an italian project for exposure reduction in an urban area...

21
©Association for European Transport and contributors 2006 AN ITALIAN PROJECT FOR EXPOSURE REDUCTION IN AN URBAN AREA: EXPERIMENTATION DESIGN AND DSS DEVELOPMENT Massimo Di Gangi University of Basilicata, DAPIT, Potenza (Italy) - [email protected] Giuseppe Musolino, Corrado Rindone, Antonino Vitetta Mediterranea University of Reggio Calabria, DIMET, Reggio Calabria (Italy) [email protected], [email protected], [email protected] 1. INTRODUCTION Social risk can be defined as a cardinal measure of potential economic loss, human injury or environmental damage in terms of both emergency event probability and the magnitude of the loss, injury or damage (CCPS, 1995). Risk (R) can be expressed as the product of probability (P) that an emergency event occurs and magnitude (M) of the consequences in the system, defined as the product of vulnerability V and exposure (N). V is expressed in terms of the resistance of the infrastructures when the emergency event occurs; N can be defined as the equivalent homogeneous weighted value of people, goods and infrastructures affected during and after the event. Two types of measurements for risk reduction may be defined (Fig. 1) (Russo and Vitetta, 2006): prevention, which consists in reducing the level of P; protection, which consists in reducing the level of M. Prevention is possible only for some kinds of events which occur in relation to human activities (power failure, radiation leak, hazardous freight, etc.). Protection can be reduced with two classes of measures (Fig. 1): resistance, which aims to abate the level of V; evacuation, which does the same with the level of N. The former entails an increase in the ability of the infrastructure (transport and otherwise) to withstand events; the latter consists in reducing the number of users and goods that can have negative effects when emergency events occur. Evacuation measures can be defined through transportation system analysis and planning activities. The literature on transportation planning in urban areas in emergency conditions is not extensive: there are only a few papers concerning emergencies in large areas when a nuclear event occurs (Goldblatt, 1993), in urban systems (Russo and Vitetta, 1996; 2003; 2004) and in buildings (Di Gangi et al., 2000). Models and procedures specified and calibrated for transportation system analysis in ordinary conditions (Ben-Akiva and Lerman, 1985; Sheffi, 1985; Ortuzar and Willumsen, 1994; Cascetta, 2001) cannot be directly applied in emergency conditions. Users move on the network in different behavioural conditions, and do not know system congestion and reliability in real time. Specific models and procedures have to be proposed.

Upload: others

Post on 18-Oct-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

©Association for European Transport and contributors 2006

AN ITALIAN PROJECT FOR EXPOSURE REDUCTION IN AN URBAN AREA: EXPERIMENTATION DESIGN

AND DSS DEVELOPMENT

Massimo Di Gangi University of Basilicata, DAPIT, Potenza (Italy) - [email protected]

Giuseppe Musolino, Corrado Rindone, Antonino Vitetta Mediterranea University of Reggio Calabria, DIMET, Reggio Calabria (Italy)

[email protected], [email protected], [email protected] 1. INTRODUCTION Social risk can be defined as a cardinal measure of potential economic loss, human injury or environmental damage in terms of both emergency event probability and the magnitude of the loss, injury or damage (CCPS, 1995). Risk (R) can be expressed as the product of probability (P) that an emergency event occurs and magnitude (M) of the consequences in the system, defined as the product of vulnerability V and exposure (N). V is expressed in terms of the resistance of the infrastructures when the emergency event occurs; N can be defined as the equivalent homogeneous weighted value of people, goods and infrastructures affected during and after the event. Two types of measurements for risk reduction may be defined (Fig. 1) (Russo and Vitetta, 2006): prevention, which consists in reducing the level of P; protection, which consists in reducing the level of M. Prevention is possible only for some kinds of events which occur in relation to human activities (power failure, radiation leak, hazardous freight, etc.). Protection can be reduced with two classes of measures (Fig. 1): resistance, which aims to abate the level of V; evacuation, which does the same with the level of N. The former entails an increase in the ability of the infrastructure (transport and otherwise) to withstand events; the latter consists in reducing the number of users and goods that can have negative effects when emergency events occur. Evacuation measures can be defined through transportation system analysis and planning activities. The literature on transportation planning in urban areas in emergency conditions is not extensive: there are only a few papers concerning emergencies in large areas when a nuclear event occurs (Goldblatt, 1993), in urban systems (Russo and Vitetta, 1996; 2003; 2004) and in buildings (Di Gangi et al., 2000). Models and procedures specified and calibrated for transportation system analysis in ordinary conditions (Ben-Akiva and Lerman, 1985; Sheffi, 1985; Ortuzar and Willumsen, 1994; Cascetta, 2001) cannot be directly applied in emergency conditions. Users move on the network in different behavioural conditions, and do not know system congestion and reliability in real time. Specific models and procedures have to be proposed.

Page 2: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

P (Frequency/Probability)

V (Vulnerability)

N (Exposure)

Curves with constant magnitude level M

M (Magnitudo)

Curves with constant risk level R

Fig. 1 - Measures for risk reduction (Russo and Vitetta, 2006).

A general procedure for risk reduction in terms of exposure has been developed by Russo and Vitetta (2004). It is composed by a system of models able to simulate transportation systems in emergency conditions. It allows to execute risk assessment (management) through evacuation planning (design) of population from an urban area, according to a “what if” (“what to”) approach (Vitetta and Velonà, 2003). Our paper presents the first results of a research project, entitled SICURO (within the framework of the EU-funded 2000-06 Regional Operative Plan of the Calabria Regional Authority), having as general objective risk reduction in urban areas through the definition and implementation of evacuation procedures, a Decision Support System (DSS) and guidelines for public administrations. In the context of the general objectives of SICURO, three main groups of actions are defined:

i. test models and procedures to assess the effects of action to reduce risk in terms of exposure;

ii. construct a Decision Support System for planning and managing an urban system in emergency conditions where models and procedures are implemented;

iii. provide public administrations, agencies and operators with guidelines for planning and managing an urban system in emergency conditions.

Actions (i) concern the specification and calibration of the system of models for transportation system analysis in an urban area in emergency conditions and the validation of the general procedure for risk reduction in terms of exposure. The above operations are preceded by an extended experimentation phase in order to assess evacuation characteristics in an

©Association for European Transport and contributors 2006

Page 3: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

©Association for European Transport and contributors 2006

urban area after an emergency event. Experimental data will be provided from official sources and from articulated field surveys. Actions (ii) concern the design and development of a Decision Support System for planning and managing an urban system in emergency conditions. The DSS will constitute a planning tool for reducing the consequences of a emergency event, allowing the design of transport supply and management of transport demand in emergency conditions. The DSS prototype will be tested through experimentation that will be conducted in the Municipality of Melito Porto Salvo (Italy). Actions (iii) aim to define guidelines for public administrations, agencies and operators, for planning and managing an urban system in emergency conditions. The paper is mainly focused on the following phases of the SICURO project: experimentation design and the DSS development and testing. The paper has the following structure. A description of the experimentation design phase is presented in section 2. The architecture of the DSS, with the implemented models and procedures, is reported in section 3. Some preliminary results and future steps concerning the experimental phase and the DSS development and testing are presented in section 4. 2. EXPERIMENTATION PHASE The SICURO project comprises an extended experimentation phase of population evacuation from an urban area, in order to construct a complete data-base which is required to specify and calibrate the system of models for transportation system analysis in an urban area in emergency conditions and validate the general procedure for risk reduction in terms of exposure. An experimentation plan was defined on part of an urban area of the Municipality of Melito Porto Salvo (Italy). The plan is articulated in two parts: evacuation area and data detection. The characteristics of the evacuation area are reported in section 4.1. The following elements concerning data detection are presented below: identification of time phases, definition of variables, detection methods and tools. Data to be detected are classified as follows: • data related to transport supply and demand; • data detected in different temporal periods, with reference to the event

taking place: o pre-event, data connected to buildings and road links, that do not

modify their characteristics in the immediacy of, and during, the event; o immediacy of the event, data connected to the position of vehicles on

links and to their functional characteristics and data related to individuals who potentially evacuate;

o during the event, data connected to the road transport system and activity chains of individuals;

Data connected to supply (Tab. 1) are related to: a. building characteristics inside the evacuation area, subdivided according to

Page 4: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

©Association for European Transport and contributors 2006

their use (residential, public, commercial and mixed); b. road link characteristics. Data (a) and (b) will be detected through a direct survey and provided by the municipal registry office in a pre-event situation, not modifying their characteristics in the immediacy of, and during, the event. Demand data (Tab. 2) are related to: a. number and characteristics of individuals present in the different types of

buildings: residents and households (residential buildings), employees and customers (public and commercial buildings);

b. number and position of vehicles on road links; c. traffic flow variables related to predefined sections of road links and

progressive time intervals; d. activity chain connected with the evacuation of each individual/vehicle,

subdivided into: reaction activity (pedestrian egress from the buildings and vehicle boarding to undertake the motorized trip) and evacuation activity (performed on vehicles from the boarding place to the final destination).

Data (c) will be detected in the pre-event situation, provided from census data (ISTAT, 2001) and by the municipal registry office, and in the immediacy of the event by means of a direct survey. Data (d) will be detected in the immediacy of, and during, the event, and data (e) and (f) will be detected during the event by means of direct survey. A direct survey will be conducted, with the support of local administration and the police, by a group of skilled detectors and manual and automatic measurement tools. Fig. 2 shows the location of video cameras and automatic traffic detectors in the evacuation area. 3. DSS ARCHITECTURE A general procedure for risk reduction in terms of exposure is presented in Russo and Vitetta (2004). It is subdivided into several inter-related components, where each modelling component is specified for emergency conditions: risk definition, transport supply analysis, transport demand analysis, transport supply-demand interaction analysis; transport system planning and design through a “what if” and “what to” approach (Vitetta and Velonà, 2003). Supply models are composed by two sub-models: network model for simulating transportation facilities available for evacuation; model for estimating supply system reliability in relation to exogenous causes. Demand models require the definition of the decisional-maker, which can be individual (user) or public (Prefect, Mayor, etc.). Demand models have a multi-step structure and simulate the dimensions of generation, departure time, distribution, modal split and route choice. Simulation of each dimension requires the definition of the set of alternatives considered by the decision-maker and the specification of the choice model (Ben Akiva and Lerman, 1985; Cascetta, 2001). Supply-demand interaction can be simulated through static or dynamic

Page 5: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

©Association for European Transport and contributors 2006

assignment models in relation to traffic flow hypotheses. If traffic flow is assumed stationary, assignment models are static and can be based on an equilibrium approach (stochastic vs deterministic) or a system optimum approach. If traffic flow is assumed non-stationary (Ben-Akiva and De Palma,1987; Cascetta, 1989; Cascetta and Cantarella 1990, 1993), assignment models are dynamic and they can be: macroscopic, when traffic flow is represented as a continuous fluid (as in the hydrodynamic theory) and outputs are aggregated; mesoscopic, when traffic flow is represented though a group user behaviour and outputs refer to groups of users; microscopic, when individual user behaviour is directly represented and outputs refer to single vehicle trajectories.

Tab. 1 - Supply-related variables to be detected Element Type Variable P/I/D(*) DM(**)

Residential

Location Number of floors Number of residential flats Area of residential flats Number of exits Type of stairs Elevator Garage

P P P P P P P P

S S S/M S/M S S S S

Public

Location Number of floors Area of public offices Number of exits Type of stairs Elevator Garage

P P P P P P P

S S S/M S S S S

Commercial

Location Number of floors Number of commercial units Area of commercial units Number of exits Type of stairs Elevator Garage

P P P P P P P P

S S S/M S/M S S S S

Supply

Building

Mixed

Location Number of floors Number of residential flats Area of residential flats Number of commercial units Surface of commercial units Surface of public offices Number of exits Type of stairs Elevator Garage

P P P P P P P P P P P

S S S/M S/M S/M S/M S/M S S S S

Page 6: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

©Association for European Transport and contributors 2006

Link Road

Initial node and final node Length Width Number of lanes Free-flow speed Capacity Parking areas

P P P P P P P

S S S S E E S

(*) P, Pre-event; I, Immediacy of the event, D, During the event. (**) DM, Detection Method; S, direct Survey; E, model Estimate; M, Municipal registry.

Page 7: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

©Association for European Transport and contributors 2006

Tab. 2 - Demand-related variables to be detected Element Type Variable P/I/D(*) DM(**)

Residential Number of residents Number of households

P/I P/I

C C/S

Public Number of employees Number of customers

P/I P/I

C/S E/S

Commercial Number of employees Number of customers

P/I P/I

C/S E/S Building

Mixed

Number of residents Number of households Number of employees in public offices Number of employees in commercial units Number of customers in public offices Number of customers in commercial units

P/I P/I P/I P/I P/I P/I

C C/S C/S C/S E/S E/S

Link Road

Number of parked vehicles right side Number of parked vehicles left side Time instant Initial distance from initial node Final distance from initial node Number of vehicles

I I D D D D

S S S S S S

Section Road

Link identification Time interval: initial instant Time interval: final instant Distance from initial node Traffic flow Temporal average speed

D D D D D D

S S S S A A

Reaction (individual)

Household/public office/commercial unit Age Sex Possession of driver’s licence Vehicle ownership Egress time from the building Location inside building and floor Decision time what to do after exit Mode choice for evacuation Arrival time at vehicle Boarding time on vehicle

D D D D D D D D D D D

S S S S S S S S S S S

Demand

Activitychain

Evacuation (vehicle)

Distance from initial node Number of doors Front and rear parked vehicles Vehicle accessibility for driver Occupancy for each vehicle Time for merging manoeuvre Arrival at intermediate road sections

D D D D D D D

S S S S S S S

(*) P, Pre-event; I, Immediacy of the event, D, During the event. (**) DM, Detection Method; C, Census data, S, direct Survey; E, model Estimate; M, Municipal registry; A, Automatic detection.

Page 8: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

©Association for European Transport and contributors 2006

Video camera

Traffic detector

Traffic closure

Evacuation area

Fig. 2 – Evacuation area and location of video cameras and traffic detectors

Page 9: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

©Association for European Transport and contributors 2006

Page 10: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

©Association for European Transport and contributors 2006

The above general procedure is implemented in a DSS for planning and managing an urban system in emergency conditions. It represents a planning tool for abating the consequences of a calamitous event, allowing the design of transport supply and management of transport demand in emergency conditions. The DDS has three main components:

• GUI (Graphic User Interface) module, that allows construction of the network model (pedestrian and road) and the demand model for the estimation of time-slice OD matrices;

• DTA (Dynamic Traffic Assignment) module, containing the mesoscopic dynamic traffic assignment procedure for simulating transport demand-supply interaction;

• OUTPUT module, that allows the collection and analysis of the results from the application of the DTA procedure.

The following sections contain the description of the GUI module and the supply and demand models (section 3.1), the description of the mesoscopic DTA procedure implemented in the DTA module (section 3.2) and the description of the OUTPUT module (section 3.3). 3.1. GUI module The GUI module of the DSS is articulated in three blocks (Fig. 3), where operations are performed to construct the network model and demand model and generate the input file for the DTA procedure. The blocks are conventionally called: • Supply, zone definition and socio-economic data acquisition; data

acquisition required for road network estimation through a supply model; pedestrian graph estimation;

• Demand, definition of origin-destination pairs and estimation of time-slice OD matrices through a system of demand models;

• Output, data from previous blocks are assembled and prepared for the DTA procedure.

Below we describe the supply model and the demand model as implemented in the DSS.

3.1.1. Supply model The supply model implemented in the DSS is a network model (pedestrian and road), consisting of a graph and cost functions. The graph comprises nodes and links (Fig. 4) and gives the possibility to simulate the activity chain connected with the evacuation of each individual, subdivided into reaction (pedestrian egress from buildings and vehicle boarding to undertake the motorized trip), evacuation (merging in the vehicular stream and travelling to the final destination).

Page 11: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

OD matrix estimation

Cartography import

Origin zone definition

Node definition

Link definition

Origin zone selection

Destination zone definition

O/D pair definition

Zonal centroid selection

Associated link selection

Pedestrian graph building

Characteristics acquisition

Characteristics acquisition

Characteristics acquisition

Data assembling

Supply

Legend

Manual operation

Manual acquisition Demand

Automatic elaboration

Conditioned elaboration Data for

DTA procedureOutput

Data

Fig 3 - GUI module Nodes are divided into three classes:

©Association for European Transport and contributors 2006

- centroid

origin, which identifies the beginning of the pedestrian trip and can be associated either to a census zone and to a single building;

Page 12: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

©Association for European Transport and contributors 2006

-

-

- - -

-

-

-

destination, which identifies the assembly centre for vehicles that have evacuated the area;

pedestrian exit, associated to each building exit and subdivided into three sub-classes related to different building uses:

private building exit, associated to residential building; public building exit, associated to public building; commercial unit exit, associated to commercial unit;

road origin of motorized trip, where individuals board vehicles to evacuate; it is the conjunction element between the pedestrian and road graph; merging, where vehicles merge with the vehicular stream on the road link; road, which simulates an intersection or significant geometrical and/or functional variations of a road segment.

Links are divided into two classes: pedestrian - egress, which simulates the egress from buildings and has the origin

node as the initial node and the exit node as the final node; - connection, which simulates the trip from the building exit to the place

where the vehicle is parked and has the exit node as the initial node and the node origin of the motorized trip as the final node;

road - merging, which simulates the merging of parked vehicles in the vehicle

stream of the road link and has a node origin of the motorized trip as the initial node and merging node as the final node;

- road, corresponding to connections between nodes allowed by the circulation scheme.

A general scheme of the above-described graph is presented in Fig. 4. Link cost (temporal) functions are of an aggregate type and can depend on attributes presented in Tab. 3.

Tab. 3 - Attributes of link cost functions Link C/NC(*) Attributes

Egress

C

Number of floors, area of public offices, number and area of residential flats, number and area of commercial units, number of exits, type of stairs, presence of elevator

Connection NC Travel time from building exit to parked vehicle, boarding time on vehicle

Merging C Width of the road link, traffic flow on the road link

Road C Length, width, number of lanes, free-flow speed, capacity, parking areas

(*) C/NC, Congested vs Non-Congested.

3.1.2. Demand model The system of demand models implemented in the DSS has a multi-step

Page 13: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

structure and simulates the dimensions of generation, departure time, distribution, modal split and route choice. In the sphere of random utility theory, the choice model for each dimension simulation from a given choice set K, assumes that an individual associates to each alternative k belonging to K, a perceived utility Uk which may be expressed as:

Uk = Vk + εk ∀ k ∈ K (1)

Legend origin egress link destination connection link private building exit node merging link public building exit node road link commercial unit exit node path to destination origin of motorized trip 1-2-3 pedestrian path merging node road node

1

2 3

©Association for European Transport and contributors 2006

Page 14: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

©Association for European Transport and contributors 2006

Fig. 4 - Pedestrian and road graph

where: Vk denotes the average, or systematic, utility of alternative k; εk denotes the random residual taking into account perception errors of decision maker as well as modelling approximations of the analyst. Different specifications can be used for term Vk, among which a linear specification is presented:

Vk = Σl αl Xk,l - Σm βm Yk,m (2)

where αl and βm are parameters greater than zero to calibrate; Xk,l are user utility attributes; Yk,m are user disutility attributes. Tab. 4 shows criteria for choice set generation and attributes for systematic utility specification in choice models.

Tab. 4 – Criteria for choice set and attributes for systematic utility Systematic utility Choice

dimension Criteria for choice

set generation Utility attributes Xk,l Disutility attributes Yk,m

Generation Emergency Pre-arranged

Residents, employees, customers, households, building areas

Departure time

Constrained Free

Accessibility, resident, employees, customers, households

Congestion

Distribution Constrained Free

Residents, employees, students

Distance, travel time

Modal split Vehicle, Pedestrian, Emergency services

Vehicle owner, available mode

Travel time

Route choice Exhaustive, Filtered Travel time 3.2. DTA module Some highlights concerning the main features of the DTA mesoscopic procedure implemented in the DSS are reported. An initial description of the whole structure of the model is presented in Di Gangi and Velonà (2003), and Di Gangi et al. (2003a).

3.2.1. Main hypotheses The approach used refers to discrete time intervals, which are supposed to be of constant amplitude (without any loss in terms of generality). Let δ be the amplitude of the generic interval t and τ the current time within the interval, τ∈[0, δ] . A packet of users P(h, k) can be defined once the followed path k (hence the origin/destination pair connected by path k) and the departure interval h have been identified. Flow conditions are considered homogeneous along a link and constant for the entire duration of each interval, and the link variables flow (q), density (ρ) and speed (v) within interval t are subject to the following relationship:

Page 15: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

©Association for European Transport and contributors 2006

-

qt = ρt·vt (3) Outflow characteristics are calculated at the end of the interval; if the amplitude of each interval is sufficiently short, they can be considered valid for the subsequent one. Thus, in order to better evaluate flow characteristics, time intervals can be subdivided into evaluation sub-intervals. Once outflow characteristics on a link are known for a generic interval (or sub-interval), the movement of a point (representative of a generic packet of users) can be traced on the link, depending on the definitions of the link model and on the movement rules adopted.

3.2.2. Demand model Let D(h, u) be the demand vector made up by the set of users belonging to class u which moves between the several origin/destination (in the following o/d) pairs leaving during interval h, and let K(i, u) be the set of paths connecting the i-th o/d pair followed by the users belonging to class u. A choice probability π(k) is associated to every path k ∈ K(i, u). Every departure interval h (of amplitude δ) is subdivided into n sub-intervals that, for the sake of simplicity, are supposed of equal length λ = δ/n, and let η be one of these sub-intervals. The generic packet P(η, k, u) will be composed by number x(η, k, u) of users, belonging to class u departing during the sub-interval η and following path k, defined by assuming as uniform the departure distribution within interval h.

3.2.3. Supply model The generic link of graph a of length La, is functionally divided into two segments by section S whose abscissa, xS

a, can assume values between 0 and La, as shown in Fig. 5. The part of the link with x ∈ [0, xS

a [ is named running segment, while the other part, with x ∈ [xS

a, La], is named queuing segment. Evaluation of the abscissa of section S is performed at the end of each sub-interval.

3.2.4. Point movements Let, at time τ of interval t, P(h, k) [with h ≤ t] be a packet of users represented by a point p located at abscissa x of link a belonging to path k. Let va

t be the current speed on the running segment during interval t, Qa the capacity of the final link section (capacity can depend on t) and ρa the maximum density on the link. − If x < xS

a, point p moves on the running segment with speed vat and, within

interval t, can reach at least abscissa xSa and then, the distance that can

be covered on running segment is given by: min{ xSa – x, (δ − τ ⋅ va

t }. If it occurs that xS

a – x < (δ − τ ⋅ vat, point p enters the running segment of link

a, at time τ’ = τ [ xSa – x ) / va

t]. If x ≥ xS

a, point p moves on the queuing segment; outflow on this segment is ruled by the capacity of the final section of link a. The queue length covered by point p until the end of the interval is given by δ = [(δ − τ ⋅ Qa] /

Page 16: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

ρa. If it occurs that x + δ > La, point p exits link a during interval t at time τ” = τ + [ (La – x ) ⋅ ρa ] / Qa .

La - xsa xs

a

queuing S running

Fig. 5 - Functional scheme of a link

3.2.5. Spill-back management Once point p reaches abscissa La before the end of the interval (either τ” < δ if xS

a < La or τ’ < δ if xSa = La), it is ascertained whether the length of the running

segment of link a+ following link a on path k is not null, that is xSa+ > 0. If so,

point p can enter link a+. Otherwise, it means that the whole length of link a+ is occupied by a queue and point p remains on link a until the queue length on link a+ is lower than the whole length of the link, that is for all the time until condition La+ - xs

a+ < La+ → xsa+ > 0 is not verified.

3.3. OUTPUT module The OUTPUT module implemented in the DSS allows the collection and analysis of the results from the application of the DTA procedure. Two sets of indicators can be defined: the pedestrian system and road transport system. The former set is not currently defined; the latter set has been defined and can be divided into three main classes (Di Gangi et al., 2003b):

©Association for European Transport and contributors 2006

link indicators, for each network link: vehicular flow on each link section, temporal average speed, spatial average speed; flow/capacity ratio at the exit link section, time profile of queued vehicles on the link; demand-supply interaction indicators, for global analysis of the road transport system: total vehicular travel distance on the network (Ltot), average travel distance per vehicle (La), total vehicular travel time on the network (Ttot), average travel time per vehicle (Ta), vehicular evacuation time or time at which the last vehicle evacuates from the network (tlve), vehicular average speed on the network (Va); evolution indicators, for the analysis of transport system evolution during evacuation: time profile of number of vehicles waiting to enter the road network (Nwait), time profile of number of vehicles on the network (Nnet), time profile of number of vehicles that reach each destination (Nsafe).

4. PRELIMINARY RESULTS

Page 17: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

©Association for European Transport and contributors 2006

Some preliminary results and future steps concerning the experimental phase and the DSS development and testing are presented in this section. 4.1. Experimental phase An experimentation plan was developed for the evacuation area defined within the Municipality of Melito Porto Salvo (Italy). Melito Porto Salvo is a town in the Province of Reggio Calabria, in the south of Italy. It lies on the Ionian coast, 20 km from Reggio Calabria. The municipality has an area of 35.33 km2, a resident population of 10,483 and 2,352 in work (ISTAT, 2001). The evacuation area (Fig. 2) is part of the central area of the town, where public offices (town hall, etc.), residential and commercial activities are located. It consists of four census zones and has an area of 0.702 km2 (1.98% of the total area in the municipality), 774 residents (7.38% of the municipality population) and 346 employed (14.7% of those in work). Tab. 5 presents some census data for the evacuation area. The plan is to simulate evacuation during one working day and detect all the presented variables concerning the conditions in the immediacy of the event and during the event. We are currently in the pre-event phase and we have detected all the related variables concerning socio-economic conditions and transport supply (facilities, services).

Tab. 5 - Census data for the evacuation area

Data Evacuation area

Area [km2] 0.702 Residents 774 Households 269 Employees 346 Buildings 165 Flats 370 Area of flats [km2] 0.327

4.2. DSS development and testing The DSS presented is currently a prototype. Several tests on the defined evacuation area were performed for its general validation, using preliminary specifications and calibrations for the system of models of supply, demand and dynamic assignment. Subsequent, more detailed specifications and calibrations of the system of models will be possible after the conclusion of the experimentation phase.

Page 18: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

©As

Below we present the simulation of evacuation activity of the population from the defined evacuation area, performed with vehicles. The definition of transport supply and demand, through road network and demand models, is described and some output data provided by the DTA procedure are reported. Simulation of reaction activity is yet to be performed. 4.2.1 Transport supply Definition of the supply model started with the zoning of the evacuation area. This phase has led to the identification of 7 origin zones and 2 destination zones (assembly centres). The road graph was then defined, consisting of 9 nodes (7 origin centroids and 2 destination centroids, located at the border of the evacuation area) and 10 links (Fig. 6). The cost functions related to road links were preliminarily specified and calibrated according to the geometric and functional characteristics of road facilities, on the basis of previous experience in similar urban areas (Musolino and Vitetta, 2003; Di Gangi et al., 2003). 4.2.2. Transport demand Transport demand flows are simulated through the application of the generation, distribution and route choice models. Departure time and modal split are not applied because, respectively, we are considering an emergency evacuation (individuals evacuate as soon as they receive the evacuation alarm) and we assume that individuals evacuate with private vehicles. The number of generated users to evacuate was estimated from census data (ISTAT, 2001) for the evacuation area. These data were distributed among the buildings of the evacuation area in proportion to their surface area. Data for each building were subsequently aggregated with reference to the 7 origin zones. 5

1

2

4 3

sociation for Eu

5

ro

6

pean Transport an

7

d contributors 2006

Page 19: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

©Association for European Transport and contributors 2006

-

-

Fig. 6 – Road graph in the evacuation area Demand flows OD were estimated by means of a multinomial Logit distribution model, that provided the distribution percentages of the demand generated in every origin (r) to the destinations (s), with reference to time period h, p[s/r, h]:

p [s/r,h] = exp (Vs) / Σ s’=1….n exp (Vs’) (4) where V s = - β X rs, (dis)utility associated to the choice of assembly centre s; X rs, free-flow minimum time on the network between origin r and assembly centre s. Two simulation scenarios were considered: Day, for morning peak hour; Night, for a generic night hour. Experimentation will be carried out in the diurnal period of a working day, so it will allow the system of models to be directly calibrated for the Day scenario. The generation of users to evacuate was carried out assuming that the following categories of users are present in the evacuation area:

Day: employees, residents under 5 and over 65 years old, housewives, unemployed, a given percentage of working residents; Night: total residents.

The distribution of generated users among assembly centres was carried out applying the model (4) with parameter β= 30, calibrated in aggregate way to take into account user behaviour in reaching the nearest assembly centres. 4.2.3. Output data Output data provided by the application of the DTA procedure concern the road transport system of the defined evacuation area inside the Municipality of Melito Porto Salvo. Some road transport system indicators provided by the OUTPUT module of the DDS are presented. Demand-supply interaction indicators for Day and Night scenarios are reported in Tab. 6. The Day scenario has lower vehicle travel times and vehicle evacuation times than those of the Night scenario, due to the larger number of vehicles generated in the Night scenario.

Tab. 6 – Demand-supply interaction indicators. Day and Night. Indicator Unit Day Night

Total vehicle travel distance on the network (Ltot) vehic*km 304.2 391.7 Average travel distance per vehicle (La) km 0.72 0.77

Page 20: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

©Association for European Transport and contributors 2006

Total vehicle travel time on the network (Ttot) vehic*min 2867.4 4367.7 Average travel time per vehicle (Ta) min 6.79 8.58 Vehicle evacuation time (tlve) min 19.0 24.0 Average vehicle speed on the network (Va) km/h 6.36 5.38

4.3. Future steps In the context of the SICURO research project, the following future steps have been planned and scheduled. Experimentation in the evacuation area, defined through the evacuation plan, will be performed in November 2006. The plan is to carry out the evacuation of the population in the evacuation area during the diurnal period of a working day: people will be notified of the event and will be asked to reach a pre-established assembly centre. All variables concerning the conditions in the immediacy of the event and during the event will be detected by field surveys performed with manual-automatic detection tools and video-cameras. After the conclusion of the experimentation phase, detected variables will be used for calibrating the system of models and validating the procedure implemented in the DSS. REFERENCES Ben Akiva, M. and Lerman, S., (1985) Discrete choice analysis: theory and

application to travel demand, MIT Press, Cambridge, Mass. Ben Akiva, M. and De Palma, A., (1987) Dynamic models of transportation

networks, Proceedings of the 15th PTRC Summer Annual Meeting, University of Bath, England.

CCPS, (1995) Guidelines for chemical transportation risk analysis, American Institute of Chemical Engineers, New York.

Cascetta, E., (1989) A stochastic process approach to the analysis of temporal dynamics in transportation networks, Transp. Research 23 B, 1-17.

Cascetta, E., (2001) Transportation systems engineering: theory and methods, Kluwer, Academic Press.

Cascetta, E. and Cantarella, G.E., (1990) A Day to Day and Within Day Dynamic Stochastic Assignment model, Transp. Research 25 A, 277-291.

Cascetta, E. and Cantarella, G.E., (1993) Modelling dynamics in transportation networks, Transportation Science.

Di Gangi, M., Luongo, A. and Polidoro, R., (2000) Una procedura di carico dinamico per la valutazione dei piani di evacuazione, Proceedings of 2nd Scientific Seminar on Methods and Technologies of Transportation Engineering, Università di Reggio Calabria, Franco Angeli.

Page 21: An italian project for exposure reduction in an urban area ...web.mit.edu/11.951/OldFiles/oldstuff/albacete/Other_Documents/Eur… · planning and managing an urban system in emergency

©Association for European Transport and contributors 2006

Di Gangi, M. and Velonà, P., (2003) Use of a mesoscopic dynamic assignment model for approaching the evolution of an urban transportation system in emergency conditions, In Environmental Health Risk II, WIT Press, pp. 227-236.

Di Gangi, M., Russo, F. and Vitetta A., (2003a). A mesoscopic method for evacuation simulation on passenger ships: models and algorithms. Pedestrian and Evacuation Dynamics 2003. Proceedings of the 2nd International Conference. E.R. Galea (ed.). CMS Press, London.

Di Gangi, M., Musolino, G., Russo, F., Velonà, P. and Vitetta, A., (2003b) Analysis and comparison of several urban road transportation assignment models in emergency conditions. In Environmental Health Risk II, Brebbia C. A. and Fayzieva D. editors, WIT Press, pp. 247-258.

Goldblatt, R., (1993) Development of Evacuation Time Estimates for the Davis Nuclear Power Station, T. R. 329, KLD publication.

ISTAT, (2001) Census data of the Province of Reggio Calabria. National Institute of Statistics, Rome.

Musolino, G. and Vitetta, A. (2003) Microsimulative approach for the evaluation of an urban transportation system in emergency conditions, In Environmental Health Risk II, WIT Press, pp. 237-246.

Ortuzar, J.deD. and Willumsen, L.G., (1994) Modelling transport, John Wiley and Sons, 2nd ed.

Russo, F. and Vitetta, A., (1996) The road network design problem to improve the safety during exogenous flow perturbations, Proceedings of the 29th ISATA, Florence, Italy, 1996.

Russo, F. and Vitetta, A., (2003) Urban transportation system in emergency conditions, In Environmental Health Risk II, WIT Press, pp. 207-216.

Russo, F. and Vitetta, A., (2004) Models for the evacuation analysis of an urban road transportation system in emergency conditions, Proceedings of ETC Conference 2004. Cambridge.

Russo, F. and Vitetta, A., (2006) A general risk model in transportation systems for disaster prevention, Proceedings of 11th International Conference on Travel Behaviour Research, Kyoto.

Sheffi, Y., (1985) Urban transportation networks, Prentice Hall, Englewood Cliff, NJ.

Vitetta, A. and Velonà, P., (2003) Evolution of an urban transportation system in emergency conditions: analysis through a pseudo-dynamic assignment model, In Environmental Health Risk II, WIT Press, pp. 217-226.