Fast(er) rail in the North East?

Given that gauge conversion in the North East is going ahead, I thought it might be good to have a quick look at the prospects for making it faster than the planned 130kph line speed.

As I discussed in the previous post, rebuilding line to RFR standard costs around $1.5m/km. For the 200km North East project, that’s $300m. Of course, that cost doesn’t include upgrading the existing SG line beyond its current 130kph limit, but given the line is already getting concrete sleepers and other works, a signalling and level crossing upgrade may well be all that’s required to make the line 160kph ready (but then again much more work may be required). The same should apply for the SG line from Seymour to Melbourne.

Buying rollingstock to service the line would also be necessary. The original Vlocity contract put the cost of each motor car at $7.04m (525/ 76) in 2002 dollars – or $8.05m in 2007 dollars. The most recent contract extension put the cost of each motorised trailer at $4.75m (38/ 8 ) in 2007 dollars. The project will see 15 carriages converted, as I understand it to be marshalled into 3 5 car sets. In Vlocity terms, that’s 6 motors and 9 motorised trailers, at a total cost of $91.05m.

RFR infrastructure standards and trains (but not service standards) are achievable without a great deal of difficulty, but it doesn’t mean they are worth it. There’s still only 5 trains per day going beyond Seymour – hardly a lot of patronage to justify additional investment. The reality is that freight is driving the investment on the line, and the industry simply isn’t demanding line speeds above 130kph.

There is little point in investing money to get line speeds over what freight demands for a measly 5 trains per day to towns where air isn’t a serious competitor. The only reason to go above 130kph would be as part of a bigger strategy for frequent high speed rail from Melbourne to Sydney. But 160kph would be woefully inadequate for such a service if it is to compete with air. Line speeds of 200kph would be required at minimum, all the way to Sydney.

Whilst 200kph line speeds and diesel traction still wouldn’t deliver the requisite 4 hour journey times (as discussed a while ago on Riccardo’s blog) to beat air CBD to CBD, they might still be a good start. The Deutsche Bundesbahn upgraded many conventional mainlines to this standard in the 70s, which made the introduction of ICE services that bit easier. But it has to be part of a broader plan for high speed passenger rail between major cities (not one horse towns) – in absence of such a plan, I’d much rather have the money go to increasing freight capacity by double tracking Albury – Junee and Seymour – Melbourne.


Standard Gauge on the North Eastern Line – is it really going to cost $500m?

Last Friday’s announcement that the North Eastern broad gauge would be converted to standard gauge between Seymour and Albury was very welcome, if somewhat unsurprising news. The project has a great deal of merit indeed, and it is good to see it finally happening.

However, I was somewhat surprised to see that it’s going to cost $500m. I’m not advocating a lower standard upgrade – ultimately the whole line from Sydney to Melbourne should have high speed rail up and running – but we should still be aiming to get value for money. The reason the cost surprised me was that $500m for 200km of 130kph track seems excessive when the (heavily criticised) Regional Fast Rail track upgrades cost $750 million for 500km of 160kph track (plus other costs of $96m, operating costs of $73m and $550 million for rolling stock). Clearly this isn’t where all the money is going, as the ARTC are generally very good at project management. As I conclude later in the post, a lot of the cost seems to be reannounced spending for other projects, meaning the actual standardisation works will cost a lot less.

What exactly is the $500m buying?

The project includes quite a bit more than just gauge converting 200km of track. The $500m provides for the following projects:

Wodonga rail bypass

The long awaited Wodonga rail bypass is part of the project. The state government had already set aside $85m for the project, and the feds $45m (see media releases here and here). This is old money – the $45m federal commitment Anthony Albanese talked about is actually Howard era spending. But it still takes up $130m of the project cost.

Extra passing loops between Melbourne and Seymour

The media release is pretty vague about what this entails, saying only that it includes “Upgrades to the standard-gauge line between Melbourne and Seymour, including the construction of new passing loops as well as upgrade works already being undertaken by the ARTC”. The ARTC is quietly sinking over $2 billion into Melbourne-Brisbane rail upgrades, including $420m for passing lanes and $400m for concrete resleepering. I don’t know whether much new work is going to be done Melbourne – Seymour on top of this investment. The Albion-Jacana-city line needs a big upgrade if the project is to work, but I don’t know whether this will happen and how much of the upgrade is existing works and how much is reannounced spending.

Refurbishing 15 carriages and converting them – along with 3 locomotives – to standard gauge

Gauge converting the locos and carriages should be reasonably straightforward and inexpensive, and I was under the impression that carriage refurbishment was already happening anyway. I couldn’t find any reference for N-set refurbishment though, but even if it were new spending, it shouldn’t cost that much. Using the $10m budget for refurbishing 55 H carriages as a guide, we see that it costs about $182 000 to repaint and retrim a carriage. Given that 15 carriages are being refurbished under the project, we’re looking at a bit under $2.75m all up.

New platforms for stations on the standard gauge line

Only 7 stations will need new platforms (assuming Benalla and Wangaratta have their short SG platforms replaced). The most logical way to go is to put new side platforms to the west of the existing stations, in the former goods yards. This would mean the closure of the existing SG track to the east of the existing BG stations.

It’s a sensible idea for two reasons: firstly, it eliminates many curves in the existing SG by taking it straight through at the stations; and secondly, the new platforms would all serve up trains, meaning you could probably get away with just a platform and a bus shelter.

Gauge conversion and upgrade of 200km broad gauge track

Without knowing to what extent the existing SG upgrades have been reannounced in this project, it is difficult to know how much this part of the project will cost. Given that the Wodonga rail bypass is going to take up $130m, there’s only $370m left over for the rest of the works, a substantial portion of which may simply be reannounced SG passing lanes and concrete sleepering.

A rough calculation puts the capital cost of RFR at $1.5m/km (750/500) not including rolling stock. That’s $300m for 200km of new track. Given the potential for such a substantial portion of the NE standardisation costs to be for other things (like upgrading the existing SG track) and the known cost of $130m for the Wodonga rail works, it seems the actual conversion works are going to cost considerably less than $500m. This is a good thing, because I would have been rather worried if they cost more per km than RFR.

I hope to post shortly on the prospects for increasing track speed beyond 130kph and buying new rolling stock using the RFR costs as a guide, and whether such a project would be worthwhile at all.



Tipping Points in Public Transport Modal Share?

I was reading through an old post on Peter Parker’s excellent blog about service frequency and its capacity to change travel behaviour; and it got me thinking about the marginal modal share return to increases in service provision. Here, quality is fairly broadly defined – I’m basically looking at a combination of frequency, ease of use, comfort, etc. Working out how patronage and modal share respond to investment in better quality public transport is an important factor in determining the optimal level (and nature) of the investment and the consequent service standard provided.

Current government policy is to make small, incremental improvements to public transport and hope that more people use it, both in actual trips made and as a proportion of total trips (modal share). This was a big part of Melbourne 2030 – government policy was for 20% of motorised trips to be taken by public transport by 2020, basically a doubling of its modal share. Now given that total trips are increasing – largely because the population is growing and employment is increasing – we will expect to see public transport patronage to increase even when it’s treading water in modal share.

So why did the government set such an ambitious target backed up by only moderate levels of investment. There are two explanations that immediately come to mind: firstly, that they were never serious about 20% by 2020 (probably the truth); secondly, that their expectations of marginal returns to investment with respect to modal share were that it was more or less constant. This would see every extra dollar spent on improving the quality of public transport improving modal share by the same amount every time. This is represented graphically below:

So has this happened? Whilst patronage on Melbourne’s public transport – and especially the trains – has been increasing substantially, the modal share for public transport has been going nowhere. Basically, the investment hasn’t done much to improve modal share.

An alternative Explanation?

Clearly, the state government’s story doesn’t explain what’s driving transport habits in Melbourne (pun intended). I’d propose an alternative explanation – modal share isn’t going anywhere because public transport ultimately has to compete with the car. Even though there’s been investment in public transport, and an argument could be made that the quality has improved slightly over the past decade, modal share isn’t going anywhere because public transport simply isn’t as convenient as the car for many trips. For modal share to increase substantially, public transport has reach the point where it is a better quality service than a car. This is the tipping point I alluded to in the title. This is expressed graphically below:

So, what are the implications if we accept this idea? Well, basically it means that public transport investment needs to be targeted to meet the needs of drivers rather than looking at public transport in a vacuum. An empirical example would be the Metropolitan Bus Upgrade Program, where lots of money has been put into upgrading bus frequency and operating hours. But if a bus route has been upgraded from an hourly frequency to a half hourly frequency (making it twice as good), there’s still no reason why choice passengers will use it – it’s still not as good as a car. Until they reach the tipping point where they are as efficient (from a user perspective) as cars, buses are unlikely to be a roaring success.

It’s not all doom and gloom and you don’t necessarily need vast sums of money to get anywhere – small improvements can make a difference and big improvements needn’t be that expensive (improving off peak rail frequency for instance). I suppose my point is that policymakers need to be mindful of what the investment is targeting and how it affects quality relative to its competitors. Spending $1billion on Myki won’t get us much closer to that modal share tipping point, but spending $1billion on frequency upgrades probably would.

For a most interesting in depth look at providing quality public transport, have a look at this document from Canada, kindly sent to be by Jason.

Extending the Epping line to South Morang/Mernda and Aurora

Let’s face it – last week’s state budget was bad for public transport. Very little money was spent, and worryingly, some of the money spent was wasted. I’m talking about the $10 million that’s been earmarked for ‘design and development works’ for the South Morang rail extension. This is an absurd amount of money to be spending on design when the DoI secretly costed the whole project at $18 million in 2003. Of course, the government lied about the whole thing and claimed the outrageous price tag of $348 million. Similarly, the Aurora line was costed by DoI at $76 million, but the government claimed a $300 million price tag to wriggle their way out of building it. Below is the Melbourne 2030 plan for rail expansions to Epping North (Aurora) and South Morang – Mernda.

So, in this post, I’m going to have a look at the options for getting rail to South Morang (and beyond) as well as Aurora, how much they would cost, and what to do with the Epping line to give these new extensions a fast and frequent service. I should point out that I think it’s terrible planning policy to allow sprawl along these corridors, but if the government is insistent on allowing it to occur (quite a lot has already been built), then rail projects in the area should be considered.

Basic extension costs

Admittedly, the $18m and $76m construction costs for South Morang will only buy basic extensions, and are unlikely to include Keon Park to Epping duplication. Using the per km construction cost of $13.7m from the Mandurah line, we find that the 3.25 South Morang extension (to Civic Drive) would cost $44.5m, and that the 5.9 Aurora extension would cost $80.8m. The Mandurah line costs are – if anything – rather too high (that project included some inner city tunnelling, underground stations, freeway widening etc which is not a factor here), but it is fair to say that the DoI costs were somewhat too low for South Morang and about right for Aurora. They are nothing like the $300m + figures the government made up.

A need for extra works?

Defenders of the state government claim that the true costs are much higher because Keon Park – Epping needs duplicating and a host of other non reasons. If you want proof, have a look here – I love ZH836301’s It’s metal on slices of concrete FFS comment. But let’s look at this logically, if we are extending services further, why should more trains necessarily be running on the existing line? (assuming that Aurora is run as a shuttle) The reality is that the single track wouldn’t be much more of a problem for South Morang services than it is for present day Epping services. Beyond that, the single track isn’t even at capacity in peak hour – have a look at the diagram below – taken from a 2004 DoI report.

Now I’m not saying we shouldn’t duplicate Keon Park – Epping (I think we should), it’s just that it isn’t really an impediment to getting rail to South Morang and Aurora. It’s only 5km of single track anyway
– even if you tore the whole lot up and completely rebuilt it to Mandurah standards, it would still only cost $68.6m! My gold plated estimate is still only 32.5% of what the government claimed it would cost!

There’s probably already the need to build rail all the way to Mernda – another 8.2km from the South Morang terminus at Civic Drive. Again, using Mandurah costs it would be in the order of $112.3m. Total cost for high standard double track rail to Aurora and Mernda, as well as complete reconstruction of Keon Park – Epping would be 44.5+80.8+112.3+68.6 = $306.2m. We should remember that this is an extreme upper bound estimate and it still comes in below the government’s made up number. Coincidently, this cost forecast is exactly the same as for the Craigieburn bypass.

Service standard

I posted on a service standard for the Clifton Hill group a while back, advocating 6tph to Epping, with every second train running express Clifton Hill – Jolimont. 6tph is probably fine but if the line is to go all the way to Aurora and Mernda, it’s clear that every second train expressing 4 stations on tracks limited to 55kph won’t really cut it. Trains going this far out really need to run express beyond Clifton Hill as well. Either a two tier service is required, or some stations could be closed.

I would generally recommend against even considering inner city stations for closure in Melbourne, even when there is a tram line nearby. I’m generally in favour of keeping both heavy rail and street trams on the same corridor because they are sufficiently differentiated to not be very good substitutes for one another. But the Epping line is different, it has not one, but two tram lines running basically right next to it as far as Thornbury. Furthermore, one of them (the 112) is ripe for upgrading to high standard light rail. So for this section, we have a somewhat closer substitute for heavy rail as well as a traditional street tram.

On top of this, when we look at the AM peak patronage data for the line, it’s clear that there’s not really much patronage between Rushall and Preston at all anyway. This is probably because of such strong competition from the high frequency 86 and 112 trams. Upgrading the service standard for the Epping line would go some way to fixing this, but whatever happens there’s still going to be two tram lines competing with the rail.

Given these factors, there’s probably a case for removing Merri, Northcote and Croxton stations and upgrading the St. Georges Rd. track to do their job. To make this work, there would need to be a proper train/tram interchange at Thornbury. The 86 and 112 would use the Preston Workshops track to cross over each other – that is the northern section of the 112 would connect with the 86 and the northern section of the 86 would connect with the 112. Thornbury station would be rebuilt directly below the tram line to provide fast connections to heavy rail services. A new station like this shouldn’t really cost more than $15-20 million. The light rail line would need to run every six minutes or better in order to get average wait time down to an acceptable three minutes.

St. Georges Rd. would need to become proper light rail – that means real traffic light priority, fewer stops built to a higher standard (under cover island platforms), a decent track speed and larger light rail vehicles (like the C2 class). This concept could be taken further by building about 700 metres of tram track (for the cost of around $7.6 million) along the old inner circle rail alignment between Nicholson St. and St. Georges Rd. As at Thornbury, the lines could be swapped, with the northern section of the 96 going down St. Georges Rd. and Brunswick St. and the 86/St.Georges Rd. tram using the Nicholson St. reserve track. As above, the reserve track would need to be properly upgraded. This would better segregate street trams from potential high quality light rail, as well as better integrating inner-northern tram services with one another. The full map of what I’m considering is shown below.

I’m not completely comfortable with closing three heavy rail stations in Northcote, but even if this didn’t go ahead, the tram reforms I’m proposing are absolutely worthwhile. The cost would be low (700 metres of new track plus building/rebuilding five junctions and proper traffic light priority) and upgrades like platform stops and new trams are going to happen anyway – it’s just a matter of targeting them to the right places.

UPDATE – Below is an 800m pedshed for St. Georges Rd. as discussed. It takes in everything between Clifton Hill and Preston.

Express running – hero or villain? Part 2

In part one, I concluded that the growth in express running has outpaced the growth in trip length, and that this was done more or less to make rail more more competitive with the private car.

Has it been the best way make rail competitive though? The DoI thinks so, and contend that “express trains are a highly valued part of the metropolitan train timetable”.

The problem is that express trains are very path hungry when they have to share the same tracks as trains which stop all stations. So unless there’s more than two tracks, line capacity – and consequently frequency – is lowered. This isn’t just a problem for stations which are expressed, it reduces capacity for the whole line.

Of the 430 km of rail lines in Melbourne, 335 km are double track, 65 km are single track, and only 30 km are triple track or better. This means that on 78% of the network, there is a significant trade off between frequency and express running, and that 15% of the network (the single track bit) can’t handle either frequent or express services.

The DoI knows this, but claims that while “Reducing the number of express trains would help to increase capacity, but would significantly increase travel times from the outer suburbs and may dissuade commuters from using the train at all.” Really? In Melbourne, the rule of thumb is that every station expressed will save 1 minute compared to if the train called at the station. But you need to run sweeper services in between the expresses to serve the stations which are expressed. So, on the 78% of Melbourne’s network that is double track, every station expressed brings you 1 minute closer to the stopper in front, reducing the headway. Of course, stopping all stations trains can come in behind the expresses at whatever the signal headway is.

The upshot of this is that express running will increase average wait time on double track rail. Given that total travel time = time on train + wait time + walk to and from stations, the benefits of express running are mitigated. For more information on this, have a look at Riccardo’s training track page on true end to end journey time. Furthermore, we need to remember that lower frequency means lower capacity, meaning that the line can transport fewer passengers.

I’m not saying express running is bad per se, but rather that we need to be aware of the costs and benefits. Often, express services are a good idea, but generally on long, busy lines with the infrastructure (preferably 4 tracks) to support frequent services. I should note that where some express running is fixed into the timetable (like V/Line trains in Melbourne) then metropolitan expresses may be easier.

Consequences for Melbourne

There is a case for express running on parts of the Melbourne rail system. If expresses are needed anywhere, it’s on the Belgrave/Lilydale, Frankston and Cranbourne/Pakenham lines – lines which are closer to commuter than metro on their outer stretches. These lines do have a quite a bit of amplification to make it easier – Belgrave/Lilydale has three tracks to Box Hill, Frankston has three tracks to Moorabbin, and Cranbourne/Pakenham has a third track planned to Dandenong (who knows whether that will happen though).

Triplication adds some capacity, and worked ok for the sort of commuter service standard that was envisaged during construction in the 60s, 70s and 80s – but that sort of service standard is poor and ultimately unsuitable for Melbourne today. Most lines need 6-10 trains per hour all day every day. If express services are to be used, they should run at the same fixed pattern, all day every day – giving these lines a two tier service. The focus should be on shorter, more frequent expresses that can be delivered reliably within the parameters of the service standard.

What does this mean in terms of running patterns? For Belgrave/Lilydale, stoppers should run from Box Hill to Flinders St, and expresses stop only Box Hill – Camberwell – Glenferrie – Burnley – Richmond – Loop. For Frankston, stoppers should run from Moorabbin or Cheltenham to Flinders St, and expresses stop only Moorabbin/Cheltenham – Caulfield – South Yarra – Richmond – Loop. For Cranbourne/Pakenham, stoppers should run from Oakleigh to Flinders St, and expresses stop only Caulfield – South Yarra – Richmond – Loop. Additional infrastructure would be required, and ultimately 4 tracks to the intermediate termini is desirable. Here’s what it might look like to commuters:

Hurstbridge and Werribee currently see express running. For more detailed information on my plans for these lines, see here and here. For lines with V/Line trains, there needs to be a concerted effort at segregation, for a more detailed look at what I’d recommend in the western suburbs, have a look at this post.

Issues with Epping-Chatswood in my recent rail construction costs post

You may recall that in my recent post on Calculating rail line construction costs in light of the Eddington report, I compared the projected Eddington costs to empirical data (namely Mandurah and Epping-Chatswood). Well, Greg posted today that these costs have in fact blown out quite substantially. Total cost for the project is now much higher than the figures that I used.

So I had a more detailed look at how much the Epping-Chatswood line is costing, and the results are somewhat more murky. According to the NSW Audit Office, the total cost was originally forecast to be $1.6 billion (in 2000 prices), but these did not include $447 million interest on borrowings. The total final forecast cost of the project was revised upward by another $241 million in July 2006. Total project cost is now in the order of $2.3 billion.

But that doesn’t really answer the question, the $447 million of interest has nothing to do with how much it actually costs to build something if you’re not borrowing money, so we can disregard this. Total cost in current dollars is then $1.97 billion (indexed from 2000 prices to 2007 prices) + $247 million (indexed from 2006 prices to 2007 prices) is about $2.22 billion.

But the issue still isn’t resolved – the cost I estimated was based on information from Thiess – the project’s principal contractor. They were building their part of the project for $860 million (indexed by me to $984 million). My mistake was to assume this was the total project cost – it isn’t. According to the Transport Infrastructure Development Corporation, The Thiess Hochtief Joint Venture (THJV) is responsible for tunnelling, station excavation, and rail systems works; RailCorp is responsible for all rail works associated with the connection of the new line to the CityRail network; and A W Edwards is responsible for the fit-out of the new stations, construction of the station entrances and the urban domain works. I note on Thiess’ webiste, their cost has increased to $980 million – which is nearly exactly the same as my indexation of $984 million (or $78.72 million per km).

The upshot of this is that the per km cost that I worked out was basically accurate – but only if you want to cost tunnel with rail and station shells, but no station fit-out or station construction at the line termini. The problem is that this figure – whilst really good for costing new rail projects – is basically useless for comparison to Eddington because Eddington costs are all inclusive. The all inclusive Epping-Chatswood cost is $177.6 million per km. This is still only 51% of the Eddington cost.

Anyway, I hope that clears a few things up and makes my costs as accurate as possible – thanks for the tip Greg! The revised per km costs are set out in the table below.


Underground rail cost

Surface rail cost









Epping-Chatswood (Thiess)


Epping-Chatswood (total)


* does not include stations
# includes land acquisition
^ includes only tunnel, rail and station shell cost

Express running – hero or villain? Part 1

I was browsing through some Eddington stuff a while back and found some interesting information in Chapter 3 – Public Transport on a Roll. These sort of reports often contain great data on the rail system, and Eddington has been no exception. In chapter 3, there’s good information on the growth in average trip length and increases in express running. I’d been thinking about to what extent expresses are a good or bad idea in Melbourne, and this information gave me enough to get at least some good insights into the pitfalls and possibilities of express running.

Trends in express running – 1940 to today

The most important information from the report on this is shown below. It can be found on pages 74 and 75.

So what to these statistics tell us? Basically they indicate that between 1940 and today, average trip length has increased from 11km to 18km (around 64%), but that peak express running has in many cases increased even more substantially – 814% between Burnley and Camberwell, 106% between South Yarra and Caulfield, 72% between Caulfield and Cheltenham. Footscray-Newport, Clifton Hill-Heidelberg and Caulfield-Clayton had no express trains in 1940, but have many now.

Granted, this comparison is very very dodgy for a couple of reasons. Firstly, I’m comparing an overall increase in trip length across the network to evening peak express increases on just a few lines, overstating the increases in express running vis a vis trip length. The lines with the biggest increases in express running are the ones which are likely to have increases in trip length substantially higher than the metropolitan average.

Secondly, some of the express numbers in the second diagram are misleading. For instance, the 57% of trains that run express Newport to Footscray are Werribee services – but in 1940 the Werribee line wasn’t a suburban service at all, and wasn’t electrified until 1984! There’s also no mention of changes in the level of express running in key sections like Jolimont-Clifton Hill and Camberwell-Box Hill.

Even with these (substantial) problems, still basically reasonable to accept this stylised fact that express running has increased at a faster rate than trip length. The DoI claims that “Express trains were introduced partly as a response to competition from the rise in car ownership.” This is probably a reasonable explanation of why express running was expanded over and above trip length increases.

In part 2, I’m going to consider whether or not this has been a good thing for rail in Melbourne.