The Eddington Strikes Back

Sorry about my blog absence of late – I’m in the middle of exams at the moment so haven’t had much time for anything interesting. I saw an article by Melissa Fyfe that appeared in the Sunday Age a couple of days back and it reminded me (as did Tom) that I never finished my Alternatives to Eddington series from a few months ago. The article suggests that Eddington (and the government’s forthcoming $20bn response) puts us on the brink of a transport revolution in Melbourne. For as long as I can remember, Melbourne has been on the brink of some kind of transport revolution – it just never seems to actually happen.

Cynicism aside, I was going to wrap up the Eddington series with a look at the prospects for a north south rail tunnel, but I’d already looked at the issue pre-Eddington and my ideas hadn’t changed. Moreover, we had some good discussion on the issues here, here, here and here; and Riccardo did a great series of posts on improving capacity on the Pakenham line. Recently though, I’ve been considering something more radical – getting rid of the city loop altogether.

Getting rid of the City Loop?

Although it is admired by many, the fundamental concept of a loop railway is flawed. For a detailed look at the substantial problems with the city loop, I recommend reading RIccardo’s excellent analysis of the issue. While critical of the merits of the loop, I’ve only ever seriously considered tinkering around the edges (by increasing through routing, making the loop unidirectional etc.) But ultimately the loop infrastructure is a sunk cost (pun very much intended), and we should look very carefully at what is most efficient now, rather than accepting existing operating practices at face value.

I’d never advocate closing the loop stations – that would be a terrible waste – but perhaps there is an argument to be made for substantial redesign. Instead of going around the city, the tunnels could be reconfigured to operate as a through route – with trains going from Richmond to North Melbourne and beyond without ever passing through Flinders St. This would require substantial redesign of the loop portals and would cost quite a lot of money, but it provides a potentially higher capacity and more efficient use of city rail infrastructure, allowing for a better range of trips and making the rail system useful for more than peak hour long haul commuting. A speculative plan of what I’m talking about is shown below:

As you can see, North Melbourne and Richmond become even more important for changing trains – for such a scheme to work properly both stations would need to employ effective cross platform interchange, which would require more flyovers. Of course for the plan to work all lines would need to operate at high frequency so that interchange wait times were as small as possible (not more than a minute or two on average).

Importantly, the link allows the north south rail line to operate independently of the rest of the network – a feature not present in the Eddington report. Furthermore, it makes the loop tunnels operate more efficiently than they are at present. Sadly, political considerations make it difficult – politicians seem obsessed with single seat journeys and bureaucrats seem to favour existing operating practices over new ones which suggest they had been wrong in the past. Still, with oil prices going nowhere but up, perhaps the transport revolution might eventually happen…

Once exams are finished I’ll finish the whole network map based on these ideas and should get it uploaded as a pdf soon.



UPDATE – To better show the broader plan I’m advocating, I’m uploading the whole map, available here. Apologies to Harry Beck. Keep in mind it’s only about half finished – I should have it done by Tomorrow or Thursday.

UPDATE 2 – I’ve got the map finished, it’s available here.

UPDATE 3 – As requested by James, I’ve made another version of the map showing new and existing lines. It’s available here.


54 Responses

  1. Glen Waverley to Geelong? (Or Tarneit?)
    I like it!

  2. Thanks Oben,
    I had Glen Waverley connecting with Werribee via Fishermans Bend – the main issue with this is that the reletively isolated GW line becomes exposed to the Werribee line with all its problems stemming from flat junctions, Vline trains and single track. These would have to be fixed. The lines could be through routed with other lines (I’m not sure which options are best), but the diagram does at least illustrate conceptually what I’m suggesting we do with loop infrastructure.

  3. Interesting. Have you got anything proposed for the Sandringham and Upfield lines in particular?

    Something I have thought about in the past was what it would be like if it was built as a two-way loop. Something like trains would run North Melbourne Flagstaff Melbourne Central Parliament Flinders Street Southern Cross North Melbourne – lines would follow the same direction around it all the time. Not something I’d advocate, just a thought I had.

  4. Ok I see.
    I also think Upfield should through route all the way (or further down) the Sandy Line via North-South Tunnel. Creating a nice trunk route through the Swanston/St Kilda corridor (avec Doncaster.

  5. Thanks Somebody and Oben. For the Upfield line, my view is that it should be part of the north-rail tunnel to Balaclava via Swanston St/St Kilda Rd. The other branch on the line would be to Doncaster. For Sandringham, I’d take the conservative approach of through routing it with WIlliamstown and Laverton locals. But – as always – I’d be interested to hear alternative suggestions.

    Somebody, interesting idea re. the 2 way loop – I’ll have to give it some thought.

  6. Just to break up the interesting conversation – Phin, your illustrations and diagrams are getting very impressive.

  7. Many thanks Lauren – I have discovered the wonders of Adobe Illustrator. On that note, I’ve uploaded the whole map – albeit half finished. It’s available here:
    I should get it finished in the next couple of days.

  8. An interesting extrapolation of the through-route concept to its logical conclusion. Why shouldn’t we end up with a London style arrangement – they were planning like this when their city was the same population as ours is now (or earlier)….

    Realistically I think fewer people than expected might need to change trains, as a lot of the City stations are already pretty close together – passengers might have to walk an extra block. Most should cope.
    And changing isn’t that hard – at Richmond on any given morning the 8/9(?) platform is chockers with interchange between FSS and Loop trains, so many have already caught on to the idea (probably ahead of the planners and pollies….)
    The Clifton Hill group gets a little orphaned, but it was always like that (especially in the old platform 14-16 days) and I can’t see an easy solution (if it is indeed a problem).

    Be nice to extend the Chadstone line up through the old Circle reservation (still exists as linear park, in a lot of places) up around to the Victoria Park interchange or Doncaster line… but that’s not just some new portals and operational changes : )

  9. Phin, I’ll post more later, but I have to say that that’s one very stylish map.

  10. Sounds like a great idea, but I can see one problem in particular. Currently, the longer lines have the worst punctuality and reliability. Here in effect we are making super long lines, up to 100 minutes from end to end, lots of time for little delays to add up.

  11. Yes, I reckon the Sandy/Upfield/Doncaster/Willy/Laverton metro/through route concept could work excellently as a separate entity. There is a real chance to turn this into an extensive metro style service which branches out quite widely.
    I’m thinking a DLR style concept (eg. Trains come in from the different branches and meet at the city – and in this case they would branch out accross town).

    I will develop this idea more and I’d like to see what you think when I’ve done a bit more.


  12. Jezza, this is true, but if combined with higher frequencies shouldn’t be too important. If frequencies remain the same then it would be a serious issue.

  13. Quickpost this image to Myspace, Digg, Facebook, and others!

  14. [URL=][IMG][/IMG][/URL]

  15. I’d hate to see a metro system implemented but built too far into the suburbs (ie commuter rail lengths) – like a Melbourne equivalent of Sydney’s proposed Rouse Hill line. Some people want every major rail line converted into a “metro” because it sounds trendy – I’d like metro’s for the inner tier on major lines, and commuter rail for the further out places.

  16. Keep in mind this is a very rough plan. The proposed lines are a guide of where I think metro style services could run efficiently and effectivley (I’m still wondering about the Willy/Laverton section).

    I plan to do much more research.

  17. No comments from me – except to echo Somebody’s call for metros not too long.

    Phin prefers a different approach to mine re linking Sandy and Upfield – but would be very happy if his plan went ahead instead!

    My limit would be 30 minutes, beyond that you have a comfortable commuter service layer, or a separate metro system (probably not justified in Melbourne, but could be justified around Parramatta)

    Done a fairly strongly worded post against DBowen for those interested.

  18. No comment from me, all good, agree with Somebody. My rule is 30 minutes of Metro, after that need Commuter layer (seats, express running) or a separate metro (not justified in Melbourne, probably though in Parramatta, although in this case would be a ‘lighter’ metro).

  19. This post is from Riccardo

    Phin, I like you idea for rerouting the loop.

    Wondering how it could be done. Presumably it would be too hard to dig into the deep underground and link the Northern Loop at Flagstaff say, with the Burnley Loop so that one line runs straight through.

    It owould probably be best to close northern loop and some trackage between FSS and Richmond for 6 months while you dig between the two – say using one Caulfield portal to reach the Northern loop but keep the other open (the point of connection suitably boarded up of course!)

    Then when that line is finished do a similar job with the Burnley loop at North Melbourne, use one portal to dig for the Burnley loop while keeping the other in use for the Northern Loop.

    When that is done, close the Clifton Hill loop and dig from that to a new portal at North Melbourne and from the City Circle loop (mostly unused) to a new portal facing Richmond.

    Is that right or have I got myself confused?

    Spoil could be carted out by rail at night.

    No track ends up redundant under this model, although some sets of points underground become emergency use only.

  20. This post is from Riccardo
    I should add, to make it neat, you could reuse the FSS portals for North Melbourne and City Circle for other routes you might have in future, or even just to help Clifton Hill trains reach platforms further south than normal.

  21. Many thanks for the comments all.

    Dave, you’re right that the Clifton Hill group is in some ways the loser from this plan. But they do get the interchange with the new metro line at Vic Park, ultimately sill have to change 3 times to get to Parliament or Flagstaff though. I’m not sure what else to do though – perhaps have a 2 way city circle running?

    Jezza, agree that putting long lines together to create even longer lines can be a big problem for reliability – that is a weakness of the plan for sure. But as Matt says, it becomes much less of a problem if service frequency is high and an even gap between trains can be maintained

    Oben – interesting plan! Am I reading it correctly in assuming lines 2 and 3 end up at Caulfield?

    Somebody and Riccardo, Completely agree about metros having a practical 30 (or perhaps even 40) min limit from the city. The perennial problem is how do we untangle a system like Melbourne’s which uses ‘suburban’ trains to cover all manner of sins. It may mean that some inner lines (say Jolimont – Clifton Hill) end up commuter and outer parts of others (like Craigieburn) end up metro. It’s a tricky one, and admittedly I don’t have the answer.

    Riccardo, no doubt it would be a tricky job to do, but I like your suggestion. Definitely worth keeping all the old portals. Nothing wrong with closing part of a line for a few months to get major works done. Melbourne seems to fear anything more than a weekend shutdown (perhaps because we seldom do major work) but London seems to cope quite well with shutting down lines like Waterloo and City and East London for months on end. What’s your view on creating a new 2 way city circle line as part of the works?

  22. I agree with Lauren. They should definitely contract you to make the new rail maps.

    Actually, have they made new Zone 1 & 2 only maps yet? I just assumed they hadn’t because they haven’t been put up on the Frankston line, but that’s not always a reliable indication…

  23. New Zone 1 and 2 maps are on some trains, I think they’re being progressively rolled out.

  24. In my opinion, this is a major step forward in the progress of the Melbourne rail system.
    At balaclava, run metro to sandy blue line to Oak/Mordi with emergency tracks so that they can run straight through.

    Run St Kilda/Port Melbourne as a light rail. At claremont st, tunnel portals with stop for interchange.

    Run underground to platform 10 at fss, removing heavy rail wiring.

    Run the Malvern corridor as a loop line to Werribee enabling version of smart C-P interchange with burnley/richmond/south yarra.

    Possibly kill Jolimont and run Clifton Hill lines through Richmond.

    Metro line run from balaclava through St Kilda (LR station), Alfred, Domain, Southbank, FSS.

    Anyway enough from me

  25. No criticism intended Calembeena, but how would Clifton Hills run through Richmond. Interested.

  26. Sorry Phin, you mean what do I think of having the loop go Richmond to NM with no FSS and SCS?

    Definitely, what they should have done from the start.

    And while you are correctly retrofitting it to make it work the way it should…I would not have built 4 tracks if working from scratch, only 2, and put the other 2 somewhere else. Would have been marginally more expensive by requiring new station boxes on this other route, but the station boxes could have been much shallower, and other savings such as not building the city circle chord, the extra portals and so on.

    FWIW, my comments on your plan, which is excellent, are:

    -Willy or Altona as a shuttle (but wouldn’t die in a ditch over this point)

    -not shown in your map at the moment (and probably not possible) but would provided different service levels for dual and single track sections, for example shuttles from Greensborough to Hurstbridge, Lilydale to Mooroolbark, Belgrave to FTG, Cranbourne and wherever else, pending duplication on these routes.

  27. Yes Phin, those lines end at Caulfield. The reason two lines join up is to create a large trunk route through the busiest corridor (in this case, Caulfield group to Swansty/St Kilda Rds and University etc) and where patronage is not as high (in this case Upfield Line), having only one “metro” line servicing it. Making significant infrastructure investments in one go would make it easy to implement higher frequencies when the need arises.

    I’m thinking whether the Willy/Altona lines (of the map above) should just terminate at North Melbourne or Footscray with good interchange facilities, and of course, a Newport-City tunnel to allow for the Geelong services.


  28. Thanks for the comments.

    Calembeena – I’d been considering undergrounding Port Melb/St Kilda light rail under the Yarra to Flinders St, but thought it might be too expensive. It would be interesting to see how much it would actually cost.

    Riccardo, agree that 2 tracks would have been plenty for this type of line, and that the other 2 should have gone under Swanston St. I was also was also wondering if the old city circle line should be reinstated – but running bidirectionally. But not I’m not so convinced – it would only be of any real use between Flinders St and Parliament.

    Oben and Riccardo, I could very well be convinced about running Altona and Williamstown as shuttles to Newport or Footscray. After all, they aren’t really busy, whereas Sandringham is. I’ll have to give this one some more thought…

  29. The map is now finished – it can be downloaded at

  30. The reason I would put CHL lines into richmond is it enables a lot more cross-platform interchange meaning that they don’t lose out as much.
    If you realigned West Richmond, you could redo the tunnel to go straight south under Hoddle St with a set of undergrounds at Rich.
    Even better would be more viaducts (essentially the line from WRich-VicPk) heading south with an island platform 11&12 probably at the angle of Brunton Av.
    The only reason that Jolimont is useful is the proximity to the MCG (easy walking distance to Richmond) and Wellington Pde (a tram corridor).
    This is really me trying to help out the CHL group as it needs all the help it can get.

  31. You forgot on your map to label Vic Park.

  32. And Malvern is Hawksburn.
    I really don’t have much to do today!
    I would suggest (as briefly detailed before)
    C/P/F to Werribee via loop and Fishy bend
    GW to Craigieburn via viaducts
    The reason I put GW to Craigieburn is hopefully by the time this ever gets completed, the NE Vic will be SG so Craigieburn would be isolated.
    I would duplicate Albion to Jacana for SG.
    New SG tracks Jacana-Craigieburn.

  33. As I don’t have much time, Phin, my only comment is that I think your Balaclava to Upfield/Clifton Hill line should be extended east to Caulfield.

  34. Thanks Calembeena and drwaddles.

    I’ve updated the map again to fix the errors/omissions (thanks for pointing them out).

    Calembeena, I see your rationale for sending the Clifton Hill group to Richmond, but it would be pricey. I wonder what the prospects would be of swinging the loop tunnels under Jolimont and putting platforms there would be — might be even more expensive.

    Ultimately, as a regular Hurstbridge line passenger, I wouldn’t mind if the trains just went to Flinders st, so long as the link from Victoria Park was up and running and services were frequent enough that changing twice to get to Parliament didn’t matter.

    Drwaddles, I’m in two minds about going from Balaclava to Caulfield. In its favour, the marginal cost of tunnelling a bit further may not be that high if undertaken as part of the bigger project. I’ll have to give this issue some more thought.

  35. I have to agree with Drwaddles about a Caulfield extension. It just opens up more interchange possibilities and destinations.

    Map looks very good.

  36. My intuition tells me that the metro should go through to sandy and the Willy/Laverton locals should go from Balaclava-Caulfield-Mordi/Oak as a local service over there.

  37. Interesting Ravenhall proposal in that map (although I disagree that any station between Sunshine and Ardeer is needed), surprised you would add something that you have not blogged about. Would love to see a post about the services to Sunbury/Melton/etc and the commuter-length interurban services!

  38. Thanks Oben, Calembeena and Somebody.

    The advantage of metro going to Sandringham is that it’s ripe for metrofication, whereas Williamstown/Laverton may be more at home with a S-Bahn shuttle off the main line.

    The Ravenhall proposal is something I looked at very briefly here:
    To be honest, I don’t know what to do with the Melton line There is a strong argument for electrification but I fear that it will just lead to contiguous sprawl all the way there – just as is happening with Pakenham. I was vaguely thinking of a S-Bahn style service to Ravenhall (with a bit of medium density development clustered around the line) but going no further so as not to tempt further land releases. Ultimately electrification will probably go all the way to Melton or nowhere though. I may well do a post about it in the near future – just have to work out exactly what I think about it!

  39. Somebody – also forgot to mention a disadvantage of going only to Ravenhall, namely that it’s only two stations from Melton. If we think that Melton should have a more frequent service than the rest of the Ballarat line (and I think it really should), then building only to Ravenhall would mean that the extra V/Lines to Melton would be only serving two stations – something of a waste.

  40. Riccardo ( – running shuttles for lines like Greensborough/Eltham would also be logical as for Greensborough-Hurstbridge in particular, silly to send long trains up there. Would you run the shuttle service as an ‘independent’ service as in connects with the trains from Melb but not guaranteed in order not to pass on the single track delays?

    Down of Eltham in particular is unlikely to ever require duplication so having some small train (or even a bus) running services up there could be a permanent solution for the loadings. Although, as I whinged about on this link, the loadings of my off-peak ex-Eltham to FSS train could have fitted in a single Tait on Friday – patronage for that entire line isn’t fantastic outside of peaks, let alone the stupid bit at the end.

  41. Hi Phin, long time reader, first time poster. Just wanted to say thankyou for that great map.

    If I may suggest another idea for a map. Could you make the same map with all the current lines coloured in black and the new lines coloured in red?

    Nothing I can add to the post except to say I sent this link to and told them to read it.

  42. Thanks Somebody and James.

    Somebody, personally I’m not convinced of the need for any further infrastructure works on the Hurstbridge line down of Westgarth, with perhaps the exception of doing something about the Heidelberg tunnel (although you could probably get away with keeping it as is). A while ago I proposed and off peak frequency of 10 minutes to Greensborough, with every second train going on to Eltham. Eltham-Hurstbridge would be run as a shuttle on the existing 40 minute frequency. The post can be found here:

    James – good idea re. making a version of the map to signify new vs. existing lines. I’ve uploaded it here:
    There may be a few omissions, but the idea is there.

  43. If duplicating the Heidelberg tunnel is not an option due to cost-benefit ratio then one option would be to duplicate the bridge over Burgundy St and as far as the tunnel which would mean that the capacity is increased without the major expense of a new tunnel.

  44. Thanks for the new map, it looks great.

  45. I’ll echo James there, that new one really makes clear what your overall proposals are and how they integrate together.

  46. Riccardo, could you give me some links to info about the Neubahnstrecke? Google is returning 1 result in english.


  47. Phin, I like your Greenborough-Eltham idea. Every second is about what the infrastructure could take.

    Certainly no promises of better services should be made until the infrastructure is improved

    I won’t go so far as Somebody does, but I would say that beyond Eltham is a ‘market’ that suburban rail doesn’t really cover well, and travel times are poor. It has more in common with say Healesville or Stony Point. And if Lilydales could belt it out down the line to the terminus, I would not be surprised if you could running a connecting service in a little over an hour to Healesville.

    I also estimate 1:15 to Stony Point, but that’s for another day

  48. Thanks for the comments.

    Tom, likewise if you went from the Rosanna side as far as Brown St, you could remove 1/3 of the single track for almost nothing.

    James and Matt – no worries. Matt, if you can only get German language info. I’m happy to translate for you.

    Riccardo, running the Hurstbridge line this way would be the most efficient way of improving services for the majority of its users – it’s just not good politics because Eltham people would get all up in arms about it. Another example of ‘fairness’ interrupting sensible solutions. I do wonder about the prospects for Healesville – would be interesting to look into it in more detail.

  49. Matt, this website (in English!) might help:

  50. Thanks Phin – I’m a personal fan of reopening the Healesville line, but it would be a stretch to argue it is good policy without more settlement

  51. I have just been reading the PT tech supplement to EWLNA from the DoI website.
    They explore the option of through-routing Northern and Burnley survices through the loop but only use the lower level.
    Have a read and see what you think

  52. I can see why you want to stop routing so many trains round the loop. But I can’t see why you’d prefer to wreck the Clifton Hill lines service than to send any trains at all around the loop.

    Not only does terminating at Flinders Street give poor access to some parts of the CBD, it also means passengers from long distance trains have to change twice to get to anywhere on the line. The only good thing that could be said about it is that it’s better than diverting the trains to Richmond!

    Surely this would be better:

  53. this is a really great idea. it looks like a proper metro system! you should show this to the morons who are currently doing the transport planning

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: