Real Journey Times Project
Preamble
A group of Transport for West Midlands (TfWM) staff were enjoying a drink after work in a city centre bar, celebrating the birthday of one of the team when in walked the Chief Executive.
"I can't stay long," she said, after giving her birthday wishes, "I have to get to an event at Edgbaston", meaning the cricket ground a couple of miles south out of town. "How long will the bus take?" she asked.
Around the table were experts in the bus business; network designers, timetable planners, transport modellers - all had smartphones with access to apps and APIs aimed at just such questions. But none of them were looking at their phone, they were already debating the answer.
Meanwhile the Chief Executive was using her phone. "I've looked it up online," she said. "I can get there in 15 minutes on a Number 45". Instantly the others stopped debating and agreed that "Oh no, you really need to leave a good half-hour at this time of day".
Anyone who uses a bus route regularly knows that the real journey time (RJT), is often much longer than what the timetable promises. So why isn't it shown on any app?
Background
Our population is becoming more urbanised. By 2050, 70% of our global population will live in cities; up from just over 50% now. In the UK, population growth in our large cities is similarly outpacing growth in many other places. This will put more pressure on our transport networks.
Journey times by road in UK cities like Birmingham, Manchester, and Leeds have continued to increase in recent decades; rising congestion is a major problem in most of our towns and cities. Since, in most cases, buses share road space with cars and lorries this has impacted heavily on bus journey times. In Birmingham, average bus speeds have dropped by 1% per year for each of the last ten years.
You can see this on the 126 service from Birmingham to Wolverhampton and how it's changed since the 1990s. In 25 years, timetabled journey times have risen by 25%. The number of buses and driver hours needed to maintain service frequency have risen to match. Bus fares have had to rise to cover the costs of drivers and buses crawling through traffic, and to make up for the fall in passenger numbers as people choose not to take the bus.
In the West Midlands there is a clear political desire to accommodate future economic growth by increasing public transport use rather than further increases in private vehicle use.
The bus passenger experience
The annual number of bus passenger journeys in the UK is trending downwards. Total year-on-year journeys made across England are down 1.5%. Some cities and regions, such as Reading, Nottingham, and London have bucked the national trend but in the West Midlands patronage has declined in line with the national trend.
Many transport operators and local authorities have invested significantly in live real-time arrivals services, helping customers to understand when the next bus will arrive at their stop. The RJT dataset builds upon this and offers a solution to a further customer problem; rather than when will my bus arrive at this stop customers can accurately plan when will I arrive at my destination.
What is Real Journey Time (RJT)?
You've probably already used something like RJT, but not on the buses. If you search for directions on Google Maps you'll see a 'typically takes' time. We are creating the same dataset for buses.
RJT can be used to make an informed prediction about when a customer is likely to arrive at their destination. RJT therefore delivers an enhancement to the customer experience of existing real time bus data.
What we built
The RJT tool is available now at www.realjourneytime.co.uk. It may be a bit slow while we focus on adding features instead of optimising performance, but if you're patient it will work.
The implementation is complicated, but the basic idea is very simple. We use the excellent TfWM API to ask for live departure information for every bus stop in the West Midlands every 5 minutes. Then we use that information to track every bus service on every route as it makes its journey across the city.
Over the past 5 months we've collected 25 million bus departures, with thousands more being added every minute.
Analysing this data to recreate a version of the bus timetable for the whole of the West Midlands based on real journey times is the hard part. To predict how long an individual journey will take we look at every single journey on that route at that time over the past 5 months.
We have decided to set our real journey time at the 95 percentile, meaning we discard the 5% of journeys that take the longest time, and report the time of the worst journey that's left. We produce graphs showing how travel time changes over the course of a day and timetables in ATCO-CIF and GTFS format so that travel time accessibility maps can be created in tools such as TRACC and Open Trip Planner.
The Real Journey Time system lets us analyse any journey on any high frequency bus routes in The West Midlands. Here we've picked buses from Dudley to central Birmingham.
We look at over 10,000 bus journeys from Dudley to Birmingham, how long they took, and how long they were timetabled to take. What's instantly obvious is that many buses run to time.
Looking more closely we see that buses run slower than the timetable at peak times, and that the timetable has already been changed to reflect this.
In the morning peak it is best to leave 90 minutes for the 70 minute bus journey from Dudley to Birmingham.
We've already learned a lot. First, that the majority of buses run on time. Second, that the delays at peak times (into town on weekdays and out of town on weekdays and Saturdays) are very large on most bus routes. They often add 50% to journey times, meaning a 20 minute journey takes 30 minutes. Delays on some routes are typically 100%, meaning a 20 minute journey takes 40 minutes. Most surprising to us was the effect of the school run on our data. Birmingham's rush hour starts at 3:30pm, and is worst at 4pm. And this school-time traffic affects traffic all across the West Midlands, not just the city centres as with rush-hour worker traffic.
Keeping bus operators on side
Everyone knows that buses run late. And everyone knows that its not always the fault of the bus operators. In fact, most of the time its not their fault.
An important part of this project was working with bus operators to produce a tool that helped them and their customers, and not a tool to blame them for the delays over which they have limited control. This was especially important because of the role of Traffic Commissioners within the UK and their ability to fine operators for poor performance.
Transport for West Midlands and National Express West Midlands (the majority bus operator) have a positive relationship. Working with other operators and stakeholders they recently formed the West Midlands Bus Alliance to work together on issues and innovations such as this. By working with the Bus Alliance we have been able to make sure that RJT data is used to improve services, not criticise operators.
We are already identifying where investments should be prioritised to improve bus services. And National Express West Midlands are already a primary user of the RJT dataset, with plans to incorporate the data in their new look journey planning (and mobile ticketing) app - due for release in late 2018.
Next Steps
A key part of the project involved the development of future potential uses and applications of the data. We worked together to highlight a high level activity roadmap in four main areas:
- to find the causes of delay to inform plans and projects to fix them, worst first.
- to make the case for funding projects to remove delays.
- to make the case for more devolution of powers to manage delays.
- to immediately better inform passengers about (currently) unavoidable delays since not all delays can be controlled, removed, or bypassed.
We are working with TfWM to get the most they can out of the RJT data by using the four applications listed above, as well as others.
We're focused initially on helping to communicate this information to passengers. We think it is important that customers know how long their journey is likely to take. We think that more accurate journey planning is especially important for customers who make multi-modal journeys - taking the bus to catch a booked train for example.
Our final next step is to bring RJT to another big UK city. We've already seen how it can help in Birmingham and were sure it can help in other cities with multiple large operators, large populations with declining bus patronage, and increasing congestion.
If you want to know more about the RJT tool, the project or the data, get in touch with Tom or Neil.
