The importance of contingency plans

I could moan on about Friday night’s experience with the replacement buses (a combination of poor night-time service provision and inaccurate passenger information) but you can read about that on Twitter, so instead let’s look at Monday night’s south eastern train problems.

It often takes multiple factors to result in a real mess.

So it was on Monday.

  • The Frankston and Cranbourne/Pakenham lines were closed for scheduled works, replaced by buses between the City and Caulfield
  • Many passengers had switched to the Sandringham line to avoid problems with those buses
  • Around 4pm, tragically a pedestrian was struck by a Sandringham line train between Brighton Beach and Hampton, forcing authorities to suspend services
  • And… an infrastructure issue at Elsternwick meant it was difficult to terminate trains halfway along the line – so the entire Sandringham line had to be closed.

This last one was the icing on the cake, resulting in huge queues on Princes Bridge and around Federation Square for buses.

In the past, they would have been able to run trains from the City out as far as at least Elsternwick.

If this had been possible, it would have had numerous benefits.

  • At least half the passengers could still catch a train to their destination
  • Far fewer passengers would need to be moved by the limited buses available
  • Buses wouldn’t have got stuck in inner-city traffic
  • Bus round trips would have been quicker, allowing them to move more passengers more quickly

Could they have run trains as far as Brighton Beach? It’s not clear. Some AM peak trains originate there, but sources say trains can only terminate on platform 1, which is not normally used, and is fenced off. This in turn would require the train to be emptied at Middle Brighton, where there are no staff – and this is not a great location to change large numbers of people onto bus replacements.

Elsternwick station

So what was the problem at Elsternwick?

Until recently, terminating trains at Elsternwick for planned works or unplanned incidents was a common occurrence.

But back in January, Metro flagged to their people that it was no longer possible, except in specific circumstances, and required extra measures to be taken.

Interviewed on Tuesday by ABC’s Jon Faine, PTV boss Jeroen Weimar gave an indication of the complications.

We have a turning point at Elsternwick, but it’s a manual turning point, it requires people to go out under safeworking orders. It’s something that you can’t activate on the spur of the moment. … It’s an old-fashioned manual set of points. That introduces a number of risks for people working in that environment. … It requires people to be working on a live railway, and when you do that without going through a safe and effective planned process, mistakes get made, and things get a lot worse.

Jeroen Weimar, 16/4/2019 on ABC Radio Melbourne (46 minute mark)

So is it really more difficult to get the right staff on-site at Elsternwick to go through the process, or to conjure up hundreds of buses to move thousands of train passengers, knowing they’ll never cope with demand?

In light of Monday’s mess, are they going to fix it? It turns out, yes. Jeroen Weimar again:

There’s some work that we’re doing next month at that location in Elsternwick, so there’s an upgrade program in place, which … will replace and update that set of points into an automated system so that we can turn around trains…

Jeroen Weimar, 16/4/2019 on ABC Radio Melbourne (48 minute mark)

I doubt this upgrade is directly in response to Monday, but rather with the knowledge that during May and June, they’re going to close the Sandringham line for more metro tunnel works.

The upgrade is good news. It makes sense to make Elsternwick capable again of terminating trains.

Obviously the less of the line gets closed, the less impact and the easier it is to manage.

Skyrail driver training - a 3-car Siemens train leaves Caulfield towards Oakleigh, Sat 16/6/2018

How many points?

What of the broader network?

In recent years, Metro have been removing points around the network, on the basis that less complicated track layouts result in fewer locations where trains need to slow down, and also mean fewer infrastructure faults.

This has some merit, but some level of flexibility is needed. (Recent changes at Caulfield mean trains from Dandenong can only terminate on platform 3, for instance. This has contributed to long delays in peak during bus replacements.)

Unfortunately, accidents will happen. An accident near one end of the line shouldn’t close the entire line – especially at the City end.

Ultimately, when incidents or planned closures take place, the infrastructure and the broader network needs to be able to cope.

  • Lead pic: James Oaten, ABC
  • April 2014: A similar scenario, nicknamed “Trainageddon” knocked out the Sandringham, Cranbourne, Pakenham and Frankston lines between the City and Elsternwick/Caulfield

PPTN – a good idea – in principle

The PPTN is the Principal Public Transport Network. The state government describes it as follows:

The Principal Public Transport Network (PPTN) reflects the routes where high-quality public transport services are or will be provided.

It supports integrated transport and land use planning by encouraging more diverse and dense development near high-quality public transport to help support public transport usage.

In other words, it’s Melbourne’s attempt to tie good public transport into planning and land use.

For instance, recent changes to the state planning provisions include relaxed requirements for parking near the PPTN (section 52.06-5), and discourage major facilities from being too far from the PPTN (section 17.02-2S).

And it’s specific about what form the PPTN should take (section 18.02-2R):

  • A metro-style rail system.
  • Extended tram lines and the establishment of a light rail system.
  • Road space management measures including transit lanes, clearways, stops and interchanges

It all sounds very positive. The theory of concentrating major development around good public transport, improving services, and allowing reduced parking can lead to good outcomes.

In effect it’s trying encourage development that will lead to people using public transport over cars, at least some of the time.

Melbourne Principal Public Transport Network (May 2017)

Residential parking

Residential car parking can get people very riled up. My reading of the current requirements is there’s currently no difference to off-street parking for residents, whether the development is on/near the PPTN or not. In both cases it’s:

  • 1 car park for each one or two bedroom dwelling
  • 2 car parks for each three or more bedroom dwelling

The only difference is that for developments not near the PPTN, there’s an additional requirement of one visitor car park per 5 dwellings in the development.

All this is a minimum. Property developers are able to build more car parks if they think their customers will buy them.

The problem is that car parks are expensive, especially the underground car parks normally associated with medium density developments of the type popping up around railway stations. Cost of parking adds to the cost of housing. And lots of car spaces encourages car ownership, and inevitably, car usage.

There are cities in the USA where they are completely removing the parking requirement for some homes around mass transit.

There’s no provision for this yet here, but the City of Moreland is seeking to override the rules, by allowing developments with no off-street parking in certain locations close to good public transport.

It’s not a requirement for no parking. Again, it’s up to the developers to figure out what the market wants.

These areas already have low rates of car ownership – so why not improve housing affordability for people who don’t want to own a car, or want to own fewer cars? As long as local streets have parking restrictions that prevent people parking lots of cars there instead, it can work.

(By the way, the Age’s story opens with: Itโ€™s a brave piece of social engineering – as if seeking to imply that requiring lots of parking isn’t engineering car dependency.)

Bent St, Bentleigh

Issues with the PPTN

I think the theory is good. But there are some problems with the PPTN as I see it:

The PPTN isn’t dense enough. It’s basically the existing train and tram networks, and some bus routes. If you assume that only the PPTN routes will be the ones that are really competitive with car travel, by providing long hours and fast, frequent service, there are some vast gaps.

Okay, so it’s just the principal network. Not every route can be part of it. But for instance, the PPTN includes no routes at all south of Frankston – not even to Mornington. That just seems illogical.

(More on this topic from Peter’s excellent Melbourne On Transit blog.)

The PPTN’s implementation is patchy. In fact, some of it doesn’t exist at all.

The PPTN mapping tool shows, for instance, a bus route along East Boundary Road, between Centre and North Roads. Good idea, except than no such bus route exists. One will be introduced in June, but it’s unclear how frequently it will run.

Even where the PPTN does exist, service quality varies widely, and it’s improving only very slowly.

It’s quite reasonable to assume that people living close to a railway station will (at least on the better served lines) be able to often use that service instead of driving. At the very least, they might be able to own fewer cars in their household than they would elsewhere.

The same might even be the case along some tram and frequent Smartbus or other high quality bus routes. But it’s highly unlikely along the many less frequent routes on the PPTN.

Buses can be improved, and relatively cheaply compared to major infrastructure projects. The bureaucracy has, or at least had, a plan to make them better. But it’s never been fully published, and there’s no hint of funding to make it happen.

This is not helped by literally billions of dollars instead being built on new motorways – the West Gate Tunnel (which includes substantial government funding, despite it being privately built) and the North East Link – which will only serve to help grow car traffic.

Bus stop underneath the skyrail at Carnegie

Great potential

The PPTN, and the associated planning provision, have great potential to help coordinate public transport and urban development.

But a lot more effort needs to be put into planning a more extensive PPTN, and of course actually implementing it.

Parking in the bus zone

I’ve mentioned this in passing before, but what the heck.

This is a bus stop on Jasper Road in Bentleigh, opposite Our Lady of the Sacred Heart College.

Normally it’s used by local bus route 701. It’s also used when there are planned rail replacement buses running, as the designated stop for Patterson station. These ran this weekend just gone, and will run again over the coming Easter long weekend.

Bus stop, Jasper Road, Bentleigh (Patterson)

Back in 2006, under a government plan called Meeting Our Transport Challenges, route 701 was one of the first to get extended hours, with services until about 9:30pm, and the introduction of a Sunday service – only hourly, but better than nothing.

When the rail replacement buses run, they are scheduled until about 1am outbound, with services all night on weekends.

Here’s the bus zone for this stop:

Bus zone in Jasper Road, Bentleigh (Patterson)

These hours apply on both sides of the road.

Not only can someone park their car in the bus stops after 7:30pm, two hours before the last bus, but they can also park here all day on Sunday and on public holidays, blocking buses.

This could make it near impossible for someone with mobility difficulties to board or alight the bus.

Most other stops in the area have been changed to full time bus zones. It’s only this one (one of few that actually gets 24-hour buses during rail replacements) that hasn’t been updated.

Some local stops have just a bus stop sign – no specific bus zone is marked. This is actually better than having a part time bus zone – regulation 195 prohibits stopping/parking there, regardless of the time (though plenty of people don’t understand this rule).

I’m not even sure whose responsibility it is to update the signs to reflect bus timetables. In this case it’s not the council; it’s a Vicroads-managed road.

Over the years I’ve raised this multiple times with various people via back channels, but today I’ve tried putting it into Vicroads directly. Will see what happens.

But I’m amazed it’s gone unnoticed for more than 12 years. And I’m wondering how many more there are like this?

  • Update 1pm. I sent an enquiry about this to VicRoads. They have referred me to PTV, so I have forwarded the enquiry to them. (Obviously the merger has only just been announced!)

Train Pain

If you were following me on Twitter on Monday morning, you’d know that I made a decision I came to regret: I travelled to work on the train to Caulfield and the express rail replacement bus into the City.

In short: the buses were hopelessly swamped by people. It was a long wait (43 minutes) in the bus queue; then a bus trip from Caulfield to the City that took more than an hour.

The trip that normally takes about half an hour from Bentleigh to the City instead took two hours.Read this Twitter thread for the grisly details.

In the cold light of day, here are some thoughts on this.

Flyer for April 2019 rail works

The good stuff.

The information has been pretty good. Lots of signage. Lots of staff making announcements on the train, on the station before the works started. It was pretty clear where to go.

There’s also been no shortage of advance warning: signs at local and CBD stations (the latter perhaps a little overwhelming), flyers handed out at stations, announcements on stations and trains, TV and radio advertising, press calls, web site updates.

There have been some errors along the way, and some of the media messaging was overly alarmist – leaving for instance some people assuming the Sandringham line is closed every day. It’s not – it’s only closed on some weekends. A simple calendar might have helped.

On the ground at Caulfield, they seem to have tweaked the verbal messaging from announcers to say “to the City”, which is much clearer (and more correct) than the previous “to Flinders“.

The massive tents at Caulfield are a smart idea. Ideally you don’t want long queues. But if you do have them, you especially don’t want people in long queues in the rain.

It reminded me a bit of the queues for rides at an amusement park. But not as fun.

I found passengers were remarkably calm while queuing. Maybe they were all posting furious messages on social media, but nobody got visibly shouty or upset, nobody jumped the queue, and (gods be praised) nobody lit up a cigarette inside the tent.

Metro was taking the operation seriously. I know this because I saw Metro CEO Raymond O’Flaherty walking around checking things out, with a suitably serious look on his face.

It’s school holidays and university mid-semester break (at least for some). This helps reduces travel demand, especially in the morning peak. The trains were quieter than usual. If these disruptions have to occur, now is the right time.

Running a stopping-all-stations every 7 minutes service from Frankston to Caulfield made sense to keep things simple and minimise passenger waits for trains. It’s similar on the Pakenham line, though Cranbourne passengers have to change at Dandenong, which adds to their travel time.

Happily for outbound passengers, trains on the Dandenong line run at least every 10 minutes until almost 10pm; until about 8pm on the Frankston line. High frequency makes outbound connections off the buses fairly easy.

On the bus there were delays due to traffic, but it seemed minimal delays due to just traffic lights themselves – apparently Vicroads was making some effort to cut delays to buses at lights, which was good… except…

Buses still faced long delays getting across Princes Highway because someone decided that cars on the highway are more important than scores of buses packed with people.

Edit: As far as I can make out, the express buses were split between Federation Square (via the M1) and Arts Centre (via Dandenong Road/St Kilda Road).

Caulfield station: a crowd exits the train to change to rail replacement buses

The problems

Train delays approaching Caulfield. Ours stopped for about six minutes, but this was nothing compared to some of the delays on the Dandenong line, with some trains stuck for an hour or more.

A big factor in this was crowding on the platforms. Caulfield station is not a modern design. Especially on platforms 2 & 3, the gates and ramps cause a bottleneck. (With the metro tunnel likely to increase Caulfield’s interchange role, an additional concourse would be a welcome upgrade.)

The track layout is also problematic – sometime in the past few years, the layout has been altered so that trains from Dandenong can only terminate on platform 3. The less busy Frankston trains could use platforms 1 & 2.

The sheer number of people meant crowding in the subway. The crowding in the subway backed up onto the ramp, and then up onto the platform, preventing empty trains from departing.

But the biggest problem was bus throughput. Standing in the queue for that 43 minute wait, there would be times when no buses at all would arrive for 5-10 minutes. Then a few would appear and fill up.

The crowds could have been handled much better if the stream of buses was constant.

Bus delays

Some express buses earlier in the morning had departed with spare capacity aboard, despite people still queuing. (This was actually shown in the timelapse in Channel 7’s news story.)

This needs fixing urgently – apart from making some in the queue needlessly wait, it wasted bus capacity that could have been used later in the morning to keep more people moving. Express buses shouldn’t leave unless full if people are waiting. Only stopper buses need to leave with space for more passengers.

Ch7 9/4/2019 - Still from timelapse of trip from Caulfield to City

Bus throughput was also affected by severe traffic delays – buses despatched to the City had a slow trip in, limiting the number of trips they could run.

  • Unbelievably, Normanby Road in Caulfield was still available to general traffic in both directions, with buses having to negotiate a sea of bollards. Did they seriously not think to just close it off?
  • There were also delays on Sir John Monash Drive for buses trying to enter and leave the precinct. Given the numbers of passengers, it would easily be justified to close down that section of the road (from Queens Avenue to Dandenong Road) to other traffic – or allocate an extra bus lane that can be used to get onto Burke Road.
  • At the very least, more time should have been given for buses getting across Princes Highway.
  • Each morning peak the M1 has been clogged. On my trip was a crawl from joining the freeway at Burke Road, up until about Toorak Road. (Can the automated signage on the “managed motorways” be programmed to show bus lanes? If not, maybe this should be factored into designs for future rollouts.)
  • After leaving the freeway there was another crawl along Batman Avenue, mostly caused by the sheer number of buses.

At this point on my trip I was looking on Google Maps at the traffic further along and onto Flinders Street and thinking/hoping they’d route some buses over to the Arts Centre to avoid it – as has been the case during previous rail closures.

Unbelievably, they didn’t. Instead our bus and all the others headed straight into a kilometre-long queue of buses and other traffic.

You know how some people claim that unlike rail transport, buses can easily adapt to changing travel demand and traffic patterns. Yeah. Not unless someone tells the drivers.


Eventually bus drivers started letting passengers out to walk the rest of the way. As I walked, I found a long queue of buses all the way to Federation Square.

Authorities implied it was because of the vegan protestors occupying the intersection at Swanston and Flinders Street. I don’t buy that.

  • Any westbound traffic delayed by the protest was separate from the bus route, because the section of Flinders Street used by the buses (from Exhibition to Russell Streets) was closed to other vehicles
  • Exactly the same problem happened again on Tuesday morning – with no protestors. And again on Wednesday morning.

To me it looked like the bus jam was due to a slow turnaround at Federation Square. They should have been getting buses unloaded and out again as quickly as possible so they could get back to Caulfield and pick up more passengers. Again the despatch procedures seem to need reviewing.

The lack of a dedicated bus lane on Batman Avenue didn’t help either, though it wasn’t as big a problem as on the M1.

For a bus and it’s driver, the round trip would have been at least 60 minutes inbound plus 30 minutes outbound. No wonder there were queues at Caulfield.

Buses queued at Caulfield during rail shutdown

Is the strategy right?

The government says 600 buses and coaches are being used this week for rail replacements, and presumably most of them are on the Caulfield to City run. But are they being used in the best way?

Getting everyone to Caulfield by train and then using lots of buses to the City worked during quiet times on the weekend when passenger numbers were lower. It struggled a bit with football crowds, but it really struggles with peak hour commuters.

It’s a firm reminder that buses can’t replace trains. It re-created the mess that occurs every weekday on the Doncaster corridor, where buses try to do that job.

The Doncaster buses are incredibly messy, but at least they have bus lanes on the freeway inbound, and along Hoddle Street and Victoria Parade. This was specifically ruled out by authorities – who chose instead to rely on traffic light priority (but then didn’t provide it crossing Princes Highway).

If they’re not going to give hundreds of packed buses the priority they deserve, would a better strategy be to shift more people onto the neighbouring rail lines, and run extra train services?

  • A bus connection from Moorabbin to Brighton Beach on the Sandringham line (as used in the past) for people from stations further south
  • Another from Westall to one of the Glen Waverley line stations for the Dandenong line passengers

With reduced holiday demand, and extra services, would they cope? (The Sandringham line is running extra services anyway – the usual 8 trains per hour in peak is 10 per hour during the works. These have been pretty crowded.)

There would be some level crossing impacts from more trains of course, but the shorter route for buses would mean fewer delays for them, and if enough extra trains could be provided, they would cope with passenger numbers so much better. (A packed train would be about 20 bus loads of people.)

For passengers closer in, the mix of stopping and express buses from Caulfield to the City would get a better run thanks to reduced passenger numbers and congestion – in the station at Caulfield and on the roads into Federation Square.

This sort of operation might use fewer resources, making the overall rail closure cheaper.

Boosting regular routes

A more radical change would be to run the above Westall and Moorabbin shuttle bus connections, but also deploy some of the hundreds of buses being used to boost local cross-suburban routes that connect the train lines, more widely distributing passengers across more stations.

Some passengers are seeking out these local tram and bus routes anyway, including some prompted by advice from Metro staff who knew about the chaos unfolding at Caulfield.

There have been reports of crowding on the 630 bus (North Road) and 67 tram (Glen Huntly Road) thanks to no extra services and (some) use of small trams. Presumably this is also an issue on the inner-suburban tram routes in the southeast.

For me, I used this method to get home on Monday. It was 58 minutes from Flinders Street to Bentleigh (via Middle Brighton and the 703), and that included a 17 minute wait for a late bus. Far better than the two hour trip in the morning.


Boosting those local bus routes and encouraging their use would be making the best use of buses – which are not really suited for long haul high capacity routes. It would also help educate passengers on their alternate routes during future unplanned shutdowns.

Obviously it’s a fine balancing act to try and make sure the swarms of passengers displaced by the rail shutdowns don’t swamp other services and replacement buses.

A good balance has not been reached this week.

Queue at Caulfield for replacement buses, Monday morning

The first weekday is always the worst. The queues at Caulfield have moved better on subsequent days. But that’s not an excuse.

It’s like when they say the first week of the university year is worst. Waiting for people to give up and try something else (such as driving) is not a good outcome. (Fair enough, for universities, sometimes travel patterns genuinely change after the first week or two.)

As of day three, the bus delays haven’t gone away – on the M1 and Batman Avenue, and around Caulfield, they have continued each peak hour.

The strategy used for this works period needs to be reviewed.

With continuing (and very welcome) works on level crossings, the metro tunnel, and upgrades for the new trains, these major rail closures will keep happening. While they do, the city and its transport network need to continue to function.

The monetary cost of driving

One of the issues that contributes to excessive car use is that it’s not straightforward to calculate the cost of driving.

Once you have the car, the cost of each additional trip you take in it is obscured. Apart from tolls and fuel costs, many might see an already-paid-for car sitting in the driveway as “free”, making it an easy option. This is why good alternatives are not based around park and ride, but instead aiming to replace the entire trip, enabling households to own fewer cars.

I thought I’d have a go at calculating the total cost per kilometre of my car.

Obviously there are a lot of variables, so each person’s result will be a bit different.

Depreciation: The Lancer I bought last year cost me $18,000 new. If I assume it’ll be near-worthless by the time it gets 200,000 kms on the clock (I’m not actually likely to drive it that much while I have it, but later owners might), then that’s depreciation of 11.1 cents per kilometre.

Some of the other costs are annual fees, so the cost per kilometre will vary according to how much I drive. The Australian average is 15,000. We’re well below that in my family, though it’s increasing a bit since one son got his P-plates last month. I’m going to use 8,000 km as an estimate.

Insurance: $708.20. (It’s definitely gone up since having a P-plater behind the wheel!) That’s 8.9 cents per kilometre.

Southland Station - shopping centre car park

Registration: having just bought the car, it was paid for the first year, but ongoing annual cost is $816.50. That works out to 10.2 cents per kilometre.

Servicing: This will vary a lot, and will get more expensive as the vehicle gets older. But for now, because I bought a brand new car, this is capped at $230 for each of the first three years = 2.9 cents per kilometre.

Petrol: The car’s got an information display which can tell you various things. One is how many litres per 100 kilometres it’s burning. From my observation this usually varies between 6 and 10, depending on whether we’re driving on country freeways with little traffic, or start-stopping along a busy city road. (On the bright side it never gets driven in commuter peaks.) The official government number for a 2017 Lancer is 7.4, but let’s be a little pessimistic and use 8.

How much does petrol cost? The Australian Institute of Petroleum reckons in Victoria the average price in 2017-18 was 135.2 cents per litre, which seems roughly right, though I wonder if it’s creeping up.

So every 100 kilometres we’re using $10.816 of fuel, or 10.8 cents per kilometre.

Thankfully this is not my car

What about tolls? We only use tollroads occasionally, perhaps about $50 per year, so I think I’ll exclude this for now.

So the cost for me is: Depreciation 11.1 + insurance 8.9 + rego 10.2 + servicing 2.9 + petrol 10.8 = 43.9 cents per kilometre.

(Contrast: public transport within Melbourne is generally $4.40 for any individual trip of up to 2+ hours, with a cap of $8.80 per day, but it gets a lot cheaper if you buy a Pass and use it regularly.)

Obviously there are a lot of costs that motorists don’t pay for directly. Driving is very heavily subsidised.

But having a number, even if it’s only an estimate, means I can quantify how much it’s costing each time we use the car.

Did I miss anything, or mess it up? What’s the cost in your household? Leave a comment!