Precursor of hydroigen induced glass lining chipping by AE monitoring.
Glass lining serves as corrosion-resistant barrier and is composed of two layers, i.e., a porous ground coat and a dense top coat. Pores in the ground-coat serves for moderating the miss-matches of mechanical and thermal properties of steel substrate and brittle glass layer, and also for trapping the hydrogen gas induced via the substrate. Chipping is a sudden fly-off of the lining layer and cannot be repaired on site. We monitored precursor of the chipping by AE and laser surface-acoustic wave (LSAW) technique. We monitored AE signals before half of the time to cause final large scale chipping. The time of first AE coincides the time when hydrogen ion reaches the ground coat. AE are supposed to be produced by the cracks connecting the pores in the ground coat. LSAW technique was demonstrated to be useful in monitoring the precursor of the chipping.
Keywords: Glass lining, Chipping, Hydrogen, Laser ultrasonic, Source dynamics, Waveform matching
Glass linings (GLs) are widely used for corrosion-resistant equipment in food and chemical industries, but tend to suffer various damages due to their brittleness. Among the damage of GLs, hydrogen-induced damage, generally known as chipping, is a serious problem since no effective countermeasure exists . Utilization of steels with a trace of titanium is recognized to be effective for trapping the hydrogen gas in the steels, but is not the essential countermeasures.
Chipping sometimes occurs after a couple of days or after a few years of service. The worst case is the chipping in the manufacturing locations. Previous research suggested that the chipping was induced by diffusible hydrogen, which was induced into the substrate steel from water in the clay, milling water, and cleaning by acid solutions. No method to evaluate the susceptibility to chipping has been established [1-3]. Glass linings are generally composed of two layers; porous ground coat near the steel substrate and dense cover coat. Pores in the ground coat are designed to moderate the stress mismatch between the steel substrate and GL, and to trap the hydrogen gas and hence contribute to reduce the chipping.
We attempted to study the precursor of chipping by AE and LSAW (Laser Surface-Acoustic Wave) monitoring. AE was demonstrated to detect the micro-cracking of the porous ground coat (GC hereafter) from early times, and the LSAW the change of acoustic properties associated with the micro-cracking of the GC.
2. Specimen and Acoustic Properties of Glass Lining
Specimen is a corrosion-resistant GL. It is composed of 0.2-mm thick ground coat (GC) and 0.8-mm thick cover coat (CC). This GL was deposited on carbon steel (JIS SS400) of 3-mm thickness, 61-mm width and 100-mm length. Figure 1 shows a typical cross-sectional structure of the GL. The GC contains a number of large spherical pores of 50 to 200 [micro]m diameter while the CC isolated fine pores of 10 to 50 [micro]m. Porosity by a laser microscope was measured as 3.5 % in the CC and 48 % in the GC. Chemical composition and baking conditions are not reported.
[FIGURE 1 OMITTED]
[FIGURE 2 OMITTED]
First we measured the SAW velocities of the GL using a laser ultrasonic system [4, 5]. Figure 2(a) shows the waveforms of dispersive Rayleigh waves at a propagation distance of 12.12 mm in the axial direction of the specimen. The Rayleigh velocity, determined by the zero-crossing time method, was from 3191 m/s to 3214 m/s. The velocities of SSCW (Surface skimming compressive wave) were measured as 6530 [+ or -] 40 m/s. Estimated acoustic and Young's modulus are shown in Table 1. Figure 2(b) shows the wavelet contour map of LSA waves. It indicates that the velocity at higher frequencies above 4 MHz or in the surface layer less than 0.8 mm is non-dispersive, suggesting a homogenous structure of the GL. Late-arrival lower frequency components (surface depth over 0.8 mm) indicate that the GC layer has a lower velocity than the steel substrate, possibly due to the large voids in the GC.
3. AE Monitoring of Chipping
Figure 3 shows AE monitoring method from the GL chipping during hydrogen charging. Hydrogen was supplied to the steel substrate in an area of 20 mm square by cathodically charging at current density of 0.05 A/[cm.sup.2] in a 0.5 kmol/[m.sup.3] sulfuric acid solution with 4 kg/[m.sup.3] C[H.sub.4][N.sub.2]. Eight small AE sensors (PAC, Type PICO with center frequency of 0.45 MHz) were mounted on both surfaces of the specimen. Sensor outputs were amplified 40 dB and digitized by an A/D converter. The AE system with 40 dB amplification did not detect any noise from hydrogen gas bubbles. A plot of cumulative AE counts with charging time is shown in Fig. 4.
AE events increased from 38 ks (10.5 hr). A large-scale chipping, shown in Fig. 5, of about 30 mm x 50 mm (1200 [mm.sup.2]) suddenly occurred at 48.6 ks (13.5 hr). A number of fragmented glass pieces from 0.5 [mm.sup.2] to 200 [mm.sup.2] were launched from the substrate with large sounds. We observed at the right edge of the exfoliated GL a porous GC of 2 mm x 4 mm square. First arrival time of diffusible hydrogen at GL/substrate is calculated as 11.3 ks by using the hydrogen diffusivity coefficient of 5.22 x [10.sup.-7] [cm.sup.2]/s in the steel. Thus the time lag of 11 ks between 11.3 ks and 22 ks, at which the first AE was detected is supposed to be the time during which the hydrogen gas pressure increases in the voids of the GC. In another word, this is the time during which the voids serve as the gas container.
Figure 6 shows typical AE waveforms detected at 43407 and 48646 s. As AE signals were detected as Lamb waves with weak So-component, the source locations were estimated using the first arrival sheet velocity of the So-wave. The sheet velocity of the substrate and GL is approximately the same since both the Young's modulus and density of the substrate are approximately 3 times that of the GL. In spite of this fact, we observed some strange waveforms for channel 1 and 2. For the event at 48646 s just before the final large chipping, the arrival of the So-packet to the Ch.-1 and -2 sensors on the steel substrate were very weak, while the amplitudes of So-component from Ch.-1' and -2' sensors on the GL were much stronger.
[FIGURE 3 OMITTED]
[FIGURE 4 OMITTED]
Strange characteristics observed here may be due to the damage location in the GC. Strong Lamb waves propagated along GL layer, of which GC was partially exfoliated from the substrate. Source location was performed for 40 events with sufficient amplitude on all channels and results are shown in Fig. 5. Most sources were located in a small "fish-scale"-type chipping area. Surface and transverse photos in this area show cracks connecting the voids. Crack openings are measured as 0.01 to 0.04 mm. These values are surprisingly large compared to those of delayed fracture of the high strength steels.
[FIGURE 5 OMITTED]
From these data, it is concluded that 1) voids in GC can hold the hydrogen gas pressure, 2) fish-scale chipping is a typical delayed fracture and occurs when the hydrogen gas pressure reaches a critical pressure, 3) AE signals are produced by the generation of micro-cracks connecting the voids in the GC. Thus, AE can be used to monitor the precursor of the large sudden chipping of the GL.
[FIGURE 6 OMITTED]
[FIGURE 7 OMITTED]
4. AE Source Analysis
Another experiment, approximately the same as that in the previous section, was attempted. In this test, damages monitoring by the LSAW system was simultaneously performed. As shown in Fig. 7, LSA waves were monitored on the surface of the GL. Figure 8 shows the cumulative AE counts with the charging time. A large-scale final chipping of 2000 [mm.sup.2] occurred at 60842 s (17 hr). AE counts rapidly increased at 20 ks just before the final chipping. Timing of both the first AE and final chipping agreed with those in Fig. 4. Twenty-seven events are located at the origin of the chipping, as shown in Fig. 9.
[FIGURE 8 OMITTED]
Fracture dynamics were estimated by the waveform matching of the first-arrival So-component. Detail of the waveform simulation can be found elsewhere . The overall transfer function was determined utilizing the pulse-laser breakdown of silicone grease in a slit as the source. Source parameters of the break-down was first determined by the deconvolution of the out-of-plane displacement of a large block measured by a laser interferometer with the theoretical Green's function of the second kind . Source parameters of chipping AE signals were estimated by matching the first So-mode waveform computed using the overall transfer function to the detected waveform. Figure 10 compares experimental and computed waveforms of the So-packets. Source parameters are shown in Fig. 11. The average crack volume 3.5x[10.sup.-16] [m.sup.3] appears to correspond to the volumes of the cracks connecting the voids in the GC.
[FIGURE 9 OMITTED]
[FIGURE 10 OMITTED]
5. Monitoring of Lining Degradation by LSAW
As mentioned in the previous section, the low-frequency components of the dispersive Rayleigh wave represent the characteristics of GC. Thus, we monitored periodically the SAWs of the GL using the laser ultrasonic system. Figure 12 shows the change of the SAWs with time. Slight waveform difference can be seen in the later times. Cross-correlation was determined and is shown in Fig. 13. It decreases with the charging time, but shows three discontinuous changes at around 11 ks, 45 and 55 ks. Here, the 10 ks is the time when the hydrogen reaches the GC. At around 50 ks as shown in Fig. 8, AE events increased rapidly. Cross-correlation of SAWs can predict the minute changes in the GC when the SAW is measured at the same position.
[FIGURE 11 OMITTED]
[FIGURE 12 OMITTED]
[FIGURE 13 OMITTED]
AE and LSAW were monitored to study the precursor of hydrogen-induced chipping of the glass lining. Results obtained are summarized below:
1) Acoustic properties of Glass lining were studied using an LSAW system. Both the Young's modulus and density of the cover coat are approximately 3 times smaller than those of the steel substrate, but the sheet velocity was found to be the same for the dense cover coat and the substrate. Group velocity dispersion of the Rayleigh waves represents the structural change in the porous ground coat.
2) Large scale chipping of the cover coat, deposited on 3-mm thick carbon steel, occurred at 49 ks to 60 ks after the hydrogen charging to the substrate. AE signals are produced by micro-cracking, connecting the voids in the ground coat. These are detected from less than half the time of the final chipping. Hydrogen gas pressure in the ground coat produces micro-cracking when its pressure reaches the critical value.
3) Source parameters of micro-cracks were estimated by the waveform matching of the first arrive So-mode, using the overall transfer function experimentally determined. Breakdown of silicone grease in a narrow slit is useful in simulating the Mode-I crack normal to surface in the thin ground coat. Crack volume and rise time were estimated as 3.5 x 10-16 [m.sup.3] and less than 1 [micro]s, respectively.
4) LSAW measured on the lining showed slight changes reflecting the structural changes in the ground coat. Cross-correlation factor of the SAWs showed discontinuous changes corresponding to those observed for AE count change.
 C.G. Bergeron, J. Amer. Ceramic Soc. 36 (1953), 373.
 Y. Nakazato, H. Kubiminato, N. Fukuda, T. Sonobe and H. Nagaishi, Kawasaki Seitetsu Giho in Japanese, 7 (1977), 70.
 I. Takahashi, A. Yasuda. K. Ito and N. Ohashi, Kawasaki Seitetsu Giho in Japanese, 5 (1975), 53.
 H. Cho, S. Ogawa, K. Yamanaka and M. Takemoto, JSME Int. Jr. Series A, 41 (1998) 439.
 Y. Mizutani, M. Takemoto and K. Ono, J. Acoustic Emission, 16 (1998) S115.
 H. Suzuki, M. Takemoto and K. Ono, J.Acoustic Emission, 14 (1995) 442.
 H. Suzuki, M. Takemoto and K. Ono, Progress in AE, VII (1994), JSNDI, Tokyo, p. 457.
KOHEI MURAKAMI AND MIKIO TAKEMOTO
Faculty of Science and Engineering, Aoyama Gakuin University, Fuchinobe, Sagamihara, Kanagawa, 229-8558 Japan.
Table 1 Acoustic and elastic properties. [rho] [C.sub.p] [C.sub.s] [C.sub.r] E kg/[m.sup.3] mf s m/s m/s Gpa GL 2460 6539 3447 3201 76.4 SS400 7870 5920 3255 3010 214
|Printer friendly Cite/link Email Feedback|
|Author:||Murakami, Kohei; Takemoto, Mikio|
|Publication:||Journal of Acoustic Emission|
|Date:||Jan 1, 2005|
|Previous Article:||The origin of continuous emissions.|
|Next Article:||Real-time executing source location system applicable to anisotropic thin structures.|