Labor Metronet review

With the state election fast approaching, Labor has put forward its plans for public transport. These are packaged under the term Metronet. This involves the construction of several heavy rail lines.

PLANNED LINES

The Northern Circle Line (NCL) will run over a new railway along Reid and Tonkin Hwys, as well as employing existing tracks along the Joondalup (to Stirling) and Midland (to Bayswater) lines to form a loop. New stations will be at Balcatta (Erindale Rd presumably), Wanneroo Rd, Mirrabooka Av, Alexander Dr, Noranda (Benara Rd presumably) and Walter Rd.
A line will also be constructed to Ellenbrook, which will branch off the NCL after Noranda, with stops at Bennett Springs (probably the intersection of Reid Hwy and Beechboro Rd), Whiteman Park and Ellenbrook.
The map also shows a line to Wanneroo, but this is not mentioned in text. I don't know of any good rail corridor to Wanneroo, and money for a tunnel would better be spent in more inner areas.
The Southern Circle Line (SCL)will incorporate an airport line previously proposed by the Liberals (from Bayswater to Forrestfield), as well as electrification and amplification of a freight line for passenger service from Forrestfield to Fremantle. New station will be at Airport West (near the current domestic terminal site, with domestic to be consolidated at the international terminal and the site redeveloped as a business park, AFAIK), Perth Airport, Forrestfield, Wattle Grove, Kenwick Junction (interchange with Armadale Line), Thornlie, Nicholson Rd, Ranford Rd, South Lake (interchange with Mandurah Line), Yangebup and Coogee, then looping back using the existing Fremantle Line tracks, through the city to the Midland Line, to Bayswater.
A dedicated airport line will also be built paralleling the SCL between Forrestfield and Shenton Park, perhaps with specialised rolling stock.
The Joondalup Line will be extended from Clarkson to Yanchep, and the Armadale Line to Byford (although the map shows a line all the way to Pinjarra!)

OPINION

GOOD

SUBURBAN RAIL

This plan focuses on delivering more cross-suburban rail, an very good idea. According to the Public Transport for Perth in 2031 plan, the inner city (CBD, West Perth, East Perth, Northbridge) only accounts for 18% of jobs in the metropolitan area (see page 12). This is only set to increase, as congestion in the CBD and plans such as the Stirling City Centre and Murdoch Mix draw more employers into secondary CBDs. Metronet will add Kewdale (the 2nd largest employment area), Canning Vale (7th largest), Malaga (10th), Balcatta (11th), This is more efficient, as instead of having full trains heading into the city in the morning, but near-empty trains heading out, those empty trains will have more passengers on them, which is better for the environment and for the cost recovery of public transport.

SOCIAL

Rail will be delivered to neglected corners of suburbia, affording better access to amenities in the CBD and along existing railway lines (eg. shopping, universities, employment), and perhaps spawning new amenities to these places, drawn by rail access. This will help combat issues such as youth unemployment and unrest.

FINANCIAL

As the plan consists of surface heavy rail, it is much cheaper than tunnelled alternatives, while still offering the speed and capacity of heavy rail. Greater accessibility, as alluded to above, would bring greater investment, as a result of higher land value, and increase access to high paying jobs, which are concentrated in the CBD.

BAD

TRACK CAPACITY

However, there is a disturbing number of lines converging on the Midland Line (five, six if you count the phantom Wanneroo line;). With regular 20 tph signalling, the maximum frequency for each line during peak hour is 4 tph, or every 15 minutes, what is being run in the off-peak. An upgrade to 30tph signalling could see 6tph, but this is what the overcrowded Midland Line receives during peak hour currently, leaving little room for growth other than traffic from Perth to before Bayswater being put onto the other lines, platform extension or running other lines at 4tph. This is not to mention that punctuality will be compromised, as each train will have to arrive at Bayswater at the exact scheduled time, or else trains after that one will be delayed.trains running even 45 seconds late may result in delays and reduction in capacity. The Midland Line will have to be quadruplicated up to Bayswater.

You could run the NCL right through to the SCL, instead of running to the city on the Midland Line. This would remove 2 lines from the Perth to Bayswater pinch point, cutting the number of lines there to 3. Each may run every 9 minutes (6 2/3 tph)on current signalling, or every 6 minutes (10 tph) with an upgrade, or double the figure with quadruplication. A new interchange station where the Midland Line intersects Tonkin Hwy would have to be built though, and airport trains would bypass this stop. SCL passengers will be able to have a zero walking distance transfer by changing to the airport line at an airport station, where both the SCL and Airport Line will call into the same platforms. There may be an extra charge on airport trains though, like in Sydney and Brisbane, which may be only be checked at entrances or enforced on board airport trains too. In addtion, reducing the number of lines does not change the fact that a large number of passengers will converge on the Midland Line between Perth and Bayswater.

LOCATION

Also an issue is the location of the NCL and SCL; the former in a highway median, the latter in primarily industrial lands.
We've yet to develop a good TOD at any of our existing freeway stations (Cockburn Central is alright, but cut off from the main commercial area (Gateway SC) by heavy traffic and car-oriented development along Beeliar Dr, as it is used to access the Kwinana Fwy), and for good reason - car traffic and pollution do not make good real estate marketing points, and the extensive park n' rides that follow car dependence do not make for quick and pleasant walks to the station. While there may be creative solutions, it is much simpler to build TODs outside of freeways. In some cases, non-highway TODs could harness suburb names that are already better than many highway-side suburbs (without naming anywhere).
As for industrial areas on the SCL, this is better, as there is no big, hulking highway to separate suburbs from their eponymous stations, but suburbs will still have to be radically different from what they are now. While it is good to serve industrial areas and their large employee numbers (as mentioned before, Kewdale is the area with the largest number of jobs out side of the inner city), industrial areas are typically large-scale and car/truck-dependent, and this must be changed if train lines serving such areas are to be well used. To kick out industry for residential and commercial development in response to this problem is also an issue; industrial areas cannot be simply pushed further and further out. The continued need for freight rail without delays to passenger trains will require wide rail corridors, without corresponding two-tier operation, frequency or existing urbanity.

COST

The costing of Metronet has been quite a contentious issue. First there was the Labour costing of $3.8 billion, then the $6.4 Liberal costing, and the Treasury costing sits in between at $4.335 billion ($5.2 billion by the time the project is finished). Let's see for ourselves what the plan might cost, in my mini - costing.

To be generous, I've thrown in a few extras, such as the aforementioned quadruplication to Bayswaters, and 4 km of tunnel for the airport section (the lack of this attacked by the Liberals in radio ads)

NCL - 17 km of surface rail
SCL - 42 km of surface rail (mostly amplified over the existing freight line, but new infrastructure is needed for passenger trains), 5 km of underground rail
Ellenbrook - 11km of surface rail
Butler to Yanchep (Clarkson to Butler is already under construction) - 13 km of surface rail
Armadale to Byford - 8 km of surface rail
Perth to Bayswater - 8 km of amplified surface track
Total - 91 km of surface rail, 5 km of underground rail, 8 km of amplified track

Approximate costs (using statistics from Martin 2010)
Surface rail - $20 million per km
Underground rail - $221.5 million per km
Track Quadruplication - $27.25 million per km

To try to account for inflation (I don't know how to do it properly), and also be conservative, I will add to these figures.
Surface rail - $25 million per km
Underground rail - $250 million per km
Track quadruplication - $30 million per km

The cost of Metronet should therefore come to -
$25 x 92 + $250 x 5 + 30 x 8 (mil)
= $2275 + $1250 + $240 (mil)
= $3769 (mil)
= $3.769 billion
Let's round that up to $3.8 billion

So even with a couple extras of my own, we come up with a costing less than anyone else's, but close to Labor's (which does not include quadding) Perhaps this is accounted for by the extra promises, such as the 17 000 extra parking spaces, the Circle Freeway or 144 new trains. Even so, the quoted costs did not include my extras, which were quite expensive. We really need public transport, so we need to make sure that money spent on it yields the greatest gains possible, by controlling our cost. What happened to our experience with the Mandurah Line, which despite its complexity, was one of the cheapest heavy rail projects in Australia of its time.

UPDATE: In line with my Liberals post, I've added an extra kilometre of tunnel, and taken that corresponding section from the surface rail part of the equation. My costing has been changed to reflect this.

The case for metros and tiering

Train systems in Australia often come as one product to serve a whole metropolitan area, which I think is best described as a suburban railway. This means that governments only have to maintain one rail-based product, saving resources, but this can cause problems as cities grow and public transport use grows faster than population.
To make trains attractive to suburban passengers, railway lines often skimp on stop spacing in inner areas to get passengers from the suburbs to the CBD quicker. This does not serve inner city areas adequately, even of there is already a railway line in the area. For example, the Mandurah line does not make any stops between the city and Canning Bridge, despite passing through dense or significant areas like South Perth or Como.

To deal with overcrowding capacity can be increased by measures that make use of existing rolling stock such as replacing seating with space for standees, such as longitudinal seating (along the sides rather than in pairs). However, this is unpopular with long distance commuters from suburban areas to the CBD who may be forced to stand for long periods of time. This seating is already common on A series trains, which do not serve far destinations except for the Armadale Line, but this modification is present at the ends of B-series trains, and is being rolled out on other parts of those sets, which are used on the 70 km Mandurah Line as well as the 30 km Joondalup Line, which is constantly extended to follow suburbia.
The solution here is to split the suburban railway into two rail modes; the metro for inner areas, and the commuter railway for outer areas. Metros will have close stop spacing to serve inner suburbs thoroughly. Their use will be higher and service will be more frequent (TUAG) because inner areas are more conducive to public transport use, being built when cars were not available or not widely used. The trains may be driverless, or operated automatically by computer, to provide this service cheaply, and will feature full longitudinal seating, because average trip length will be shorter.
On the other hand, commuter rail will have much wider stop spacing, for quick journeys from the outer suburbs to the city centre. Their frequency will be dictated by levels of use, which likely aren't as high as further in. But, of course, users of commuter rail will usually get a comfy seat.
Perth is part-way there, with the two tiers operating on the Armadale corridor (commuter to Armadale, metro to Thornlie) Monday to Saturday 6am to 12am, and supplemented on Sundays by an all stops service to Armadale. However, the rollingstock is still the same on both lines, the metro section (at least) should be grade separated, and of course the tiering should be full time, with the Thornlie line a bit more frequent.
The Midland and Fremantle lines are probably short enough to be metros, despite the express running in the peak, but again, grade separation is needed, and frequency could be better.
The Clarkson and Mandurah lines have inner tiers to Whitfords and Cockburn Central operating on the peak shoulder (formerly off-peak and the height of the peak before service cuts and fleet expansion respectively). This is completely grade separated but the stations are in freeway medians, and so are widely spaced and pedestrian/bike-unfriendly, the tier is part-time only and frequency needs a boost.

Why Park n' Ride isn't the answer

Park n’ Rides, or stations with large amounts of car parking, are usually considered good public transport policy. They encourage public transport usage in low density regions. Perth has them at most new stations, and we’re not doing too badly from them. However, maybe we could be better off with a different strategy.

Park n’ rides consume large amounts of land and discourage walk-up patronage. While our Park n’ Rides are mostly at freeway stations on the Joondalup and Mandurah Lines, where walking to the station is already hard, the large parking lots surely aren’t helping. There is also insatiable demand for them, just like roads. They often fill up in the morning peak, so they encourage peak usage over off peak usage. Instead of replacing car usage with PT usage, it lets car usage continue along with PT, which is an easy and fairly good outcome for now but not optimal.

In place of these alienating Park n’ Rides we could allow residential and commercial development, or TOD (Transit-oriented development), buildings that will generate much more traffic that is also more balanced than in Park n’ Rides, while earning money rather than costing money. To compensate for the loss of the car park feeder buses should be run much more frequently.

We could also place Park n’ Rides just beyond walking distance of the station (about 400m for most people) where the attraction of the station is lesser and so land cheaper, with links to the station by feeder bus. This would be the best of both worlds, but it would be difficult to get the land in our existing suburbs because it would probably be housing, and new suburbs should be optimised for walking, cycling and taking a feeder bus to the station, so large Park n' Rides aren't necessary.

In conclusion, we should reduce our dependence on Park n' Rides, although the idea suggested in the last paragraph of car parks beyond walking distance of the station deserves consideration.

The case for good off-peak public transport

Public transport is often used as an alternative to peak-hour traffic, but it should be embraced just as much during off-peak times.

Getting more passengers on off peak and weekend services gets the most out of capital investments already spent, such as railway lines, bus stops and vehicles. A bus has already been purchased so there is no extra cost to run it during off-peak times except for the driver and fuel, but the trip will earn money from fares.

Good off-peak services also gives peace of mind to peak hour commuters that should they need to come to work earlier or later, go home early, stay behind or do an errand that services are available for them at that time.

It also offers an alternative to driving to the city because while traffic isn’t bad, it will become an issue in the future, and parking is still a problem. The city is easy to serve well by public transport.

Trains run every 15 minutes all day everyday and some trips are standing room only, but most buses only run every hour during off-peak and so are unattractive to travellers with cars. Even major bus routes have a long way to go on weekends with only services along Beaufort St and on the Circleroute between Fremantle and Southlands running every 15 min on weekends.

In conclusion, off-peak public transport is a good use of infrastructure and fleets that have already been bought, and although Perth is doing well on this matter with its trains, we could improve, especially on buses.

Why widening and building more roads is not the answer

The usual ‘solution’ to a congested road network is to widen roads or build more freeways. However, it has been proven that building more road space does not relieve congestion in the long term. It just encourages longer or extra trips, with demand for even more roads, or induced demand. Building lots of roads makes cities bland and alienating, and difficult to walk, bike or take PT in, unfair to people who can’t or don’t drive. Also, the amount of roads needed to satisfy every possible travel need is impractical, consuming large amounts of land, and, unbelievably expensive

On the other hand investment in public transport or active transport (walking and biking) are cheaper and carry more passengers than road investments. They also take up less land and are future-proof; they not only can cater for population growth but will stay useful with global warming and peak oil.

In conclusion, building lots of roads will not solve congestion problems, which reminds me of a good quote by Orlando, Florida traffic engineer Walter Kulash, “Widening roads to solve traffic congestion is like loosening your belt to cure obesity”. If drivers want more roads then at the very least they should pay for them.

Why PRT is not the answer

Another commonly suggested solution to the problems of our current cars is PRT (Personal Rapid Transit), where small rail vehicles offer direct journeys not shared with others. They are usually very light and can operate very frequently, so some think of it as being better for the environment and having more capacity than public transport beating it at its own game. However, I would disagree.

If a PRT vehicle comes every second (very optimistic), then 60 vehicles will come in a minute and 3600 per hour. Single occupancy in cars is normal, so if there is single occupancy in PRT that would be a paltry 3600 passengers per hour. Even with 6 per vehicle (PRT-pooling?), the capacity is 21 600 per hour. If a railway line has trains that carry 1000 people every 2.5 min 24 000 passengers can be carried in an hour, and more is possible. If trains for 1500 come every 2 min, 45 000 can be carried, in which case PRT vehicles would need to carry on average 12.5 people, even with a nearly impossible one min headway.

If PRT can be made lighter than trains, why can’t trains be made even lighter? In addition, PRT infrastructure on every street (necessary for it to be useful) would be very expensive.

In conclusion, while PRT could work in an expensive private community, it is not suitable for serious transport needs in large cities.