July 2011, Volume 5, No. 7 (Serial No. 44), pp. 596-605 Journal of Civil Engineering and Architecture, ISSN 1934-7359, USA

Using Magnetic Barkhausen Noise Technology and Finite Element Method to Study the Condition of Continuous Welded Rails on the Darwin-Alice Springs Line Ralph (Wei) Zhang1, 2 and Helen Wu2 1. Thermit Australia Pty Ltd, Somersby, N.S.W. Australia 2. School of Engineering, Edith Cowan University, Joondalup 6027, Western Australia Abstract: The Magnetic Barkhausen Noise (MBN) technology is a non-destructive method to measure the neutral temperature of the CWR track. A series of in-field verifications and data comparison on Australian mainline tracks have shown the results from that system are highly accurate and reliable. The system can be an accuracy and cost-effective tool to prevent the potential buckling and break of CWR rails. The physical backgrounds and features of the system are represented in this paper. The Darwin-Alice Springs Line is a newly constructed main line in Australia which is linked from the north and middle of Australia. Originally, this rail line is designed and constructed in a “cost-effective” way to a lower price, and the key parameters are relatively low. To maintain the stability of the CWR tracks in a very harsh environment, some new technologies such as the MBN technology were utilised. From the results of neutral temperature, it is found that the majority of them are very high. Combined with the calculation and finite element analysis, these problems evidenced that it is caused by the low toe load fastening system and high sleeper spacing. After that some suggestions are given to improve the stability of the CWR on the railway line. Key words: Magnetic Barkhausen Noise, finite element method, condition, CWR track.

1. Introduction The 1420 km (880 miles) line between Alice Springs and Darwin in Australia’s Northern Territory was opened on January 2004. Completing the national rail network, the line connects the deepwater port of Darwin with the south of the country and created a freight land-bridge to Asia and the rest of the world. Continuous welded rails (CWR) on concrete sleepers are used throughout and the line is being built for 23 ton axle loads with clearances for double-stack containers and a maximum speed of 115 km/h (71 mph).

Corresponding author: Ralph (Wei) Zhang, senior engineer, research fields: railway track and structures. E-mail: [email protected].

The rationale for the Darwin extension was primarily based on freight flows. From 2008 six mixed Adelaide – Darwin freight services run weekly in each direction, some extends to Melbourne. Within the Northern Territory these were jointed in May 2006 by dedicated mineral trains of manganese ore from Bootu Creek mine. Having secured about 90% of the freight carried between Adelaide and Darwin in 2008 the railway line expected to carry 800,000 tons of general freight in 2009. It was a massive logistical exercise to overcome the difficulties of the track maintenance work on a new railway across some of the most inhospitable terrain in the world, with summer temperature reaching 50ºC and a three-month monsoon season. To keep stability of the CWR track for the keep growing freight revenue, BJB Joint Venture Rail

Using Magnetic Barkhausen Noise Technology and Finite Element Method to Study the Condition of Continuous Welded Rails on the Darwin-Alice Springs Line

Maintenance which is contracted for the permanent of way of this line found they are facing great challenges from high cost, shortage of man-power, hazard environment conditions and without effective and reliable inspection equipments. Serious of derailment has happened caused by the misalignment that related from CWR buckling, and a investigation report that conducted by the Australian Transport Safety Bureau has shown “greatly increased the risk of misalignment and buckling” [1] that caused by the original track laying method and wrong setting of neutral temperature. According to the agreement between BJB and Thermit Australia, Thermit Australia was given permission by BJB to perform in-track measurement using the RailScanTM neutral temperature measuring device along Alice Springs - Darwin line. Testing was conducted on the mainline track on June, 2009.

2. Physical Backgrounds The magnetic Barkhausen noise technology was equipped within a smartly designed machine–RailScan system to carry out the measurement of longitudinal stress distribution in the CWR track. Based on the micro-magnetic theory, every part of ferromagnetic materials contributes to the uniform magnetization. The internal magnetization is not uniform down to the microscopic scale. Many magnetic domains are magnetized in a different direction. The magnetization inside each domain is made up of many atomic moments which in a different direction. The magnetization inside each domain is made up of many atomic moments which are lined up by the action of their exchange force. The schematic drawing of the domain structure is shown in Fig. 1. The crystallites are limited by grain boundaries, the magnetic domains by the Bloch-walls. The domain structure was first predicted by Weiss (1907) and experimentally verified by Barkhausen (1919). In 1932, Bloch described that the boundary between the domains is not sharp on an atomic scale but is spread

597

over a certain thickness wherein the direction of spins changes gradually. The effect of the existence of such Bloch-walls is shown in Fig. 2. Two kinds of Bloch-walls are defined in the figure–the 90°- (BW1) and the 180°-Bloch-wall (BW2). If a field is applied, the two kinds of domains whose magnetization directions are closest to the field direction increase their volume and finally cover the whole volume (Figs. 3b and 3c). If the field is increased further, the magnetizations in each domain rotate from their easy directions toward the field direction, and finally the saturation magnetization is reached (Fig. 3d). This process is typical and one important feature of ferromagnetic substances. They exhibit a fairly complex change in magnetization upon the application of a magnetic field. This behaviour can be described by a magnetization curve possessing three distinct regions (Fig. 4). Starting from a demagnetized state, the magnetization increases (broken curve) and finally reaches the saturation magnetization. In the region “I” the process of magnetization is almost reversible. That is, the magnetizaion comes back to zero upon removal of the field. The domain walls move reversibly and will return to their original position if the field is removed. Beyond this region the processes of magnetization are no longer reversible. In the low-field region “II” the

Fig. 1 Domain structure of ferromagnetic substances: 1-crystallite, 2-grain boundary, 3-Weiss’s domain, 4-Bloch-wall.

598

Using Magnetic Barkhausen Noise Technology and Finite Element Method to Study the Condition of Continuous Welded Rails on the Darwin-Alice Springs Line

Fig. 2 Change of the magnetization vector in Bloch-walls. 90°-Bloch-wall (BW1-upper figure) and 180°-Bloch-walls (BW2-lower figure).

Fig. 3 Magnetization in a [110] direction of Si-Fe. A: domain structure, B-D: orientation process under the action of an increasing magnetic field.

Fig. 4 Hysteresis loop: Magnetization regions I, II and III starting from a demagnetized state and field dependent magnetization processes by Bloch-wall types BW1 and BW2. In the domain of irreversible wall movement the hysteresis loop has the shape of a stair and Barkhausen jumps occur in the form of micro- eddy currents (enlarged increments on the left side of figure).

domain wall movement occurs irreversibly as walls overcome barriers presented by pinning centres in the

microstructure. In the highest field region “III” only little domain wall motion occurs and further magnetization is predominantly due to rotation of the magnetization vector within individual domains. It is essential for further understanding to know the interactions of the different Bloch-walls during the magnetization. Bloch-walls of the first kind (BW1) are walls which separate magnetic domains in which the residual stresses do not vanish in adjoining domains. In iron all (100)-90° BW’s are of this type. Bloch-walls of the second kind (BW2) produce only residual stresses within the Bloch-wall itself. In iron all 180° Bloch-walls and the (110)-90° walls are of this type. BW2’s separate areas with the same magnetostrictive behaviour and consequently no elastic energy will be changed during their movement. Therefore BW2 do not interact with macro stresses RS1 and micro stresses RS2. Beside BW1 all rotation processes (RP) are stress sensitive. BW1, BW2 and RP exist at different magnetic field strengths to the magnetization process (Fig. 4). Following previous work, in polycrystalline ferrous materials BW2 contribute to the magnetization in a dominant way around the coercivity field strength Hc. Outside of this field region BW1 becomes more and more dominant. If magnetization becomes saturated, BW density decrease and finally remagnetization processes take place by RP’s (Fig. 3). Because of these micromagnetic interactions with the microstructure, parameters can be recorded with one set of micromagnetic quantities. Hereby stresses are measured by values mainly determined by BW1. For the micro-structure, the magnetization process consists of many small discontinuous flux changes, which correspond to the irreversible movement and flopping of Bloch-walls and domains. During this process, micro eddy currents are induced in the volume (Fig. 4), enlarged increments of hysteresis loop. These so-called Barkhausen jumps can be measured by an appropriate detector. Along the hysteresis loop, the MBN shows two maxima around the coercivity field strength Hc. The effective value of

Using Magnetic Barkhausen Noise Technology and Finite Element Method to Study the Condition of Continuous Welded Rails on the Darwin-Alice Springs Line

MBN is generally recognized as a measure for quantitative non-destructive evaluations. It contains the generated pulses and has a noise-like spectrum. The stress sensitivity of BW1 and RP can be used to measure the longitudinal stress in the rail. In any load stress applied, the permeability for the applied magnetic field changes. Tension leads to an increase of the permeability. The higher the longitudinal stress, the higher the increase of permeability. The rail becomes more and more easily magnetisable. When compression stress is applied, with increasing compression the rail becomes magnetic hard. The permeability for the magnetic field decreases. This so-called Villari-effect is measurable by the help of the MBN signal. The MBN contains the eddy currents of the stress sensitive parameters BW1 and RP and depends significantly of the longitudinal stress in the rail. Tension increases the amplitude of the MBN, whilst compression leads to decreases. The higher the longitudinal stress, the higher the signal amplitude of the MBN. MBN has the sensitivity with stress condition in metal materials has been found by the physicists since the first half of last century, and it has been equipped with different types of devices to measure the stress condition in CWR rails for decades.

3. Functional Principle The RailScan system is designed to measure the stress distribution and the Neutral Temperature of continuously welded rails (CWR). The device operates by means of non-contact gauging using the magneto-elastic principle and allows fast measurement and documentation of the actual Neutral Temperature of rails. The longitudinal stress and the Neutral Temperature are determined by measurement of characteristic magnetic values. The rail is magnetized by the application of an alternating magnetic field. Interaction of the magnetic field with the magnetic microstructure is orientation dependant and can be measured with an appropriate probe. The measured signal contains the pulses

599

generated in the rail and has a noise-like spectrum. The amplitude of this so-called Magnetic Barkhausen Noise (MBN) depends significantly of the longitudinal stress in the rail. Tension increases the amplitude of the MBN, whilst compression leads to decreases. The higher the longitudinal stress, the higher the signal amplitude (Fig. 5).

4. Measurement Device and Operation The RailScan system consists of a manually operated railcar, a central unit, and a pair of probe (Fig. 6). The central unit contains the computer-operated measuring electronics. A separate battery provides the power supply for the equipment. The probe consists of two yokes that are pressed around the rail-head with the help of a Bowden wire and springs. The probe, which is replaceable, is geometrically adapted to the rail type. The rail temperature is measured with an integral infrared thermometer. Before performing the measurement the rail will be marked. The measurement is made after positioning the device above each mark in turn. The rail is energized by a magnetic field in the acoustic frequency range, and the level of the Magnetic Barkhausen Noise at the surface of the energized region is measured. One measurement series consists of 50 marked points. Evaluation is performed automatically after the measurement. The measurement result can be stored and visualized on a PC via data transfer. Data can be read and processed by Microsoft Office format.

Fig. 5 Stress dependence of the MBN reconstructing a MBN-stress relationship.

used

for

600

Using Magnetic Barkhausen Noise Technology and Finite Element Method to Study the Condition of Continuous Welded Rails on the Darwin-Alice Springs Line Calibration Curve of One Steel 50kg/m Plain Carbon Rail

Magnetic Parameter [Beta]

1.6

1.4

1.2

1.0

0.8

0.6

0.4 -80

-60

-40

-20

0

20

40

60

Longitudinal Load Stress [MPa] Stress (x) vs. Beta (y)

Fig. 7 Calibration curve of One Steel’s 50 kg/m plain carbon rail.

Prior to the measurements being taken, a calibration curve for One Steel’s 50 kg/m plain carbon rail was prepared in Thermit Australia’s laboratory using three rail samples. The measurements taken in track were compared against this curve. The calibration curve from these three rail samples are shown in Fig. 7. All the neutral temperature of the RailScan measurement are processed and calculated based on the above calibration results.

6. Data Analysis After completing the measurements the raw data are downloaded to a laptop for further evaluation. The final results were obtained by evaluating and plotting the measured values of the magnetic parameter and rail temperature vs. the longitudinal coordinate and measuring point number and further by depicting the load stress determined by means of the averaged magnetic parameters and calibration curve (Fig. 8). The neutral temperature is calculated by means of the following equation:

TN =

σ E ×α

+ TRail

Magnetic parameter (Beta) triangles up

5. Calibration

50 4 40 3 30

2 20

1

10

0

Rail temperature [degree C] black dots

5

RailScan unit.

0 0

10

20

30

40

50

Measuring point number

1.6

Magnetic Parameter [Beta]

Fig. 6

1.4

1.2

1.0

0.8

0.6

0.4 -80

-60

-40

-20

0

20

40

60

Longitudinal Load Stress [MPa]

Fig. 8 Stress sensitive parameter and rail temperature, plotted vs. longitudinal coordinate (up), longitudinal load stress determined by means of the rail specific calibration curve (bottom), CWR module at 2551 km Left rail on the Alice Springs-Darwin Line.

Using Magnetic Barkhausen Noise Technology and Finite Element Method to Study the Condition of Continuous Welded Rails on the Darwin-Alice Springs Line

Where: TN –The neutral temperature before de-stressing Trail – Rail temperature at the stress-free condition α –Thermal coefficient of rail, using 11.5×10-6 (m/°C) E – Elastic modulus, using 2.07×105 MPa All the neutral temperature results are calculated following the above procedure and methodology. Since the year 2006, more than 200 km CWR track measurement has been done on Australian heavy haul operated railway lines. The in-field measurement results have been compared with some widely acknowledged method/equipment to confirm the accuracy and reliability of the RailScan system. The results represented in this paper are from two in-field measurements in 2008 on two Australian heavy haul railway lines. One is on the Australian Rail Track Corporation’s (ARTC) North Coast Line (from Sydney to Brisbane) in Gloucester, and the other one is Sydney-Broken Hill line in Ivanhoe, Menindee, and Broken Hill. A series of in-field verifications and data comparison on Australian heavy haul tracks (North Coast Line and Broken Hill Line) have shown the results from that system are highly accurate and reliable. The system can be an accuracy and cost-effective tool to prevent the potential buckling and break of CWR rail. 6.1 RailScan Results and Further Analysis The in-field measurements were conducted on Alice Springs-Darwin Line from 2477 km to 2747 km, close to the stations of Berrimah, Adelaide River, and Pine Creek. It is a single railway line. The gauge is the 1435 mm international standard width. The rail type of the track is One Steel’s 50 kg/m Plain Carbon rails, which are manufactured later than year 2000. For the sleepers, pre-stressed Austrak concrete sleepers with 2.4 m length are installed and assembled with the rails by Pandrol “Fast clip” (as shown in Fig. 10). The spacing of the concrete sleepers is approximately 720 mm (about 1389 sleepers/km) on

601

straight track and curves of 1200 m radius and over, and 700 mm on tighter curves [4]. The ballast thickness is 150 mm minimum and generally in good condition. 17 of the overall 20 modules are tangent track or its majority part is tangent track (such as tangent track and partly of its adjacent transition part of a curve). The other 2 modules are tangent track combined with curve track (left and right rail of kilometrage 2747.5 km). Among these 20 modules, 6 of them are short section of tangent track (length from 40 m to 120 m) between two curves; 9 modules directly connected with curve track on one of their ends; Two of them is located behind concrete girder bridge. The neutral temperature results from RailScan measurement for the 20 CWR modules are tabulated in Table 1. Obviously, one of the significantly impression from Table 1 is generally the neutral temperature values are high and significantly higher than the original neutral temperature which is about 48°C. To confirm the results from RailScan again, BJB compared the results with the “cutting rail and gap measurement method”, the CWR module at 2533 km was been cut (Fig. 9). The rail is the left side rail of the track and the unclip length is 100 m. The rail temperature at cutting is 24°C. After vibrating the final rail gap is measured as 63 mm. Put the data into the equation. ΔL x TN = Trail + αLx Where: ∆Lx – The changing of length before and after cut the rail Lx – The length of CWR module Trail – Rail temperature at the rail cut

TN = Trail +

0.063 ΔL x = 24 + = 78.8°C 11.5 × 10 −6 × 100 αLx

The result is about 10°C higher than the result from RailScan measurement. This CWR module was stress readjusted last July,

602

Using Magnetic Barkhausen Noise Technology and Finite Element Method to Study the Condition of Continuous Welded Rails on the Darwin-Alice Springs Line

that was one year ago. The original set-up neutral temperature is 48°C. The results from “Cutting rail and gap measurement method” and RailScan has a 9.8ºC deference. Because the measurements are carried out at different ambient condition. RailScan measured at 33.7ºC and “Cutting rail method” at 24°C, the difference of rail temperature is 9.7ºC. It is very close to the difference of neutral temperature which is 9.8ºC. Studying the CWR resistant load from the fasteners, the basic concept of fastener design is the longitudinal resistance from the fastener must be higher than the ballast resistance on the sleeper, hence, to ensure the rail will not creep on the top surface of the rail seat of sleeper. Therefore, the sleeper, fasteners, and ballast can work together as a frame to provide adequate resistant force for the stability of CWR track. Table 1

From the designer of the “non-insulated FAST clip” that specially designed for this line, the nominal vertical hold-down force (toe load) per clip shall be 9 kN with a total force of 18 kN per rail seat (As shown in reference [5]). For the longitudinal axial resistance of concrete sleeper, according to the study carried out by Dr. Arnold Kerr in USA, and using the references that tested by Indian Railway for the standard gauge and concrete sleeper track, if the ballast is in regular good condition, the longitudinal ballast resistance are 12.93 kN and 13.28 kN for 1310 sleepers/km and 1540 sleepers/km respectively. Hence, it can be predict that, for the Alice Springs – Darwin line, for its 1389 sleepers/km track condition, the longitudinal resistance that provided by the ballast on each concrete sleeper should be around 13 kN.

The neutral temperature results from RailScan measurement on the Darwin - Alice Springs Line.

Kilometrage of CWR Module

Length of CWR module (m)

Location of CW rails

Rail(ambient) temperature (ºC)

Neutral temperature (SFT) results from RailScan (ºC)

2724.8

400

Right

39.1

60.8

2747.5

500

Left

43.3

68.5

2747.5

500

Right

44.4

67.8

2508.5

400

Left

24.4

58

2508.5

400

Right

31.9

59

2510

200

Left

37

64

2510

200

Right

40.1

58.0

2510.5

120

Left

37.5

72.5

2510.5

120

Right

40.1

75

2511.5

40

Left

41.7

66

2511.5

40

Right

41.0

60.1

2531

350

Left

37.5

71.5

2531

350

Right

35.5

70

2532

350

Left

24.9

70

2532

350

Right

31.5

70

2533

50

Left

33.7

69

2533

50

Right

36.2

63

2551

400

Left

40

57.6

2551.5

250

Left

40.8

54.5

2551.5

250

Right

42.4

49.4

* The kilometrage of a CWR module is defined by the location of the middle of the module. ** Look towards Darwin direction, left side is defined as the Left rail, the other side is Right rail.

Using Magnetic Barkhausen Noise Technology and Finite Element Method to Study the Condition of Continuous Welded Rails on the Darwin-Alice Springs Line

Fig. 9

603

The CWR module of 2533 km.

which is 13 kN for concrete sleeper with 720 mm spacing. That means the long rail can creep longitudinally along the rail seat of sleeper, when the rail temperature is different with the neutral temperature. 6.2 Numerical Study Using Finite Element Method

Fig. 10 Austrak concrete sleeper assembled with Pandrol Fast clip fasteners.

Studying the resistance on the two rails that provided by the four “Non-insulated FAST Clips” on the distance of one concrete sleeper spacing: Rf = Ft × 4 × μ Where: Rf – The resistance that provided by the fasteners on the rails at one concrete sleeper spacing Ft – Toe load of one fastener μ – Friction coefficient factor, for the steel to steel contact between rail and steel fastener, μ = 0.2. Rf = Ft × 4 × μ = 9 × 4 × 0.2 = 7.2 kN The 7.2 kN resistance load is significantly lower than the requirement of per sleeper ballast resistance

To understand the amount of influence of ambient temperature change on longitudinal stress distribution on the CWR track on the Alice Springs-Darwin line, a finite element model has been created to carry out the numerical study (Figs. 11 and 12). The widely used finite element software ANSYS is used to create the FEA model. The literatures [4, 10, 13] are referenced for this numerical study. To ensure the high accuracy of the FEA model and minimise the influence of its boundary condition, a full scale model of the 100 m curve track is created. Rail and concrete sleepers are simulated using beam element with the cross-section input data that are same as its original shape profile. Non-linear node-to-node contact elements are applied to simulate the contact and interaction of “Non-insulated FAST Clip” fasteners and steel rails. These types of contact problems involve small relative sliding between contact surfaces, the geometric nonlinearities are utilised. The toe load of fasteners are input as the “normal stiffness” kN. The “sticking stiffness” kS represents the stiffness in the tangent direction and a input data 0.22 is chosen as the coefficient of friction between the fastener and steel rail plus coefficient between steel and rubber pad.

604

Using Magnetic Barkhausen Noise Technology and Finite Element Method to Study the Condition of Continuous Welded Rails on the Darwin-Alice Springs Line

Lateral and longitudinal resistances of ballast on each concrete sleeper are simulated by using the combined spring-damper with different specified equivalent input stiffness values. The torsion capability of the spring-damper element is fixed. These elements are directly connected with the beam element of concrete sleepers in the longitudinal and lateral directions. The boundary conditions of the CWR rails are assumed to be fixed-fixed on one end and fixed in the Y direction on the other end [4]. The spring elements that representing the ballast lateral resistance are fixed on their field side directions, which same as the condition of ballast where outside of concrete sleepers. For the nonlinearities of the majority materials in the FEA model, multi-liner Kinematic hardening and von Mises equivalent stress theories are followed. The thermal load that is equal to 10ºC was applied on the end of the 50 kg rails. From the FEA solution it can be found that the results of the longitudinal stress changing along the two rails under the thermal load is averagely equal 8.2ºC neutral temperature change. This is confirmed that the 9 kN lower toe load of the fastener could not provide enough resistance load to control the originally set-upped neutral temperature.

And it can be seen as another evidences to show the high accuracy of the RailScan system. Based on the previous experience, the short tangent CWR module between two curves usually has higher neutral temperature value (For example, the two modules at 2510.5 km). Our previous studies have found that – studying the neutral temperature of each point along a curve, it can be found that the neutral temperature is not evenly distributed. From its moving average the neutral temperature is lower at the circular curve and higher at its transition curves and significantly high at the adjacent tangent track (see detail in Ref. [9]). Bending stress of the rail caused by the curvature while installed on the sharp curves can influence the thermal stress distribution in the longitudinal direction and it seems it is sensitive with the radius of the curve, the smaller the radius the greater the decrease in longitudinal stress (see detail in Ref. [9]). For the 2533 km CWR module, the curvature of curve track is small, the influence of neutral temperature caused by the bending stress in the longitudinal direction is not significant compare with the sharp curve tracks (for this CWR model, the value is about 1.5ºC which is determined by another finite element model’s study, its method see detail in Ref. [9]).

Fig. 11 Combined spring-damper element (COMBIN14) used to simulate the ballast resistance on concrete sleeper.

Fig. 12

Finite element model of the 100 meter length CWR module of 2533 km at Alice Springs-Darwin line.

Using Magnetic Barkhausen Noise Technology and Finite Element Method to Study the Condition of Continuous Welded Rails on the Darwin-Alice Springs Line

7. Conclusions

[2]

Plenty of results from RailScan measurements and testing works in the recent years have shown that the

[3]

RailScan system is a cost-effective and sophisticated equipment, it provide accurate and reliable data to monitor the condition of neutral temperature and stress distribution in a specified railway section, and it is a very useful tool for track maintenance and

[4] [5]

research on CWR. For the railway track on the Alice Springs-Darwin line, on many section, the structural strength of the

[6]

CWR track are relatively low, because of the higher sleeper spacing (720 mm) and lower toe load (9 kN) of “non-insulated FAST Clip” can not provide enough resistant force to control the creep of the long rail

[7] [8]

under the very harsh environment. Therefore, the neutral temperature of the CWR track can be changed during one day from cold night to very hot noon time.

[9]

It can be resourced as the major reason that caused misalignment of many of the CWR tracks at the Alice

[10]

Springs-Darwin line. To ensure the stability of the Alice Springs-Darwin railway line, there is no doubt that the CWR track system need to be strengthened. Increasing the toe load of the fasteners is the basic way for this strengthening work. In addition, considering the condition of the resistance of fastening system is lower than the ballast resistance, the neutral temperature should be designed and determined based on the fastener resistance for the track maintenance work. Moreover, the nonlinear finite element model is a very useful tool to carry out the analysis of the CWR track at different specified locations. Some further parameter studies will be carried out when the FEA model is further improved.

References [1]

N. Adlam, Wonky train tracks no laughing matter, Northern Territory News, June 29th, 2009.

[11]

[12]

[13]

[14]

[15]

[16]

605

ANSYS Co., Theory reference for ANSYS and ANSYS workbench, Release 11.0. ANSYS Inc. and ANSYS Europe Ltd., 2007. Y. Bao and E. Barenberg, Three-dimensional nonlinear stability analysis of tangent continuous welded rail track under temperature and mechanical loads, Transportation Research Record 1584, 1997, pp. 31-40. S. Chikazumi, Physics of Magnetism, John Wiley & Sons, New York, 1964. T. Dawson, N. Underhill and R. Niall, Track and formation design for the Alice Springs-Darwin railway project, in: Core Conference on Railway Engineering, June 20-23, Darwin, 2004. S. Douglas and K. Sherwood, The supply of a cost effective concrete sleeper and rail fastening solution for the Alice Springs to Darwin railway, in: Core Conference on Railway Engineering, June 20-23, Darwin, 2004. A. Kerr, Fundamentals of Railway Track Engineering, Simmons-Boardman Books, Inc., USA, 2003, p. 229. A. Kish, Guidelines to best practices for heavy haul railway operations–Infrastructure construction and maintenance issues, International Heavy Haul Associations, IHHA, June, 2009, pp. 71-87. Metal Elektro, Neutral rail temperature measurement with RailScan, Hungary, 2004. R. Moller, P. Radmann and R. Zhang, Using magnetic Barkhausen noise technology and numerical method to study the condition of continuous welded rails on Australian heavy axle track, in: 9th International Heavy Haul Conference, Shanghai, China, June 2009. P. Wang and X. Liu, The Theory of Calculation and Design Method of CWR Track within Turnout, The Xinan Jiaotong University Publisher, 2007, Chengdu, China, pp. 230-235. (in Chinese) A. Wegner, Non-destructive determine-action of the stress free temperature in CWR tracks, in: International Rail Forum, Madrid, November, 2004. A. Wegner, Prevention of track buckling and rail fracture by non-destructive testing of the neutral temperature in cw-rails, in: 8th International Heavy Haul Conference on “High Tech in Heavy Haul”, Kiruna, Sweden, June, 2007. Y. Yang and A. Gu (2006). Using ANSYS software to study the stability of CWR track, Railway Engineering, Sep., 2006. (in Chinese) W. Zhang, Installation and maintenance method of the railway turnout, The Chinese Railway Publisher, Beijing, 1998. (in Chinese) W. Zhang, Across-stations continuous welded rails railway, The Chinese Railway Publisher, 2000. (in Chinese).