2014 ITM Presentation 4/29/2014 High Speed Rail Passenger Demand Forecasting: A New Approach Based on Network Scale with Real-world Applications in China Dr. Xiushan Jiang Associate Professor School of Traffic and Transportation Beijing Jiaotong University Dr. Lei Zhang Associate Professor Yijing Lu Graduate Research Assistant Department of Civil and Environmental Engineering 1173 Glenn Martin Hall, University of Maryland 1
Outline 1. Introduction 2. Objectives 3. Methodology 4. Application 5. Conclusion 22
Introduction The HSR network in China is rapidly expanding 3 3
4
Introduction The HSR travel demand forecasting is a vital issue. From the view of macroscopic planning perspective: Evaluate the feasibility of construction project Foresee the overall trend of the rail network. From the view of microscopic operation perspective: Determining the technical index Rail track length Rail network operational scheme HSR revenue management and so on. 5
Objectives Develop an improved four-step demand model for HSR operators and planning authorities. Demonstrate the methods through case studies in China Explore the transferability of the model to the other countries. 6
Methodology The Improved Four-Step Demand Model Comprehensively consider the characteristics of HSR passenger flows: Temporal dynamics and spatial network effects. Build economic-population model to predict induced travel demand. Use RP/SP joint survey of modal split. Assign travel demand to multiple paths based on utility function. 7
Traffic zone division and determination of OD Determine travel demand of each OD in the base year Traffic generation and distribution in different Developing stages Trend passenger flow model Economy population induced passenger flow models Total travel demand of each OD in different developing stages Traffic modes division HSR travel demand split rates of each OD in different developing stages Traffic assignment based on utility function multiple paths traffic assignment model based on utility function passenger travel demand along each path in different developing stages 8
Application: A case study Traffic Analysis Zones and Base Year ODs by Travel Modes 9
Application: Data The highway data was computed according to passenger departure schedules. The airline data was obtained from China's Traffic Yearbook. The HSR data was deprived from the Ticketing and Reservation System. Regional economic data was employed to acquire the trend and the induced passenger demand. Mode preference survey at different mode stations in Beijing, Tianjin, Shanghai, Guangzhou. 10
Tianjin Qinhuangdao Shijiazhuang Jinan Qingdao Hefei Xuzhou Suzhou Yangzhou Hangzhou Ningbo Datong Shanxinabu Luoyang Yichang Yingtan Zhuzhou Fuzhou Shantou Zhuhai Jinzhou Dalian Siping Neimenggu Ningxia Chongqing Nanning Chengdu Qinghai Xizang Results: Trip generation and distribution 3000.00 2500.00 2000.00 1500.00 3000.00 2500.00 2000.00 1500.00 1000.00 500.00 0.00 1000.00 2010 2015 2020 500.00 0.00 2011 2015 2020 11
Tianjin Cangzhou Qinhuangdao Shijiazhuang Dezhou Jinan Zibo Qingdao Bengbu Hefei Xuzhou Nanjing Shanghai Hangzhou Ningbo Wenzhou Taiyuan Zhengzhou Luoyang Wuhan Yichang Yingtan Changsha Zhuzhou Fuzhou Xiamen Shantou Guangzhou Shenzhen Jinzhou Shenyang Dalian Chanchun Haerbin Xian Chongqing Guiyang Kunming Chengdu Lanzhou 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 Results: High speed rail mode split rates 0.00 2020 2015 2011 12
Results: Comparison thousand 600000 500000 400000 300000 200000 100000 0 2009 2011 2015 2020 Predicted Rail Trips The Observed or the official predicted Rail Trips 13
2011 14
The HSR Route Choice Model 2015 15
The HSR Route Choice Model 2020 16
Conclusions The HSR travel demand will continue increasing with reduced growth rate. With the growth of the HSR network, The induced passenger demand becomes more important. The trips are concentrated in the developed regions with large population and high employment. In different HSR developing stages, passenger demand increases with different growth rates. 17
Future steps Further modify the calibration method of the parameters in order to improve prediction accuracy. Establish utility functions specific to traffic zones. Consider complementary effects and substitute effects among multiple modes. 18
Thank you Email: xshjiang@m.bjtu.edu.cn 19
Introduction Traditional model Historical Average model Exponential smoothing model Grey forecasting Autoregressive integrated moving average (ARIMA) Elastic coefficient method Box - Jenkins approach and so on. The modal demand method The logit model Four step model 20