NDT.net • Aug 2005 • Vol. 10 No.8

Screening of Switch Blades using Guided Waves

Dr. M. Evans* and Dr. K. Vine
Guided Ultrasonics (Rail) Ltd.
17 Doverbeck Close, Ravenshead
Nottingham NG15 9ER

*Corresponding Author Contact:
Email: Mark@guided-ultrasonics.com, Internet: www.guided-ultrasonics.com
Published in Proceedings of RAILWAY ENGINEERING-2005, The Eighth International Conference
“Maintenance & Renewal of Permanent Way; Power & Signalling; Structures & Earthworks”, Internet: www.railwayengineering.com

ABSTRACT

Switchblades are considered as a very high priority for non-destructive evaluation for several reasons. Firstly, the blade is inherently weaker than the parent rail due to its reduced cross section, reduced mechanical support and high lateral dynamic loading. Secondly, a rail break at a switch is very likely to cause derailment of a train and thirdly, switches are very prone to rolling contact fatigue.

The inspection of switchblades is inevitably more difficult than plain rail inspection due to the continually changing profile and shape of the blade as it tapers from root to tip. Traditional ultrasonic methods such as U3 (walking stick) cannot reliably inspect the tip section where the head width tapers down from 50% to 0% (of nominal head width). Electromagnetic methods have been employed to determine the pocket length of surface emanating cracks on the running surface for the entire length of the switch. Neither method claims to detect defects which are outside the head section of the rail.

The G-Scan rail screening instrument by Guided Ultrasonics (Rail) Ltd uses guided waves which travel tens of metres down the rail and reflect from changes in cross section. Successful trials of the equipment have been carried out on Alumino-thermic welds, plain rail defects and level crossing rails. It would be attractive to employ the instrument for switchblade testing but this is potentially difficult due to the varying section along the blade. This paper describes experiments carried out on a number of switches with cracks and deliberate saw cuts at various locations. The results are very promising and indicate that defects of 5% cross sectional area and above are detectable in the base of the blade.

INTRODUCTION

In this paper guided wave results are given from a series of defects in a switch blade. The target defect type in this study was a simulated crack (saw cut) emanating from the toe of the blade. The switch was 113A rail profile type C geometry manufactured in 1980. The switch was decomissioned in 2004 and taken to Carillion Rail Test in Crewe specifically for this project. The switchblade was in very good condition showing no sign of wear or GCC. All bolted attachments such as check rail, spacer chogs, stretcher bars and locking mechanisms were still attached as were the pandrol clips and chairs. A photograph of the switch being tested is given in Figure 1.


Figure 1
: The switchblade under test



Figure 2: The cross section of the blade at the defect location

Current NDT techniques are sensitive to surface eminating defects on the running surface but do not inspect the rest of the rail cross section. Even visual inspection cannot reliably detect the presence of cracks in the base and toe of the rail without removing the blade, cleaning and using dye penetrant inspection.

GUIDED WAVE INSPECTION OF RAIL

The G-Scan instrument has been developed by Guided Ultrasonic (rail) Ltd. for the guided wave screening of rail. The instrument is dry coupled to the rail using a mecanical clamping and a pneumatic system. Once attached the guided waves are generated by an array of piezoelectric transducers. These waves travel away from the instrument in both directions and are reflected from changes in cross section of the rail. The reflected signals are received and processed to calculate which modes, directions and distances the reflections have come from. From this information an 'A scan' trace is generated which indicates the amplitude of reflections versus distance from the instrument.

Currently the G-Scan instrument is in the final stages of product approval for use on Network Rail infrastructure. The applications which are currently being considered are for the testing of aluminothermic welds, weld repaired defects, plain rail defects, level crossings and switchblades.

Currently aluminothermic welds are not routinely inspected (other than visual inspection) as the ultrasonic procedure (U6) is widely regarded as being unreliable and cumbersome. However, trials have indicated that weld defects of greater that 10% cross sectional area can be reliably be detected using G-Scan which makes this a very attractive application. In many cases the corrosion defects which occur at level crossings cannot be detected reliably by current NDT techniques but trials have demonstrated that base and toe corrosion defects of as little as 5% cross sectional area can be reliably be detected using G-Scan. Crossings could be inspected without lifting the roadway or stopping road or rail traffic thus reducing the time and cost of inspections dramatically.

The application of guided wave inspection to switchblades is the topic of this paper and this will now be discussed in detail.

APPLICATION TO SWITCHBLADES

An example trace is given below for the switchblade before any defects were introduced. Reflection peaks can be seen at the root and tip of the blade and some smaller reflections from the spacer blocks (chogs). The reflection from the root end of the blade is a single isolated peak as is expected but the reflection from the switch tip is an series of peaks over a 2m range. This corruption of the signal due to the tapered geometry of the switch causing additional mode conversion and dispersion of the guided wave modes. In order to maximise the switch tip reflection, the locking mechanism was unbolted from the tip of the switch. This should not be necessary in all cases as the amplitude of the switch tip reflection is not used for defect severity measurement.


Figure 3: A schematic of the switch and stock rails with the corresponding G-Scan trace. No defect has been introduced at this stage.

The G-Scan trace in Figure 3 shows no refelected signals between the last chog and the switch tip which indicates that there are no detectable defects within this section of the blade. The time taken to perform this test was approximately 10 minutes and no surface preparation of the rail was required.


Figure 4
: A photograph of the toe defect used in this study.
In this case 18mm deep.

TOE DEFECT RESULTS

An example photo of the toe defects used for this study is shown in Figure 4, this is viewed looking vertically down on the rail. The cross sectional loss caused by the defect has been estimated, as a percentage of the total rail cross section at the location of the defect (see Figure 2). These calculations are given in Table 1 along with the reflection coefficient measured from the G-Scan traces.

Defect depth (mm) Cross sectional loss (%) Reflection Coefficieint (%)
11 2.1 5
18 3.4 7
24 5.2 9
30 6.5 11
36 8.4 14
Table 1: The cross sectional loss and reflection coefficient calculated for each defect depth.

These values are also plotted as reflection coefficient vs. cross sectional loss in Figure 5 and the actual G-Scan traces for each defect size are shown in Figure 6. The data point at 45% cross sectional area was a result measured from an actual base defect which was found in a switch which was had just been removed from service.


Figure 5: Reflection coefficient vs. cross sectional loss for a toe defect.

The results show a broadly linear relationship between reflection coefficient and cross sectional loss for the range of the defects tested. The detection threshold has been measured to be 5% reflected amplitude which indicates that toe defects of 3.4% and larger can be detected reliably. The 2% defect gave a measureable reflection but it was too small to be reliably distinguised from the noise floor.

At this location along the blade 3.4% cross section equates to a toe defect of 18mm deep but since the rail cross section decreases towards the tip of the blade, so the physical size of a 3.4% defect decreases in proportion. Further experiments will be carried out with defects closer to the tip of the blade to quantify this effect.


Figure 6
: The G-Scan traces with the corresponding defect depths. The red cursor shows the root of the switch and the blue cursor shows the defect location.

CONCULSIONS

The G-Scan instrument is capable of detecting foot defects in switches of 3.4% of cross section and greater. At the location used for this study that equates to a crack emanating from the toe of the rail for a length of 18mm. There is no other NDT technique currently in use which can detect defects of this type and size reliably. A similar study into the sensitivity of the G-Scan instrument to head defects in swithches is underway and the results of this will be published at a later date.

© NDT.net |Top|