IDENTIFYING the precise location of variations in track stiffness can inform railways and infrastructure managers about potential sources of track degradation and the appearance of forces and moments in track structure well before they become an issue, helping to improve the effectiveness of track maintenance activities.
Studies have found that track stiffness variations at wavelengths of less than 3m are a dominating factor in the generation of environmental vibrations compared with ordinary track, while large bending moments associated with hanging sleepers and soft track are an important source of rail crack growth which can result in rail breaks.
Several methods are used by a handful of railways around the world to detect these phenomena, yet they often rely on the use of specialised machines which are expensive to buy and slow to carry out the tests, which can be disruptive to everyday traffic.
Swedish infrastructure manager Trafikverket is currently testing a new method of detecting track stiffness and deflection measurements by combining these tests with those already carried out for measuring track geometry by a track recording car.
Trafikverket is working in partnership with Infranord on the project, which has prepared three track recording cars for the tests which are estimated to cost 90% less to carry out than those by a dedicated vehicle for stiff track measurement.
A patent is pending for the method which is based on the fact that a longitudinal level measurement of a track is subjected to a loaded axle comprising two parts.
The first part relates to level variations found because of irregularities present in the unloaded track, while the second part relates to the extra deflection which is created by the loaded axle. An inertial measurement of level (sIn(x)) at position x should according to standards and common practice always be loaded (sL(x)), and thus consists of two parts - unloaded track geometry (sU(x)) and the effect of loading (w(x,x)) according to the following equation.
sin (x) = sl (x) = su (x) + w (x)
Level can also be measured using a chord/versine method and the simplest form is the three-point chord method.
Figure 1 shows a theoretical example of the different parts of the equation.
In part A the unloaded level is shown as a solid blue line. When a sinusoidal deflection is added (the black dash line), the sum of these is the loaded level (red dash line). Part B illustrates a single axle loading the track at position 0. A three point chord measures the level when the loaded measurement wheel is placed in the middle (C2) and the outer, possibly lightly loaded measurement wheels (C1, C3) (or non-contact lasers) are placed at the same distance from the middle wheel.
Studies have found that it is possible, with the help of sensor fusion, to extract the deflection by using one inertial and one chord measurement system for longitudinal level. Deflection is further related to track stiffness by the wheel-rail force.
During the modernisation of the track recording car fleet by Infranord, which is the contract holder for periodic measurements in Sweden, their three Plasser & Theurer EM80 track vehicles were rebuilt and renamed Infranord Measurement Vehicle (IMV) 100.
The rebuild involved installing a modern inertial measurement system to allow for flexibility while complementing the old mechanical three-point chord. So without planning for track deflection/stiffness measurements per se, the necessary building blocks for the innovation are in place.
Currently measurements are being used in two projects. In the EU-research project Rivas, Trafikverket is studying the combined effect of level and stiffness on environmental vibrations. In another project Trafikverket is looking at implementing results from the EU-research project Innotrack, where the main focus is seasonal variation of track stiffness on the iron-ore line in northern Sweden.
For this project track stiffness/ deflection was measured on track section 118 on three occasions in 2012 - February, April and August - when conditions varied greatly; in February ground frost often penetrates deeper than 2m depending on material and frost insulation; in April ground frost starts to thaw, and if snow has melted, the ballast layer can be a few degrees above freezing; and in August no ground frost remains.
Trafikverket has made track recording data of the section of line available from 2009 to 2013 which makes it possible to follow trends in longitudinal level and to make comparisons with the deflection measurements from 2012.
In total 26 longitudinal level measurements are available and the degradation can be clearly viewed from Figure 2 where the top-to-top value (maximum value - minimum value) is calculated at the defect for each measurement. Degradation is moderate up until early 2012 when the degradation rate increases. There is an improvement in the autumn of 2012. However, there is no record of any maintenance in the official database of Trafikverket.
Deflection measurements during 2012 are shown in Figure 3. Already in February the top-to-top deflection (1-25m) is 6mm. The trend from this point shows fast degrading track with as high as 13mm deflection (1-25m) of the rail in August. This kind of defect is rather hard to maintain in a planned way, as the degradation rate changes and very soon results in a need for acute maintenance. Track deflection measurements reveal the defect as a "suspect" and already at the start of faster degradation.
So far the method has shown that there is no obvious correlation of bad spots with rapid degradation and infrastructure assets such as switches and crossings, bridges and insulated rail joints. This indicates that problems with drainage and soft soil are a more common cause than expected. The development of a low-cost stiffness measuring method creates new opportunities to find the root causes. The next step for Trafikverket will be to identify the spot failures that degrade rapidly on the lines with high annual traffic. Correlating these spot failures with information about stiffness will enable the best possible remedial action to be taken. Ideally, the most important lines should be measured with the results incorporated into the existing track condition data bank.