TRAFFIC FLOW MODELLING ON MULTILANE DIVIDED HIGHWAYS

Abstract

With the rapidly growing economy, industrial growth, increasing number of central business districts (CBDs) and special economic zones (SEZs) in India; the movement of traffic between cities, industrial centers and trading centers has increased tremendously due to increase in the number oftrips. As a result, the Government of India took up National Highway Development Program (NHDP) to provide high density and high speed corridorsespecially multilane highways (more than one lane for unidirectional traffic flow), to meet the present and future needs of growing traffic and to develop a rapid, efficient, safe, and well connected highway network. In the wake of these projects, the length of multilane highways which is 12% of the National highway network at present, is going to be significantly increased with the 4/6 laning ofthe existing highways. The mixed traffic behavior on multilane highways remains scarcely explored. Most of the studies on multilane traffic flow pertain to homogeneous traffic scenarios. These studies have very limited applications to mixed traffic conditions due to their incapability in satisfactorily explaining the complexities of mixed traffic behaviour. Also, the available literature on mixed traffic mainly deals with the single and two-lane roads and there is no comprehensive study for understanding traffic flow behavior on multilane highways. IRC: 64-1990 also provides detailed guidelines for capacity of single-lane, intermediate lane and two lane rural roads; leaving multi-lane highways almost unattended. Due to lack of standard codes for multilane highways in India, it is difficult for traffic engineers, policymakers and planners to take accurate decisions with respect to planning, design, and operations of these highways. The present research work aims at investigating the mixed traffic flow behaviour on fourlane divided highways for varying conditions of traffic volume and shoulders, and developing a simulation model for the observed traffic flow to estimate the capacity under u these conditions. To understand the traffic flow behavior on four-lane divided highways under mixed traffic conditions, the arrival patterns of vehicles, speed characteristics, lateral placement of vehicles and overtaking behavior were analyzed. The arrival data of vehicles were collected at 20 different sections selected on different four-lane divided highways. Speed data were collected at 24 sections and lateral placement data at 9 sections. Acceleration behaviour of a total of 750 vehicles of the different categories was studied to determine the overtaking and lane changing characteristics of mixed traffic on four-lane divided highways. Traffic composition and traffic volume did not cause any direct impact on patterns of vehicle arrivals. However, deviations of arrivals from the mean rate of arrival successfully defined the fitting of a specific statistical distribution irrespective of traffic volume. A parameter called coefficient of variation (COV) was defined to explain the arrival patterns of vehicles. It is defined as the ratio of variance of arrivals from the mean value and mean arrival rate. Poisson distribution defined the arrival patterns for COV values between 0.865 and 1.153, inclusive of both values. Binomial distribution fitted the vehicle arrival distributions for COV values less than or equal to 0.865 and negative binomial distribution satisfactorily explained the vehicle arrivals for COV greater than or equal to 1.078. Both negative binomial and Poisson distributions fitted the arrival data for COV values between 1.078 and 1.153. Similarly, both binomial and Poisson distributions defined the arrival data for COV value of 0.865. Normal distribution, gamma distribution and log normal distribution failed to define the observed speed characteristics, based on the results of chi-square test and K-S test. The concept of unimodal and bimodal curves is introduced in the present study to define the speed pattern on multilane highways under mixed traffic conditions. A statistical parameter, speed dispersion ratio (SDR), is introduced to explain the unimodality and bimodality in the speed data. In the case of sections with unpaved shoulders, the speed iii distributions curves were bimodal for SDR less than or equal to 1.786 and unimodal for SDR greater than 1.786. In the case of sections with paved shoulders, the speed curves were unimodal at all the 4 sections with SDR value ranging between 1.741 and 2.164. It is observed that the speed distribution curve tends to be bimodal with the increase in percentage of NMVs; although it is not necessarily the accurate measure of bimodality. Percentage of trucks significantly influences the speed distribution patterns on sections with paved shoulders. Normal approximation could not define the observed placement data of 'all vehicles', cars and heavy vehicles; when the data for total pavement width were considered. However, the normal distribution successfully defined the lanewise lateral distribution of these vehicle categories. In the case of two-wheelers, both lognormal and normal distributions fitted the observed data. Lognormal distribution solely defined the placement behavior of both nonmotorized vehicles and auto-tractor category. Placement behavior of non-motorized traffic (NMT) is considerably affected by the provision of paved shoulders. On sections with paved shoulders, about 85% of NMT shifted on to the shoulders. Heavy vehicles on these sections observed a shift of 41% towards shoulder lane, followed by two- wheelers and auto-tractor with average lateral shift of 21 % and 18.4 %, respectively. Car traffic was almost unaffected by the type of shoulder. At comparable traffic volumes, outer lanes carried higher percentage of traffic on sections with paved shoulders as compared to that on sections with earthen shoulders. Overall usage of shoulder lane increased with increase in traffic volume for both types of shoulders. The acceleration characteristics of the different types of vehicles on four-lane divided highways in India are evaluated. It is observed that the shoulder conditions (paved and unpaved) influence the acceleration behavior of vehicles. In the case of highways with paved shoulders, two-wheelers rarely overtook the test vehicle (car) as they mostly used IV shoulder lane for their movement. An inverse correlation is observed between the overtaking speed and the acceleration rate. A simulation program is developed in Visual Basic 6.0 to simulate the observed traffic flow behavior on four lane divided highway. The analysis of the observed data for vehicle arrival, speed, lateral placement and acceleration characteristics was used to develop the simulation program. Based on simulation results, a capacity of 4770 vph and 5290 vph is estimated in each direction for 'all cars' situation for four-lane divided highways at sections with earthen and paved shoulders, respectively. In the case of sections with earthen shoulders, it changes to 1300 vph, 877 vph, 2970 vph and 13520 for 'all heavy vehicles', 'all tractors', 'all autos' and 'all 2Ws' cases, respectively. In the case of sections with paved shoulders, it changes to 1349 vph, 892 vph, 3097 vph, and 15088 vph for 'all heavy vehicles', 'all tractors', 'all autos' and 'all 2Ws' cases, respectively. Simulation runs are also performed for car and one of the remaining categories of the vehicle in its different a proportion and their impact on capacity, speed and vehicular interaction (in terms of PCU) is evaluated. In the case of highway sections with earthen shoulders, capacity decreases with the introduction of tractors, autos or heavy vehicles (HV) in the traffic stream. The reduction in capacity is the maximum in the case of tractors because of their large size and low speed. However, 2Ws increased the capacity of the highway due to their smaller size and good operational characteristics. The speed of car traffic dropped invariably with the introduction of a second vehicle type in the stream. The extent of drop was dependent on the type of the second vehicle, its proportion in traffic stream and traffic volume. PCU values for a vehicle type decreased with the increase in volume-capacity (V/C) ratio and its proportion in the traffic stream. In the case of HVs, the PCU values show an increasing trend with increase in its proportion. An increase of 10-12% in capacity of four-lane divided highway sections is observed withthe provision of paved shoulders.