Jun 4, 2013 - Martin Wedderburn. Independent Transport .... the way pedestrians choose their routes (Ciolek, 1978, Guy, 1987, Helbing &. Molnar, 1997 ...
NETWORK, NETWORK, NETWORK: PEDESTRIAN MOVEMENT ANALYSIS AND ACTIVITIES Martin Wedderburn Independent Transport Consultant Alain Chiaradia Researcher in Spatial Design Economics, University of Cardiff
1. INTRODUCTION Walking was long regarded as the invisible mode in transport analysis. The fact that every single journey we make involves an element of walking meant that traditionally walking was taken for granted or ignored. And as pedestrians (at least the most able-bodied ones) are arguably the most flexible transport mode, they were often treated as an afterthought and accommodated where space allowed. However, in recent years there has been an increasing realisation of the benefits of providing high quality public spaces to encourage walking. This forms part of a wider cultural shift in transport planning towards demand management as an alternative to predict and provide planning. As a result, tools to analyse walking have become an integral part of mainstream transport planning practice in just a few years. The term pedestrian movement analysis covers a range of tools to model patterns of pedestrian movement. Many transport planners may be familiar with the colourful heat maps that are used to visualise concentrations of movement, although they may not be familiar with the different techniques available. However, as transport practitioners we need to understand the relevance of pedestrian planning tools for the design of the wider public realm beyond station surroundings, for example for town centres and neighbourhoods as a whole. This paper provides an overview of several types of pedestrian movement analysis tools, present some recent examples and highlight several issues for the future. 1.1.
Pedestrian behaviour
Research and experience has shown that walking patterns are predictable but that pedestrian movement differs from other transport modes:
Walking speed is largely unaffected by the infrastructure provided. Variation in speed is observed by age group, journey purpose and size of city (the larger the city the faster people walk!). Distance clearly matters but cognitive and environmental quality dimensions are also important. People can stop at any time and they do not have to park. Most socioeconomic transactions are pedestrian based.
People use both perceptual information (what they can see, hear, etc.) and inferences (guesses about things they cannot directly perceive) to construct mental maps of an area. These mental maps then inform route choice plans across a movement network, and they change in response to new information encountered. These ‘way-constructing’ and ‘way-finding’ processes allow people to organise public spaces and their attributes into a safe and easy spatial pattern to navigate. Beyond mental maps, route choices are also affected by people’s perception of their own physical strength and endurance. Age and gender will affect the journey range greatly. It can also be seen as the journey budget and can be defined either as a time or distance radius (within 400m or 5 minutes walking time, 800 min or 10 minutes). Spatial accessibility is the interface between an individual’s ability to navigate through an urban environment and the configurative qualities of that environment. This element of individual ability is constantly evolving, increasing as spatial information becomes more complete. As a result, the more an individual resides in a location, the more their spatial abilities increase. In fact it is possible to categorise at least three levels of spatial ability in individuals (Meilinger, 2008, Montello, 2005). The first level is landmark knowledge. Persons with landmark knowledge are able to recall the characteristics (cue and function) and location of a place. The second kind of geographical knowledge is route knowledge, where people are able to link landmarks with directions for getting from place to place. Route knowledge includes directions for navigation (sometimes called procedural knowledge). The third level is map knowledge, survey knowledge or configurational knowledge. Persons with this ability know the interrelationship of places and routes with each other. Map knowledge often includes information about distances (proximity) and angles between features (legibility) and the capacity to project the possible route (sometimes called inference knowledge). The levels of crowding or congestion of transportation facilities can be measured using a 6-digit scale from A to F, with A indicating free flowing conditions and F indicating extremely crowded (Fruin, 1970). Strict application of this approach is inappropriate for pedestrian projects since user perceptions of congestion and crowding depends on the type of environment and the types of activities they are engaged in (Transport for London, 2010). 1.2.
Pedestrian networks
From the perspective of a professional, understanding the nature of pedestrian networks is essential to understanding how people move around on foot. The emphasis is on the role that various aspects of the built environment have on influencing pedestrian movement dynamics. Network abstraction is the process of making pedestrian paths visible and simple to understand. The title of this paper emphasises the importance of networks in the design process, and will illustrate the challenges in getting the level of network detail right for the type of analysis.
One way to illustrate the types of network and their relevance for pedestrian route choice is to consider the difference between a labyrinth, a grid and a maze. Figure 1: Thinking of city layouts as labyrinth, grid and maze
A labyrinth is defined as a construction that leads from a starting point to a goal by a single path, with no branches or dead-ends.”, people will “never” get lost in a labyrinth because of its unicursal character.
Ancient Greek 7 coils labyrinth
Many Underground stations can be thought of as labyrinths since irrespective of the complexity of the path there is one way in and out. In term of pedestrian route choice preference, this is easy for most pedestrian software to deal with. Issues are capacity and crowding management, i.e. maintaining a good level of service. In a perfect grid street plan, streets run at right angles to each other. A grid is multicursal and calls for the traveller to make a series of decisions, which affect how quickly the destination is reached.
Pure grid
If it was not for the architecture as landmark every block would be the same. We would not know where we are if it was not for a system that distinguishes the blocks, such as a street name system or a numbering system. Understanding pedestrian route choice preference is therefore not easy. A maze, is also multicursal and calls for the traveller to make a series of decisions, which affect how quickly the destination is reached. It is apparent that a maze is designed to make the task of navigation as difficult as possible. This is mainly due to the multiple dead ends and multiple rotations that cause people to lose their sense of overall direction. Mazes are fine for fun fairs but not ideal for the design of a city. European cities are not pure grid, labyrinth nor mazes, but city design mixes these three spatial layouts.
Ancient Chinese maze Taken from Zhang L, Chiaradia A, Zhuang Y (2013)
2. Three levels of pedestrian movement analysis Methods for pedestrian behaviour modelling can be classified by size and scale of analysis. There are essentially three types of tools that can be classed as pedestrian movement analysis. These are:
Multivariate analysis - referring to all forms of analysis that predict
movement patterns through a statistical relationship between urban layout variables and observed movement (e.g. Space Syntax). Assignment modelling – referring to all forms of analysis that estimate movement between origins and destinations, and assign this demand to routes. Micro-simulation – referring to agent-based modelling tools to simulate in detail the interaction between pedestrians.
The following table summarises the relevance of each of the three levels of analysis. Table 1: Three levels of pedestrian movement analysis
Multivariate analysis
Assignment modelling
Micro-simulation
Purpose
Establish a statistical link between network configuration and movement
Incorporate pedestrian movement into a traditional transport modelling and evaluation framework
Understand pedestrian comfort and safety at a detailed level
Role in the design process
Option generation and testing
Planning, feasibility, appraisal
Detailed planning and design
Scale
Area or neighbourhood wide
Area or neighbourhood wide
Individual station or junction
Method
Calculation of the statistical relationship between activity density and network
Calculation of change in trip production / attraction
Simulation of pedestrian movement and interaction
Calculation of potential movement Cost
Low
Pedestrian route assignment model Medium
Calculation of density measurements
High
In recent years there has been much attention focused on micro-scale modelling, in particular evacuation scenarios (Xiaoping et al, 2009) associated with stations and sports stadia, both subject to stringent guidelines on crowding and safety. These models are effective in predicting crowd
behaviours in high density, confined areas with known origins and destinations. Local scale software tools have been also developed for very limited urban areas (Papadimitriou et al. 2009), although these models are not designed to be applied to more open ended environments. Another aspect of interest is how these tools are integrated in the design process. How feasible is it to integrate the tools at the early stages of the design process where extensive data may not be available, yet where most design added value occurs and where risk can be high. All three forms of analysis therefore have a distinct role to play for transport practitioners and are relevant for particular tasks and at specific scales. In the planning of a new urban extension with a new public transport interchange, for example, multivariate analysis could be used to test layout options during the feasibility stage and to inform the Design and Access Statement. A pedestrian assignment model could be built alongside other traffic models during the design phase, and its outputs used to evaluate design options in a consistent fashion. Micro-simulation can subsequently be employed to test safety and comfort issues in the interchange and at a key junction in the network. 3. KEY CONCEPTS This paper presents some of the key concepts involved in very large scale pedestrian movement analysis. Based on recent examples of good practice some of the key issues are presented. 3.1.
Pedestrian route choice
Route choice is one of the processes that may be described by general choice theory. A route is described as a chain of consecutive nodes joined by links, connecting trip origin and destination (Bovy & Stern 1990). A wide variety of algorithms have been developed in transport modelling to represent the route choice decision-making processes for different transport modes (highways, public transport, cycling). Figure 2: Example of four vehicular routes between the same origin (top) and destination (bottom)
The fastest route (time)
The shortest route (length)
The route with fewest turns
The route with fewest intersections (via motorway and merging)
Adapted from Bovy and Stern (1990), Rogers and Langley (1998) and Piet and Stern (2009)
Although route selection strategies are largely subconscious (Hill, 1982), several researchers have formulated theories on this behaviour. Distance is not only an important factor on which route choice is based, it also influences the way pedestrians choose their routes (Ciolek, 1978, Guy, 1987, Helbing & Molnar, 1997, Lausto & Murole, 1974, Seneviratne & Morrall, 1985, Verlander & Heydecker, 1997). Different types of distance are therefore distinguished in literature. Khisty (1999) distinguishes between ‘perceived’ distance and ‘cognitive’ distance which include the assessment of the geometric complexity of the routes. Directness is defined in relation with visibility in relation to inferred destination direction that is pedestrians walk straight towards a visible destination, unless they are hindered by obstacles, other pedestrians, or diverted by other attractions. When considering pedestrian route choice, it is then important to differentiate between the distance of a route and the complexity of that route. Route distance can be measured as the metric (Euclidean) between two points through a network. Previous research has shown that there are specific factors that may affect pedestrians route choice such as distance or time, the number of obstacles or interactions with other pedestrians along the route, the directness of the route (i.e. the number of directional changes), the level-ofservice provided by the roadway and traffic environment, the overall attractiveness of the environment, and so on (Hoogendoorn & Bovy, 2004). The measurement of route complexity seeks to reflect people’s observed preference for the simplest route. For example, the simplest route can be defined as the one that minimises the number of intersection points, the number of turns or the sum of the angular turns.
Figure 3: Comparing the metric and angular catchments of the two railway stations in Croydon From West Croydon Station
The maps show in olive green the metric distance catchment area within a 10 minute walk from West Croydon Station and East Croydon Station (shown as pink dots). 10 minutes’ walk 0
1,000
2,000
Metres
From East Croydon Station
The maps compare the metric catchments to the street network that is easily legible, the most direct from the same location (geometric or angular distance) 0 1 (>0° 90 ° 180 °