Doctoral Thesis: How frost heave in the railway track can be mitigated
A new doctoral thesis at NTNU finds that almost 70 percent of isolated track defects along the Trondheim–Støren line are caused by frost heave rather than settlement, with the culverts beneath the track as the main cause.
The main suspect: an open culvert in a railway embankment in midwinter. The pipe carries water through the embankment – and cold air into it, so the frost reaches soil that would otherwise have stayed unfrozen. ( Photo: Rui Tao)
For the first two years of his doctoral work, Rui Tao did not know what he was looking for. A railway track can cause train accidents in dozens of ways, from cracks in the rail head and loose fastenings, the components that hold the rails to the sleepers, to uneven settlement deep in the substructure. The literature pointed in every direction at once.
"For the first two years of the project, I was somehow lost in so many track defects. I had to find out which one was the most critical for the Norwegian railway," Tao says.
He came to Norway in August 2021 with a background in water conservancy and hydropower engineering from Hohai University and Wuhan University in China. He had little prior knowledge of the Norwegian railway system. The way he eventually approached the problem was inspired by how an investigator works on a criminal case.
"The process is really like the way investigators work in a detective novel. You have some clues, and then you try to find out who is behind it," he explains.
The clues had to be gathered from three fields that rarely talk to each other: geotechnical engineering (the study of ground conditions), geomatics (the discipline that uses satellites and advanced measurements to map terrain), and railway engineering. The thesis was submitted at the Department of Civil and Environmental Engineering at NTNU in the spring of 2026 and funded by the Centre for Green Shift in the Built Environment. Tao is now a postdoctoral researcher at NGI in Trondheim.
The line between Trondheim and Støren, where Tao analysed four years of measurement train data. ( Photo: Rui Tao)
131 defects along 50 kilometres
Bane NOR operates the Roger 1000 measurement train, which records track geometry as it runs and covers the section between Trondheim and Støren twice a year. Tao analysed vertical rail displacement from 2016 to 2020 using a continuous wavelet transform, a signal processing method that separates isolated anomalies from the gradual wear along the line.
Along those 50 kilometres he found 131 isolated defects. Almost 70 percent of them were heaving, not settlement. Around 70 percent of the heaving defects returned after the track had been maintained. Degradation was typically between 0 and 3 millimetres per year, with extreme cases of 6 to 7 millimetres.
That the heaving defects keep returning says something about what maintenance actually achieves.
"The maintenance is mainly done with the tamping machine, which corrects the geometry at the surface. If the cause lies in the substructure, the problem comes back again and again and again," Tao explains.
To quantify how serious a heaving defect is, he simulated an NSB Class 73 train running over idealised track defects. Two quantities are critical. If the contact force between wheel and rail falls toward zero, the wheel loses contact with the rail and the risk of derailment increases. If the car body's vertical acceleration increases, passengers feel it as a jolt. The simulations showed that the heaving defects both increase operational risk and reduce passenger comfort.
Tao then inspected the areas affected by heaving himself.
" We found that the locations affected by heaving coincided with culverts, and it was not random. Almost everywhere the track had heaved, there was a drainage channel, a culvert, or another structure carrying water," he says.
The physical model test: a scaled culvert in a freezer room at NTNU. Rui Tao (right) with Professor Gustav Grimstad from NTNU, who helped him build the setup. ( Photo: Private)
Condoms as a membrane
A culvert is an open-ended pipe carrying water through the railway embankment. They are very common in Norway, and many of them were originally installed to drain farmland and channel rainfall into rivers and streams. In winter, those open ends draw cold air through the pipe. The cold, therefore, reaches the soil well inside the embankment, and the frost penetrates considerably deeper than it otherwise would.
At that point three conditions are met at once, and they are exactly the three that frost heave requires: temperatures below zero, frost susceptible soil and enough water.
"In Norway we have long winter and a lot of frost-susceptible soil, and the groundwater table is often shallow. Unfortunately, all three conditions are easily met along the railway lines," Tao points out.
But two things occurring together are not the same as one causing the other. The field observations showed where the heaving defects were, not what happens in the soil around the culvert over the course of a winter. Tao needed to find that out in order to move forward.
Instrumenting a culvert in service is close to impossible, so Tao built a scaled-down version in a freezer room at NTNU: minus six degrees for the air above the ground and inside the pipe, plus six degrees in the water reservoir below, displacement sensors at the surface and temperature sensors at several depths.
The budget was 10,000 Norwegian kroner.
"That is nothing. I almost had to mobilise everyone in the lab to make it happen," he says.
The soil in the model was fully saturated, and water could not be allowed to enter the pipe itself. The solution was a latex membrane at each end, secured with O-rings.
"We used the cheapest solution we could find," Tao says with a smile. He collected almost 20 condoms from the free supply in the main building at NTNU.
The experiment showed why a pipe can lift a railway track. When water freezes, it expands by around nine percent, and that alone does not produce a large heave. The difference is that frozen soil retains more water. Around each soil grain lies a thin film of water that does not freeze, and it migrates upward toward the freezing front, the boundary in the soil between frozen and unfrozen ground, and adds to the ice there. Flat layers of ice, known as ice lenses, grow in this way, and they keep thickening as long as water arrives from below.
" The final conclusion is that ice lenses are the main cause of frost heave. Their growth could be seen with the naked eye during the physical model test," Tao says.
Rui Tao defended his doctoral thesis at NTNU in 2026 and is now a postdoctoral researcher at NGI in Trondheim. ( Photo: NTNU)
Only thermal insulation works
As Tao understands it, Bane NOR plans to install numerous culverts to increase drainage capacity and reduce flood risk along Norwegian railway lines. The thesis concludes that frost protection should be built into the design and construction of the culverts themselves to avoid frost heave.
Tao tested three measures in a computer model, one against each of the three frost-heave conditions. Thermal insulation against the cold, prestressed geogrids against the heaving soil, and an impermeable film against the water.
Only the thermal insulation worked. It reduced differential heave, the unevenness that makes the track rough, by up to 90 percent, while the film and geogrids had very little effect.
"Our conclusion is that only the thermal solution works. The problem with the hydraulic solutions is that the Norwegian winter is too long. They have some effect at first, but over time, with a lot of cold, the heave returns to its original level. The mechanical solution also fails because the difference in stiffness between the geogrid and the frozen soil is too small for the geogrid to be effective. " Tao says.
After the defence: Rui Tao (centre) with his main supervisor Yutao Pan from NTNU (left) and the head of his section at NGI, Jean-Sébastien L’Heureux (right). ( Photo: Private)
Satellites that watch the track move
All of this applies to new railway construction. For track already in place, digging up the embankment to install insulation is too expensive. There the heaving has to be kept down with the tamping machine, and inspections determine where and when it is deployed. Demand for inspection already exceeds capacity.
"We need better guidance for maintenance, so that people are sent to the right place before a major problem or an incident occurs," Tao says.
He therefore tested satellite data. The method, called PS-InSAR, measures ground motion by comparing radar images from two passes. Sentinel-1 was chosen because the images are free and open. The method is widely used to monitor soil moisture and flooding, but is little used on railways.
"Whether it could be used on the railway, and on the Norwegian one in particular, we had no idea," he says.
The raw data could not be used directly, partly because some measurement points were recorded in the wrong locations.
"At the Gaula river, you see many InSAR measurement points lying inside the river. They should have been on the railway bridge deck," Tao says.
He therefore developed a processing method that suppresses both time-series noise and InSAR PS point misplacement. Along a 52-kilometer section, four of seven anomalies matched incidents reported by train crews, and two of the remaining three were confirmed by measurement train data. The last one was located in an area with very few InSAR measurement points.
The method is therefore sufficient to indicate when something is about to happen. The distance from documenting this in a scientific paper to the routines of an operator such as Bane NOR is nevertheless considerable.
"The paper was published last year and received almost 30 citations in one year, so it has received certain attention in the academic field. But how it can be used in daily work is perhaps what I care most about. The true measure of our research lies not in its publication, but in its application to real-world challenges,” Tao says.
For Tao, the value of railways goes far beyond his research and their practical function:
“Compared with air and road travel, railways may seem slow and inflexible. Yet their tracks are always there, quietly carrying passengers and goods across the country. Along these fixed rails, we can always find our way home, to our families, our friends and the people we love,” he concludes.
