Post implementation reviews (PIRs) are an important step to seeing whether projects achieve what they were intended to and to learn what can be improved for future projects. Every year the NZTA conduct PIRs on a sample of projects and for a few years now the NZTA have been publishing the results online. I’ve covered them before here and last year Peter has looked at the trends in them here. Now the results of from 2012/13 and 2013/14 are online and they make for interesting reading. Below are some of the interesting points I’ve picked up from a few of the ones I’ve read.

SH20 Manukau Harbour Crossing

This project duplicated the existing Manukau Harbour Bridge allowing for 8 lanes of traffic plus bus shoulders as well as widening approach roads and a few other changes. The PIR says the business case for the project forecast that 97% of the benefits would come from travel time savings, 1.6% from reduced crashes, 1.3% from vehicle operating cost savings and 0.1% from reduced emissions (quite how you can get emissions reductions when traffic volumes increase is beyond me).

The report says travel times between the airport to Hillsborough Rd were forecast to reduce by 43% however based on their estimates actual travel time savings are more like 37%. They then that’s a good result as traffic volumes increased by 14%. Unfortunately that in itself is way less than expected as shown by the graph below which looks remarkably similar to some other predictions we’ve seen.

Manukau Harbour Crossing Traffic Volume actual vs prediction

Traffic growth predictions ended up ~45% above the actual volumes of 102k vehicles. I can only assume that had traffic predictions been met that that the travel time savings predicted might have evaporated away. While a few reasons are given for the traffic predictions being wrong, the main blame is aimed at the traffic models and the assumptions within them.

There are some positives though, safety has improved with the annual crash rate falling by 28% which is above the regional average of 24% – although they note this isn’t statistically significant. The biggest positive is the cost which at $218 million was 19% less than estimated ($268m) which is in part due to the project being completed seven months early.

SH20 Mount Roskill Extension

Closely related to the Manukau Harbour Crossing and built partially at the same time, the Mt Roskill Extension saw SH20 extended from Hillsborough Rd by 4.5km though to Maioro St. Again the vast majority of expected benefits (99.34%) came from travel time savings.

The report notes that compared to the 2011 traffic predictions, the actual volumes were 9% lower – although that has grown in 2011-13. Importantly these motorway projects are often partially justified as taking traffic off local roads. The report notes that of the limited monitoring available, traffic volumes on local roads in the area were 21% higher than forecast meaning just a 6% reduction in volumes when it should have been a 22% decrease. So less traffic on the motorway and more stayed on local roads than expected – that’s not a great outcome. Amazingly they also note that no detailed project specific traffic model was produced, instead relying only on high level strategic modelling. This isn’t the kind of behaviour you’d expect for a project costing hundreds of millions of dollars.

There is also quite an interesting comment around the provision of PT. They note that the original documents including the paper that went to the board for funding approval noted that 3m shoulders would be added that could be used as bus lanes if needed. However as built the shoulders are “currently of variable width or absent at some points”.

Another area where the project didn’t perform well was with construction. The project ended up costing $227 million which was 22% higher than the $186m forecast which is blamed on ground conditions, associated earthworks and other items not fully specified or considered in the original tender and consent process. In addition the project was completed one year later than initially expected.

Lake Road Widening

This one is a local road project and involved widening Lake Rd on the North Shore between Esmonde Rd and Hauraki Corner which was said to be to relieve congestion. The reviewers say that in their view, the main issue with Lake Rd is that traffic queues on Esmonde Rd to get on to the motorway but that fact wasn’t recognised in the brief. I guess the planners and engineers involved had their horse blinkers on when working on the project, only focusing on the direct aspect they were looking at.

Economic benefits for the project primarily came from travel time savings (63%), reduced driver frustration (23%) – not sure how this is measured or impacts the economy. Like the other project traffic volumes haven’t been realised and the data available suggests it has actually declined.

Lake Rd Widening Volumes

The project was originally conceived as a single project but ended up being split in two stages. The total cost was initially $11.8 million however the figure was revised to $16.5 in a new assessment after the small first stage was completed. It ended up coming in 2% less than the revised figure but that was still 37% higher than initially expected

Bethlehem Township Improvements – Tauranga

The NZTA fast tracked the widening 400m of SH2 through the Bethlehem Township after the Tauranga City Council received approval to build a roundabout at the western end. They predicted that would “substantially exacerbate existing congestion problems at Bethlehem unless the highway was widened”. Travel time savings made up 88% of the BCR for widening the road.

Despite the widening travel times have actually increased and embarrassingly have done so at the same time that traffic volumes have decreased – as shown in the charts below.

Bethlehem Township Travel times

Bethlehem Township Traffic volumes

Perhaps things would have been much worse without the widening but it certainly doesn’t seem to have lived up to expectations.

Mangatawhiri Deviation

Not all projects have had bad outcomes though. One of the good ones is the Mangatawhiri Deviation which replaced a notoriously dangerous section of SH2. Despite the project cutting 3km off the existing route, 61% of the benefits came from improved safety with travel time savings making up 21%. The impact of the project on crashes is clearly shown below. They also not there have been three minor crashes on the old road which is now a local road.

Mangatawhiri Deviation Crash record

The shorter and straighter length of road has reduced travel times (although they say there is no reliable base line info to compare with) however they also point out that the exception to this is over holiday periods when there are huge increases in traffic volumes.

The project also came in more than 6 months early and $2.9 million (6%) under the original funding allocation. I’ve long viewed the Mangatawhiri deviation as an example of a project we should be using in other places such as instead Puhoi to Wellsford. It’s also a shame that the current focus on building the RoNS has delayed similar projects east and west of this one.

All up these and some of the ones I haven’t covered highlight some significant issues with our planning and assessment processes. That doesn’t create a huge amount of confidence in me that the same issues won’t be repeated with projects such as the East-West Link or an Additional Waitemata Harbour Crossing – projects that have costs orders of magnitude larger than the ones above. In the case of the latter I certainly don’t think the NZTA should even be contemplating trying to predict things like traffic volumes until Waterview has opened and they have a good handle of the impacts it has.

Share this


  1. The problem with selectively picking results without understanding the wider ramifications of the changes implemented is that you can make any stats ‘prove’ anything you like. The SH20 volumes are lower than expected due to the fact that the southbound route is avoided by savvy motorists as much as possible from midday onwards and especially on Fridays when there is gridlock where SH20 meets the southern motorway. If that poorly designed piece of roading were sorted then SH20 volumes (southbound) would be higher. As it stands from Redoubt Rd going south to Manurewa there are some 7 lanes of traffic converging into 2 within a kilometer or less on the motorway. This design was pure madness that resulted in daily gridlock. Before adding more lanes to an already oversubscribed thoroughfare the logical ting to do is alleviate the initial issue, in this case by adding the extra lane or lanes all the way from Manukau to the Bombays southbound. I and others were stunned by this, as even the most basic water bucket logic test says you can not keep adding water without increasing the size of the holes at the bottom. In an endeavor to ‘sort’ the issue traffic lights were installed – on SH20 – a motorway. Total madness, but nothing to do with the SH20 work itself.

    As for the airport to Hillsborough times – again nothing negative to do with SH20 works or the bridge – but more to do with the traffic lights and gridlock of Kirkbride road – impacting traffic to and from the airport (hopefully soon to be eliminated by the building of the new underpass thus avoided the stopping of vehicles).

    The point I am trying to make is to look at the bigger picture and ask yourself why certain results are what they are (and why expectations were not met) and look (as I suspect the engineers do) at what other factors have come into play or are likely to come into play as a result of the changes made.

    1. One the other hand you could read it as the never-ending cycle where as soon as the main source of congestion is fixed it means the next one down the road is now the cause, which needs to be now fixed…

    2. Did you happen to work on those projects? Perhaps you should direct your anger at those reviewing the projects – but seeing as I assume they know what they’re doing and they ask for comment from those involved in the project I’ll take their view over yours.

    3. Surely a review of a project should reference the initial business case. The examples you outline above tend to indicate that they were flawed. It’s hardly the reviewers fault that they found the project expectations weren’t met.

    4. So Ricardo, what you are saying is that it was a bad project design that didn’t fix the real problem, and that the predicted benefits didn’t eventuate as a result? Or in other words, hasn’t done what they said it would?

    5. Still hasn’t done what they said it would. So your point? Spend another couple of billion to fix the problems and move the congestion 100m down the road?

    6. “The SH20 volumes are lower than expected due to the fact that the southbound route is avoided by savvy motorists as much as possible from midday onwards and especially on Fridays when there is gridlock where SH20 meets the southern motorway. ”

      Is there a reason why the original analysts expected there to be no issue? Was there a contemporaneous fix planned that never happened? I don’t recall this being the case.

  2. Interesting that they don’t review the projected BCRs. Do any other reviews note the effects on cycling? The Mill St, Hamilton review says, “The cycle lanes provided are of minimum width (1.2 metres), and from site observations, appear to be unattractive to some cyclists. A proportion of cyclists cycled on footpath to either approach or bypass some junctions; more generous cycle lane widths may have encouraged more cyclists to use them.”

  3. As a Devonporter I can say that Lake Road has been a congested road in both directions for over ten years. It gets hit by morning/evening, school and weekend sports peak traffic volumes. It isn’t helped by a lack of connected and frequent public transport services, but the principal reason is indeed the lack of a motorway onramp at Esmonde Road, recently made worse by traffic lights. It is normal to spend 20-30 minutes in crawling traffic waiting to enter the motorway, which is just generating pollution and wasting everybody’s time.

    The answers lie in more frequent and cheaper public transport, and in fixing that wretched motorway onramp. Some local folks are proposing further widening of Lake Road but logic and experience suggests that this would be an expensive folly.

    1. I also live there and I don’t understand your comment on Esmonde Road on ramps. There are on and off ramps in all directions there. Do you mean they are in the wrong place?

      The money for widening Lake Road is gone (at least partly because NIMBYs have ensured very little population growth by opposing sorely needed intensification) and there is currently a traffic behaviour study under way to determine how the traffic is being generated. I imagine it will show a lot of education and Navy related traffic.

      I agree 1000% about needing dedicated PT lanes for the buses. Not sure where the width for that will come from. I wonder if a centre tidal bus lane would be the answer.

      I also think a bridge from the end of Francis Street to the church on Esmonde Road would allow both a link to SkyPath via SeaPath plus Akoranga busway station for walking and cycling. It could also allow a link through to Takapuna by upgrading the existing path alongside the mangrove swamp opposite the church to Auburn Reserve. The exisying Green Route is a joke after Francis Street. The local board is investigating that bridge and a budget has been put aside.

      Considering the $2m cost of the Bayswater pipeline bridge, I imagine the Francis Street bridge will cost $2-3m.

      I really think that would encourage a lot more people to consider those options. Once you are at Akoranga busway station, there are fast frequent buses to the city and beyond. Though the main problem may be that they are all full by the time they get to Akoranga.

      1. A Francis St link is probably a good idea, but there is a link already from Hart to Harley and then Eldon. Is this a significant detour?

        1. How do you get from Hart to Harley? Not a local but I see no link on the map. In any case, still seems to add almost a kilometre extra distance. Maybe not the most urgent new estuary boardwalk, but still looks quite useful

  4. The SH20 extension to Maioro St has changed the behaviour on local roads as these are now much easier to use. For example, I travel fairly regularly to New Lynn from my home in Kingsland. If I go for social purposes I will get the bus or the train so I can have a drink, but for other purposes I used to drive either via Mt Albert or the NW motorway. With the improvements around SH20 it is now much nicer to go via Sandringham Rd, over the motorway, and through the Tiverton Rd improvements. This also cuts out the Avondale roundabout which I hate. Also the new Countdown supermarket must have had an effect on Richardson Rd.

  5. Makes you wonder who is pushing the various projects.
    Looking at SH20, which I use fairly regularly, the choke points are at the ends at present, the junction with SH1 at Manukau and the Maioro Street interchange. There is no reason to believe the latter would sort itself out when the Waterview Tunnel is open simply because a large percentage of the traffic using that interchange is heading west towards New Lynn down Wolverton etc streets. Other than rush hours, the motorway is fairly empty.
    As someone who has worked on SH1 north of the Johnson Hill tunnels, I am firmly of the opinion that widening and three laning the existing road would be adequate, probably all the way to the Brynderwyns. That road is not all that busy outside one or two days a year, and a system, similar to the Pacific Highway in New South Wales where the overtaking lanes change sides every two kilometres, with prominent signage, would probably work. The holiday rush hour could be catered for with movable barriers to two lane the entire length from Warkworth to Puhoi, with four lanes in areas such as Windy Ridge where there is room to do so.

  6. Transport models appear to be substantially broken. They clearly miss significant factors; they are consistently over predicting driving demand and under picking Transit demand. This is probably because there has been a substantial shift in preferences that neither the models nor the modellers seem able to capture.

    We live in an age of discontinuity and extrapolation simply cannot pick the discontinuous.

    This is extremely important because billions and billions of dollars are spent on the basis of these models,

    Take the case of the CCFAS study that predicted mad outcomes like people still trying to drive when the traffic isn’t moving and there’s a fast new underground train service available as an option. Those conclusions already look silly in a city with 20+% rail uptake growth and traffic VKT declining per capita. And we haven’t built the tunnel yet!.

    New models please, ones that accept more uncertainty around our willingness to drive and match the era we now live in.

    1. The problems are simple, at the moment roading funding is pretty easy to obtain, figures are conservatively worked out e.g over estimated by design as traffic flow and capacity is the important factor. Costs are justified by time so to get the project funding you need to be conservative to ensure less time is spent in traffic, and therefore more time savings and greater benefits are obtained. When it comes to rapid transit/public transport, funding is rather hard to obtain, figures are conservatively worked out e.g underestimate by design as operating costs is the important factor. Cost are justified by the number of people using the service and the cost per person, if the figures are not conservatively worked out it is torn apart from critics saying it will never reach those figures. (Having people waiting as buses or trains fly past full is considered success, where this is considered a failure on the roading network)

      We need to some how align these, either both roading and rapid transit are conservatively worked out in terms of time savings, and its better to overestimate, or we go for the more unideal situation and underestimate to ensure we are not overspending. It all comes down to how we estimate the benefits, and at the moment they are not truly aligned.

  7. “quite how you can get emissions reductions when traffic volumes increase is beyond me”

    This is because the Economic Evaluation Manual focuses too heavily on the benefits to individuals and doesn’t reflect benefits or dis-benefits to the community as a whole.

    Claiming emissions reductions is only true if you restrict the model just the transport project itself. In the case of the Manukau Harbour crossing, such an approach implies that because cars are travelling faster, each car produces less emissions per kilometre than if they were travelling slower, so that is a benefit to *the individual* motorist.

    This approach is flawed though of course, since the total volume of vehicles increases both within and outside of the project scope, due to the newly created capacity inducing demand for fossil fuel based travel. The total volume of emissions from the transport system as a whole would necessarily increase (since we are adding more capacity for single occupant cars), compared with not doing the project at all.

    1. It’s actually not totally absurd. The emissions reductions are calculated relative to an alternative (“do-minimum”) scenario in which there is a similar amount of aggregate traffic, just moving more slowly and hence emitting more. So from the perspective of those two scenarios, it makes sense to talk about a net reduction in emissions.

      I.E. the modelling says that emissions have risen, but not as much as they otherwise would have.

      But, as so often happens, the models don’t necessarily make contact with reality. They fail to account for (substantial) induced traffic effects – i.e. people choosing to drive more, or live further out, because it’s easier. Induced traffic can lead to some perverse outcomes – while it’s individually rational for everyone to recycle travel time savings into different travel patterns, doing so means more negative externalities from transport.

      This process is actually bad for drivers as well as the environment. The economic literature has found that wider roads generally fill up with more traffic and revert to the same travel speeds as before. This means that there are more people experiencing the same level of delay and frustration on the road – hardly a win for the poor beleaguered motorist!

  8. My understanding was the Lake Rd widening was mostly for the benefit of cyclists by adding in a cycle lane. Not sure what the count is now but last I checked there were only 500 cyclists on a week day using Lake Rd. $16M for 500 cyclists a day. Could have been better spent elsewhere.

    1. The report itself summarises the project nicely. You should read it. This isn’t a $16m cycle way.

      1. I did read the report. I never said anything about a cycle way. Just a cycle lane. Much of the widening appears to be to fit the cycle lanes in. Thus it was a cycling project. Of course they did stick in a median which is generally a good safety measure so maybe it was a safety project in disguise.

        Regardless, I don’t actually see how they could improve vehicle travel time with what they said they did. And how the heck could they get that vehicle growth if nothing is getting built in the area?

        1. Did you miss this outline of the project? Doesn’t sound like the cycle lanes were the expensive bit.

          Widening of Esmonde Road approach Lake Road to two lanes, with flaring to three lanes immediately before the Lake Road signals.

          Provision of two left turn lanes from Lake Road into Esmonde Road on an improved alignment.

          Improvements at Lake Road/Esmonde Road intersection, including provision of pedestrian, cycle and bus priority facilities.

          Road widening on Lake Road to retain two lanes in each direction between Hauraki Road and Esmonde Road.

          Traffic management restrictions on side road turning movements.

          A central painted median (2.5 metres) and right turning bays to provide a safer driving environment and easier right turns for vehicles entering side roads or properties.

          Recessed bus stop bays on either side of Lake Road.

          On-road kerbside cycle lanes (1.5 metres) in each direction on Lake Road.

          Pedestrian crossing facilities at Lake Road/Hauraki Road signals and introduction of Lake Road ‘mid-block’ signalised pedestrian crossing between Napier Avenue and Cameron Road.

        2. “Regardless, I don’t actually see how they could improve vehicle travel time with what they said they did. And how the heck could they get that vehicle growth if nothing is getting built in the area?”
          Yes this is exactly the point, often these projects are sold with all kinds of claims that don’t mean stack up

        3. Actually it was widened from Esmonde Road to Hauraki Corner so that there could be two vehicle lanes each way. Nothing to do with cycling.

          The cycle lanes were installed in 2007 for about $400k from NZTA. If they are removed that money has to be paid back.

          For that price, 500 cyclists a day is not bad. According to AT, the daily traffic (I assume this excludes bikes) count on Lake Road is 8527 (

          Based on that, people cycling make up 5.8% of total vehicles on Lake Road on a weekday. In the context of the English speaking world that is spectacular. Especially considering the cycle lanes just disappear at the Hauraki Corner intersection to allow two car lanes.

          Actual evidence is so much less fun than ideological knee jerk reactions based on nothing, isn’t it?

    2. My recollection was that initially the council planned to simply remove the cycle lanes and then back down, there were never any plans for install a proper protected cycle lane, and the project would have simply widened the road to provide more car lanes but kept the same on road painted lanes for people on bikes. As such, attempting to claim it was a 16 million spend on cycling is disingenuous.

      1. “As such, attempting to claim it was a 16 million spend on cycling is disingenuous.”

        Yet part for the course.

        With the cycle lanes being on the outside edge of projects, they are easy targets especially when widening is proposed – “Just drop the stupid cycle lanes and we can keep our front yard”. No recognition that the road space gobbling-up of property is in 9 out of 10 cases done for cars, with the remaining maybe 1 out of 10 cases being for buses, bikes or pedestrians.

    3. Even if it was $16M for 500 cyclists a day, it is a bargain compared to the next harbour crossing. Only $32,000 / vehicle vs $108,055 / vehicle for the harbour bridge.

      NZTA Harbour Bridge Counter May 2015: 158,650
      Without Harbour Crossing: +30% 47,595
      With Harbour Crossing: +65% 103,122 (additional 55,527)

      $6,000,000,000 / 55,527 = $108,055 / vehicle

      $16,000,000 / 500 = $32,000 / vehicle

  9. Is there really an inverse relationship between traffic speeds and emissions? If so then the laws of physics need reviewing, and so does the more usual message to drivers is that you can save fuel by easing off the accelerator.
    Energy is required to move, and more energy is required to move more rapidly, in order to overcome the greater resistance-forces. It cannot seriously be argued that going faster in all cases reduces emissions.

    Where traffic is stop-start with a lot of idling I can understand fuel-use and emissions being elevated. However a break-even speed must soon be reached at which the inefficiencies of very low speed and the greater energy-requirement for higher speeds balance out.

    What is this optimal speed, and is it at a level that substantiates the claims made by road promoters that this or that scheme will reduce emissions? Or is the general principal of an optimally-efficient driving-speed being misused to promote faster roads which will actually push vehicle speeds into less-efficient regions.

    Given the various other ways in which roading-project benefits are deliberately overstated, I would be very surprised if this factor is not also being deliberately misapplied.

    1. steady speed is a lot more efficient than having to stop start etc. As you can see with car fuel efficiency ratings typically a highway fuel consumption figure might be 7l/100km while city will be 12l/100km (and this is optimum, in reality the highway figure is reasonable on the open road but the city figure is usually a lot higher so you end up with approximately double the consumption). The whole aerodynamics sure does play a part but only above 50km/h really below that it doesn’t matter enough so 50km/h smooth is a lot more efficient than 20km/h stop-start. 100km/h does use more fuel than 80km/h however again the amounts are not a huge difference as most cars are designed to run economically at 100km/h (if you think about an old style 5 speed manual transmission at 100km/h in 5th most cars will be doing 2500RPM and at 80km/h they might be doing 2300RPM. However at 120km/h they’ll be doing 3000RPM which is a big jump for the extra 20km/h compared to from 80-100km/h.

      1. In your example wouldn’t it be 2000rpm at 80km/h if you’re still in 5th, or were you implying they’d change gears? Not that RPM alone matters – it’s really the power output over time ala the amount of throttle which you need to increase to maintain a higher speed based on the drag.

        Not that this matters really for hybrids and the like – most hybrids I’ve seen have a higher fuel consumption on the open road than in urban environments, which I assume is the exponential effect of drag with speed outweighing the losses from stop/start traffic. Given this change in vehicle technology, should we be adding a cost to projects that increase speed?

        1. Yes my Prius is much more fuel efficient around the city as it gets a chance to recharge the battery when I brake or go down slopes. If I put the engine braking setting on as I coast down the AHB, I can fully recharge the battery.

          Hybrids really are a fantastic piece of technology as well as being really nice to drive and very safe. I would hate to go back to driving an ordinary car as our fuel bills are ridiculously low.

  10. Were these reviews prepared by the same people who advise the minister of Bridges/transport?
    If so, is it possible to start a movement toward requesting a CBA of all their existing recommendations on the basis that previous recommendations are outside acceptable parameters?

  11. “The biggest positive is the cost which at $218 million was 19% less than estimated ($268m) which is in part due to the project being completed seven months early.” So the entire bridge project was built for $218m a decade ago so with inflation about $250m in todays money. How much is that kirkbridge road project costing again? Something like $170m? That is ridiculous that a simple small bridge is costing almost as much as a large motorway bridge over a harbour!

  12. You can widen the bridge or motorways as much as you like so more vehicles can travel faster, but they still have to park somewhere when they get into town and that isn’t part of any equation at present.

Leave a Reply

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