JOURNAL OF TRANSPORTATION SYSTEMS ENGINEERING AND INFORMATION TECHNOLOGY Volume 9, Issue 2, April 2009 Online English edition of the Chinese language journal

Cite this article as: J Transpn Sys Eng & IT, 2009, 9(2), 141 146.

Received date: Sep 7, 2008; Revised date: Nov 23, 2008; Accepted date: Nov 29, 2008 *Corresponding author. E-mail: zhou.yue@pbworld.com Copyright © 2009, China Association for Science and Technology. Electronic version published by Elsevier Limited. All rights reserved. DOI: 10.1016/S1570-6672(08)60060-4

RESEARCH PAPER

Pedestrian Simulation Modeling for World Expo 2010 Shanghai ZHOU Yue*, WANG Jiangyan, HUANG Di, SUN Shengyang Parsons Brinckerhoff Engineering Technology (Beijing) Co. Ltd, Beijing 100062, China

Abstract: It is expected that World Expo 2010 Shanghai will attract as many as 600,000 spectators in a normal peak day. The complexity of pedestrian circulations after the Expo’s opening will be a challenge to the Expo’s efficient operations. A plat in the Expo site was selected for a microscopic pedestrian simulation modeling study, consisting of multiple pavilions with high attractiveness, a variety of spectator service facilities, such as shopping centers, restaurants, and restrooms, and different kinds of pedestrian facilities. The Legion computer simulation package was used for the study. Simulation parameters were calibrated basedon statistical analyses on relevant field data sets. Characteristics of Expo spectators’ touring actions were presented. Through a comparative analysis on circulations of the Expo spectators and those of game spectators in a stadium and passengers in a metrostation, a modeling approach was proposed for the Expo spectators, which was different from the approach to modeling circulations of the game spectators and metro passengers. Simulation results were analyzed and interpreted from the perspective of pedestrianplanning and operations for the use of relevant authorities. The simulation study and results helped to prompt measures to improve the study site’s planning, engineering design, and operations from the pedestrian circulation perspective. The methodology can be applied to other desired areas of the Expo site with propagation of the preparation for the major event.

Key Words: World Expo 2010 Shanghai; pedestrian circulation; pedestrian simulation modeling; simulation study

1 Introduction

It is expected that World Expo 2010 Shanghai will have a spectator demand of 70,000,000 for the whole event duration[1]. It is also expected that the Expo will attract as many as 600,000 spectators in a normal peak day and 800,000 in an extreme peak day[1]. The Expo site will have multiple types and complex layout of pavilions and facilities of activity, transportation, landscape, and other supporting utilities. Moreover, the Expo spectators will have complicated and random actions during their tours. Therefore, it is desirable to conduct a simulation modeling study for the complex pedestrian circulations, at current planning and engineering design stage of the Expo site. The study should aim at examining the pedestrian circulations based on current site planning and engineering designs, identifying potential problems and inefficiencies, and helping to prompt measures to improve the pedestrian circulations and recommendations to operations of the pedestrian circulation system after the

Expo’s opening. The Expo site is composed of more than 40 plats containing

different pavilions and facilities. Spectators’ circulations within a plat are subject to the layout and design of the pavilions and facilities, composition of the pedestrian circulation system, nature and attractiveness of the pavilions, space for spectator actions and so forth. Moreover, taking into account that touring actions of the spectators are characterized by randomness, the spectator circulations within the plat will be very complicated. Although it would not be practical to conduct pedestrian simulation modeling study for all the plats of the Expo site, a typical plat was selected for the simulation study based on its current site planning and engineering designs.

2 Background

2.1 Description of study site The study site is a plat located in the Pudong part of the

Expo site. The plat consists of pavilions of multiple European

ZHOU Yue et al. / J Transpn Sys Eng & IT, 2009, 9(2), 141 146

countries that are believed to be of high attractiveness to the spectators. The plat also contains several shopping centers, restaurants, and restrooms, as well as greens and site maintenance facilities that are not accessible to the spectators. In addition, several piers of the Lupu Crossing are also enclosed in the plat. Puming Road, Beihuan Road, Xiying Road, and Changqing Road are to the north, south, west, and east of the plat, respectively. An elevated pedestrian walkway is hanged into east part of the plat and is connected to the plat ground by stairs and escalators. There is a major pedestrian corridor going through the plat in an approximately horizontal direction (west-east) and with pavilions distributed at both sides. The corridor connects the Xiying Road by its west end and the ground entrance of the elevated pedestrian walkway by its east end. The layout of the plat and surrounding roadways are illustrated by Fig. 1, where capitals A to H represent different pavilions and letters a to e represent spectator service facilities, such as shopping centers, restaurants, and restrooms. In general, the plat is a very typical site with high spectator demand, miscellaneous types of pavilions and spectator service facilities, diverse pedestrian circulation facilities, and complex plat layout. 2.2 Characteristics of Expo spectator tour actions

The Expo spectators have the following characteristics for their touring actions (refer to description of the study site presented earlier): (a) the spectators usually have multiple tour destinations after entering the plat and before leaving the plat; (b) compositions of spectators’ tour destinations can have a considerable amount of possibilities; (c) the spectators have plenty of choices for places to enter and leave the plat since the plat is an open area, in addition, there is no limitation on how many times they can enter and leave the plat. The characteristics of Expo spectators put their tour circulations in a good contrast to circulations of game spectators in a stadium and passengers in a metro station. A detailed comparative

analysis on circulations of Expo spectators and those of game spectators and metro passengers is given in Section 3.2 to derive modeling approach for Expo spectator circulations.

3 Simulation modeling study

3.1 Legion model and parameter calibration

A computer simulation package Legion was used to conduct the pedestrian simulation modeling for this study. The Legion is an advanced and well-developed microscopic pedestrian simulation platform, and it has been applied to assist in pedestrian planning, facility design, and operations for many large-scale sport events and transportation terminals[2]. The model behind the Legion software was developed by Still[3]. On the basis of the field data sets collected from worldwide locations, the Legion model was well calibrated and validated[4].

To make the Legion model more compatible and suitable for projects in China, the PB team conducted a series of pedestrian data collections in sites of Beijing Olympic Testing Games, metro stations in Beijing and Shanghai, and other kinds of places, and the team also performed statistical analyses on the collected data. On the basis of the statistical analyses, the Legion model was calibrated by the data sets collected in China. Through the calibration, important model parameters were determined and compared with values obtained by relevant studies conducted abroad. For example, it was found that the mean pedestrian speed on flat corridors of sites of the testing games and metro stations is 1.3 m/s, greater than 1.2 m/s recommended by the ITE (Institute of Transportation Engineers) of the US, and that the mean speeds of pedestrians moving upstairs and moving downstairs are 0.73 m/s and 0.77 m/s, respectively, which are comparable with results from surveys conducted abroad. As site of this study is an open area and contains flat ground and outdoor

Fig. 1 Layout of the study site and surrounding roadways

ZHOU Yue et al. / J Transpn Sys Eng & IT, 2009, 9(2), 141 146

020406080

100120140160180200

[0, 0.2) [0.4, 0.6) [0.8, 1.0) [1.2, 1.4) [1.6, 1.8) [2.0, +)

Speed interval (m/s)

Fig. 2 Speed distribution of pedestrian on flat ground of an open area

020406080

100120140160180

Speed interval (m/s)[0, 0.2) [0.4, 0.6) [0.8, 1.0) [1.2, 1.4) [1.6, 1.8) [2.0, +)

Fig. 3 Speed distribution of pedestrian moving upstairs

Fig. 4 Speed distribution of pedestrian moving downstairs

stairs, statistical analyses were performed on data sets obtained at an open area’s flat ground and outdoor stairs[5].The statistical analyses indicated that the mean pedestrian speed on the outdoor flat ground is 1.06 m/s, with a standard deviation of 0.22 m/s; the mean pedestrian speed moving upstairs at the outdoor stairs is 0.83 m/s, with a standard deviation of 0.21 m/s; the mean pedestrian speed moving downstairs is 0.80 m/s, with a standard deviation of 0.13m/s. The distributions of the pedestrian speeds of the outdoor flat ground, moving upstairs and moving downstairs on the stairs, are illustrated in Figs. 2, 3, and 4, respectively. 3.2 Analysis and modeling methodology

Because of the characteristics of the Expo spectators’ tour actions, the method used to model their circulations in the study site should be different from that used to model circulations of game spectators in a stadium and passengers in a metro station. To derive modeling approach for the Expo

spectator circulations, a discussion is provided on comparing circulations of the Expo spectators to those of the game spectators and metro passengers.

Game spectators enter the stadium via proper entrances according to their tickets and then move to correct levels and zones to view the games. The number of origin-destination pairs of their circulations is very limited, or, in other word, the number of possible combinations of stadium entrance and spectator zone is limited. Furthermore, game spectators normally do not have any major interim destinations after they enter the entrances and before they arrive at the spectator zones. Therefore, their circulations are relatively simple and clear to be identified. Metro passengers enter the station from their desired entrances, go through the security screening and ticket checking-in, and then arrive at correct platforms to wait for their desired trains. In this process, the security screening area and ticket checking-in gates can be viewed as important interim destinations. As all passengers need to “stop at” the same interim destinations and the number of ground entrances of the metro station is even smaller than the number of the stadium’s entrances, passenger circulations can be easily identified and the number of possible circulations is very limited.

In contrast, the Expo spectators have many more origins and destinations since they can enter and leave the plat at a plenty of possible locations, and they also have many more interim destinations since there are quite a few pavilions and spectator service facilities in their tour process. Furthermore, the number and composition of interim destinations differ from spectator to spectator and therefore have a large number of possibilities. Consequently, the spectator circulations are very hard to be completely identified due to the large number of possibilities. From this perspective, Expo spectator tour circulations can hardly been completely coded in the simulation model by simply following modeling approach for game spectators and metro passengers.

On the basis of the above-mentioned analysis, for this study, an alternative approach was proposed to model the large number of spectator circulations in a fairly realistic and practical way, taking advantage of features of the Legion model and the “random” nature of microscopic simulation. The alternative procedure taken by this study is essentially presented as follows: First, all desired paths in the study site that would be used by spectators to enter and leave the pavilions and service facilities were identified and coded. Then, at proper locations of the study site, “decision areas” were determined and coded into the model. The “decision area” was defined as within the area a spectator would be able to and willing to make decision of whether or not to enter an interim destination (pavilions and spectator service facilities) or which interim destination to enter in case of multiple ones. The “decision area” coded in the model selected a portion of

ZHOU Yue et al. / J Transpn Sys Eng & IT, 2009, 9(2), 141 146

Table Fruin pedestrian level-of-service scheme

Threshold (pers./m2)LOS

Walkway Step

A 0–0.31 0–0.54

B 0.31–0.43 0.54–0.72

C 0.43–0.72 0.72–1.08

D 0.72–1.08 1.08–1.54

E 1.08–2.17 1.54–2.69

F >2.17 >2.69

spectators passing the area and then assigned them to different interim destinations according to predefined proportions. All selected spectators were picked by the “decision area” randomly. Spectators who had finished their tours in a pavilion would return to the desired paths outside the pavilion to continue their tours, whereas those who decided not to enter a pavilion would step forward directly along the desired paths onto the next “decision area.” Furthermore, logical functions were coded into the “decision areas” as basic assumptions to enhance the reality of the model and are presented as follows: (a) a pedestrian would not enter a pavilion or service facility more than once; (b) when the number of spectators in a pavilion had reached its capacity, spectators who decided to take a tour would have to wait outside the pavilion in a predefined “waiting zone”; (c) once the number of spectators in the “waiting zone” had also exceeded the capacity, no more pedestrians would be allowed to join waiting. By properly programming these “decision areas” throughout the whole study site, modeling of the large number of possible spectator circulations was achieved in a fairly realistic and practical way. 3.3 Simulation results analysis

The simulation described spectator actions in the plat qualitatively and quantitatively. On the basis of the video outputs of the simulation model, the relevant authorities were able to directly observe spectator circulations in the plat and therefore to achieve qualitative overall knowledge of the

circulations. Quantitative simulation results included cumulative max pedestrian density map and cumulative high pedestrian density map. Pedestrian density is defined as number of pedestrians per square meter and is an important criterion in determining pedestrian level-of-service (LOS). This study used the LOS scheme developed by Fruin[6] which is most widely applied in the field of pedestrian planning. The Fruin LOS scheme is presented in Table.

The cumulative max pedestrian density map recorded the maximum pedestrian densities experienced by all locations in the study site for the whole simulation duration (Fig. 5). The map is fairly straightforward in that the darker the color is, the higher the pedestrian density is and vice versa. The map helped to identify the pedestrian congestion locations in the plat, i.e. bottlenecks of the pedestrian circulation system. The map also illustrated pedestrian circulation routes and their interactions in the major corridor of the plat. The simulation study was able to detect problems that could not be easily identified by ordinary analytical methods and to reveal the complexities and severities of problems that could often be underestimated or ignored by ordinary analytical methods. For example, as shown in the map, congestions were so severe in front of the entrances of pavilions E and F that the two congested areas had tremendously undermined effective pedestrian circulations in the major corridor. Ordinary analytical methods could hardly illustrate the severity of this problem. On the other hand, it was illustrated by the map that pedestrian interactions at the west and east sides of the pavilions were relatively minor. Correspondingly, it was recommended not to set entrances on the pavilions’ sides facing the major corridor, and if such design would not be avoided, to set spectator waiting zones at the pavilions’ west and east sides instead and then to guide the spectators to entrances from the waiting zones by proper separation devices. For another example, the map indicated that pedestrian interactions were very critical in the area surrounded by pavilions A, B, C, and D, and it was thus recommended not to plan any landscape structures in the area, such as a fountain, so as not to obstruct the heavy pedestrian circulations.

Fig. 5 Cumulative max pedestrian density of the study site

ZHOU Yue et al. / J Transpn Sys Eng & IT, 2009, 9(2), 141 146

Fig. 6 Cumulative high pedestrian density of the study site

Cumulative high pedestrian density map recorded cumulative durations of pedestrian densities higher than a predefined threshold experienced by all locations in the plat. In this study, a threshold of two persons per square meter was adopted. As in Fig. 6, straightforwardly, the darker the color is, the longer the high pedestrian density experience was and vice versa. The map helped to identify which locations in the study site would experience long cumulative duration of pedestrian congestions, to locate bottlenecks that would occur repeatedly or would not discharge easily and automatically. Using the two maps jointly helped to identify the most “problematic” part of the study site’s pedestrian circulation system, i.e. those locations with worst congestions and longest durations. The most effective measures to improve the worst situation were proposed accordingly and therefore an optimal use of limited resources (e.g. space, facilities, human power, funding, and so forth) was achieved. For example, results of the two maps indicated that congestion at the entrance of pavilion B was very critical and lasted for a long time. By examining the simulation video, it was found that the problem arose from the current setting of the direction of the pavilion’s entrance which resulted in critical speed drops of the entering spectators and therefore the formation of bottleneck. It was recommended to reprogram the entrance of the pavilion so that the spectators would get into the pavilion smoothly.

4 Conclusions

This paper presented a microscopic pedestrian simulation modeling study for a plat in the site of the World Expo 2010 Shanghai. The study analyzed characteristics of the Expo spectators’ tour actions. Parameters of the simulation model were calibrated based on statistical analysis on relevant field data sets. On the basis of the comparative analysis of the Expo spectators’ tour circulations and circulations of game spectators and metro passengers, a modeling approach for this study was proposed. Simulation results were presented and

analyzed from the perspective of pedestrian planning and operations, and the use of the results by relevant authorities was also introduced.

The simulation modeling study on the typical plat identified potential problems and inefficiencies of the pedestrian circulation system under current site plan and engineering designs, and consequently helped relevant authorities to improve the spectator circulations from perspectives of facility planning, engineering design, and pedestrian operations. Because of the representativeness of the study site, the methodology can be applied to other desired areas in the Expo site featuring notable spectator activities with the propagation of the preparation for this major event. This study can serve as an effective technical support for relevant tasks in the preparation and, furthermore, as the scientific basis for pedestrian operations after the Expo’s opening. Moreover, the study methodology can be applied to other large-scale events as a valuable reference for relevant pedestrian planning and simulation work.

References

[1] www.expo2010china.com, Accessed on November 15, 2008. [2] www.legion.com, Accessed on November 15, 2008. [3] Still G K. Crowd Dynamics, PhD Thesis of University of

Warwick, Coventry, United Kingdom, 2000.

[4] Berrou J L, Beecham J, Quaglia P, et al. Calibration and validation of the Legion simulation model using empirical data. Pedestrian and Evacuation Dynamics 2005, Springer Berlin Heidelberg, 2007, 167–181.

[5] Shi J G. Research on Pedestrian Characteristics in Special Events, PhD Thesis of Beijing University of Technology, Beijing, China, 2007.

[6] Fruin J J. Pedestrian and Planning Design. Metropolitan Association of Urban Designers and Environmental Planners, New York, United States, 1971. .