When does walking beat buses?

It's a decision nearly every passenger who doesn't live next to a railway station has to face; is it quicker to walk home or wait for the bus?

The calculation is particularly important for Melbourne because though a minority of people live within 1km of a railway station, most of the rest live within 5km of one. To a large extent patronage growth depends on the abilty to attract new passengers from this 1 to 5 km belt.

There will be a distance below which walking is always quicker and is the preferred access mode. Planners call this the primary pedshed. This will be within about 400 metres of stops or station entrances. Walking is usually also fastest within a secondary pedshed, or up to about 800 metres from entrances.

At very long distances (eg 10km) the bus will always be quicker, even if the passenger has just missed one and needs to wait for the next one.

As explained before, it is the in between distances that are most crucial to patronage growth. The answers here are not clear cut, as it is not possible to determine which is faster without consulting timetables and making some estimates.

I will call the distance above which the bus is faster than walking the critical distance. This will vary according to factors like travel speed, distance, and for the bus, waiting times. The latter is set by service frequency or co-ordination and any late running caused by other traffic.

Let us assume a walking speed of 5km/h. For simplicity, we'll assume that the bus' speed is infinite. This will skew the numbers towards the bus, but only reinforces the main point that on local routes speed is worthless without frequency.

The following table shows how the critical distance is inversely proportional to service frequency, but can be made a small and constant amount with timed transfers.

Wm | Wavg | Dcrit (Wm) | Dcrit (Wavg) | Dcrit (Wco)
120 min ----- 60 min ----- 10 km ----- 5 km ----- 0.83 km
60 min ----- 30 min ----- 5 km ----- 2.5 km ----- 0.83 km
40 min ----- 20 min ----- 3.3 km ----- 1.66 km ----- 0.83 km
30 min ----- 15 min ----- 2.5 km ----- 1.25 km ----- 0.83 km
20 min ----- 10 min ----- 1.66 km ----- 0.83 km ----- 0.83 km
15 min ----- 7.5 min ----- 1.25 km ----- 0.62 km ----- 0.83 km
10 min ----- 5 min ----- 0.83 km ----- 0.42 km ----- n/a

Wm = max wait (service frequency)
Wavg = average wait (assuming random arrival)
Dcrit (Wm) = maximum critical distance
Dcrit (Wavg) = average critical distance
Dcrit (Wco) = critical distance co-ordinated services (10 min wait)

What does the table mean? If the bus runs every hour and one has just been missed, then walking is faster than waiting in all areas up to 5km from the station. This is the 'worst case' example; a critical distance of 2.5 kilometres (assuming a random arrival at the bus stop) is more representative where services are unco-ordinated (as in Melbourne).

This gap between the station's pedshed (800 metres) and the critical distance (in this case 2500 metres) represents those areas that could potentially benefit from the train, but cannot since feeder buses are ineffective and the station is beyond walking distance*. The bigger the gap, the more people miss out, and the smaller the gap the greater the station's catchment area and thus potential patronage.

Ideally the critical distance should equal the pedshed so that there is no gap at all. With an 800 metre pedshed, this situation is reached when service frequency is every 10 minutes. Assuming the worst case (bus just misses the train) the maximum wait is the same as the pedshed distance and the average is much less.

Where a 10 minute frequency cannot be achieved, the critical distance can still be lowered greatly by harmonising headways and carefully co-ordinating timetables. In this way every second train will have a good connection with the bus so people can plan a shorter journey time if they so wish. Even with a generous 10 minute connection time the difference between the pedshed and the critical distance reduces to insignificant proportions.

Hence a transfer time of 10 minutes is probably acceptable though not seamless. With improved pedestrian access to bus stops and stricter timed transfer arrangements (eg 4-5 minute connections) the critical distance could be lowered to 400 or even 300 metres. Such a reduction would most benefit activity centres that are slightly away from the railway station but becomes less important for low-density residential areas.

What conclusions can be drawn? The first is that there are currently large gaps between station pedsheds and critical distances. This creates large 'donuts' immediately outside station pedsheds where both walking and bus access to the station can be considered poor (especially when compared to driving).

Secondly, these poorly served donuts contain huge populations. If 20% of people live within station pedsheds, assuming 20 to 30% more (ie 600,000 to 900,000 people) just outside is not unreasonable.

As a result a strategy to fix transport beyond the pedshed by introducing Perth-style bus-train route and service co-ordination is likely to boost patronage, achieve modal share goals and improve utilisation of a largely empty bus fleet.

(*) It should not be overlooked that there is also such thing as a 'bikeshed', with a distance of around 2000 metres. Bike + train is almost certainly faster than bus + train without service co-ordination. However due to crowding, it is unwise to encourage bicycles on 'normal flow' peak hour trains. The alternative of using bicycle lockers at stations helps some, but if the destination is beyond the station's pedshed then the lack of a bike reduces ease of access. This gets back to improving bus access as described above.

Proposed V/Line timetables

Substantial service increases on RFR lines plus Seymour:

New timetables

Sure the express trains aren't quite as fast as originally promised, but this is more than outweighed by the greater number of services. To misquote Paul Keating, 'These are the timetables that bring home the bacon'.

Car worlders, pragmatists and hostages

A set of guidelines (pdf 114k) for intending commercial sponsors provides an interesting insight into how Metlink, Melbourne's public transport marketing body, sees the world.

In brief, Metlink divides society into those who are 'car only' and those receptive to using public transport. This second group is divided into those who choose to use public transport and those who use it for lack of other choices.

This is pretty standard stuff. It identifies extremes of people who (i) will always drive no matter how good public transport is and, (ii) captive riders who will always take public transport no matter how bad it is. The critical middle are the 'pragmatists' or 'choice passengers' who can be wooed either way.

However the guidelines also contain some assertions that I think are shaky. There is also no mention of other factors at least as important as the coarse age/sex demographic groups used.

For example, 'pragmatists' (which must be a key target group for any marketing) are described as being in the lower income range. Unlike 'car worlders' they are not described as having higher than average educational attainment.

No one disagrees that 'captive passengers' are likely to be low income. For these people public transport's social role is critical.

However I do not think that the explicit income or implicit education generalisations about 'pragmatists' hold much water. Disturbingly this positions public transport as a a residual service for low income earners with limited appeal for others.

I would encourage anyone interested in the topic to get an ABS Social Atlas and look at the following three maps: (i) Percentage of people with degrees, (ii) Higher income households, and (iii) proportion of people who commute by public transport. Train and tram network maps will also be helpful.

What do these maps show? Most striking is that there is a fair degree of overlap between education, income, transport services and patronage. People with high formal education live in much the same suburbs as the high income people, ie within 15km of the CBD.

Given high (and increasing) average incomes of inner-suburbia residents can't all be 'hostages'. Given their public transport usage habits they can't be 'car worlders' either. Therefore, they must be 'pragmatists', willing to use it when it suits their needs. And because there are more services in inner suburbs than outer suburbs, this is the case for a greater percentage of trips. Which is reflected in the modal share statistics. Thus there are heaps of 'pragmatists' that are likely to be neither lowly educated or low income.

Now onto other important factors.

One has got to be family structure. The economics of car ownership are worst for single people and best for families. A family can be driven for the same cost as one person, whereas public transport fares are largely 'per capita'. Then there's the practicalities, including strict time deadlines or carrying equipment. In contrast the single person for whom car costs are a larger proportion of income may find public transport more attractive. The concentration of single people in inner suburbs and families in outer suburbs only reinforces this due to better transport nearer the city.

Another is geography. I have already discussed the inner suburbs. Outer suburbs in Melbourne comprise a mixture of the following (i) old rail-based hubs with large populations of low socio-economic 'transport hostages' in suburbs up to 3-4 km away (eg Werribee, St Albans, Broadmeadows, Noble Park, Dandenong and Frankston), and (ii) Newer car-based suburbs with no rail and only a limited bus service. These areas have low rates of transit usage, have many people with trades qualifications and higher average incomes than the first group. Examples include Knox and newer parts of Frankston. Though not strictly outer, the City of Manningham is also car-dominated. Manningham excepted, these suburbs are neither highly degreed or earn the top incomes. If all these areas comprise a large number of 'car-worlders' then factors other than income or education must be more significant.

What are my conclusions?

The first is that Australians tend to be pragmatic by nature. Where public transport is good (inner suburbs) they will use it. Where public transport is poor (outer suburbs) they won't. Our pragmatism is good news since it reveals potential for patronage growth if choices are broadened.

Secondly, 'pragmatism' is distributed across all income and education groups and there are no grounds for it to be regarded as a lower income trait. Doing so is selling public transport short. However I would link pragmatism with intelligence!

Thirdly, classifying people by age, sex and income, though old staples of market research, may give inadequate results compared to geographic and household living patterns which may have a greater bearing on transport use.