Comment

Dwell time reductions: Good design plus people management

Alex Barron is an associate director and head of metro benchmarking at the Railway and Transport Strategy Centre (RTSC) at Imperial College London. The RTSC has been conducting metro benchmarking for more than 20 years through the CoMET and Nova metro benchmarking groups, where 33 metros across the world compare performance and share good practice. Here, he discusses some recent research he led into dwell time management. 

Many railways and metro systems are investing in additional capacity through a variety of technical solutions, including new or upgraded trains and signalling systems and, increasingly, greater automation. To achieve maximum capacity, a system approach is required: capacity is equal to train frequency multiplied by train capacity, and can be constrained by train design, train operations and capabilities, signalling systems, terminal capacity, and station dwell times. 

While investments most often address technical constraints, dwell times may be the single most significant remaining constraint to maximising capacity. In fact, extended or inconsistent dwell times may prevent the promised increases in capacity resulting from these investments from being achieved. Station dwell times are a key determinant of line capacity, passenger journey times, and resource requirements (e.g. trains/drivers). 

Context 

Recent research by Network Rail suggests that dwell times are a growing cause of delays, and research conducted by the Railway and Transport Strategy Centre (RTSC) at Imperial College London via the CoMET and Nova metro benchmarking groups has shown that many metros across the world find that train throughput is reduced because of dwell times. However, dwell times have not received as much attention as more engineering-focused topics (such as rolling stock performance or signalling). This might be because very small factors (seconds, or even fractions of a second) can make a difference, and because the challenges of managing dwell times are more about people than technical factors. 

Based on my experience working with metro and railway operators with the RTSC, the good news is that while improving dwell times is not easy, it can be inexpensive; the solutions often involve better measurement and management of existing practices (and making incremental changes to those practices) rather than complex engineering solutions. 

Extended or inconsistent dwell times lead to longer train journeys, longer passenger journeys, and irregular headways. These problems are more common where passenger volumes and train frequencies are higher, especially in places operating at or near capacity, such as the South Western Main Line into Waterloo or the core parts of the London Underground network. But even on quieter sections of networks, considerable time and capacity can be wasted if dwell times are not well-managed. The resulting impacts on punctuality (lower fleet and staff utilisation), regularity (bunching and crowding), and service quality (causing passenger frustration) ultimately reduce total capacity and lower customer satisfaction. 

On average, metro trains spend 25% of their time stopped at stations, but typically less than half of that time (only about 40%) is actually taken up with passenger movement (boarding/alighting from trains) – the remainder is function time (door operations/ waiting for signals to clear). Unfortunately, peak dwell times are often longer than planned for, which demonstrates both the problem and the fact that many operators are either not well-informed about dwell times or are not equipped to take action to address the problems. 

People 

Dwell time management represents the ‘human factors’ side of metro or railway capacity, focusing on both passengers and staff. Unfortunately, what is best for the individual passenger (such as ‘the stand in the doorway’ mentality) is not what is best for all passengers or for the overall service. Increasing numbers of passengers and trains at key times and locations are making this problem worse. A 2013 study by the CoMET metro benchmarking group, of which I was the lead analyst, identified the key passenger issues as boarders not letting alighters off first, passenger crowding holding or blocking doors, passenger crowding doors and vestibules, and passengers bunching on platforms, leading to uneven loading. The primary solutions to address these issues are not necessarily new, but it is critical that operators review and renew their approaches based on changing conditions to ensure dynamic management of the problems. 

Traditional solutions often involve campaigns, signs, and announcements, the best of which demonstrate to passengers why they should move along platforms or not hold doors. These methods should focus on the idea that bad behaviour impacts all passengers, and that it is in the passenger’s own best interest to comply. 

However, evidence suggests that static signage and repeated announcements will quickly blend into the background and be ignored by passengers. Instead, these messages must be dynamic; campaigns must be changed often, and announcements must be made live by staff, using a variety of phrases that are appropriate to the situation. Effective platform staff can make announcements to reduce crowding and reduce the risk of repeat door closures, to encourage more efficient passenger movement, and to assist passengers with reduced mobility (to help limit any dwell time impacts). 

Evidence from metros worldwide suggests that effective platform staff can reduce average dwell times by 1-2 seconds at busy platforms. Although this may sound insignificant, aggregating these savings over several stations in a critical network segment can ensure reliable and consistent operations, and enable additional trains to operate.

Signage and announcements must evolve with technology to convince passengers to change their behaviour. As more real-time data and better systems enable a better understanding of train loading and crowding, more must be done to manage the situation. Traditional signs like the one at Vauxhall on the Victoria Line (‘Front of train less crowded’) or repeated announcements to ‘use the full length of the platform’ have historically been used to get passengers to comply, but by using new technology it may soon be possible to show the space available on the next train in each carriage, providing evidence to passengers and increasing compliance.  (If you’ve travelled through Heathrow Airport Terminal 5, you might have seen something similar showing the average wait times at the north and south security lines.) If there were digital signs on platforms showing which train carriages were particularly busy, passengers would be more likely to move along the platform to find a more spacious carriage. 

Despite the critical role that staff play in managing dwell times, the 2013 CoMET study found that nearly all metro operators surveyed didn’t even mention staff performance as a cause of dwell time problems. All staff involved with dwell times must understand the importance of every second (or even fraction of a second), which will create an ‘every second counts’ culture throughout the organisation. To do this, it is necessary to measure and demonstrate problems and use targets/incentives to reward good behaviour. The increasing availability of data can help; for example, reports that track dwell times for specific drivers or guards, or at specific stations during platform staff shifts, could be a useful tool. 

Design 

Although managing dwell times is largely a people problem, the design of trains and stations can significantly influence passenger flow/behaviour and staff performance; the best people management cannot overcome design limitations. Door design and operations are the most important factor. The RTSC’s benchmarking research has suggested that a usable opening of at least 1.4m is needed to enable two simultaneous passenger flows. If doors are wider than 1.6m it is likely better to have an additional door (very wide doors can take longer to open or close and can therefore be easier to block or hold). 

It is also important to consider the total door interface available on trains: research has shown that the average for metro trains worldwide is for doors to be approximately 29% of total train length. 

For UK railways, the standard design of two 1.4m doors per 20m carriage is only about 14%; although this may be appropriate for outer suburban, regional, and rural applications, it is clearly time to rethink this standard model for inner-suburban services. 

Fortunately, the new Class 700 trains for Thameslink will have slightly wider (1.5m) doors, and the Class 345 trains for Crossrail will have three doors per car. Unfortunately, many new trains recently built or upgraded for busy London-area services will retain the traditional design for many years to come. 

In addition, it would help to make door closing more clear, perhaps by using flashing lights around the door frame (as proposed for London Underground’s New Tube for London design) or even countdown timers similar to those at pedestrian crossings. 

Perhaps even more important than the doors themselves is the design of the space immediately adjacent to doors, known as standbacks. Modelling recently completed by the RTSC has shown that the presence and size of standbacks have significant effects on boarding and alighting rates, primarily because they enable the full door width to be used (as illustrated on p19). Although additional open space adjacent to doors takes away valuable seating, it is critical that the full benefits of these spaces relating to dwell times are considered when such design decisions are being made. 

The future – technology and automation 

Although managing dwell times is primarily about managing passengers and staff, there are ways that new technology can help. Following the adage that what is not measured cannot be managed, it is essential for operators to capture detailed data about dwell times. By using systems to generate regular reports and analysing the data to look for trends at specific platforms, with specific train types, and specific staff members, it is possible to better target programmes and investments, including campaigns, training, and the placement of staff. 

There are other key opportunities to use new technologies in dwell time management. Using real-time data to communicate with passengers about train crowding is now possible with either weight- or CCTV-based systems. To assist with staff performance, increasingly automated supervision systems to manage train services (including setting dwell times) can be linked to driver or guard assistance systems, which is a good way to reduce the human variability inherent with staff determining the length of dwell times. Increasing levels of automation in metro systems has been associated with more consistent dwell times.  

At the far end of the spectrum, the use of full automation further reduces the unpredictable element, using platform doors and trains running at grade of automation four (unattended train operation – UTO). Although this introduces some additional challenges, such as how to calibrate dwell times and responding to sudden changes in demand (e.g. surges following special events), it has been shown to produce very consistent dwell times and to improve passenger behaviour. The graphic from RTSC research into the impacts of UTO on metros (see p19) shows the regularity of dwell times (stop times) and headways for systems with manual driving (GoA1), automatic driving (GoA2), and full UTO (GoA4). 

Conclusion 

To address dwell time, it is necessary to break down the problem, consider every element, and aggregate incremental improvements that can be consistently achieved. 

It is clear that a combination of management strategies and new technology can help metro and railway operators to respond to increasing and inconsistent dwell times, to ensure that train services perform at their best, and that the benefits from the significant investments in other systems can be fully realised.

''The challenges of managing dwell times are more about people than technical factors.”

Tell us what you think - Have yopur say below or email

[email protected]