing the stop time in isolation is useless information. Of course it will be longer on the near side if there is a red light. This is perfect example of statistics that can be used and misused to prove anything. Thank you Richard for sharing this.

Sure... but if the bus was going to the far side stop, many times it would hit the red light anyway and wait 30 seconds (or whatever) to move on.

In the near side example, the red-light-factor is part of the stat.. in the far side example, the red-light-factor is completely ignored.

[lrt's friend]

Originally Posted by HighwayStar

Sure... but if the bus was going to the far side stop, many times it would hit the red light anyway and wait 30 seconds (or whatever) to move on.

In the near side example, the red-light-factor is part of the stat.. in the far side example, the red-light-factor is completely ignored.

Exactly. You need to consider both the time to cross the intersection which may include a red light cycle and the time while at the bus stop regardless of whether it is near side or far side. This is the only way that we are comparing apples to apples. It may still prove that a far side stop is better but the study didn't do this. Both a near side and far side stop may include a wait for a red light cycle and additional time at the bus stop itself. Signal priority may provide an advantage to a far side stop but signal priority for transit has not been implemented very often.

[TransitZilla]

Originally Posted by Richard Eade

Studies, like the one that the referenced article is based on, really need to be understood for what they are. This one Except, nowhere in the study did they take into account the most fundamental reason for a stop lasting longer before the bus crosses a signal-controlled intersection – that a red light was involved.

The study doesn't quite say what you say it says. It actually differentiates between signalized and non-signalized intersections. The figure you quoted (near side stops taking 5 seconds longer) is for near side stops not at signalized intersections.

Near-side stops located at signalized intersections are even worse: 9.2 seconds slower than non-signalized stops (Page 10). As you say, this extra time can be attributed to the traffic signal being red while the bus is at the stop.

In the study conclusion, when they say that the route could have 9 extra runs, they use the 5 second number, not the longer signalized number.

As you say, traffic signals could delay the bus with both near side or far side stops, but the study is saying that even ignoring the extra delay caused red lights, near side stops still take longer. And on top of that, far side stops give you the ability to implement signal priority to reduce signal delay.

[Richard Eade]

Quite right, Brad. I tried to summarize too much and mis-quoted the report.

As you point out, the 4.2 (Portland) and 5 (Montreal) second increases in stop time are the averages for all near-side stops (at both signalized and unsignalized intersections) of the routes evaluated in those cities. If we just consider the signalized intersections, those delays increase to 7.2 (Portland) and 9.2 (Montreal) seconds, on average. Their conclusion used the 4.2/5 second increase because it is the overall average. To be more precise, they could have done a count of the number of near-side stops at signalized and unsignalized intersections for each route and come to specific delays for each route.

That correction said, it does not change the point that I was making; that the study does not include any time crossing the intersection for the far-side stops. That intersection delay could happen at signalized or non-signalized intersections.

For example, if the bus arrives at a near-side stop at a signalized intersection, it needs to stop for passengers. It can’t leave that stop until it has a green light. That might mean that it can leave immediately, or it might mean waiting for most of a cross-stream phase. Regardless of how much a bus needs to wait on a specific trip, without any signal modification (bus priority) there will be runs that take longer due to waiting for a green signal. At the end of the day, the total of the time taken at that stop for each run divided by the number of runs (i.e., the average time per stop) will be longer than the shortest stop duration (which, presumably, was not influenced by a red signal) – accounting for all other variables, of course. Basically, the bus is going to be delayed sometimes by having to wait for a green signal and that increases the average stop time at signalized intersections.

Similarly, at non-signalized intersections, there are going to be times when the bus can not leave its stop and cross the intersection because of cross-traffic in the intersection, turning vehicles in front of the bus, or pedestrians crossing in front of the bus. Just as traffic signals increase the average time at a near-side stop, factors such as these delay buses leaving near-side stops at non-signalized intersections. Thus, they also will have a longer average stop time.

However, none of the above-mentioned delays are taken into account by the study when it calculates its far-side stop time. Without accounting for the intersection-caused delays for far-side stops, there is no fair comparison of the stop times of near-side verses far-side stops.

If I assume some numbers; 15 seconds for the actual transit stop with all variables held constant, and a simple intersection that alternates between west-east and north-south green signals every 60 seconds; I would have the following:

In effect, with the study’s methodology, far-side timing would remain at 15 seconds for each time the bus reached the stop, while near-side timing would vary between 15 seconds (the bus arrives and leaves within a green phase) and 75 seconds (the bus arrives 15 seconds before the phase changes to red).

However, if I was going to make it a fair comparison, I would need to change it to the following:

A far-side stop could not be reached until the bus can get through on a green phase, so, the far-side stop time would be 0 to 60 seconds for the intersection PLUS the 15 seconds for the stop. This makes the timing for a far-side stop vary from 15 seconds to 75 seconds. That is the same as a near-side stop, EXCEPT that a near-side stop can have the stop time and the signal time overlap – something that the far-side stop can not do.

If the bus arrives at the intersection at exactly the moment that the light turns red, it will have to wait 60 seconds before it can proceed across the intersection. If there is a near-side stop, the 15 seconds for the stop can start and be finished long before the bus can move on. In that case, the bus will wait a total of 60 seconds at the near-side stop. If the stop were situated across the intersection, the bus would still have to wait the 60 seconds for a green light before it could get to the stop to wait 15 more seconds; for a total of 75 seconds.

Because the time at a far-side stop must be successive to the time required to get through a signalized intersection, the average time for a far-side stop must be longer than the average time at a near-side stop where the stop time and the intersection time can be concurrent.

In the case of a non-signalized intersection, the bus leaving a near-side stop must find a clear path through the intersection, thus the intersection is added to the stop time, just as it is for a far-side stop. Ergo, for an unsignalized intersection, there should be very little difference in stop times for the two locations.

[lrt's friend]

Thank you Richard for working through the logic that suggests that the study did not fairly consider all factors in making their conclusions.

What you have concluded as a result of including signal cycle time for both near and far side stops is consistent with the 'gut feel' of some posters on this board that a near side stop would be more efficient.

I would like to see similar comparisons and studies by bus type, regular buses versus articulated verses double decker. I feel there may be some false economies created by the use of larger buses especially on routes with frequent stopping. Also, service efficiency losses for the rider because larger buses will need to stop more frequently when handling larger volumes of passengers.

[TransitZilla]

Wow, I really started something here, eh?

Richard, you've made lots of good points. It would be interesting to see another study that does look at the full time to serve near side or far side stops and cross signalized intersections. The McGill study doesn't consider everything.

I'm not sure what other studies exist on this but I'm sure there are some.

Originally Posted by Richard Eade

If I assume some numbers; 15 seconds for the actual transit stop with all variables held constant, and a simple intersection that alternates between west-east and north-south green signals every 60 seconds; I would have the following:

In effect, with the study’s methodology, far-side timing would remain at 15 seconds for each time the bus reached the stop, while near-side timing would vary between 15 seconds (the bus arrives and leaves within a green phase) and 75 seconds (the bus arrives 15 seconds before the phase changes to red).

However, if I was going to make it a fair comparison, I would need to change it to the following:

A far-side stop could not be reached until the bus can get through on a green phase, so, the far-side stop time would be 0 to 60 seconds for the intersection PLUS the 15 seconds for the stop. This makes the timing for a far-side stop vary from 15 seconds to 75 seconds. That is the same as a near-side stop, EXCEPT that a near-side stop can have the stop time and the signal time overlap – something that the far-side stop can not do.

If the bus arrives at the intersection at exactly the moment that the light turns red, it will have to wait 60 seconds before it can proceed across the intersection. If there is a near-side stop, the 15 seconds for the stop can start and be finished long before the bus can move on. In that case, the bus will wait a total of 60 seconds at the near-side stop. If the stop were situated across the intersection, the bus would still have to wait the 60 seconds for a green light before it could get to the stop to wait 15 more seconds; for a total of 75 seconds.

I hear what you're saying, but what is the likelihood that a bus caught at a red light will happen to have pulled right up to the stop when the light is red so that it can load/unload? I'd say is is highly unlikely.

With a near-side stop, it is more likely that the bus (even with a near-side stop) will be queued behind other vehicles short of the bus stop. It won't be able to advance to the stop until the light has turned green, and then by the time it is done loading/unloading, the light will have turned red again and it will be stopped for another light cycle.

In your example, the worst case scenario for a far side stop is about 75 seconds. With a near-side stop, you could easily have a situation where the bus is caught for 30 seconds in the queue at a red light, but not yet at the bus stop. Then it advances, spends 15 seconds at the stop, but by that time the light has turned red, so then it has to wait another 60 seconds. Now we are at 105 seconds.

I've had exactly this situation happen several times on the eastbound 85 at Carling/Merivale. We've approached the intersection as the light is green. But it is a near-side stop, so we have to stop short of the intersection. By the time we are ready to go, the light turns red, and then we are stuck for a full light cycle before being able to proceed. If that stop was on the far side (which I think it should be), we would have breezed through the light, stopped for 20 seconds or so, then been on our way.

Because the time at a far-side stop must be successive to the time required to get through a signalized intersection, the average time for a far-side stop must be longer than the average time at a near-side stop where the stop time and the intersection time can be concurrent.

Technically this could be true, but the worst case scenario I've outlined for a near-side stop is worse than the worst case scenario for a far side stop. The likelihood of the stop time being exactly concurrent with the red light time is low. And with far side stops, it is technically very easy to implement bus-detection technology to hold green lights so that buses will never be caught at the intersection just as the light turns red. This reduces the potential signal delay. This isn't really possible with near-side stops because the green would have to be held much longer and it's not really predictable how much longer.

The McGill paper speaks to this on page 13 where it discusses the increase in stop time variability (9%) on near-side stops. Variability is bad if we want buses to stick to a consistent schedule. Far-side stops with signal priority make consistency much easier to achieve.

[CityTech]

Originally Posted by bradnixon

In your example, the worst case scenario for a far side stop is about 75 seconds. With a near-side stop, you could easily have a situation where the bus is caught for 30 seconds in the queue at a red light, but not yet at the bus stop. Then it advances, spends 15 seconds at the stop, but by that time the light has turned red, so then it has to wait another 60 seconds. Now we are at 105 seconds.

I've had exactly this situation happen several times on the eastbound 85 at Carling/Merivale. We've approached the intersection as the light is green. But it is a near-side stop, so we have to stop short of the intersection. By the time we are ready to go, the light turns red, and then we are stuck for a full light cycle before being able to proceed. If that stop was on the far side (which I think it should be), we would have breezed through the light, stopped for 20 seconds or so, then been on our way.

This is very true. The high probability of being stuck at a red light twice in moderate to heavy traffic is the main argument as to why far side stops are better.

That said, that problem could be mitigated with queue jump lanes on the near side (which would allow the bus to proceed to the stop even if there's a red light blocking cars ahead from moving, preventing this specific problem).

Originally Posted by bradnixon

Technically this could be true, but the worst case scenario I've outlined for a near-side stop is worse than the worst case scenario for a far side stop. The likelihood of the stop time being exactly concurrent with the red light time is low. And with far side stops, it is technically very easy to implement bus-detection technology to hold green lights so that buses will never be caught at the intersection just as the light turns red. This reduces the potential signal delay. This isn't really possible with near-side stops because the green would have to be held much longer and it's not really predictable how much longer.

The McGill paper speaks to this on page 13 where it discusses the increase in stop time variability (9%) on near-side stops. Variability is bad if we want buses to stick to a consistent schedule. Far-side stops with signal priority make consistency much easier to achieve.

This is the real advantage of far side stops: they minimize the maximum downside and thus create more consistent travel times. Consistency is arguably even more important than travel time for local buses.

The only situation where near side isn't ultimately worse is one where there's queue jump lanes/bus lanes, and no signal priority. Then the downside of near side (the double-red-light problem) is taken away and the upside of far-side (signal priority) is not given.

[CityTech]

Another issue is dwell times. I don't know if I'm just crazy, but it seems to me that people in Ottawa take a lot longer to board/exit buses than people in other cities. If you compare riding a bus in Ottawa to riding a bus elsewhere, it seems that buses stay stopped at stops for way longer. In Toronto it seems like the bus can turn over half its seats in a few seconds, whereas here stops take an eternity if there's more than a handful of people using them.

[PHrenetic]

Originally Posted by CityTech

Another issue is dwell times. I don't know if I'm just crazy, but it seems to me that people in Ottawa take a lot longer to board/exit buses than people in other cities. If you compare riding a bus in Ottawa to riding a bus elsewhere, it seems that buses stay stopped at stops for way longer. In Toronto it seems like the bus can turn over half its seats in a few seconds, whereas here stops take an eternity if there's more than a handful of people using them.

Good Day.

When I moved here from Montreal, this is EXACTLY what I found..... and I attribute it to four factors :
- 2+2 seating with NARROW aisles here .vs. 2+1 seating with WIDE aisles almost everywhere else, except on specific long-distance routes,
- front-door boarding and rear-door debarking is just plain easier, esp. with #1,
- properly queued wait lines almost everywhere else.vs. the chaos scramble jam at the front-door to pay boarding,
- not waiting for unload before jamming on to load, blocking everybody.

IE: - a general lack of mobility-enhancing features, and non-courtesy undoubtedly exacerbated by the OC-T penchant for disaster bus congas at overloaded stops.

Definitely NO Joy !

[Uhuniau]

Originally Posted by CityTech

Another issue is dwell times. I don't know if I'm just crazy, but it seems to me that people in Ottawa take a lot longer to board/exit buses than people in other cities. If you compare riding a bus in Ottawa to riding a bus elsewhere, it seems that buses stay stopped at stops for way longer. In Toronto it seems like the bus can turn over half its seats in a few seconds, whereas here stops take an eternity if there's more than a handful of people using them.

A very large part of the dwell-time problem, at least on the truly urban routes in Ottawa, stems from OCTranspo's continuing refusal to run the right equipment (i.e., artics) on the right routes, and the crappy practice of way too many drivers of not automatically opening up all doors of high-capacity buses to boarding, instead of just the front door.

Routes like the 19, 7, and 12 would run so much more smoothly and efficiently with these changes. The 7, in particular, is a notorious victim of Transpo's favouritism towards suburban ass-meet-seat service at the expense of urban routes.

[Richard Eade]

Originally Posted by bradnixon

. . .Technically this could be true, but the worst case scenario I've outlined for a near-side stop is worse than the worst case scenario for a far side stop. The likelihood of the stop time being exactly concurrent with the red light time is low. And with far side stops, it is technically very easy to implement bus-detection technology to hold green lights so that buses will never be caught at the intersection just as the light turns red. This reduces the potential signal delay. This isn't really possible with near-side stops because the green would have to be held much longer and it's not really predictable how much longer.
. . .

Well, we can both cherry-pick certain cases that look good for either near-side or far-side stops. Instead, let’s see if we can work it through and find a real answer.

I’ll assume that it takes exactly 15 seconds for the bus to service a stop. The intersection is signalized with 25 seconds of green, 5 seconds of orange, and 30 seconds of red. The bus driver is very law-abiding and only proceeds into the intersection if the light is green – so no crossing on orange. There is no signal priority applied (just as there was none in the study).

For the 60 seconds of the traffic cycle, the bus could arrive at any second, so I’ll make a table showing the delay for each location of stop for each of 60 arrival times:

Time (s) 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Signal ....Gr Gr Gr. Gr Gr. Gr Gr. Gr. Gr. Gr Gr. Gr Gr. Gr Gr. Gr Gr. Gr Gr. Gr Gr. Gr Gr. Gr Gr. Or Or. Or Or. Or
Near (s) 15 15 15 15 15 15 15 15 15 15 50 49 48 47 46 45 44 43 42 41 40 39 38 37 36 35 34 33 32 31
Far.. (s) 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 50 49 48 47 46

Time (s) 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Signal ....Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re
Near (s) 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15
Far.. (s) 45 44 43 42 41 40 39 38 37 36 35 34 33 32 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16

You will notice that the numbers are the same, but there is a phase-shift between the two delay curves. A graphic will show this better:



The total of the delays for each of the curves is 1530 seconds. This tells us that, all other things being equal, it should not matter if the bus stop is placed near-side or far-side.

OK. So, if there is no benefit to near-side or far-side in the case of random arrival at a signal-controlled intersection, what things actually do make a difference?

You mentioned holding a green signal for an approaching bus that is heading to a far-side stop. What happens when a bus on the cross street also has the green held for it? Given that the over-all cycle length is probably going to stay the same, doesn’t that mean that you are potentially robbing green for a later bus. Over time, will things not even out to be of no real benefit?

Also, predicting when to extend a green for an approaching bus could be difficult. Is there a costly transmitter and receiver involved? What do you do if the bus is creeping toward the intersection in bumper-to-bumper traffic? OC Transpo has a hard-enough time trying to predict when a bus will arrive at your bus stop. What makes you think that it would be any more precise predicting a bus’s arrival at an intersection? Also, why not extend the green and allow near-side buses to leave up to 5 seconds after their signal should have gone red? It is easy to detect a bus that is already stopped at a stop.

[Catenary]

Originally Posted by Richard Eade

Well, we can both cherry-pick certain cases that look good for either near-side or far-side stops. Instead, let’s see if we can work it through and find a real answer.

I’ll assume that it takes exactly 15 seconds for the bus to service a stop. The intersection is signalized with 25 seconds of green, 5 seconds of orange, and 30 seconds of red. The bus driver is very law-abiding and only proceeds into the intersection if the light is green – so no crossing on orange. There is no signal priority applied (just as there was none in the study).

For the 60 seconds of the traffic cycle, the bus could arrive at any second, so I’ll make a table showing the delay for each location of stop for each of 60 arrival times:

Time (s) 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Signal ....Gr Gr Gr. Gr Gr. Gr Gr. Gr. Gr. Gr Gr. Gr Gr. Gr Gr. Gr Gr. Gr Gr. Gr Gr. Gr Gr. Gr Gr. Or Or. Or Or. Or
Near (s) 15 15 15 15 15 15 15 15 15 15 50 49 48 47 46 45 44 43 42 41 40 39 38 37 36 35 34 33 32 31
Far.. (s) 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 50 49 48 47 46

Time (s) 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Signal ....Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re
Near (s) 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15
Far.. (s) 45 44 43 42 41 40 39 38 37 36 35 34 33 32 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16

You will notice that the numbers are the same, but there is a phase-shift between the two delay curves. A graphic will show this better:



The total of the delays for each of the curves is 1530 seconds. This tells us that, all other things being equal, it should not matter if the bus stop is placed near-side or far-side.

OK. So, if there is no benefit to near-side or far-side in the case of random arrival at a signal-controlled intersection, what things actually do make a difference?

You mentioned holding a green signal for an approaching bus that is heading to a far-side stop. What happens when a bus on the cross street also has the green held for it? Given that the over-all cycle length is probably going to stay the same, doesn’t that mean that you are potentially robbing green for a later bus. Over time, will things not even out to be of no real benefit?

Also, predicting when to extend a green for an approaching bus could be difficult. Is there a costly transmitter and receiver involved? What do you do if the bus is creeping toward the intersection in bumper-to-bumper traffic? OC Transpo has a hard-enough time trying to predict when a bus will arrive at your bus stop. What makes you think that it would be any more precise predicting a bus’s arrival at an intersection? Also, why not extend the green and allow near-side buses to leave up to 5 seconds after their signal should have gone red? It is easy to detect a bus that is already stopped at a stop.

On your last point, it is virtually impossible to extend the green for a bus at the stop, and can be counter productive to do so. If you extend the green and the bus still doesn't make it through, the bus now has to wait longer at the light than if you didn't extend the green.

The other point that was mentioned earlier is the situation where a bus is stuck in a line of cars. If there's even a couple of cars ahead of the bus at a red light, the bus may not be able to serve the stop. When the bus finally makes it to the stop, it is then there on the green, and is likely to have a red by the time it is ready to go.

In the end, far-side stops are best practice for a reason, and it isn't because of bunch of speculation on the internet.

[roger1818]

Originally Posted by Richard Eade

Well, we can both cherry-pick certain cases that look good for either near-side or far-side stops. Instead, let’s see if we can work it through and find a real answer.

I’ll assume that it takes exactly 15 seconds for the bus to service a stop. The intersection is signalized with 25 seconds of green, 5 seconds of orange, and 30 seconds of red. The bus driver is very law-abiding and only proceeds into the intersection if the light is green – so no crossing on orange. There is no signal priority applied (just as there was none in the study).

For the 60 seconds of the traffic cycle, the bus could arrive at any second, so I’ll make a table showing the delay for each location of stop for each of 60 arrival times:

Time (s) 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Signal ....Gr Gr Gr. Gr Gr. Gr Gr. Gr. Gr. Gr Gr. Gr Gr. Gr Gr. Gr Gr. Gr Gr. Gr Gr. Gr Gr. Gr Gr. Or Or. Or Or. Or
Near (s) 15 15 15 15 15 15 15 15 15 15 50 49 48 47 46 45 44 43 42 41 40 39 38 37 36 35 34 33 32 31
Far.. (s) 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 50 49 48 47 46

Time (s) 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Signal ....Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re Re
Near (s) 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15
Far.. (s) 45 44 43 42 41 40 39 38 37 36 35 34 33 32 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16

You will notice that the numbers are the same, but there is a phase-shift between the two delay curves. A graphic will show this better:

The total of the delays for each of the curves is 1530 seconds. This tells us that, all other things being equal, it should not matter if the bus stop is placed near-side or far-side.

OK. So, if there is no benefit to near-side or far-side in the case of random arrival at a signal-controlled intersection, what things actually do make a difference?

You mentioned holding a green signal for an approaching bus that is heading to a far-side stop. What happens when a bus on the cross street also has the green held for it? Given that the over-all cycle length is probably going to stay the same, doesn’t that mean that you are potentially robbing green for a later bus. Over time, will things not even out to be of no real benefit?

Also, predicting when to extend a green for an approaching bus could be difficult. Is there a costly transmitter and receiver involved? What do you do if the bus is creeping toward the intersection in bumper-to-bumper traffic? OC Transpo has a hard-enough time trying to predict when a bus will arrive at your bus stop. What makes you think that it would be any more precise predicting a bus’s arrival at an intersection? Also, why not extend the green and allow near-side buses to leave up to 5 seconds after their signal should have gone red? It is easy to detect a bus that is already stopped at a stop.

The big problem with your random arrival analysis is it is assumes there is an equal probability that a bus will arrive at the stop at any time within the cycle. This is not the case for either Near-side or Far-side stops.

In the case of Near-side stops, unless there is a queue jump lane, traffic will block the bus from reaching the stop within a few seconds of the light turning red, so the vast majority of buses will arrive at the stop while the light is green. Since the dwell time is almost as long as the green signal, the light will most likely be amber or red when it is time to leave.

In the case of a Far-side stop, since the buses can't travel through a red light, all buses will arrive anywhere from a few seconds after the light turned green until a few seconds after the light turned amber (it takes a few seconds for the bus to get from the intersection to the stop).

As you can see, in both cases, the vast majority of buses will arrive at the stop at the same point in the cycle, but in the case of a Near-side stop, almost all of them will then have to wait for a red light to turn green after the stop dwell time, where as in the case of a far-side stop, they can all proceed normally.

[Richard Eade]

Originally Posted by Catenary

. . . The other point that was mentioned earlier is the situation where a bus is stuck in a line of cars. If there's even a couple of cars ahead of the bus at a red light, the bus may not be able to serve the stop. When the bus finally makes it to the stop, it is then there on the green, and is likely to have a red by the time it is ready to go. . . .

Yup, you can pick specific cases that favour far-side stops, but I could suggest even more cases where a near-side stop wins. Look at the graph I provided above. Your example is shown at, say, 11 seconds into the cycle. This provides the most delay for a near-side stop and the least delay for a far-side stop. But that is one example. If the bus had arrived 1 second earlier, there would have been no difference in the delays between the two locations. If it had arrived 15 seconds later, the near-side stop would have been faster.

Your example mentions a line of cars that cause the bus to take more than one signal cycle to get to the intersection/stop. This, of course, is irrelevant. Having the stop on either side of the intersection will have no effect on a bus getting TO the intersection. A bus heading to a far-side stop will take exactly the same number of signal cycles to get to the intersection or stop. That is why I referenced 'the last time a bus stops before the intersection' in my previous post.

. . . In the end, far-side stops are best practice for a reason, and it isn't because of bunch of speculation on the internet.

Sorry, but your declaration (whether on the Internet or not) is insufficient to support far-side stops as being superior. Do you actually have any analysis that you can provide?

It seems to me that there must be a reason for the number of near-side stops that exist (except, maybe, in Ottawa). For example, if I look at the study that started this movement and check the routes that it evaluated, we get the following information:

In Montreal; route 45N - out of 43 stops, 34 are near-side; 45S - 37 of 43 are near-side; 67N - 37 of 38; 67S - 37 of 41; 161E - 37 of 44; 161W - 44 of 48; 165N - 25 of 31; 165S 26 of 35; 439N - 9 of 19; 439S - 15 of 23; 467 - 15 of 16; 467S - 17 of 18; 470E - 19 of 24; and 470W - 17 of 23 are near-side stops.

It seems to me that the fact that the majority of stops are near-side stops would indicate that the industry practice is to put stops on the near-side.

[TransitZilla]

Originally Posted by Richard Eade

Yup, you can pick specific cases that favour far-side stops, but I could suggest even more cases where a near-side stop wins.

Richard, the point that I and others have been making is that the probability of higher delay for a near side stop is greater. We're not just picking specific cases. The likelihood that a bus will be able to pull up to the correct place to serve the near-side stop and still make it through on the green is low. Is it possible? Yes, but it will not be the case the majority of the time.

Once stop that does work well as a near-side stop in Ottawa is at Heron WB & Prince of Wales. But that is because there is a queue jump lane, so buses are always able to drive right up to the stop and serve it even if the light is red. In locations without a queue-jump lane, buses will typically be caught in traffic and will not be able to pull up to the stop until the light changes to green. In that case the light is likely to have changed to red by the time the bus has finished serving the stop.

BTW: Translink agrees that far-side stops are better.