Carnegie Mellon University

Image of Jingxiao Liu

May 11, 2018

Simulations, Sensors Provide Insight into Health of Transportation Infrastructure

By Alexandra George

Alexandra George
  • College of Engineering
  • 412-268-2579

Carnegie Mellon University researchers are working across many disciplines to create technologies that will improve and extend the life of transportation infrastructure.

Researchers in the Department of Civil and Environmental Engineering (CEE) are leaders in indirect structural health monitoring, or developing a low-cost, low-maintenance approach in which vehicles use sensors to indirectly monitor roadways, railways and bridges while traveling.

Jingxiao Liu, a CEE Ph.D. student advised by CEE Associate Professor Mario Berges and CEE Assistant Professor Hae Young Noh, is using ANSYS simulation software to model how a track geometry train car moves and collects data.

The project is a collaboration between Berges, Noh, Jacobo Bielak, the Hamerschlag University Professor in the Department of Civil and Environmental Engineering; Jelena Kovačević, the Hamerschlag University Professor and Head of Electrical and Computer Engineering and Professor of Biomedical Engineering; James H. Garrett, Jr., dean of the College of Engineering.

Track geometry is the measurement of a system in three dimensions. The simulation models track parameters for a light rail system to predict when the railway's health is degrading and is in need of repair, or when the geometry indicates a dangerous situation could cause a derailment. Currently, the track geometry car runs once or twice a year to collect data through a partnership with the Port Authority of Allegheny County. Because it's not feasible for the car to run continuously, the predictive model provides a complementary way to continuously monitor infrastructure systems.

"We have a simpler mathematical model, but we are trying to solve this problem by a physically -guided data-driven approach, which needs instructions from physical insights of the train-track interaction system and a good statistical inference from the data. This is why the ANSYS software is so valuable," says Liu said.

Liu's current simulation project is a bogie-track interaction system, which simulates a bogie, or wheel and axle system, accelerating over a railway.

Figure 2
Before moving the system, Liu conducts finite element analysis on the model as it sits in place. He analyzes the points of contact on the model. In many instances points of contact are frictional. This particular model is set up with the wheel bearing as frictionless and the contact between the wheel and track as frictional.
Figure 3

An example of frictional contact

Figure 4

Liu analyzes the joints.

Figure 5

Then he divides the model into separate finite elements in a process called meshing.

Figure 6

In the next step, loading, Liu applies forces equivalent to the weight of carrying a train.

Figure 7

Liu conducts a static structural analysis and models the train while it is stationary.

Figure 9

The final step of the process is conducting transient structural analysis while the simulation is running in order to gather data on how the system moves.

The model predicts railway health based on previously recorded data by allowing the researchers to test physical aspects of the bogie-track system. While they cannot change the parameters on the actual physical track, they can change the model. Testing the effects of different parameters informs when part of the track may have a problem. Rather than waiting for an imminent disaster, they can use that information to do track maintenance.

To simulate how the system moves along the track, they study the forces applied to different elements and the resulting deformations. This helps the researchers understand how the system behaves physically under different pressures and situations. The final step is transient structural analysis, in which the system is analyzed as it is in motion.

"The value of this software is that we can make our simulation approximately realistic to the real-world, and well-understand the train-track interaction system," Liu said.

Animation of simulation

The team also is investigating how to use acceleration data from vehicles on roads and bridges for similar purposes. Using data from previous experiments, they are developing a track geometry prediction algorithm inferred from the finite element model to predict a structure's future health.

"Our task is to improve the model, to make it robust to environmental factors and other influences," Liu said. "That's a lot less expensive than the track geometry car. We are trying to improve the monitoring of the huge railway network or even the infrastructure network with lower cost, less employee exposure, and more robustness, and reduce incidents and accidents, such as derailments."

This project is supported by the National University Transportation Center Mobility21 at Carnegie Mellon. Mobility21 is a member of the university's Metro21: Smart Cities Institute, which seeks to research, develop, and deploy solutions to improve the quality of urban life.