|NDT.net - August 2000, Vol. 5 No. 08|
Corresponding Author Contact:
|2nd International Conference on NDE in Relation to Structural Integrity for Nuclear and Pressurized Components, New Orleans May 2000.|
|TABLE OF CONTENTS|
To tackle this task a concerted action was initiated in Europe for Aging Materials Evaluation and Studies by Non Destructive Techniques (AMES-NDT) and was carried out during the past years (1). The aim of the concerted action was to identify the needs for further development in view of NDT-methods for measuring aging phenomena in materials caused by load (temperature and pressure) as well as neutron irradiation embrittlement.
For the characterization of materials due to aging we have made an approach with Magnetoinductive Harmonic Analysis (MHA). Selected results of these investigations are being presented. Due to the fact that the base materials of RPV are of ferritic nature magnetoinductive methods are applicable. Experiments have been performed primarily with JRQ-steel (ASTM A-533-BCl.1) because of its international use for analysis and comparison of results. First the working principle of MHA is described and then results are presented.
Fig 1: Magneto-mechanical analogy
The working principle of MHA is based on the magneto-mechanical analogy schematically shown in Figure 1. Its fundamental reason is the correlation between the mechanical and magnetic moments of the elementary atomic magnets. In the macroscopic range of materials the results of the analogy are reflected in the correlation of macro-mechanical properties of ferromagnetic materials, like tensile strength, yield strength, hardness, shelf energy, anisotropy etc. on the one hand and magnetic properties, characterized by the shape of the magnetic hysteresis loop and the quantities describing the shape, like differential permeability, coercivity, remanence and saturation magnetization on the other hand. This correlation is often applied by magnetic methods, like Barkhausen Noise, where the coercivity is determined and related, for example, to hardness or other mechanical quantities. In our opinion, however, to measure the complete hysteresis curve or parts of it and then evaluate it is not the most efficient and elegant way to determine mechanical properties.
Fig 2: Working principle of MHA
Magnetoinductive Harmonic Analysis, we think, is much better suited for this purpose. How it works is explained in Figure 2. A coil system driven by a sinusoidal current generates a sinusoidal magnetic field which is transmitted into the material under investigation. The hysteresis loop of the material is a finger print of it's characteristic properties. The alternating magnetic field of the coil excites the hysteresis loop and by this the signal is modulated by the shape of the hysteresis loop, which acts - electrically speaking - as a transfer function. Thus the signal now received by a sensor contains the shape of the hysteresis loop in form of higher harmonics. These harmonics are determined by a Fast Fourier Transformation (FFT) and correlated to the mechanical properties of the material obtained from destructive tests. This "calibration procedure" has to be performed once for one material in order to correctly apply the method and has to be repeated for different types of materials. As can be seen very clearly from figure 2 work hardening affects the shape of the hysteresis loop remarkably and is immediately reflected in the signal and the higher harmonics amplitude and phase spectrum of the receiver signal. Due to the symmetry of the hysteresis loop only odd higher harmonics (220.127.116.11.9. ...) are measured.
Fig 3: Measuring System for MHAW
A typical measuring system is schematically shown in Figure 3. Basically it consists of a computer with a special plug-in board, an amplifier module and a coil system containing the transmitting and receiving coil. The plug-in board, specifically designed for this application, generates a sinuso-idal signal, which is sent to the trans-mitting coil. the modulated magnetic flux induces a measuring voltage within the receiving coil, which then is amplified, digitized and analyzed. The harmonics of the received magnetic flux signals result from a FFT Spectral Analysis, which gives the amplitudes and phases of the signals at the harmonic frequencies (2). Evaluated are the odd harmonic frequencies from one to eleven. For setting up the "calibration curve", i.e. the correlation between mechanical-technological values (tensile strength, yield strength, hardness, etc.) and measured magnetic values (amplitudes and phases of higher harmonics), the procedure is as follows: By variation of the measuring frequency and the magnetic field strength of the transmitting signal it is possible to generate a great number of parameters (measured values). If there are several measured parameters which correlate linearly with a material property, the latter can be described mathematically as a function of several parameters by means of a multidimensional linear regression analysis. This procedure increases the accuracy of the method considerably. Corresponding to the number of parameters used is the dimension of the regression analysis.
In view of the measuring time it can be stated, that the overall time including signal data acquisition, spectral analysis and calculations giving the mechanical values of the material is around one to two seconds. Therefore the method is well suited for on-line and in-line application. Furthermore the sensor system, normally consisting of coils, can easily be adapted to specific tasks of power plant components and a robust construction is made for rugged use in the field. The electronic and the computer system can also be made portable. Many applications have already been performed in the field (3) with success.
A quantified statement of the performance of the measurement technique with respect to the associated degradation mechanism should be elaborated. This should include sensitivity to the mechanism, tolerances and insensitivity to other mechanisms. There is another important aspect in particular regarding RPVs. The present technique to survey RPV is to use Charpy test samples irradiated in the vessel to measure upper shelf energy by destructive tests. This reduces the number of survey samples available. Therefore an interesting aspect would be to monitor at least some samples by NDT in discrete periods (RPV revision),measure eventual changes in material properties by NDT, relate it to upper shelf energy and put the sample back into the RPV until the next test. Thus surveying samples would be available for a longer time of plant operation. Also because of this reason MHA-measurements have been made with Charpy Test samples using encircling coils. For the inspection of the RPV itself surface coils are being used.
(1) Feasibility Study
The objective of the feasibility study was to investigate whether MHA would be able to measure characteristic mechanical-technological values in the presence of an austenitic steel cladding. Based on extensive know-how of MHA-application in steel works (3) feasibility tests were made with high strength steels (Steel types TStE 690 and TStE 960) by measuring tensile strength using MHA and correlating it to tensile strength from destructive tests.
Figure 4 shows the result of tensile strength values measured without austenitic cladding. To characterize the degree of correlation between values measured destructively and those measured by MHA, the correlation coefficient is used. Correlation coefficient equal to zero percent means there is no correlation at all and 100 percent means perfect correlation. To specify the tolerances, the absolute standard deviation in N/mm2 is used as well as the relative standard deviation in (%). As can be seen from the results without cladding the correlation coefficient of 99% is very high and therefore 95% of all data fall into the ±3% band of standard deviation, named class 1 (CL.1). Figure 5 shows the results of measurements made under the same conditions but with an austenitic cladding of 6mm. The austenitic steel plate was put on top of the ferritic base material without welding. As can be seen from figure 5 the correlation coefficient is still 97.3% resulting in an absolute deviation of 16.8 N/mm2, twice as high as in the case without cladding. None of the data has a maximum standard deviation (class 3) of larger than 6% which is a very good result.
||Fig 4: MHA without cladding
||Fig 5: MHA with cladding
The conclusions drawn from this feasibility study are the following:
|Table 1 : Chemical composition (weight %) of JRQ-steel|
The comparison to all measured data was made to fresh material (as-delivered) of JRQ-plate 7 which served as reference material. The samples were distributed to different laboratories participating in the AMES-NDT concerted action. Thus, different techniques could be compared on this occasion. The JRQ plate material was manufactured by the KAWASAKI Steel company, Japan. The heat treatment of JRQ was:
|Normalization||900° C - Air cooled (AC)|
|Quenching||880° C - Water quenched (WQ)|
|Tempering||665° C - 12 hours - Air cooled|
|PHWT||620° C - 40 hours - Furnace cooled|
Fig 6: Preparation of Charpy Test samples
|Fig 7: JRQ plate 10|
|Fig 8: JRQ plate 11|
According to the ASTME E 23-96 the standard Cv-notch sample size is 10x10x55 mm. As shown in Figure 7 and Figure 8 the heat treatment of the plates was performed. In order to produce brittle material all the bars coming from plates 10 and 11 were thermally treated at 700° C in a vacuum environment during eighteen hours and then quenched in water. This treatment was performed by NRG, Petten, NL. The mechanical behavior of the non-aged and thermally treated material has been determined by Charpy impact testing of the steel. The impact energy is determined as a function of the Charpy test temperature in order to draw the so-called "Charpy impact curve". In order to characterize the embrittlement stage of the treated material Charpy tests were performed at JRC - AM, Petten, NL, with a 300 J hammer following the standard ISO/CD 14556. Impact tests were carried out on fresh JRQ material (plate 7) and on the treated material from JRQ plates 10 and 11. Results are shown in Figure 9. For the fresh material the upper shelf energy (USE) is 191 J and the ductile-to-brittle transition temperature (DBTT) at 41 J is 19.6° C. For the thermally treated material USE is 120 J and DBTT at 41 J is 103.0° C.
In addition hardness measurements by Hardness Vickers 10 tests were performed on the specimens. These hardness measurements were made on the surface of the bars and on the broken samples from the Charpy tests. As could be seen the hardness of Cv-specimens (measured next to the notch) is less than the hardness of the bars. This occurs because the thermal treatment, in particular the quenching, causes a hardening of the surface. Therefore additional hardness measurements were performed at different depths. The nearer to the surface the harder the material . This effect must be taken into account when using NDT-methods at the surface or with different penetration depths.
Fig 9: Charpy impact curve for fresh and thermally treated material
The microstructure of the JRQ material has been examined by optical microscopy and scanning electron microscopy (SEM) to obtain information on the microscopical changes that caused the shift in DBTT and reduction in USE. Specimen no. 3 of bar number 10.1 (see figure 7) has been used for the examinations. The microstructural sample was taken from the end face of a broken Charpy Specimen. Cross sections were made parallel to the fracture surface of the Charpy specimen. In the microstructure the a-bainitic matrix with isolated carbides and segregation bands are clearly visible. On the former austenite grain boundaries a phase has formed. This is in contrast to the fresh, as-delivered, condition, where no phases have formed on the prior austenite grains. The formation of the phase is continuous throughout the cross section. In the segregation bands the phases are almost continuous on the former austenite grain boundaries. In the matrix a semi-continuous phase formation has occurred. This result is due to the differences in chemical composition (i.e. carbon content) between the matrix and the segregation bands. Although 700°C is below the Ac line of this material , prolonged annealing at 700°C might result in the formation of austenite in certain parts of the microstructure due to local differences in micro-chemical composition. This transformation seems to start at the former austenite grain boundaries. During quenching the formed austenite will transform into martensite. Therefore it is expected that the phases found on the former austenite grain boundaries are martensitic phases. This was confirmed by SEM investigation of the etched samples. The phases show a very fine martensitic lath structure. The martensitic phase has an average size of 4 ± 2 micrometer. The microstructural analysis was performed by NRG, Petten, NL.
(2) Experiments with Charpy test samples using different coil types.
The objective of these experiments was to investigate whether the MHA-technique is able to distinguish between fresh and thermally aged samples and with what sensitivity. The coil setups used are shown in Figure 10.
Fig 10: Type of coils and scans used in the experiments
Bars of 10 by 10 mm and 210 mm length were used. A scan along the length of the bar was made by encircling coil as well as with surface coil. The characteristic magnetic parameters, sensitive to microstructure of steel - as explained above - are the amplitudes and phases of the higher harmonics. In these experiments the first, third and fifth harmonic were evaluated in respect to amplitude and phase.
The results for the first harmonic (fundamental mode) are shown in Figure 11.
|Fig 11: Amplitude and Phase of Harmonic, Encircling Coil, f = 234 Hz|
The scans were performed for three different samples:
(a) fresh sample (no 7.9) from JRQ-plate 7 (as- delivered material).
(b) aged material sample (no 10.5) from plate 10
(c) aged material sample (no 11.2) from plate 11.
The measured values of the fresh material sample (7.9) is the reference curve. On the left hand side the amplitude values of the first harmonic and on the right hand side the phase values are shown along the length of the sample. f is the measuring frequency. The following results were found for the first harmonic:
For the third harmonic the corresponding curves are shown in Figure 12. The following results were found:
The same type of experiments were performed using surface coils for scanning the bars (see figure 10). The results are essentially the same, as it should be, although the surface coil is less sensitive due to the geometrical setup. The equivalence of the results is also demonstrated by Figure 14.
Fig 10: Hardness measured by MHA and destructive tests
Here the hardness values measured by MHA are plotted versus the hardness values from Vickers Hardness tests performed at the samples as described before. There is a linear correlation curve obtained by a two dimensional regression analysis using only the third and fifth harmonic amplitude as parameters.
From the results obtained so far it can be concluded that for the conditions described, the fresh and aged samples could easily be distinguished. This shows the high potential of the MHA-technique regarding analysis of thermal aging and embrittlement of Charpy test samples in particular if encircling coils are used.
(3) Tests under simulated RPV conditions.
In order to investigate the use of MHA-technique for direct application at the pressure vessel component two simulation experiments were performed.
|Fig 15: Scan of JRQ-plate 13, f = 234 Hz|
JRQ Plate 13
Measuring Position: Upper side
Fig 16: Cladding attenuation at measuring frequency of 234 Hz
The results shown in Figure 16 give quantitative attenuation coefficients for the higher harmonics at a frequency of 234 Hz for the materials considered. Having a 6 mm austenitic steel cladding, which is in the range of RPV cladding, the attenuation of the first and third harmonic is approximately a factor of five in amplitude whereas for the fifth the factor is about ten. By a proper choice of measuring conditions the MHA-signals are still large enough for the analysis of mechanical-technological quantities of the base material beneath the cladding. This can be seen from the following results. From the scan of JRQ-plate 13 using the first harmonic amplitude it could be concluded that the homogeneity of the material was good. Now, if the third harmonic amplitude is plotted the picture looks a bit different as shown in Figure 17.
Fig 17: 3. Harmonic amplitude scan of plate 13, f=234Hz|
Fig 18: 3. Harmonic amplitude scan of plate 13, f=234 Hz
Cladding: 6 mm
Fig 15: Scan of cladding welded to RPV base material
|© NDT.net - firstname.lastname@example.org|||Top||