Scholarly article on topic 'Mixing two French tools for delivery areas scheme decision making'

Mixing two French tools for delivery areas scheme decision making Academic research paper on "Civil engineering"

Share paper
OECD Field of science
{"Urban freight" / "urban policy" / modelling / simulation / "decision-making support"}

Abstract of research paper on Civil engineering, author of scientific article — Loïc Delaître, Jean-Louis Routhier

Abstract Delivery area schemes are constructed in a phased manner to answer local problems without knowing the consequences at the global level in the city. An unsuited or poorly distributed scheme rapidly decreases the efficiency of deliveries in the city. If the number of areas is not enough to match the demand in deliveries, the amount of illegal parking to deliver the goods increases, which slows down the flow of vehicles. But a scheme with too many delivery areas causes unused public space. To achieve a balance, the transport authorities need tools to diagnose according to a holistic view of the logistical problems in urban areas. In Urban Freight Transport (UFT), two French tools have been developed to improve knowledge in this field with their own drawbacks. In mixing the two, we create a synergy and an efficiency of use. Finally, to illustrate our work, we propose to implement the combined tool in La Rochelle.

Academic research paper on topic "Mixing two French tools for delivery areas scheme decision making"

Available online at


Procedia Social and Behavioral Sciences 2 (2010) 6274-6285

The Sixth International Conference on City Logistics

Mixing two French tools for delivery areas scheme decision making

Loïc Delaîtrea*, Jean-Louis Routhierb

aMINES ParisTech, CAOR, 60 boulevard Saint Michel 75272 Paris Cedex 06, France bLaboratoire d'Economie des Transports, ISH, 14 Av. Berthelot, 69363 Lyon Cedex 07, France


Delivery area schemes are constructed in a phased manner to answer local problems without knowing the consequences at the global level in the city. An unsuited or poorly distributed scheme rapidly decreases the efficiency of deliveries in the city. If the number of areas is not enough to match the demand in deliveries, the amount of illegal parking to deliver the goods increases, which slows down the flow of vehicles. But a scheme with too many delivery areas causes unused public space. To achieve a balance, the transport authorities need tools to diagnose according to a holistic view of the logistical problems in urban areas. In Urban Freight Transport (UFT), two French tools have been developed to improve knowledge in this field with their own drawbacks. In mixing the two, we create a synergy and an efficiency of use. Finally, to illustrate our work, we propose to implement the combined tool in La Rochelle. © 2010 Elsevier Ltd. All rights reserved

Keywords: Urban freight; urban policy; modelling; simulation; decision-making support

1. Introduction

Urban Freight Transport (UFT) takes an increasingly important place in the urban mobility strategy because it produces higher costs by increasing emissions of pollutants (NOx), greenhouse gases (CO2) and noise. Considering the proportion in freight in traffic, the flow of goods contributes in a disproportionate manner to pollution and thereby reduces the quality of life in the city.

The set of environmental problems and accessibility, whether for the carriage of passengers or for the distribution of goods in the city, endangers the viability and sustainability development of urban areas. Moreover, the effectiveness of goods distribution in the city itself is hampered by congestion, which may be more or less compounded by the actions of public policy for reducing other nuisances.

Currently, the impacts of UFT on overall traffic are partially known. In fact, surveys and analyses are used to quantify the flows of freight vehicles travelling, and therefore to evaluate the impacts of their presence on all traffic.

* Corresponding author. Tel.: +33-14-051-9028; fax: +33-14-051-9110. E-mail address:

1877-0428 © 2010 Published by Elsevier Ltd. doi:10.1016/j.sbspro.2010.04.037

However, an important part of the street occupation is generally forgotten: the on-street parking for delivering. Moreover, in the dense urban areas of European cities, the lengthening of on-street parking for delivering goods is twice as the lengthening of running vehicles (Bonnafous, 2000). But, the impact of obstructing parking, or even illegal (infractions to the road rules) is not documented. This kind of parking has several causes, but the most important is the lack and the unavailability of delivery areas (IAURIF, 2004).

Matching delivery requirements (flows and duration) with spaces set aside for this purpose for goods vehicles can reduce these impacts. To undertake such an approach at global level, a city must first establish a diagnosis of the needs and the existing situation and then determine an optimal scheme of delivery areas. A scheme of delivery areas is provided by the spatial distribution of its delivery areas at city scale.

This paper aims to provide a tool able to simulate the operations of delivery areas in order to diagnose and to help decision making processes to develop adapted scheme for parking in a city by estimating the number of goods vehicles which could be troublesome and then, in order to estimate the inconvenience on the overall traffic of the city. To reach the goal, two tasks are distinguished: the generation of data and the simulation of delivery areas. The French tool FRETURB©, initially presented in (LET, 2001), developed by LET (Laboratoire d'Economie des Transports) is used for the first step. It calculates an evaluation of the delivering times for each establishment in a area at any time of the day and for any various types of vehicles (light vehicles, rigid lorries, articulated trucks). The French tool DALSIM© (Delivery Areas and Logistics SIMulation) simulates scenarios of the distribution of the delivery areas. DALSIM allows diagnosis of a scheme of delivery areas, taking into account their impact on global traffic flows (Delaitre et al., 2008). DALSIM is divided into two modules: a module for the simulation of delivery areas, and a module that highlights the spread of the obstruction on the overall traffic. Both parts are based on the queueing theory included in the Operations Research and the Systems Dynamics respectively. The resulting tool, by mixing the two described before, is applied in the French city of La Rochelle to analyse the feasibility of such a modelling approach.

2. Problem Statement

Within the organization of the UFT system, delivery areas have a specific role (Delaitre et al., 2007): these are stop areas, and not suitable for parking, where a driver can stop their vehicle to perform freight loading and unloading operations without disrupting traffic flow to the commercial and industrial activities in a limited radius. As a result, delivery areas have a strong impact on local traffic and the flow of goods.

Although the duration of stops is very short (30% of all deliveries take place in less than 5 minutes, 60% in 10 minutes), more than 50% of the deliveries take place in unauthorized spaces in Paris (IAURIF, 2004). The reasons are numerous and the most frequently quoted concern the unavailability of the delivery areas, regularly occupied by the shopkeepers, craftsmen and even local residents, or the remoteness of the locations regards to the delivery points (parking of delivery persons being always closer to delivery points). A total of 75 % of deliveries are done in ten minutes in the agglomerations of Bordeaux, Marseille and Dijon (Ambrosini et al., 1997, 1999a, 1999b). In fact, just over 10% of the recipients do not have a delivery area nearby.

Given this situation, there is a real problem and to solve it local carriers, merchants and freight owners must agree. Local authorities as a "conductor" must measure the impact of the ineffectiveness of the location and the size of delivery areas in the city to be able to suggest improvements. Thus the problem lies for policy makers around the optimization of location and the size of delivery areas.

The list of cases is obviously not exhaustive, other cases (derivative cases) can be broken especially considering improvements to roads such as bike paths or even the presence of sidewalks, the number of lanes, and the types of parking vehicles (in the middle lane, half Lincoln, on a traditional parking space). In this paper, we focus on the diagnosis of a scheme of delivery areas in a city, i.e. we calculate the number of obstructing vehicles (vehicles which did not find any available place on a delivery area) and assess the impact of these vehicles on global traffic. The case 3 is unusual in the sense that it presupposes an illegal situation regards to citizens (occupying illegally an area). We choose not to discuss case 3 (a private vehicle occupies the area of delivery) or its derivative cases, as it theoretically should not happen. In this context, only cases 1 and 2 are considered (Figure 1).

Case 1:

No delivery area

The goods vehicle stops on the street, slowing down or stopping the flows back (two vehicles in this picture)


¡■¿■¿■¿J Obstructing goods vehicle Goods vehicle

Vehicle : car... etc K///^ Delivery area

--------- Street

Case 2:

One delivery area near the delivery point

ii . i . i I . i . M

The goods vehicle stops on the street, slowing down or stopping the flows back (two vehicles in this picture), because the area is occupied by another goods vehicle delivering an activity.

Case 3:

One delivery area near the delivery point



The goods vehicle stops on the street, slowing down or stopping the flows back (two vehicles in this picture), because the area is occupied by another vehicle (shopkeepers, private car...).

Figure 1 Uses patterns of a delivery area (Source: Delaître et al., 2008)

3. Existing Approaches

Although the problem of sizing and positioning delivery areas is common, we observe a lack of knowledge in the interaction between the goods vehicles parking and the flow of circulating traffic. This issue is a source of potential optimization of vehicles movements in the city since the zones without an effective plan of delivery areas generate negative impacts on network capacity and security (Aiura and Taniguchi, 2005).

In this section, we describe three main types of approaches as well as their advantages and limits to tackle this problem, which have been developed in France.

3.1. A pragmatic rule

The first approach deals with an efficient rule whose objective is to determine the number of delivery areas among to the number of deliveries in the areas' neighbourhood. This consists of considering the total time whose delivery area could be used and dividing it by the mean of delivery durations. The output is a number of deliveries that could be allocated to a delivery area. Obviously, this rule is quite simple to implement but requires data difficult to collect.

3.2. The Freturb model

The Freturb model in its 3rd version is able to estimate the delivery demand for each delivery area (Routhier and Toilier, 2007). It is considered as a policy-oriented simulation model of Urban Goods Movement (Sonntag and Meimbresse, 2007). It is a regression model calibrated with the results of thorough coupled 4,500 establishments and 2,200 drivers surveys carried out in three different sized towns (Routhier and Patier, 2008). The statistical observation unit is not the goods (classical approaches), but the movement of vehicles (that is the event of goods delivering or pick-up carried out when a vehicle enters the establishment). As input data, it requires a local establishment database but no large local surveys. The reason is that the authors showed that the coefficients are not changing considerably from one French city to another.

The Freturb model makes possible to simulate several data on zones or road sections of the urban area, such as:

• the number of movements of vehicles each week, according to 45 types of activity, the proportions of shipments/receptions, the average time spent for delivery or pick-up, three types of vehicles (LGV <3.5 t., rigid lorries, articulated vehicles). (module 1 in the next figure);

• the road occupancy by the delivery vehicles in the study area (module 2: trip generation per zone);

• the parking time (on-street or otherwise) of the vehicles in each zone (module 3: loading-unloading time generation);

• an approximate breakdown of the deliveries in the study area (module 4: Distribution of vehicle movements each half an hour a day), and

• the trip distribution from a zone to another (module 5: O/D Trips distribution).

Figure 2 Structure of the global Freturb model (V.3)

3.3. The Dalsim model

Finally, the third concern is addressed by the Dalsim model which diagnoses a delivery areas scheme to help decision making process in urban design.

Dalsim aims to simulate different scenarios considering configurations of delivery areas to evaluate the impacts on traffic flows in a global manner. It simulates the use of each delivery area to estimate the temporal distribution of

obstructing vehicles. These kind of vehicles are said obstructing, because they could not find any available area to deliver the goods and consequently have to stop on the street. Then the tool simulates the spread of the obstruction on the global traffic.

This is a hybrid tool, a delivery area is modelled by a queuing system and the impacts on traffic overall are given by a systems dynamic model. It is therefore, divided into two sub-models as illustrated in Figure 3.

Distribution of the overall vehicles in the city Calibration parameters

Figure 3 Structure of the Dalsim tool (Source: Delaitre et al., 2008)

The first module considers each delivery area as a queue with a capacity, a neighbourhood radius (the maximum distance between the place to deliver and the delivery area), an arrival rate of goods vehicles and a departure rate. Then, three types of vehicles are considered to build the second module among their status on the network, the obstructing vehicles, the obstructed vehicles and the others vehicles (which are neither obstructing nor obstructed). The obstructing vehicles are the goods vehicles which do not find any legal place to deliver their goods, i.e. when all the delivery areas are full or which deliberately do not use facilities. The obstructed vehicles are the vehicles affected by the impacts of the obstructing vehicles, which results in a decrease of efficiency. This tool allows the study of the relationship between the obstructing vehicles and the obstructed vehicles and puts forward the spread of the obstruction (Delaitre et al., 2008).

4. Proposed Tool

The proposed tool emerged from two main critical analyses about the Freturb tool and the Dalsim tool. By using the modules 1, 3 and 4 of the Freturb Model (see Figure 4), it is possible to provide a diagnose the urban freight movements in the study area. The Freturb tool is able to simulate the behaviour of economic activities in terms of deliveries and pick-ups in the neighbourhood of the delivery area (NDA):

• From module 1, it calculates the number of deliveries and pick-ups carried out on the day in each NDA,

according to the activities located in this NDA. The articulated vehicles are not accounted, because they don't use the delivery areas.

• From module 3, it makes possible to make an appraisal the average parking time in the study area according to the type of vehicle.

• From the module 4, it makes it possible to obtain the number of vehicle occurrence for the entire day.

Without any specific surveys, the Freturb model makes possible to estimate the distribution along the day of the vehicles and the parking time demand in each NDA (each half an hour).

The Dalsim tool requires a lot of data that are difficult to obtain or could only be gained with too long and too expensive surveys.

Figure 4 Structure of the Freturb tool used in "Freturb-Dalsim"

With these two facts, we gather these two tools among to the following figure to simulate scenarios in changing the location and the size of delivery areas, adding new areas or deleting ones. The proposed tool is illustrated in Figure 5.


Output/ Input


■ w MS Excel

Change scenario Diagnostic Satisfying? ^

*Location and size of

areas 1 Yes

*Adding or deleting areas

r End

Figure 5 Structure of the proposed tool and process of use

The process of simulation begins with a national French file collecting all economic activities of any city (the SIRENE File). This file is available for 0.12€ per line, i.e. per activities. From this file, the Freturb model is able to give an MS Excel file which provides per considered zones (for us, a zone is a street), the number of movements per vehicle type per half an hour. This built file is given as input to Dalsim which then simulates the occupation of delivery areas for each zone, i.e. for each street. In this paper the second module is not studied because we are focused on describing the linkage between the Freturb and Dalsim models.

The proposed tool, is not used as a decision support system in this paper, but is evaluated in order to know if the mix of the two models described below is feasible or not and gives interesting results to avoid costly surveys or not.

The theoretical basis of the proposed tool is not described because the overall basis of each tool is summarized. As a consequence, the theoretical framework is the sum of the theoretical basis of Freturb and Dalsim. The link between the two is provided by the MS Excel File. Nevertheless, we advise the reader to consult the previous papers dealing with the two models in order to have a better and a precise view (Routhier and Toilier, 2007; Delaitre et al., 2008).

5. Results in La Rochelle

5.1. Geographic area

This case study is a validation of the feasibility with the combined tool. It concerns the city centre of La Rochelle representing 80,055 inhabitants (10,827 in this area) and more than 6,000 economic activities (almost 2,000 in this area).

Figure 6 Case study area

5.2. Diagnostic of delivery areas scheme of La Rochelle city centre

We present two diagnostic overviews of the delivery areas scheme of La Rochelle city centre, the first is using only the Dalsim model (called Dalsim calibrated by surveys), the second is using the combined tool Dalsim and Freturb (called Dalsim calibrated by Freturb).

Using Dalsim calibrated by surveys, the Dalsim model has been already calibrated and validated in La Rochelle (Delaitre et al., 2008). To be calibrated, the model required quite expensive surveys for each activity in the city centre, which represent the number, the duration and the time windows of deliveries. For La Rochelle, the surveys have been led in collecting all the activity deliveries (number, duration and time windows) very near to a delivery area (a maximum distance of 10 metres). This threshold was found by asking carriers what would be the maximum distance between delivery area and delivery point to use a delivery area. In La Rochelle, the carriers want to benefit by travelling a minimal distance by foot. Figure 7 explains geographically the activities involved or not in the surveys.

Part of the city

On this basis, Dalsim simulates the arrival on and the departure from each delivery area of goods vehicles and gives the number of obstructing vehicles (vehicles being not able to find any available area to deliver the goods). For example, the Figure 8 shows the number over time during one week. This is an undervalued representation of reality, because for instance, case 3 is not considered even if it is not neutral. To compare these results from the simulation, we conducted a small number of surveys to observe what happen in reality in the streets, and are represented in the graph by "observations" (Figure 8).

The variation between the simulation and the surveys is quite low. Figure 8 shows that the simulation, which does not take into account the obstruction of illegal parking of vehicles, reflects the trend of observation on 10 delivery areas (the Market place is considered special, so we did not include it).

The most relevant streets are "Rue Saint Jean du Pérot", "Rue Chef de ville", and "Rue Fleuriau" which are shown in Figure 6. For budgetary reasons, only half of the areas have been observed for the validation of the simulation, bringing the threshold of study from 15 obstructing vehicles by week. We have also postponed the number of obstructing vehicles generated by illegal occupations of the area by private vehicles. Such observation helps to know the biases caused by the case where private vehicles are parked illegally.

The major drawback of this methodology is that we do not control the evolution of the data of calibration, i.e. the surveys whose objective is data generation. This can be improved in using the Freturb model to generate data for calibration and simulation of Dalsim.

Using the Dalsim calibrated by Freturb. Starting from the SIRENE file, we generate with Freturb the data required by the Dalsim tool. The Freturb tool supplies the number of movements per week, the percentage of shipments and reception, the type of vehicles and the management mode for each category of activity. The Dalsim tool requires the temporal delivery distribution for each activity. As a consequence, the interesting results to be kept from Freturb are the number of movements twice the proportion of reception. Indeed, the expeditions and the management mode are not interesting for the delivery area diagnostic and analysis. The results are shown in Table 1.

Number of obstructing vehicles per week

■Q £

160 140 120 100 80 60 40 20 0

obstructing vehicles from observation — ■■ - - Number of obstructing vehicles from simulation

\\ / \ ж

ж \ x ч .

■ Ï-.J.--1' --1-1-1-1-1-1-1-1-1-

~ ' Ж ^

^ <-40 Л

K& ^ r^ <f «


Figure 8 Simulated distribution and observations of obstructing vehicles in La Rochelle (Source: Dalsim)

Table 1 Number of obstructing vehicles simulated by the proposed tool

Zone Streets Number of Number of obstructing vehicles from DALSIM tool Number of

delivery areas DALSIM calibrated with surveys in La Rochelle DALSIM calibrated with Freturb obstructing vehicles observed in the zones (observations)

1 Saint Jean de Perot 5 77 92 85

2 Rue du Palais 2 20 45 16

3 Rue Leonce Vieljeux 1 2 6 -

4 Place Foch 1 4 2 -

5 Rue Chef de Ville 3 49 62 56

6 Rue Dupaty 1 7 25 -

7 Rue Villeneuve 1 14 45 12

8 Rue Gambetta 2 11 18 15

9 Rue Saint Louis 1 3 5 -

10 Rue Amelot 1 8 12 -

11 Rue du minage 1 - 43 -

12 Rue Gargoulleau 1 17 19 15

13 Rue Fleuriau 1 35 15 29

14 Place du marché 1 22 15 80

15 Rue Thiers 1 15 56 14

16 Rue Emile Normadin 3 14 48 -

17 Quai Valin 2 21 26 22

18 Quai Louis Durand 1 8 8 -

19 Rue Albert 1er 2 8 6 -

Mean 1,63 18,61 28,84 34,40

This table gives for each street the numbers of obstructing vehicles from the simulation of the combined tool (Freturb + Daslim), from the simulation of Dalsim calibrated by surveys and from what we have observed in reality still called "observations".

In a global view, each studied street has problems with their delivery areas, more or less, from at least 1 obstructing vehicles to 77 for the street "Rue Saint Jean du Perot". The mean number of obstructing vehicles for each street is more than 18.61 vehicles, which is quite important for Dalsim calibrated by surveys and 28.84 for Dalsim calibrated by Freturb. The mean 28.84 indicates that Dalsim calibrated by Freturb generates more obstructing vehicles than Dalsim calibrated by surveys. These results are coherent because Freturb considered all the activities of each street while the surveys limit the number of activities within a maximum distance (10 metres around the delivery area) as shown in Figure 9.

Figure 9 Activities considered in surveys and activities considered by Freturb

Then, the mean number of observations is quite high but it is quite normal because initially, all streets with a simulated number of obstructing vehicles under 15 were not selected to be observed in reality. Figure 10 shows curves from simulations and observations.

Without any surprises, the proposed tool, i.e. Dalsim calibrated by Freturb, overestimates the number of obstructing vehicles. It is due to mainly a huge number of activities. The sensibility with this parameter (number of considered activities per delivery area) increases when the number of area is equal to one and when the length of the street is quite important. This is the case for the streets "Rue Villeneuve" and "Rue Thiers": these two streets are very long and comprise only one area each. It means that in the simulation of Dalsim calibrated by surveys, a major part of the activities in these streets are not considered while it produces goods movements. In that case, the Freturb calibration is closer to reality.

The market place "Place du marché" is a special case to analyse because one day a week, delivery areas are used to service the commercial market and it produces on this day a huge number of obstructing vehicles. On the other days, the "Place du marché" has no particularly reasons to generate obstructing vehicles as shown by the simulations. Finally, Table 2 calculates the gap between simulations and observations.

A comparison between the means of the gaps could be one way to evaluate their performance. Globally, Dalsim calibrated by surveys underestimates reality (because the mean of gaps is positive) and Dalsim calibrated by Freturb overestimates reality (because the mean of gaps is negative). But in absolute values, the proposed tool is better than Dalsim calibrated only by surveys because the mean of gaps is lower (4.9 compared with 6.3) and the tool is more efficient.

Number of obstructing vehicles from simulations and observations


Figure 10 Graphics of simulations and observations

Table 2 Gaps between simulations and observations

Zone Streets Gap between observations and Dalsim calibrated by surveys Gap between observations and Dalsim calibrated by Freturb

1 Saint Jean de Perot 8 -7

2 Rue du Palais -4 -29

5 Rue Chef de Ville 7 -6

7 Rue Villeneuve -2 -33

8 Rue Gambetta 4 -3

12 Rue Gargoulleau -2 -4

13 Rue Fleuriau -6 14

14 Place du marché 58 65

15 Rue Thiers -1 -42

17 Quai Valin 1 -4

Mean 6,3 -4,9

6. Conclusion

We briefly presented two French models of urban freight: one dealing with the generation of vehicles movements (Freturb), another dealing with the simulation of delivery areas (Dalsim). The added value of this paper was, first, to combine the two models because they are complementary, and second, the proof of feasibility in mixing the models

for a practical case. Indeed, Freturb generates and gives data to calibrate the Dalsim tool, without needing expensive surveys. Obviously, this is currently limited to a case study but it shows the potential benefits.

The main result is that the combined tool, i.e. the Dalsim tool calibrated by Freturb, is more efficient to generate the number of obstructing goods vehicles and is estimating more realistically what happens around delivery areas than the Dalsim tool alone (calibrated by specific surveys). The results of the proposed models are challenging the robustness of our observations because we have collected data during one week and this is perhaps not representative enough, since the models use means, which are statistically more representative.

As a consequence, there are two ways of development: the Dalsim tool could be used without surveys which would avoid costs, and/or the Freturb model could add a supplementary module to simulate delivery areas.

As the link between the tools is still manual, future work will be concerned at first on the development of an automatic link between them. Then, we will put forward the advantages of such a tool in a decision process. Indeed, the tool is able to simulate various scenarios. Scenarios are built by the user in changing some parameters such as the location and the dimension of areas and the distribution of economic activities for prospective, or in adding new areas to overcome an excessive number of deliveries at a specific place in the city. By iteration, the user can analyse the various scenarios they implement which helps the decision making process for urban planning. The results of each simulation is the temporal distribution of the number of vehicles which have not been able to find any available areas to deliver goods (obstructing vehicles) and the estimation of the impacts of these obstructing vehicles on the global traffic in terms of obstructed vehicles.


Aiura, N., Taniguchi, E. (2005). Planning on-street loading-unloading spaces considering the behaviour of pickup-delivery vehicles and parking

enforcement. Proceedings of the 4th International Conference on City logistics, Langkawi, Malaysia, 2005. Ambrosini, C., Patier, D., & Routhier, J.-L. (1997-1999). Transport de marchandises en ville, enquêtes quantitatives de Bordeaux, 226p.

enquêtes quantitatives de Marseille, 120p, 1999a, enquêtes quantitatives de Dijon, 122p, 1999b. Bonnafous, A. (2000). Les marchandises en ville : le problème méthodologique de l'appréhension statistique, in « L'intégration des marchandises

dans le système des déplacements urbains ». Etudes et Recherches LET n°15 (pp. 85-93), 2000. Delaître, L., Awasthi, A., Breuil, D., & Molet, H. (2007). An entropy based characterization of medium sized cities for selecting urban freight

solution. National Urban Freight Conference, December 4 - 5, 2007, Westin Long Beach Hotel Long Beach, CA 90802. Delaître, L., Molet, H., & Breuil, D. (2008). Outil de simulation pour le diagnostic des aires de livraison : Le cas de La Rochelle. Revue

Internationale de Recherche Transport Sécurité (RTS). IAURIF (2004). Note rapide sur les transports n°362. Le transport de marchandises par véhicule utilitaire léger en Ile de France. Octobre 2004. LET (2001). Mesurer l'impact du transport de marchandises en ville — le modèle FRETURB version 1. DRAST ADÈME, p51-78. Routhier, J-L., & Toilier, F. (2007). FRETURB V3, a policy oriented software tool for modelling urban goods movement. 11th WCTR, Berkeley, CA, USA.

Routhier, J-L, & Patier, D. (2008). How to improve the capture of urban goods movement data? Resource paper International Conference on

Survey Methods In Transport, Annecy, France, May 2008. Sonntag, M. (2007). Bestufs Work Package 3 Urban freight data harmonisation and modelling. 3rd round table 2007, Brussels.