If you read all the hysteria, we are on the cusp of a massive transport disruption that will entirely transform our current system through gadgets like driverless cars, transport-as-a-service, connected vehicles, taxi bots, various fancy apps and goodness knows what else. Sitting behind all of this will be incredible data processing capability that’s currently almost beyond our imagination – with each driverless car generating thousands of gigabytes of data a day.

Source: https://www.networkworld.com/article/3147892/internet/one-autonomous-car-will-use-4000-gb-of-dataday.html

This is an exciting future in some respects, if we get it right. But it’s also a future that seems an incredibly long way away whenever I’m unnecessarily waiting at a red light and nothing’s coming the other way – whether I’m walking, on my bike, catching the bus or driving.

Auckland’s traffic lights are controlled by a system called “SCATS”, Sydney Coordinated Adaptive Traffic System. It sounds kind of fancy when you read about it:

The Sydney Coordinated Adaptive Traffic System, abbreviated SCATS, is an intelligent transportation system that manages the dynamic (on-line, real-time) timing of signal phases at traffic signals, meaning that it tries to find the best phasing (i.e. cycle times, phase splits and offsets) for a traffic situation (for individual intersections as well as for the whole network). SCATS is based on the automatic plan selection from a library in response to the data derived from loop detectors or other road traffic sensors.

SCATS uses sensors at each traffic signal to detect vehicle presence in each lane and pedestrians waiting to cross at the local site. The vehicle sensors are generally inductive loops installed within the road pavement. The pedestrian sensors are usually push buttons. Various other types of sensors can be used for vehicle presence detection, provided that a similar and consistent output is achieved. Information collected from the vehicle sensors allows SCATS to calculate and adapt the timing of traffic signals in the network.

In reality though, the system doesn’t work – especially for pedestrians, public transport and at off-peak times. It’s frustrating and I think it hugely undermines the faith people have in Auckland Transport – after all it’s natural to get grumpy when “the system” is unnecessarily keeping you waiting. It’s also very dangerous, as pedestrians frequently and understandably simply run out of patience and just cross the road.

Digging into the detail of why our “dumb” traffic lights are so damn annoying is challenging, but seems to come back to a few key changes that we need to make around shortening signal cycle lengths and shifting to truly “smart” technology.

Shortening Cycle Lengths

An intersection’s “cycle length” is how long it takes all phases of the traffic light to complete.  In Auckland most cycle times are 120 seconds or more, whereas industry best practice is 60-90 seconds. Shorter cycle lengths are great for urban areas, walkability and also reduce the frustration and danger of a long and potentially unnecessary wait at a red light:

Though often invisible to the public, traffic signal cycle lengths have a significant impact on the quality of the urban realm and consequently, the opportunities for bicyclists, pedestrians, and transit vehicles to operate safely along a corridor.

Long signal cycles, compounded over multiple intersections, can make crossing a street or walking even a short distance prohibitive and frustrating. This discourages walking altogether, and makes streets into barriers that separate destinations, rather than arteries that stitch them together.

As a pedestrian, and if you know the light sequence for an intersection, there’s often nothing more frustrating than just missing your phase in the sequence. As a driver, cyclist or PT user, while shorter cycle times might mean you’re more likely to not make it through in a single phase and more likely to hit a red light, it also means much shorter waits and a much lower likelihood of unnecessarily waiting.

Moving to shorter cycle lengths is a great example of understanding that human psychology is different to the numbers a computer spits out and – even though it might mean slightly less throughput along a major corridor – the shorter waits and reduction in unnecessary waiting means that the public are almost certainly going to be happier.

Wellesley 5:32pm

Using “Smart” technology

The engineering and maths of traffic lights seem pretty complicated and I’m sure a lot of people spend a lot of time meticulously analysing how to sequence different sets of lights together so that a car can plough through 3 seconds faster. But in this day, when adaptive artificial intelligence and machine learning is becoming a reality, it seems positively archaic to take such a “hands on” approach – especially one that seems to so quickly fall apart if something unusual happens, or doesn’t work for you at all if you’re a pedestrian, or leaves you waiting at a red light with nothing coming if you’re travelling off-peak.

Surely, if we are going to be anywhere near ready for our much hyped transport technology revolution, then we can at least have properly adaptive traffic lights that use smart cameras to know exactly how many vehicles and people and buses and bikes are at the each intersection, know how long they’ve been waiting, know what’s happening at nearby intersections and can run thousands of scenarios in half a second before picking the one that’s most efficient and best gives effect to whatever strategic goal we have asked it to perform. This probably means a complete overhaul of the current SCATS system, but I’d be enormously surprised if there aren’t a whole heap of much smarter software solutions out there which can do this task.

Overall, we should not accept the dumb traffic light system we have in place today. Contrary to many letters to newspaper editors, improving traffic lights will not solve all our transport challenges. However, I’m confident that shifting to shorter cycle times and smarter technology with a relentless focus on finding a better way to optimise across all road users, will make a big difference. Perhaps most importantly for Auckland Transport, a major focus here is a good way of building trust with the public and getting rid of those enormously frustrating and downright dangerous situations of unnecessary delay.

Share this


  1. A huge bear bug for me is the disdain Auckland Transport treats the pedestrian or cycle phase, firstly requiring a beg button, but secondly whenever it is pressed the lights will cycle through all the cycle for vehicles before getting back to wherever the cross signal would get slotted in. In comparison, whilst waiting you often see a car arrive at another part of the intersection and have to stand and wait whilst they get instant priority and a green light. It’s clear that vehicles and inserting vehicle phases are priorities over active modes. The fact that pedestrians have to even press the button to even get a phase is another joke, with many central city lights cycling automatically between greens for each car direction, even in the absence of any cars at all, meanwhile pedestrians have to arrive press and wait.

    1. Yes, and while Council is trying to improve pedestrian amenity in order to rejuvenate town centres, the traffic light phasing hinders this work. Residents have been asking for better phasing at a local crossing since last century. Council reports mention it time after time. Yet nothing happens.

    2. I’ve never really understood why beg buttons get so much hate. They can occasionally be mildly annoying, but I they’re better than any of the alternatives. In Christchurch they’re starting to put in infrared sensors so you don’t have to push a button, but they have their problems – like thinking you’ve disappeared and cancelling the phase, just because you’ve inadvertently stepped outside of the invisible area that triggers the sensor. Calling all pedestrian phases every cycle would be another alternative, but then that would make the signals really annoying at times when there’s no pedestrians around e.g. middle of the night. What would you want instead of buttons?

      1. “What would you want instead of buttons?”

        What much of the civilised world does – automatic pedestrian phases! They activate every round. Like, you know, those for cars.

        Its an elementary fairness – at least in areas where pedestrians are likely to use the crossing 80-90% of every phase anyway, we should not expect pedestrians to have to ask.

        In NZ however, we even force them to ask if their movement does not conflict with ANY other movement. I.e. crossing Hobson Street on the north side of Union Street. All the cars are stopped there every phase, and because its a one-way southbound, pedestrians could go automatically. But no. Beg button. Peds are peasants.

        1. Agree Ari – car phases do not get called every cycle, only when there’s cars there (that’s the whole point of SCATS). Same with bike phases, bus phases and pedestrian phases. I just think calling all ped phases every cycle would make sense at certain intersections at certain times of day, but you’d still want some sort of sensor (button or otherwise) for the times when there aren’t many pedestrians around.

        2. As someone who has worked with signal designers a lot I can tell you that cars do NOT have anything equivalent to beg buttons:

          a) *every* intersection has automatic phases for cars. They do *not* get called by traffic loops. In converse, I can tell you maybe 5 out 200 intersections I have worked on have automatic pedestrian phases.

          b) in fact, most intersections do not even *have* detector loops for the straight through movements. Only for turning traffic

          c) Drivers whose movements necessitates a loop do not have to push a button. They are detected automatically. Let them lean out of the window to push a button if you want to make it fair.

          So no, its not an elementary fairness.

        3. How old are you Damian? Signal phases have been called by loops for over 50 years. One of the major costs of signals is the need to cut loops on every lane and duct the wires back to the local controller. Scats requires a presence loop at the limit line on every lane and some designers get carried away an add in advanced loops as well.
          The problem is that SCATS is quite dumb as Mat says. It doesn’t make very good use of the data as it puts all intersections in a subsystem onto a common cycle time ( the cycle appropriate to the worst intersection) then it leaves up greens on the stretch phase of every other intersection. But there are most definitely loops and car phases are called..

        4. You are partially correct, and I was partially wrong, mfwic. I did have a look at a few plans, and its true that I misremembered b) somewhat – most through lanes (but no, not all) on recent signals designs I had some association with have loops.

          So B) partially – as for A) and C) I’ll stick with my comments.

        5. If it is connected to SCATs then it has to have loops on every lane that is signalised. SCATS can’t work otherwise. The presence loops don’t count cars but work out the saturation level on each approach or the % of time the loop is occupied. SCATS needs that on throughs as well as turns as an input to its cycle time model and its splits or phase allocation model. In the past only some loops were used as an input to its offset or coordination plan choice module. Even if it is not shown on a plan someone will cut a standard 4.5m loop when the lights go in.
          (The loop length is why scats loops are not that good for counting. Cars often have gaps of less than 4.5m when it is very busy and slow and the loop sees them as one vehicle. But as I said it is looking at occupancy of the loop not counts).

        6. Put detectors in that detect me automatically and trigger instantly if there’s no conflicting movements (ie that emulate the convenience of car detector loops) and that might approach fairness. Instead I have to stop and wait every time, often for an entire cycle. Often twice, since cars get a phase for every movement and pedestrians don’t, except for a very, very few Barnes dances.

          (Hey, that’s a thought. Automatic triggering for straight-through movements and a beg button to request a Barnes dance sequence for other movements. Just like cars get!)

        7. 1st post. some bad info here:

          Ped auto recall is included in most if all softwares. I understand it is used in Auckland (Queen St CBD intersections. It’s also used at some of the CRL roadwork intersections where the beg button is unaccessible.

          a) Not every intersection has automated phases for cars. The vast majority have an arterial (aka reverted) phase but there’s quite a few (mostly motorway interchanges) that rest in last phase until a detector is activated.

          b) Completely wrong. Please name 1 intersection that doesn’t have limit line nor approach loops.

          c) AFAIK There’s a trend away from cycle loops. How do you detect a carbon road bike, and how do you notify the cyclist that he has triggered the call? Seems to be a shift to grab rails with light in the beg button box to notify that demandl is lodged.

          You can set different master isolated thresholds on SCATS so that intersections in the same subsystem can run on independent cycle times.

          Loops don’t get cut if they’re not on the construction plan. The design review process is pretty robust and missing loops should be added prior to construction plans

          “(Hey, that’s a thought. Automatic triggering for straight-through movements and a beg button to request a Barnes dance sequence for other movements. Just like cars get!)” Symonds St in the middle of uni has something similar I think.

        8. Scats loops both count vehicles and measure saturation. The counts can be used to trigger jumps in cycle times but that’s not used on all intersections. Info on use of degree of saturation data is correct but note that in general a rolling average of the last three phases are used.

        9. +1 to one way streets! The pedestrian phase should *always* be active when the one way has a red; no beg button, no delay, just green, green, green.

        10. I partially agree. The lights should be active for most of the time the parallel through traffic is occurring to allow for better walking conditions. However, where there is a not uncommon right turn situation without a turn arrow, there should be a period of the phase where the pedestrian phase is not active to allow turning traffic to safely complete their turning maneuver. This only needs to be a short period of time, but also important that pedestrians don’t dash across when they are on a red light themselves. This is where both pedestrians and motorists need to get better at following the rules.

        11. Don’t press the button. If the red ped light isn’t illuminated you can legally cross whenever it is clear because you are unsignalized. If you press the button, then the light illuminates instantly.

    3. Yes, I had some very frustrating conversations with AT once after sending in a suggestion that one particular well-used ped/cycle phase could usefully trigger automatically. Apparently we need to ‘balance’ the needs of different road users, and that’s why two types of users need to be made to wait the cycle around if they arrive just too late to press the button in time…

    4. Well said BBC. It’s nuts we have beg buttons at all. At have told me it’s to reduce the length of the phases so they don’t have to allow for pedestrians to cross. Does not make sense if our phases are so long..

    5. In the 1980s I noted pedestrians were detected at some French traffic lights by a device that appeared to be a form of radar. As I recall it involved two arms set at an angle of about 45 degrees sloping up over the general area where pedestrians would cross. seemed to work perfectly the few times I used these intersections. I recall one set at Dunkirk.
      Are they still in use? Anyone aware of their drawbacks?

  2. The (surburban) pedestrian traffic lights I use most often stop the traffic almost immediately. Now admittedly this is not an intersection just a crossing(s) but if they are prepared to stop traffic on a busy arterial every time I push the button – why not elsewhere?

    Could it be that the crossings are all near schools?

    1. That is unusual, I’ve used a few of these and they all let me wait for a few minutes before giving green light.

      1. Unusual but they exist and are good, and yes I think I’ve only seen them mid-block. They make you wait if there was a pedestrian phase recently. Otherwise, they give you a green immediately (or rather, once the traffic has a red). There is no pedestrian phase unless it is called.

        The problem I’ve had with them is that they are often associated with having to stand on a block and be sensed as well as having to press the beg button. This doesn’t work for lots of people who are in conversations – they just don’t read the sign saying it, and they don’t necessarily wait on the block. Also, the blocks stop working. And people don’t report them because they’re so used to long wait phases, they give up on the wait and cross anyway, rather than realising it’s a failure.

        1. There weren’t. You’d see those other pedestrians while you’d be approaching the light. A while ago I was at one in Pt Chevalier, and the nearby traffic light went through two (!) cycles while we were waiting.

          There’s one on Northcote Road near the motorway, for that one the wait is so long I didn’t figure out for a long time whether it actually worked or not. You should never have to wait for more than 2 minutes at such a pedestrian crossing.

        2. Theres one on Border rd, Henderson where Project Twin Streams crosses the road. Almost instant green light for people on foot or bicycle. Alternatively one on Te Atatu Peninsula when you are crossing the onramp to go on bridge is only activated if cars are banking up on bridge- no cars no lights for walkers or those on bikes.

        3. Onewa Road and Mokoia Road have one each. I have waited over two minutes at both of them. Often stationary traffic waiting for upstream intersections or no traffic due to reds at upstream intersections, but still two minute waits.

    2. Probably headway timing has terminally expired vehicle approach due to gaps in traffic. If that happens you get a shift to the ped phase immediately. Very common on midblock ped crossings where there’s no continuous ped demand and plenty of gaps in the traffic

  3. I agree with this, I think NZ has one of the worst setups for traffics lights in the world. This based on growing up in UK and visiting a large number of both European and Asian counties over the years.

    The SCATS systems suppose to be pretty smart, but i don’t see any evidence of this in NZ.

    As a young adult in the late 90s in the UK, i used to ride bicycle late at night & if you got the ride speed & not many cars around, you could trigger the sensors to give you a green light thought the junction without stopping. As soon as you had gone thought the junction the lights would also change back to red, interestingly if there no other vehicles, all lights would remain red. The same would happen if you pressed the beg button, if traffic was very low it would more or less change instantaneously.

    1. You are comparing SCATS to the UK SCOOT system which is superior in most ways. SCOOT has always has an online model to adjust signal offsets based on current conditions. It is far and away better for anything except maybe a single arterial corridor (which SCATS was originally designed for). NZ went with SCATS as it was cheap (free in the beginning).

      1. It’s funny mfwic because I recall SCOOT was looking at accommodating a phasing system like SCATS and SCATS is looking to move towards a movement based system like SCOOT. I honestly don’t think SCOOT could do *much* better in Auckland. At least not for the price tag to switch. And yes, all Siemens products are hellish expensive. Heck, they tried to buy SCATS. I think roughly SCATS has 1/3 market share and SCOOT has 1/3 market share so the New South Wales government marketing team must be doing something right.

        1. Yes they have used NSW taxes to fund their R&D. We got SCATS because Manukau City wanted it for Pakuranga Rd. It made sense there, but it didnt make much sense in the CBD. The evaluation was made by the National roads board who did a cheap deal for SCATS. We have been stuck with it since. I curse it everytime I sit at a red light and there is no traffic on the green.

        2. Maybe we should buy something better, then, using a regional fuel tax? Justified on the [crappy] logic of ‘we need more money, this is an easy way to raise some’. And other cities could pay to use it, perhaps raising money with a regional fuel tax there too? Congestion charge being preferable, of course, but that seems a while away, and meanwhile we have a pedestrian being hit every day on average in Auckland.

        3. Seriously Heidi, probably the worst part about any regional fuel tax is it will kill congestion pricing dead. People argue there is no need if you are already paying a fuel tax. Then we miss out on the benefit side of charging people. If we had congestion charges on cordons and allocated the money raised on each corridor across the cordon in for improvements to transport on that specific corridor ie cycle lanes or bus lanes of bus subsidies, then the people paying get a win of fewer people holding them up, and the people who can’t, won’t, don’t pay, get a win of improved options.

    2. I noticed moving from Christchurch to Auckland how frustrating the long phases were and everyone I know from Christchurch thought the same. So it is not necessarily a NZ problem then an Auckland one.

      Is my impression right? Does anyone know if Christchurch uses shorter phasing?

      1. As mentioned elsewhere here on the threads, its a combination of allowing filter turns or not (with some impacts on safety and behaviour) and with having different levels of traffic volumes, and simple different philosophes, it seems.

        It certainly makes sense that you can/would have shorter cycle times if you have filter turns, as it removes (or reduces the incidence/length of) some separate turn phases.

      2. I don’t think Christchurch cycle times are any shorter than Auckland – there’s plenty of 120 second cycle times here as well. But there are probably fewer quirky intersections in Christchurch, with most intersections just being stock-standard 4-way cross-roads, which are easier to get to work efficiently. It is true that Christchurch allows people to do filter right turns far more than Auckland does, which reduces queues and delays for everyone, but the trade-off is more crashes.

        Christchurch also uses SCATS so has all the same limitations associated with that.

      3. My distinct dislike of traffic lights in Christchurch is the lack of right turn phases with right right turning traffic having to sit through a number of phases while waiting to get a break in the oncoming traffic to be able to turn right.

        Again, as with Auckland, replacing traffic lights with roundabouts (and separate zebra pedestrian crossings) would make a big improvement in traffic flow, making things more user-friendly for both motorists and pedestrians.

        1. That’s interesting. As a visitor from Dunedin, I find the lack of right turn phases in Christchurch refreshing. In Dunedin, there is often a red arrow stopping you from your right turn, despite there being no traffic. And it means that there are often 2-4 extra phases each cycle which makes for a multi-minute wait.

        2. The grass is always greener on the other side I guess. I never liked driving in CHCH because the lack of filter turns, but it mostly seems to work because people are used to it.

        3. I like the lack of right turn arrows in Chch. Not hard to go 1 block past and left left left.

        4. I had the pleasure of driving in Napier recently and it was joyous to be able to turn right, on the green phase when there was no traffic coming the other way; no nasty red arrow keeping me waiting for no one to pass like in Auckland

        5. The problem with having or not having RT phases comes down to appropriateness of use. It sounds like here you have cities that either put them in everywhere or nowhere. When I used to design signals, the base case was no turn phase, filter, called “permitted”. Then there were very specific thresholds for turning movements when you needed to add the exclusive turn phase, but kept the filter, so “protected-permitted” and then even higher thresholds (or sight distance limits) for when you eliminated the filter and relied on “protected only”. If you follow those guidelines, you don’t have overly complicated signal phasing, and you don’t have people waiting through multiple cycles to turn or making crazy dangerous turns. Filter is also nice because it allows you to enter the intersection on the green ball and if you don’t get a break, you can legally complete your turn on the clearance interval.

  4. How amazing would it be if the SCATS data actually informed the AT Metro journey planner!? Or if waits for pedestrian phases were so short you didn’t have to worry about them…

    The app shows it’ll take 1min to cross Ti Rakau Dr to catch a connecting bus, but the quickest I’ve covered the 130m this week is 4mins. 3 of those minutes just stood waiting to cross at two sets of lights. With the connecting bus only running every 30mins it’s not worth the risk, so I’ve switched back to a slower route with one less transfer.

    This is the biggest issue with the New Network for me – a system designed around transfers with seemingly little thought put into making transfers easy or plausible outside major interchanges.

  5. I’m not sure that shorter cycle times in general is necessarily the answer. There is nothing more frustrating than being on a bus and taking three phases to get through a simple intersection because the phase is so short – especially when you’re worrying about whether you’ll make your train or not. Agree that there are many situations where shorter cycle times would definitely be more efficient, but I don’t think it’s a hard-and-fast rule.

    Conversely, where there are huge amounts of traffic to shift, short cycle times are extremely inefficient because of the need for many cars to accelerate first from a stop. I think that the answer is “smarter”, not “shorter”

    1. Probably that’s true. Pedestrians having phases twice per cycle, for example, but the traffic phases still being a reasonable length so that the gaps between phases doesn’t introduce inefficiencies.

      1. Let’s assume 30secs per exclusive ped phase. That’s 60secs. If it’s a crossroads you’ll probably need 4 phases to meet demands. So if you want a max of 120s cycle times you have 15 secs for each phase. Take off the 5 secs intergreens and you’re down to 10 secs of green time per phase. If that’s not enough then you’re going to need 140+ cycle times.

        Then there’s the question of at what cycle length do you introduce the repeat exclusive ped phase. Due to intergreens you probably can’t double it until cycle time is 100secs. So at a 99sec cycle time you have 70 secs wait between ped cycles, and then at 101 cycle time you have 20secs wait between cycle times. So how do you manage this transition. There are ways to bandaid it but no real good solutions.

        Maybe better to have a single exclusive ped and set max cycle time lower

    2. All else being equal, there is no difference between waiting through three 120 second cycles or four 90 second ones.

      However given that turn pockets are usually pretty short (two or three cars long) , and turning traffic queues usually back up into the kerbside bus lanes, a faster turnover of shorter phases might actually reduce the traffic impacts on buses.

      1. There is a difference though because of the wasted “churn” time known as intergreen between each phase change. Basically, one line of traffic can’t stop instantaneously and another start immediately, so you have a reduction in throughput every time the lights change. So if you have more cycles (and hence more light changes) in the same time period, then more of that time will be sub-optimal = reduced total throughput = greater delay.

        1. I was referring to DavidByrnes comment about the frustration of waiting in a bus, and not wanting to wait for more cycles. Doesn’t really matter if you wait for more when they are shorter.

        2. 120s cycle time, 4 phases, 30s each, 4 amber/red periods of 5s = 20s total = 17% of the cycle cars are meant to be stopped.
          80s CT, 4 phases, 20s each, 4 amber/red periods of 5s = 20s total = 25% of the cycle cars are meant to be stopped.

          Shorter cycles means lower capacity means more waiting.

        3. Except, as Nick R said above, when there are short lanes. In that case a longer cycle results in a reduction in capacity. See SH18 at Caribean Drive for a good example. The right turn queues back and blocks the through lane. It would be more efficient to shorten the cycle or extend the short lane. Critical lane length Dc=jqr/(n(1-y). Where j is the queue space per vehicle, q is arrival flow in veh/s, r=c-g, and y is the q/s ratio.

        4. That’s a very hairy question…

          Imagine a simple T intersection of 2 streets, both 2 lanes wide. And traffic lights with a very simple cycle (green for the main street, and then green for the side street).

          In theory that gives us some capacity, XX% green time, YY cars per cycle. And so on…

          Now imagine approaching this T intersection on the main street, and from your point of view, turning into the T street would be a cross turn:

          The first car at the green light goes straight. The second has to wait for a gap to turn. In practice that often means waiting until the light turns red. So the actual capacity in your direction is a couple of cars per cycle, regardless of cycle length.

          I’ve seen that a lot before I moved to NZ. When it comes to dumb traffic lights we’re doing OK.

          The variant you see over here is what mfwic is describing. I think this is also what occasionally jams up Birkenhead Avenue / Glenfield Road for a couple of kilometres.

    3. I found the shorter cycles excellent in the UK. As the bus lanes mean the bus always get to the front of the queue, you want to light to go green quicker, and are not concerned about whether you will get through in that cycle, your at the front you always get though.

  6. The cycle length of the traffic lights in Auckland was something that surprised me after moving from the UK. It’s no wonder so many people run amber lights when the alternative is to sit and wait for another 2 minutes. If we shorten the cycle time, can we then get rid of those awful racing track starts where a single lane branches out into two at the traffic lights, then back to one again afterwards? This is presumably to shorten queues but shortening the cycle time would do the same thing and remove the merging battles that always result on the other side.

    1. While I agree that the two-lanes merging back into one is bad for a lot of reasons (including removing space needed for footpaths, cycleways, bus stops…), shortening the cycle time doesn’t make signals work better for qeueus of cars – the opposite is true, which is why we got so long signal times. See my longer response below for explanation.

    2. Actually, the silly racetrack extra lane is specifically to trick the traffic model into showing greater capacity and therefore higher LOS. There seems to be an extreme reliance on traffic modelling here.

  7. One important fact – WHY cycle times in NZ are so long – seems to be missing?

    Its about cars. Like almost everything in NZ traffic, it is about optimising for maximum car troughput.

    That is because every time a traffic light (within the “once around” cycle) switches from letting one street go to giving the green to the other street(s), it needs something called the “intergreen time”.

    Basically, “intergreen” is the time for cars to clear from the middle of the intersection (And to let those red-light runners through without killing *too many* pedestrians… the cynic in me says)

    So every, say 60 second long total cycle that has two intergreen times of 5 seconds (as a simplified example) loses 17% of the whole cycle in terms of capacity – nothing can (or is supposed) to enter the intersection.

    Now increase your cycle time to 120 seconds, because your intersection is so congested that every driver takes several cycles to go through anyway. You still only got 2 x 5 seconds intergreen, but now that “lost time” is only 8% – you have just increased the car throughput of the intersection quite substantially! The signals modeller gets a pat on the back, and pedestrians have to wait a lot longer.

    You got to understand that it’s not just “dumb” technology as to why our signals are so bad for pedestrians.

    Its by design.

    1. Yes, it is about cars, but it’s just as much as a problem also for buses. At the majority of intersections there’s no advance lane for buses, so the impact of short cycle times on buses is just the same as on cars.

      Ultimately, these problems are related to congestion more than they are related to cycle times. Tinkering with cycle times, or the use of smarter traffic light technology, may produce some benefits but the real solution to the problem is to dramatically reduce the number of cars on the road. And that won’t be achieved by switching to driverless vehicles, but by a radical shift to PT.

      1. Yes, we need to “to dramatically reduce the number of cars on the road” as top priority. And in addition to needing a radical shift in priority to PT, a radical shift is needed so walking and cycling become nice experiences.

        I would agree with Damian that it’s by design. Or the way I’d put it is, it doesn’t matter how much clever technology we have, if the design decisions are governed by car bias.

    2. When I was designing signals 90 s was considered the max desirable cycle length, especially in urban areas, and 120 s was only to be used in extreme circumstances like high-speed suburban arterials or extremely complex signals.
      Part of the problem in Auckland is the excessive complexity of phasing. Too many split phases, too many exclusive RT phases, too few concurrent ped phases and the silly LT phases. Generally the minor turn (RT in the States) phases were EXTREMELY frowned upon and ONLY used in locations with very high turning movements or odd conflicts, maybe at 5% of intersections. Then again, turning drivers there are required to give way to peds and actually do so. Another difference is that, outside of New York City, you can generally make the minor turn on a red phase.
      There are some other designs that I find quite quirky here, like providing an exclusive turn phase for a movement that is shared with a through movement, so people trying to turn cannot use the arrow because they are blocked by through traffic.

  8. I’m disgusted at how little filter right turns there are in Auckland. In Melbourne you could say nearly 80% are, whereas here it’s probably close to 5%? In most cases, people will drive just past the intersection and do a quick U-turn and turn left. For example the right turn from Botany Road to Millhouse Drive. At least have a filter right turn at night time….

    1. AT considers them (filter turn signals) to be much less safe. I’m not up to speed on the research as to whether it is true.

      Filter turns at night – I think I wouldn’t like that. Allow it or not allow it. Don’t change how you expect a traffic signal behaves in such a fundamental way by time of day… people will get confused and/or do the manoeuvre even during prohibited times…

      1. If people can’t tell the difference between a red arrow and no arrow they probably shouldn’t be driving a car. No reason not to change the way they are used at different times of the day.

        Most lights that allow a filter turn will still have a red arrow some of the time and drivers seem to cope with this.

        1. I’d expect it is likely less about attention than about permissiveness – “I filter turn here at night, so I feel entitled to do it during the day”.

      2. We have filters in Hamilton that change within a *phase* let alone a day. It works well. Start with a green for that turn, go to red as traffic starts the other way, then go to filters. Simple.

        1. Would be interesting to see a decent sized study. Surprised it wasn’t part of the High Risk Intersections Guide

    2. The very same dude who brought SCATS in and tuned it up at Pakuranga Rd was the main guy who lobbied for years to get red arrows. Once we had red arrows they go put on almost every right turn whether it was needed or not. Not only does it slow everything down and require longer cycle times but they can lead to blocked through lanes when the right turn lane is too short.

    3. This article has the most blatantly misleading headline in the world, but hidden away in the middle of it is this “A draft road safety report, which was presented to the Canterbury Regional Transport Committee last week, also found a lack of traffic light controls for right-hand turns contributed to a high number of crashes involving right turning vehicles.”


  9. Just to correct you Matt there are not people going through any detailed analyses of lights on an ongoing basis. The people running them just make ongoing changes when they think it might help. Sometimes for all the wrong reasons. The long cycle times for example sometimes actually lead to a lower efficiency. If there is a short turning lane that gets full and queues back and blocks the through lane then a shorter cycle is more efficient. But they dont run them that way, I don’t know why. Also since the cycle is set in common for many intersections then many of them will simply have a cycle that is too long.
    The problem is SCATS was designed for a single corridor, not a full network. The method they decided on made sense for a single corridor. ie detect at the stop lines, calculate and set the cycle at the busiest intersection and make that common, then calculate and set the splits or phase times, then final manually set an offset or linking plan between the intersections in the peak direction and let the computer decide when to change to another preset plan. Problem is it is just awful in a network as 4 way coordination can never be achieved. The result is lots of sitting at green lights when noting is coming. It would probably be better to turn the damn thing off and run each intersection isolated in many cases. When things are overloaded then all a signal system can do is ration a bit for each direction.

    1. Is it being taught better at university, now, do you know, mfwic? I remember suggesting a traffic light phasing improvement for pedestrians in a project about Kepa Rd, and the marker’s response was to tell me why it could never work. It’s quite common now…

    2. I would be really interested to know what would happen if they ran everything in isolation in the peak. I’ll make sure I take the train that day.

      I totally agree that there is too much waiting on a side road 8pm at night for no reason. My biggest annoyance. At peak times it doesn’t matter because its clogged up in all directions.

  10. One thing I can say as an occasional driver especially inter-peak etc times around suburb parts of Auckland I notice just how fast & unhindered with perfect flow my drive is. It’s quite a contrast after doing more walking, biking, busing or training. Slip lanes & stuff make me think, wow this really is designed for driving. This is something in the past I would never have noticed. I was one of those people who would get annoyed by peds pressing a crossing, or stepping out not really seeing you or bikes holding me up for a few secs & traffic lights telling me to stop. Auckland’s car priority trained me well in this thinking (and perhaps the fact that I’m normally running late/last minute everywhere 😉 ). I also learned to drive in rural areas by and large so traffic was a novelty.

  11. Auckland’s traffic system is faaaaaar from perfect, but there are so many factors and so many trade offs to consider which most people aren’t aware of.
    It is fair to say pedestrians have to wait a long time some times. Not so fair to blame the system being used to say it can be solved with lower cycle times.

    ——– Purpose of traffic lights ————
    Traffic lights are there for SAFETY. Remove them all and traffic would flow much better. Along the main road. And forget about pedestrians, they get bridges or tunnels or nothing. And there would be so many more deaths and injuries.

    Traffic lights are trying to trade off efficiency with safety.
    The more you trade off for safety, the more you lose in efficiency.

    ——— pedestrian beg buttons ————–
    The reasons pedestrians have buttons is because their hands can reach the button. The beg button for cars is under the road. So not sure what what peoples point is.

    If you want the pedestrian phase to run automatically even if no one is there(like Queen St) by that argument should all the car phases run too if no one is there? Why do you run a phase you don’t need? That just makes the problem worse! Also consider the fact that pedestrians crossing phases are very long because you have to cater slow people as well. There is really a lot of time wasted catering for slow pedestrians for safety reasons.

    The trade off here is that the pedestrian crossing only runs if someone pushes the button which frees up time for other traffic (such as the following pedestrian phase)

    ———- Using US as an Example??———
    You cannot compare the US with NZ(or even the UK). It’s apples and oranges.

    ————-Number of phases——-

    In the US you usually only have two phases: North/South and East/West and the pedestrian crossings just run in those phases. Each phase gets 30s to fit the crossing in and you get a 60s cycle time. It is child’s play to run a low cycle time with just two phases.
    Most US cities have huge one way networks where the phasing again is extremely simple and actually much safer for pedestrians. (one way networks are why NYC was able to roll out a huge cycle network so quickly without enormous impact to car traffic)
    The trade off here is two way roads are more convenient for access, but one way roads are more efficient in moving traffic so you need higher cycle times to move the same amount of traffic.

    ———-turning arrows———-

    In the US they don’t often use turning arrows and they also are quite happy ban lots of turns everywhere.
    In Auckland we cater to every direction with safe phases for every turn in every direction and are reluctant to ban turns (ie Mt Albert). This means we have to have several phases at every section. So either each direction gets 5s of green or you have to have a cycle time ~120s to get something manageable.
    The trade off here is that we don’t ban movements because we want to keep cars on arterial roads and we don’t want to divert traffic down minor residential roads or encourage dangerous manoeuvres at other points along the road(ie u turns).

    In the US you also have the equivalent of Left Turn On Red, where cars can just turn on a red if the coast is clear. This is bad for pedestrians and cyclists. NZ doesn’t allow this for safety reasons, instead we get green arrows as the trade off of separating traffic from each other. This requires higher cycle time.

    —— Barnes Dances —-
    I would also point out that you would be hard pressed to find many Barnes Dance’s in the US because they are pretty inefficient. You can’t cut cycle times to 90s if you have Barnes Dances all over the place like in Auckland. I seriously think Auckland is a strange anomaly with so many of them and probably a result of our high cycle times. If you have one Barnes Dance along a road, it makes it really difficult to move stuff along the road.

    The trade off here is pedestrian safety at the price of higher cycle time to move the same amount of traffic.

    ——–Filter Turns——–
    The trade off here is safety against efficiency. Filter turns usually results in crashes at busy intersections. Christchurch has filtering everywhere because they don’t have the volumes of traffic we do. I’m also told that the old councils all had different policies for filtering because different engineers ran each one, so we still have some inconsistencies. Some areas never filter, some always filter, some by time of day. Removing filter turns means you have to give them a phase. Thus higher cycle times.

    —– Cycle phases —–
    Now we are getting all these cycle ways with their own phases for safety reasons which also adds cycle times.

    —-Other trade offs—–

    There are other trade offs which I won’t go into like intersection size, intersection geometry, crash patterns at an intersection, slip lanes, inter green times, pedestrian crossing times, pedestrians behaviour, sensor technology, broken sensors, road works, vehicle occupancy, green waves, coordination, intersection delay vs whole journey delay, lack of traffic engineers, incompetent engineers, Auckland Transport strategy. The list goes on and on.

    I’m sure you will note that except of sensor stuff, these trade offs have NOTHING at all to do with the SCATS system itself. Instead they are design choices made for various reasons and most these of why we have such high cycle times compared to some other countries. These are limitations placed upon SCATS. You cannot except any system to perform well no matter how smart it is when you have all these trade off and constraints placed upon it.

    1. Beg buttons – Perhaps people have come to hate them because they are so bad for cyclists. And since our cycling infrastructure is almost non-existent in most places, cyclists have to swap between car infra and ped infra and it’s a pain in the backside to have to try to reach a beg button from a bike.

      Barnes Dance – BD is necessary in a country where there are enough drivers oblivious to other road users. Change that, and we might not need the BD for safety. But by then it’ll be necessary for efficiency because with increased cycling mode share, BD is is really efficient. (See “Why We Cycle” for footage of its efficiency and discussion of the mental health benefits.)

      1. Push-buttons aren’t needed for cycling; we don’t use them in Christchurch, it’s mostly in-ground loop detectors and now some overhead detectors. On many of the Major Cycleways, the detectors are about 30m ahead of the intersection and the lights have changed for you by the time you get there (only a couple of busier roads where it’s not always so instantaneous).
        BTW, there’s a proposal to trial a shared ped/cycle Barnes Dance in Christchurch in a route currently under design. Shared mainly because it’s between two schools and I can’t imagine one kid on a bike waiting while their walking mate crosses (or vice versa).

    2. Re Beg buttons. The problem is not with the button, that’s fine. The problem is that there’s usually at least one phase every cycle (and sometimes more than one) where it is perfectly safe to cross as a pedestrian as there is no conflict. Simply light up the green person on those phases and everyone will be happy. The ped beg button then only prompts for those conflict-free phases, just like the car loops do.

      1. AT is onto it. The Urban Street and Road Design Guide says:

        “Opportunities to provide a walk indication should be maximised wherever possible. Vehicular movements should be analysed at every intersection in order to utilise non-conflicting phases to implement walk intervals. For example:

        1 Pedestrians can always cross the approach where vehicles cannot turn at a four leg intersection that includes a one-way street when the primary street has a green indication.
        2 Introduce concurrent pedestrian phases within signal cycles that also include an all-pedestrian phase.
        3 Introduce concurrent pedestrian phases at intersections that have slip lanes and an all-pedestrian phase.
        4 Use double phase Barnes dance (two pedestrian phases each cycle) where long cycles cause excessive delays for pedestrians.”

      2. You normally need either a one way steet, a slip lane, or a staggered ped for there to be a phase where a ped can run every cycle without conflict.

        But lets say there’s no ped waiting for the fully protected crossing, but a ped waiting to cross another part of the intersection. If the protected ped is set to walk for green (ie runs each phase) then the minimum phase time increases because you can’t terminate the phase until the pedestrian crossing has completed it’s red man, so the pedestrian who’s on the other crossing gets delayed.

        The common way around this is ped reintroduction, which is commonly used all over Auckland.

        Slip lanes are an interesting topic in themselves. When you take one out you lose the ability to reintroduce peds and you generally decrease pedestrian LOS based on time. With a slip lane, a ped has to cross an unsignalised slip lane crossing (usually a single lane) then gets a low delay fully protected crossing for the rest of the journey. Without a slip lane the pedestrian gets a less frequent crossing with only partial protection. I sometimes wonder if those who advocate for removal of slip lanes actually understand the effects of the removal. I know which one of those I’d prefer my kids to cross the road at, and it’s at the slip lane.

    3. “You can’t cut cycle times to 90s if you have Barnes Dances all over the place like in Auckland.”

      Of course you can Ari. You are just making a generalised value judgement based on your view of relative usage and wait times. You’re simply assuming the delays to traffic would be unacceptable, and I assume you’re not considering or counting pedestrian delay in any way. But in many cases the traffic engineers default view that all the people are driving and anyone crossing on foot is just a disruption to the major flow can be quite wrong.

      Take Queen Street for example. At any of the barnes dance intersections there are ten times more people crossing the intersection on foot than driving through it in vehicles. That’s not figure of speech, it is literally ten times more pedestrians that drivers.

      So indeed, you could easily set the centre of the CBD up on 90 second cycles with Barnes dance crossings and you’d be reducing net delay overall. With 30 seconds of ped phase and 60 seconds for all the car phases, you’d still be over prioritising traffic sevenfold. In fact you could swap that and double phase the barnes dance with 60 seconds for ped phase and 30 seconds for traffic per 90 second cycle, and you still wouldn’t be optimising for minimised net delay.

      1. Last time I looked, Queen St had 2 Barnes Dance per cycle and very low vehicle green times. Priority on Queen St (at least in the lower section) is 1/Peds 2/PT

        Despite what the conspiracy theorists on here wrongfully assume about general vehicles being prioritised in Auckland over other modes, in most town centre intersections, the only reason general vehicles get any priority is because PT requires priority and PT and cars are using the same lanes. You can’t move a bus stuck in a line of cars without moving the cars (or without providing a bus lane).

        1. Actually they have cut the second barnes dance per cycle, due to hand wringing over traffic flows around the CRL construction works. Sorry but they literally took out pedestrian phases at the busiest pedestrian intersection in the country to try and make cars flow better.

  12. Here’s a dangerous situation. I ambled up to a ped crossing. No red indication telling me I must not cross. Another pedestrian was waiting on the same island to cross at 90 degrees from me. Just as I reached for the button, the buzzer sounded so I stepped off and was almost struck by a car approaching at 40 km/hr + from my right which had a green light. It turned out that the buzzer had been triggered by the other pedestrian, and because I hadn’t yet pressed my button, I got no indication that crossing was forbidden. I think the lack of a red indication at all forbidden times is a cost cutting measure by Council. There are occasional news reports of ped injuries and deaths in the city. I would like to know how many are due to this.

    1. That’s a particularly dangerous situation, given you didn’t have a red. But the audio signals sounding for each direction are a danger even without that. Especially for kids. And especially for kids on bikes, when the adults can’t hold their hands. I wonder if there is any research money going into solving this? Speakers in the middle of the crossing? Different audio sounds? Light bars set into the footpath? And Barnes Dance wherever this isn’t possible?

    2. It was done to save power in the 70s. Buzzers are for the visually impaired, not for other people. Everyone else must look for a green man and check that it is safe to proceed. When you push the button the red man comes up which intuitively tells you that your request was registered. In Australia you could stand around for ages before realising that you need to push the button.

      If you had the red man up all the time to forbid crossing, then you would never know if your request went through or not when you pushed the button.

      If the red man light doesnt outright work, then that needs to be reported.

      In the US they have two different noises for the two crossing from the same corner to reduce this problem but I dont know of any research into the scale of the problem.

      I think a few people have been run over by cars running the red, but I have not heard of a case in Auckland recently where someone got run over by accidentally crossing at the wrong time. I have seen it several times though. Zombie people on their stupid phones.

        1. Yep, also people in conversation.

          I noticed yesterday that the pedestrian crossings on Richmond/Ponsonby road have gone silent, at least on some legs. That was a spot where you’d frequently get people mistaking the buzzer for their crossing

      1. I wasn’t listening to my phone in a zombie like state. I was alert but this lack of an explicit Don’t Cross is dangerous. I wonder whether we still need to save power with LED lights.

        1. It’s not forbidden to cross if the light isn’t red actually. You can cross whenever you like if you don’t push the button!

        2. Yep, nice to be able to legally cross when there’s no traffic rather than waiting for the lights to cycle. Unintended benefit of being so cheap as to not be able to keep a light bulb lit.

        3. It’s not about the keeping the lightbulb lit, it’s about telling you if the beg button has been pushed or not.

      2. I got caught out as described in Australia last year. Wandered up, saw that the red man was already displayed so didn’t push the button. When the traffic stream next to me went green but I still didn’t have a green man, I realised that it didn’t give you a ped phase automatically unless someone pushed the button – and you didn’t know from the display if that had already happened yet…

      3. I am used to the ‘blank until activated’ behaviour and appreciate the visual feedback acknowledging that the pedestrian crossing will be triggered.

        What I don’t like is that bicycle signals appear to use the Australian style ‘red by default’, which leads to the missed phases you described.

        1. Only a problem if they don’t have bike detectors – which should be standard these days; the technology is straightforward.

  13. In central Melbourne (Hoddle grid, six-lane streets) most intersections have 80-90 second two-part cycles. In each 90 seconds pedestrians get 20 seconds of green man and 20 seconds of flashing red man. Right turns filter (actually hook turns to avoid holding up trams, but the concept is the same). It works.
    As a pedestrian strolling in inner Melbourne (20 seconds of green man each 90 seconds) it took me a while to realise why it was so much more pleasant than Sydney (typically, 10 seconds of green man each 120 seconds).

  14. AT’s Urban Street and Road Design Guide says “Best practice guidance suggests that pedestrian wait times in urban areas should be no longer than 30 seconds.” And they quote NZTA for this.

    Looking forward to these changes, AT. I’m assuming you want to use best practice, and not just continue with car bias?

    1. That 30 sec is clearly wishful thinking. The Victoria/Albert at some stage was 105 seconds, Symonds St by Wellesley Overbridge – close to 120 seconds (I think I timed 140sec at one stage). What’s particularly frustrating is that generally there are 3 stages at each intersection:
      1. Hardly any traffic – lights painfully cycle through empty phases, but pedestrians have to wait for their ‘start’ signal, since only the backup-up pedestrians are allowed to cross, if you arrive later – tough luck, wait another 2 minutes for your chance.
      2. Regular traffic – just the regular long wait for pedestrians
      3. Heavy traffic – cars backed up through the intersection, nothing is moving. Pedestrians wait even longer than usual as the lights are trying to clear the intersection.

      Whether that’s incompetence on AT side or wilful action – doesn’t matter, it forces many people on daily bases to take chances with traffic. As a regular PT user I found that nothing is most frustrating that seeing your bus leave because you happen to be on the wrong side of the intersection.

  15. Our traffic light phase is inefficient during off peak. The induction detector can only detect either have cars waiting or not. It doesn’t detect how many cars are queuing. So it is often one car on a intersection is given same priority over long queue of waiting cars that has momentum.

    The traffic light needs to be able to detect how many cars are waiting, what is their current speed/momentum, and also how many people are waiting in the intersection. I think all of these can be done using camera and motion detection software. With these information, a more efficient light phasing can be done om realtime.

  16. Walking around CBD a thought occurred to me that some intersections seem to be backwards in their car vs pedestrian priority.
    Some intersections have far more pedestrians than cars but require pedestrians to wait.
    Is there any scope to flip this so that pedestrians have right of way, and cars have red lights that trigger the phase change when they drive up?
    Eg the pedestrian crossing behind Britomart.

  17. In some other cities such as Vancouver, traffic lights become giveaways during offpeak by flashing orange, this is more efficient on suburban road when there is not a lot of traffic.

  18. There seems to be a frustrating feature that was introduced some years ago whereby there is a dead-time for the loop signal to trigger a right-turn arrow. For example any vehicle arriving within say the last 5 seconds before the right-turn cycle begins, does not trigger the right-turn arrow.

    It used to be that arriving in a right-turn lane any time up until the right-turn cycle began would trigger a turning arrow.

    So you arrive in a right-turn lane just before the cycle begins, you then sit on a red arrow, and watch the cars coming in the opposite direction get a right-turn and a straight through green, and you need to sit through a full cycle before you finally get to turn.

    The logic for this one escapes me.

  19. Why do we accept many of the “dumb” decisions Auckland Transport make in general – they are the single biggest recipient of ratepayer funding in Auckland and their decisions affect everyone every day in Auckland. Traffic lights are but one of the many transport issues in Auckland needing attention but are certainly one of the primary causes of traffic congestion in Auckland.

    Traffic lights stop the traffic flow which causes traffic to bank up and causes congestion. Many roads which have traffic light controlled intersections would work much better with roundabouts, which slow traffic down but keep it steadily moving and are safer.

    This link below to the Washington State Department of Transportation explains the benefits of roundabouts:


    This link to youtube shows a short video which explains the benefits of roundabouts in a simple, easy to understand manner:


    This link to youtube shows a short video showing a before and after example of an intersection in the US which has been converted from traffic lights to a roundabout, clearly demonstrating how much better the traffic flows with a roundabout:


    In Auckland, there are many people who want roundabouts but Auckland Transport’s traffic engineers and their predecessors seem intent on only installing traffic lights. There were some very heated public meetings over Auckland Transport’s plans to install the first set of traffic lights in Pukekohe not so long ago, which the people of Pukekohe have long opposed with the former Franklin District Council having a policy of installing only roundabouts which work very well.

    Another example was on Tamaki Drive in central Auckland where locals wanted Auckland Transport to install a roundabout at a problem intersection on this busy road, but Auckland Transport disregarded this and went ahead and have installed traffic lights:


    In Warkworth, NZTA continue to stand by the use of traffic lights at the notorious State Highway 1 / Hill Street intersection, where the traffic lights are clearly the cause of the problems, despite a former traffic engineer planner having come up with a double roundabout solution which he says will sort the problems here:


    One of the worst places for poor use of traffic lights which are frustrating for both motorists and pesdestrians, is Papakura. The main north-south route Great South Road has traffic lights at nearly every intersection along it through the Papakura town centre, as does the bypass route around the town centre. Roundabouts could replace nearly every set of traffic lights in Papakura and would work far more effectively, together with separate stand alone ‘zebra’ pedestrian crossings to cater for pedestrian movements, resulting in much better traffic flows and travel times.

    There needs to be a shift away from the use of traffic lights if Auckland’s traffic congestion is to ever be seriously addressed (together with providing more train and tram services across the region). It has been internationally proven that roundabouts move more traffic through them in the same time period as one controlled by traffic lights, they are safer, more cost effective and efficient and require less maintenance and don’t rely on electricity to operate.

    Pedestrian movements could be better catered for on busy roads and high pedestrian areas with ‘zebra’ pedestrian crossings, which give pedestrians priority and only stop the traffic flow when there is actually pedestrians waiting to cross, meaning (often impatient) pedestrians don’t have to push a button and wait as they would with traffic lights and don’t end up making motorists needlessly wait at red lights after the pedestrian has decided to cross before getting the cross signal when no traffic was coming.

    1. For some reason lots of pedestrians and cyclists get run over a lot at roundabouts. Great for cars, unless the traffic flow is unbalanced then only one direction gets to go the whole time.
      Zebras can have poor safety records if vehicles are speeding a lot and you can’t do really have them at a busy intersection.

      1. I’ve actually copied Robin’s comment to use to demonstrate that although decades of investment in roads for drivers means a major shift is now required, there are people who continue to try to make decisions on behalf of pedestrians and cyclists from behind their steering wheels. I really appreciate the time he took to put it all down. Sometimes people working in transport forget the need for education.

  20. Well I for one am truly stunned we have any system for our traffic lights because I always thought some “traffic engineer” took a lucky dip approach to settings along with Auckland Councils cheapo parts bin special light systems that appear to be made in North Korea. And it is in any case that the lights sure as shit do whatever the hell they want anyway, to the continual detriment of traffic flows.

    Sit at red lights with the conflicting lane on green and empty forever

    Sit a a red arrow knowing that despite the opposing lanes being empty you have to wait for the entire cycle to do its thing again.

    Wait an eternity at a turning arrow only for it to let through 3 cars, one of those on a yellow to red. Right turns from Khyber Pass to Park Rd are hopeless for short phasing that never clears the backlog, buses most especially included.

    Approach a green light on otherwise empty roads only for it to go red and yet find the conflicting lane now has a green on empty lanes. The no demand green light is a signature of Auckland’s traffic lit intersections.

    Leave on a green only for the next intersection to go red. Customs St is a classic for that one.

    In short our useless traffic light logic, if it can be called that, will be a very good reason why there is so much red light running.

    Thing is if we had an automatic give way to pedestrians on left turns without the constant infantile need for traffic lights everything would flow all the better. And left turns on reds or no reds when there is no need to wait as per most other mature countries.

    And more and better designed roundabouts. And if you cannot handle a roundabout, hand your licence back to NZTA!

    1. Customs St: if peds are prioritised in town centres areas then it’s a difficult argument to get progression on a road corridor through a highly pedestrianised CBD area.

      And the issue on right turn Khyber to Park. Priority to PT trumps general vehicle.

  21. Another related issue with traffic lights is the NZTA’s use of on ramp signals to motorways across Auckland, which has been a disaster with traffic banking up back into suburban streets, which then impacts negatively on the surrounding local neighbourhoods with the noise and pollution from idle traffic, as well as making it difficult for people to get around local streets and rat racing occurring.

    The NZTA decision to install traffic lights on the new Waterview section of the State Highway 20 South Western motorway connection was met with much opposition and negative reception:


    Putting traffic lights on motorways is just traffic planning madness. The whole idea behind this incredibly expensive motorway tunnel project was to provide a seamless alternative motorway route with the Western Ring Route which would be a free flowing alternative to the congested State Highway 1 route via Spaghetti Junction through central Auckland.

    The NZTA have already created a congestion nightmare with the traffic lights they installed at the southern end of the State Highway 20 South Western Motorway section of the Western Ring Route where it joins the State HIghway 1 Southern Motorway at Manukau. The congestion regularly banks back as far as the eye can see on all southbound approaches of both motorways as well as the southbound on ramp from Redoubt Road, and both southbound and northbound all the way south to Papakura for much of the day now.

    Well designed motorways in the UK and Europe have roundabouts at the end of off ramps rather than traffic lights in NZ, which keep the traffic flow moving

    Both the South Eastern Highway and Te Irirangi Drive don’t flow as freely or work as effectively as they could do due to traffic lights being installed at all the intersections along them, rather than having grade separated intersections which they ought to have. Both these key arterial routes need to have their traffic lights replaced with grade separated intersections (along with installing light rail along the middle of Te Irirangi Drive).

    With Te Irirangi Drive upgraded into an expressway with no traffic lights and light rail along the centre, a seemless cycle path could also be developed alongside of it between Botany and Manukau.

    1. The ramp lights on Fanshawe St onramp in the morning are so disruptive they cause a queue back to Quay St that takes 20 to 30 minutes to go from there to the motorway.

      It is astounding how NZTA’s desires completely stuff Aucklands traffic flows!

  22. I once worked with a chap who said he worked on programming SCATs.

    He told me it was all nothing but a massive bunch of totally obsolete spaghetti FORTRAN code, and this was in the days when FORTRAN and spaghetti code were long since frowned upon and no longer talked about at all – well at least, not in polite society.

    I’m sure its been modernised a lot for basic system supportability reasons if nothing else since then.
    But like those hopelessly out of date traffic models that AT always uses for its general traffic and PT modelling. That show traffic growing forever without constraint and no ones using PT at all.

    You suspect that deep down, all those original bad design decisions, trade offs and whatever else still “lurk in the basement of the machine”. So its simply become ingrained into its DNA.

    SCATs has long since passed the point of diminishing returns with regards optimising traffic flow for each intersection. Now all its basically doing is letting AT rearrange the deckchairs on the Titanic, as they wave their collective hands at the problem all the while allowing them to hope it all goes away – once some radical future tech solves the problems for them.

    I always wondered why we don’t have traffic light countdown signs to show how long we have to wait for the next green. Since clearly SCATs knows this well in advance.

    If we did, then I’m sure the red light running frequency would massively increase once motorists realised the truth of how long they were having to actually wait at each red light!

    1. I think it is still dodgy FORTRAN. Its what you get when some traffic engineers in the 70s decided they wanted their own traffic system and wrote their own code. Half the aussie states tried to develop their own systems too.

      The way I understand it is that you can’t use countdown lights with SCATS for vehicles unlike pedestrian crossings with a set crossing period that doesn’t change. The green time given to a phase can vary at any second. It could start at 20 then suddenly drop to 5 and end the phase early. Or it could jump to 30 because the next expected phase isn’t needed. Same thing with the red. It might start at 60 then drop to 30 then drop to 0. Or I think in a worst case scenario it could just sit on 30 if a late bus or the prime minister is coming through with some lights being held on green.

      So unlike some 3rd world countries that just have fixed timers and no sensors you cant show that number because it would jump around all the time. It provides little benefit to drivers other than to encourage them to race. But I can see how if you had a counter, drivers could see how long they are actually waiting.

      1. Those “third world countries” include large parts of the US that simply operate fixed time signals (SCATS, SCOOT, etc haven’t made much of a dent there); hence they can display a true ped (and driver) countdown time.

        As you say, SCATS can dynamically adjust phase times on the fly depending on traffic arriving. So for example, the phase might have got down to the final 2 seconds before it plans to change, and then another car shows up and extends it for (say) another 5 seconds. So a countdown timer would be jumping all around the place.

        The other factor that makes our signals sometimes seem a bit dumb is that, by and large, the SCATS detectors are right up near the stop lines. You as a human can see that there is no traffic coming for 200m (or that there is a another vehicle just approaching the intersection), but SCATS is essentially blind until those vehicles get to the intersection, so its logic has to “back-calculate” somewhat.

  23. I can’t understand why i get orange lights over 5 times in a row when heading straight down a main road. Am I not going fast enough between intersections? I’m usually doing a little over
    50km h
    And I know when I get the 1st red light I will also get the next 4 red lights after.
    Can’t lights work as a network and group cars together especially when heading along main roads.
    And drivers turning on to the main road will wait at the light and then get grouped at the next so there will be no more orange lights 5 times in a row.
    Also we need timers instead or the orange lights.
    Maybe roundabouts with lights to help even the flow when you have a busy road crossing a less busy road

    1. They can be synchronised (and plenty do; come and visit the one-way green waves through central Christchurch for example – operating like that for over 40 years). But like all things it requires someone to program them to do this. You also have to appreciate that if you have a number of different corridors all intersecting, it will be well nigh impossible to maintain a green-wave on all of them at the same time (esp. for those drivers who are at the very start or end of their green phase).

    2. Could it be that you’re going against the main flow of traffic? The green wave can only be timed in one direction.

      1. No, it is possible to have green waves in both directions of a road concurrently; it just starts to get trickier when you have to factor in green waves for crossing roads too…

  24. Stupid things about NZ’s traffic lights that I would change:-

    1) Buttons that not only make you beg, but make you wait ages too:
    I would alter the ‘beg’ button logic so that the pedestrian phase initiates as soon as the button is pressed, after which a time-delay must elapse before the next ped phase can begin. This is how it’s done in many countries.
    But the stupid NZ way is for the button to initiate the time-delay first, and this has to elapse before the ped phase can begin. Of course the peds get tired of waiting and dash across before their phase, then when the traffic finally gets stopped there are no peds there.

    2) Bicycle detection that doesn’t work:
    Where a minor road intersects a more-major road, the traffic lights will often hold the minor road forever at ‘stop’ until a ground loop senses a vehicle there. If that vehicle is a bicycle the odds are it won’t be detected and the lights will never change. This is particularly common at night when there may be no motor vehicles on the minor road to properly trigger the loops.
    The cyclist must legally wait for possibly ages until a car comes along to trigger the loops. Or else just go through the red.

    3) Traffic that is permitted to drive through a ‘green-man’ pedestrian phase:
    In many civilised countries, a ‘green man’ indication will guarantee to pedestrians that traffic is barred from proceeding through their crosswalk.
    But not in NZ. Here it is common for turning traffic to be given a green light through an already active pedestrian phase. Two green-phases on a collision-course. How stupid is that? The only ‘protection’ provided for pedestrians may be sign under the big green light saying “Turning traffic give way to peds”, if it gets noticed or is even there.

    4) Red filter arrows:
    You’re a driver waiting at a big red light which suddenly goes green. You instinctively move off, but then belatedly realise there is a little red arrow prohibiting your move.
    Safety 1.01: Never should the “go” indication be more-prominent that the “stop”.
    In less stupid jurisdictions, the big light would stay at red until it was safe for all to go. Permission to filter would be given by a little green arrow, a failure to observe which would simply prompt a honk from behind rather than a potential collision. Dumb dumb dumb..

    1. You seem to be talking about things that can be done as best practice but are not always, probably due to lack of understanding/training by the relevant signal designers. E.g. bicycle detection at an intersection is a relatively trivial problem of loop tuning and appropriate on-road marking. Mind you, an answer to problem 3 is to have a red arrow first, which you then complain about in problem 4…

      1. GlenK, I’m not sure you have understood my concerns about the red arrow.

        I am suggesting that if *any* traffic movement is to be held at stop while other movements are permitted to go, the aspect of the major lantern should be red, with permitted movements receiving a minor green arrow only.

        What I am concerned about is where the aspect of the major lantern is green, even though some movements are not permitted to go, and only a minor red arrow communicates this.

        A small red arrow beside an overall green is much easier to misinterpret dangerously than a small green arrow beside an overall red.

        No difference in the permission given to traffic, just communicated in a way that is less-hazardous if misinterpreted.

        1. I agree. I’ve been in cars when people new to the country or back from OE in places where this doesn’t occur have mucked up and not noticed the red.

  25. Auckland’s traffic congestion problems are largely due to the overkill default use of traffic lights at intersections instead of using roundabouts. Traffic lights stop the traffic flow. They are more dangerous and are more costly to operate than roundabouts – and are a nightmare when there is a power cut and the lights go out.

    If you look around cities and towns in New Zealand such as Hamilton, Rotorua, Taupo, Pukekohe, and around the world particularly in Europe and the UK, which primarily use roundabouts instead of traffic lights, the traffic flows much better.

    Even the NZTA are now installing roundabouts at major intersections on State Highways in rural locations, because they are safer and more practical. But in Auckland they put traffic lights on the on ramps and off ramps which cause the traffic flow to stop and bank up.

    In Auckland, there are many people who want roundabouts but Auckland Transport’s traffic engineers and their predecessors seem intent on only installing traffic lights.

    There were some very heated public meetings over Auckland Transport’s plans to install the first set of traffic lights in Pukekohe, which the people of Pukekohe have long opposed with the former Franklin District Council having a policy of installing only roundabouts which work very well.

    On Tamaki Drive, locals wanted Auckland Transport to install a roundabout at a problem intersection on this busy road, but Auckland Transport disregarded this and went ahead and have installed traffic lights:


    In Warkworth, NZTA continue to stand by the use of traffic lights at the notorious State Highway 1 / Hill Street intersection, where the traffic lights are clearly the cause of the problems, despite a former traffic engineer planner having come up with a double roundabout solution which he says will sort the problems here:


    One of the worst examples of poor use of traffic lights is in Papakura. The main north-south route Great South Road has traffic lights at nearly every intersection along it through the Papakura town centre, as does the bypass route around the town centre. Roundabouts could replace nearly every set of traffic lights in Papakura and would work far more effectively, together with separate stand alone ‘zebra’ pedestrian crossings to cater for pedestrian movements, resulting in much better traffic flows and travel times, and making it quicker and easier for pedestrians to cross roads rather than having to wait at traffic lights.

    1. The idea that you somehow have all the definitive answers to a problem that has been debated by traffic engineers for years is laughable. There are pros and cons for both, traffic lights generally handle high volumes the best, and use less land.

      Personally I would replace a lot of Auckland’s lights with Dutch style pedestrian and cycle friendly roundabouts, but I don’t think that would be popular once people saw the impact on traffic flows.

      I’m not sure the examples you give of rural intersections and smaller towns have any relevance to Auckland as they will generally be in places with much lower traffic volumes and cheaper land costs. Also the fact that Pukekohe has a policy of no roundabouts doesn’t automatically mean it is a sensible one.

  26. In the US, which has long been a country with a history of almost extensively using traffic lights, they are now moving to converting traffic light controlled intersections to roundabouts for the many benefits they bring.

    This link below to the Washington State Department of Transportation explains the benefits of roundabouts:


    This link to youtube shows a short video which explains the benefits of roundabouts in a simple, easy to understand manner:


    This link to youtube shows a short video showing a before and after example of an intersection in the US which has been converted from traffic lights to a roundabout, clearly demonstrating how much better the traffic flows with a roundabout:


    There needs to be a shift away from the use of traffic lights if Auckland’s traffic congestion is to ever be seriously addressed (together with providing more train and tram services across the region). It has been internationally proven that roundabouts move more traffic through them in the same time period as one controlled by traffic lights, they are safer, more cost effective and efficient and require less maintenance and don’t rely on electricity to operate.

    Pedestrians can be better and safely catered for on busy roads with dedicated ‘zebra’ pedestrian crossings where necessary. This means the traffic flow only has to stop when there are actually pedestrians waiting to cross. Zebra type pedestrian crossings are also better for pedestrians as they do not have to wait for the traffic lights to change, as they can cross as soon as the traffic stops with pedestrians having right of way on pedestrian crossings. This means avoiding the unnecessary and frustrating delay to motorists when an all too common impatient pedestrian presses the button to cross at a traffic light controlled intersection and then crosses when there is no traffic coming without waiting for the cross signal, leaving motorists needlessly stopped at a red light for the pedestrian crossing phase.

    1. Where would you locate these pedestrian crossings relative to the roundabout? Would you put them right on the edge of the roundabout or say 50m down the road? If it were the former I think you would be surprised how much impact this had on the traffic flow, if it were the latter you are asking a pedestrian to walk 200m extra just to get to the opposite side of the intersection.

    2. Bear in mind that historically the US has used signals in places where a roundabout would have been more appropriate in the first place. That doesn’t mean that all signalised intersections are bad; there are plenty of busy ones that would just lock up if you tried to convert them to a roundabout (esp. at peak hours). And multi-lane roundabouts still present all kinds of awful for any peds or cyclists using them (unless you spend a lot on grade separation). Horses for courses; the US finally learned to use a different horse…

    3. If everyone walked, cycled or took public transport, there would be virtually no congestion. There would also be hundreds of lives saved every year. So trying to make cars move more quickly doesn’t solve the ultimate problem of too many cars and not enough land/money to build more roads.

      Traffic signals are for safety. They are far safer for pedestrians and cyclists. Roundabouts only work below a certain volume of traffic, then they grind to a halt. If you have a constant stream of pedestrians, the roundabout grinds to a halt. If you have unbalanced flows of traffic, the roundabout grinds to a halt. This doesn’t happen with traffic lights because everyone gets their fair share of time. It may be more inefficient with time, but they are generally safer and balance things out.

      If we had the laws like in Holland where the driver is always at fault if they hit a pedestrian or cyclist, then I guess Auckland could put more roundabouts in.

      1. The driver is not always at fault if they hit a pedestrian or cyclist in Holland. That is a myth. It is true however that if your car hits a cyclist or pedestrian you’re usually liable for any damage. See https://bicycledutch.wordpress.com/2013/02/21/strict-liability-in-the-netherlands/

        This is a cultural difference. Over here it’s the most mundane thing to be on collision course with a pedestrian with the throttle wide open. In Holland, and in most of the surrounding countries, you’re expected to avoid collisions with pedestrians.

  27. I worked for Auckland Council in the mid 70’s and SCATS was just being installed then in a minor format, its been developed of the years and improved in terms of the inner city, however the management of traffic flow and intersection management in greater Auckland is a dammed disgrace and defies description.

    Traffic flows aren’t properly manged in greater Auckland, it it all happens organically or by accident. Phasing of signals is antiquated and is completely out of sync with the traffic flows at any given time, the traffic signals are of no help at all as they are operated, they are an impediment to sound and safe traffic management.

    We’d all like to think someone at Auckland Transport would be reading the sort of material shown below and working proactively to solve the chaotic traffic congestion in greater Auckland given the ratepayer funds AT sucks up each and every year at an indecent rate, however its apparent to all but the blind they aren’t and the traffic congestion is being added to and compounded by AT as the simply cant deal with it they have lost the plot and the “where with all” to drag the cities traffic management from the early 20th Century into the 21st Century.

    The other key thing we need to do is get people in from places who know what the hell they are doing and show that by how their cities function traffic management wise and let them show us how to do it correctly, clearly we aren’t so its about time someone grew up and asked for help !

    Remind me doing the same thing over and over again and expecting a different outcome, is what ?????

    Some reading links below that might be of interest and help, no one and I mean no one has cornered the market on good ideas and it seems plain crazy we aren’t tapping into the good ideas that are working and copying them






    If we don’t act now and sort this cities traffic shambles out before 2020 we will be just another chaotic third world farce in that regard.

    Rgds, Grant

  28. As for the plans to get us all out of our cars – it isn’t going to happen people, not unless you live and work in the Inner City. It takes us a decade to build a few kilometers of an underground rail loop for goodness sake, it will be 50 years before Auckland has the brains to build a mix of heavy and light rail that works for us all, so how do we all get around until then ? On Scooters I suppose – the family that scoots together gets wet and run over together ?

  29. Wow lots of comments.
    Question to ask is where do people get trained to deign signals or operate them using Scats?

    RMS, suppliers of Scats have not ran a course in about 10 years and even then it’s rubbish.

    At uni how much design work is done here? any courses? i understand they are all theory and don’t cover much especially in the real world!

    I got trained in UK where they have many courses all year for everything, had people around me who had 20 years experience at least. We had 20 staff that covered larger population than auckland, double the traffic and did ALL our own designs too. While in auckland they have too many people but most under 5 years experience, same in Christchurch. all the guys that have have left the operations centres or are too old!

    We don’t design well at all in NZ, just leave it up to the operation teams and Scats to sort out issues, we retrofit sites later on as we have no idea what we are doing.

    Movement based design and operations is the way to go, we don’t do that with Scats and hence why we have a poor system.

    If the foundations are not good it just falls down…

  30. Drive in the biggest cities in Europe who actually have the latest intuitive traffic light systems (The Netherlands & Germany to name but two) and you will instantly understand how antiquated SCATS is and how backward AT is as they pretend they are managing traffic flows across greater Auckland.

    It’s all smoke and mirrors in reality

    Traffic management in Auckland is a shambles and given the money drivers of cars and trucks pay the returns in this regard are laughable

    These are the sort of people AT and NZTA need to be talking to – see below


  31. I think you will find most of these countries actually have very poor fixed time systems! But have finally gone modern. In the system used in NZ Scats it is getting an update this year and will overtake systems like this once again.
    The performance of a system is not always at fault especially in NZ where other modes are poor and have had little investment until recent times but is kind of too late.
    The take up of other modes is huge in Europe, like buses and bikes, has been for years.

  32. The biggest required change is much more fundamental. It is to derate the parameters for maintaing vehicle flow in favour of enhancing the parameters that provide pedestrian amenity. It is this general lack of pedestrian amenity that is inhibiting so much possible gains in the quality of our city.

  33. I think this article overstates the “dumbness” of the traffic signals… The entire point of SCATS is to optimise the flow of all traffic through an intersection, the greatest emphasis on being “maximum throughput” as opposed to “quickest time”. Although these two phrases sound like they imply the same thing, they don’t. An intersection that will get more traffic through it by having longer phases because there is a brief period of a few seconds between each phase which the intersection is largely at a standstill. However this does mean that users at the front of the queue on the phases that are “on red” (notably peds) will likely have longer wait times, especially if they’ve only just “missed the green”.

    The other important thing to remember with SCATS (or any other similar system like SCOOT etc) is that it cannot manufacture intersection capacity. All it can do is optimise the time available across different phases based on the traffic/peds detected. Perhaps the irony here is that when an intersection reaches capacity, SCATS won’t help at all, and in some cases will make the problem worse by extending phases to unreasonable levels which indeed will cause the safety issues of peds just running across between cars etc.

    On the whole, I’ve found most of the major NZ cities traffic signalling to be pretty good… as long as it’s not rush hour because during that time, many intersections will simply be over-capacity and the only fix for that is to either spend lots of money on upgrading the intersection (more lanes), or modest improvements like reducing the number of phases (eg: turn restrictions).

    1. “The entire point of SCATS is to optimise the flow of all traffic… However this does mean that users at the front of the queue on the phases that are “on red” (notably peds) will likely have longer wait times… Perhaps the irony here is that when an intersection reaches capacity, SCATS won’t help at all, and in some cases will make the problem worse by extending phases to unreasonable levels which indeed will cause the safety issues of peds just running across between cars etc.”

      Maybe a better way to put it is that the entire point of SCATS is to optimise the flow of all vehicle traffic, rather than “all traffic”. Pedestrians, ill-counted, and ill-valued, are traffic too.

      “many intersections will simply be over-capacity and the only fix for that”… is to reduce vehicle traffic volumes.

      1. Reduce vehicle volumes?! That’s too radical a suggestion Heidi. That’s like car genocide! That’s illegal!

        It’s true though. SCATS can’t manage what it can’t count. Having said that, if they are lucky enough to have a crossing at an intersection pedestrians do get a lot of time to cross the road. So proportionally they probably get their fair share of time because of the sheer volume of people in cars/buses and so few pedestrians. When you average out the delay, it’s pretty fair.

        Of course even though it isn’t measured, pedestrian delay can be estimated to be pretty high all across the city because pedestrian delay is equal in value to driver or bus passenger delay. If AT valued pedestrian delay higher than other modes, they should reduce that delay for pedestrians. I’d argue that pedestrians should be valued the same, except at town centres, schools and major PT hubs where they should get significantly more priority over other modes.

        The only place I can think of where the pedestrians don’t get a fair share of time is probably on Queen St where they should probably get 90% of the time and only get half that.

  34. Its not the dumb signals that is the problem its the culture. We all drive and never like to use other modes hence why we struggle as most roads are over capacity. Mad max movies say it all, we love petrol, Holdens and Fords!
    Also in NZ there is no training for signals in operations or design of so no wonder we are where we are. Scats don’t even offer training! for a software provider that’s really stupid..

    1. “we love petrol, Holdens and Fords!”
      I don’t know what circles you move in. But I’ve never loved any of those things and neither does anyone I know.
      I know a fair few automobile enthusiasts. But I don’t know any with any especial love for Holdens and Fords. Holden won’t even exist anymore in 5 year’s time.

      Maybe I’m wrong but I don’t think that driving is as much culturally-ingrained as people think.

Leave a Reply

Your email address will not be published. Required fields are marked *