Transport modelling research across Australia | 27 Oct 2022

The AITPM Transport Modelling Network (TMN) brings to you the second in a series of webinars aimed at sharing the wonderful research that is being conducted around Australia at our academic institutions. In this seminar we will hear from four researchers from QUT, PATREC and Victoria University. Each presentation will be for 20 minutes followed by a 30 mins Q&A.

Join us to hear from four researchers:

Speaker #1: Professor Alex Paz, QUT and Prithvi Bhat Beeramoole, QUT | Estimation of discrete outcome models simultaneously considering multiple objectives and complex data characteristics

This study focuses on the discrete outcome estimation problem, which involves testing multiple hypotheses that can affect both interpretability and predictive accuracy. While a few studies have proposed a mathematical programming approach to assist with hypothesis testing and estimation, there is limited knowledge regarding the effect of also considering both in-sample and out-of-sample model performance criteria during the search for a specification. To address this knowledge gap, a multi-objective optimization framework is proposed including both in-sample goodness-of-fit and out-of-sample predictive accuracy to generate multiple unique specifications and perform extensive hypothesis testing considering simultaneously potential explanatory variables, nonlinearities, heterogeneous effects, and correlations. A metaheuristic was designed and implemented to solve a proposed multi-objective nonlinear mixed-integer mathematical programming problem. Experiments including various datasets and discrete outcomes were used to illustrate the efficacy of the proposed framework to find specifications that are either very similar or dominate those reported in literature, considering both interpretability and predictive accuracy. Important insights regarding potential explanatory factors and heterogeneous preferences, which were not reported in literature, were captured using the proposed framework.
In addition, for the datasets used in this study, the proposed framework enabled the discovery of three distinct clusters considering specification type and model performance in terms of interpretability and predictive accuracy. These clusters suggest that mixed-Logit models with correlated parameters perform significantly better than those without correlation, while multinomial-Logit models showed the worse performance. 

A comparative analysis including multiple performance measures suggests that model evaluation using in-sample Bayesian Information Criterion (BIC) and out-of-sample mean absolute error (MAE) enables estimation of specifications with better interpretability and generalizability compared to those estimated using maximum log-likelihood and minimum number of model parameters. In addition, a mostly linear relationship was observed between in-sample and out-of-sample log-likelihood, indicating that the latter does not provide any additional information regarding prediction compared to the in-sample estimates. These results showed the value of using an optimization framework to support the estimation of discrete outcome models by enabling extensive hypothesis testing and including multiple performance criteria as well as complex data characteristics to discover important and generalizable insights.

Speaker #2: Dr James Lennox, Victoria University | Spatial economic dynamics and transport project appraisal

Transport infrastructure is very long-lived. Over time, changes in accessibility delivered by major transport projects are likely to affect the distribution of population and employment. Dynamic Spatial Equilibrium (DSE) models can capture these land use responses as they unfold over time, in a forward-looking spatial general equilibrium. These dynamics are crucial. Firstly, the construction phase of projects can be explicitly represented. Secondly, land use changes are often of direct interest to policy-makers, so their dynamic evolution is of interest per se. Thirdly, slow land use dynamics have important implications for welfare analysis because future benefits are discounted and adjustments are costly.

In this paper, we present a flexible DSE model incorporating dynamics of internal migration and occupation choice, and intra-period spatial linkages via commuting and trade flows. We calibrate the framework to Australian data and illustrate its application by modelling a hypothetical fast express rail service in South-East Queensland. In analysing the results, we highlight the roles of both general equilibrium effects within periods and costly dynamic transitions between periods. Transport cost changes are exogenous inputs to the DSE model in our simulation. However, we describe how the model could be linked to existing four-step strategic transport model to enable dynamic Land Use -- Transport Interactions (LUTI) simulations.

Speaker #3: Dr Chao Sun, PATREC | Applying perimeter controls to larger urban regions to regulate traffic at a system-wide level by optimizing flow in each region

Perth has begun to embrace Smart Freeway technology with the creation of the new Kwinana Freeway Northbound and in-progress Mitchell Freeway Southbound systems. Ramp metering is one of the cornerstone solutions, which in simple terms means regulating traffic inflow at on-ramps to prevent flow breakdowns on the freeway so it can remain reliable while delivering higher throughput. This iMOVE project investigates the possibility of applying a similar type of perimeter control to larger urban regions to regulate traffic at a system-wide level by optimizing flow in each region.

Perimeter control (also known as gating) works by dividing the network into zones and regulating their flow exchange at the boundaries. It aims at load-balancing between zones across the network to achieve a stable and optimum operation at the global level. Controllers prevent overflow of traffic into busy zones by leveraging spare capacities in less busy zones as temporary storage space. This contrasts with local congestion relief strategies that focus on individual pinch points, which can result in pushing too much traffic downstream and creating another bottleneck.

Effective implementation of perimeter control requires a good understanding of the behaviour of each zone. Macroscopic Fundamental Diagrams (MFDs) are commonly used for such purposes. They describe underlying relationship between a zone’s speed, flow, and density at the aggregate level and are believed to be stable under different traffic demands. The accurate measurement of MFDs has become possible in recent years with the advent of ‘big data’ and iMove project 1-003 has demonstrated that Perth can be divided into zones with well-behaved MFDs.
This project investigated the application of MFD-based perimeter control to the Perth CBD road network using computer simulation. The significant findings of this research are:

• Well-behaved MFDs can be generated using Perth CBD simulation data if there is some level of congestion in the simulated network
• Perimeter control can be applied to reduce congestion in a protected zone with the cost of increased delays at the entering boundary
• The type and strength of perimeter control can be modified to alter the balance between reduced congestion inside and increased delays to enter
• If the increased flow inside the system outweighs the delay imposed at the perimeter, then the network performance can be improved using perimeter control
• We have found that network flow can be increased by 39% and trip completions can be increased by 17%

By keeping the whole network at a steady state, gating has the potential benefit of maximising the total productivity and reducing the need for local capacity expansions.

This provides social, economic, and environmental benefits by improving traffic conditions for drivers using existing infrastructure which reduces required future infrastructure spending.
This research is funded by PATREC and the iMOVE CRC and supported by the Cooperative Research Centres program, an Australian Government initiative.