Navigation – Plan du site
2019
854

Intermodal mobility analysis with smart-card data. Spatio-temporal analysis of the bus-metro network of Rennes metropole

Cyprien Richer, Etienne Come, Mohamed Khalil El Mahrsi et Latifa Oukhellou
Traduction de Alvin Harberts
Cet article est une traduction de :
La mobilité intermodale par les données billettiques. Analyses spatio-temporelles du réseau bus-métro de Rennes Métropole

Résumé

This paper aims to analyse the intermodal practices of mobility in the bus-metro network of Rennes metropole. Intermodality being strongly linked to the use of urban public transport networks, the Rennes Métropole bus and metro network data provided by Keolis-Rennes provides a very significant part of daily intermodality. To compensate for the lack of information on destinations and correspondances in the ticketing data a reconstruction of trips was made on the basis of hypotheses provided by the literature. The research carried out in the "Mobilletic" project considerably deepened the understanding of the temporal and spatial dimensions of intermodal mobility within Rennes Métropole’s urban public transport network.
daily mobility, public transport, data science, data analysis, intermodal transportation

Haut de page

Texte intégral

The authors would like to thank Keolis Rennes for providing the ticketing data on the bus and metro network.

Introduction

1This article focuses on the analysis of intermodal practices on the scale of the Rennes metropole bus-metro network. The challenge is to deepen knowledge of intermodal mobility using ticketing data collected from user validations. Indeed, these practices remain difficult to measure despite the importance of the intermodality lever for the organisation of mobility systems.

2This contribution is part of the work of the Mobilletic project (2013-2016). The aim of this research is to enhance ticketing data through an approach that combines the human and social sciences with data sciences. The question of the valuation of big data for decision-making is indeed central. Based on digital traces collected throughout the public transport network, it is possible to develop analysis and visualization tools to better understand people's mobility.

3Even if the implementation of ticketing systems does not primarily meet the objective of studying urban mobility (but rather the problems of combating fraud), transport operators and mobility organising authorities have high hopes for the use of the data collected. Like any source of information, it has advantages (number of precise continuous records) and limits that must be overcome: in particular the fact that validation is only carried out at the entrance to the network and that it is therefore necessary to rebuild the validation chains to work on intermodality.

4The objective here is to study the potential of ticketing data to analyse intermodal mobility with two main questions: to what extent can ticketing data improve knowledge of intermodality at the scale of the urban network of the Rennes metropole? What depth of spatio-temporal analysis can be achieved with enriched ticketing data?

5The article will be divided into three parts. The first part reviews the state of knowledge on data and methods of analysis of intermodality. The second part deals more precisely with ticketing data, their potential and the processing of the data provided by Keolis-Rennes transport operator. Finally, the third part presents some results from the spatio-temporal analysis of Rennes ticketing data.

Data and methods for intermodality

Mobility and the challenges of intermodality

6The concept of mobility demonstrates the value of analysing transport demand through the prism of uses and lifestyles, in addition to the more technical approach to "transport" that was largely predominant in the 20th century. The mobility paradigm as presented by Amar (2006) is accompanied by a desire to finely understand the diversity of potential mobility between individuals (accessibility) and individual and collective travel choices. The stakes are high since mobility questions are regularly identified as issues at the heart of social, environmental, economic or territorial challenges.

7Faced with these challenges, public policies are challenging the hegemony of personal automobile use, particularly in urban areas. With the emergence of new deregulated offers, local authorities' actions are helping to build increasingly multimodal cities. They aim to promote sustainable mobility practices intented in particular to reduce the use of private cars and developithe use of public transport (CT) alone or in combination with other modes of mobility such as walking, cycling, car sharing or carpooling.

8Intermodality, which refers to the "possibility of switching from one mode of transport to another" during the same journey (Margail, 1996), may appear to be one of the levers of sustainable mobility policies. In theory, the construction of an intermodal mobility system can make it possible to meet environmental challenges (reducing car use) by guaranteeing a high efficiency of the transport system (from "mass transit" to "door-to-door"), within the limits of economic constraints (optimisation of the various modes of transport in their field of relevance rather than new offers or infrastructures). Intermodality is a necessary condition for the sustainability of the transport system (Szyliowicz, 2003). Like an ecological system, Replogle affirms that a transportation system that depends on only one or two modes is much more susceptible to inefficiency, disruption and failure than a system that combines many different modes of transportation (Replogle, 1991).

9However, for intermodal passenger transport to work well, it is not enough to build transit systems; it is also necessary to provide appropriate information structures that allow seamless travel (Szyliowicz, 2000). However, intermodality has to face a paradox: it is a force for the optimisation of transport networks, but it is often also a fragility for the mobility of people. Complementarity of transport offers makes it possible to link networks and strengthen their performance, as demonstrated by many accessibility indicators (for example, those based on graph theory: Dupuy, 1985; Stathopoulos, 1997; Chapelon, 2003; L'Hostis and Conesa, 2008). However, the "load breaks" or connections induced by these intermodal organizations often increase the hardship for the traveller (Wardman and Hine, 2000; Litman, 2008; Zhan Guo and Wilson, 2011). It is the paradox of intermodality that remains a major issue for large urban systems. A good knowledge of intermodality is necessary to reconcile the points of view of the organisation of transport networks and mobility practices. However, the quantification and evaluation of intermodality remains difficult in the "absence of an overall statistical description of these phenomena" (Margail, 2002).

Intermodal mobility, a use that is still unknown?

10Studies and research indicate that there is a lack of accurate data on travel patterns and travel in non-motorized modes, such as cycling and walking or on combinations of modes used in intermodal journeys (Mark A. Miller, Camille Tsao, 1999; Michael Duncan, David Cook, 2014; Gandit, 2007). Figures on intermodal practices are much more disparate and difficult to obtain in any country (Gandit, 2007). For example, the annual surveys of the population census in France are limited to the question of the "most commonly used mode" for "habitual journeys to work" while the traditional "modal shares" are based on a hierarchy of modes of transport that masks combined uses.

11A panorama of intermodal mobility is nevertheless possible with the data from Mobility-Surveys in France (Richer, Meissonnier, Rabaud, 2016). According to the traditional use of Mobility-Surveys, travel is intermodal if it consists of at least two journeys (walking excluded by convention). The trips must be linked for the same reason (e. g. work). If there is an intermediate reason (e. g. small purchases around the transfer station), it corresponds to several trips. Intermodality according to Mobility-Surveys does not take into account the temporal criterion: the connection can be immediate or very long, what counts is the sequence of journeys for the same activity.

12According to these definitions, intermodal mobility can be studied on a weekday at the level of an agglomeration. It’s been observed that intermodal travel can reach 10% of daily mobility and that the number of people making at least one intermodal trip during the day can reach 18% (this maximum rate outside the Ile-de-France region is reached in the Lyon conurbation - EMD 2016- figure). The analysis of Mobility-Surveys data opens up interesting analytical opportunities, particularly to compare territories, measure major changes or cross-reference mobility practices with household socio-economic variables. The use of these data also makes it possible to broaden the definition of intermodality by taking into account walking as an integral part of the travel chain (Rabaud, Richer, 2015).

Towards a new look at intermodality with ticketing data

13A majority of the work carried out so far is based mainly on the analysis of these Mobility-Surveys. These data present - like any data source - assets (collection of all modes, all motives and information on individuals and their households) and limitations (surveys carried out at best every 10 years, relatively expensive, with sampling rates of 2 to 3%). Other sources are now available (ticketing data, GSM, Wi-Fi or Bluetooth traces, geolocation of our publications on social networks generated during our travels...). These devices were not initially designed for mobility analysis but can be very rich in terms of traceability. These digital traces can be used to set up renewed and complementary observation and modelling approaches for urban mobility, in particular multimodal and/or intermodal mobility.

14Ticketing data in particular, open up an important field of analysis to improve knowledge of individual mobility in complementarity with other data sources. Compared to ticketing data, Mobility-Surveys collect a small number of records every ten years on all daily trips rich in socio-economic metadata. Conversely, ticketing data provide a large number of records per minute on a small part of daily mobility which is poor in metadata. Some experts present the exploitation of ticketing data as the transition to "HD film" (with the implication that Mobility-Surveys would be a "Black&White photo"). The image may seem overstated because it should be made clear that the "HD film" covers a small part of daily mobility (10 to 15% in large urban areas, 20% in Île-de-France since only validations are "captured" in all or part of the public transport network) while the B&W photo is representative of all daily mobility (including immobility, moreover). The first authors to explore these data as a new source of information for mobility analysis (Bagchi, White, 2004) had already identified several constraints such as the lack of validation upon arrival, the absence of the reason for travel and the sample of the population which was not necessarily representative.

15Few studies use them today, partly because of the relative novelty of these information sources, the lack of consolidated methods to exploit them jointly, and the low number of feedback from social science researchers (geography, sociology, economics, psychology) on possible treatments based on these new data. This is the challenge of the Mobilletic project (2013-2016), which aimed to enhance ticketing data through a multidisciplinary approach to improve knowledge of intermodal mobility.

The potential of ticketing data

Ticketing data and intermodality

16Research has been conducted on the use of ticketing data for the analysis of travel in public transportation or with shared mobility systems (Côme et al., 2014; Vogel et al., 2014). With regard to public transport, a first part of the work carried out in this field has focused on improving ticketing data, in particular the reconstruction of journeys (estimation of destinations and detection of connections) (Barry et al., 2002; Trépanier et al., 2007; Chu et al., 2008). Another part of the work focused on the exploitation of these data through data mining tools (Morency et al., 2006; Trépanier et al., 2012; Lathia et al., 2013; Ma et al., 2013; El Mahrsi et al., 2016). Among the objectives of the latter work, there was obtaining an improved and synthetic representation of the profiles of public transport users, the study of individual travel practices and also the analysis of congestion in a transport network.

17Ticketing data seem particularly appropriate for dealing with issues related to public transport networks. However, intermodality is a very common practice in public transport networks (Zhan Guo, Wilson, 2011). Intermodal transfers are "endemic" to public transport systems, particularly in large multimodal networks (Vuchic, 2006). In France, the modal share of public transport is the main determinant of intermodal mobility, while ¾ daily intermodal trips are the result of transfers within urban public transport networks (Richer, Rabaud, Lannoy, 2015). This is why ticketing systems, which collect almost all trips made on the public transport network, provide a very significant part of daily intermodality.

18The window for observing movements is important, the acquisition frequency is almost continuous and the information collected is distributed along the network's linear features. These data have interesting intrinsic advantages such as exhaustiveness (or near exhaustiveness), spatial and temporal fineness, and a very low response bias compared to those found in the survey data. However, there are several difficulties in exploiting these sources of information (El Mahrsi et al., 2015):

  • missing data: for systems such as the metro or bus, only the original trip data are available (the user validates his ticket only when boarding).

  • the large volume of this data (approximately 250,000 validations per working day in Rennes) is a source of wealth, but it raises the problem of its storage and processing.

19These anonymized digital traces provide a flow of longitudinal data. However, socio-economic data relating to the user are not available, even if it is possible to have information on the user's profile when the type of subscription (full fare, student, social fare, etc.) or transport ticket is known.

Keolis-Rennes network data sets

20Rennes metropole is the organising authority for the mobility of a public transport network called STAR Bus-metro. It covers 38 municipalities and serves more than 400,000 inhabitants with more than 70 regular bus lines and a metro line. The operator of the Keolis urban transport network produces and operates the STAR Bus-metro data, owned by Rennes metropole. As part of the Mobilletic project, Keolis-Rennes has made available anonymised ticketing data (user identifiers are periodically anonymised in order to make it impossible to cross-reference with other information). Over one month of data (April 2014), it contains information on 5,404,096 validations, 80% of which were carried out by approximately 135,000 smart cards (called KorriGo cards). Half of the validations of the entire network are carried out on the metro line. When you take into account the eight major bus lines, 85% of Star network trips are covered.

Figure 1: Location of the "Rennes metropole" conurbation

Figure 1: Location of the "Rennes metropole" conurbation

Source : Authors, background map Geoportail

21In the Rennaise conurbation, ticketing data provide about 12% of daily mobility, or 170,000 trips/day made by bus and/or metro out of 1,482,000 trips/day (according to the 2007 "Certu standard" Mobility-Surveys for Rennes metropole). Out of 1,482,000 trips, 60,000 are intermodal, which is slightly more than 4%. Excluding trips made only on foot, intermodality accounts for 6% of mechanized trips. This corresponds to 12% of people making at least one intermodal trip during the day (i.e. 88% not making an intermodal trip).

22Of the 60,000 intermodal trips, 45,700 involve at least one bus+metro (16,800), metro+bus (17,400) or bus+bus (15,700) connection (double counting trips with several connections). The intermodality informed by the validations on the STAR network reaches - in theory - (because in practice not all validations are usable) more than 75% of the daily intermodality on the scale of the Rennes metropole. In addition to Mobility-Surveys, ticketing data thus provide a large amount of information to improve knowledge of a large proportion of intermodal mobility at the local level.

23The ticketing data used in the Mobilletic project "captures" approximately 224,000 validations per day, or 177,000 trips per day. The STAR bus-metro network claims a peak of 280,000 trips/day on its bus-metro network, which is equivalent to about 220,000 trips/day today. The ticketing data enriched in the Mobilletic project correspond to approximately 80% of daily trips on the STAR bus-metro network. About 75% of validations are carried out with the KorriGo card and therefore 25% with paper tickets, while fraud is estimated at 10% on the Rennes network. The share of intermodal travel (intrinsic to the bus-metro network) is very similar depending on the sources and methods (between 24 and 27%).

Figure 2: Comparison of intermodal mobility by data source

Rennes metropolitan area (STAR bus-metro network)

Sources

STAR bus/metro network mobility

Bus/metro intermodality

Share of intermodal trips in total bus/metro network trips

Mobility survey

2007

215,700 trips stage/day

170,400 trips/day

45,700 intermodal trips /

49,900 correspondences

26,8 %

Operator data

Statistical annual report 2011

71 million trips/year

Connections rate

1,240

24,0 %

Ticketing Data

2014 Ticketing Data (enriched in the Mobilletic project)

224,000 validations/day

177,000 trips /day

43,000 intermodal trips /

47,000 correspondences

24,3 %

Table reading note - Clarification on the definition of concepts (travel according to EMD, travel according to Ticketing, travel, validation, travel):
A "trip" according to the Mobility-Surveys is equivalent to a movement in space for a purpose (e. g. work, study, purchase, return home...). It may consist of one or more "trips stage" corresponding to the use of one or more modes of transport to make this journey.
Since there is no information on the reasons, a "trip" obtained through the ticketing data is similar to an origin-destination movement obtained following the reconstitution of the validation chains. Validations" are here the equivalent of the notion of "trips stage" or "voyage" for operators, since they correspond to an elementary segment of the journey, which may be a journey by metro or on a bus line.
An intermodal trip according to the usual measure of Mobility-Surveys is therefore composed of several mechanized trips; an intermodal trip according to the Mobilletic project method is therefore composed of several validations (or trips stage) linked according to a time criterion and a spatial criterion. In both cases, the linked bus+bus trips stage or validations count as an intermodal trip.

Source : Authors, after various data

Data enrichment and validation chain reconstruction

24The raw data set of the Keolis STAR bus-metro network in the Rennes metropole contains the following information:

  • anonymous card number (for validations made with a card),

  • date and time of the validation,

  • validation station,

  • type of ticket

  • direction of travel (for buses).

25As in most French urban transport networks, no information on travel destinations is available in this raw data set since the user does not have to validate when leaving the network. The dataset does not specify whether the user is in correspondence.

26To compensate for the lack of information on destinations and to detect matches, hypotheses and enrichment methods validated in the literature were used. Barry et al (2002) developed a methodology that estimates Origin-Destination (OD) matrices for all metro stations in New York City based on two assumptions: first, a large proportion of users return to the destination station of their previous trip to make the next one and second, they finish their last trip of the day at the station where they started their first trip of the day. In another article, Chu et al (2008) pay particular attention to the detection of correspondences over time and distance between two validations, which may lead to the conclusion that they are the same trip.

27The potential destinations of the registered validations were therefore estimated by searching for the station on the current line closest to the user's next departure station. If this station is located more than 500 meters from the next validation station, we have not assigned a destination. Corresponding validations were detected as follows: when the origin and destination of a validation are estimated, the time of correspondence with the next validation is calculated by subtracting from the start time of the next validation the estimated arrival time of the current validation. This arrival time is simply estimated by adding to the validation date the theoretical travel time as defined by the time tables available in Rennes in GTFS (General Transit Feed Specification) format. When this time is less than 30 minutes, we consider that the two validations form a single trip because they are linked to the same reason of travel. In this case, the two validations are grouped together in the same trip.

28The reconstruction of the validation chains according to the Mobilletic project enrichment method was therefore based on a time criterion (validation within 30 minutes after the previous destination) and a spatial criterion (correspondence within a radius of less than 500 m). This measure of intermodality therefore differs slightly from other definitions of intermodality: with Mobility-Surveys, the focus is on the sequence of trips for the same reason, while network operating rules refer to correspondence if the second validation takes place less than an hour after the first. Let's take a user who takes a train at 8:20 am, arrives at a transfer station at 8:35 am, takes advantage of his stop to do some shopping in the neighbourhood, and takes the bus back at 9:10 am. He will be able to validate in correspondence if the rules of the network allow it, but this sequence of trips will not be considered as an intermodal trip with traditional Mobility-Surveys analyses (because he stops to do some shopping in between) and will not be counted in correspondence in the enriched data of the Mobilletic project for a different reason (the connection time exceeds 30 minutes).

Figure 3: Available ticketing data in relation to the number of trips made by the network

Rennes metropolitan area (STAR bus-metro network)

Travel / day

Estimated trips per weekday on the network (source Keolis Rennes, 2014)

280 000

Of which about 10% fraud (trips not validated)

28 000

Of which validations with paper ticket

62 000 (25% travel excluding fraud)

Of which Korrigo card validation

190 000 (75% travel excluding fraud)

Mobilletic Project

Data available for the Mobilletic project

224 000 (80% of network travel)

Data after enrichment

160 000 (72% of the data available)

Source: Authors, based on Keolis Rennes ticketing data

Spatial and temporal analysis of bus-metro intermodality in the Rennes Métropole

The weekly and daily rhythm of intermodality

29The number of validations per hour in the STAR network over a typical week shows a cyclical behaviour of the network strongly related to the time and day of the week. There is a classic weekday profile with the traditional morning and evening peaks that correspond to the busiest times of public transportation. The number of validations per hour during these peak periods (between 15,000 and 20,000 validations) is at least twice as high as during off-peak hours (around 7,500). The lunchtime peak (around 10,000) is a little lower except on Wednesdays. This day is distinguished by three peaks of equal importance (around 15000). Weekend validation behaviours are different: the Saturday validation peak reaches the weekday off-peak validation level (around 7500) while Sunday reflects diffuse use with a low significance afternoon peak (around 2500).

Figure 4: Number of validations per hour in the STAR transport network in Rennes (week of the 7th to 13th April 2014)

Figure 4: Number of validations per hour in the STAR transport network in Rennes (week of the 7th to 13th April 2014)

Source: Authors, based on Keolis Rennes ticketing data

30As far as connections are concerned, their share on the Rennes bus-metro network varies slightly between 20 and 25% from one week to the next. In a given week, the median share of intermodality on a weekday is about 23% of total network trips. On weekends, the share of intermodal travel is lower, although it is far from negligible: the median is around 19% on Saturdays and 17% on Sundays. On a daily basis, the gross variation in the number of intermodal trips follows the same trend as the overall rate of validations.

31The share of intermodal trips during the day shows differentiated trends: there is a very high proportion of intermodality in the morning corresponding to longer trips that start earlier (probably from peripheral municipalities). Intermodal journeys, which are always significantly longer than those without connections, reach an average distance of 8 km during the morning rush hour.

32At 7am, on weekdays, 40% of trips are made with a connection. The rest of the day, the curve is uneven but remains in a fairly stable range between 20 and 25% intermodality. Another feature that can be seen every day of the week is the slight trough at noon in terms of the share of connecting trips. We find the same dropout between 12pm and 2pm of about 5 points. It can be assumed that users who use public transport during the meridian break are those who do not have to make a connection: either they can go directly home or they are looking for places (for example, restaurants) that are accessible without breaking the load.

33These results attest to an intermodality that is more suffered than chosen, because it concerns the longest journeys that start earlier for constrained reasons. Conversely, when a choice is available to the user, at lunchtime during the week or during weekends, intermodality is much less frequent.

Figure 5: Number of validations with and without correspondence on a Friday

Figure 5: Number of validations with and without correspondence on a Friday

Source: Authors, based on Keolis Rennes ticketing data

Figure 6: Share of validations in correspondence according to different weekdays.

Figure 6: Share of validations in correspondence according to different weekdays.

Source: Authors, based on Keolis Rennes ticketing data

Figure 7: Average distances of trips (with or without correspondence) by time of day

Figure 7: Average distances of trips (with or without correspondence) by time of day

Source: Authors, based on Keolis Rennes ticketing data

The orientation of bus - metro connections

34During the week, the most frequent travel chains correspond to the shift from bus mode to metro (37% of connections). The other metro-bus and bus-bus combinations are slightly less frequent (31.5%). This difference is even more pronounced in the morning during the week, since 41% of intermodality is in the bus-to-metro direction, while the three types of combination are balanced during the evening rush hour. There is therefore no absolute symmetry in the mobility chains. This is due to the ease of connecting to the metro, which is almost immediate due to the high frequency of the metro line, while connecting the other way can be longer. The median connection time from the bus to the metro is between 2 to 3 minutes, while it reaches 12 to 13 minutes the other way around. A user can thus chain bus and metro in the morning but not in the evening, preferring to walk or ride a bike at the exit of the metro rather than wait for a bus.

Figure 8: Distribution of connection times in the bus-to-metro direction

Figure 8: Distribution of connection times in the bus-to-metro direction

Source: Authors, based on Keolis Rennes ticketing data

Figure 9: Distribution of connection times in the metro-to-bus direction

Figure 9: Distribution of connection times in the metro-to-bus direction

Source: Authors, based on Keolis Rennes ticketing data

35In the bus-metro centres of the Rennes metropolitan area network, the bus station always has a higher connection rate than the eponymous metro station. Regardless of the station pair and time, the number of validations in the bus is increasingly dependent on connecting users. For example, at the Henri Fréville station, depending on the time of day, 55 to 65% of the validations in the bus station come from connecting users, compared to 23 to 32% for the metro station.

36However, these intermodality rates conceal other realities, as we can observe at the level of the Villejean-University bus-metro interchange centre: if, in relative terms, bus use is more dependent on connecting users (mainly leaving the metro), in absolute terms, at the morning rush hour, the bus provides three times as many users to the metro (221 metro users come from the Villejean-University bus station) as the opposite (77 bus users come from the Villejean-University metro station). The situation is reversed during the evening rush hour but in different proportions.

Figure 10: Orientation of connections by time of day at the Villejean-University metro-bus interchange centre

Morning rush hour 6am-10am

Villejean-University bus station

Villejean-University metro station

Validation/hour

153

530

Correspondence/hour

96

221

Share of connections

63 %

42 %

Source of correspondence

77 come from the metro station and 19 from the bus station

221 from the bus station

Evening rush hour 4pm-8pm

Villejean-University bus station

Villejean-University metro station

Validation/hour

407

846

Correspondence/hour

226

134

Share of connections

55 %

16 %

Source of correspondence

207 come from the metro station and 19 from the bus station

134 come from the bus station

37The metro being a very attractive mode in the Rennes transport network, the use of the bus is very much linked to the possibility of taking the metro on arrival. Buses pour into the metro in the morning, although, given the large number of users who come on foot (and to a lesser extent by bicycle or car), the proportion of metro users who come from the bus is lower than the opposite. The design of the lines involves a very intermodal operation of the bus, very dependent on the possibility of taking the metro on arrival or completing the journey after taking the metro. In other words, the metro is self-sufficient for a majority of users, whereas the bus alone would have limited relevance.

Intermodal catchment areas of metro stations

38Spatially, the places where bus-metro connections are made are concentrated on a few intermodal points. A small number of stations in the metro-bus network concentrate intermodal practices. Thus, 4% of the stations concentrate 91% of the connections made on the network and 46% of the validations. During the morning rush hour on weekdays, out of 676 bus/metro network stops, 587 have a value of less than 1 transfer / hour.

39Nearly 80% of daily intermodality is organised around, from or to the metro line. The main intermodal hubs are bus stops connected to a metro station. The functioning of Rennes intermodality is based on two main types of transfer points:

  • feeder exchange centres at the entrance to the city and at the ends of the metro (in the North-West, Villejean-University and in the South/South-East: Henri Fréville, Triangle and La Poterie) which combine both metro and bus, generally associated with a car park and relay;

  • a central exchange district, mainly around the République metro, which alone receives half of the connections of the Rennes metropolitan area's bus-metro network. During the busiest periods, there is almost one bus-metro connection every two seconds in the République interchange centre. This exchange district can be extended to Sainte-Anne, which offers a fairly fragmented intermodal operation.

40Other metro stops have either a local service character (Charles de Gaulle, Clemenceau, Italy, Le Blosne) or intermodal potential but with other modes not included in the available ticketing data (P+R to JF Kennedy; TER to Pontchaillou). It should be noted that the train station in Rennes has a very limited intermodal role due to the fact that it does not take into account intermodality with the other networks that connect to it (Ille-et-Vilaine buses, TER and TGV).

Figure 11: Intermodal catchment area of metro stations

Figure 11: Intermodal catchment area of metro stations

Figure 12: Characteristics of the main intermodal catchment areas

Intermodal catchment areas (average data for weekdays 6-10am)

Average volume of intermodal travel (per hour)

Average distance of bus journeys from intermodal journeys (km)

Number of municipalities concerned (excluding Rennes)

  

République

390,2

4,1

12

  

Villejean-Université

221,7

5,5

6

  

Henri Fréville

141,4

7,6

9

  

Sainte-Anne

130,8

2,9

4

  

La Poterie

109,3

5,8

5

  

Gares

55,5

2,1

1

  

Triangle

52,5

6,1

3

41To visualize the spatial configuration of Rennes' intermodality, we propose to observe the intermodal catchment areas of the metro line stations. These are shown in Figure 7 and are constructed by collecting, for each station, the destinations of the trips connecting from the metro. The estimated drawdown flows (and averaged over the entire day) are represented by solid lines between the drawdown metro station and the destination station of the trip.

42The most intermodal stations are generally those with the largest intermodal catchment area. Due to its central location and its very important bus offer, the République metro station attracts bus users from a large part of the urban area: in the East (and North-East) to Acigné and Thorigné and in the West (and South-West) from Bruz to Cintré. From east to west, the origin of those who make a connection to the République metro station goes as far as the edge of the Rennes metropolitan area.

43The main bus-metro connection stations each have a large catchment area up to the edge of the transport perimeter: Henri Fréville, by attracting the South sector from Bruz to Bourgbarré, has the highest average number of bus journeys in drawdown (7.6 km). Villejean-University attracts the whole North-West sector (even beyond Gévezé); La Poterie the South-East sector from Corps-Nuds to Nouvoitou. These two major intermodal stations have an average of 5.5 and 5.8 km of travel. The distance is significantly less than Henri Fréville because Villejean-Université and La Poterie also have a local attractiveness in Rennes districts, which reduces the average distance.

44The Sainte-Anne metro station has one of the narrowest catchment areas (average distance travelled: 2.9 km). Apart from Betton, its influence extends mainly within the city of Rennes (Ballangerais, Maurepas, Longs-Champs districts). In this sector, also in the form of a feeder line at the République station, the construction of the second metro line should profoundly transform the catchment areas of the République and Sainte-Anne metro stations.

45It should be noted that there are strict cuts in the drawdown basin and other more vague separations: for example, part of the commune of Bruz falls back to the République station, and the other part to Henri-Fréville. Finally, we can also note the overlapping catchment areas: the attractiveness of the Kennedy metro station covers part of the municipalities in the Villejean area. The same goes for the catchment area of the Triangle metro station, which doubles that of Henri-Fréville, which is much wider.

The morphology of places of exchange defined by practices

46Ticketing data provides the ability to change the zoom level for a "close" analysis of more localized phenomena. This is the case, for example, of the functioning of transfer points, which cannot be described in detail in a Mobility-Survey because they are not statistically representative. This allows us to observe more closely the characteristics and orientation of bus-metro transfers at the main transfer stations. Since data enrichment identifies the origins-destinations of each trip, it is possible to have a precise idea of the user's intermodal path between the station where he gets off and the transfer station. This negative mapping opens up prospects for analysing the morphology of the transfer points through the intermodal practices that take place there.

47The intermodal practices described with the ticketing data demonstrate the solidarity of uses between the stations in the network, whether or not they are grouped together in the same exchange centre. The traces of the validations in correspondence draw a particular morphology of the different intermodal places of public transport in Rennes metropole. These reticular forms differ according to two main criteria:

  • the level of compactness of transfers, illustrated by the length of the links: the shorter the links, the more direct the transfers are and the more compact the transfer point can be considered. We empirically established a distinction around a distance of 100 meters. However, it should be noted that this estimate remains partial because the physical distance between two stops does not make it possible to fully determine the "difficulty" of the connection, which also depends on the quality of the routes, the signage, the presence of stairs.

  • the number of interacting stops that make it possible to distinguish between "unipolar" intermodal locations such as bus-poles, "bipolar" with a pair of interdependent stations (often bus-metro), or "multipolar" with several stops composing the intermodal zone. The proportion of correspondences at the level of each link makes it possible to precisely measure the shape of the exchange pole. If a single link between two stops captures more than 90% of the connections, this pole is considered bipolar.

48We have identified five contrasting situations (a sixth "exploded-unipolar" could be added if we consider the vertical transfer distances in multi-level poles of exchange). However, this classification is more of a continuum where each intermodal station could be positioned according to the number of connected stops[x] and the average distances of the links between connected stops in the[y] pole. For explanatory purposes, it seemed preferable to us here to present the results in the form of a typology.

Figure 13: Distribution of metro stations according to the morphology of intermodal locations

Figure 13: Distribution of metro stations according to the morphology of intermodal locations

49The exchange hubs of the Rennes conurbation are mainly organised around a very interdependent pair of bus-metro stops. In most situations, more than 90% of bus transfers come from the same name metro station and vice versa. They are therefore hubs of exchange where bus stops are grouped in a single node near the metro station, more or less directly. It can be a simple stop or a bus station with several platforms (e. g. Villejean, Fréville, La Potterie) directly connected, which refers to the "compact bipolar" type.

50The "exploded bipolar" type characterizes longer interdistances with connections that take place in public spaces. The Anatole France hub is quite representative of the lack of immediacy in bus-metro connections. Even if the distance is not insurmountable, the lack of intervisibility between buses and the metro can increase the cognitive load of the connection and require guidance by appropriate signage.

51The intermodal sites of Sainte-Anne or Charles-de-Gaulle illustrate situations that can be compared to the idea of an "exchange district" (Menerault, 2006). Intermodality is shared between several bus stations in a fragmented way. The Charles-de-Gaulle metro station illustrates an atypical situation where connections are made in equal parts from the Charles-de-Gaulle and Plélo-Colombier bus stations (approximately 35 to 40% of the origin of the connections), while the latter station does not physically serve the Charles-de-Gaulle metro. This intermodality saves users from having to go to the République pole to take the metro.

52The main, relatively compact République exchange node is dominated by a privileged relationship: more than a third of the connections of this hub take place specifically between the "République" Bus stop and the metro station of the same name. Nevertheless, the République exchange node is largely multipolar: four bus stops (Nemours, Jaurès, Pré Botté, and République) to which we can add "Les Halles" compose with the metro a very dense exchange node.

53The distribution of stations in this typology through the analysis of large amounts of data provides accuracy and processing capacity for a large number of exchange nodes of various configurations. The qualitative knowledge of the network by the mobility authority or operator is thus enriched: the quantifiable differences between the empirical knowledge and the result of these measures are valuable in an operational logic. This work can encourage the authorities to question in particular the layout of the connecting points (for example, bringing remote stops closer together or improving routes, etc.).

Conclusion

54This article illustrates the contribution of ticketing data to the analysis of intermodal mobility. The series of analyses proposed attests to the possible spatial and temporal enrichment of the knowledge of intermodality. Ticketing makes it possible to broaden the scope of traditional analysis of this type of mobility by the precision and quantity of data offered.

55The study of bus-metro intermodality in the Rennes urban area offers an overview of possible research and opens up many perspectives for spatio-temporal analysis. We have demonstrated the possibility of studying, in a general way, the overall functioning of the bus-metro network with the shopping areas that link bus lines with a metro station, and in a more precise way, the flows within the exchange centres that intermodal behaviour shapes. With ticketing data, these two spatial analysis scales are systematically combined with temporal analysis possibilities that we have outlined in a simple manner. It is indeed possible to follow the evolution of the catchment areas during the day just as it is possible to have a dynamic observation of the rhythms of an exchange node almost by the minute.

56However, it should not be assumed that ticketing data are sufficient for all intermodal analyses and that they do not present any constraints. It should be remembered that no single data source can answer all the questions. For issues centred on the organisation of a public transport network and the regulation of its intermodal locations, ticketing data are high-quality resources. It should be noted, however, that the value of this data will depend on the often tedious pre-treatment and enrichment work. The sample of analyses proposed in this report is not "turnkey" and it is often difficult for network operators and their organising authorities to achieve these analysis possibilities.

57In addition, the proposed analyses provide insight into part of intermodality. Admittedly, connections between urban public transport modes dominate in all analyses of intermodality in daily mobility. This is the largest volume but not the most problematic from the user's point of view. Connections between buses and the metro are facilitated by the simple fact that they are managed by a single operator (Keolis Rennes in our case): it is therefore the same fare, the same information, and the complementarity is organised by a single authority (Rennes metropole in this case). The challenge of intermodal coordination is more acute in the case of relations with other operators: with the train (TER, TGV), with the intercity bus network or even with new mobility offers (private coaches or car sharing for example). One of the prospects for analysis via ticketing data would be to integrate data from park and ride and self-service bicycles (same players as buses and metros) and to complement it with departmental and regional public transport offers. The common support of the Breton "Korrigo" travel card could thus be combined with the prospect of joint analysis of mobility data.

Haut de page

Bibliographie

Amar G., 2010, Homo mobilis. Le nouvel âge de la mobilité, éloge de la reliance, Limoges, FYPéditions.

Bagchi M., White P. R., 2004, “What role for smart card data from bus systems”, Proceedings of the Institution of Civil Engineers, Municipal Engineer, No.157, 39-46.

Barry J.J., Newhouser R., Rhabee A., Sayeda S., 2002, “Origin and destination estimation in New York City with automated fare system data”, Transportation Research Record, No.1817, 183-187.

Chapelon L., 2003, “Evaluation des chaînes intermodales de transport : l’agrégation des mesures dans l’espace et le temps”, Actes du colloque Technological innovation for Land transportation, TILT, Lille.

Chu K., Chapleau R., 2008, “Enriching archived smart card transaction data for transit demand modeling”, Transportation Research Record: Journal of the Transportation Research Board, No.2063, 63-72.

Côme E., Oukhellou L., 2014, “Model-based count series clustering for bike sharing system usage mining: A case study with the vélib system of Paris”, ACM Transactions on Intelligent Systems and Technology, Vol.5, No.3, Article n°39, septembre 2014. URL : https://dl.acm.org/citation.cfm?doid=2648782.2560188; doi:10.1145/2560188

Duncan M., Cook D., 2014, “Is the provision of park-and-ride facilities at light rail stations an effective approach to reducing vehicle kilometers traveled in a US context?”, Transportation Research Part A, Vol.66, 65–74

Dupuy G., 1985, Systèmes, réseaux et territoires. Principe de réseautique territoriale, Paris, Presse de l’ENPC.

El Mahrsi M.-K., Briand A.-S., Côme E., Oukhellou L., 2015, “Utilité des données billettiques pour l’analyse des mobilités urbaines: le cas rennais”, in: Mattei M.-F., Pumain D. (dir.), Données urbaines, Economica Anthropos, coll. Villes.

El Mahrsi M.-K., Côme E., Oukhellou L., Verleysen M., 2016, “Clustering Smart Card Data for Urban Mobility Analysis”, IEEE Transactions on Intelligent Transportation Systems, No.99, 1-17.

El Mahrsi M.-K., Côme E., Baro J., Oukhellou L., 2014, “Understanding Passenger Patterns in Public Transit Through Smart Card and Socioeconomic Data”, In Proceedings of 3rd International Workshop on Urban Computing (SigKDD), New-York, 24 août 2014.

Gandit M., 2007, Déterminants psychosociaux du changement de comportement dans le choix du mode de transport. Le cas de l’intermodalité, Thèse de doctorat, Psychologie Sociale Expérimentale, Grenoble, Université Pierre Mendès France.

Lathia N., Smith C., Froehlich J., Capra L., 2013, “Individuals among commuters : Building personalised transport information services from fare collection systems”, Pervasive and Mobile Computing, Special issue on Pervasive Urban Applications, No.9-5, 643-664.

Litman T., 2008, “Valuing Transit Service Quality Improvements”, Journal of Public Transportation, No.11-2, 43-63.

L’Hostis A., Conesa A., 2008, “Définir l’accessibilité intermodale”, in : Banos A., Thévenin Th. (dir), Systèmes de Transport Urbain, Paris, Hermès-Lavoisier.

Margail F., 1996, “De la correspondance à l’interopérabilité : les mots de l’interconnexion”, Flux, No.25, 28-35.

Margail F. (dir), 2002, Intermodalité et interfaces : Comprendre les usages pour guider les décisions, Rapport contrat DRAST N° 98MT28.

Menerault Ph. (dir), 2006, Les pôles d’échanges en France. État des connaissances, enjeux et outils d’analyse, coll. dossiers du CERTU n°172, Lyon, Certu.

Miller M., Tsao C., 1999, Assessing Opportunities for Intelligent Transportation Systems in California’s Passenger. Intermodal Operations and Services, Review of the Literature, Interim Report for MOU 375.

Morency C., Trépanier M., Agard B., 2006, “Analysing the variability of transit users behaviour with smart card data”, The Ninth International IEEE Conference on Intelligent Transportation Systems, Toronto, Canada, September 2006.

Rabaud M., Richer C., 2015, “Walking : the missing link of intermodality ?”, International conference MUP-UPM “The Intricacy of Walking and the City: Methods and Experiments”, Paris-Marne la Vallée, 21-23 January 2015.

Replogle M.A., 1991, “Sustainable transportation strategies for Third World development”, Transportation Research Record, No.1294, Washington DC, National Research Council.

Richer C., Rabaud M., Lannoy A., 2015, “L’intermodalité au quotidien. Un panorama de la mobilité intermodale en France”, in : Armoogum J., Guilloux T., Richer C. (dir), Mobilité en transitions. De la connaissance à l’aide à la décision, Lyon, Cerema, Coll. Rapport d’études et de recherches, 123-134.

Richer C., Meissonnier J., Rabaud M., 2016, “Quelles intermodalités dans les mobilités quotidiennes ?”, in : Chapelon L. (dir) Transport et intermodalité, Londres, ISTE editions, Coll. Sciences, société et nouvelles technologies, 261-288.

Stathopoulos N., 1994, “Effets de réseau et déséquilibres territoriaux dans la structure de l’offre ferroviaire à Paris”, Flux, No.18, 17-32.

Szyliowicz J., 2003, “Prise de décisions, transport intermodal et mobilité durable : vers un nouveau paradigme”, Revue internationale des sciences sociales, Vol.2, No.176, 207-220.

Trépanier M., Tranchant N., Chapleau R., 2007, “Individual trip destination estimation in a transit smart card automated fare collection system”, Intelligent Transportation Systems, No.11, 1-14.

Trépanier M., Habib K.-M., Morency C., 2012, “Are transit users loyal ? Revelations from a hazard model based on smart card data”, Canadian Journal of Civil Engineering, No.39-6, 610-618.

Vogel M., Hamon R., Lozenguez G., Merchez L., Abry P., Barnier J., Borgnat P., Flandrin P., Mallon I., Robardet C., 2014, “From bicycle sharing system movements to users: a typology of Vélov cyclists in Lyon based on large-scale behavioural dataset”, Journal of Transport Geography, No.41, 280-291.

Vuchic V.R., 2006, Urban Transit Systems and Technology, Hoboken NJ, John Wiley & Sons.

Wardman M, Hine J., 2000, Costs of Interchange: A Review of the Literature, Leeds, University of Leeds, Institute of Transport Studies.

Yeh Chao-Fu, 2009, Intermodalité et coûts des déplacements urbains dans les mégapoles - Les cas de Paris, Shanghai et Taipei, Thèse en Urbanisme, Université Paris-Est, 506 p.

Guo Z., Wilson N., 2011, “Assessing the cost of transfer inconvenience in public transport systems: A case study of the London Underground”, Transportation Research Part A, No.45, 91-104.

Haut de page

Table des illustrations

Titre Figure 1: Location of the "Rennes metropole" conurbation
Crédits Source : Authors, background map Geoportail
URL http://journals.openedition.org/cybergeo/docannexe/image/33189/img-1.png
Fichier image/png, 638k
Titre Figure 4: Number of validations per hour in the STAR transport network in Rennes (week of the 7th to 13th April 2014)
Crédits Source: Authors, based on Keolis Rennes ticketing data
URL http://journals.openedition.org/cybergeo/docannexe/image/33189/img-2.png
Fichier image/png, 7,7k
Titre Figure 5: Number of validations with and without correspondence on a Friday
Crédits Source: Authors, based on Keolis Rennes ticketing data
URL http://journals.openedition.org/cybergeo/docannexe/image/33189/img-3.png
Fichier image/png, 112k
Titre Figure 6: Share of validations in correspondence according to different weekdays.
Crédits Source: Authors, based on Keolis Rennes ticketing data
URL http://journals.openedition.org/cybergeo/docannexe/image/33189/img-4.png
Fichier image/png, 166k
Titre Figure 7: Average distances of trips (with or without correspondence) by time of day
Crédits Source: Authors, based on Keolis Rennes ticketing data
URL http://journals.openedition.org/cybergeo/docannexe/image/33189/img-5.png
Fichier image/png, 96k
Titre Figure 8: Distribution of connection times in the bus-to-metro direction
Crédits Source: Authors, based on Keolis Rennes ticketing data
URL http://journals.openedition.org/cybergeo/docannexe/image/33189/img-6.png
Fichier image/png, 27k
Titre Figure 9: Distribution of connection times in the metro-to-bus direction
Crédits Source: Authors, based on Keolis Rennes ticketing data
URL http://journals.openedition.org/cybergeo/docannexe/image/33189/img-7.png
Fichier image/png, 25k
Titre Figure 11: Intermodal catchment area of metro stations
URL http://journals.openedition.org/cybergeo/docannexe/image/33189/img-8.png
Fichier image/png, 492k
Titre Figure 13: Distribution of metro stations according to the morphology of intermodal locations
URL http://journals.openedition.org/cybergeo/docannexe/image/33189/img-9.png
Fichier image/png, 175k
Haut de page

Pour citer cet article

Référence électronique

Cyprien Richer, Etienne Come, Mohamed Khalil El Mahrsi et Latifa Oukhellou, « Intermodal mobility analysis with smart-card data. Spatio-temporal analysis of the bus-metro network of Rennes metropole », Cybergeo : European Journal of Geography [En ligne], Systèmes, Modélisation, Géostatistiques, document 854, mis en ligne le 25 octobre 2019, consulté le 18 novembre 2019. URL : http://journals.openedition.org/cybergeo/33189 ; DOI : 10.4000/cybergeo.33189

Haut de page

Auteurs

Cyprien Richer

Cerema, Direction Nord-Picardie, Lille, France ; Chargé de recherche ; cyprien.richer@cerema.fr

Etienne Come

Ifsttar, Grettia, Marne-la-Vallée, France ; Chargé de recherche ; etienne.come@ifsttar.fr

Mohamed Khalil El Mahrsi

Ifsttar, Grettia, Marne-la-Vallée, France ; Chargé de recherche contractuel ; mohamed-khalil.el-mahrsi@ifsttar.fr

Latifa Oukhellou

Ifsttar, Grettia, Marne-la-Vallée, France ; Directrice de recherche ; latifa.oukhellou@ifsttar.fr

Haut de page

Droits d’auteur

Licence Creative Commons
La revue Cybergeo est mise à disposition selon les termes de la Licence Creative Commons Attribution - Pas d'Utilisation Commerciale - Pas de Modification 3.0 non transposé.

Haut de page