3D discrete-continuum simulation of differential settlement in ballasted railway transition zones

What is the topic of your Doctoral Thesis?

My thesis, titled "3D discrete-continuum simulation of differential settlement in ballasted railway transition zones" focuses on modeling and understanding the complex behavior of these critical track sections.

A transition zone is where the track's stiffness changes significantly, like the area where a soft railway embankment meets a rigid structure like a bridge or tunnel. These zones are notoriously prone to differential settlement, where one part of the track settles more than the other, leading to track degradation and high maintenance costs.

The core of my research was to develop a novel 3D hybrid modeling approach to simulate this problem. This model uniquely combines three different computational methods:

• The Discrete Element Method (DEM) to model the granular materials like ballast and sub-ballast at the particle level.

• The Finite Difference Method (FDM) to model the continuous components, such as the rail beams and the subgrade.

• The Finite Element Method (FEM) to simulate the dynamic interaction between the moving train and the track.

This integrated framework allows us to realistically simulate both the short-term dynamic forces from a passing train and, most importantly, the long-term, progressive buildup of settlement over repeated axle loads. The simulation process can be seen in YouTube from this link: youtu.be/h1o8B9umRFo

Why did you choose this topic?

My motivation was driven by the significant challenge railway transition zones pose to sustainable infrastructure. These zones are a major vulnerability in the railway system. The abrupt stiffness changes lead to accelerated degradation, which in turn demands frequent and costly maintenance. For high-speed rail, in particular, even minor irregularities can be amplified, compromising safety and ride quality. This creates a cycle of high life-cycle costs and increased environmental impact from repeated maintenance interventions.

I saw a gap in the available modeling tools. While DEM is excellent for understanding granular materials like ballast, it has significant limitations in modeling continuous structures like rails or the subgrade. To truly understand the problem and develop effective, preventive design strategies, we needed a more holistic, hybrid approach. My goal was to build that tool—one that could accurately predict long-term performance and help engineers design more durable and resilient transition zones.

What are the most important results?

I'd group the most important results into two categories: methodological and physical.

On the methodological side:

• We successfully developed techniques to make these massive simulations computationally feasible. We showed that particle scalping (removing the smallest fine particles from the sub-ballast model) can reduce simulation time by up to 90% without compromising the accuracy of the shear behavior.

• We implemented a Periodic Cell Replication Method (PCRM), which allows us to efficiently build very long, realistic track models by generating one small, representative "brick" of compacted ballast and then replicating it.

On the physical side, our hybrid model revealed the specific failure mechanism in these zones:

• The model confirmed that the abrupt stiffness gradient is the root cause, amplifying dynamic wheel-rail forces right at the transition.

• This amplification leads to void formation, or gaps, developing beneath the first few sleepers on the softer, ballasted side of the transition.

• As these sleepers become "suspended", the train's load gets redistributed onto the adjacent sleepers further down the track.

• This load shift is the crucial finding: it progressively pushes the peak settlement point about 3 meters into the softer track, away from the structure itself. This is what worsens the track irregularity over time and highlights the need for a gradual, rather than abrupt, stiffness change.

Did you come across something unexpected during your thesis research?

Yes, absolutely. A significant and unexpected finding emerged when I was investigating the reliability of the DEM simulations.

I observed that when running 2D simulations, my results would vary significantly even if I used the exact same set of particles and material properties. Initially, I was concerned there was a flaw in my model. However, after a systematic study, I discovered that this variability was caused by the initial random arrangement of the particles. The unexpected part was the dramatic difference between 2D and 3D. In 2D models, this initial random packing had a significant influence on the results. But in the 3D models, the outcomes were much more resilient and consistent.

It turns out that the richer, more complex spatial framework and contact network in 3D help to stabilize the granular assembly. The 2D models, with their inherent geometric limitations, just couldn't capture this stable behavior and were overly sensitive to small, random variations. This finding was what solidified my decision that 3D modeling, despite being more computationally expensive, was absolutely essential to get reliable and realistic results for this problem.

Who will benefit from your results? What kind of impact may it have on surrounding society?

The primary beneficiaries are railway engineers, infrastructure owners, and maintenance planners, such as the Swedish Transport Administration (Trafikverket), which financially supported this research.

My hybrid model provides them with a practical, predictive tool. They can use it to:

• Test new designs for transition zones before they're built.

• Evaluate mitigation strategies, like using under-sleeper pads, reinforced ballast, or adjusting sleeper spacing to create a more gradual stiffness transition.

• Develop more targeted and cost-effective maintenance strategies by understanding exactly where and why the degradation is occurring.

The broader impact on society is directly tied to achieving more sustainable and durable infrastructure. By designing transition zones that last longer and require minimal maintenance, we can significantly lower the life-cycle costs of our railways. This translates to:

• Lower costs for taxpayers.

• Fewer service disruptions for passengers and freight.

• A reduced environmental impact from fewer maintenance activities and material replacements.

Ultimately, this research contributes to building a safer, more reliable, and more sustainable high-speed railway network.

What will you do next/where can one reach you?

The logical next steps for this research are to build upon the framework we've established. This includes incorporating more advanced ballast degradation models and validating the long-term predictions against full-scale field data. Another exciting avenue is integrating AI models with these detailed simulations to predict maintenance needs for railway tracks.

This research has also laid the foundation for a post-doctoral project, which will be carried out soon by another colleague to explore these areas further.

As for me, I currently work in the industry at Sweco Sverige AB. My goal is to combine the academic knowledge from my Ph.D. with practical industrial experience to contribute more effectively to real-life projects and challenges.

You can reach me via my LinkedIn profile: www.linkedin.com/in/alireza-ahmadi/