We are testing a new beta website for gov.scot go to new site

Costs of Congestion: Literature Based Review of Methodologies and Analytical Approaches



3.1 A review of literature has revealed a persistent view that there is no single definition of congestion, although the concept is commonly understood and the term is used widely by academics, policy makers and laypersons. There are a number of examples of substantial research projects considering aspects of congestion at national and European level which do not include a definition of congestion as part of the work. In fact, the definition of congestion will vary according to the context (urban, interurban) and can be both an objective state of the transport network and a subjective condition for the transport network user. As a result a number of definitions exist and these are outlined below. Following a summary of the three states of congestion, research on congestion from a transport users perspective is given, followed by rather more formalised definitions which lend themselves towards quantified measures.

Types of congestion

3.2 The three types of congestion are outlined by Brownfield et al (2003) as Recurrent congestion, Non-recurrent congestion and the Pre-congestion state, as shown in Table 3.1. These types are based upon the frequency and predictability of the congestion - factors which will impact on driver behaviour. The costs associated with each type of congestion are likely to be different. Non-recurrent congestion costs may be more difficult to quantify due to the inherent sparseness of adequate amounts of data needed - it may be argued that the costs could be higher as drivers have not been able to take the possibility of congestion into account in planning their journey or alternatively the costs may be less dramatic as drivers pre-developed strategies for coping with congestion will not have come into play. Some routes are increasingly subject to non-recurrent congestion however, for example with accident black spots. In these cases drivers may 'learn' an expected cost in terms of likely delay and successful contingency routes. The Pre-congestion state will carry some costs similar to those of congestion, including loss of control over drivers' environment, deterioration in the environment and other impacts.

Table 3.1 - Summary of types of congestion

Congestion Type


Recurrent congestion

Occurs at regular times at a site. It can be anticipated by road users that normally use the route during those times. Examples of recurrent congestion are morning or evening peak hour congestion, or congestion due to a regular events such as a street market on a particular day each week

Non-recurrent congestion

Occurs at non-regular times at a site. It is unexpected and unpredictable by the driver and is normally due to incidents such as accidents, vehicle breakdowns or other unforeseen loss of carriageway capacity

Pre-congestion (Borderline congestion)

Occurs where free-flow conditions breakdown but full congestion has not yet occurred. This may occur either side of the time period when congestion occurs or upstream or downstream of congestion that is already occurring.

Notes to table
Source: adapted from Brownfield, 2003

Perceived congestion

3.3 Perceived congestion is as important a concept as the formalised definitions. The need to introduce policy measures to improve congestion may be driven at least in part by the political considerations arising from a large number of transport systems users that believe congestion is a problem in the parts of the network they use. This may in turn relate to the historical state of the network, which is likely to influence users expectations of their journey and finally their perception of whether the current levels of efficiency reflect a congestion problem or not. Transport system users in a geographical area which has a history of slow, unreliable and delayed journeys may have a different perception of (and greater degree of tolerance towards) levels of congestion than those in areas with a recent history of relatively free flow conditions. A number of studies have therefore looked at perceived perception i.e. the state of the traffic system from the users' subjective interpretation. These studies have highlighted a number of ways of viewing what congestion is, which have relevance for measures aimed at putting users at the centre of transport policy.

3.4 A qualitative study carried out for the Department for Transport ( DfT, 2001) reported alternative definitions of congestions based on the perceptions of drivers. Group discussions took place with 83 drivers of cars and light commercial vehicles from six areas of England (covering a variety of locations and possible congestion difficulties). In addition to eliciting views on definitions of what congestion is, a range of other traffic problems were discussed and a small number of different indicators of congestion reviewed for public value and acceptability. In general, the concept of congestion was widely understood but there was considerable variation on how it may be specifically described. Three main themes arose:

  • 'stationary or near jam conditions'
  • 'loss of speed due to weight of traffic'
  • 'slow progress'

3.5 The latter is related to vehicle density, even though traffic may not be at a standstill. For example having to drive at 40 or even 50 mph on a crowded motorway is perceived by some drivers as congestion, even though traffic is still progressing at reasonable speed. The most favoured definition related to delays rather than density though, with the description of stationary or very slow moving (<5mph) traffic prevailing. Two formal definitions were presented as follows:

  • 'Traffic is congested if there are so many vehicles that each one travels slower than it would do if the other vehicles weren't there'
  • 'Traffic is congested if there are so many vehicles that they are brought to a standstill or can only crawl along'

3.6 Interestingly, these two definitions reflect the two fundamental approaches to interpreting congestion. The first is strongly related to the principles behind marginal costs of congestion ( discussed further in chapter 6) and the second reflects a 'traffic engineering' perspective which underlies many of the indicators and measures of congestion ( summarised in chapter 4).

3.7 Comparing definitions in the urban and interurban contexts, the ability of traffic to proceed through junctions is seen by some as the defining characteristic of congestion. Leonard (1993) defines congestion in urban areas as being:

'the condition when the free movement of traffic through junctions starts to break down',

proposing a five-point scale of congestion graded from free-flowing to gridlock conditions. In the motorway environment however, speed is more likely to be the factor defining congestion. Recent work on perceived congestion on motorways includes that of DfT (2005), where respondents were able to indicate their own definition of congestion. Just over half of all respondents said they thought congestion on a motorway was defined by a traffic jam with complete stops of 5 minutes, this being the dominant response. Less than half of respondents (45%) considered a motorway to be congested if they had to travel at less than 20 mph and 39% if they experienced stop/start traffic for more than 15 minutes. Less than 20% considered the motorway to be congested if they had to travel at around 50mph.

Formalised definitions of congestion

3.8 Formalised definitions of congestion begin to express congestion more rigorously and in terms which may be strongly related to indicators or form the basis for quantified measurement. The definition given by the Highways Agency ( DMRB, 1997) captures the wide understanding of congestion and relates it to characteristics of the network. This states that congestion is

'the situation when the hourly traffic demand exceeds the maximum sustainable hourly throughput of the link.'

3.9 At this point, traffic is likely to experience one or more of the following: flow breakdown with speeds varying considerably, average speeds drop significantly, the sustainable throughput is reduced and queues are likely to form. The definition forms the basis for the Congestion Reference Flow, which is a quantified measure of congestion and described in chapter 4 below. According to Goodwin 2004:

'Congestion is defined as the impedance vehicles impose on each other, due to the speed-flow relationship, in conditions where the use of a transport system approaches its capacity'.

3.10 This expresses congestion as a phenomenon which involves the interrelation of vehicles and the idea of impedance arising to others from an additional vehicle on the network. This particular definition dovetails with the economic approach to measuring marginal congestion costs described in chapter 6.

3.11 Research into the relationship between congestion and accident risk for the DfT (Brownfield et al, 2003) sought to define congestion in a way which was quantitative and easily measurable, considering the urban and interurban contexts separately. Twenty sites from across England were used as a basis to gather evidence, including four motorway sites, nine peri-urban sites and seven urban sites. For an interurban (or peri-urban) link, the following definition was derived:

'An interurban or peri-urban link is defined as being congested when the point average speed taken over 3 minutes is below 50% of the speed limit'

3.12 This was based on broad agreement with empirical evidence found from other studies, which indicated that highly congested roads with speed limits of between 30 and 40 mph have an average speed of approximately 20mph and observations of motorway speed flow curves that showed flow breakdown occurring between 30mph and 40 mph. For the urban links, the following definitions were applied:

'An urban link (with a signalised exit) is defined as congested when traffic cannot exit the link within one cycle. An urban link with an unsignalised exit is defined as congested when traffic cannot exit the link within a time equivalent to one signal cycle (the cycle time equivalent was calculated by estimating what the cycle time would be if the link exit was signalised.'

3.13 The two definitions are in line with the previous findings by DfT, 2001 and Leonard, 1993, with speed forming the criteria for interurban congestion and stops forming the criteria in the urban case. The supporting basis for these urban definitions included the fact that if traffic is consistently delayed by more than one cycle, the junction is likely to be close to saturation (and therefore congested), which implies a high volume/capacity ratio.

3.14 In summary, despite the past research into congestion and frequent use of the term, the state of congestion is often understood but not formally defined. Perceived congestion is an important factor alongside more objective definitions in driving the need for policy measures. Definitions vary according to two major dimensions - the traffic engineering perspective and the economic cost driven dimension which in fact relate to two major efficiency objectives i.e. system efficiency and economic efficiency. Users' perceptions were generally consistent with one or other of these dimensions. Congestion in urban areas can be distinguished from that in the interurban context as it can be recognised by the inability to exit a link within a traffic cycle. Congestion in an interurban context may be defined through speed of travel (or ultimately stopping). Both perceived and formalised concepts of congestion lend themselves to more objective measurement and indicators, which are described in chapter 4.