A Stochastic Simulation Based Approach for Transportation Demand Forecast and Safety

Abstract

Reliable traffic forecasts are critical for successful planning and financing of transportation projects. However, the accuracy evaluation results of the predicted traffic volumes often indicate significant discrepancies between the actual and predicted traffic volumes. These discrepancies are mainly due to both limitations and uncertainties in traffic demand forecasting models; therefore, researchers are continuously attempting to estimate the error and bias in traffic forecasts. To enhance these endeavors, a Monte Carlo simulation (MCS)- based approach is proposed in this paper as an alternative to the stochastic traffic demand forecast. This approach uses a sequential process to capture the effects of model inputs and variables on the error and bias in traffic demand forecasting. The approach proposed in this study was developed in three steps. First, the key factors causing errors and biases in traffic demand forecasts were identified based on a comprehensive review of traffic forecasting practices. Second, the effects of each variable on traffic volumes were quantified using MCS. Lastly, the statistical approach was constructed to provide an interval estimation of traffic volumes based on the findings of MCS. Through these steps, the inherent uncertainties in socioeconomic variables and heterogeneities in passenger behaviors with regard to traffic demand modeling were considered. Then, simulation experiments were conducted to investigate the applicability of the proposed approach to a real-world network. This approach is expected to reasonably capture the stochastic nature of future traffic volumes and quantify the risks associated with the error and bias in traffic demand forecasting. The findings of this study indicate that traffic forecasting practitioners can use the proposed approach with ease. The analysis results showed that the proposed approach clearly captured the changes in link traffic volumes owing to the statistical changes in the value of time (VOT). The experiments indicated that the uncertainties in VOT perceived by road users can be used to estimate the risks of toll road financing. The elasticity of traffic volume to VOT for each link, which can be used as a key input to the road project feasibility study, can be derived from the MCS-based analysis implemented in this study.

Keywords

• demand forecast
• Monte Carlo simulation
• origin-destination (O-D)
• stochastic model
• transportation
• value of time