NDT.net - June 1998,
Table of Contents ECNDT '98
Copenhagen 26 - 29 May 1998
Ultrasonic Assessment Of Residual Stresses In The Rim Of Railway Solid Wheels
M. Gori 1, M. Certo 1, G. Patelli 2, L. Aruta 3
1 CISE SpA - Segrate (MI) - I
2 Lucchini Siderurgica SpA, Brescia - I
3 FS SpA, ASA Materiale Rotabile Trazione, Laboratorio CND, Firenze - I
Tensile stresses of unsafe magnitude can be occasionally found in the rim of railway wheels, as a result of drag-braking under severe conditions. Such stresses are an hazard for human, environmental and economical reasons, since they may drive cracks up to wheel failure. Railway organizations and wheel manufacturers thus pointed out the demand for an NDE method enabling the non-invasive assessment of rim residual stresses.
A joint work was carried to validate for such task an ultrasonic system, ECOMAT, which can evaluate residual stresses through the acoustoelastic effect, measuring the birefringence of two shear waves produced by a non-contact dual-coil EMAT probe.
The authors compared birefringence and strain gauge data on new wheels, during wheel sectioning, and observed that stresses were differing for less than ±10 MPa. Additional trial tests were performed, which could allow to optimize the evaluation of travelling time, in order to grant reliable stress measurements also on used wheels. ECOMAT is currently operating in a wheelset inspection and maintenance plant, and a program is in under way to integrate the system in a wheel production line.
Railway freight cars use block brakes acting on the wheel tread. The brake shoe friction considerably heats the tread and, under some operating conditions, tensile stresses can be induced in the wheel rim. To hinder a swift increase of tensile stresses, prevent crack propagation and wheel rupture, compressive stresses are added to modern solid wheels by rim chilling. Notwithstanding the effectiveness of such protection, the non-invasive evaluation of wheel stresses has attracted wheel manufactures, who could document the compressive stresses added in production, and railway organizations, which would test that severe braking has not reversed them to a tensile unacceptable magnitude. The acoustoelastic birefringence of shear waves was then selected by "European Rail Research Institute" as the NDE method best suited for stress measurements in the rim of railway solid wheels , and recently introduced, by "Union International des Chemins de fer", to assess the condition of operating wheels .
To provide a testing apparatus suitable for the above objectives, the italian rail and wheel manufacturer Lucchini Siderurgica, and Italian Railway FS, have contributed to the improvement of the ECOMAT ultrasonic system, that CISE has been developing for the nondestructive evaluation of residual stresses in industrial components.
STRESS EVALUATION BY ACOUSTOELASTIC BIREFRINGENCE
The acoustoelastic birefringence was discovered in 1959 by Benson and Raelson , who detected a slight stress dependence of shear wave velocity from polarization direction, and then extensively investigated to exploit such phenomenon for its potential as a nondestructive technique .
The acoutoelastic birefringence (hereafter B) is the anisotropy between the velocities of two shear horizontal (SH) waves, generated and detected in the same measuring site, and polarized along orthogonal directions. B can be evaluated by the relative difference between the travelling time of the above said waves:
B _ TR - TC / TR (1)
(C and R are the labels of polarization directions, circumferential and radial respectively). In a stress-free condition, however, B is seldom null since materials usually have an inherent texture anisotropy B0. Hence, only the stress-related birefringence (B - B0) linearly depends from the difference between principal stresses (C - R) via the acoustoelastic constant A:
(B - B0) * A = C - R (2)
In the presence of a biaxial stress field, slightly varying along the thickness, the method can reliably evaluate the average difference of principal stresses. In railway solid wheels, both the safety compressive stresses, set in by rim chilling, and the hazardous tensile stresses that may originate in used wheels, are known to have a predominant hoop orientation: as principal stress R is negligible, equation (2) can allow an evaluation, nearly absolute, of the circumferential stress C.
ACCURACY REQUIREMENTS FOR EXPLOITATION TO RAILWAY WHEELS
To examine the performance we should provide to enable reliable measurements, standard deviation ( can be derived from equation (2). As a first approximation, this will be a function of the travelling time magnitude T and accuracy T, and of the acoustoelastic constant which establishes the stress sensitivity for a selected material:
The factor 2 takes into account that we have to measure two travelling times, which will have nearly the same magnitude and accuracy. In equation (3) the effects of the potential scatter of the constant A (about 10%) and the contribution of texture, have been neglected.
The reliable evaluation of residual stresses needs, as an imperative requirement, accuracy repeatability and reproducibility in the measurement of travelling times. In a typical steel wheel, with rim thickness of 140 mm, a stress accuracy of ±2.5 MPa could be ensured if T were about 1 nanosec. To measure T with such accuracy, suitable algorithms can be adopted, such as correlation of multiple echoes or selection of arrival time of first backwall echo; averaging can help the control of erratic and unacceptable fluctuations of the mean value of T, and possibly be used to reject false measurements.
Even if exceeding practical needs, an accuracy of few MPa is required to warrant the reliability of the system and endorse its application. On used wheels, for example, texture anisotropy B0 could not be assessed, and we will have to tackle its unpredictable contribution. If B0 had a magnitude of 0.4 10-3, a stress-equivalent error larger than ±50 MPa would eventually label the accuracy of our stress prediction: the only possibility to allow the final user in tolerating such uncertainty is, at least, to keep the inherent accuracy of the system within the magnitude, of few MPa, that we have stated above.
THE ECOMAT SYSTEM
ECOMAT is the CISE ultrasonic system based on EMATs (ElectroMagnetic Acoustic Transducers), which can evaluate residual stresses using the following hardware and software:
The recent ATI unit has two separate channels with on-board digital multiplexer, to allow a rapid stress evaluation at each selected probe position, and optionally drive a mechanized rig to automate scanning.
It is worth noticing that such stress measurements can be performed either by conventional transducers (PZT), which give strong signals, or using the weaker non-contact EMATs . PZTs, however, need much more care since a variable thickness (or temperature) of the coupling layer is known to produce erratic contributions in time measurements. Correlation of multiple echoes or a strict control of the said coupling characteristics can be used to minimize or suppress such effects.
- AMB100, ISA electronic card plugged into an AT-compatible personal computer, for acquisition conditioning and digitalization of ultrasonic signals;
- ATI, external pulser/preamplifier unit, frequency range 3 - 5 MHz;
- EMAT-DC, 4 MHz dual coil probe, to provide two bulk shear horizontal waves of orthogonal polarization;
- DOS or Windows 95 software package, for AMB100/ATI managing and data analysis.
SYSTEM VALIDATION ON NEW WHEELS
At the Lucchini Siderurgica plant in Lovere a blind test was performed, to validate ECOMAT for wheel stress evaluation. CISE measured B on three new wheels, which Lucchini cut in sections of decresing amplitude while monitoring the stress relieved by means of strain gauges. The R7 steel wheels (NW1-=820 mm, NW2-=920 mm, NW3-=1250 mm) were heat treated by rim chilling to produce hoop residual stresses.
Fig.1 New wheel NW1 ( = 820 mm). Average B values as a function of distance from tread surface. Error bars show oscillations along hoop direction.
Fig.2 New wheel NW2 ( = 920 mm). Average B values as a function of distance from tread surface. Error bars show oscillations along hoop direction.
Fig.3 New wheel NW3 ( = 1250 mm). Average B values as a function of distance from tread surface. Error bars show oscillations along hoop direction.
Fig.4 Residual stresses released during cutting of new wheels NW1, NW2 and NW3. Strain gauge measurements performed after each cutting step.
Fig.5: Correlation of residual stresses for 3 new wheels. Most of the experimental data are within the confidence region ± 10 MPa.
Fig.6 ECOMAT stress measurements performed on a set of 34 wheels (3 new and 31 used). FS wheelset inspection and maintenance plant (Firenze).
For each wheel, three distinct radial sections were dressed, each one with a set of three strain gauges, glued in the middle of rim faces and tread surface. Sectors of decreasing amplitude (180°, 60°, 20°) were cut to liberate the above said distinct sections. Eventually, 10( sectors had been machined, which could be used, as reasonably stress-free samples, for characterizing the texture contribution.
Radial scans were performed close to each strain gauge set, few additional scans were done elsewhere to increase statistics. Travelling times were evaluated by correlation of multiple echoes. Fig.1 represents B values (hoop average) for NW1, before and after stresses were released. The graph shows the sudden decrease of B after bisecting the wheel. Similar results were obtained for NW2 (fig.2) and NW3 (fig.3). The standard deviation of the hoop average shows some oscillation of variable amplitude over the whole rim.
The radial dependence of B, nearly unaffected due to sectioning, was associated to the typical wheel texture. The average >B0 of 10° sectors was 0.2‰ for wheel NW1, and -0.2‰ for wheels NW2 and NW3, with oscillations of same magnitude. Such values were used to account for texture when deriving the original residual stresses by equation (3).
To measure the acoustoelastic constant, during tensile tests, 6 T-shaped samples were machined, 2 of different orientations for each wheel. Different values were measured, all within the range [-110000 - -137000] MPa. Since no relationship was found with sample orientation or wheel type, which could explain such scatter, to compute the original residual stresses the mean value 123500 MPa was adopted.
Then Lucchini disclosed its strain gauge measurements (fig.4). The large amount of stress (70-80%) suddenly released when bisecting the wheel, confirmed the long-range characteristics of the compressive stresses added by rim chilling.
The amount and progress of the stress relieve with cutting was similar for both methods, and the data agreement was very impressive. The graph shown in fig.5 considers the residual stresses existing in the three wheels, before each cutting step and for the sections where the strain gauge sets were located, thus giving a total amount of 27 experimental points. Most points (25 out of 27) were found within or close to the confidence region of ±10 MPa, providing the successful validation of the system.
STRESS MEASUREMENTS IN A WHEELSET NDT LAB
An ECOMAT system was delivered to the FS wheelset NDT lab in Firenze, where additional trial tests were performed. It was assessed that a significant number of false measurements occurred on several wheels, when correlating multiple echoes having a very different S/N ratio. To overcome the problem, a suitable algorithm was set up, to measure travelling times by selecting the arrival of first backwall echo. This latter method could retain the accuracy shown, when using the former method, in the frame of the validation work.
FS evaluated the upgraded system by scanning the rim of different wheels along three radial lines, the formers spaced by 40 mm, and the latter 400 mm from the formers. Few scans were repeated to assess reproducibility. Since texture was unknown, a prudent set of parameters was used to evaluate stresses: acoustoelastic constant A of -140000 MPa for maximum stress sensitivity, tensile stress threshold of 150 MPa (instead of 200 MPa) to avoid stress underestimation in the presence of a compressive-like texture contribution.
Fig.6 shows residual stresses for a sequence of 3 new and 31 used wheels (left to right), with minimum, median and maximum stress values of each inspected wheel. Again, for the first 3 new wheels compressive stresses of about -100 MPa were detected. The majority of the other used wheels showed compressive or tensile stresses of magnitude lower than 100 MPa. The last two wheels showed tensile stresses larger than 200 MPa, with peaks close to 300 MPa. Fig.7 shows the stress profiles along the said radial lines (B1, B2, B3) for one of the inspected wheels. The negligible scatter observed after the repetition of line B1 (B1b), is a demonstration of the optimal reproducibility of the measurements.
Fig.7 Residual stresses in the rim of wheel 11B. One scan was repeated to assess reproducibility. FS wheelset inspection and maintenance plant (Firenze).
The start up of ECOMAT in the wheelset inspection and maintenance shop took place in March 1997. To the present date, more than 600 wheels had been examined: all of them could be evaluated in terms of residual stresses, and less than 2% had to be dismissed for the presence of tensile stresses exceeding the allowed threshold.
The ultrasonic system ECOMAT using a non-contact dual-coil EMAT probe can evaluate residual stresses in the rim of railway solid wheels, with accuracy and performance suitable for an industrial exploitation.
Following the results of the validation and optimization work described in the present paper, a system was installed in a wheelset inspection and maintenance plant, where FS now performs routine stress measurements on used wheels.
A program is currently under way to integrate a new ECOMAT system into the automated wheel production line, at Lucchini Siderurgica plant in Lovere; two additional systems will be also delivered to FS for setting additional stress measuring stations.
- Standardisation des essieux. Mיthodes de surveillance des roues monobloc en service. Mיthode aux ultrasons puor la dיtermination non destructive des contraintes rיsiduelles dan le jantes de roues monobloc. Report n( ERRI B 169/RP 6, 1995. Prepared by ERRI B 169/3. European Rail Research Instute, Utrecht, NL.
- Matיriel remorquי. Roues et essieux montיs. Conditions concernant l'utilisation de roues de diffיrent diamטtres. Code UIC 510-2. Union International des Chemins de fer.
- R.W.Benson, V.J.Raelson. Prod. Eng. (NY), 30, 56, 1959.
- Y.Pao, W.Sachse, H.Fukuoka. Acoustoelasticity and Ultrasonic measurements of Residual Stresses. Physical Acoustics Vol. XVII, Academic Press NY -1984.
- R.E.Schramm, J. Szel__ek, A.V.Clark Jr. Report n( 30 - Dynamometer-induced residual stress in railroad wheels: ultrasonic and saw cut measurements. NISTIR 5043, March 1995. National Institute of Standards and Technology, Boulder, Colorado.
Copyright © NDT.net,
/DB:Article /SO:ECNDT98 /AU:Gori_M /AU:Certo_M /AU:Patelli_G /AU:Aruta_L /CN:IT /CT:UT /CT:characterization /CT:stress /CT:wheel /CT:Railway /ED:1998-07