Quantification of the effects of non-motorised transport and roadside activities

Abstract

The Roads & Highways Department (RHD) of Bangladesh is introdueing the Highway Development and Management Model (HDM-4) for effieient road planning and management. In order for HDM-4 to be implemented and used successfully in Bangladesh, it must first be calibrated and validated. However, this is problematic due to the non-lane based heterogeneous characteristics of the traffic stream and frequent interaction between various roadside activities and vehicular traffic. This thesis presents the results of a study to quantify the effects of Non-Motorised TranspOI1 (NMT) and interaction of roadside activities. To this end the following models were developed/evaluated: free speed models, speed-flow relationships under heterogeneous traffic system, the effects of proportions of NMT and roadside activities on speed-flow, NMT time and operating cost models. In order to develop speed-flow models under heterogeneous traffic, dynamic data related to vehicles and static data for roadway characteristics were collected involving 53 sites from a wide variety of roads in Bangladesh. Data on static vehicle parameters were gathered from approximately 3,000 vehicles. Time and operating cost data for NMT were obtained by conducting a comprehensive questionnaire survey in three phases involving approximately 6,500 respondents. The results of this research demonstrate that, as far as human driven NMT is concemed, the probabilistic limiting velocity model approach used in HDM-4 for motorised transport, can easily be extended to model free speeds for NMT under combined road characteristics. The research adopted a unified approach to develop speed-flow relationships for heterogeneous traffic by employing separate individual speed-flow relationships for each vehicle type and assigning passenger car space equivalent (PCSE) values on the basis of critical space. Speed-flow relationships were more or less successfully applied to heterogeneous traffic systems using three-zone models. It was apparent in the research that capacity estimates by Greenshields model were not compatible with the concept of separating static and dynamic attributes of speed- flow relationships by Hoban (1987). It was observed in the research that single regime generic models were not only relatively easy to calibrate, but also provided a relatively better representation of speed-flow relationships in comparison to that of three-zone models. It was observed that NMT speeds are generally not a function of traffic Ilow volumes up to capacity. There is no road in Bangladesh. which does not carry NMT. Consequently, it was not possible to calibrate directly the boundary points of the three-zone speed-Ilow model with respect to the proportion of NMT and roadside friction. The problem was overcome by employing linear mixed models using analysis of variance (ANOY A). The ANOY A models showed no significant impact of car speeds on the proportion of NMT in the traffic. This might be attributed to the separation of static and dynamic characteristics of speed-flow in this research. However, the proportion of NMT in the traffie mix was found to have signifieant effect on motorised transport Ilow and capacity. Again, direct analysis of the effects of roadside friction showed that this has a signi ficant impact on both speed and capacity. The lindings demonstrate the inadequacy of the different approaches. which adjust only the free speeds while keeping constant the boundary points of the three-zone speed-Ilow models. In order to estimate NMT operating costs, the values of in-vehicle time and waiting time were evaluated out using Stated Preference (marginal logit or probit) and Transfer Price models, respectively. The robustness of the value of time by marginal models with respect to "within individual" or "between individual" variation was investigated. However, no significant difference was obtained. In order to overcome the difficulty of calibration, and to assess relative merits of different modelling techuiques for NMT operating costs, eight different types of models were developed; (i) aggregate linear model, (ii) aggregate logistic model, (iii) discrete stated preference model, (iv) discrete transfer price model, (v) elemental cost model under constant working hours, (vi) elemental model cost under constant travel distance, (vii) existing HDM- 4 model, and (viii) modified HDM-4 model. These were compared and validated with respect to the actual fares charged or willingness to pay under different operating environments. Out of the eight models considered, only two, the existing HDM-4, and the elemental cost under constant travel distance, produced some inconsistent results when compared to the actual fares charged. However, the problem with the existing HDM-4 model appeared to be minor and was possibly due to double counting of some crew related costs. The assumption of constant annual distance utilisation was found to be unrealistic for commercial operation of NMT. Aggregate models were proved to be more comprehensive for commercial NMT operations where fare data exists. Discrete models with a small number of criteria were found to produce consistent results for both commercial and private operations of NMT. These models also proved to be the only viable alternative for the development of operating costs for pedestrians where no tangible operating costs or fare data existed. The modified HMD-4 models demonstrated the importance of inelusion of waiting time for operating cost estimation. Since total operating costs generated by the models under different trip conditions were equivalent to the actual fares charged according to road condition, this proved that NMT operate under long-run marginal equilibrium conditions.