MELBOURNE:CityLink Problems

Fullscale tolling is also a ways off. We had good talks with a bunch of people in Melbourne during a recent visit to see the world’s second fully automated toll road. Seems to be agreement that the front end of the toll system is working quite well. There are 300k transponders on issue. Gantries over the 2x4-lane Tullamarine Fwy, widened as part of the northern end of the Western Link, have been in operation over a year with all kinds of traffic and weather. For 4 months now they’ve had traffic with and without transponders running under gantry equipment all down the Western Link. Contracts provide for 90% license plate imaging, and they say they’re up to 97% with the possibility of improving slightly more. (When they first opened up they got barely 50% but improved lighting, filters, new camera timings, and other tweaking of equipment got it up to the present 97% level.)

Saab’s stereoscopic video tracker and profiler system works well too, officials say, as does the transponder tracking and communications of the readers. This is the first major test of the Saab Combitech open road toll system and so far it has performed to requirement, officials who otherwise disagree, say.

Unlike the world’s first automated toll road, 407-ETR in Toronto, which has about 128 toll points on the ramps of its 29 interchanges, MCL has some 18 toll points located over the mainline or travel lanes between the interchanges. This is a virtual barrier system tolling point by point as compared to the Toronto virtual ticket system that is based on trip computation. (Cross Israel is following the MCL point toll format though with 407 equipment.) In the Melbourne system from Saab considerable processing is performed at each toll point. So-called ‘white lists’ of vehicle license plate numbers of patrons who have purchased video day passes are uploaded regularly to the toll points on the road and checking of vehicle class as profiled by the video profiler against the transponder-encoded class information at the toll point equipment on the roadside. In theory this reduces the volume of data that needs to be moved to the central toll management system and the volume of computations that have to be performed there.

The last sentence started ‘In theory...’ because in practice MCL has superimposed important elements of trip tolling on top of straightforward point tolling. Modifying the various point tolls as the vehicle passes under mainline gantries is a trip toll cap ($2.40 for cars) for each continuous trip on the toll road. This trip cap requires a process they call ‘trip reconstruction’ to be conducted for every journey by the central toll management system. It requires massive computation using an Oracle database system. There is some dicey use of inference in place of measurement.

Beating Gamers

Without any equipment over the ramps there is no way to measure trips. Trips have to be inferred from records of the timing of passes of adjacent toll points. If the time between passing toll point 2 and 3, 1600m (1-mi) apart, say, is more than say 6-mins in rush hours, you assume speed of less than 16km/hr (10mph) and infer the motorist must have departed the off-ramp, done some business off the toll road, and re-entered the on-ramp. How to allow for variable speeds especially when you have to expect occasional congestion and stop & go traffic and incidents? The time allowed has to be varied by time of day or even measured traffic conditions. How to avoid couriers and other intensive users developing pickup and delivery schedules that exploit the time they are allowed before a continuous trip with a toll cap is converted into two separate trips with up to double the toll?

Another serious problem arises with the occasional non-read of a transponder, which may occur, say, once in 1000 passes or say once in 300 trips. Then a motorist traveling under toll points 2,3,4,5, and 6 will, once in 300 trips, be read as having traveled say 2,3,4 and 6, 5 having been a non-read. This non-read artificially ‘breaks’ the trip and generates two trips and two tolls, instead of one toll as specified by the trip toll cap.

The terms of the toll concession combined with the concessionaire’s need to maintain good customer relations demand that anomalies be resolved in favor of the customer, but officials say that it has been difficult to develop procedures to do this without opening up the system to gaming by toll evaders. They say think they just about have the problem beaten, and are running quantities of real traffic data through the system to test it.

There have also been major delays developing the system to generate correct invoices and statements and the required financial reports, and the communications interface with the mailing house that sends out toll bills. Some say these problems have loomed larger than the trip reconstruction issue.

MCL’s western link opened Aug 15 and if the toll management software had been ready tolling would have started then too. Saab was responsible for system integration at the tolling points and Translink, a joint venture of Transfield and Egis has responsibility for delivery of the troublesome toll management system. Translink in turn subcontracted the writing of code to Computer Sciences Corp (CSC) of the US. This is the first toll management system CSC has attempted (and we’d wager: the last!)

But there seems to be some inexperience all around. Transurban the concessionaire produced a clumsily complex toll system requirement. Those on the system delivery side complain that Transurban’s requirements, for which they have been writing code, have been constantly changing. 46% of the code in the present system has been written this year alone, reflecting changes made in requirements in the last year, though writing of the code goes back 3 years. In the period Oct to Dec 98 there were 40 change orders issued for the toll system.

Some of the changes, the owner says reflect discovery of problems the original code had in meeting specified requirements, so it will be difficult to untangle where responsibility lies – not an academic issue since the system supplier is liable for damages for uncollected tolls that are the product of its failure to deliver.

A senior official admitted to us: “If we had it over we would have done things differently. We had not thought through all the implications in enough detail. The trip toll cap did add a level of complexity we had not anticipated. It has been pretty frustrating that the toll system has been so difficult to develop. But we are very close to beating most of the problems.”

They have learned the hard way the costs of a complex toll scheme.

Why did they want to implement this bastard mix of point and trip tolling? If you want to toll trips, why not stick the equipment on the ramps and measure entries and exits? An official influential in the MCL implementation told us the equipment won’t work adequately placed on ramps. In slow traffic the cameras miss too many front license plate numbers because they get obscured by the vehicle ahead. He says that they concluded one meter (40") vehicle spacing is the break point. When vehicles are 1m or closer to one another there is no camera angle that will get you a legible picture of the license plate. His reasoning was that with the off-ramps of MCL delivering traffic onto busy inner city streets there would be rush hour backups on those ramps that would have many vehicles getting closer to one another than the critical 1m spacing, so ramp based equipment like 407-ETR would miss too many license plate numbers.

MCL ramps are reasonably long, though perhaps not as long as 407, and MCL’s off-ramps are mostly 2-lane ramps (407 has 2 and 3-lane ramps) if only to allow sufficient storage of exiting vehicles to prevent backups as far as the mainline. But the design of MCL is definitely tighter than 407 given the tighter constraints of its inner city location. Ramp located gantries would also have cost a bit more, approx 36 toll points compared to 18.

But why then not go with simple point-tolling with the equipment over the mainline in a virtual barrier system? Why complicate the point tolling with the darned trip toll cap? The answer we got was that traffic and financial modeling showed major benefits to be gained from favoring longdistance trips over short trips, the through-traffic over quick on-and-off traffic. Government officials too favored a toll system that encouraged the long-distance trips. So giving them a financial break was the reason for the trip toll cap.

Whatever the validity of these propositions the mixed toll scheme they are attempting to operate in Melbourne seems a highly complex one, heavily at odds with veteran tollsters invocations to “KISS” (Keep It Simple, Stupid). It also requires heavy data movement and heavy computing loads, and these seem rather liable to throw up anomalies and problems that in turn will generate customer complaints and PR issues. Time will tell whether the problems are serious.

In the first segment, the western link, only a small proportion of trips will hit the trip toll cap, so not too much will be demanded of the system, and tolling could start shortly. 407-ETR had system problems that delayed its start 129 days. Dec 23 MCL would surpass 407’s record for a toll road to be open without toll collection!

Other toll system issues concern a day pass for people without transponders. For a flat $4.50 prepaid ($5.20 postpaid by the following midday) motorists are entitled to unlimited use of the system for the day. There are loud complaints that this is a heavy toll for making the occasional trip into Melbourne from country towns to the immediate northwest, an hour’s drive beyond the airport. The conservative government that ran the state until October elections fell in a surprise upset with a huge swing against it in country seats, in part it is said because of resentment that programs such as MCL were rigged against country people.

Actually the day pass is a bargain for someone making a bunch of trips on the one day, so Transurban says no more than 12 may be obtained per vehicle per year. That’s another item to be policed. The concessionaire is apparently giving the day pass a rethink.

The day pass system, of course along with toll enforcement, depends on license plate imaging. The MCL imaging system is presently set up to handle some 70k images per day, about 12% of the expected 600k toll point passes expected when the whole system is tolling. As with 407-ETR’s start-up there are some concerns on MCL that image processing could overtax the system early on, before transponders are fully distributed.

Against that, CityLink has done one heck of a job getting transponders out – 300k even before opening. That must be a world record for pre-toll placement of transponders.

Grand Gateway

The western link represents a grand new gateway to Melbourne from the Airport, and indeed from points north, including Sydney. A new vista of the city’s tall buildings, the docks, the Bay and bridges unfolds. It’s rather like the spectacular approach into Chicago curving in around the line of the lakeshore from Indiana along the Chicago Skyway. At the beginning of the CityLink Transurban built one of the world’s largest pieces of modern sculpture with the massive leaning beams... that leave me cold, at least in daytime. Under lights at night the bright yellow of the overhanging beam and the counterpoised red posts make a decent spectacle. But just south of this the CityLink sound tube, a hightech steel frame pergola of sweeping oval section, that encloses the whole 6-lanes of the elevated motorway is an original piece of design genius that elegantly serves an important environmental function – allowing sound-absorbing panels to be placed to protect residents of nearby highrise apartment blocks – as well as adding interest to the journey. It gets our ‘Design of 99’ Award.

The near-finished Domain Tunnel, is a well-equipped pleasantly spacious 3-lane facility for city-bound traffic that had the finishing touches of lane markings being applied and variable message signs scrolling through final tests when we visited.

Outside the posted speed limit on the MCL will be 100km/h (63mph) but in the tunnels 80km/hr (50mph). The tunnels turn gently in both horizontal and vertical planes so views ahead should be good and there is none of the mesmerizing effect of long straight tunnels. 3-lanes have a far more spacious feel than normal 2-lane tunnels. Lighting intensity in the MCL tubes is controlled to vary according to outside conditions and to minimize transition stress for drivers with higher illumination levels near the portals. There seem to be plenty of message signs, and all the regular modern surveillance, alarm and escape stuff.

Bubbling Burbling Burnley

The construction problems are in the longer and deeper Burnley Tunnel which takes outbound traffic from the city – the opposite traffic direction from the Domain Tunnel. 70m below ground level, the watertable provides about 60m (200') of head above the middle section of the Burnley tunnel, generating pressures of 70 atmospheres, about the same in tons/sq meter. The initial grouting and membrane of the tunnel was inadequate in many places and water entered at a rate of 100 liters/sec – “like a waterfall” they said – compared to 2 liters specified in the contract. The water pressure lifted and broke seven massive 20m x 11m x 1.8m unreinforced concrete floor panels. 250m of the floor is being replaced if damaged and tied down to the rock with new rockbolts to prevent further movement. Tests on another 500m have yet to determine whether more replacement and repairs will be needed in adjacent sections. Some describe this as a region of unexpectedly fissured basalt located below an aquifer, though others say that over-dramatizes conditions. Grouting will eventually stem most of the water inflow. Already the inflow has been reduced below 20 liters/second. In hindsight the tunnelers should have done heavy grouting ahead of excavation, so that they were excavating in a zone in which the fissures had been filled first. That’s expensive if it turns out not to be needed. More detailed core sampling might have warned of the highly porous fissured rock. But tunneling is a risktaking exercise, an argument for having the private sector take the risk. This time the risk didn’t pay off, and the Transfielde-Obayashi joint venture constructors will have to spend an estimated extra $20m to $30m fixing the water problems. And they may be up for more than that in toll revenues lost.

Traffic & $s

Traffic and revenue forecasts are always difficult but a hunch is that this is a strong project, offering major motorist benefits for the tolls charged. Modeling done for the initial public offering of Transurban shares published in 1996 estimated that the toll schedules proposed would divert an average of 23% of projected free traffic away from the toll road. It suggested for year 2001 662k/toll passes/weekday, rising to 731k in 2011 and a 0.5% increase annually thereafter. This was based on 6 toll zones. There are now 9 toll zones, one extra because an additional leg was added to the project (Exhibition St Connector) and a couple of other zones have been subdivided. The pre-opening forecasts are that traffic in the first year will generate “about 600k” toll point passes/weekday. With an average 2.4 passes per trip that’s 250k trips/wd. On the separate legs weekday traffic was forecast in the range 99k to 132k 2001 and 113k to 134k in 2011. Commercial vehicles are forecast at 21% of the traffic of which heavy vehicles are 5.4%.

The mainline of the roadway is 2x4 lanes for the highest volume leg the Tullamarine Fwy. The rest is 2x3 lanes. These are sufficient for projected traffic volumes, consultants say, but some time in the concession period some ramps will become overloaded. There is already congestion in the north of the CityLink section of the Tullamarine Fwy especially in the PM peak. Just beyond the north end of the concession two interchanges (Bulla Rd and Calder Hwy) are inadequate and heavy weaving movements slow traffic to a crawl and cause backups way down the Western Link, which will detract from timesavings on the MCL. Another serious bottleneck could occur in the very center of MCL. This is a 3km stretch of the Westgate Fwy east of its IC with Western Link and connecting to the Southern Link. Transurban may have made an error in not negotiating for control of this pre-existing stretch of elevated motorway which has two further ICs (at Montague St and Kingsway/Power St). Unpriced the intruding ‘freeway’ leg may represent a serious traffic bottleneck between the two parts of the Transurban concession. Already the state roads people think just that and have refused to allow one of the Transurban ramps to be opened.

Toll revenue was projected in 1995 to be $120m in 2001 and $141m in 2011, based on earlier toll rates. Under the concession terms – it’s a 34 year concession – maximum toll rates may rise by the CPI or 4.5% each year whichever is the greatest.

Officials talk of exploring differential toll rates by time of day or level of congestion at some point in the future, but they obviously have more urgent issues on hand.

Troubles Ahead

The first years of MCL could see continued troubles. First priority for Transurban is to get a decently acceptable toll system running to generate revenue for the investors and to service debt. The odds look at least even that the cumbersome toll scheme they are opening with will generate anomalies and a level of complaints and problems that will require them to transition to a simpler toll scheme. That may require difficult negotiations to amend the concession deed. A new state Labor government is now in power without any stake in the detailed development of the concession, and it is liable to take populist anti-business positions that could add to Transurban’s problems. There’s also a ‘killer’ press locally that is reflexive in amplifying bad news. Pioneering’s tough.

Structure

One of the successes of the MCL project has been the way the concession has been structured. There is Melb City Link Authority (MCLA), a small specialized team that closely monitors the project representing the state which has issued the toll concession contract, and an Office of the Independent Reviewer (OIR), jointly funded by the state and the concessionaire. The OIR has independent regulatory powers to lay down testing requirements and must sign off on separate pieces of the project, both civil works and systems, before they can be commissioned. OIR is a kind of institutionalized independent arbitrator. Australia’s leading engineering company Sinclair Knight Merz, Parsons Brinckerhoff, and Wilbur Smith Assoc are working for the OIR. Commercial issues between the state and the concessionaire Transurban are handled separately by MCLA.

Present Problems to Pale...

Whatever it’s present problems Melbourne CityLink will be a great toll road system. It represents good value for the resources invested. 90 lane-km of difficult inner-urban motorway built in four years for a project cost of $1300m ($14.5m/lane-km.) The closest US project to MCL in construction and environmental difficulty is Boston’s Central Artery/Tunnel project undertaken by traditional governmental processes and tax funds - 260 lane-km taking 10 years to build and costing $11b ($42m/lane-km.) The investor approach in Melbourne is delivering a project at about a third the unit cost of the extravagant state operation in Boston.

The MCL concession is held by Transurban, a publicly traded company which has a 34 year concession in return for financing and building the facility. In Boston a huge debt is burden is left to the taxpayers of Massachusetts, while in Melbourne investors foot the bills, take the risk (and rewards, if they do well.)

MCL will transform mobility over large parts of the 3m pop area of Australia’s second city, providing faster, less stressful, safer, more economical, and visually attractive trips. Covering 22km it consists of:

(a) The Western Link

(1) widening 7.5km of Tullamarine Fwy from 2x2 to 2x4 lanes along the Moonie Ponds Creek

(2) 2x3-lanes of elevated concrete segmental box girder construction for 5km linking the Tullamarine Fwy to the Westgate Fwy, including a new bridge over the Yarra River.

(b) The Southern Link

(3) Burnley Tunnel 3.4km eastbound 3-lanes under city gardens, the Yarra River and the inner suburb of Burnley

(4) Domain Tunnel 1.6km westbound 3-lanes coffer dam construction under the Yarra River and driven under the city gardens

(5) Widening the riverside Southeast Fwy 2km for westbound traffic from Burnley to the Domain Tunnel portal

(6) Widening the Southeast Fwy from 2x2-lanes to 2x3 lanes for 5km

(7) Exhibition St Connector from the SE Fwy to the east end of the CBD in a new bridge over railyards

Linking three Orphans

It links together three previously unconnected radial motorways that petered out several miles before reaching the city center. The roadway provides better connections to and from the city for the southeast, the west and the northwest, and the first high quality freeflowing bypass of the central area for through traffic. As well as helping commuters – and it will help as many commuters on freed-up surface arterials as it will users of the tollway – it is very much a commercial facility. The manufacturing and hightech employment areas are in the southeast of Melbourne, the ports and railheads on the immediate west of the city and the airport and most important interstate destinations are northward. MCL efficiently integrates those three centers for the first time. Melbourne already has the leading container port of Australia and CityLink will enhance the economies of the port and the airport, and the city’s pre-eminence in the country’s trade and distribution. Melbourne’s rival to the north, which is building its highways in a more piecemeal fashion, is at least 10 years away from having a network to match that provided by MCL.

The project is an unusual exercise in bipartisanship in a ‘take-no-prisoners’ political culture. The project was conceived and begun by Labor, and after a change of government, was implemented by the conservatives. It is difficult enough getting agreement to build any major new central city highway, let along one putting together a project that can be funded entirely by investors and therefore without any burden on taxpayers. The mistakes being made along the way pale into insignificance compared to the overall accomplishment. (Contacts Andrew Manton Transurban 61 3 9612 6999 amanton@transurban.com.au, Alf Smith Melb CityLink Auth 61 3 9655 6601 asmith@mcla.vic.gov.au, Salahdin Yacoubi Translink 61 3 9291 8103 syacoubi@translink.com.au)