Rethinking peak hour with new transport models | KPMG | AU
close
Share with your friends

Rethinking peak hour with new transport models

Rethinking peak hour with new transport models

As Australia’s population grows and peak congestion gets worse, more data-driven decision making and a human-centred approach to transport planning and modelling is needed.

1000

Also on KPMG.com

Melbourne skyline at dusk viewed from the Studley Park overpass

Traffic congestion experienced in global megacities, such as London, Los Angeles and New York, is steadily winding its way into major Australian cities. In Los Angeles’ 'peak hour' starts notably earlier and ends later than Melbourne and Sydney, as shown in Figure 1.

Figure 1: ‘Width’ of the peak in Melbourne, Sydney and Los Angeles

Source: Victorian Integrated Survey of Travel and Activity, New South Wales Household Travel Survey, US National Household Travel Survey.

Note: Shows number of people en route to work (by car or public transport) relative to the peak for every 15 minute period from 5:00am to 11:00am.

This extended rush is known as 'peak spreading' – when people leave earlier or later in an attempt to avoid congestion. It’s a logical, human response and something that naturally occurs in most urban areas as populations increase. The main causes of peak spreading are described in Figure 2.

Figure 2: Causes of peak spreading in a growing city

Leave home earlier As congestion increases, some people will leave home earlier in order to get to work at the same time. For example, someone who leaves home at 7:15am to arrive at work at 8:00am may need to leave home at 7:00am to maintain the same arrival time. They may also leave home earlier to avoid congestion and get better run in the traffic, even if it means that they arrive at work earlier than the desired arrival time. This causes the peak to start earlier.
Arrive at work later Others might leave home at the same time (e.g. 8:15am), but due to increased congestion, arrive at work at 9:15am instead of 9:00am. They may also leave home later to avoid congestion, even if it means they arrive at work later than their desired arrival time. This causes the peak to end later.
Live further from work Cities may also expand in its urban footprint, particularly if much new development occurs on the urban fringe. This means that on average, people live further away from where they work. This additional travel distance means that some will need to leave home earlier, and others will arrive at work later. This causes the peak to both start earlier and end later.

According to the Victorian Government’s household travel survey, currently 28 percent of all car driver trips on the road at 8:00am in Melbourne are non-work related1. As Melbourne grows and congestion increases, some of those 28 percent of trips may begin to shift outside of the peak. The remaining 72 percent of work trips are becoming more flexible due to changes in the 9:00am-5:00pm work environment, contributing to some of the trips beginning to occur outside of the peak.

Peak spreading is not considered in infrastructure planning

As Australian cities continue to grow rapidly over the coming years, how we plan for and invest in transport infrastructure will be fundamental to our continued economic and societal success.

However, the models that Australian transport agencies currently use to help predict demand do not fully consider peak spreading, one of the important travel behaviour responses to congestion. The traditional, trip-based models assume that the shape of the peak in future years is the same as the shape of the peak today.

Predicting peak spreading

In 2017, Infrastructure Victoria adopted KPMG’s Melbourne Activity and Agent Based Model (MABM) to prepare advice to the Victorian Government on Melbourne’s transport policy and infrastructure needs. There are two aspects of MABM that make it suitable for modelling of peak spreading behaviour:

  1. MABM models people and their activities, rather than trips in isolation

    KPMG’s MABM is based on an approach known as activity-based modelling. Activity-based models represent how people trade-off spending more or less time at activities and/or being late for activities against being stuck in congestion or paying higher fares or prices during peak times. The MABM also represents how each individual makes these trade-offs in response to their own circumstances, including their daily schedule and constraints, their income and other personal factors. This is in contrast to traditional transport models which model volumes of trips, and may not specifically account for how different people make these trade-offs.

    Activity-based models have been widely used by planning agencies in the US and Europe over the last decade. Many metropolitan planning organisations in the USA use activity-based models, including the largest metropolitan areas of New York, Chicago and Los Angeles. Activity-based models are also used by planning agencies in Europe, including in the Netherlands, Belgium, Germany and Switzerland. KPMG’s MABM is the first metropolitan scale activity-based model to be applied in Australia.

  2. MABM models every minute of the day

    KPMG’s MABM uses a continuous timescale for modelling, allowing people to make fine adjustments to departure or arrival times to avoid congestion. By contrast, traditional models used for infrastructure planning typically do not allow for any adjustment in departure or arrival times, meaning they are not suited to directly modelling the peak spreading effects.

Peak spreading in Melbourne

We are already seeing peak spreading in Melbourne, as explored in a June 2018 article2 in The Age.

Our cities are expected to grow rapidly over the coming decades. By 2050 Melbourne and Sydney’s population is predicted to nearly double, bringing it closer to the size of the Los Angeles metropolitan area today. This population growth will make for significant widening of the peak. Projections from KPMG’s MABM showing the width of the peak in 2046 compared to today for the Melbourne metropolitan area is provided in Figure 3.

Figure 3: Projected ‘width’ of the peak in Melbourne 2015 vs 2046

Source: KPMG’s Melbourne Activity and Agent Based Model, Victorian Integrated Survey of Travel and Activity.

Note: Shows number of people en route to work (by car or public transport) relative to the peak for every 15 minute period from 5:00am to 11:00am.

Some examples of how people’s behaviour changes in growing cities, leading to peak spreading, is shown in Figure 4.

Table 4: Current and future behaviour of hypothetical Melbourne residents

Person Zoe Sam Lindsay
Lives in Werribee Caroline Springs Mount Eliza
Works in Melbourne Airport Melbourne CBD St Kilda Road
Behaviour today Leaves home at 6:55am by car and arrives at work at 7:30am. Leaves home at 8:15am by car and arrives at work around 9am. Leaves home at 7:30am by car and arrives at work around 8:30am.
Behaviour in future More congestion means Zoe has to leave home at 6:45am to arrive at work by 7:30am. Sam has flexible hours, meaning he can leave home at 8:45am and arrive at work at 9:30am to avoid the worst congestion. Leaves home at 7:00am and arrives at work by 8:00am to avoid the worst congestion.

The risks of making the wrong infrastructure decisions

Peak spreading is a natural way of managing demand, and can help us to get more value from our existing infrastructure. At the same time, excessive peak spreading may cause inconvenience, deterioration in quality of life and adverse economic impacts.

A lack of data driven, human-centred insight that attempts to capture 'real-world' behaviours like peak spreading can therefore lead to under or over investment in infrastructure – infrastructure that can take years to build and cost billions.

We cannot keep building our way out of congestion. The time is now for a smarter approach to transport demand modelling, which better understands human behaviours, and can help shape more innovative, cost-effective solutions that more suitable for the community.

Adopting agent and activity based models such as KPMG’s MABM that more closely reflect people’s travel behaviour is one way we can better understand the implications of peak spreading for transport policy and infrastructure, and as a result make smarter decisions.

To understand more about the activity and agent based modelling such as KPMG’s Melbourne Activity and Agent Based Model and how it can help with issues including peak spreading and infrastructure planning, please contact our team.

Connect with us

 

Request for proposal

 

Submit