AE Source and Relation between AE Activity and Rate of Corrosion of Oil Tank Bottom Plate on Acidic Soils

Transcription

1 Materials Transactions, Vol. 46, No. 11 (2) pp. 249 to 2496 #2 The Japan Institute of Metals AE Source and Relation between AE Activity and Rate of Corrosion of Oil Tank Bottom Plate on Acidic Soils Sosoon Park 1; *, Shigeo Kitsukawa 2, Kenji Katoh 3, Sigenori Yuyama 4, Hiroaki Maruyama and Kazuyoshi Sekine 1 1 Division of Materials Science and Engineering, Faculty of Engineering, Yokohama National University, Yokohama 24-81, Japan 2 High Pressure Institute of Japan, Tokyo 11-2, Japan 3 Nippon Steel Corporation, Futtsu , Japan 4 Nippon Physical Acoustics Corporation Ltd., Tokyo 1-11, Japan Japan Oil Gas and Metals National Corporation, Tokyo 261-2, Japan In Service Inspection by Acoustic Emission technique offers the user significant advantages, such as the capability of monitoring corrosion damage of the bottom plates of an oil tank without opening it. In this study, the mechanism of AE generation due to the corrosion of tank bottoms on strong acidic sand was examined. During the corrosion process, the corrosion rate and the AE activity were estimated and the relation between these two factors was examined. The sources of AE due to the corrosion of tank bottoms on an acidic soil were specified. Here, the physical foundations of a global diagnosis technique based on the AE method for evaluating the corrosion damage to tank bottoms were presented. (Received June 1, 2; Accepted September 16, 2; Published November 1, 2) Keywords: oil storage tank, acidic sand, corrosion rate, acoustic emission activity, cavitation 1. Introduction Pitting is an important cause for oil leakage accidents in oil tanks. Therefore, cost-effective and in-service diagnosis techniques are increasingly needed for detecting and evaluating corrosion damage of such tanks. The acoustic emission (AE) method is considered to meet the requirements of an in-service inspection technique. In Europe and USA, this method has been adopted for evaluating corrosion damage to tank bottoms by empirical procedures based on field data, even though the relation between AE and the corrosion damage to tank bottoms in the field is not yet clear. 1 7) In this paper, corrosion of the tank bottoms in an acidic environment was studied. Acidic environment is encountered in some parts of tank bottoms during their life span. Two possible AE sources due to corrosion can be considered. One is cavitation in an acidic atmosphere. It has been reported that water droplets adhere to the undersurface of tank bottoms in the gap formed between the tank bottom and the soil in response to repeated expansion and shrinkage of metal. 8 11) If chlorine ions are present in the soil, the undersurface of the tank bottom becomes gradually more acidic and the corrosion is accelerated in a part of tank bottoms. 8 12) The other possible source is cracking of corrosion products. This paper is concerned only with the former one. The electrochemical and chemical reaction that occurs during the corrosion of iron under acidic conditions is shown by eqs. (1) and (2). In this case, hydrogen gas is generated by a cathodic reaction but no corrosion products are produced. Therefore, cavitation of hydrogen gas can occur and it can be considered as an AE source due to the corrosion. During this corrosion experiment, the corrosion rate and the AE activity were estimated. From the results, the *Graduate Student, Yokohama National University relation between AE activity and corrosion rate on acidic sand was examined. From the AE activity and the parameters of AE wave, the nature of the AE source during the corrosion on the acidic sand was discussed. Fe! Fe 2þ þ 2e (anodic reaction) ð1þ 2H þ þ 2e! H 2 " (cathodic reaction in acidic environment) ð2þ Fe þ 2H þ! Fe 2þ þ H 2 " (total reaction) 2. Experimental Procedures This study was carried out in three parts. The first part involved the measurement of AE activity and estimation of the AE source. This was followed by an investigation on the corrosion condition of the bottom plates, including the rate of corrosion as obtained from measuring the polarization resistance (Rp). In the third part, the relation between the AE activity and the corrosion rate was estimated from the results of the first two parts. Figure 1 shows the configuration of experimental setup for estimating the corrosion rate and measuring AE. Table 1 shows the corrosion conditions and the reference names for each experiment. For example, ph1fe3 is the reference name for an experiment that was conducted using an SS4 plate in contact with sand containing ph1 acidic solution (hereafter, acidic sand). The moisture content of the acidic sand was 3% of its mass (hereafter, 3 mass%). The moisture content was defined as the volume (or mass) percentage of acidic solution mixed into the dried sand. The corrosion conditions shown in Table 1 were decided on the basis of the acidity observation of corroded tank bottoms in the field. 13) To compare the AE characteristics at different corrosion rates, SS4 plate and zinc-galvanized SS4 plate were used. The corrosion area

2 Acoustic Emissions during Corrosion of Bottom Plate of Oil Tank on the Acidic Soils 2491 A SI 1287 for Electrochemical Interface B SI 126 for Impedance/Gain-Phase Analyzer C µ DispTM Digital AE System with 4 Channels Pre-amplifier (4dB) PC A B R. E. AE Sensor Specimen C. E. Acidic sand PC C Z-win & Z-Plot State of Sealing Up Room Temperature Wave Analysis by Signal Processing Fig. 1 Experimental set up for corrosion conductance and AE measurement. Table 1 Corrosion conditions and reference names. SS4 Zn galvanized SS4 Moisture contents (mass%) ph1 ph1fe2 ph1fe2 ph1fe3 ph1fe3 ph1zn2 ph1zn2 ph1zn3 ph1zn3 ph2 ph2fe3 ph2fe3 ph2zn3 ph2zn3 ph3 ph3fe3 ph3fe3 ph3zn3 ph3zn3 ph4 ph4fe3 ph4fe3 ph4zn3 ph4zn3 (hereafter, corrosion surface), that is, the undersurface of the specimen in contact with sand, was 1 cm 1 cm. The side area and the top surface of the specimen were covered with Teflon tape to protect from corrosion. Specimens were placed on the acidic sands and the corrosion environment was sealed to retain moisture. 2.1 Measurement of polarization conductance Polarization resistance during corrosion was measured by electrochemical impedance spectroscopy (EIS) with Z-Win and Z-Plot software (Solartron Co. Ltd.), Solartron Instruments (SI 1287 for electrochemical interface and SI 126 for impedance/gain-phase analyzer). To analyze the frequency response characteristics, a sinusoidal signal with a frequency of.1 Hz to 2, Hz within the voltage 1 mv was applied to the corrosion system (See Fig. 1). Polarization resistance (Rp) was calculated from the EIS (Nyquist plot) by frequency response analysis. Polarization conductance (Yp) and corrosion current density (i corr ) were given by the equation, Yp ¼ 1=Rp, i corr ¼ K Yp. K was the proportional constant determined by corrosion system. Therefore, Yp is the representative value used for estimating corrosion rate as well as i corr. The corrosion potential (E corr ) was measured with a saturated Ag/AgCl electrode as a reference electrode (R.E.) and Type 34 stainless steel was used as a counter electrode (C.E.). The corrosion state could be estimated from the values of Yp and E corr. 2.2 Measurement of weight loss and estimation of K To estimate the corrosion proportional constant K, the weight loss (W) and Yp were measured at every hour during the corrosion progress. From W, i corr ðwþ was calculated by Faraday s law of electrolysis. From the plotting of i corr ðwþ and Yp at the same corrosion time, the slope value of the approximation curve gives the value of K. In this experiment, the corrosion surface was cm cm. The side and the top surface of the specimen were covered with Teflon tape to protect from corrosion. Weight loss (W) at corrosion time (t 1 ) was defined as the weigh loss due to the corrosion from the start to t Measurement of AE A digital AE system, DiSPÔ with a 4-channel board (Physical Acoustics Corporation Ltd.) and two kinds of AE sensor (1 khz with a narrow band and 1 MHz with a wide band), was used for acquisition of AE waves. AE signal acquisition was conducted with a threshold of 3 db. AE sensors were set on the top surface of the specimen. After denoising the AE waves detected during corrosion, various values for AE activity were estimated. MHRt (Mean Hit Rate

3 2492 S. Park et al. Imaginary Part of Polarization Impedance, R / Ω ph1fe3 Rp =R 1 -Rs Rs R 1 Real Part of Polarization Impediance, R / Ω Corrosion Current Density from W, icorr / ma cm Fig. 2 Electrochemical impedance spectroscopy (Nyquist plot). Fig. 4 Weight loss and corrosion current density by W of ph1fe3. Rp Rs Cd Polarization Conductance, Yp / 1-4 S cm Fig. 3 Open-circuit of corrosion system during entire AE acquisition time), MHR1hr (MHR at every 1 h step during AE acquisition) and Acc Slope were estimated as representative AE activity values. The Acc Slope value was measured from the slope of the approximation line at every 1 h step of the accumulated AE hit curve. MHRt appeared to be suitable for long-term acquisition at low corrosion rates. MHR1hr appeared to be suitable for shortterm acquisition at high corrosion rates. Acc Slope could be estimated regardless of the interval length of the measurement time and the corrosion rates. The centroid frequency [khz] by the fast Fourier transform method and the RMS (AE Root Mean Square value of the effective voltage [V]) of significant AE waves were also analyzed. All significant AE waves were classified by the value of the centroid frequency and the RMS. 3. Results and Discussion 3.1 Estimation of corrosion behavior on the strong acidic soil As a typical result, EIS by Nyquist plots during the corrosion of ph1fe3 is shown in Fig. 2. Rp was calculated from the equation Rp ¼ðR 1 RsÞ, as shown in Fig. 2, and Yp was calculated from Rp. From Fig. 2, the open circuit of this corrosion system could be estimated as shown in Fig. 3. Rp is equivalent to the impedance due to the corrosion of a SS4 plate. Rs is equivalent to the impedance due to the acidic sand. An electrical double layer, which was constructed in the gap between the specimen and sand, works as Fig. Polarization conductance of ph1fe3. a condenser and its capacitance (Cd) causes the resistances to change depending on the frequency of the alternating current. From the result of the bode plot and phase analysis, the left part of the half-circle of the Rs was considered to be the noise originated in the deviation of phase and was not considered as important. From measurement of W, i corr ðwþ was calculated and is shown in Fig. 4. Figure shows Yp according to corrosion time. i corr ðwþ and Yp decreased with increasing corrosion time. From the slope value of the approximation curve of i corr ðwþ versus Yp (from 1 to 24 h), the corrosion proportional constant K was estimated. From the corrosion proportional constant K, i corr ðypþ was estimated. Figure 6 shows the generating rate of hydrogen gas, as calculated from i corr ðypþ or i corr ðwþ. From these results, by using Faraday s law and eqs. (1) and (2), the amount of hydrogen gas generated due to corrosion was estimated. The generating rate of hydrogen gas decreased with increasing corrosion time. From this result, it is assumed that the hydrogen gas generated during corrosion was related to the cavitation that emitted AE. E corr during corrosion is shown in Fig. 7. For ph1fe3, the corrosion rate, Yp and i corr decreased sharply from the start to. h. From. to 1 h, Yp and i corr increased very rapidly. Yp and i corr decreased again very rapidly from 1 to h. After h, Yp and i corr increased again very slowly. This tendency continued till 24 h. E corr decreased very quickly

4 Acoustic Emissions during Corrosion of Bottom Plate of Oil Tank on the Acidic Soils 2493 Generating Rate of Hydrogen Gas, log(mole / 1-1 mol cm -2 sec -1 ) From Icorr( W) From Icorr(Yp) Acidity, ph / ph Temp Temperature, T / C Reacting Time, t / h Fig. 6 Generating rate of hydrogen gas during corrosion, ph1fe3. Fig. 8 Acidity of sand according to time and without corrosion of iron. Corrosion Potential, Eorr vs.she / mv Fig. 7 Corrosion potential of ph1fe3. from the start to. h. At the 2nd h and the 4th h, E corr was at its peak. From 7 to 24 h, E corr continued to increase very slowly. These results mean that the corrosion system changed to a state of cathodic inhibition very rapidly for the first. h. From. to. h, the corrosion system changed to one of anodic inhibition. This means that the rate of corrosion was very high but it declined very rapidly for the first. h. Here, the change of the ph and temperature on the surface of sand was measured using digital ph meter (HM-2P, manufactured by DKK-TOA) and the result is shown in Fig. 8. The ph of acidic solution used in this additional experiment was 1 and the moisture content was 3 mass% of the sand. The increase of ph which was caused by the reaction between the hydrogen ions and the element materials composing sand was observed. After h, the ph on the surface of sand changed from 1 to 4.2. As a result, the corrosion system changed to the state of cathodic inhibition and the corrosion rate i corr decreased. 3.2 Analysis of AE activity due to corrosion on the various acidic soils AE activities (MHR1hr, MHRt, and Acc Slope), which were detected by 1 khz sensor, at each condition of corrosion shown in Table 1 were analyzed. Figure 9(a) shows the dependency of MHRt on the moisture content of sand. The results obtained with 1 MHZ sensor are not shown here, however they were similar to those obtained with 1 khz sensor. At a moisture content lower than 2 mass%, no significant AE was detected and at a moisture content of 3 mass%, the MHRt was very low. However, it increased considerably at a moisture content of 3 mass%. Figure 9(b) shows the dependency of MHRt on the ph of sand. At the same moisture content, higher MHRt were obtained at lower ph. Moreover, the MHRt of zinc galvanized SS4 was (a) 2 (b) 2 MHRt, hit rate / Hit h Zn Fe MHRt, hit rate / Hit h Zn Fe Moisture Contents, mass percent / mass% Moisture contents Acidity, ph / - ph Fig. 9 Dependency of hit rate on the moisture content and ph of acidic sands.

5 S. Park et al. (a) Acc Slope, hit rate /Hit. -1 h Acc Slope vs Yp MHR1hr vs Yp MHR1hr, hit rate /Hit h -1 (b) Acc Slope, hit rate /Hit h Acc Slope vs Yp Acc Slop = 6.474*Yp R 2 =.6817 MHR1hr vs Yp R 2 =.7161 MHR1hr = 7.773*Yp MHR1hr, hit rate /Hit h -1 Polarization Conductance, Yp/1-3 S.cm -2 Polarization Conductance, Yp / 1-3 S.cm -2 at the same time after 3hr passed Fig. 1 Relation between Yp and the AE activity measured 3 hours later from the measurement of Yp (ph1fe3). higher than that of ungalvanized SS4, although the dependency of MHRt on ph was similar for both the ungalvanized SS4 and zinc galvanized SS Relation between AE activity and corrosion rate Figure 1(a) shows the relation between the MHR1hr (before classification of AE waves into two patterns) and the Yp at the same corrosion time, which was derived from the result of the AE activity and corrosion rate of ph1fe3. There was no relation between the two factors and the relation between the Acc Slope and the Yp also showed no dependency. However considering the fact that cavitation is likely a main mechanism of the generation of AE in acidic environment and that it takes time for hydrogen gas to be released and form a bubble, there could have been some time lag between AE activity and the Yp. Hence, Yp was compared with the AE activity measured some time later from the measurement of Yp. Figure 1(b) shows that there was a positive relation between the Yp values of ph1fe3 and the MHR1hr measured 3 hours later. The relation between the Acc Slope and the Yp showed the same tendency. Because Yp represents corrosion rate and corrosion conditions were controlled to be similar to those ones in the field, it is clear that the corrosion rate of tank bottoms in the field can be estimated by AE activity parameters such as MHR1hr and Acc Slope. 3.4 Source of AE due to corrosion of tank bottoms From Figs. 9 and 1, it is clear that both corrosion rate and AE activity are higher at lower ph. However, under a condition of low moisture AE activity was extremely low, even if ph was low. This fact indicates that the chemical reaction of generating hydrogen gas was not detected as an AE signal because the difference among these corrosion experiments at the condition of ph1 acidic sand was only moisture content. When the moisture content was 3 mass%, acidic solution appeared up to the surface of the sand, but it did not appear at a moisture content of 3 mass%. From these observations, the following process is proposed: As a first step, when considering the surface of the acidic sand at the 3 mass%, the gap between the sand and specimen is filled with acidic solution. Then the hydrogen adhered to the corrosion surface of the specimen and accumulated to form a bubble according to the corrosion progress. In the second step, the bubble broke out between the acidic solution and the specimen. This phenomenon is called cavitation. Cavitation can be detected easily because of its energy and the amplitude are higher than those created simply by hydrogen generation. Under moisture conditions lower than 3 mass%, there was not enough solution as cavitation was caused by hydrogen gas. This may be the reason why the AE signal was not detected at a moisture condition lower than 3 mass%. Therefore, the origin of the AE signals detected in this experiment is likely to have been the cavitations of hydrogen gas. In the field, if the tank bottom had been stuck neatly on its foundation, the sand could not be expected to have a moisture content of 3 mass%. However, if the chemical conditions around the tank bottoms were to correspond with these mentioned in the introduction, the regional acidification of the undersurface of tank bottoms and regional immersion of it probably should also happen. 8 11) Figure 11 shows the Acc Hit (accumulated hit count of AE) and distributions of centroid frequency. AE waves detected from ph1fe3 and ph1ze3 were classified into two patterns by centroid frequency. To clarify the AE source of these two patterns, the MHR1hr of each one was analyzed and the result is shown in Fig. 12. The characteristics of pattern 1 were as follows. 1) The centroid frequencies were lower than those of pattern 2. 2) These AE waves were detected for the first 8 h and the centroid frequencies had similar values. 3) MHR1hr of ph1fe3 was high for the first h (in the case of ph1zn3, for the first 3 h) and afterwards it became very low. 4) As the ph after h was 4.2, the density of hydrogen ions on the surface was very low. The last two characteristics show that AE waves of pattern 1 presumably originated in the plastic deformation and movement of dislocation by the solution of hydrogen ions ) However, the AE source of pattern 1 is not yet clear and needs to be verified by future discussions and additional experiments. The characteristics of pattern 2 were as follows. 1) The centroid frequencies were higher than those of pattern 1. 2) AE waves of pattern 2 were detected from. to 12 h. 3) The

6 Acoustic Emissions during Corrosion of Bottom Plate of Oil Tank on the Acidic Soils 249 Acc Hit, Hit /count Line-Circle : Pattern 1 Dot-chanined Line-Circle : Pattern 2 Moisture Content 3wt%, 1kHz Centroid Frequency, Frequency /khz ph1fe3, Acc Hit ph1zn3, Acc Hit ph1fe3. Centroid Frequency Corrosion Time, t /h ph1zn3, Centroid Frequency. Fig. 11 Accumulated hit and centroid frequency distribution with corrosion time (ph1fe3 and ph1zn3). (a) (b) MHR1hr, hit rate /Hit h ph1fe3 before classification pattern 1 pattern 2 MHR1hr, hit rate /Hit h ph1zn3 before classification pattern 1 pattern Corrosion Time, t /h ph1fe Corrosion Time, t /h ph1zn3 Fig. 12 AE activity before and after classification, MHR1hr. MHR1hr of pattern 2 increased from. to 4 h and then decreased. It increased again after 6 h. Here, E corr decreased from 1.6 to 4 h. The amount of hydrogen gas attached to the undersurface of the specimen is explained by the decrease in E corr. After cavitation occurred, E corr increased again from 4 to h. The MHR1hr of pattern 2 decreased again during this time. This is assumed to be originated in a decrease in the amount of hydrogen gas causing cavitation. 4) Figures 4, and 6, show the fact that the amount of hydrogen gas generated due to corrosion was proportional to the corrosion rate, i corr ðwþ, and i corr ðypþ. Considering that the hydrogen gases take time to cause cavitation, these characteristics of pattern 2 can be specified as representing AE signals caused by cavitation. In accordance with these discussions, the AE waves detected during this experiment was due to the corrosion of tank bottoms and cavitation is considered to be a source of AE due to the corrosion in the field. From the fact that AE activity has a positive relation with the corrosion rate, it can be considered to be a factor related to the corrosion rate or corrosion damage of tank bottoms. The reasons for this conclusion and the discussion from Figs. 4 to 1 respectively are as follows: 1) The corrosion rate increased with decreasing of ph, as is generally known; 12) 2) The corrosion rate of zinc is higher than that of iron in acidic solutions, as is generally known; 12) 3) AE activity during this corrosion experiment showed similar behavior with corrosion rate; 4) Corrosion condition similar to those in this experiment could occur in bottom tanks. 8 13) 4. Conclusion The corrosion rate and the AE activity during the corrosion of metal plates sitting on acidic sands were measured and the relation between these two factors was estimated. And the sources of AE due to the corrosion of tank bottoms sited on the acidic sand were also examined. The results can be summarized as follows: (1) Active corrosion under low ph conditions can be detected by the AE method. (2) Cavitation by the accumulation of the hydrogen gas generated from cathodic reaction is one source of AE signals due to the corrosion of tank bottoms. (3) AE activities vary according to the acidity of sand. There is a positive relation between AE activity and corrosion rate. These results confirm the fact that corrosion damage can be detected by the AE method and the corrosion rate of tank

7 2496 S. Park et al. bottoms in the field can be estimated using this method. This indicates that the AE method could be used to develop a global diagnosis system for evaluating the corrosion damage of in-service tank bottoms in the field. This use of AE method has the possibility of reducing the cost of maintenance and repair of oil tank. REFERENCES 1) Y. P. Kim, M. Fregonese, H. Mazille, D. Féron and G. Santarini: NDT&E International 36 (23) ) A. Kamiya, K. Morofuji, K. Enuma, M. Yamada and S. Yuyama: J. High Pressure Institute of Japan 4 (22) ) S. Yuyama: Boshoku Gijutsu 3 (1986) ) G. Dai, W. Li, Y. Zhang and F. Long: Mater. Evaluation 6 (22) ) I. Morita, T. Arakawa, H. Hatanaka and M. Hagiwara: J. High Pressure Institute of Japan 4 (22) ) M. Yamada, S. Kitsukawa, S. Yuyama, A. Kamiya, K. Sekine and H. Maruyama: J. High Pressure Institute of Japan 4 (22) ) S. Yuyama, M. Yamada and K. Sekine: J. High Pressure Institute of Japan 4 (22) ) K. Katoh, S. Itoh, H. Ishimoto and T. Yashiki: Zairyo-to-Kankyo 1 (22) ) K. Katoh, K. Suzumura, H. Ishimoto and T. Yashiki: J. High Pressure Institute of Japan 41 (23) ) K. Katoh, H. Ishimoto and T. Yashiki: J. High Pressure Institute of Japan 41 (23) ) K. Katoh, C. Katoh, H. Ishimoto and T. Yashiki: J. High Pressure Institute of Japan 42 (24) ) The Japan Society of Corrosion Engineering: Corrosion Handbook, (Maruzen, Tokyo, 2) pp. III-4-1 III ) S. S. Park, S. Kitsukawa, K. Katoh, S. Yuyama, H. Maruyama and K. Sekine: The 17th International Acoustic Emission Symposium, (24) pp ) H. Hagi: The 8th International Acoustic Emission Symposium, (1991) pp ) H. Hagi and T. Hirose: The 9th International Acoustic Emission Symposium, (1993) pp ) H. Hagi: JSME International Journal 39A (1996) ) H. Hagi: JSME 6A (1994)