PAPER Hollow Ultrasonic Crack Detection for Conventional Railway Vehicles Kazunari MAKINO Researcher, Jiro YOHSO Senior Researcher, Hiroshi SAKAMOTO Hiromichi ISHIDUKA Assistant Senior Researcher, Senior Researcher, Laboratory Head, Vehicle & Bogie Parts Strength Laboratory, Vehicle Structure Technology Division An ultrasonic testing technique was studied for a hollow axle with a 30 mm bore diameter manufactured on a trial basis for conventional railway vehicles. To compensate for the decrease in crack detection sensitivity due to the small bore diameter, a piezocomposite focal probe was designed. It has been demonstrated that the ultrasonic testing equipment thus developed could detect artificial flaws with a depth of 0.15 mm at the non-fitted central part and those with a depth of 0.3 mm at the inner end of the wheelseat (fitted part). The accuracy of axle inspection for conventional railway vehicles equipped with such hollow axles will match that of Shinkansen vehicles. Keywords: conventional railway vehicles, hollow axle, ultrasonic test, piezocomposite focal probe 1. Introduction On railway vehicles, every component is inspected in compliance with a fixed standard in order to prevent breakdown or damage. Since a failure may lead to a serious accident, axles need to be inspected at the regular intervals specified. For Shinkansen vehicles, hollow axles with a bore diameter of 60 mm have been used to reduce the unsprung mass since the debut of the Series 300 Shinkansen cars in 1992. These axles are inspected automatically through the bore with ultrasonic angle beams (see Fig. 1). On the other hand, solid axles are used mostly for conventional railway vehicles, though there are some cases where hollow axles with a bore diameter of 60 mm are used, for example on JR Hokkaido Series 183 and JR Shikoku Series 2000 diesel railcars. Recently, there have been calls for further automation and labor saving in the inspection of conventional railway vehicles. If fine cracks in an axle can be detected with the wheelset assembled, part of the magnetic particle test that is carried out by dislocating or dismounting wheels and other fittings can be omitted, thereby Fig. 1 60 mm diam. Angle-beam technique Brake disk (or gear) head (Scanning in axial direction with rotation) Bore Wheel Transducer : : Ultrasound Hollow axle ultrasonic testing technique for Shinkansen vehicles improving axle inspection efficiency. The technique 1) that uses grazing SH waves to detect fine cracks is difficult to apply because a high viscosity couplant is needed in order to transmit ultrasound. Therefore, based on over 10 years experience with hollow axles used on Shinkansen vehicles, we developed a hollow axle and ultrasonic testing equipment that are applicable to conventional railway vehicles. This paper provides the results of tests on hollow axles, manufactured on a trial basis, carried out using this equipment. 2. Solid axle ultrasonic testing technique for conventional railway vehicles The solid axle ultrasonic test for conventional railway vehicles is now carried out with a combination of normal beam, longitudinal wave angle-beam, and angle-beam techniques. An outline of the method is described in Fig. 2. The normal beam technique positions a normal probe on an axle end face and tests the axle with ultrasound parallel to the axial direction. However, it cannot test the positions where the ultrasound does not reach, such as the outer end of the wheelseat. In addition, cracks at a depth of more than 10 mm at the non-fitted central part might Difficult part to inspect using normal beam technique Normal beam technique Wheelseat Gear seat (or brake disk seat) Fig. 2 Non-fitted central part Brake disk (or gear) Angle-beam technique Inner end Flaw Wheel Outer end Wedge Longitudinal wave angle-beam technique Flaw depth Solid axle ultrasonic testing technique for conventional railway vehicles 78 QR of RTRI, Vol. 46, No. 2, June. 2005
not be detected because of the beam spread or attenuation of the ultrasound. The longitudinal wave angle-beam technique positions a wedge at an appropriate angle between a normal probe and an axle end face according to each part to be inspected such as the journal, the inner and outer ends of the wheelseat, the gear seat and the non-fitted central part. Then, each position is tested with an ultrasonic longitudinal wave, the refraction angle of which is normally 4 to 28 degrees. Figure 2 also shows the test set-up for the inner end of the wheelseat. The longitudinal wave angle-beam technique can detect shallower cracks than the normal beam technique, but the inspection accuracy is limited because the path length (the propagation distance to the ultrasound reflection source) is longer than that of the angle-beam technique explained below. The automatic ultrasonic equipment for testing solid axles using the longitudinal wave angle-beam technique has been introduced at some workshops for conventional railway vehicles. The angle-beam technique transmits an ultrasonic shear wave whose angle of refraction is larger than that of the longitudinal wave angle-beam technique (normally 37 to 55 degrees) from an axle surface to test the opposite side of the axle surface. The path length is shorter than those of the normal beam and longitudinal wave anglebeam techniques, and the shear wave with a shorter wavelength is used, therefore the inspection accuracy is good. Since the probe cannot scan the positions where parts such as a wheel, brake disk and gear are fitted, however, probes at various angles of refraction are needed to make the ultrasound transmit from the surface where there are no fittings or differences in level. As mentioned above, there are both merits and demerits in each solid axle testing technique for conventional railway vehicles. To improve the efficiency and accuracy of the axle inspection drastically, it is necessary to discuss the introduction of hollow axles on both conventional railway and Shinkansen vehicles. 3. Bore diameter of axles for conventional railway vehicles journal diameters on conventional railway vehicles are mostly either 110 mm or 120 mm. If an axle journal with a 110 mm diameter is bored to a diameter of 60 mm, as has been experienced in Shinkansen axles, then the bending stress at the journal is about 1.10 times that of a solid axle. This does not pose a serious problem. However, the deflection and its angle at the journal will increase because of the decreased stiffness of the axle as a whole. Then, fretting damage at the part where the axle is fitted with the journal bearing inner ring or the back cover may increase as the axle rotates. It is desirable to make the bore diameter small, in order to minimize fretting damage. In addition, in the case of a journal bearing structured to install the front cover to the axle end face with bolts, it is not possible to machine the axle to a bore diameter of 60 mm because there are bolt holes at the axle end face. For these reasons, we adopted a 30 mm bore diameter for hollow axles on conventional railway vehicles (see Fig. 3). Stamp Fig. 3 Bolt hole end face Front cover Fretting damaged part Journal bearing Bolt 30 mm diam. 110 mm diam. End cover Outer ring Inner ring Schematic view of hollow axle for conventional railway vehicles 4. Ultrasonic testing probe improvements 4.1 Improvement of ultrasound directivity Bore The probe that is most commonly used for the ultrasonic testing of hollow axles on Shinkansen vehicles has a flat transducer with a 50-degree angle of refraction. Figure 4 shows a schematic view of the ultrasound propagation oscillated with a flat transducer. When the bore diameter is 30 mm, the ultrasound is focused on and radiates from a point behind the part to be inspected (axle surface). In order to assure inspection accuracy, it is necessary to shape the transducer into a three-dimensional curved surface and focus the ultrasound on the part to be inspected. We therefore designed the shape of the transducer to focus the ultrasound on the part to be inspected 2). The design concept of the focal transducer in Fig. 5 is outlined below. (1) A lattice at the bore surface of an axle is assumed. (2) A straight line that connects each lattice and the focal point at the part being inspected, which is the ultrasound propagation path, is defined. The lattice and the straight line guide the angle of refraction. (3) The angle of incidence of the ultrasound is calculated from the ratio of the ultrasound velocity in an axle to that in a wedge and the angle of refraction defined in (2). (4) A straight line is defined from each lattice toward the transducer with the angle of incidence calculated in (3). (5) When the ultrasound propagates from a focal point to- Flat transducer center 30 mm diam. bore Ultrasound focal point Ultrasound radiation Part to be inspected (axle surface) (a) Bore diameter: 30 mm Flat transducer center 60 mm diam. bore Ultrasound focal point Part to be inspected (axle surface) (b) Bore diameter: 60 mm Fig. 4 Schematic view of ultrasound propagation oscillated with flat transducer QR of RTRI, Vol. 46, No. 2, June. 2005 79
Optimized shape of transducer (focal transducer) Flaw echo Press-fit echo Flaw echo + Press-fit echo Wedge Angle of incidence Lattice Bore surface surface (part to be inspected) Normal transducer Angle of refraction Focal point Fig. 5 Design concept of focal transducer ward the transducer via each lattice, a set of the positions where each ultrasound arrives at a certain time is determined as the temporary shape of the transducer. (6) The phase shift by refraction or reflection is calculated when the ultrasound is oscillated from each position of the transducer shape determined in (5). The temporary shape of the transducer is corrected to agree in the phase of the received wave at all positions of the transducer, and the final shape of the transducer is determined. 4.2 Control of ultrasound wave shape Piezoelectric ceramics such as lead zirconate titanate (PZT) are conventionally used as the material for transducers. In this case, the damping effect of the transducer is small and the flaw echo becomes a waveform accompanied by several waves. On the other hand, fitted parts such as a wheelseat generate the press-fit echo consisting of multipeak waves, because there are very small cavities caused by the surface roughness of the axle and fittings that make the ultrasound reflect. When a flaw is small, it is difficult to distinguish the flaw echo from the press-fit echo. If the waveform is controlled and the ultrasound accompanying a small number of waves can be transmitted, the shape of flaw echoes becomes sharp, making it more easily separated from press-fit echoes. It is possible to control the waveform by such methods as placing damping material at the back of the transducer or making adjustments to electrical circuit damping, but the sensitivity of the transducer decreases at the same time. We therefore tried to increase the sensitivity to flaw echoes in comparison with press-fit echoes by using a piezocomposite transducer that combines the piezoceramics and a polymer (damping material) and by making the number of accompanying waves nearly one, as shown in Fig. 6. The piezocomposite transducer is composed of piezoceramics. Its edges, which are several dozen micrometers in length, are wrapped and formed by a polymer in a latticed shape. It can reduce the number of accompanying waves with minimizing the decrease in sensitivity. This transducer is flexible, can be easily formed into any shape, and is suitable for the focal transducer mentioned above. In addition, its acoustic characteristics are close to those of a wedge for an angle probe that is generally made of acrylic resin, and increases the sensitivity of an angle probe in particular. Piezocomposite transducer Fig. 6 Transducer types and flaw and press-fit echo waveforms 4.3 Piezocomposite focal probe Figure 7 shows the piezocomposite focal probe manufactured on a trial basis, and Table 1 gives its specifications. A test piece on which several kinds of artificial flaws were machined was prepared and tested with a ceramics probe, a piezocomposite probe and a piezocomposite focal probe, each having a 40-degree angle of refraction. A comparison of the heights of flaw echoes has proved that the flaw sensitivity increased about 6 db when a piezocomposite probe was used instead of a ceramics one, and about 6 db more when a focal one was used. Polymer Piezoceramics 50 µm Fig. 7 Piezocomposite focal probe and its transducer Table 1 Specifications of piezocomposite focal probe Item Transducer material Transducer size Nominal frequency Nominal angle of refraction Shape of contact surface Specification Piezocomposite 15 mm x 12 mm 5 MHz 40 degrees Cylinder of 15 mm radius 5. Ultrasonic test on hollow axle with small bore diameter 5.1 Testing equipment and test method Figure 8 shows the ultrasonic testing equipment for hollow axles with a small bore diameter in which a piezocomposite focal probe with a 40-degree angle of refraction is installed. Figure 9 shows the system structure. The probe head is inserted into the bore and moved in the axial and circumferential directions, and the ultrasound transmitted from the bore surface propagates forward (the direction to the axle center) and backward 80 QR of RTRI, Vol. 46, No. 2, June. 2005
Flaw detector PC Motor controller Oil pump Model axle head Motor Elevator Fig. 8 Ultrasonic testing equipment for hollow axles with small bore diameter I/O signals from motor, encoder and sensor θ-axis motor Flaw detection gate output head X-axis motor Waveform Control signal PC RS-232C To axle Couplant (oil) Oil pump Flaw detector Display A/D converter board Motor Motor controller board controller Digital I/O board Head position output Fig. 9 Ultrasonic testing equipment system Counter length: 1,900 gear side Journal bearing Model gear K J I HG FEDC B A K φ 192 J φ 150 I HG FEDC B A Gear side 100 Model wheel Bearing inner ring 100 Counter gear-side test zone Gear-side test zone 1,000 1,000 (mm) Fig. 10 Model hollow axle with small bore diameter (the direction to the axle end). With this equipment, an ultrasonic test was carried out on a model hollow axle with a small bore diameter that was manufactured on a trial basis. The model axle has artificial flaws machined in positions where cracks may arise and is used to confirm the inspection accuracy of the testing equipment. Figure 10 shows an outline of a model hollow axle and Table 2 shows the positions and shapes of 25 artificial flaws. The flaws on fitted parts such as the position I-I were machined at a position 3 mm inside the fit edge. After artificial flaws had been machined, a model gear and inner rings of gear bearing were fitted on the gear side of the axle and a model wheel and journal bearing were fitted on the counter gear side. Table 3 shows the conditions under which automatic testing was carried out on the model hollow axle. The movable range of the probe head in the axial direction was about 1,350 mm, and each half part of the axle length was tested from both ends of the axle. The position of the probe head and the testing sensitivity were adjusted using the 1S flaw (a square flaw of ) in the position D-D on the non-fitted central part. The flaw detec- φ 195 5 φ 150 φ 110 φ 30 0 Symbol Part Table 2 Artificial flaws of model hollow axle with small bore diameter Distance from gear-side axle end [mm] Distance from counter gear-side axle end [mm] EDM-machined artificial flaw(s) A-A Inner end of gear-side wheelseat 436 1464 0.3S, 0.6S, 1S; 1E B-B Inner end of gear seat 625 1275 1S C-C Non-fitted central part 950 950 0.5A D-D Non-fitted central part 1000 900 1S E-E Non-fitted central part 1050 850 0.6S F-F Non-fitted central part 1100 800 0.5N G-G Non-fitted central part 1150 750 0.3S H-H Non-fitted central part 1200 700 0.15S I-I Inner end of counter gearside wheelseat 0.3E, 0.6E, 1E, 3E 0.5N, 1N; 0.3S, 0.6S, 1S; 1464 436 J-J Outer end of counter gearside wheelseat 1607 293 1S K-K Fit edge of counter gearside journal bearing 1736.5 163.5 0.5N; 0.3S, 0.6S, 1S Depth: a Depth: a Depth: a Notched flaw: N Square flaw: S 1) The types of artificial flaws are described as A (all-round flaw), N (notched flaw), S (square flaw) and E (semi-elliptic flaw). 2) The number in front of flaw types (A, N, S, E) indicates the flaw depth [unit: mm]. 10 Length: 2c a/2c = 0.35 Semi-elliptic flaw: E QR of RTRI, Vol. 46, No. 2, June. 2005 81
Axial scan pitch: 2 mm Data recording point W 1 W 2 W 3 W 4 Angle of refraction: 40 deg. Data recording pitch in circumferential direction: 2 degrees Fig. 11 1 Spiral scan of probe head Table 3 Automatic test conditions Item Setting value Test zone 100-1000 mm from both axle ends Working sensitivity 18 db above the sensitivity of 80 % echo height of 1S flaw on the position D-D in the non-fitted central part Scan of probe head Spiral scan Rotation speed of probe 100 rpm (maximum: 300 rpm) head Axial scan pitch 2 mm per one rotation Data recording pitch in 2 degrees circumferential direction Trigger gate range The range including the path length at any position on the axle surface from the minimum diameter (journal) to the maximum diameter (wheelseat) Flaw detection gate 2 mm width from the first echo that exceeded range the threshold level of the trigger gate Recording information 1) The position of probe in the axial and circumferential directions 2) Maximum echo height and path length within the range of flaw detection gate tor amplifier gain when the echo height of this flaw was 80 percent was defined as the specified sensitivity, and the working sensitivity was adjusted to the specified sensitivity +18 db to compensate for the decrease of flaw sensitivity caused by the rotation of the probe head. The scan of the probe head was in the shape of a spiral, as shown in Fig. 11. The axial scan pitch per one rotation was 2 mm, the data recording pitch in the circumferential direction was 2 degrees, and the rotation speed was 100 rpm. 5.2 Flaw detection gate settings Center of ultrasound beam W i : Calculated path length to axle surface of each part : Flaw detection gate (set around axle surface) Echo height [%] Fig. 12 Normal flaw detection gate settings 100 80 60 40 20 0 Flaw echo Path length of flaw echo detected. 90 120 Path length [mm] Spurious echo Height of spurious echo incorrectly detected. Flaw detection gate at wheelseat Fig. 13 Waveform at position I-I In ultrasonic axle testing, flaw detection gates are normally set at every test position such as a wheelseat or a non-fitted central part, as shown in Fig. 12. Flaw detection gates are set around the calculated path length to the axle surface where particular attention needs to be paid, and only the waveform indicated within the gate range provides useful information. Figure 13 shows a waveform tested at the inner end of the counter gear-side wheelseat (position I-I) when a normal-type flaw detection gate is used. When this position is tested, spurious echoes from the corner of the wheelseat and its fillet are generated in addition to the flaw echoes, as shown in Fig. 14. Even if the flaw echo and the spurious echo are separated, the echoes will appear simultaneously within the range of the flaw detection gate when the difference of their path lengths is short. In addition, the height of the spurious echo is incorrectly detected as the echo height in the flaw detection gate if the spurious echo is more pronounced than the flaw echo. Therefore, as shown in Fig. 15, a trigger gate was set in a range including the path length at any position on the axle surface from the minimum diameter (journal) to the maximum diameter (wheelseat). A flaw detection gate of 2 mm width was set at the first echo that was above the trigger gate threshold level. If we use this method, only the flaw echo whose path length is shorter than that of the spurious echo can be detected separately when the flaw echo and the spurious echo appear simultaneously at the positions where spurious echoes are generated. The path lengths of echoes detected in scanning were compared among the several data recording points adjoining in the circumferential direction. If there were recording points whose path lengths varied over the range Calculated path length 40 deg. Reflection at flaw Ultrasound Center of ultrasound beam Bore Reflection at corner or fillet Fig. 14 Ultrasound reflection at inner end of counter gear- side wheelseat 82 QR of RTRI, Vol. 46, No. 2, June. 2005
Angle of refraction: 40 deg. Center of ultrasound beam : Trigger gate (constant in testing) Spurious echo Flaw echo 5.3 Test results Figure 16 shows the results of testing the non-fitted central part (positions from D-D to H-H) and the inner end of the counter gear-side wheelseat (position I-I, fitted part) when the ultrasound was transmitted forward with the probe head inserted from the axle end of the counter gear side. This Figure shows the maximum echo height in the flaw detection gate at every probe position in scanning. All artificial flaws at the non-fitted central part, and all artificial flaws except those of semi-elliptic shape whose depth was equal to or smaller than 1 mm at the inner end of the wheelseat, can be detected in a state that spurious echoes are eliminated and the signal/noise ratio is sufficiently great. Flaw detection gate (following first wave) Echo height and path length of flaw echo correctly detected. Trigger gate (a) Case of existing flaw echo and spurious echo Detected path lengths compared and whether flaw echo exists or not decided. Flaw detection gate Spurious echo Echo height and path length of spurious echo detected. Trigger gate (b) Case of existing spurious echo only Fig. 15 Echo detection technique using trigger gate of 1 mm by less than 15 degrees, it was decided that their data contained those of flaw echoes that appeared ahead of the spurious echoes. 6. Comparison of inspection accuracy Table 4 shows the minimum flaws that can be detected at every part of an axle when axles are tested by the technique mentioned in this report, by the longitudinal wave angle-beam technique for conventional solid axle railway vehicles using automatic testing equipment, and by the angle-beam hollow axle technique that also uses automatic testing equipment for Shinkansen vehicles 3). With the ultrasonic test from a bore of 30 mm diameter machined in an axle for conventional railway vehicles, the inspection accuracy becomes higher than was the case in tests using the conventional technique with a longitudinal angle-beam from the end face of a solid axle. We could therefore obtain an inspection accuracy equal to that of hollow axles for Shinkansen vehicles. 7. Conclusions On the basis of experience gained with hollow axles that have been used on Shinkansen vehicles, a hollow axle with a small bore diameter of 30 mm applicable to con- Notched flaw of 0.5 mm depth Circumferential position [deg.] 0.3 mm depth Square flaws 0.15 mm depth 0.6 mm depth Distance from axle end face [mm] Displayed range Echo height [%] Echo height [%] Displayed range Notched flaws 0.5 mm depth Square flaws 0.6 mm depth 0.3 mm depth Semi-elliptic flaw of 3 mm depth Circumferential position [deg.] Notched flaw of Distance from axle end face [mm] (a) Non-fitted central part (b) Inner end of counter gear-side wheelseat (fitted part) Fig. 16 Ultrasonic test results of model hollow axle with small bore diameter QR of RTRI, Vol. 46, No. 2, June. 2005 83
Table 4 Comparison of inspection accuracy Angle-beam technique used on φ30 mm hollow axles for conventional railway vehicles (this report) Longitudinal wave anglebeam technique used on solid axles for conventional railway vehicles Fit edge of journal bearing 0.3S 1N 0.6S Outer end of wheelseat 1S 10N 1S Inner end of wheelseat 0.3S,3E 1N 1S Inner end of gear seat 1S 3N 1S Non-fitted central part 0.15S 3N 0.15S Angle-beam technique used on φ60 mm hollow axles for Shinkansen vehicles ventional railway vehicles was tested using ultrasonic testing equipment in which a piezocomposite focal probe with a 40-degree angle of refraction was installed. As a result, a square-shaped artificial flaw 0.15 mm deep at the non-fitted central part and another 0.3 mm deep at the inner end of the wheelseat (fitted part) could be detected. By actively introducing hollow axles also into conventional railway vehicles, it would be possible to improve the efficiency of inspections and increase the flaw detection accuracy of automatic axle inspections. Acknowledgements We would like to thank Prof. H. Toda of Wakayama University for guiding us in the design of focal probes. References 1) Yohso, J., Sakamoto, H., Makino, K. and Ishiduka, H., "Inspection of Rolling Stock s by Using the Grazing SH-wave Ultrasonic Method (in Japanese)," RTRI Report, Vol. 16, No. 5, pp. 35-40, 2002. 2) Hachiya, M., Toda, H., Murata, Y. and Yohso, J., "Characteristics of Ultrasonic Transducer for Bored Type (in Japanese)," Proceedings of the 9th Symposium on Ultrasonic Testing, pp. 75-76, 2002. 3) Yohso, J., "Development of Automatic Ultrasonic Testing Equipment for General and Bogie Inspection of Shinkansen Hollow," Proceedings of the 11th International Wheelset Congress, Vol. 2, pp. 47-50, 1995. 84 QR of RTRI, Vol. 46, No. 2, June. 2005