Integration of Street Vendors in Footpath Design Guidelines for an Indian City

Transcription

1 Integration of Street Vendors in Footpath Design Guidelines for an Indian City Ashish Verma 1, Shirin Mary Antony 2 1, 2 Dept of Civil Engineering, Indian Institute of Science (IISc) Bangalore, Bangalore , Karnataka, India 1 ashishv@civil.iisc.ernet.in; 1 rsashu@yahoo.com; 2 shirin.antony@gmail.com Abstract-Given the heterogeneous structure of Indian society and urban areas, street vendors are not only necessary but also inevitable on urban streetscape, as they provide services to all commuters with cheap and easily available goods. They have always been a characteristic feature of footpaths in Indian cities and towns. Therefore, instead of usual Indian approach of eliminating them from the pedestrian facilities, it is prudent to integrate them into the pedestrian policy and facility design guidelines. So far, there has been no attempt in India in this direction and any such pedestrian policy and design documents are silent in this aspect. The draft national policy on street vendors prepared recently in India argues that needs of this section are vital for urban planning purposes. Considering this background, this paper proposes a pedestrian micro-simulation based approach for preparing footpath design guidelines integrating street vendors in these facilities. The micro-simulation model was calibrated and validated using data collected on selected locations in Bangalore city. The model was then used for two purposes; (a) to demonstrate strategies to improve level-ofservice at the study locations in Bangalore with re-organized vendor spaces, (b) to develop general footpath design guidelines (integrating street vendors) for different footpath widths and pedestrian flows. The paper demonstrates the effectiveness of pedestrian micro-simulation approach to develop such guidelines for footpath design integrating street vendors. Keywords- Street Vendors; Pedestrian; Footpath; Design Guidelines; Bangalore; India I. INTRODUCTION Street vendors are broadly defined as persons who offer goods and services for sale to the public without having a permanent built-up structure but with a temporary static structure or mobile stall (or head load). They are called by different names in different parts of India such as hawker, pheriwalla, rehri-patri walla, footpath dukandars, sidewalk traders, etc. [1]. According to the survey done by National Association of Street vendors of India (SVI) in Bangalore, the average money spent in Indian Rupee (INR) per month by upper income group on street vendors is between 2,500 to 3,000 and lower income group with 800 a month. In Bangalore, 83% of the consumers prefer mobile vendors as it is convenient and provides fresh and cheap vegetables at their door steps [2]. Bangalore houses about 30,000 street vendors. The number of street vendors in the country takes up 2% of the total population of our country [1]. For every 30 to 50 persons or for every 6-10 families, there is one vendor [3]. Street vendors have been a characteristic feature of Indian cities and towns from time immemorial. However, there has been limited attempt in India to integrate them into any formal infrastructure policies, market design policy or pedestrian policy and facility design guidelines. The government of Indian State of Karnataka is keen to pursue a formal integration policy towards street vendors. Some of the triggers for this initiative are the Indian Supreme Court s judgment on Sodhan Singh versus New-Delhi Municipal Council (NDMC) case of 1989, and the Central Ministry of Housing and Poverty Alleviation policy of 2009 on street vendors, which followed the guidelines of the Supreme Court judgment. While the integration of street vendors is emphasized, it is equally important to ensure and improve the level-of-service (LOS) of facilities like footpath that are primarily meant for pedestrians. The importance and need for focusing on pedestrian facilities can be summarized as follows: Sustainable travel mode: Walking mode is one of the most sustainable modes of transport. National Urban Transport Policy (NUTP): The NUTP that was released in 2006 by the Ministry of Urban Development (MOUD), Govt. of India strongly promotes non-motorized transport. Considering NUTP as the guiding document, the government has been providing funds for improvement of pedestrian facilities through schemes like Jawaharlal Nehru National Urban Renewal Mission (JnNURM). Pedestrianization would encourage people to participate more in walking and adopt healthy lifestyle. Public transport system relies on supporting walking environment since the access and regress trips from a transit stop are predominantly made by the walk mode. Pedestrians are the most vulnerable section subjected to road safety hazards. The main reason is explained by the lack of walkways, cross walks etc. in Indian cities. In India, little attention has been devoted to integrated design of urban areas with due consideration for walking as a mode of movement. This needs construction of pedestrian facilities like exclusive walkways or footpaths, or marking of space on the road for their safe movement if no such dedicated facility can be provided along the roads. Due to inadequate exclusive

2 facilities available for their movement, there exists a constant conflict between the pedestrian and the vehicles in sharing the limited space available on a road. There are several types of frictions experienced by a pedestrian in mixed traffic conditions. These include friction with nearby pedestrians, parked vehicles, moving vehicles, and with road side developments. Considering the above background, a sponsored study was undertaken by the authors to integrate street vendors in pedestrian policy and facility design guidelines, with the case study of Bangalore city, the capital of state of Karnataka. The followings are the objectives of the study: 1. To understand the social aspect of presence of street vendors on pedestrian facilities. 2. To understand the pedestrian and road users perspective on the advantages/disadvantages of having street vendors on the pedestrian facilities in Bangalore city. 3. To understand the requirement of street vendors (type and extent) in different types of localities within the Bangalore city (office complex, commercial area, residential area etc.). 4. To develop improvement strategies on provision of space for street vendors on pedestrian facilities, taking case example of a specific location within Bangalore which demonstrates the basic amenities and infrastructure of a typical city. 5. To quantify the impact of provision of space for street vendors on pedestrian mobility and level-of-service, using pedestrian micro-simulation. 6. Finally, to develop pedestrian facility design guidelines for Bangalore city. While understanding of social aspects of street vendors is equally important, it may not be of interest to the readers of this journal and is therefore not included in this paper. This mainly involves conducting qualitative surveys to understand the social background and needs of street vendors and the opinions of road users and pedestrians about street vending and its usefulness. Based on these surveys, an attempt has been made to create policy guidelines regarding aspects including, type of vending in different areas, formal mechanisms to support street vending, space requirements for street vendors etc. This paper basically presents the work done to fulfill the objectives 4 to 6, as stated above i.e. developing pedestrian facility design guidelines using pedestrian micro-simulation. In India, the Indian Roads Congress (IRC) is the responsible body to recommend the guidelines on pedestrian facilities (footpaths) for an urban area (Table 1). Basically, IRC guideline no , gives the design recommendations on footpaths; however it is silent on the provisions for street vendors on footpaths [4]. Besides this, the other points observed from the IRC-103(1983) guidelines are as follows: 1. The widths of footpaths are recommended based on flow of pedestrians only. It does not consider the surrounding environment, the main obstructions like trees, electric poles, etc. 2. It is observed from the recommendations that widths are proportionate with the flows. But, in general, they do not follow a linear distribution. 3. There is no recommendation on how to evaluate pedestrian facilities like determination of level of service etc. TABLE 1 IRC RECOMMENDATIONS FOR WIDTH OF WALKWAY Width of Sidewalk (m) Capacity in number of persons per hour All in one direction In both directions A very recent draft revision of the above code (IRC: ) has been released in May 2012, in which the issue of street vendors has been addressed mainly in qualitative terms but it still lacks any quantitative guidelines on how they can be integrated in footpath design guidelines considering different LOS criteria[5]. The recommendations of the Highway Capacity Manual of Transportation Research Board, with regard to the quantitative analysis of sidewalks aim at facilitating the movement of pedestrians and accommodating a number of users of a given facility that is compatible with the desired LOS. Considering the above background, it is therefore important to take up such study for integrating street vendors in pedestrian facility design guidelines. II. PAST STUDIES This section reviews past studies on street vendors, pedestrian facility design and evaluation, and understanding and simulating pedestrian behaviour in various situations. The proposed methodology in this paper intends to build upon the understanding of existing literature in the related aspect of the proposed work. Since, the work focuses on integration of street vendors in footpath design guidelines, it is important to see what have been the focuses of past studies on street vendors with respect to policy and design, and what approaches (if any) have been used to integrate street vendors in footpath design. Since, pedestrian micro-simulation model has been used in this work, it is also important to study different levels of pedestrian behaviour models, including micro-simulation models reported earlier, and understand their advantages and disadvantages

3 A. Studies on Street Vendors Nduna [6] investigated the struggles between communities of street traders and the municipal authorities of Umtata, capital city of Transkei. The study spans the period Themes of concern in the paper are objections to street traders and the adoption of repressive policies towards them, the resistance of street traders and the changing attitudes of the Transkei government and Umtata local authorities towards the operations of hawkers. The paper by Smart [7] examines, through the medium of a case study, the social and economic effects of a government policy decision to establish permitted places where Hong Kong hawkers might trade legally. B. Studies on Pedestrian Facility Design and Evaluation Rakesh [8] studied the existing status of pedestrian facilities in the area under consideration and gave recommendations to promote planning of pedestrian facilities based on a qualitative evaluation. In the work by UTTIPEC [9] design guidelines were developed for Delhi comprising of goals for Integrated streets for Delhi, street hierarchy of Delhi, and design toolkit. Under the Pedestrian Transportation Program [10] Portland s Pedestrian Design Guide was developed to integrate the wide range of design criteria and practices into a coherent set of new standards and guidelines that, over time, promotes an environment conducive to walking. Throughout, the guidelines attempt was made to balance pedestrian needs with the design needs and constraints. Penna de Araujo [11] tested the application of a method for making a qualitative evaluation of pedestrian crossings, based on the methodology of Khisty [12]. The study identifies the performance measures such as, comfort, safety, system and continuity with their respective attributes including waiting time, space available while waiting to cross, number of pedestrians, one-way or two-way street, state of the road surface, road width, vehicle speed, visibility, lighting conditions, guardrails, absence of obstacles in vicinity, state of sidewalks, lowered kerb, pedestrian signals, central island, which may be utilized in the evaluation. Sahani and Bhuyan [13] carried out a study to define pedestrian LOS for urban off-street facilities in developing countries having heterogeneous traffic flow conditions. The LOS was developed using data from two Indian cities with population less than one million using Affinity Propagation cluster algorithm. C. Studies on Modeling Pedestrian Movement Previous research on pedestrians movement in urban environment is extensive and ranges from macroscopic pedestrian flow modelling to individual pedestrians behaviour (Asano et al. [14], Chattaraj et al. [15], Chattaraj et al. [16], Shah et al. [17], etc.). In order to model pedestrian movement, it is important to take into account the activity agenda of pedestrians and incorporate the interactions between pedestrians and their environment (roadway, traffic, street vendors, other obstacles and crowd). A complicated decision-making process is involved, in which pedestrians perceive and assess their environment, decide their strategy and adapt it accordingly, if necessary. However, pedestrians behaviour may not always be based on a simple stimulus-response process, but may also be strongly related to human factors. Moreover, unlike vehicles flows, pedestrian flows are characterized by a significant degree of randomness [18]. Most of the traditional pedestrian studies in the past have been carried out on a macroscopic level. Fruin [19, 20] first suggested the macroscopic pedestrian modelling which was followed by many researchers and was also adopted by TRB [21] and ITE [22]. The main concern of macroscopic pedestrian studies is space allocation for pedestrians in the pedestrian facilities. Pedestrian interactions are aggregated into macroscopic flow-speed-density equations. Using these sets of equations, the minimum requirement of pedestrian facilities is roughly estimated. The macroscopic approach, however, fails to explain certain observed phenomena that more efficient pedestrian flow can be reached with less space when a different set of rules regarding pedestrian movement etc. is used. While macroscopic approach is still widely used because of its simplicity, research on more comprehensive microscopic pedestrian model is emerging [23]. Microscopic pedestrian models are based on a detailed representation of space, individuals and consideration of personal abilities and characteristics [24]. It treats every pedestrian as an individual and the behaviour of pedestrian interaction is measured. Helbing and Molnar [25] explain how pedestrian motion can be described by a simple social force model for individual pedestrian behaviour. In this model, an individual is subject to long ranged forces and his dynamics follow the equation of motion, similar to Newtonian mechanics. Computer simulation of crowds of interacting pedestrians shows that the social force model is capable of describing the self-organization of several observed collective effects of pedestrian behaviour very realistically. However, the model implementations have lacked an origin-destination matrix, and so the models work best in situations where undirected exploratory movement is likely to occur. Also, the models rely on recalculated visibility graphs to store visual data from locations within the system; these graphs use a large amount of computer memory. Further, in the social force model, the pedestrian is considered as a point or a particle in a 2D environment. Borgers and Timmermans [26] formulated and tested a micro-level simulation model of pedestrian route choice and allocation behaviour within city centres. The model was developed to predict the likely effect of transportation plans and retail planning measures on pedestrian behaviour and hence on the profitability of shopping streets. The model was tested in the city of Maastricht. Dijkstra et al [27] developed a multi-agent system that can be used for visualizing simulated pedestrian activity and behaviour to support the assessment of design performance. This system is based on cellular automata and agent

4 technology. Agents represent objects or people with their own behaviour, moving over a pedestrian network. Each agent is located in a simulated space, based on the cellular automata grid. The system was applied for a T-junction with simple model and limited user-agents (pedestrians) with restricted behaviour. Even if the CA based approach is generally better understood than analytical models by experts in different application domains, and more easily applied to model related scenarios, both these approaches share the limit of considering individuals as homogenous entities, and generally do not provide elements of edibility and dynamism, like changes in behaviour of individuals. Further, Bandini et al [28] presented a Multi Agent Systems (MAS) approach to crowd modelling, based on the Situated Cellular Agents (SCA) model. This is a special class of Multilayered Multi Agent Situated System (MMASS), providing an explicit representation of spatial structures and different means of agent interaction. Heterogeneous agents are obtained through the definition of different agent types, specifying different behaviours and perceptive capabilities. The model is rooted on some basic principles of Cellular Automata (e.g. the definition of adjacency geometries), but also takes into account the autonomy of modelled entities, with their own internal architecture. Hoogendoorn et al [29] describes an approach to model individual route choice behaviour in continuous time and space under uncertainty. Numerical solution approaches were proposed to approximate the dynamic programming equation. The approach is applicable to predict path choice behaviour in infrastructure facilities of realistic size, as was shown by application of the approach to pedestrian behaviour modelling in Schiphol Plaza. This example was used to illustrate the effect of uncertainty on pedestrian choice behaviour. One of the main observations is that when uncertainty increases, pedestrians tend to prefer routes which give them wider berth. Hoogendoorn et al [30] contributed a new approach to determine microscopic pedestrian data from digital video measurements. The approach consists of six steps, including pedestrian detection and tracking, and was applied to data collected during walking experiments conducted at Delft University of Technology. It was found that the approach is suitable to detect all pedestrians in the detection zone, and track them with a high accuracy. False detections are identified by two performance measures. The newly developed detection and tracking techniques can also be applied to other types of video footage (using alternative detection techniques) and under different circumstances. As such, the collected data are either used to develop new theories and models, or will be used to calibrate and validate existing walker models. The work by Chae [31] aims to improve the understanding of and quantify the interactions between pedestrians and vehicles in the vicinity of single lane roundabouts. The focus was on the development of data extraction and simulation tools (ITRE-mv). The simulation model was calibrated and validated and was then extended to analyse and predict the effect of policy and technological treatments on the operational effectiveness of both vehicular and pedestrian traffic. William et al [32] examined the relationship between walking speed and pedestrian flow under various flow conditions and the bi-directional pedestrians flow effects for the signalized crosswalks in Hong Kong. Kretz et al [33] proposed a method to make agents in a microscopic simulation of pedestrian traffic walk approximately along a path of estimated minimal remaining travel time to their destination. Teknomo [23] developed a Microscopic Pedestrian Simulation Model (MPSM) of pedestrian movement where every pedestrian is treated as an individual. The results of the simulation have rejected the direct relationship assumption of space and flow in the macroscopic level. The paper by Hughes et al [34] describes the integrated use of modelling and visual simulation in efforts to provide preliminary estimates of treatment effectiveness. Verkehr In Städten - SIMulationsmodell (VISSIM) was adapted to study pedestrian-vehicle interactions at roundabouts and where photo-realistic Three Dimensional (3D) simulation was used in support of design efforts intended to improve access for visually impaired pedestrians at complex intersections. The example in the paper serves to demonstrate how the essential elements of pedestrian crossing behaviour can be modelled using operational field data and how this can be used to characterize the decision-making performance of different classes of pedestrians (those with normal vision and those who have pronounced visual impairments). Asano et al. [14] developed a microscopic model of pedestrian behaviour using a two-player game and assuming that pedestrians anticipate movements of other pedestrians so as to avoid colliding with them. The results of the simulation model were compared with experimental data and observed data in a railway station. Using experiments on single file pedestrian motion, Chattaraj et al. [16] pointed out differences in pedestrian flow in India and Germany due to cultural differences. The same was used to devise modification to the Blue-Adler model of pedestrian flow. Although there is plenty of literature available on pedestrian flow modelling, however as far as the study of street vendors and applications of these models are concerned, not much work has been done particularly on designing pedestrian walkways considering the street vendors. III. PROPOSED METHODOLOGY This section presents the methodology adopted for developing pedestrian facility design guidelines, integrating street vendors. As it has been observed in India, the elimination of the street vendors is impractical due to various social and economic reasons; therefore the only way to tackle the chaotic situation among the footpath users is to bring about a facility for the footpath users that co-exist with the street vendors. Therefore, the aim is to develop guidelines for providing walking facility for the pedestrians for a desired level of service with due provisions for street vendors. With this objective in view, a quantitative methodology is formulated to develop pedestrian facility design guidelines. Since, there is pedestrian-obstruction

5 (street vendor) interaction at micro-level, understanding the pedestrian behaviour in the presence of street vendors is essential for developing the design guidelines, therefore it is found prudent to use pedestrian micro-simulation approach for the study purpose. Fig. 1 presents the proposed methodology, which includes the following: Identification of study area (areas of peak pedestrian flow with street vendors along the walkway) Data collection of pedestrian movement using video graphic surveys Pedestrian studies and analysis Developing pedestrian micro-simulation model using VISSIM o Model validation and calibration o Modelling of various LOS scenarios Development of strategies to improve level-of-service at the study locations in Bangalore with re-organized vendor spaces. Development of general footpath design guidelines (integrating street vendors) for different footpath width and pedestrian flow The knowledge of the pedestrian traffic system mainly comes from observations and empirical studies. Pedestrian studies can be divided into pedestrian data collection and pedestrian analysis and simulation. The data collection consists of the task associated with the observation and recording of pedestrian movement data while pedestrian analysis is focused on the interpretation of the data in order to understand the observed situation and to plan and design improvements. The next section presents the case study. IV. CASE STUDY At least one study stretch from each of the three different commercial, industrial, and residential areas in the city of Bangalore, Karnataka State, India is selected for the study. The names of the locations are given below: COMMERCIAL AREA - Gandhi Nagar RESIDENTIAL AREA - Gandhi Bazaar, Madiwaala INDUSTRIAL AREA - Peenya 2nd Stage, Koramangala

6 IDENTIFICATION OF STUDY AREA DATA COLLECTION PEDESTRIAN STUDIES GEOMETRIC DETAILS OF THE STUDY AREA SOCIO ECONOMIC CHARACTERISTICS OF STREET VENDORS PEDESTRIAN ALYSIS PEDESTRIAN PEDESTRIAN SIMULATION SIMULATION USING USING VISSIM VISSIM MODEL ON THE REAL TIME DATA STRATEGIES FOR IMPROVED (LOS) AT STUDY LOCATIONS MODEL CALIBRATION AND VALIDATION ON REAL TIME DATA GENERAL DESIGN GUIDELINES FOR INTEGRATING STREET VENDORS IN THE PEDESTRIAN FACILITY MODEL OUTPUTS AND RESULTS Fig. 1 Proposed Methodology The selected locations are thickly populated with presence of street vendors and considerable pedestrian flow during the peak hours. Often the pedestrians are forced to use the carriageway as there is little or no space for them along the walkways owing to the presence of street vendors. In all the study locations, the formal shops co-exist with the informal street vendors. The stalls are very well patronized by shoppers. Consequently the remaining space is barely adequate for a person to walk. The footpaths are wide enough in the study locations, on an average about two and a half meters, for the pedestrian flow observed; but due to the presence of street vendors and impediments like lampposts, electricity junction boxes, telephone pillar boxes etc., the actual space available is less and varies considerably. In places like Gandhi bazaar (Fig. 2) and Madiwala in Bangalore, the major part of footpath width is occupied by the vendors. The vendors have even built temporary structures to set up stalls for vending. Due to all this, the pedestrians are often forced to occupy the carriageway for movement and the pedestrian-vehicle conflict is commonly observed in these areas. The next section discusses the data collection done in study locations

7 Walkway Journal of Intelligent Transportation and Urban Planning Jul. 2014, Vol. 2 Iss. 3, PP Fig. 2 Almost the whole of footpath occupied by vendors in Gandhi bazaar V. DATA COLLECTION Collection of data for modelling the pedestrian flow is an important step to ensure quality results. As in case of vehicular flow, pedestrian modelling also requires a high level of detail in data collection in order to be useful, including basic parameters like, volume, speed and density. As in the case of vehicular traffic where for calibration of microscopic models require detailed information on the interaction between vehicles to improve modelling of car-following and lane-changing behaviour, for microscopic pedestrian modelling also detailed information on microscopic behaviour of pedestrians and complete pedestrian trajectories is required. The data collection for the study area is classified into two: footpath inventory survey and pedestrian movement data collection using videography. The footpath inventory survey comprises of the geometric details of the study locations, such as width of the carriageway, width of the footpath, footpath surface condition (raised/continuous/smooth/cracks or bumps), total length of the study stretch, location of street vendors, length of stretch occupied by street vendors, area occupied by street vendors, land use on both sides of the road, area used for pedestrian movement etc. The pedestrian movement data for the analysis is collected using video graphic technique. A longitudinal trap of 15 meter (m) to 20 meter is made on the road by self-adhesive tape (for clear visibility) for measurement of volume, density and speed. A video camera is mounted at an elevated place so as to cover the entire test section (Fig. 3). The data for movements in both the directions are collected during evening peak hours on a typical weekday. Length of walkway Pedestrian Trap Width of walkway Fig. 3 Pedestrian data collection area

8 VI. EXTRACTING VIDEO DATA FOR PEDESTRIAN MICRO-SIMULATION The procedure for extracting the relevant data from the video files is based upon standard approaches to identify pedestrian speed, flow, and density. The concept involves marking out a rectangular box on the video monitor screen, set to approximate the dimensions of a virtual box on the ground of set breadth and length. The width of the foot paths in the sites selected varied depending on the dimensions of the walkway surveyed, but the data is standardized to match the Highway Capacity Manual (HCM) definitions by dividing the observed flow data by the width. The following sub-sections summarize the data extraction process. A. Determining Speed of Pedestrians Pedestrian speed is the average pedestrian walking speed, generally expressed in units of meters per second. A virtual longitudinal trap, the length (15 m) and width of which is already known is marked on the ground for the observation of data through videos. The following are the steps involved in extracting the speed data: 1. Select a random pedestrian about to enter the trap and track him/her through the trap length. 2. Note the pedestrian s entry and exit times in and out of the trap area (Figs. 4 and 5). 3. Pedestrian walking time is thus obtained by subtracting the time of entry into the rectangular box from the time of exit. 4. Walking speed is then derived by dividing the known length of the box by the walking time, previously calculated. From Figs. 4 and 5, the total time taken by the randomly chosen pedestrian to cross the trap length of 15m is 9 seconds (sec.). Thus the speed of this random pedestrian is recorded as 1.67 m/sec. The average speed of random pedestrians taken at different intervals of time is recorded as the pedestrian speed at that particular location. At least one-third of the total pedestrians were sampled at each location to obtain the average pedestrian speed. B. Determining Density Pedestrian density is the average number of pedestrians per unit of area within a walkway or queuing area, expressed as pedestrians per square meter. The following are the steps involved in extracting the pedestrian density data: 1. Rewind the tape back to when the test subject was in the middle of the rectangular box. 2. As the selected pedestrian is in the middle of the study area, pause the tape and count the total numbers of other pedestrians in the rectangular box with the selected subject. 3. Dividing the total number of pedestrians in the box obtained from the previous step by the area of the rectangular box gives the pedestrian density. Density of the area is obtained by counting the total number of pedestrians in the pedestrian trap (Fig. 6) and dividing it by the area of the pedestrian trap. Here, as shown in figure, the total number of pedestrians is 20. The area of the trap length is 69 m2. Thus the density of the study area was found to be 0.29 ped/m2. The average of densities recorded during the peak hour is reported as the density of the study area. At least 30 data points were collected during the study hour to obtain the average densities. Fig. 4 Pedestrian entry time

9 Fig. 5 Pedestrian exit time C. Determining Pedestrian Flow Rate Fig. 6 Pedestrian density Pedestrian flow rate is the number of pedestrians passing a point per unit time, expressed as pedestrians per 15 minutes or pedestrians per minute. The word point refers to a line of sight across the width of a walkway perpendicular to the pedestrian path. Pedestrian flow per unit of width is the average flow of pedestrians per unit of effective walkway width, expressed as pedestrians per minute per meter. The following are the steps involved in extracting the pedestrian flow rate data: 1. The volume of pedestrians crossing the trap for every 15 minute (min.) of the peak hour is counted manually. 2. The peak 15 min. volume is observed. Dividing it by the width of the footpath gives the flow per meter. This flow value per minutes gives the flow rate in pedestrians/min/m. The flow of the pedestrians is obtained by taking manual counts of the pedestrians entering and leaving the pedestrian trap for every 15 min. of the peak hour. The peak 15 min. volume is reported as the flow rate in pedestrians/min/m

10 D. Pedestrian Space Pedestrian space is the average area provided for each pedestrian in a walkway or queuing area, expressed in terms of square meters per pedestrian. This is the inverse of density, and often a more practical unit for the analysis of pedestrian facilities. Similarly, other parameters like dwell time are also determined for simulation purpose. For this study, the time spent by a pedestrian at a street vendor location (for availing services or buying goods) is called as the dwell time. VII. DATA ALYSIS AND FIELD OBSERVATIONS A quantitative assessment of the existing pedestrian facility is carried out after the data collection process. The data for pedestrian movements in both the directions is collected during evening peak hour on a typical weekday. The peak hour for pedestrian movement varies between 4:00 pm to 6:00 pm at different study locations. This data obtained from the field is then analysed to understand the suitability of the facility in terms of level of service measurement for the recorded pedestrian volume. Fig. 7 shows the prevailing condition of the footpaths in the field at Gandhi Nagar. Similar is the situation at other study locations as well. The geometric details of the study locations and the pedestrian characteristics after extraction from the videos are given in Table-2 and 3, respectively. Seen from Table 3, the pedestrian flow was observed to be highest in Gandhi Bazaar (44.96 ped./min/m) followed by Gandhi Nagar and Peenya. Since, there are no LOS guidelines specified in IRC codes, the HCM-2000 guidelines, as given in Table-4, are used in this study [35]. The pedestrian behaviour is modelled in this study using micro-simulation. The next section briefs some of the challenges in doing so. TABLE 2 GEOMETRIC DETAILS OF THE STUDY STRETCHES Gandhi Nagar Gandhi Bazaar Madiwala Koramangala Peenya Width of footpath in meter (m) Footpath Surface Raised Raised Raised Raised Raised Length of study stretch (m) Width of roadway (m) Width on carriageway used by pedestrians (m) Number of mobile vendors Time spent by mobile vendors at particular location (min.) TABLE 3 PEDESTRIAN CHARACTERISTICS FROM VIDEO SURVEY Study Location Speed (m/sec) Density (ped /sq.m) Flow Rate (ped./min/m) Dwell time min-max LOS Gandhi Nagar C Madiwala A Gandhi Bazaar D Koramangala B Peenya C VIII. MODELING PEDESTRIAN MOVEMENT To be able to develop and calibrate a microscopic pedestrian simulation model, a number of variables need to be considered. The parameters those are typically gathered and used for the simulation includes: geometric details of the study locations, the pedestrian volume per hour in both the directions, the obstacles along the walkway etc. However, specifically focusing on studying the effect of street vendors on pedestrian flow introduces additional complexity and requirements for modelling purpose, for example, the space occupied by the vendors along the walkway. Additional factors of pedestrian behaviour come into picture in the presence of street vendors that need to be captured and built in into the simulation model. If a pedestrian does not intend to get engaged in street vending then such spaces more-or-less acts as simple obstacles, however if they wish to get involved in street vending, then they have interaction with these spaces, in terms of stopping at these locations for seeking intended service or buying items. Based also on their personal characteristics and their familiarity with the area, they will have certain pattern of identifying the street vendors and route choice and will spend certain time at these street vending locations. The purpose of walking (shopping, commuter etc.) will also affect certain simulation parameters like, dwell time at vending location. The proportion of mobile vendors compared to stationary vendors at study locations will also affect the randomness level and ease of movement for pedestrians. Above all, these factors will have impacts on speed, flow, and

11 density of pedestrian movement. Considering the unique application of pedestrian micro-simulation model for studying street vending, there is nothing much in the past literature to guide on standard ways of capturing all these behavioural aspects and therefore authors have relied on their own understanding to develop ways of capturing them, which is through combination of field observation, questionnaire survey, video data, and understanding of the study locations. A. Simulation of Pedestrians Using VISSIM Fig. 7 Prevailing conditions on footpath at Gandhi Nagar The pedestrian micro-simulation tool available in VISSIM is used in this study. The movement of pedestrians in VISSIM is based on the Social Force Model [25]. The basic idea is to model the elementary impetus for motion with forces analogous to Newtonian mechanics. From the social, psychological, and physical forces a total force results, which then sums up to the entirely physical parameter acceleration. TABLE 4 PEDESTRIAN WALKWAY LOS BY HCM-2000 (ADAPTED FROM: TRANSPORTATION RESEARCH BOARD [35]) S.No. LOS Category Pedestrian Space (m 2 /p) Flow Rate (p/min/m) 1. A > B > > C > > D > > E > > F 0.75 Varies The forces which influence a pedestrian s motion are caused by his intention to reach the destination as well as by other pedestrians and obstacles. Thereby other pedestrians can have both an attractive and a repulsive influence [25]. This simulation model is validated in a threefold way: firstly, macroscopic parameters are calculated and compared to empirical data; secondly it is assured that microscopic effects like lane formation in counter flow situations and stripe formation in crossing flow situations are reproduced, and thirdly a realistic impression of resulting animations is in the focus. For pragmatic reasons, the behaviour of pedestrians is often classified hierarchically into three levels [29]: On the strategic level (minutes to hours), a pedestrian plans his route. He generates a list of destinations. On the tactical level (seconds to minutes), the pedestrian decides on the route between the destinations, making a rough routing decision. On the operational level (milliseconds to seconds), the actual movement is performed. This includes evading opposing persons, moving through a dense crowd, or simply continuing the movement toward the destination. In essence, the hierarchy of modelling pedestrian behaviour ranges from macro- to meso- to micro-simulation. The macroscopic models are generally based on traffic flow or queuing theory, or on fluid or continuum mechanics. Hunt and Griffiths [36] developed a macroscopic model for delay acceptance in pedestrians movement in relation to vehicles traffic volumes. Hughes [37] proposed a continuum theory for pedestrians flow in large crowds, in which the crowd is seen as an entity that behaves rationally under the aim at achieving immediate goals (rather than an overall goal) in minimum time. Daamen et al [38] calibrated the fundamental diagrams of traffic flow for pedestrian crowds inside and upstream bottlenecks, and proposed that a disaggregation of the crowd upstream the bottleneck into homogenous crowds may allow them to be described by fundamental diagrams. Early meso- and microscopic pedestrian models were mainly developed in Cellular Automata, where pedestrians move on a grid of cells; a set of rules defines the state/occupation of a cell in dependence of the

12 neighbourhood of the cell, and cell states in successive time steps are updated using a transition matrix. The works of Gipps and Marksjo [39], Borgers and Timmermans [26], Blue and Adler [40], Weifeng et al [41] etc. led to further advancements in pedestrian micro-simulation models. The Social Force Model [36] controls the operational level and parts of the tactical level, whereas the strategic level is defined by the user input. There are three types of evaluations possible in VISSIM; cross-sectional measurements, travel time measurements, and area-based measurements. Further, pedestrians can change their behaviour (e.g. preferred speed) with respect to space and time. The animation during simulation in VISSIM can be done using 2 & 3-dimensional (2D and 3D) animation as well as a LOS display based on HCM Also, the simulation can be recorded directly as AVI video animations or as ANI data for later evaluation. For editing support background images of various formats (e.g. bmp, dwg, dxf, ecw, jpg, png, shp, sid, tif, wmf) can be imported. B. Development of a Pedestrian Micro-Simulation Model In the present study, the pedestrian micro-simulation model is developed and used for three purposes : assessment of existing condition on study locations, improved and alternate LOS scenario for study locations, developing general footpath design guidelines (integrating street vendors) for footpaths of different widths and pedestrian flows. The existing pedestrian condition is simulated in VISSIM using the real time data obtained from the field data collected. The steps involved in development of the pedestrian simulation model in VISSIM include input and initialization, pedestrian generation and placement, running of simulation for the stipulated time, and evaluation of simulation results. The following points detail out some of the important aspects of building the pedestrian simulation model for the present study: 1. Run-time parameters Length of simulation period, time-steps, etc. Graphical output and animations. 2. Network topology Size of the walking area (length and width of footpath ) Location of special infrastructure (electric poles, trees, etc.), and preferred walking areas. Location and geometry of obstacles (space occupied by street vendors, any temporary structures used by vendors) Exact area occupied by each street vendor along the length of the study stretch Mobile vendors along the study stretch (assumed as obstacles) 3. Parameters describing walking behaviour Speed, density, flow, etc. Purpose of walking (shopping, commuter, etc.) Preferred distance to edge of footpath and obstacles. Space used for pedestrian movement (on footpath, on carriageway) 4. Activity scheduling and route choice parameters Origin areas (locations where pedestrians enter walking infrastructure) Activity areas (areas where pedestrians perform activities, such as buying a commodity from the vendors, waiting, etc.) Time spent by pedestrians at a particular vendor Activities (description of activities and the relevant activity areas) Activity patterns (sequences of activities) Route taken by the pedestrian along the study stretch (path taken from origin to destination with intermediate stops if any at the vendor stalls) 5. LOS is used as the parameter for evaluation of simulation results. The objective of model calibration was to obtain the best match possible between model performance estimates and the field measurements of performance. It may be noted that there are no universally accepted procedures for conducting calibration and validation process. The responsibility lies with the modeller to implement a suitable procedure which provides an acceptable level of confidence in the model results. Given the situation, focus was kept on parameters like, speed, density, flow, LOS, dwell time etc. that can be obtained from simulation as well as that can be practically measured in field. During VISSIM calibration key input parameters were altered based on the observations from the field and model outputs were compared against field data to determine if the output was within acceptable levels. It was found that not all default VISSIM input parameters represented study area conditions and needed to be adjusted to replicate reality. The routes followed by the pedestrians while waiting near a street vendor and time spent by them at a particular vendor were refined to give the desired output results. A set of aggregate parameters like speed, density, and LOS are observed from the video (Table 3) and are compared with the simulation model results (Table 5) for the purpose of validation. The calibrated and validated simulation model is then used for generating and evaluating various LOS scenarios for an improved pedestrian movement incorporating the street vendors in the pedestrian facility

13 IX. DEVELOPMENT OF STRATEGIES FOR IMPROVED LOS AT STUDY LOCATIONS The geometric data collected in the field and data from video graphic survey are analysed to determine the level of service of the pedestrian facility at which it is currently operating. An improved level of service is attained by modifying the existing layout of the footpath with proper re-orientation of street vendors and testing its effectiveness using the developed simulation models of each location. The obtained strategies below revealed that by maintaining the same width of the footpath as observed in the field, and proper rearrangement of street vendors more systematically and symmetrically, considerable improvement in the level of service (at least one level better) is obtained. The following sub-sections discuss the improvement strategies identified for specific study locations. TABLE 5 VALUES OF SPEED, DENSITY AND LOS OBTAINED FROM THE MODEL Study Location Speed (m/sec) Density (ped /m 2 ) LOS Gandhi Nagar C Madiwala A Gandhi Bazar D Koramangala B Peenya C A. Gandhi Nagar The total length of the study stretch at Gandhi Nagar is 187 m with a footpath width of 2.6 m. The level of service at which the facility is currently operating is found to be C. An improved level of service is observed when the street vendors are more systematically and symmetrically positioned along the footpath, without reducing their number. The recommended strategy for the study stretch along Gandhi Nagar showed an improvement in the existing pedestrian facility. Fig. 8 shows the strategy obtained for study location in Gandhi Nagar, the pink colour block indicates the street vendor stalls and the lighter shade indicates the spacing between the stalls which can be used for pedestrians to halt at a particular street vendor location. Landmarks: Anyamma temple, Towards Towards KG road Central complex, Venus lodge B. Gandhi Bazaar Fig. 8 Design specification-gandhi Nagar The length of the study stretch at Gandhi bazaar is 247 m and it has a width of 3.4 m. Fig. 9 shows the strategy obtained for study location in Gandhi Bazaar, which shows that proper geometric re-arrangement of the street vendor at this location has resulted in the improved level of service for the same number of street vendors. C. Koramangala The length of the stretch at Koramangala is 625 m and the width is 2.5 m. For every 200 m stretch, the arrangement of the street vendor stalls, as shown in Fig. 10, is to be provided in order to obtain a better level of service in this area. The darker blocks indicates the street vendor stalls and the lighter blocks the spacing for the pedestrians to halt when waiting to buy goods from the street vendors

14 D. Peenya For a stretch of 126 m length and 3 m width at Peenya, the strategy, as shown in Fig. 11, is to be provided to obtain a better level of service for the current pedestrian facility. The green blocks in the figure are the street vendor stalls and the blue blocks indicate the spacing between the hawkers that can be used as pedestrian halting areas while buying goods from vendors. Since the study location Madiwala is already operating at LOS A, no strategy is proposed for it. Table 6, summarizes the strategy obtained for each study location. It is expected that by adopting the proposed improvement strategy on ground for each study location mentioned above an improvement in pedestrian LOS by at least one level can be practically achieved at the study locations. Towards Ashram Towards K R road Landmarks: Pai Vinod deluxe hall, Citibank ATM, Bus stand Fig. 9 Design specification-gandhi Bazaar Sarjapur main road Towards Inner ring road and St Johns hospital Landmarks: Koramangala water tank, Union bank of India, Police public school, 1 st block Fig. 10 Design specification-koramangala Towards Hegganahalli Towards Peenya 2 nd stage Bus stand Fig. 11 Design specification-peenya X. GENERAL FOOTPATH DESIGN GUIDELINES After obtaining the specific strategies for each of the study location, the calibrated pedestrian micro-simulation model is further used to develop general pedestrian facility design guidelines integrating street vendors, which can be used in conjunction with the defined pedestrian policy (obtained from the initial part of this study) to improve the pedestrian LOS at any location or design new pedestrian facility at any location

15 Location Gandhi Nagar Gandhi Bazaar Koramangal a Peenya Existing LOS C D B C Improved LOS B C A B TABLE 6 STRATEGY FOR IMPROVED LOS AT STUDY LOCATIONS Hawker Stalls Length (m) x width (m) 6.8 x 0.8 hawker stalls 2 Nos.- bottom edge 1.5 x 2m spacing 6 Nos.-bottom edge 1.5 x 2m spacing 8 Nos.- top edge 2.5 x 2m spacing 5 Nos.-bottom edge 5.0 x 2m spacing 5 Nos.-top edge 5.0 x 3m spacing 4 Nos.-top edge 5.0 x 3m spacing 5 Nos.-bottom edge 25 x No.- bottom edge 8 x spacing 1No.- top edge 5.0 x 30m spacing 3 Nos.- top edge 1.5 x 0.7@ 2m spacing 4 Nos.-top edge 1.5 x 0.7@ 2m spacing 4 Nos.- bottom edge 1.5 x No. bottom edge 6 x 2 1 No. top edge Total No. of stalls along the stretch Length(m) x Width(m) x x Nos. for every 200m stretch 625 x x 3 For doing this, one of the basic inputs is the preferred space requirement for street vendors, which is obtained as 0.36 to 0.54 square meter (4 to 6 square feet). The general design guidelines are then developed for different footpath widths (as typically recommended in IRC design guideline) and for different LOS levels (as per HCM-2000). The idea is to define in the guideline, how many street vendors of a given size can be geometrically accommodated in every 20 m block length of footpath for different LOS levels. Since, each LOS level in HCM corresponds to a flow rate range, so it takes care of different pedestrian volumes. Also, proper geometric positions of street vendor stalls are used so that they can be easily incorporated and explained in general design guidelines, and which were obtained after simulating and testing different scenarios of geometric positions of the street vendors. So, for example, the authors suggest the arrangement of street vendors along the footpath have a width of 1.5 m corresponding to the flow rate for LOS A, B, C, D, E, and F and, similarly for footpath widths of 2, 2.5, 3, and 3.5 m (Fig. 12 to Fig. 16). Here, the flow rate that cannot be accommodated for particular width of footpath is avoided. For example: 1.5 m width cannot accommodate a flow rate for LOS C, D, E and F, while also integrating street vendors, hence it is excluded. Table 7 summarizes the general design guidelines prepared for pedestrian facility, integrating street vendors

16 TABLE 7 GENERAL FOOTPATH DESIGN GUIDELINES (INTEGRATING STREET VENDORS) FOR FOOTPATH BLOCK LENGTH OF 20 M AND FOR DIFFERENT LOS Width of footpath in m 1.5 LOS A B C Hawker stall size and positions for 20m length 7.5m from either ends of the footpath- 1.2m x 2.5m spacing No. of Hawker stalls 2- bottom edge D E A 2.0 B C 6.5m from either edge of the footpath- 2 x 3m spacing hawker stalls 2- top edge D E A 3.0 B C a) 8.5 m from top left edge of foot path- 1.5m x 0.5m hawker stalls b)6.5m from either bottom edges with 1.3 x 4.4m spacing hawker stalls 7.0m from top left edge and bottom right edge of the footpath- 1.3 x 0.4 hawker stall 1- top edge 2- bottom edge Total 3 1-top edge 1-Bottom edge Total 2 D E A B 3.5 C 6.5m from either edge of the footpath- 2 x 3m spacing hawker stalls 2- top edge D E 7.5m from left top edge of the footpath 2.0x 0.5m hawker stall 1- top edge Fig. 12 Width of footpath- 1.5m, Flow Rate: > 16-23p/min/m, LOS B

17 Fig. 13 Width of footpath- 3.0m, Flow Rate: > 16-23p/min/m LOS B Fig. 14 Width of footpath- 3.0m, Flow Rate: > p/min/m LOS C Fig. 15 Width of footpath- 3.5m, Flow Rate: > p/min/m LOS C Fig. 16 Width of footpath- 3.5m, Flow Rate: > p/min/m, LOS D Based on the assessment of pedestrian flow and density at an existing or a new location and the LOS level that the local civic bodies, who are responsible for providing footpaths, have decided to provide, they can decide a particular width of footpath, and number and arrangement of street vendors on it. For designing footpaths at new locations, the pedestrian flows