Scholarly article on topic 'CLASS: A DSS for the Analysis and the Simulation of Urban Freight Systems'

CLASS: A DSS for the Analysis and the Simulation of Urban Freight Systems Academic research paper on "Civil engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
Transportation Research Procedia
OECD Field of science
Keywords
{"decision support system" / "urban goods movements" / "city logistics" / "urban freght demand" / "shopping demand" / "restocking demand."}

Abstract of research paper on Civil engineering, author of scientific article — Antonio Comi, Luca Rosati

Abstract The complexity of methods and models for the investigation of urban freight systems has pushed researchers to develop decision support systems (DSS) for the analysis and simulation of the effects of city logistics measures and the effects of exogenous scenario changes, as land-use, demographic and socio-economic characteristics. In fact, today these types of systems provide to support decision-makers to understand and simulate the structure of freight urban system and to compute some indicators that compared with target and benchmarking values allow to identify its level of service. Such a DSS, named CLASS (City Logistics Analysis and Simulation Support System), has been recently proposed and in this paper some advancements are presented. Then, two application examples for the simulation of urban freight transport in a large urban area and the assessment of freight distribution activity location in a medium size urban area are presented.

Academic research paper on topic "CLASS: A DSS for the Analysis and the Simulation of Urban Freight Systems"

Available online at www.sciencedirect.com

ScienceDirect

Transportation Research Procedía 5 (2015) 132- 144

Transportation

Procedía

www.elsevier.com/locate/procedia

SIDT Scientific Seminar 2013

CLASS: a DSS for the analysis and the simulation of urban freight

systems

Antonio Comia* and Luca Rosatib

a Department of Enterprise Engineering, Tor Vergata University of Rome, Rome, Italy b Department of Civil Engineering and Computer Science Engineering, Tor Vergata University of Rome, Rome, Italy

Abstract

The complexity of methods and models for the investigation of urban freight systems has pushed researchers to develop decision support systems (DSS) for the analysis and simulation of the effects of city logistics measures and the effects of exogenous scenario changes, as land-use, demographic and socio-economic characteristics. In fact, today these types of systems provide to support decision-makers to understand and simulate the structure of freight urban system and to compute some indicators that compared with target and benchmarking values allow to identify its level of service. Such a DSS, named CLASS (City Logistics Analysis and Simulation Support System), has been recently proposed and in this paper some advancements are presented. Then, two application examples for the simulation of urban freight transport in a large urban area and the assessment of freight distribution activity location in a medium size urban area are presented.

© 2015TheAuthors.Publishedby Elsevier B.V. Thisis an open access article under the CC BY-NC-ND license (http://creativecommons.Org/licenses/by-nc-nd/4.0/).

Peer-review under responsibility of the Societa Italiana dei Docenti di Trasporti (SIDT).

Keywords: decision support system; urban goods movements; city logistics; urban freght demand; shopping demand; restocking demand.

1. Introduction

The growing necessity to improve city sustainability and livability has pushed local administrators to look at city logistics measures. They mainly managed operational actions in order to reduce interferences with other vehicles and inhabitants (e.g. using Limited Freight Traffic Zones) and the pollutant emissions (e.g. using constraints based on Euro emission standards and new urban distribution centers). Recently, local administrators are also looking at other

* Corresponding author. Tel.: +39 06 7259 7059; fax: +39 06 7259 7053. E-mail address: comi@ing.uniroma2.it

2352-1465 © 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.Org/licenses/by-nc-nd/4.0/).

Peer-review under responsibility of the Società Italiana dei Docenti di Trasporti (SIDT). doi:10.1016/j.trpro.2015.01.007

classes of measures of medium/long term type, such as urban land-use development governance: allotment of retail area among small, medium and large retail activities; location of retail activities with respect to warehouses and distribution centers.

The proper choice of a set of city logistics measures (city logistics scenario) to be implemented has to be based on the simulations of the specified scenario (i.e. using a "what if approach"), computing some effect indicators able to quantify the expected results in terms of internal and external, direct and indirect cost variations (Nuzzolo and Comi, 2014a). Generally, these indicators are obtained from the road network performances and impacts forecasted using procedures that require as input the shopping car and the freight vehicle O-D matrices. These matrices are assigned to the road network, obtaining link flows, which in turn are used as inputs of the other models that, for example, allow determination of pollution emissions, energy consumption, road accidents and so on.

Several methods and models have been developed for the estimation of shopping car O-D matrices and the restocking freight vehicle O-D matrices (Nuzzolo et al., 2013a). The complexity of these methods and models has pushed researchers to develop decision support systems (DSS) for the analysis and simulation of the effects of city logistics measures and the effects of exogenous scenario changes, as land-use, demographic and socio-economic characteristics (Taniguchi et al., 2012; Munuzuri and Gonzelez-Feliu, 2013). In fact, today, these support systems are able to support decision-makers to understand and simulate the structure of freight urban system and to compute some indicators that compared with target and benchmarking values allow to identify its level of service (Browne et al., 2012; Stathopoulos et al., 2012). Many of these tools are based on empirical relationships that well describe the current state of the system (Sonntag, 1985; Boerkamps and van Birsbergen, 2000; Lohse, 2004; Routhier and Toiler, 2007; Gentile and Vigo, 2013) but they could have some limits when new city logistics scenarios (before implementation) are simulated and assessed: Wiver and Viseva-W, GoodTrip, Freturb, CityGoods. Based on this statement, a new DSS, called CLASS (City Logistics Analysis and Simulation Support System), has been presented in Comi and Rosati (2013). The earlier version of CLASS is mainly devoted to the identification of critical stages and the simulation of shop restocking. Then, some advancements were implemented in order to link shopping and restocking mobility, including a shopping model sub-system that allow to take into account that shop restocking is mainly due to satisfy the demand of end consumers that buy at shops. CLASS implements new advanced demand models that provide to capture the effects on actors' behavior of city logistics measures, land use scenarios and demographic changes (Nuzzolo and Comi, 2014a). The DSS architecture and the advancements in demand modelling are described in Section 2, while Section 3 reports the results of two application examples: simulation of urban freight transport in the inner area of Rome and the location assessment of freight distribution activities in Padua, a medium-size urban area. Finally, some conclusions and further developments of this study are given in section 4.

2. CLASS: the proposed decision support system

CLASS has been developed to satisfy the requirements of:

• City Logistics Managers that want to identify the main characteristics and the critical stages of the actual City Logistics System (CLS);

• Land Use planners that desire to assess and to compare demographic, socio-economic and land use scenarios for defining optimal spatial distribution and accessibility of urban freight facilities (e.g. shopping malls, warehouses, distribution centers) able to improve city sustainability.

To meet the above aims, CLASS provides the computation of a set of indicators related to:

• demographic and socio-economic changes, able to describe the trends in shopping attitudes and preferences including the impacts of new way of purchasing (e.g. e-shopping);

• land use, able to describe the commercial land use characteristics of the study area (e.g. total number and ratio between employees and residents);

• freight demand and supply, able to describe the freight moved within the study area according to different reference units (i.e. quantity, deliveries and vehicles), transport services (e.g. on own account or by third parties) and type of vehicle used for restocking (e.g. light, medium and heavy goods vehicles);

• logistic profile, able to identify areas homogeneous with respect to some specific logistics needs (Macário et al., 2008);

• road network performances and impacts, able to estimate network transportation costs (e.g. travel times and operative costs), traffic pollutant emissions, and road accidents involving both passenger and freight vehicles.

2.1. The logical architecture

The logical architecture of CLASS consists of: input database, road network module, demand module, assignment module and output module (Figure 1).

The input database contains information on exogenous variables of the scenario, as resident and employees of each traffic zones, traffic management measures, number of loading/unloading areas, presence and number of logistic infrastructures, like Urban Distribution/Consolidation Centre, Nearby Delivery Area or Transit Point, and information about goods vehicle road accidents.

Figure 1. CLASS framework (Comi and Rosati, 2013)

The road network module comprises the graph of the main road network and relative link cost functions specific to both passenger and freight vehicles. The module simulates the relevant aspects of travel demand as a function of the activity system and road travel costs.

The demand module implements an advanced a multi-stage modelling system that considers a discrete choice approach for each decisional level. The shopping and restocking flows (in terms of passenger trip, freight quantities, deliveries and private and commercial vehicles) are simulated through two model sub-systems able to estimate the shopping mobility O-D matrices, the restocking quantity O-D matrices, the delivery O-D matrices and the

restocking vehicle O-D matrices. This module hence gives the O-D matrices which are the input for the subsequent assignment sub-system.

The assignment module includes path choice models and network loading models for both passenger and freight vehicles. The network loading model simulates how O-D vehicle flows load the paths and the links of the road network, and estimates the link flows, i.e. the number of cars and freight vehicles on each network link.

The output module allows to estimate, using data and results of above modules, the city logistics scenario indicators. It includes four sub-modules: land-use indicators, demand and supply indicators, logistic profile and road network performances.

In the land-use indicator sub-module allows to describe the commercial land-use characteristics of the study area. Several indicators can be computed and used for characterizing each traffic zone. CLASS provides the evaluation of the following metrics: total number and ratio between employees and residents; total number and density of retailer outlets; total number and density of retail employees also disaggregated for freight type; average number of retail employees also disaggregated for freight type.

The demand-supply indicator sub-module allows to compute freight demand and supply indicators. As regards demand, CLASS calculates different metrics, for example:

• the freight quantities and deliveries produced and attracted by zones also disaggregated for freight type and type of retail outlets;

• the average delivered quantity in relation to each types of transport service (e.g. on own account and third party), freight and retail outlets (e.g. small, medium and large);

• the deliveries time slice distribution; according to time, the delivery freight flows are computed and hence split per time period (e.g. one in the morning between 8:00 and 10:00 am and one in the afternoon).

For what concerns, supply indicators, CLASS estimates for each freight types:

• the services for transport quantity and deliveries transported by different types of transport services (i.e. on own account and third party);

• the vehicles fleet used for freight restocking characterized in terms of vehicle types (e.g. light or medium or heavy), equipment (e.g. refrigerator), emission standards (e.g. Euro I, Euro II) or type of fuel (e.g. gasoline, diesel).

The logistic profile sub-module provides indicators useful for the identification of the areas homogeneous with respect to some specific logistics needs:

• commercial density and homogeneity through the total number of retail outlets and/or of the retail outlets types per block (i.e. small, medium or large);

• logistic accessibility through the congestion level on the streets serving the zone, the presence of traffic management measures for the zone (e.g. time windows, zone 30, Limited Traffic Zones, pedestrian streets), the number of loading/unloading zones and the average access distance among loading/unloading zones and retail outlets, the presence of logistic infrastructures as Urban Distribution Centre (UDC), Nearby Delivery Areas (NDA) or Transit Points (TP);

• characteristics of the products destined to retail outlets, like fragility, perishability or cooling needs;

• actor (e.g. retailers, wholesalers, carriers) needs according to particular requirements, like urgency of deliveries, frequency of deliveries, quantity to be delivered and time slice of deliveries (e.g. morning, afternoon or night).

In the road network performance sub-module, the link flows (output of assignment module) are used for the estimation of several scenario indicators that, in turn, are used for computing the new scenario effects:

• network transportation costs as travel times and operative costs;

• traffic pollutant emissions, using average emission functions that allow pollutant emissions to be estimated in relation to average link kinematic variables (e.g. vehicle speed and acceleration); for example as implemented within COPERT (COmputer Programme to calculate Emissions from Road Transport; Eggleston et al., 2000);

• road accidents involving both passenger and freight vehicles, using accident prediction models; the probability of accidents are evaluated in relation to vehicle flows and road characteristics, location and characteristics of infrastructures (e.g. junction), control system (e.g. traffic lights, crosswalks), and other standardization variables (e.g. reference period, environmental conditions).

2.2. The demand model

The implemented and used demand modelling system allows to capture the effects of city logistics measures on actors' behavior (Nuzzolo and Comi, 2014a and b). It allows to analyze shopping mobility as a component of freight mobility and considered that changes in shopping attitudes or actions impacting on purchasing behavior of end consumers (e.g. location of shopping zone, transport mode to use for shopping) can also affect shop restocking mobility. It consists of two main steps (Figure 2):

• shopping model sub-system; it allows us to simulate end-consumer behavior for shopping and to estimate quantity bought by end consumers in order to satisfy their needs, and hence to identify the freight flows attracted by each traffic zone;

• restocking model sub-system; given the quantity attracted by each traffic zone, it allows us to estimate the restocking quantity origin-destination (O-D) matrices characterized by freight types and type of vehicle used.

The shopping model sub-system allows us to point out the effects arising from implementation of medium/long-term actions on the location of retail outlets and places of residence, and due to changes in the characteristics of the population (e.g. demographic and socio-economic changes).

The restocking sub-system includes models for the simulation of the freight distribution process from the freight centers to the retail zones, and can be also used to determine the medium/long-term effects arising from implementation of actions on the location of logistic establishments (e.g. warehouses and distribution centers).

SHOPPING RESTOCKING - Quantity

Figure 2: Urban freight modelling structure

2.2.1. Shopping model sub-system

Assuming that the decision-maker (i.e. end consumer) is in zone o, the choice dimensions involved are: the number of trips (x) for shopping, the type of shop (k; e.g. small, medium, large) and destination (d), and the transport

mode (or sequence of modes; m). The global demand function can be decomposed into the product of sub-models, each of which relates to one or more choice dimensions. The sequence used is the following:

D'od [skm] =D'o [s] ■ p' [dk / so] ■ p' [m / dkso] (1)

where:

• D'od [skm] is the weekly average number of trips with origin in zone o undertaken by the end consumer belonging to category i (e.g. families with one or more components) for purchasing goods of type s (e.g. foodstuffs) in the type of retail outlet k (e.g. small, medium and large retail outlets) located in zone d by using transport mode m;

• Dio[s] is the weekly average number of trips undertaken by end consumers belonging to category i for purchasing goods of type s with origin in zone o, obtained by a trip generation model;

• p'[dk/so] is the probability that users, undertaking a trip from o, travel to destination zone d for purchasing at shop type k, obtained by a shop type and location choice model;

• p'[m/dkso] is the probability that users, travelling between o and d for shopping in shop type k, use transport mode m, obtained by a mode choice model.

Finally, the quantities required by each zone to satisfy end consumer needs can be obtained by introducing a quantity purchase model (Russo and Comi, 2012). This model gives the probability that the end consumer, arriving in a given zone, purchases something of a certain size. Therefore, the total quantity of goods type s attracted by retail outlet k in zone d, QTd[sk], can be calculated as:

QTd [sk ] = £Q [sk ]+QEd [sk ] = £ £ D0d [skm]-p' [dim / mks ]-dim+QEd [sk ] (2)

i ' o,m,d'm

where:

• Qid [sk] is the goods quantity bought/sold in retail outlet k in zone d given by the demand of end consumers belonging to category i living in a zone within the study area;

• QE'd [sk] is the goods quantity bought/sold in retail outlet k in d given by the demand of end consumers belonging to category i living in a zone external to the study area;

• dim is the dimension of purchases, e.g. expressed in kg;

• p'[dim/mks] is the probability that a trip concludes with a purchase of dimension dim conditional upon undertaking a trip from zone o to retail outlet k for a purchase of goods type s using the transport mode m.

2.2.2. Restocking model sub-system

Once the quantity of goods bought/sold in retail outlet k in zone d, QTd[sk], has been estimated by the above model sub-system, the restocking quantity flows characterized for vehicle type v, QVod, departing from zone o can be obtained as follows:

QVod [skv] =QTd [sk] • p [a / dks] • p [v / dk] (3)

where:

• QVod[skv] is the freight quantity of type s bought/sold in retail outlet k in zone d transported on pair od by vehicle type v;

• p[o/dks] is the probability that the freight attracted by outlet type k located in zone d comes from warehouse zone o, obtained by an acquisition model;

• p[v/dk] is the probability that the freight attracted by outlet type k located in zone d and coming from zone o is transported by vehicle type v, obtained by a vehicle choice model.

3. CLASS application for analysis and simulation

CLASS was used for the analysis and simulation of the current City Logistics System (CLS) of the inner area of Rome and for assessing the effects of freight activity location strategies upon transport costs in the medium-size urban area of Padua in northern Italy. Below, the results for the two application test cases are reported.

The models described in the earlier section and implemented within CLASS were calibrated using data from two different surveys. One, carried out for the inner area of Rome in 2008, allowed to investigate and set up restocking model sub-system. It consisted of traffic counts of commercial and other vehicles at the border of study area, with about 600 interviews of truck drivers in order to investigate the supply chain of freight distribution within the study area, and about 500 interviews of retailers in order to investigate the retail trade in the study area for each freight type (Nuzzolo and Comi, 2014b and c). The other supported the calibration of shopping model sub-system. It consisted of more than 900 interviews to end consumers (more than 300 households) living in the city of Rome. Each developed model belongs to discrete choice theory and was developed within random utility model. For more details, refer to Nuzzolo and Comi (2014a and d).

3.1. City logistics analysis and simulation

The study area of Rome (Figure 3) is of about 9 km2, with about 50,000 inhabitants and 24,000 employees related to trade. The study was supported by some surveys carried out in 2008: traffic counts of commercial and other vehicles at the border of study area, with about 600 interviews of truck drivers in order to investigate the supply chain of freight distribution within the study area, and about 500 interviews of retailers in order to investigate the retail trade in the study area for each freight type. The study area is a mixed land-use area (CBD, residential, commercial, tourist) which is mainly affected by attraction freight flows (Nuzzolo and Comi, 2014c), while the origins of freight flows take place mainly in the peripheral areas of Municipality. The analysis highlights freight movements in the study area amounting to about 15,000 tons per day and more than 66% is destined to retail or food-and-drink outlets. In terms of freight segmentation, 36% consists of foodstuffs (about 16% is dispatched to restaurants and cafe, and 20% to retailers), 61% consists of other end-consumer products (e.g. household and health products), and the remaining 3% are goods related to services. To analyze the system, the area of the municipality of Rome was divided into 99 traffic zones with a level of detail which increases as the inner area was approached. The inner area consisted of 16 traffic zones.

Figure 3. Study area in Rome and Padua

The following Figures picture the results of CLASS application in terms of CLS indicator estimation. In fact, a GIS system is integrated with CLASS offering a quick and intuitive graphical interface. The output indicators are separately described for land-use, demand, transport supply, logistic profile and road network performances. Figure 4 shows retailers spatial density as thematic map and, for each traffic zone, the distribution of retailer types according to freight type (i.e. foodstuffs, home accessories, stationery, clothing, household and personal hygiene, building, other), as histograms. The northern zones have the highest density of retailers, mainly related to foodstuffs and clothing while the southern zones are characterized by a low retailer density. Figure 4 also shows a thematic map of the daily deliveries with graphs describing the distribution of deliveries between the two types of transport services (i.e. on own account and 3P) and the used vehicle type (i.e. light and medium). It emerges that decreasing the density of daily deliveries, the share of own account increases.

Retailer types

retailer_types H Foodstuff □ Home accessories

■ Household and personal hygiene /

■ Stationery

Retailer density

retailer_density «retailer sJkm2 □ 0

□ 333,333 ■ 866,667

□ 1000

Q 1333,333 a 1666,667

a 2000

Daily deliveries

Delivery types 0 3p light

■ 3p medium

■ owner own light

■ owner own medium

Daily deliveries

\ Deliveries □ 0 iiN f □ 400 II 800

a 1200

■ 1900

■ 2000 ■ 2400

-V \\>

___ w~

Figure 4. Freight distribution performance in the inner area of Rome

An analytic study of land use, freight demand and supply indicators have allowed CLASS to define for each zone the logistic profile zones according to classification proposed by Macario et al. (2008). CLASS identified 3 different logistic profiles for the study area (Figure 5). The profile B, that includes areas with hotels, restaurants, small grocery stores, small neighborhood markets mainly characterized by the perishability of the products, was applied to 7 traffic zones for a total extension of 4.7 km2. These areas contain the 50% of residents and the 65% of employees with a ratio of employees on residents equal to 0.80 and there are the most touristic zones of the city. The logistic profile C (i.e. business center characterized for high commercial density and low homogeneity with low logistic accessibility) was applied to 3 zones of study area for a total extension of 1.1 km2. In these areas, there are many office buildings, and therefore the ratio of employees on residents is equal to 1.26. Finally, the logistic profile E (i.e. area with local trade characterized for low commercial density and homogeneity with low logistic accessibility) was applied to 6 traffic zones with a total extension of 3.4 km2. These areas contain the 42% of residents and the 19% of employees with a ratio of employees on residents equal to 0.29.

■ Profile A

I Profile B

I Profile C

■ Profile D

I Profile.E

Profile A, cluster of shops specialized in one specific type of service/product characterized for high commercial density and homogeneity and low logistic accessibility;

Profile B, hotels, restaurants, small grocery stores, small neighborhood markets mainly characterized by the perishability of the products (ho.re.ca);

Profile C, business center characterized for high commercial density and low homogeneity with a low logistic accessibility;

Profile D, large commercial stores mainly characterized by a good logistic accessibility and a big amount of freight to be delivered;

Profile E, residential areas with local trade characterized for low commercial density and homogeneity with a low logistic accessibility;

Figure 5. Logistic profile examples for the inner area of Rome

The road network performance indicators were evaluated considering the whole main road network of Rome municipality in order to evaluate the total amount of transport costs related to CLS of the inner zone. The road accident module provides some analysis based on location of road accidents involving freight vehicles, as pictured in Figure 6. In 2012, there were 219 accidents in the study area that involved freight vehicle and mainly concentrated within the zone with the highest density of deliveries. Similarly, the Figure 6 shows the freight vehicle link flows required for identifying the main routes followed by trucks delivering within the study area.

Figure 6. Freight distribution performance for the inner area of Rome: accidents involving goods vehicles (left side) and freight vehicle link flows - LGV in blue and MGV in grey - (right side)

3.2. Location of warehouse and retail facilities in a medium-size urban area

Below, the application of CLASS to a medium-size urban area of Padua (Figure 3) for assessing freight facility location effects is briefly described. The assessment was performed comparing three different land use scenarios. The definition of scenarios to be assessed can be considered of the long-term results of some different land-use

governance measures that local administrators could promote in order to improve the sustainability of the city. According to the main characteristics of the identified study area, the study area was considered as a closed system and then it was assumed that the goods quantity bought by end consumers living in a zone external to the study area can be neglected (QE'd [sk] = 0). The focus here is on the application of a joint modelling framework to an urban transportation networks when actions that can also change end consumers' shopping behaviors occur, a topic for which the literature is quite sparse (Crocco et al., 2010; Kawamura et al., 2012; Sanchez-Diaz et al., 2013).

Each scenario was defined hypothesizing shifting of both retail and warehouses employees among the three identified urban spaces: the central area (CA), where the density of end consumers and small retailers is usually higher, the first ring (FR), with medium end-consumer density and the presence of warehouses, and the second ring (SR), where end-consumer density is low and large shopping malls and freight distribution facilities are located. The scenario definition was in terms of percentages of retail and freight employees with respect to their total number and scenario S0 (status-quo). Figure 7 plots the shopping trip production and the delivery flow attraction. It emerges that currently the city center is mainly interested by light goods vehicle flows while the size of vehicle increases as distance from center raises.

Figure 7. Shopping trip production according to freight type (left side) and delivery flow attraction according to

vehicle type (right side)

According to the general strategies, the medium and large retail outlets and the warehouses were moved towards higher accessible areas, along major roads and junctions. Then, the different scenarios were assessed through aggregate indicators, such as vehicle-km and vehicle-hour for shopping trips and for freight distribution trips, grouped by vehicle type (e.g. cars, Light, Medium and Heavy Goods Vehicle). Table 1 reports an example of scenario assessment output where the results obtained for Scenario 3 (shifting retail and warehouse employees towards the fist ring in order to reduce distance between origin and destination of freight vehicle trips). As many outcome indicators are a function of vehicle-km, in the first instance, the total distance covered by cars and commercial vehicles was computed. Therefore, a reduction of truck-kms more than 7% with a reduction of car-kms of about 13% was obtained because of in the first ring the higher density of residents was revealed and the future facility location allows to reduce the distance between residential (consumption) and shopping places.

Table 1. Example of scenario assessment output.

S3 vi S0

freight distribution A% freight distribution LGV vehicle-km -7.6%

n. vehicle 4.7%

A% freight distribution MGV vehicle-km -6.7%

n. vehicle 0.0%

A% freight distribution HGV vehicle-km -5.7%

n. vehicle -1.9%

A% freight distribution vehicle-km -7.4%

n. vehicle -1.6%

shopping A% car shopping trips vehicle-km -12.8%

n. vehicle 0.2%

Total A% equivalent vehicle-km -12.7%

LGV: Light Goods Vehicles; MGV: Medium Goods Vehicles; HGV: High Goods Vehicle

The pollutant emissions were also estimated for all identified scenarios. The strategy for locating freight activity, according to the results of the application, seems to favor the clustering of freight distribution centers and of medium-size retail businesses in the first urban ring. This distribution could have a win-win positive effect both on the reduction of freight distribution vehicle-km and shopping trips - km made by car. This implies a reduction of total freight distribution travel time, on total shopping travel time and on pollutant emissions.

Table 2. Example of scenario assessment output: pollutant emission comparison

S3 vs S0

freight distribution LGV and MGV -8.0%

CO2 emission HGV -7.3%

shopping passenger vehicles -13.6%

total -13.5%

freight distribution LGV and MGV -8.0%

PM10 emission HGV -6.5%

shopping passenger vehicles -12.1%

total -11.9%

freight distribution LGV and MGV -8.3%

PM2.5 HGV -6.8%

emission shopping passenger vehicles -12.6%

total -12.3%

freight distribution LGV and MGV -7.8%

NOx HGV -7.3%

emission shopping passenger vehicles -13.2%

total -13.0%

freight distribution LGV and MGV -5.6%

VOC HGV -6.5%

emission shopping passenger vehicles -13.3%

total -13.2%

Total external costs due to pollutant emissions* -13.1%

LGV: Light Goods Vehicles; MGV: Medium Goods Vehicles; HGV: Heavy Goods Vehicle; * for homogenization coefficients refer to MIT (2006)

4. Conclusions and further developments

The paper proposed an overview of decision support system developed for the analysis and simulation of urban freight systems and presented two application examples: analysis and simulation of current status and assessment of freight locations. The updated version of CLASS is one of the first tools that implements advanced behavioral modeling systems, and uses an integrated approach allowing to jointly measure impacts due to changes in shopping behaviors and shop restocking. Two test cases were proposed: analysis and simulation of urban freight transport, exante assessment of land-use planning strategies (i.e. location of retail and wholesale activities).

The analysis of current scenario allows to investigate the logistics and freight transport in relation to land use, freight restocking demand and supply, logistic profile and road network performances and impacts. In particular, CLASS computes different indicators for each of the above aspects of urban logistics and freight transport. Their comparison with target and benchmarking values allows to identify its level of services and the possible critical stages. The simulation is able to point out the relations existing among city logistics measures, decision-maker choice dimensions by using a multi-stage demand model and a discrete choice approach for each decision level.

The assessment of freight facility location allowed to assess the effects of end-consumer behavior and location of retail outlets and restocking centers upon urban freight mobility. A scenario analysis is hence briefly described and the outputs obtainable by CLASS are synthetized showing the potentiality of such a system for city planning and management. In the application of the proposed system of models, three scenarios, each one related to the distribution of retail outlets and freight distribution centers, were presented and simulated. The obtained results depend from the already well planned location of retail activities in Padua municipality and the presence of the City Port distribution center in the first ring. The research demonstrates that land-use planning and, in particular, spatial distribution of urban goods facilities (shops, warehouses etc.) can minimize, with more or less influences, transport costs and their environmental impacts both for freight distribution and for shopping. The proposed assessment methodology is able to evaluate land use scenarios, in terms of transportation costs, that can be used to measure economic, social and environmental target strategic objectives. Besides, CLASS implements new advanced shopping demand models that allow to point out the effects due to changes in shopping behavior and shop restocking.

Furthermore, the study offers a focus on the impacts of land use planning policies and in particular of the location choices of retail activities and freight distribution centers on end-consumer behavior and freight distribution flows. The proposed methodology jointly analyses the freight distribution and shopping mobility segments impacts. The variation of vehicle - km of the two segments components and the variation of equivalent vehicle - km demonstrate that the shopping mobility transport costs are higher than their freight restocking counterpart and that it is fundamental to investigate the two components at the same time. The integrated approach has the advantage of stressing the interrelations between the two elements and to jointly measure their impacts.

This DSS should be considered a response to urban policy needs and is a useful tool for urban policy-makers involved in designing urban freight measures: they have to deal with a large number of trucks and vans delivering goods in the urban area whilst preserving the economic viability of city businesses and also ensuring environmental sustainability. Therefore, it is a tool developed to support ex ante assessment to simulate goods movements and capture the effects due to urban freight transport measures on end-consumer and retailer behaviour.

Some extensions of this DSS are addressed to: define for each study area specific planning strategies according to their "logistic profile", to include the e-commerce, and its impacts on shopping mobility and shop restocking taking account of home deliveries.

References

Boerkamps, J.H.K., van Binsbergen, A.J., Bovy, P.H.L., 2000. Modeling behavioral aspects of urban freight movement in supply chains. In

Transportation Research Record 1725, 17-25.

Browne, M., Allen, J., Nemoto, T., Patier, D. and Visser, J., 2012. Reducing social and environmental impacts of urban freight transport: A review of some major cities. In Procedia - Social and Behavioral Sciences 39, Elsevier Ltd, 19 - 33.

Comi, A. and Rosati, L., 2013. CLASS: A City Logistics Analysis and Simulation Support System. In Procedia — Social and Behavioral Sciences 87, DOI: 10.1016/j.sbspro.2013.10.613, Elsevier Ltd, 321 - 337.

Crocco, F., De Marco, S., Iaquinta, P., and Mongelli, D. W., 2010. Freight transport in urban areas: an integrated system of models to simulate freight demand and passengers demand for purchase trips. In International Journal of Mathematical models and methods in applied sciences 4(4), 265-273.

Eggleston, S., Gorißen, N., Hassel, D., Hickman, A. J., Joumard, R., Rijkeboer, R., White, L., and Zierock, K. H., 2000. COPERT III: Computer Programme to Calculate Emissions from Road Transport. European Environment Agency.

Gentile, G. and Vigo, D., 2013. Movement generation and trip distribution for freight demand modelling applied to city logistics. In European Transport \ Trasporti Europei (2013) Issue 54, Paper n° 6, ISTIEE, Trieste, Italy.

Kawamura, H. and Sriraj, P. S., 2012. Effect of the Built Environment on Urban Freight Movement and Operations. In Proceeding oof the 91st Transportation Research Board Annual Meeting, Washington, D. C., USA.

Lohse, D., 2004. Travel Demand Modelling with Model EVA - Simultaneous Model for Trip Generation, Trip Distribution and Mode Choice. TU Dresden, Working paper.

Macario, R., Filipe, L. N., Martins, P.M., Ries, V., 2008. Elements for a master plan in urban logistics. In Innovations in City Logistics, E. Taniguchi and R. G. Thompson (eds), Nova Science Publishers, Hauppauge NY, 499-516.

Munuzuri, J. and Gonzalez-Feliu, J., 2013. Decision-making tools and procedures for City Logistics. An introduction to the Special Issue on decision support for urban logistics. In European Transport/ Trasporti Europei 54, paper 1, Trieste, Italy

Nuzzolo, A. and Comi, A., 2014a. Direct Effects of City Logistics Measures and Urban Freight Demand Models. In Sustainable Urban Logistics: Concepts, Methods and Information Systems, J. Gonzalez-Feliu, F. Semet, J. .L. Routhier (eds.), DOI: 10.1007/978-3-642-31788-0_11, Springer-Verlag Berlin Heidelberg, 211 - 226.

Nuzzolo, A. and Comi, A., 2014b. Urban freight demand forecasting: a mixed quantity/delivery/vehicle-based model. In Transportation Research PartE 65, DOI: 10.1016/j.tre.2013.12.014, Elsevier Ltd, 84-98.

Nuzzolo, A. and Comi, A., 2014c. Urban Freight Transport Policies in Rome: Lessons Learned and the Road Ahead. In Journal of Urbanism: International Research on Placemaking and Urban Sustainability, DOI: 10.1080/17549175.2014.884976, Taylor & Francis.

Nuzzolo, A. and Comi, A., 2014d. A system of models to forecast the effects of demographic changes on urban shop restocking. In Research in Transportation Business & Management 11, DOI: 10.1016/j.rtbm.2014.03.001, Elsevier, 142 - 151.

Nuzzolo, A., Coppola, P. and Comi, A., 2013a. Freight transport modeling: review and future challenges. In International Journal oof Transport Economics vol. XL (2), Fabrizio Serra Editore, Rome, 151 - 181.

Routhier, J.L. and Toilier, F., 2007. FRETURB V3, a policy oriented software of modelling urban goods movement. In Proceedings of the 11th World Conference on Transport Research, Berkeley CA.

Russo, F. and Comi, A., 2012. The Simulation of Shopping Trips at Urban Scale: Attraction Macro-Model. In Procedia — Social and Behavioral Sciences 39, E. Taniguchi and R. G. Thompson (eds.), DOI: 10.1016/j.sbspro.2012.03.116, Elsevier Ltd, 387-399.

Sanchez-Diaz, I., Holguin-Veras, J. and Wang, C., 2013. Assessing the role of land-use, network characteristics and spatial effects on freight trip attraction. In Proceedings of the 92nd TRB Annual Meeting, Transportation Research Board of the National Academies, Washington DC, USA.

Sonntag, H., 1985. A Computer Model of Urban Commercial Traffic. In Transport, Policy and Decision Making. 3 (2), 171-180.

Stathopoulos, A., Valeri, E. and Marcucci, E., 2012. Stakeholder reactions to urban freight policy innovation. In Journal oof Transport Geography 22, Elsevier, 34-45.

Taniguchi, E., Thompson R.G., Yamada T., 2012. Emerging techniques for enhancing the practical application of city logistics models. In

Procedia — Social and Behavioral Sciences 39, Elsevier Ltd, 3-18.