Guest Speakers - Prof Terry L. Friesz& A/Prof Ke Han- Statistical Metamodeling of Dynamic Network Loading and Application to Dynamic Traffic Assignment

Guest speaker Prof Terry L. Friesz, Pennsylvania State University, USA and A/Prof Ke Han, Imperial College London, UK  presented a seminar at rCITI on Friday 1 July 2016.

Terry L. Friesz is the Harold and Inge Marcus Professor of Industrial engineering at the Pennsylvania State University. His research concerns the application of differential game theory and other mathematical formalisms to the study of transportation, supply chains, facility location, and interdependent infrastructure systems. He has been a professor at MIT, George Mason University, and the University of Pennsylvania, where he held the UPS Foundation Chair in
Transportation. His papers have appeared in Operations Research, Transportation Research Part B, Transportation Science, the Journal of Regional Science, Urban Economics and Regional Science, the SIAM Journal on Control and Optimization, Mathematical Programming, and Environment and Planning. He is the founding editor of the Springer journal Networks and Spatial Economics.

Ke Han received his BS degree in 2008 from University of Science and Technology of China, and his PhD degree in 2013 from the Pennsylvania State University. He then joined Imperial College as a Lecturer (Assistant Professor) in Transport. His research interests include traffic flow theory, network modelling, intelligent transport systems, traffic operation and control, network design, and transport resilience and sustainability. To date, he has co-authored two books and published over 60 journal and conference papers. He serves on the editorial board of Transportation Research Part B, and is a guest editor for Transportation Research Part C. He is a visiting professor in Nanjing University of Aeronautics and Astronautics. He is among the five Chan Wui & Yunyin Rising Star Fellows selected globally by Transportation Research Board.

Prof Friesz and Dr Han's presentation was on Statistical Metamodeling of Dynamic Network Loading and Application to Dynamic Traffic Assignment. Dynamic traffic assignment models rely on a network performance module known as dynamic network loading (DNL), which expresses the dynamics of flow propagation, flow conservation, and travel delay at a network level. The DNL defines the so-called network delay operator, which maps a set of path departure rates to a set of path travel times. It is widely known that the delay operator is not available in closed form, and has undesirable properties that severely complicate DTA analysis and computation, such as discontinuity, nondifferentiability, nonmonotonicity, and computational inefficiency. The paper discussed proposes a fresh take on this important and diffcult problem, by providing a class of surrogate DNL models based on a statistical learning method known as Kriging. Presented was a metamodeling framework that systematically approximates DNL models and is flexible in the sense of allowing the modeler to make trade-offs among model granularity, complexity, and accuracy. It is shown that such surrogate DNL models yield highly accurate approximations and superior computational efficiency. Moreover, these approximate DNL models admit closed-form delay operators, which are Lipschitz continuous and infinitely differentiable, while possessing closed-form Jacobians. The implications of these desirable properties for DTA research and model applications were discussed in depth.