CHAPTER 3 STUDY METHODOLOGY

Transcription

1 29 CHAPTER 3 STUDY METHODOLOGY 3.1 INTRODUCTION The mode choice does not rely on a single parameter. It is associated with many parameters. In this chapter, the methodology for estimating the mode choice used in this study is explained in detail. The factors influencing the mode choice are first identified and then quantified. The parameters which have significant impact are analyzed in detail. The analysis is done for the parameters like land use activity, station characteristics, opinion survey and travel pattern. The methodology of this work is shown in Figure 3.1. Essentially, the steps involved are data collection, analysis, model building and forecasting. 3.2 STUDY AREA Chennai city is situated on the North-East end of Tamil Nadu on the coast of Bay of Bengal with a total area of Sq. Km. It lies between ' and ' of the latitude and ' and ' of the longitude on a sandy shelving breaker swept beach. It stretches nearly Km along the Bay coast from Thiruvanmiyur in the south to Thiruvottiyur in the north and runs inland in a rugged semi-circular fashion.

2 30 Objective Study Area Selection Land Use Study of MRTS Corridor Secondary Data 1. Demography 2. Employment. Data Collection Primary Data 1. Opinion survey 2. OD survey of Mass Transport passenger (bus, MRTS) 3. Frequency, Accessibility, Travel time, and Travel Cost of Mass Transport Systems(bus, MRTS). Study of operational characteristics of the Mass Transport System Demand Matrix of MRTS/ BUS Identifying Tangible and Non-Tangible parameters influencing Mode mode choice LOGIT Logit MODEL Model Model Development and validation Artificial Artificial Neural Neural Network (ANN) Model Network (ANN) Sensitivity Analysis Formulation of scenarios and analysis Findings and Conclusion Figure 3.1 Methodology Chart Chennai has three rail corridors such as southwestern corridor connecting Beach and Tambaram, western corridor from Beach to Avadi and northern corridor from Beach to Thiruvallur. In addition to these rail transport system, MTC operates 2990 buses on nearly 640 routes serving the travel demand of the people in the city. The MRTS rail corridor in Chennai has been

3 31 taken for the study purpose. The MRTS influence area (i.e. about 1.5 Km on either side) has been taken as the study area (Figure 3.2). Figure 3.2 MRTS Influence Area

4 32 The MRTS alignment passes through the Cooum and Adyar rivers and runs parallel to the Buckingham Canal for about 10 Km. There is a good scope for the passenger patronage to MRTS from the influence area which has mixed land use activity. 35 percent are residential, 25 percent are institutional and about 15 percent are open and vacant land, 12 percent are industrial and 13 percent are commercial. The total MRTS system was planned for Km in four stages; till date only two stages have been constructed, that is, Phase I from Park to Thirumailai and Phase II from Thirumailai to Velachery. The total length of MRTS in operation is only Km. The average interstation spacing between MRTS stations is about 1 Km, whereas the distance between Beach and Fort station, Light house and Thirumailai stations is about 1.71 Km. The study focuses only on the constructed portion of MRTS. The section of the line encompassing the first three stations - Beach, Fort and Park Town, is at a ground level; after Park Town the line is elevated. All the stations after Park Town - Chintadripet, Chepauk, Triplicane, Light House, Mandaveli, Greenways Road, Thirumailai, Kotturpuram, Kasthuribai Nagar, Indira Nagar, Thiruvanmiyur, Taramani and Perungudi are elevated and Velachery station is at ground level. 3.3 LAND USE DISPOSITION ALONG THE MRTS CORRIDOR The patronage to any transportation project is highly dependent on the land use disposition along the corridor. The patronage of MRTS is also attributed to the land use development along the corridor and the socioeconomic characteristics of the people living along the corridor. The land use along the corridor and the socioeconomic characteristics of the people are discussed in this section. The economically weaker section of the society encroached on vacant land available on the banks of Buckingham Canal. They have built

5 33 huts and temporary structures for living. The area of settlement by this group is called slum. Table 3.1 explains clearly the impact of slum population on MRTS ridership. It is evident that, if the slum population is higher (10-20 percent) there is less scope for MRTS ridership. In case of Thirumailai, the slum population is percent but, it has good ridership. It is because of high density residential population and slum is available at only on one side. Hence, the slum population has significant impact on reduction of MRTS ridership. Because of the encroachment, safety and hurdles in commercial development other people avoid using the MRTS. Table 3.1 Details of Population and Slum Population Station Population in numbers 2001 Slum Population in numbers Percentage of Slum Population MRTS Patronage in numbers Beach Fort Park Town Chintadripet Chepauk Triplicane Light House Thirumailai Mandaveli Greenways Road Kotturpuram Kasthuribai Nagar Indira Nagar Thiruvanmiyur Taramani - I Taramani II Velachery

6 34 Table 3.2 and Figure 3.3 give the classification of stations by population density. Table 3.3 gives the land-use breakup of MRTS corridor. Correlating Table 3.1, 3.2 and 3.3, it is very clear the stations Triplicane and Light House come under high density zones but have 20 percent of slum population. Hence, the ridership at these stations is very low. The stations like Chindaripet, Beach, Chepuk, Thirumailai and Mandaveli come under medium density zones and hold a moderate ridership. The stations like Fort, Park, Greenways Road, Kasturbanagar, Indiranagar, Thiruvanmiyur and Velachery come under low density zones but, due to availability of institutional and residential area theses stations hold good ridership. Thus stations with high density with good mix of residential and Institutional area will have scope for higher ridership. Table 3.2 Classification of Stations Based on Population Density in MRTS corridor Phase High Density Medium Density Low Density Zone Zone > 400 ppha Zone ppha < 200 ppha Phase I Triplicane (586) Chintadripet (294) Fort (195) Light House (517) Beach (287) Park Town (180) Chepauk (236) Thirumailai (370) Phase II Mandaveli (204) Greenways Road (158) Kotturpuram (127) Kasthuribai Nagar(120) Indira Nagar (154) Thiruvanmiyur (147) Taramani and Perungudi (77) Velachery (44)

7 35 From Table 3.3 and Figure 3.3 it is clear that both ends of MRTS are having institutional area and commercial area. The residential area is sandwiched between the two ends. Hence there is a good scope for trip generation and attraction along the MRTS corridor. Table 3.3 MRTS Station Wise Land Use break up in percentage Station Name Residential Land Use break up in percentage Commercial Institution Open Road Industrial Vacant Total in percentage Beach Fort Park Town Chintadripet Chepauk Triplicane Light House Thirumailai Mandaveli Greenways Road Kotturpuram Kasthuribai Nagar Indira Nagar Thiruvanmiyur Taramani Perungudi Velachery

8 36 Figure 3.3 Population Density along the MRTS Corridor 3.4 MASS TRANSPORT SYSTEMS OPERATED IN THE STUDY AREA MRTS and Bus transport system are the two major mass transport systems operated in the study area. The operational aspects and ridership of the systems are described in this section MRTS Characteristics MRTS operation between Chennai Beach and Chepauk stations was unveiled to commuters in The stretch was built with an investment of million Indian Rupees (INR). The demand for MRTS on implementation was about 600 per day, since MRTS operated for a small distance of 5.84 Km with six stations. On extending the MRTS up to Thirumailai in 1997 with an additional cost of million INR the

9 37 demand gradually picked up from 600 to 9000 over a period of 7 years. In 2004 the MRTS was further extended upto Thiruvanmiyur with an additional cost of 6840 million INR. Because of the extension the demand increased from 9000 to and over three years it reached up to On extending the MRTS section up to Velachery in 2008 again there was rise in the demand and it is up to per day on an average. On studying the growth of demand from (Figure 3.4) it is evident that the demand increased in relation with the length of operation. After commissioning of Phase II MRTS showed remarkable increase in the ridership. Figure 3.5 shows the trend of average MRTS passenger loading per day from the year 1995 to 2008 (March). It is evident that Beach and Thirumailai station has average patronage of Beach is the only station which has the maximum patronage of 5469 per day and next to Thirumaili station the Velachery station which has the average patronage of 3656 per day. MRTS Passenger Demand per Day in Numbers Year Figure 3.4 MRTS Passenger Demand per Day from 1995 to 2008

10 38 MRTS Passenger Demand per Day in Numbers Beach Fort Park Town Chintadripet Chepauk Triplicane Light House Thirumailai MRTS Stations Mandaveli Greenways Road Kotturpuram Kasthuribai Nagar Indira Nagar Thiruvanmiyur Taramani Perungudi Velachery Figure 3.5 Average Passenger Loading at MRTS Stations The other stations like Fort, Park, Chintadripet, Triplicane, Light House, Mandaveli, Greenways Road, Kotturpuram, Kasthuribai Nagar, and Indiranagar have average passenger loading of per day. The total ridership on MRTS ranges from per day. Table 3.4 gives the average loading at each MRTS station as on March Table 3.4 Patronage at MRTS Stations as on March 2008 Name of the station MRTS Demand per day in Numbers Name of the station MRTS Demand per day in Numbers Beach 5469 Greenways Road 878 Fort 2955 Kotturpuram 825 Park Town 3156 Kasthuribai Nagar 825 Chintadripet 1191 Indira Nagar 1025 Chepauk 3346 Thiruvanmiyur 3246 Triplicane 1277 Taramani 350 Light House 2047 Perungudi 250 Thirumailai 5365 Velachery 3656 Mandaveli 585 Source: Railways Department

11 39 MRTS operates 100 services / day between Beach and Velachery and the cumulative running time from Beach to Velachery is 42 minutes at an average speed of 21 Kmph. The morning peak Period is identified between 8.00 AM AM and the evening peak period is identified between 5.00 PM 7.00 PM. The frequency varies from 2 trips per hour to 5 trips per hour. A survey of boarding and alighting passengers was carried out for 8 hours covering morning and evening peak hours to understand the current ridership and extent of line utilization. It showed that at present the system operates 5 transit trips during peak period. The maximum number of passenger trips observed during peak period is MRTS system is designed to carry 13, 00,000 trips for 15 hours, both ways but the total trips observed for 15 hours is trips both ways. Hence, the line utilization 2.76 percent. To estimate the present hourly variation in ridership in MRTS a survey on passenger boarding and alighting was carried out between 7:00 AM and 11:00 AM in the morning and 4:00 PM and 8:00 PM in the evening. At each station the total number of passengers boarding and alighting was surveyed and passenger trips per hour per direction were calculated. Figures 3.6 and Figure 3.7 below represent the same for morning and evening hours. The morning peak is observed between 8:00 AM and 10:00 AM and the evening peak is observed between 5:00 PM and 7:00 PM. 75 percent of passenger trips, south bound, is during 8:00 AM - 10:00.AM. The frequency of train is higher during this period and these trips should be mostly work trips. MRTS experiences directional patronage and Table 3.5 below clearly indicates that southbound traffic is heavy during morning hours. The directional split during the morning period is 68:32 (South: North). Table

12 40 shows that a morning passenger trip is and evening passenger trip is Peak period is observed between 8:00 AM and 10:00 AM and 5:00 PM and 7:00 PM. This value justifies that the evening passenger trips are also return work trips. Number of Passenger Trips Towards North Towards South AM AM AM AM Morning Period Figure 3.6 MRTS Passenger Loading Hour Wise Variation Morning Period Number of Passenger Trips Towards South Toward North PM PM PM Evening Period PM Figure 3.7 MRTS Passenger Loading Hour Wise Variation Evening Period

13 41 Table 3.5 Directional Split of MRTS Passengers Direction Morning Period Towards South Directional split during morning hours Evening Period Directional split during Evening hours Total Directional split for the entire day % % % Towards North % % % Total % % % Inter-station spacing refers to the distance between the two stations. It influences the journey speed of the transit between the stations. The sectional speed of MRTS is 72 Kmph. However, the current average operating speed is Kmph from Beach to Velachery. This is due to the presence of closely spaced stations. A station like Mandaveli has the least station load of 585 and it hardly contributes to passenger trips. Except Kasthuribai Nagar Thiruvanmiyur and Velachery station load in all the other stations in Phase II is less than Table 3.6 gives the inter-station spacing and the station load of all stations. In Phase 1 of MRTS, passenger movement among the MRTS stations is moderate, i.e. commuters move from Thirumailai to Chepauk, Beach for various purpose of trips. However, in Phase 2-commuter movement among the MRTS stations is completely nil. When stations are spaced so close such as 0.8 Km, the demand in MRTS has been very low.

14 42 Table 3.6 Inter Station Spacing of MRTS Origin Station Beach Fort Park Town Chintadripet Chepauk Triplicane Light House Thirumailai Mandaveli Greenways Road Kotturpuram Kasthuribai Nagar Indira Nagar Thiruvanmiyur Taramani Perungudi Destination station Fort Park Town Chintadripet Chepauk Triplicane Light House Thirumailai Mandaveli Greenways Road Kotturpuram Kasthuribai Nagar Indira Nagar Thiruvanmiyur Taramani Perungudi Velachery Inter station Distance in Km Bus Transport System The high capacity MRTS had been introduced along an alignment that has already well established Bus routes. To find the potential passengers to use MRTS, the assessment of total number of passengers making trips by bus along the MRTS influence area is presented. In addition, total number of buses operated parallel to MRTS is also reviewed in this section.

15 43 The Bus routes operating parallel to MRTS in the influence are identified based on the following criteria; Routes touching at least one MRTS station. Routes running within 1.5 Km from MRTS. Routes that originate beyond Velachery, Thiruvanmiyur and end the trips in MRTS influence area. Routes that originate beyond Beach station and end the trips in MRTS influence area. The identified routes taken up for analysis are shown in Figure 3.8. Parrys Corner, Thirumailai, Thiruvanmiyur and Velachery are the important nodes which operate buses parallel to MRTS. The list of bus routes, number of service and number of singles operated per day are shown in Table 3.7. About 195 services are operated along the MRTS corridor making 2882 trips per day. The over lap length of these buses and MRTS ranges from 5 Km to 17 Km. Routes operated parallel to MRTS were identified and total number passengers making trip per day in these routes are computed as based on the following criteria; Passengers traveling parallel to MRTS to a minimum distance of 4 Km. Passengers having origin beyond Velachery, Thiruvanmiyur and destination in MRTS influence area. Passengers having origin beyond Beach station and destination in MRTS influence area.

16 Figure 3.8 Bus Routes Running Parallel to MRTS Taken up for Study 44

17 45 Table 3.7 Passenger Movement and Bus Movement along MRTS Corridor ORIGIN Route No. Destination Number of Single/Day PP51/EXT Parrys Corner 68 Velachery 21L Parrys Corner 198 5G Thiruvanmiyur 128 1G Thiruvottriyur 36 5G Velachery 128 1C Ennore 56 Thiruvanmiyur I Thiruvottriyur 288 6E Tollgate 12 6D Tollgate Injambakkam 60 5 Parrys Corner 80 19B Soliganallur 48 Adyar 19K Sirucherri 32 21B Parrys Corner 40 21H Kelabakkam 72 M5 Soliganallur 48 M19 Injambakkam 80 3A Parrys Corner 22 Thirumailai / Chepauk 5K Taramani 144 M15 Medavakkam 48 45E Keelakattalai 48 3A Mylapore 22 5 Adayar 80 5C Taramani 84 18D Keelakattalai 44 18P Velachery Indiranagar 16 19E Kovalam 10 Parrys Corner 19G Kovalam G-CUT Srinivasapuram 28 PP19 Injambakkam 148 PP19 EXT Kovalam 74 21P Indiranagar 96 21H Kelambakkam L EXT Keelakattalai K Keelakattalai 84 Total 2822

18 DATA COLLECTION In this section, a detailed methodology is presented for conducting various primary surveys needed to quantify the factors influencing mode choice parameters like travel time, travel cost (mode wise) and accessibility to bus stops and MRTS stations. In addition, to understand the travel characteristics and travel pattern of the commuters the methodology to conduct opinion survey is also explained Opinion Survey The opinion survey helps one in identifying the actual desire of the people. It provides the base for developing any model for likelihood estimations. In this work the opinion survey was designed in such a way that it helps to identify the influencing factors for travel of the people by MRTS and bus. The opinion survey was conducted among the MRTS users, bus users and employees having offices along the MRTS corridor. The major objective of the opinion survey is to identify the reasons for not preferring the MRTS and desirable factors for switch over. To enable the process, a questionnaire was designed with multiple choice questions to collect the people opinions. Formats of questionnaire are shown in Appendix 2 and 3. In this work direct interview method was adopted with survey questionnaire to obtain the opinion. The persons to be interviewed were selected by random sampling method Estimation of Travel Time between Two Modes The travel time between two nodes includes the in vehicle time, dwell time and access time. The in-vehicle travel time is measured by using

19 47 the GPS. Conventionally the travel time was measured manually with stop watch. The major disadvantage of this method was the manual assumption involved and that may lead to mis-prediction and prone to error. Hence, GPS was used in this study. The probe vehicle and the route were selected first and the GPS was fitted to the probe vehicle. In this work the probe vehicles were MRTS and bus. The GPS will automatically record the data like altitude, speed, acceleration, latitude and longitude. Finally when the destination was reached the GPS recording was stopped. GPS Visualizer, Data Processing software were used to extract the data and the processed data is then used to compute the travel time precisely Dwell Time Survey The dwell time is the time that bus spends at stopping to serve the passengers at a specific stop. Levinson (1983) and Rajbhadri (2003) state that the dwell time can account for up to 26 percent of the total travel time. Hence a detailed study is carried out in this work. The survey location is selected along the MRTS corridor. Three persons were involved in this study, one person recorded the arrival time and departure time of buses, second person recorded the number of alighting persons and standing inside the bus and the third person recorded the number of boarding commuters in the bus. The collected data was processed and dwell time model was developed. The bus dwell time consists of the time needed for the following events to occur; 1) passenger boarding, 2) passenger alighting. If the number of passengers alighting and boarding are independent of each other, the dwell time was expressed by the sum of two random variables as shown in Equation 3.1. The marginal passenger alighting and boarding time per

20 48 passengers was assumed constant. The proposed model for estimating dwell time is given in Equation 3.1. α d = µ b + σ a + ρ s (3.1) where, α d = Bus dwell time in Seconds. µ b = Number of passengers boarding σ a = Number of Passengers alighting ρ s = Number of passengers standing inside the bus Accessibility Survey The accessibility survey is the direct appraisal survey done by a group of experts. The members assessed the status of accessibility level in terms of feeder service, approach road conditions, parking facility, bus stop locations, etc. The accessibility readings are analyzed and weightage is assigned for each category of options as discussed in section Accessibility plays an important role in any system. If a system is provided with excellent facilities and if there is no accessibility then it will discourage the passengers to use the system. Table 3.8 shows the various accessibility indices for the MRTS and bus system. Table 3.8 Accessibility Indices Attributes Nature of Attributes Weightage Assigned Feeder Service availability at Yes 1 MRTS Stations No 0 Location of Bus Stop Bus stop < 200m 2 from MRTS Stations Bus stop with in m 1 Bus stop with in >500m 0 Approach Road condition Good 1 Poor 0 Parking Facility availability Available 1 at MRTS Stations. Not Available 0

21 PROPOSED LOGIT MODEL FOR MODE CHOICE In this section the proposed Logit model for mode choice estimation is presented. The mathematical models derived from Random Utility Theory is the richest and tried extensively for the simulation of transport related choices and choices among discrete alternatives. Random Utility Theory is based on the hypothesis that every individual is a rational decision maker, maximizing utility relative to his/her choices. The underlying important assumptions are: An individual is faced with a finite set of choices from which only one can be chosen. Individuals belong to a homogenous population, act rationally, and possess perfect information and always select the option that maximizes their net personal utility. If C is defined as the universal choice set of discrete alternatives, and j the number of elements in C, then each member of the population has some subset of C as his or her choice set. Most decision-makers, however, have some subset C n, that is considerably smaller than C. It should be recognized that defining a subset C n, that is the feasible choice set for an individual is not a trivial task; however, it is assumed that it can be determined. Decision-makers are endowed with a subset of attributes x n X, all measured attributes relevant in the decision making process. In the case of only two alternatives A and B in a choice set (like Bus and MRTS) and if the alternatives have systematic utilities of say U BUS,

22 50 U MRTS, the measure of utilities U Bus, U MRTS is the function of travel time, travel cost and accessibility which may be expressed as in Equation 3.2 U Bus / MRTS = a 0 + a l x 1 + a 2.x 2 + a 3.x 2.a n.x n (3.2) where UC BUS : Utility measure of Bus transport UC MRTS : Utility measure of MRTS a 1, a 2, a 3, a n : Utility coefficients of Bus/ MRTS x 1, x 2, x 3, x n : Utility Parameters Based on the above utility equations, the proportion of travel demand is estimated using the multinomial Logit model as given in Equations 3.3 and 3.4. In this model e(i) s of choice utility function are all assumed to be independent and are identically distributed exponential function. P(BUS) P(MRTS) ( U Bus ) ( U Bus ) ( U M RTS ) (U MRTS ) (U Bus ) (UMRTS ) (3.3) (3.4) where, P(BUS) : Proportion of travel demand for Bus P(MRTS) : Proportion of travel demand for MRTS 3.7 PROPOSED ANN MODEL FOR MODE CHOICE In this study an ANN architecture which was never adopted before in literature is used to address some of the main drawbacks of existing ANN models, namely the unclear interpretation of parameters and the absence of an explicit utility function, which do not allow for elasticity analysis and, more

23 51 generally, make the generalization of the resulting models arguable. The proposed architecture separately specifies the function, assumed linear, between attributes and mode systematic utility. The function is modeled through a hidden layer between mode systematic utility and choice probabilities. This way, an explicit utility function can be specified, as commonly for Random Utility Models, the calibration of such structure allows an interpretation of input variables, while the relative weights allow to carry out an elasticity analysis. The proposed architecture contains four layers as shown in Figure 3.9; 0. Attribute (Input) layer, regarding relevant attributes, with one processing unit for each attributes 1. Utility layer, regarding systematic utility, with one processing unit for each transportation mode, 2. Hidden layer, regarding effect of utility on choices, with a number of processing units to be defined, 3. Probability (Output) layer, regarding choice probabilities, with one processing unit for each transportation mode. The whole model can be described by equations between the processing units of each pair of successive layers, as described below. Layer 0. Given the values of j- th attribute for each mode k: x kj calibration; the transfer function being the identity. Layer 1. For each processing unit k (one per mode) of the utility layer V k as in Equation 3.5 v k = j β k j x kj + ASA k (3.5) where the weights kj, and the biases ASA k are to be estimated. Layer 2. For a processing unit y n (their number being a design option) of the hidden layer, Equation 3.6

24 52 yn = φ ( k wnk vk + bn) (3.6) where the weights wnk, and the biases b n are to be estimated during the calibration; the transfer function φ( ) being a design option. Figure 3.9 Proposed Architecture of ANN Model Layer 3. For a processing unit Pm (one per mode) of the probability layer, Equation 3.7 Рm = Ψ ( n znk yn + cm) (3.7) where the weights znk, and the biases cn are to be estimated during the calibration; the transfer function φ( ) being a design option. Eventually, to compensate numerical errors, the output (probability) values Pm (non-negative in any case) are normalized so that their sum is equal to one.

25 SUMMARY In this chapter, a brief description of the methodology of the work carried out in this thesis is explained. Model development and data collection process were given more importance. The land use disposition along the corridor was analyzed and it was found that 35 percent are residential, 25 percent are institutional and about 15 percent are open and vacant land, 12 percent are industrial and 13 percent are commercial. The various travel characteristics of MRTS system was studied and it was found that Beach, Velachery and Thirumailai stations are the well-patronized stations. The travel pattern of the people was analyzed and it was found that passengers are moving along the MRTS corridor by bus and they are the potential passengers likely to get shifted to MRTS.