In view of the long service history of the Finnish railway network, what makes the subject so urgent right now? The Finnish railway network has been in use since the 1860s. In the early days, stability engineering was still a great unknown, and, in the main, embankment construction on soft soil beds proceeded through the application of the trial-and-error method.
In fact, stability computation and geotechnical studies developed in the Nordic countries largely by virtue of the construction of railways. In Finland, the Geotechnical Commission of the State Railways was established in 1919. Application of computational methods became widely accepted as late as in the 1940s, at a time when the main part of the existing railway network’s total length of approximately 6,000 kilometres had already been constructed. A great many things have changed over time. The concepts of failure safety, the determination of soil strength, and strength reduction have all been specified in more detail alongside the progress of scientific knowledge. The live loads of the railway traffic on the infrastructure have changed as well.
Quite evidently, the aim has been to increase carrying-capacity by means of increased speeds and axle weights. Increased live loads, however, require the load-bearing capacity of the railway network, including the stability of the embankments and soil foundation, to be validated before any traffic is authorised. It is also noteworthy that a substantial part of the Finnish railway network has been laid directly onto soft-soil roadbed. As well as the aforementioned factors, the increasing need for repairs due, for example, to ground frost damage necessitates rather extensive stability analyses to be conducted across the entire railway network.
Given the scale of the issue at hand, the associated cost impact will be significant. Therefore, it is imperative that both the computations and the resultant remedial schemes are based on maximally accurate and correct assumptions and premises. As part of the joint Tera research programme of the Tampere University of Technology (TUT) and the Finnish Transport Agency, the purpose of the RASTAPA project is to increase the accuracy of stability computation methods and to compare the stability improvement options in terms of economic input-output life cycle assessment.
Full-scale embankment collapse experiment
A full-scale embankment collapse experiment in Perniö, Salo, was an elementary part of the research project. The purpose of the experiment was to determine what would happen in an actual embankment collapse, and to test a number of different metering devices as to their suitability to stability monitoring of railway embankments. Planning and preparation for the experiment started at the beginning of 2009, and the experiment itself was conducted on 20 – 21 October 2009.
For the experiment, a 50-metre long low-profile embankment, with sleepers and rails, was constructed on a soft clayey soil bed, on the site of a previous embankment. The load model represented a very heavy freight train coming to a halt on the track. The "train" consisted of beams representing axles and modified sea cargo containers placed upon the beams. The containers were then loaded gradually with sand using a custom-made belt conveyor several metres in length, which allowed for the keeping of a safe distance from the embankment (Image 1).
Measurement data on the physical phenomena related to the collapse (soil movement behaviour, soil water pressure) was logged using various metering devices, from which we could mention here the automatic inclinometers, robot tachometers, pore water pressure sensors, soil pressure sensors, laser scanners, and the automatic deflection tubes developed at the Department of Civil Engineering at Tampere University (in the research area of earth and foundation structures). The instrumentation may very well have been the most extensive one ever used in such an experiment, and there were approximately 300 individual metering points. Automated logging of measurement data enabled real-time monitoring on-site.
The test started on 20 October 2009 by adding approximately one quarter of the final sand load to the containers. The test continued early in the morning on the following day in the expectation that the embankment would collapse in daylight. On the second day, the majority of Finnish geotechnical engineers were present on the site, with high hopes that they would witness an actual collapse, probably for the first time in their lives. Unfortunately, the event kept everyone waiting, and only a handful of TUT personnel were present when the final collapse took place at about 9.30 in the evening. Image 2 shows a view of the experiment site on the day following the test loading.
Shortly before the collapse, the rate of soil movement started to accelerate. During the collapse, the embankment slumped rapidly, the containers fell onto one side, and the soil adjacent to the embankment became heaved. In the experiment, the final live load of the train was nearly 22 tons per rail metre, approximately one million kilograms in total. This load is 2.5 times the maximum live load per metre allowed on the Finnish railway network. It should be noted, however, that the soil foundation conditions in Perniö may not necessarily have been the poorest possible in the railway network.
In the course of the experiment, large quantities of unique measurement data were gathered, which is being utilised in research currently in progress. For instance, extremely valuable information was obtained about the behaviour of pore water pressure during the different phases of the experiment (Image 3). Furthermore, extensive experience was gained concerning the appropriate use of metering devices suitable for stability monitoring of embankments. Appropriate use of the instruments allows for the detection of an imminent embankment collapse, and corrective action can be taken in time.
Development of computational methods
Stability analyses are routinely conducted mainly through the use of the lamella method based on the limit equilibrium principle. On sites requiring special accuracy, the finite element method can also be applied.
The shear strength of the soil bed can be modelled in both cases using the standard strength, i.e. closed-form shear strength or the effective-stability parameters. The closed-form shear strength is the most commonly used method and also the easiest to apply. However, it does involve problems concerning the determination of the correct computation value for both the sediment under, and adjacent to, the existing embankment. The values obtained using the closed-form shear strength may in many cases be too conservative. The greatest problem related to the use of effective-stability parameters is that the pore water pressure in the soil must be known. A greater pore water pressure results in weaker stability and, thus, a greater probability of soil failure. The problem is that pore water pressure increases in proportion to the plasticisation of clay bodies also where the sediment is not exposed to any actual extra stress. The research project is developing two optimally simple methods, which allow taking into account the pore water pressure due to the plasticisation phenomenon in the application of the lamella method. In stability computation based on the finite element method, this factor can already be included, in principle, in the analysis. The problem here is that the material models available are either partially deficient or very complex, including a high number of computation parameters, some of which are extremely difficult to determine.
The failure of clay bodies is also a highly time-dependent event. In the Perniö load test, for example, even a smaller load would have caused the failure, if the exposure time had been extended correspondingly. The finite element method also allows for the modelling of time-dependent soil behaviour. Image 4 presents an example calculation of the failure load of the Perniö experiment in correlation to time.
Stability remediation
Construction of counter berms is clearly the most commonly used stability remediation solution, even today. The necessity and size of the counter berm is usually determined through stability analysis based on closed-form shear strength. Using this method may, however, result in constructing unnecessarily wide counter berms in cases where excessively conservative choices are made regarding, for example, the depth of standard stability in a thick soft-soil sediment. Hence, a minor variance in the calculation premises may have a significant cost impact. It should also be remembered that the counter berms cause deflection of the railway embankment itself, and may therefore increase repair needs. For these reasons, one of the objectives of the RASTAPA project is to compare the various stability remediation solutions and to determine, for the different conditions, the most suitable reinforcement measure that will produce the optimal economic impact over the life cycle.
