Frequency and accessibility – how should we evaluate DRT?

A method for working out the value of DRT is not yet fully developed. However, with increasing numbers of services maturing, we are reaching a point where we can go back to basics on the metrics of what makes a useful service and work from there...

Buses are about getting people around. They are the most used form of public transport, but in the current climate so much of the discussion is about commercial viability and cuts rather than enabling people to travel – with all the social and economic benefits transport unlocks.

---

Hear more about HertsLynx and why it works in a free webinar on May 10, 10.30-12.00

---

In this context, we see endless debates about the value of DRT and the cost per passenger or its commercial prospects. Those from more traditional bus industry backgrounds often demand improvements to fixed line buses even as those same services are being whittled away.

The HertsLynx service now has nearly two years of data showing increasing use, regular bookings and positive customer feedback.  Working out what makes services like this successful is key

And approaching DRT in isolation - without taking into account the wider potential for carbon reduction, equity or even the possibility of combining DRT with other services in a total transport model – is limiting. 

A method  for working out the value of DRT is not yet fully developed. However, with increasing numbers of services maturing, we are reaching a point where we can go back to basics on the metrics of what makes a useful service and work from there.

There are already well-defined elements of public transport that contribute to well-functioning services, and research into how responsive people are to service changes in each category.

Jarrett Walker set out seven basic elements of useful service in his book ‘Human Transit’. Amongst them, conveying people where they want to go and when is key. There are five more, but these two are the most fundamental and also the key drivers of cost. 

‘Where’ implies fixed line services forming an intricate network so that buses pass closely enough to the start and finish of people’s journeys and ‘when’ that they are sufficiently frequent so that the bus is there when passenger needs it. Both these things are expensive in different ways.

The challenge of low-density areas 

Obviously the more frequent a service, the more buses and drivers are required to run that service. In addition, to run close buses to a high proportion of potential origins and destinations – homes, workplaces, schools and services, implies that the network will be complex – requiring even more vehicles and drivers – and it will run into areas where people are found in lower concentrations, where buses are unlikely to be running at or near capacity.

And here is the rub. Low density areas are not like high density areas. There is copious research into the increase in ridership that can be achieved by increasing frequency in high density areas. However, in low density areas, there is just not the number of people to fill buses, even if everyone wanted to jump on the bus at once. It makes it hard to focus on the metrics of accessibility and frequency because achieving the kind of metrics that can be achieved for city centres appears unrealistic from the off.

This is where developing useful measures for the efficacy of DRT comes in. We can start to look for means of evaluating whether a service can get close enough to the origins and destinations (or onward travel nodes) for journeys, and we can also develop ‘frequency proxies’. 

Developing useful measures for DRT efficacy

The former is simple, we can look at maps of services and see what proportion of the population is within a short walk of the DRT virtual stops. [illustration]

The latter is more complicated because it’s trying to assess what level of DRT service approximates to the frequency of a fixed-line bus. Obviously, fixed-line bus frequency is pretty simple to measure, the buses stop at fixed points to a timetable a number of times per hour, whilst DRT services, which only travel when and where they are requested have no such easy measurements.

Instead, in principle, it should be possible to calculate a scale of service utility for DRT by matching how closely people’s trip requests are matched by journeys offered and calculate a proxy for the relative frequency of the service. There are possible approaches using features intrinsic to DRT.

With DRT we can see the trips people want to make – and those origins and destinations tend to be scattered across the area rather than falling along lines or corridors. One approach would be to work out the fixed-line routes which would be required to offer these journeys with those travel times and project what kind of frequency would make them attractive.

Another approach would be to see how many DRT trips are requested but not fulfilled. There are several reasons that this might be the case, one is that the user is just doing research (how long would it in theory take me) and is looking for options before planning a journey (in effect they are making a request without intending to travel).

The other is that the journey is not possible because some of it falls outside the area served or the service hours. The final is that no vehicles are available for the journey requested.

These latter instances – if they can be isolated – are most relevant for calculating a ‘frequency proxy’. It’s easy to see that if trips are turned down, the service is less useful. A simple metric could just be a proportion of trip requests for which vehicles are available.

However, it’s complicated. DRT services usually have booking windows where people can request their trips – for instance from two weeks to 20 minutes before they make their trip. A service that was not available from two weeks before the intended trip would be viewed as less available than a service that was not available if requested at the last minute.

A proxy for frequency could be designed as the optimal point in the booking window to be able to book the trip. So for example, if a passenger has to book a week ahead to be guaranteed their desired trip, the service has a low-frequency proxy, and the shorter the time is between booking and that booking likely being accepted, the higher the frequency proxy.

The real-world benchmark for DRT evaluation

These approaches offer a starting point for further investigation, by which we can assess the level of service offered. However, this is where the real world comes in. Whilst these approaches have a good chance of creating a tool for standardising and comparing services, they can only be validated by benchmarking against real-world experiences. Do the results correlate with services that are busy and well-used?

The HertsLynx service now has nearly two years of data showing increasing use, regular bookings and positive customer feedback.  Working out what makes services like this successful – both the on-the-ground team, communications and customer service – as well as the hard maths around meeting demand – is key. This is the data and information needed to create better bus services that get people where they want to go when they want.