6 March 2020
The landslide-induced TGV (high speed train) accident in France yesterday
Posted by Dave Petley
The landslide-induced TGV (high speed train) accident in France yesterday
Yesterday, 5 March 2020, a TGV (high speed train) struck a landslide between Strasburg and Vendenheim in the Bas-Rhin area of France. The train remained upright, not least because it appears that it was a glancing blow rather than a direct collision, but 22 people were injured, one seriously. The best news report, with thanks to Scott Johnson, is in L’Usine Nouvelle. The article is in French, but Google Translate does a fine job.
This image of the landslide was tweeted by SNCF:-
The 5 March 2020 landslide that derailed a TGV (high speed train) in France. Image Tweeted by SNCF.
.
The landslide is a large rotational slip in a slope in a cutting. The displacement of the mid-section is quite large, but little of the debris appears to have reached the tracks. This prevented a more serious accident. The train, which had 348 passengers on board, was travelling 270 kilometres per hour (170 miles per hour) at the time of the collision.
The line is quite new – Wikipedia indicates that it was constructed in the period between 2010 and 2016. A failure on this scale will inevitably cause concern, and is surprising. News reports indicate that the landslide was triggered by heavy rainfall. Interestingly, this is being described as an “accident intolerable” – i.e. an unacceptable accident – by the local trade union.
After the accident, the train came to a stop at about 48.729, 7.514, based on matching images to Google Earth. The accident must have been to the southeast of this point. The most likely location appears to be 48.719, 7.538, but this is very tentative.
Landslide-induced train accidents occur fairly often around the world, sometimes with very serious consequences. It is very unusual for an accident to affect a modern high speed line, especially in well-designed earthworks, which would typically have been constructed with a c.125 year design life. Thus, understanding the causes of this accident will be a priority.
Dave Petley is the Vice-Chancellor of the University of Hull in the United Kingdom. His blog provides commentary and analysis of landslide events occurring worldwide, including the landslides themselves, latest research, and conferences and meetings.
Did the train run into it or did people notice it before?
I can appreciate why someone would label this as an ‘unacceptable’, but it would be very, very interesting to look at the quantified risk of such an event occurring along this alignment that coincides with a train passing by at 270 km/h. When I had the pleasure of having Oldrich Hungr as a professor in graduate school, he showed us the quantification of risk of a car being hit by rockfall along the Sea to Sky Highway – the route between Vancouver and Whistler on the west coast of BC, Canada – and not surprisingly, the risk was fairly low even with a relatively fairly high occurrence rate of rockfall that occurs along mountainous highways in the province. I would expect a similar result from looking at the risk along this particular rail alignment as well.
I would also be keen to see the results of a geotechnical site investigation for this particular length of slope. The slopes adjacent to the failure do not appear show any distress (based on the photo above). This leads me to ponder if there is a particular occurrence of soil stratigraphy at the location of the failure that provided a conduit for groundwater to raise the phreatic surface within the failure zone.
And as always, many, many thanks for your continuing effort of posting here on a regular basis. Your ongoing sharing of knowledge is very, very, much appreciated.
[An interesting point, but a TGV travelling at >200 km/h with up to 500 passengers will mean that the consequence side of the equation will look very different to a car on a highway. I anticipate that the expectation is that a failure of this scale on a cut slope in this setting is highly unlikely. I would be very surprised if the investigation does not find that something has gone wrong here in terms of ground investigation, design and/or construction. There will be lessons to be learnt; the urgency will be to determine as to whether this applies to other slopes along the alignment. D.]
Just another instance where site prep based on “historic” rainfall stats has proved to be short-sighted and injurious.
It’s not only the ‘Consequence’ but also the ‘Likelihood’ that makes a train derailment a much higher risk than a rockfall hitting a passing car. A landslide impacting the tracks, either through debris reaching the tracks or displacing the tracks, means the ‘hazard’ remains in place until someone notices it and alerts the train operator, who are then able to stop trains (provided they are many kilometres away from the hazard). In rural areas like the above, it will unfortunately most likely be the train driver that is the first person to notice the failure. This situation is not a comparable Likelihood or Consequence to the passing of a rockfall in a split second hitting a car. This sort of failure is most certainly unacceptable.
The local geology of this failed slope may be unique compared to the surrounding terrain. The failure occurs on the flank of a hillock probably underlain by sandstone with a limestone carapace. There is also dolomite and chalk in the region. The scarp shows some geometric, high-albedo surfaces that I interpret as evidence of structure in resistant dolomite or chalk. The geology also appears to differ on either side of the track … is the track on a geologic discontinuity? These subsurface geologic conditions have the makings of slope failure without sufficient engineering measures to improve the factor of safety. Additionally, surface drainage is likely the major contributing factor to triggering the failure. There is a paved road that slopes down towards the failed slope. I can’t find imagery to support observations, but there may be no drainage ditch or just a shallow one along the road. At the crest of the hillock approaching the bridge abutment to the west of the failure, the road is cut on both sides forming a channel for rainfall to accumulate and divert downslope towards the failure. Furthermore, there are no obvious diversion trenches, lined or unlined, along contour of the slope. The slope is micro-benched for shallow stability and has two or three larger benches for global stability but there is no obvious slope reinforcement. Just by looking at the imagery and a small-scale geologic map it seems obvious that this slope should have had better surface drainage, sub-drainage, and slope reinforcement. I really wonder what led the engineering team to consider that slope reinforcement wasn’t necessary. We will probably learn the geologic model was too conceptual and not site-specific enough… they probably used the subsurface exploration for the bridge abutments as the design basis for the slopes instead of drilling and trenching the slopes themselves. Failure of engineered slopes along major infrastructure is intolerable. We must accept some risk from large natural slopes as a practicality, but once we investigate, design and engineer a slope there is close to zero tolerance even if the risk of failure is never eliminated with engineering measures.
RE: Charlie’s comment on rockfall risk…
I agree with Richard…
Was Prof Hungr’s analysis about the risk of a rockfall hitting a car (as you write) or a the risk of a car hitting rockfall debris (more similar to the TGV scenario)? Agree, it is pretty unlikely that a rock fall actually strikes a moving vehicle, although it occasionally happens. Much more common is for the vehicle to strike the rock fall debris sitting in the middle of the road, or a train to hit ground debris on tracks.
Anyways, I wonder what the driver of the car in this video would have to say about Prof Hungr’s analysis: https://www.youtube.com/watch?v=uOJfcTZME0U
[I agree too. Rockfalls do occasionally hit cars (there are several videos on Youtube of this). Trains are of cource much longer, so the risk is much higher, but the true threat continues to be impact by the train on landslide debris, and derailment, as per the incident in China this week. D.]
Sentinel-2 moisture Index data shows “ponding” areas where the moisture index remains significant above the “normal” in various surficial wells along the slope. Sentinel-1 data shows activity in the area near the slide during the season. Clearly, when the precipitation levels are above the norm, and there are insufficient horizontal drainage options in the surrounding slopes — no matter how shallow .., gravity will come to town.