交流磁化プローブを用いた鋳造構造物の材質評価
公開日:
カテゴリ: 第1回
1. Introduction
Ductile cast iron is cast iron in which the graphite is present as tiny spheres: this feature permits greater strength and greater ductility thanFerrite gray cast iron of similar composition. There are so many applications of ductile cast irons, for instance, automotive-crankshafts, pistons, motor frames, levels, furnace doors, electrical fittings, switch boxesPearlite and tools for aqueducts; then the nondestructive evaluation of its mechanical properties is of paramount importance for the maintenance of various kinds of apparatus as well as structures. Two factors, the morphology of graphite and the type of matrices,
259 HB mainly determine mechanical properties of cast irons. With the goodGraphite sensitivity to graphite morphology, ultrasonic techniques have been used for the evaluation of mechanical properties in cast iron [1,2].In case of ductile cast iron, mechanical properties depend on both the nodularity of the graphite and the properties of matrices. Because it is difficult for ultrasonic techniques to assess the structure of matrices, direct observation and destructive testing such as hardness tests are usually carried out in the final analysis [3,4]. Therefore, it is essentially required to develop methods to assess the properties of matrices by means of non-destructive evaluation [5]. The matrix of ductile cast iron consists of structure in term of proportions of pearlite and ferrite. As Figure 1: Example of microstructure of 4 ductile cast iron specimens the quantity of pearlite increases, the strength and hardness of the with different hardness, 143 HB, 162 HB, 215 HB and 259 HB material increase. The proportion of ferrite and pearlite to the materials respectively. White and gray colors indicate the ferrite and pearlite, principally determines ductility and impact properties.respectively. Dark ?gspots““ indicate the graphite. Figure 1 shows the optical microscopy photos of four specimens with difference in hardness, 143 HB, 162 HB, 215 HB and 259 HB; white The AC magnetization probe used in this study includes simple data and gray colors indicate the ferrite and pearlite, respectively; moreover acquisition and good accuracy, in order to accommodate practical usage. the dark ““spots?h show the graphite: it can be seen that the pearlite ratio To emphasize the permeability variation, the AC magnetization probe increases as hardness increases. Since the ferrite and pearlite have has a resonant circuit tuning up for about three times higher than the difference in electric as well as magnetic materials, then fundamental excitation frequency. It is an effective technique to assess non-destructive ways utilizing electromagnetic field phenomena make the properties of matrices and it is expected a complete evaluation of it possible to evaluate the properties of matrices.mechanical properties of ductile cast iron. This nondestructive, as well There are some methods to evaluate the matrices in non-destructive as inexpensive approach can be realized through the combination of the way by using harmonic analysis of eddy current testing [6] and eddy present method with [6] for the matrices, and ultrasonic testing for current evaluation concentrated in the reversible region of initial graphite morphology. magnetization curve [7). This paper also considers the weakiy magnetized region with a low intensity of excitation. This means that 2. AC magnetization method eddy current signal reflects on only the two parameters, namely, The AC magnetization method consists of calculating magnetic flux conductivity and permeability around Rayleigh region. Because of density B[T) as a reaction of an alternating applied magnetic field reversible region on magnetization, it guarantees in addition the H[A/m). In the ferromagnetic materials the function B = uH is not reproducibility of experiment [8]. Considering the relation between the linear, it depends on the non-linearity of the permeability on varying of hardness and the structure of matrix, it is possible to estimate the the magnetic field. hardness of ductile cast iron efficiently. In addition, the hardness covers The initial magnetization curve is the relation between H and B for a several other properties of materials, as resistance to deformation, ferromagnetic material virgin, which is material that has not had an resistance to friction and abrasion. There are two main differences with influence of magnetic field yet. Figure 2 is a schematic diagram of the eddy current testing. One is to use relatively lower frequency, in a initial magnetization curve. Generally, it is possible to classify three range of 6-7 kHz, to catch permeability variation in each of hardness. zones of interest. The other is to employ the amplified 3rd harmonic component of sensor output voltage to easily distinguish the differences in hysteresis loops.
H(A/m Figure 2: Initial magnetization curve: the points 1,2 and 3 indicate the division of the curve in three regions.The region 0-1 is called reversible or Rayleigh region and it is considered in this investigation to guarantee the reproducibility of the measurements. In this region the material returns to its original state if a reversing magnetic field is applied. One of the principal problems is how to comprehend that the investigation is carrying out in the region. On the other hand, the regions 1-2 and 2-3 are non-reversible regions. in these regions, the residual effects of the magnetization, usually shown in the magnetic hysteresis loop (Figure 3), are exhibited, meaning that it is difficult to obtain good reproducibility without specific demagnetization preconditionings.B [T] ABoH[A/mFigure 3: Hysteresis loop: the parameters to be evaluated in a hysteresis loop are the area, the positive zero-cross value of the magnetic field, called ““coercivity and the zero-cross value of the flux density, called ““residual?h.Hysteresis curve1.5B(T)Ferrite based Pearlite based-1.5H (A/m)Figure 4: Hysteresis loops of Ferrite based and Pearlite based cast irons.The hysteresis loop is the relation between the magnetic field H and the flux density B for a ferromagnetic material under an alternating magnetic field. In this case it is possible to observe that the magnetization curve traces different ways after the initial magnetization curve is overtaken. Br, called residual, and Hc, called coercivity, represent the effects of the residual fields. The shape of the loop depends on the material and the area of the loop is proportional to the losses in term of energy.Since the hysteresis loop changes its characteristics on changing the materials, then the AC magnetization approach uses those concepts to evaluate the properties of the materials. Figure 4 shows the differences in term of hysteresis loop for the ferrite- and pearlite-based cast irons. The pearlite-based cast iron presents a wider loop, meaning that it is-148hard in terms of magnetic and mechanical properties. In this case the probe with the 3rd harmonic amplified will be more sensible and the normalized value will be higher, as we will show in the experimental results. Furthermore the maximum value of B makes it possible to identify the structure of the matrix.3. AC magnetization measurement systemThe evaluation of cast irons was carried out with experimental setup showed schematically in Figure 5. The probe used in the system consists of two coaxial coils with differential connection [9].The specimens were magnetized by supplying sinusoidal voltage to the probe at frequencies in a range of 6-7 kHz. The system presents the amplified 3rd harmonic of detected voltage by a circuit of which resonance frequency is about 18 kHz [10]. With the intention of show the differences of measurements in the non-resonance frequencies it was investigated also the 3kHz.The exciting and detecting voltages, the values of the harmonics ratio in dB and the area of loop in Lissajous graphics were recorded by using a PCMCIA National Instrument DAQ Card 6062E; the software to control the excitation and detection, and to save the data on file was realized in LabVIEW 7.0 under a Laptop with 384 MB RAM and 1.13 GHz Intel Pentium III Processor.Figure 5: Experimental Setup: it is composed of the exciter, ferromagnetic core and system to detect the Vpickup.All instrumentation weighs about 1 kg and takes up as a book: it guarantees the portability of the probe device. It consists of a DAQ Card for Analog Digital conversion, a battery and a probe which contains the coils, ferromagnetic core and some circuits to calibrate and filter the signals; the measurement is carried out by putting the probe on the specimen: each measurement takes up less than 20 seconds to permit an average of the main parameters: it makes is possible to carry out very fast inspection. It is obvious that this kind of probe is very useful and free from any additional apparatus for lift-off adjustment as the eddy current evaluation [3]; furthermore the dimension of the probe facilitates the ease of handling as shown in Figure 6.In this study 26 cast iron specimens with different hardness were prepared as the subjects. The measurements were carried out at 3 frequencies: 3 kHz, 6.8 kHz and 7.5 kHz. To show the reproducibility of the results we carried out a set of three measurements in three different days. The exitation voltage Vapplied in the measurements was 1.2 V to consider the reversible region in magnetization process.Table 1 lists the Brinell hardness of the specimens used in this investigation. The hardness measured in advance varies from 140 HB to 270 HB.Figure 6: AC magnetization measurement system.SpecimenIDTable 1: ID and hardness of specimens. Hardness (HB) Specimen Hardness (HB)ID 139169 14217171 144 16178 145179.512Figure 7 shows the output of the AC magnetization probe. In the Lissajous graphic of the waveforms, the results of specimens ID 36 (142 HB), 8 (202 HB) and 4 (273 HB) at the frequency 6800 Hz and exciting voltage 1.2 V are shown. The horizontal and vertical axes are represented by V applied and V pickup, respectively. Figure 8 plots the FFT analysis of the specimens of Figure 7 (b). This approach uses those measured data to characterize the structure of matrix.22146.525182150.51891514.1 Relation between Lissajous loop area and Hysteresis lossTaking the exciting voltage as the horizontal input and detected voltage as vertical input forms a Lissajous. An analogy is drawn between the magnetic hysteresis loop and the Lissajous from this experiment.23Is152 154 160 161 164193.5 196.5 201.5202 208.5 266.5273Loop Area trend - 3000 Hz - 1.2 V0.3519168Meas1 Meas2 A Meas34Results and discussionSpec 4 - 273 HB - Meas 1 - 6.8 kHz - 1.2 V1.510.50130150170190210 230 Hardness250270290-0.5-1(a) 3000Hz Loop Area trend - 6800 Hz-1.2 VV applied V pickup..-.-_1 (0???????????????????????025 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400(a) Output waveformsLLLissajous Graphic 60Loop AreaWORVERWARMIN1Meas 1Meas 2 A Meas 30.5-130150170190210 230 Hardness (HB)250270290ypickupSpec 36 Spec 8 Spec 40.902 -0.500.po-10.51150(b) 6800Hz50 =Loop Area trend - 7500 Hz-1.2 V-1Vapplied(b) Lissajous graphic Figure 7: Output of the AC magnetization probe. (a) Waveform of exciting and detected voltages of the specimen ID 4. (b) Lissajous graphic of which the horizontal and vertical axes correspond to exciting and detected voltages, respectively.Loop AreaMeas 1Meas 2 A Meas 3130150170190210 230 Hardness (HB)250270290FFT Analysis - 6800 Hz-Spec 36 -Spec 8 Spec 4(c) 7500 Hz Figure 9: Relation between the hardness and loop areas at 3000 Hz (a), 6800 Hz (b) and 7500 Hz (c), 1.2 V exciting voltage.wMagnitude (dB)WNWOOD12EUWA-IZUELEMESDELES ESCOLARE NOT CREATEELEIn magnetic hysteresis loop the horizontal axis is the field intensity H, and the vertical axis is the flux density B, corresponding to the exciting voltage and detected voltage respectively. The flux density is a measure of the intensity of the action of a magnetic field: there is the influence of the material and the magnetization state; while the detected voltage is a measure of the intensity of the action due to the exciting current. In this paper, the concepts of coercivity, residual, and hysteresis loss in magnetic hysteresis property, were used in AC voltage analysis as an equivalent value of AC magnetization. The equivalent values include an effect of eddy currents. The equivalent residual and coercivity are considered by ““residual like““ and ““coercive force like““, respectively. These equivalent parameters are defined as the zero-cross values of exciting and detected voltages, and the equivalent hysteresis loss HL isFrequency (Hz)Figure 8: FFT analysis of the detected voltage for the specimens ID 36 (142 HB), 8 (202 HB) and 4 (273 HB) at 6800 Hz.5490 10980 16470 21960 27451 32941 38431 43921 49411 54901 60391 65881 71371 76861 82351 87842 9333298822 9-149defined as the area of the Lissajous, as given bymaxHL=Ezvvv-EzVgdv-1Table 2: Constants of the regression curve with 1.2 V of exciting voltage Frequency (Hz)Averageerror (HB) 6800209.63 | 71.39 +10.65 0.87 7500202.71 55.34 +11.14 0.86Error of hardness estimation at 6800 Hzwhere Ve [V] and VD [V] denote the exciting and detected voltages, respectively; moreover Z is the impedance of probe.Figure 9 shows the trends of the loop area at 3000 Hz, 6800 Hz and 7500 Hz, 1.2 V exciting voltage. Three measurements were carried out in three different days to make sure the reproducibility. It is possible to observe the reproducibility of the results for both frequencies 6800 Hz and 7500 Hz in all measurements. In case of 3000 Hz, it is found that the different trend with the lack of reproducibility for values of hardness greater than 200 HB. The reason of these differences is that 3000 Hz is far to have the 3rd harmonic component intensity amplified and sensibility of probe is lower. Therefore we considered the results for 6800 Hz and 7500 Hz.Meas 1 Meas2 Meas3MUNWINError (HB)4.2 Regression curves of loop areaRegression curves based on the results of measurements in Figure 7 are formed from the following equation,Hardness (HB) (a) 6800HzHB = m X+q-2Error of hardness estimation at 7500 HzMeas 1 Meas 2Meas 3where X is the calculated value (e.g., loop area, residual like, etc.); m and q are constants determined from the measurements. Since the detected voltage depends on frequency and exciting voltage, then constants m and q are determined by each measurement condition. HB is the value of hardness in HB. Using these regression curves it is possible to estimate the hardness depending on loop area. Figure 10 plots the trend of hardness depending on loop area (HL) at 6800 Hz and 7500 Hz. The figure was obtained from the average of the three measurements as shown in Figure 9 (b) and (c).Error (HB)LEHardness trend - 6800 Hz - 1.2 V290mmHB= 209.63HL + 71.395R? = 0.8773Hardness (HB)(b) 7500Hz Figure 11: Histograms of errors of hardness estimation at 6800 Hz (a) and 7500 Hz (b). Error is defined as deviation from the regression curves in Figure 10.Hardness trendHardness (HB)Linear (Hardness trend)Hardness trend - 7500 Hz-1,2 VHardness trend0.20.30.40.5 0.6 Loop Area0.70.80.9Linear (Hardness trend)Hardness (HB)(a) 6800HzHardness trend - 7500 Hz -1.2 VHB = 167.6HL + 74.811R' = 0.9722HB = 202.71HL + 55.345R? = 0.86591300.20.30.40.5 0.6 Area Loop0.70.80.9Hardness (HB)Hardness trend(a) Loop area Error of hardness estimation at 7500 Hz.PLinear (Hardness trend)..Meas 1 Meas 2 Meas 30.20.30.40.50.70.80. 910.6 Loop AreaError (HB)(b) 7500Hz Figure 10: Trend of the hardness depending on loop area at 6800 Hz (a) and 7500 Hz (b), 1.2 V exciting voltage.0THardness (HB)Table 2 lists the constants m and q of the regression curves for the two different frequencies 6800 Hz and 7500 Hz. The R-squared value, also well-known as the coefficient of determination or correlation, reveals how closely the estimated values for the trend line correspond to the measured data: in other words it is a value that supply a ““goodness““ of regression curve. A trend line is most reliable when its R-squared value is at or closed to 1.(6) Error histogram Figure 12: Hardness trend (a) and histogram of error of hardness estimation at 7500 Hz (b). Specimens with high average error were not considered.--150The average value of the error in the estimation of the hardness from the value of loop area is defined as deviation from the regression curve. Figure 11 summarizes the errors in the histograms for the three measurements: here error is defined as deviation from the regression curve. It can be observed that only one specimen of 168 HB in hardness significantly deviates for the linear fitting curve of experimental data.Taking the consideration of the fact that Brinell hardness tests usually includes errors of about 10 HB, the present method enables estimation of hardness in ductile cast irons with good accuracy. In this case the average error is +10.65 HB and +11.14 HB for investigation at 6800 Hz and 7500 Hz, respectively.In case that some specimens with high average error are removed, it is possible to obtain R2=0.97 and average error ?’6.63 HB as shown in Figure 12.4.3 Coercive force like, residual like and 3rd harmonic component intensity trendWe analyzed also the trend of the equivalent coercivity and residual as ““coercive force like ““ and ““residual like““, respectively defined as the zero-cross values of exciting and detected voltages; and the trend of the normalized 3rd harmonic component intensity, which is defined as the ratio of intensities of the 3rd harmonic to the fundamental components. These values are strongly correlated with the loop area showing the similar results to the discussion in the Section 4.2. The error is also similarHardness trend - 6800 Hz-1.2 V290HB = 711.03CFL + 119.8R? -0.8517Hardness trendHardness (HB)Linear (Hardness trend)0.050.10.150.2Coercive Foree like (V)(a) Coercive Force likeHardness trend - 6800 Hz - 1.2 VHB = 876.38RL +99.93R= 0.8337 Hardness trend Linear (Hardness trend)Hardness (HB)130.00.020.040.060.08 0.1 Residual like (V)0.120.140.160.18(b) Residual likeHardness trend - 6800 Hz - 1.2 VHardness trendPLinear (Hardnesstrend)Hardness (HB)HB = 18.156TH + 654.55R? = 0.611-30-28 -26 -24 -22 Normalized 3rd harmonic component intensity (DB)-20(c) Normalized 3rd harmonic component intensity Figure 13: Trend of the hardness at 6800 Hz depending on ““coercive force like““ (a), ““residual like““ (b) and ““normalized 3rd harmonic component intensity?h (c). d of the hardness at 6800 Hz depending on ?gcoercive““residual like““ (b) and ““normalized 3rd harmonic sity?h (c).-151 -Figure 13 shows the trend of the hardness depending on ““coercive force like““, ““residual like““ and ““normalized 3rd harmonic component intensity at the frequency of 6800 Hz and exciting voltage of 1.2 V. It can be observed that the trends of Figure 13 correspond to that of Figure 10 (a). It is possible to observe that the values increase as the hardness increases, as anticipated in the Section 2.Figure 14 shows the trends for frequency of 7500 Hz, showing that the same trend as that in Figure 13. Table 3 summarizes the constants of the regression curve depending on frequency and on parameters, CFL, RL, and TH abbreviate is ““coercive force like““, ““residual like““, and ““normalized 3rd harmonic component intensity?h, respectively. It can be seen that the average errors are in line with loop area trend average error.Hardness trend - 7500 Hz - 1.2 VHardness trendLinear (Hardness trend)Hardness (HB)HB=733.28CFL + 87.508R? = 0.8820.050.150.1 Coercive Force like (V)0.20.25(a) Coercive Force likeHardness trend - 7500 Hz - 1.2 VHardness trendHardness (HB)Linear (Hardness trend)HB = 796.94RL + 68.213R? = 0.8562och0.050.20.250.10.15 Residual like (V)(b) Residual likeHardness trend - 7500 Hz - 1.2 VHB = 18.422TH + 715.3R?=0.7768Hardness (HB)Hardness trend Linear (Hardness trend)-34-20.32 .30 -28 -26 -24 -22Normalized 3rd harmonic component intensity (dB)(c) Normalized 3rd harmonic component intensity Figure 14: Trend of the hardness at 7500 Hz depending on ““coercive force like““ (a), ““residual like““ (b) and ““normalized 3rd harmonic component intensity' (c).Table 3: Constants of the regression curve with 1.2 V of exciting voltage for different parameters Parameter FrequencyAveraging (Hz)error(HB) CFL6800 711.03 | 119.8 +12.32 0.85 CFL7500 733.28 87.51 +9.39 RL 6800 852.11 | 100.9 +15.730.80 RL7500 796.94 68.21 +12.48 TH 6800 18.156 654.5 +23.090.61 TH7500 18.422 715.3 +18.56 0.780.880.855. SummaryIn this study, the AC magnetization method gives a novel material evaluation for cast irons. The particularity of this investigation is the amplified 3rd harmonic of the detected signals obtained in a range of frequencies between 5 kHz and 7 kHz. It could permit a high sensibility with low frequency and small dimension of probe.A set of 26 specimens was prepared with different casting. The followings can be drawn from AC magnetization measurements of specimens: 1. The AC magnetization method enables us to assessing thehardness of the ductile cast irons independently on theirchemical composition, thermal treatment and casting method. 2. The investigation presented in this paper provides anon-destructive evaluation of hardness of ductile cast irons with good accuracy in view of practical applications. This probe used in this investigation makes it possible to carry out fast evaluation of hardness without lift-off noise like in eddy current testing.AcknowledgmentsThe authors would like to thank the NDE Center of Japan Power Engineering and inspection Corporation (JAPEIC) for providing a probe. We acknowledge the specimens provided by Ube Steel Co., Ltd. We are also grateful to Professor Toshihiko Abe of Advanced System Evaluation Laboratory, Tohoku University for fruitful discussion and Takeshi Satoh of Institute of Fluid Science, Tohoku University for his technical assistance.References [1] D.N. Collins, W. Alcheikh, ?gUltrasonic non-destructive evaluation of the matrix structure and the graphite shape in cast iron?h Journal of Materials Processing Technology, Vol. 55, (1995), pp.85-90.[2] H. Kage, Y. Tanaka, ?gEvaluation of Cast Iron Quality by Ultrasonic Testing?h, Journal of the Japan Foundrymen's Society, Vol.56, No.7 (1984), pp.408-414, in Japanese.[3] K. Igawa et al., Basic & Application of Ductile Cast Irons, Maruzen, Tokyo, 1992, in Japanese.[4] C. D'Amato, C. Verdu, X. Kleber, G Regheeren and A. Vincent, ““Characterization of Austempered Ductile Cast iron Through Barkhausen Noise Measurements?h, Journal of Nondestructive Evaluation, Vol. 22, No. 4, December 2003, pp 127-139.151 T. Uchimoto, T. Takagi, S. Konoplyuk, T. Abe, H. Huang, M. Kurosawa, ““Eddy Current Evaluation of Cast Irons for Material Characterization““. Journal of Magnetism and Magnetic Material, 258-259 (2003), pp.493-496.[6] K.L. Feiste, P. Marques, Ch. Reichert, W. Reimche, D. Stegemann, A.J.M. Rebello, E.S. Kruger, ““Characterization of nodular cast iron properties by harmonic analysis of eddy current signals““, The e-Journal of Nondestructive Testing & Ultrasonics, Vol. 3, No.10 (1998), pp.245.[7] T. Uchimoto, T. Takagi, M. Kurosawa, S. Konoplyuk, T. Abe ““Material Evaluation of Cast Iron Using the Eddy Current Method ?h, Studies in Applied Electromagnetics and Mechanics 24, Electromagnetic Nondestructive Evaluation (VIII), (Eds. T. Sollier, D. Premel and D. Lesselier), (2003), pp.159-166.[8]R. M. Bozorth, Ferromagnetism, NJ: Princeton, 1951[9] W. Cheng, M. Shiwa, G Chen, H. Yoneyama and Y. Horii, ?gFinite element simulation of magneto inductive evaluation of PWHT temperature of Cr-Mo steel weld joints““, Int. J. Appl. Electromagn. Mech., Vol. 19 (2004) No. 1-4, pp. 125-130.[10] JP Patent No. 2001 - 255305.? 152“ “交流磁化プローブを用いた鋳造構造物の材質評価“ “Luca BARTOLOMEO,Hisashi ENDO,Tetsuya UCHIMOTO,Mitsuharu SHIWA,Toshiyuki TAKAGI“ “交流磁化プローブを用いた鋳造構造物の材質評価“ “Luca BARTOLOMEO,Hisashi ENDO,Tetsuya UCHIMOTO,Mitsuharu SHIWA,Toshiyuki TAKAGI
Ductile cast iron is cast iron in which the graphite is present as tiny spheres: this feature permits greater strength and greater ductility thanFerrite gray cast iron of similar composition. There are so many applications of ductile cast irons, for instance, automotive-crankshafts, pistons, motor frames, levels, furnace doors, electrical fittings, switch boxesPearlite and tools for aqueducts; then the nondestructive evaluation of its mechanical properties is of paramount importance for the maintenance of various kinds of apparatus as well as structures. Two factors, the morphology of graphite and the type of matrices,
259 HB mainly determine mechanical properties of cast irons. With the goodGraphite sensitivity to graphite morphology, ultrasonic techniques have been used for the evaluation of mechanical properties in cast iron [1,2].In case of ductile cast iron, mechanical properties depend on both the nodularity of the graphite and the properties of matrices. Because it is difficult for ultrasonic techniques to assess the structure of matrices, direct observation and destructive testing such as hardness tests are usually carried out in the final analysis [3,4]. Therefore, it is essentially required to develop methods to assess the properties of matrices by means of non-destructive evaluation [5]. The matrix of ductile cast iron consists of structure in term of proportions of pearlite and ferrite. As Figure 1: Example of microstructure of 4 ductile cast iron specimens the quantity of pearlite increases, the strength and hardness of the with different hardness, 143 HB, 162 HB, 215 HB and 259 HB material increase. The proportion of ferrite and pearlite to the materials respectively. White and gray colors indicate the ferrite and pearlite, principally determines ductility and impact properties.respectively. Dark ?gspots““ indicate the graphite. Figure 1 shows the optical microscopy photos of four specimens with difference in hardness, 143 HB, 162 HB, 215 HB and 259 HB; white The AC magnetization probe used in this study includes simple data and gray colors indicate the ferrite and pearlite, respectively; moreover acquisition and good accuracy, in order to accommodate practical usage. the dark ““spots?h show the graphite: it can be seen that the pearlite ratio To emphasize the permeability variation, the AC magnetization probe increases as hardness increases. Since the ferrite and pearlite have has a resonant circuit tuning up for about three times higher than the difference in electric as well as magnetic materials, then fundamental excitation frequency. It is an effective technique to assess non-destructive ways utilizing electromagnetic field phenomena make the properties of matrices and it is expected a complete evaluation of it possible to evaluate the properties of matrices.mechanical properties of ductile cast iron. This nondestructive, as well There are some methods to evaluate the matrices in non-destructive as inexpensive approach can be realized through the combination of the way by using harmonic analysis of eddy current testing [6] and eddy present method with [6] for the matrices, and ultrasonic testing for current evaluation concentrated in the reversible region of initial graphite morphology. magnetization curve [7). This paper also considers the weakiy magnetized region with a low intensity of excitation. This means that 2. AC magnetization method eddy current signal reflects on only the two parameters, namely, The AC magnetization method consists of calculating magnetic flux conductivity and permeability around Rayleigh region. Because of density B[T) as a reaction of an alternating applied magnetic field reversible region on magnetization, it guarantees in addition the H[A/m). In the ferromagnetic materials the function B = uH is not reproducibility of experiment [8]. Considering the relation between the linear, it depends on the non-linearity of the permeability on varying of hardness and the structure of matrix, it is possible to estimate the the magnetic field. hardness of ductile cast iron efficiently. In addition, the hardness covers The initial magnetization curve is the relation between H and B for a several other properties of materials, as resistance to deformation, ferromagnetic material virgin, which is material that has not had an resistance to friction and abrasion. There are two main differences with influence of magnetic field yet. Figure 2 is a schematic diagram of the eddy current testing. One is to use relatively lower frequency, in a initial magnetization curve. Generally, it is possible to classify three range of 6-7 kHz, to catch permeability variation in each of hardness. zones of interest. The other is to employ the amplified 3rd harmonic component of sensor output voltage to easily distinguish the differences in hysteresis loops.
H(A/m Figure 2: Initial magnetization curve: the points 1,2 and 3 indicate the division of the curve in three regions.The region 0-1 is called reversible or Rayleigh region and it is considered in this investigation to guarantee the reproducibility of the measurements. In this region the material returns to its original state if a reversing magnetic field is applied. One of the principal problems is how to comprehend that the investigation is carrying out in the region. On the other hand, the regions 1-2 and 2-3 are non-reversible regions. in these regions, the residual effects of the magnetization, usually shown in the magnetic hysteresis loop (Figure 3), are exhibited, meaning that it is difficult to obtain good reproducibility without specific demagnetization preconditionings.B [T] ABoH[A/mFigure 3: Hysteresis loop: the parameters to be evaluated in a hysteresis loop are the area, the positive zero-cross value of the magnetic field, called ““coercivity and the zero-cross value of the flux density, called ““residual?h.Hysteresis curve1.5B(T)Ferrite based Pearlite based-1.5H (A/m)Figure 4: Hysteresis loops of Ferrite based and Pearlite based cast irons.The hysteresis loop is the relation between the magnetic field H and the flux density B for a ferromagnetic material under an alternating magnetic field. In this case it is possible to observe that the magnetization curve traces different ways after the initial magnetization curve is overtaken. Br, called residual, and Hc, called coercivity, represent the effects of the residual fields. The shape of the loop depends on the material and the area of the loop is proportional to the losses in term of energy.Since the hysteresis loop changes its characteristics on changing the materials, then the AC magnetization approach uses those concepts to evaluate the properties of the materials. Figure 4 shows the differences in term of hysteresis loop for the ferrite- and pearlite-based cast irons. The pearlite-based cast iron presents a wider loop, meaning that it is-148hard in terms of magnetic and mechanical properties. In this case the probe with the 3rd harmonic amplified will be more sensible and the normalized value will be higher, as we will show in the experimental results. Furthermore the maximum value of B makes it possible to identify the structure of the matrix.3. AC magnetization measurement systemThe evaluation of cast irons was carried out with experimental setup showed schematically in Figure 5. The probe used in the system consists of two coaxial coils with differential connection [9].The specimens were magnetized by supplying sinusoidal voltage to the probe at frequencies in a range of 6-7 kHz. The system presents the amplified 3rd harmonic of detected voltage by a circuit of which resonance frequency is about 18 kHz [10]. With the intention of show the differences of measurements in the non-resonance frequencies it was investigated also the 3kHz.The exciting and detecting voltages, the values of the harmonics ratio in dB and the area of loop in Lissajous graphics were recorded by using a PCMCIA National Instrument DAQ Card 6062E; the software to control the excitation and detection, and to save the data on file was realized in LabVIEW 7.0 under a Laptop with 384 MB RAM and 1.13 GHz Intel Pentium III Processor.Figure 5: Experimental Setup: it is composed of the exciter, ferromagnetic core and system to detect the Vpickup.All instrumentation weighs about 1 kg and takes up as a book: it guarantees the portability of the probe device. It consists of a DAQ Card for Analog Digital conversion, a battery and a probe which contains the coils, ferromagnetic core and some circuits to calibrate and filter the signals; the measurement is carried out by putting the probe on the specimen: each measurement takes up less than 20 seconds to permit an average of the main parameters: it makes is possible to carry out very fast inspection. It is obvious that this kind of probe is very useful and free from any additional apparatus for lift-off adjustment as the eddy current evaluation [3]; furthermore the dimension of the probe facilitates the ease of handling as shown in Figure 6.In this study 26 cast iron specimens with different hardness were prepared as the subjects. The measurements were carried out at 3 frequencies: 3 kHz, 6.8 kHz and 7.5 kHz. To show the reproducibility of the results we carried out a set of three measurements in three different days. The exitation voltage Vapplied in the measurements was 1.2 V to consider the reversible region in magnetization process.Table 1 lists the Brinell hardness of the specimens used in this investigation. The hardness measured in advance varies from 140 HB to 270 HB.Figure 6: AC magnetization measurement system.SpecimenIDTable 1: ID and hardness of specimens. Hardness (HB) Specimen Hardness (HB)ID 139169 14217171 144 16178 145179.512Figure 7 shows the output of the AC magnetization probe. In the Lissajous graphic of the waveforms, the results of specimens ID 36 (142 HB), 8 (202 HB) and 4 (273 HB) at the frequency 6800 Hz and exciting voltage 1.2 V are shown. The horizontal and vertical axes are represented by V applied and V pickup, respectively. Figure 8 plots the FFT analysis of the specimens of Figure 7 (b). This approach uses those measured data to characterize the structure of matrix.22146.525182150.51891514.1 Relation between Lissajous loop area and Hysteresis lossTaking the exciting voltage as the horizontal input and detected voltage as vertical input forms a Lissajous. An analogy is drawn between the magnetic hysteresis loop and the Lissajous from this experiment.23Is152 154 160 161 164193.5 196.5 201.5202 208.5 266.5273Loop Area trend - 3000 Hz - 1.2 V0.3519168Meas1 Meas2 A Meas34Results and discussionSpec 4 - 273 HB - Meas 1 - 6.8 kHz - 1.2 V1.510.50130150170190210 230 Hardness250270290-0.5-1(a) 3000Hz Loop Area trend - 6800 Hz-1.2 VV applied V pickup..-.-_1 (0???????????????????????025 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400(a) Output waveformsLLLissajous Graphic 60Loop AreaWORVERWARMIN1Meas 1Meas 2 A Meas 30.5-130150170190210 230 Hardness (HB)250270290ypickupSpec 36 Spec 8 Spec 40.902 -0.500.po-10.51150(b) 6800Hz50 =Loop Area trend - 7500 Hz-1.2 V-1Vapplied(b) Lissajous graphic Figure 7: Output of the AC magnetization probe. (a) Waveform of exciting and detected voltages of the specimen ID 4. (b) Lissajous graphic of which the horizontal and vertical axes correspond to exciting and detected voltages, respectively.Loop AreaMeas 1Meas 2 A Meas 3130150170190210 230 Hardness (HB)250270290FFT Analysis - 6800 Hz-Spec 36 -Spec 8 Spec 4(c) 7500 Hz Figure 9: Relation between the hardness and loop areas at 3000 Hz (a), 6800 Hz (b) and 7500 Hz (c), 1.2 V exciting voltage.wMagnitude (dB)WNWOOD12EUWA-IZUELEMESDELES ESCOLARE NOT CREATEELEIn magnetic hysteresis loop the horizontal axis is the field intensity H, and the vertical axis is the flux density B, corresponding to the exciting voltage and detected voltage respectively. The flux density is a measure of the intensity of the action of a magnetic field: there is the influence of the material and the magnetization state; while the detected voltage is a measure of the intensity of the action due to the exciting current. In this paper, the concepts of coercivity, residual, and hysteresis loss in magnetic hysteresis property, were used in AC voltage analysis as an equivalent value of AC magnetization. The equivalent values include an effect of eddy currents. The equivalent residual and coercivity are considered by ““residual like““ and ““coercive force like““, respectively. These equivalent parameters are defined as the zero-cross values of exciting and detected voltages, and the equivalent hysteresis loss HL isFrequency (Hz)Figure 8: FFT analysis of the detected voltage for the specimens ID 36 (142 HB), 8 (202 HB) and 4 (273 HB) at 6800 Hz.5490 10980 16470 21960 27451 32941 38431 43921 49411 54901 60391 65881 71371 76861 82351 87842 9333298822 9-149defined as the area of the Lissajous, as given bymaxHL=Ezvvv-EzVgdv-1Table 2: Constants of the regression curve with 1.2 V of exciting voltage Frequency (Hz)Averageerror (HB) 6800209.63 | 71.39 +10.65 0.87 7500202.71 55.34 +11.14 0.86Error of hardness estimation at 6800 Hzwhere Ve [V] and VD [V] denote the exciting and detected voltages, respectively; moreover Z is the impedance of probe.Figure 9 shows the trends of the loop area at 3000 Hz, 6800 Hz and 7500 Hz, 1.2 V exciting voltage. Three measurements were carried out in three different days to make sure the reproducibility. It is possible to observe the reproducibility of the results for both frequencies 6800 Hz and 7500 Hz in all measurements. In case of 3000 Hz, it is found that the different trend with the lack of reproducibility for values of hardness greater than 200 HB. The reason of these differences is that 3000 Hz is far to have the 3rd harmonic component intensity amplified and sensibility of probe is lower. Therefore we considered the results for 6800 Hz and 7500 Hz.Meas 1 Meas2 Meas3MUNWINError (HB)4.2 Regression curves of loop areaRegression curves based on the results of measurements in Figure 7 are formed from the following equation,Hardness (HB) (a) 6800HzHB = m X+q-2Error of hardness estimation at 7500 HzMeas 1 Meas 2Meas 3where X is the calculated value (e.g., loop area, residual like, etc.); m and q are constants determined from the measurements. Since the detected voltage depends on frequency and exciting voltage, then constants m and q are determined by each measurement condition. HB is the value of hardness in HB. Using these regression curves it is possible to estimate the hardness depending on loop area. Figure 10 plots the trend of hardness depending on loop area (HL) at 6800 Hz and 7500 Hz. The figure was obtained from the average of the three measurements as shown in Figure 9 (b) and (c).Error (HB)LEHardness trend - 6800 Hz - 1.2 V290mmHB= 209.63HL + 71.395R? = 0.8773Hardness (HB)(b) 7500Hz Figure 11: Histograms of errors of hardness estimation at 6800 Hz (a) and 7500 Hz (b). Error is defined as deviation from the regression curves in Figure 10.Hardness trendHardness (HB)Linear (Hardness trend)Hardness trend - 7500 Hz-1,2 VHardness trend0.20.30.40.5 0.6 Loop Area0.70.80.9Linear (Hardness trend)Hardness (HB)(a) 6800HzHardness trend - 7500 Hz -1.2 VHB = 167.6HL + 74.811R' = 0.9722HB = 202.71HL + 55.345R? = 0.86591300.20.30.40.5 0.6 Area Loop0.70.80.9Hardness (HB)Hardness trend(a) Loop area Error of hardness estimation at 7500 Hz.PLinear (Hardness trend)..Meas 1 Meas 2 Meas 30.20.30.40.50.70.80. 910.6 Loop AreaError (HB)(b) 7500Hz Figure 10: Trend of the hardness depending on loop area at 6800 Hz (a) and 7500 Hz (b), 1.2 V exciting voltage.0THardness (HB)Table 2 lists the constants m and q of the regression curves for the two different frequencies 6800 Hz and 7500 Hz. The R-squared value, also well-known as the coefficient of determination or correlation, reveals how closely the estimated values for the trend line correspond to the measured data: in other words it is a value that supply a ““goodness““ of regression curve. A trend line is most reliable when its R-squared value is at or closed to 1.(6) Error histogram Figure 12: Hardness trend (a) and histogram of error of hardness estimation at 7500 Hz (b). Specimens with high average error were not considered.--150The average value of the error in the estimation of the hardness from the value of loop area is defined as deviation from the regression curve. Figure 11 summarizes the errors in the histograms for the three measurements: here error is defined as deviation from the regression curve. It can be observed that only one specimen of 168 HB in hardness significantly deviates for the linear fitting curve of experimental data.Taking the consideration of the fact that Brinell hardness tests usually includes errors of about 10 HB, the present method enables estimation of hardness in ductile cast irons with good accuracy. In this case the average error is +10.65 HB and +11.14 HB for investigation at 6800 Hz and 7500 Hz, respectively.In case that some specimens with high average error are removed, it is possible to obtain R2=0.97 and average error ?’6.63 HB as shown in Figure 12.4.3 Coercive force like, residual like and 3rd harmonic component intensity trendWe analyzed also the trend of the equivalent coercivity and residual as ““coercive force like ““ and ““residual like““, respectively defined as the zero-cross values of exciting and detected voltages; and the trend of the normalized 3rd harmonic component intensity, which is defined as the ratio of intensities of the 3rd harmonic to the fundamental components. These values are strongly correlated with the loop area showing the similar results to the discussion in the Section 4.2. The error is also similarHardness trend - 6800 Hz-1.2 V290HB = 711.03CFL + 119.8R? -0.8517Hardness trendHardness (HB)Linear (Hardness trend)0.050.10.150.2Coercive Foree like (V)(a) Coercive Force likeHardness trend - 6800 Hz - 1.2 VHB = 876.38RL +99.93R= 0.8337 Hardness trend Linear (Hardness trend)Hardness (HB)130.00.020.040.060.08 0.1 Residual like (V)0.120.140.160.18(b) Residual likeHardness trend - 6800 Hz - 1.2 VHardness trendPLinear (Hardnesstrend)Hardness (HB)HB = 18.156TH + 654.55R? = 0.611-30-28 -26 -24 -22 Normalized 3rd harmonic component intensity (DB)-20(c) Normalized 3rd harmonic component intensity Figure 13: Trend of the hardness at 6800 Hz depending on ““coercive force like““ (a), ““residual like““ (b) and ““normalized 3rd harmonic component intensity?h (c). d of the hardness at 6800 Hz depending on ?gcoercive““residual like““ (b) and ““normalized 3rd harmonic sity?h (c).-151 -Figure 13 shows the trend of the hardness depending on ““coercive force like““, ““residual like““ and ““normalized 3rd harmonic component intensity at the frequency of 6800 Hz and exciting voltage of 1.2 V. It can be observed that the trends of Figure 13 correspond to that of Figure 10 (a). It is possible to observe that the values increase as the hardness increases, as anticipated in the Section 2.Figure 14 shows the trends for frequency of 7500 Hz, showing that the same trend as that in Figure 13. Table 3 summarizes the constants of the regression curve depending on frequency and on parameters, CFL, RL, and TH abbreviate is ““coercive force like““, ““residual like““, and ““normalized 3rd harmonic component intensity?h, respectively. It can be seen that the average errors are in line with loop area trend average error.Hardness trend - 7500 Hz - 1.2 VHardness trendLinear (Hardness trend)Hardness (HB)HB=733.28CFL + 87.508R? = 0.8820.050.150.1 Coercive Force like (V)0.20.25(a) Coercive Force likeHardness trend - 7500 Hz - 1.2 VHardness trendHardness (HB)Linear (Hardness trend)HB = 796.94RL + 68.213R? = 0.8562och0.050.20.250.10.15 Residual like (V)(b) Residual likeHardness trend - 7500 Hz - 1.2 VHB = 18.422TH + 715.3R?=0.7768Hardness (HB)Hardness trend Linear (Hardness trend)-34-20.32 .30 -28 -26 -24 -22Normalized 3rd harmonic component intensity (dB)(c) Normalized 3rd harmonic component intensity Figure 14: Trend of the hardness at 7500 Hz depending on ““coercive force like““ (a), ““residual like““ (b) and ““normalized 3rd harmonic component intensity' (c).Table 3: Constants of the regression curve with 1.2 V of exciting voltage for different parameters Parameter FrequencyAveraging (Hz)error(HB) CFL6800 711.03 | 119.8 +12.32 0.85 CFL7500 733.28 87.51 +9.39 RL 6800 852.11 | 100.9 +15.730.80 RL7500 796.94 68.21 +12.48 TH 6800 18.156 654.5 +23.090.61 TH7500 18.422 715.3 +18.56 0.780.880.855. SummaryIn this study, the AC magnetization method gives a novel material evaluation for cast irons. The particularity of this investigation is the amplified 3rd harmonic of the detected signals obtained in a range of frequencies between 5 kHz and 7 kHz. It could permit a high sensibility with low frequency and small dimension of probe.A set of 26 specimens was prepared with different casting. The followings can be drawn from AC magnetization measurements of specimens: 1. The AC magnetization method enables us to assessing thehardness of the ductile cast irons independently on theirchemical composition, thermal treatment and casting method. 2. The investigation presented in this paper provides anon-destructive evaluation of hardness of ductile cast irons with good accuracy in view of practical applications. This probe used in this investigation makes it possible to carry out fast evaluation of hardness without lift-off noise like in eddy current testing.AcknowledgmentsThe authors would like to thank the NDE Center of Japan Power Engineering and inspection Corporation (JAPEIC) for providing a probe. We acknowledge the specimens provided by Ube Steel Co., Ltd. We are also grateful to Professor Toshihiko Abe of Advanced System Evaluation Laboratory, Tohoku University for fruitful discussion and Takeshi Satoh of Institute of Fluid Science, Tohoku University for his technical assistance.References [1] D.N. Collins, W. Alcheikh, ?gUltrasonic non-destructive evaluation of the matrix structure and the graphite shape in cast iron?h Journal of Materials Processing Technology, Vol. 55, (1995), pp.85-90.[2] H. Kage, Y. Tanaka, ?gEvaluation of Cast Iron Quality by Ultrasonic Testing?h, Journal of the Japan Foundrymen's Society, Vol.56, No.7 (1984), pp.408-414, in Japanese.[3] K. Igawa et al., Basic & Application of Ductile Cast Irons, Maruzen, Tokyo, 1992, in Japanese.[4] C. D'Amato, C. Verdu, X. Kleber, G Regheeren and A. Vincent, ““Characterization of Austempered Ductile Cast iron Through Barkhausen Noise Measurements?h, Journal of Nondestructive Evaluation, Vol. 22, No. 4, December 2003, pp 127-139.151 T. Uchimoto, T. Takagi, S. Konoplyuk, T. Abe, H. Huang, M. Kurosawa, ““Eddy Current Evaluation of Cast Irons for Material Characterization““. Journal of Magnetism and Magnetic Material, 258-259 (2003), pp.493-496.[6] K.L. Feiste, P. Marques, Ch. Reichert, W. Reimche, D. Stegemann, A.J.M. Rebello, E.S. Kruger, ““Characterization of nodular cast iron properties by harmonic analysis of eddy current signals““, The e-Journal of Nondestructive Testing & Ultrasonics, Vol. 3, No.10 (1998), pp.245.[7] T. Uchimoto, T. Takagi, M. Kurosawa, S. Konoplyuk, T. Abe ““Material Evaluation of Cast Iron Using the Eddy Current Method ?h, Studies in Applied Electromagnetics and Mechanics 24, Electromagnetic Nondestructive Evaluation (VIII), (Eds. T. Sollier, D. Premel and D. Lesselier), (2003), pp.159-166.[8]R. M. Bozorth, Ferromagnetism, NJ: Princeton, 1951[9] W. Cheng, M. Shiwa, G Chen, H. Yoneyama and Y. Horii, ?gFinite element simulation of magneto inductive evaluation of PWHT temperature of Cr-Mo steel weld joints““, Int. J. Appl. Electromagn. Mech., Vol. 19 (2004) No. 1-4, pp. 125-130.[10] JP Patent No. 2001 - 255305.? 152“ “交流磁化プローブを用いた鋳造構造物の材質評価“ “Luca BARTOLOMEO,Hisashi ENDO,Tetsuya UCHIMOTO,Mitsuharu SHIWA,Toshiyuki TAKAGI“ “交流磁化プローブを用いた鋳造構造物の材質評価“ “Luca BARTOLOMEO,Hisashi ENDO,Tetsuya UCHIMOTO,Mitsuharu SHIWA,Toshiyuki TAKAGI