Sydney Metro NW: higher cost than CityRail if no branching (part 1)

Branching of railway lines have a bad name from an engineering perspective (but often are a good thing from an economic perspective).

They can cause overcrowding (especially at inner urban stations) during peak hours, they can increase complexity and they can cause unreliability if they are on the surface (ie: if they are not in a tunnel or in an overpass).

They are also (wrongly) blamed for low frequencies (eg. every 15~30mins) during off peak periods.  However, as we shall see below, in fact it is the high operating costs of unbranched lines that make high off peak frequencies uneconomic.  As shown below, branching in this case **helps** improve off peak economics and is either neutral or positive for better off peak frequencies.

As for the issue of decreased reliability, this is so for surface ("flat junction") rail lines, where branches need to cross over the other rail line (of the bidirectional track pair).  But the flat junction problems disappear completely when rail lines are in separate tunnels.

Without a doubt, Sydney's network has far too much branching, with each end often branching not just to two branches, but often into three or four branches.  Clearly 3-4 branches are too much, but this does not necessarily mean a single branch point (into two branches) is bad.    In fact, around the world, numerous modern rail systems use branching - typically to a moderate degree of one branch at one or both ends.  Crossrail, RER, even Hong Kong's East Rail line all have branches.

So should Sydney Metro have branches?  Previously, a case was made that Sydney Metro SW does not need a branch.  Instead, by extending to Liverpool and then to Parramatta (via taking over the Cumberland line tracks), this would bring in enough patronage to fill up the 45,000 pax/hr capacity of Sydney Metro without branching.

On the other hand, Metro NW is a different story.  There is massive spare capacity and the many "string of pearls" employment centres strung up at Chatswood, MacPark, Castle Hill, Norwest and Rouse will all serve to reduce & even out the peaky, unidirectional passenger flows seen in Metro SW.  So for the foreseeable future to 2060 and beyond, capacity of 22,000 pax/hr will be the most that is ever needed.

Metro NW without a branch will also have very poor operating cost metrics.  The CIE (Centre for International Economics) Report released by IPART in December 2015 shows that all the hype about Sydney Metro's low maintenance driverless trains (that get rid of drivers and get rid of train guards completely), is just that - hype.  Benchmarks from around the world show that even the most "technically efficient" rail systems are only around 30% more efficient than Sydney Trains (in terms of cost of moving one carriage over one km).  For example, Singapore's Metro has a cost of $7.55 (to move one train car for one km) vs Sydney Trains $9.13.  Where the real inefficiencies lies is not lack of fancy train hardware or maintenance cost inefficiency, but in "allocative efficiency" - which is about patronage, rail catchments, service differentiation and demand side measures - ie: measures to ensure train carriages are kept full.  Allocative efficiency (ie: cost per pax trip) is where Sydney Trains underperforms by 300%, as shown below:

The Singapore benchmark of $7.55 to move one train car along one km provides an excellent way to cost Sydney Metro NW's operating expenses:

Note that the above equation:

1. Incorporates total operating costs ie: train operations, train & track maintenance, customer interfacing/ticketing systems & general overheads (but not infrastructure construction).

2. Assumes equivalence to Singapore Metro’s technical efficiency levels of $7.55 to move one train car over one km. This figure is valid for both peak & off peak. (Heavy 300 tonne trains will not change much in weight regardless of pax loading, ie: whether full or empty, the wear & tear on tracks, the electricity use, the rolling stock maintenance, general overheads, etc are all the same per car.km).

3. A high level of objective certainty exists for all above numerator figures - only the denominator involves “subjective” estimates.

4. A similar calculation applies to off peak costs (lower pax/hr denominator but also lower numerator with only 6tph instead of 15tph).  Peak hour however accounts for majority (55-60%) of total operating costs.

The table below shows what the operating costs per trip are once the patronage numbers are plugged into the denominator of the above equation (using NSW Government's NWRL forecasts as well as a +20% and -20% high/low patronage scenarios):

Note that Metro SW has much higher patronage than Metro NW, due to it not just taking Bankstown line patrons but also taking on Western line passengers for trains that will terminate at Central (passengers from these trains will then go down escalators directly from the Central platforms where the trains have terminated, and catch Sydney Metro to get to CBD stations like Pitt St, Martin Place and Barangaroo).  These Metro SW trains then travel north the full 40km - all the way to Rouse Hill and Cudgegong road.  Due to lower patronage on Metro NW, these trains will be substantially empty.  This is why the costs per passenger trip are so high on Metro NW.  In fact, it will be even more expensive than the average across the "old" Sydney Trains network.

As the problem is one of relative imbalance between Metro SW and Metro NW, densification of these corridors will actually worsen (rather than improve) the poor Metro NW cost metrics:

A second part of this post will show how this can be remedied by connecting Sydney Metro NW to a branch formed from the upper part of the north shore line (to Hornsby).  This will nearly double patronage whilst halving line length (20km to Hornsby vs 40km to Cudgegong Rd).  To give away a preview of the key outcome, the combined effect of these two measures, plus cost savings by truncating (ie: shortening) the North Shore line will overall reduce cost per pax trip on a branched Metro NW+Hornsby to as little as ~$1 per trip.  This brings a branched Metro NW+Hornsby to levels of allocative efficiency comparable to some of the best around the world.  So if you can't wait for the second part of this post, the take home message is branching Metro NW to also include Hornsby will dramatically improve allocative efficiency.  Alternatively, even higher service levels can be achieved for the same cost.