Exploration of corrosion defect detection of tank bottom plate based on electromagnetic acoustic emission

Abstract

This study investigates the feasibility of detecting corrosion-induced defects in tank bottom plates using electromagnetic acoustic emission (EMAE) technology based on high-intensity pulsed current excitation. Experiments employed intact and electrochemically corroded Q235 steel specimens simulating the bottom environment of storage tanks. Key findings demonstrate that EMAE with pulsed current excitation successfully induced distinct acoustic emission signals in corroded specimens, exhibiting significantly higher intensity compared to intact specimens. Crucially, corroded specimens displayed a pronounced Kaiser effect, characterized by rapid attenuation of signal intensity and energy within the characteristic frequency band of 46.875-62.5 kHz upon repeated loading at identical current levels, a phenomenon absent in intact specimens. Time-frequency analysis confirmed that this characteristic frequency band is uniquely present and exhibits Kaiser effect attenuation exclusively in signals from corroded specimens. Parametric analysis further revealed distinguishing waveform features associated with corrosion signals. These results validate that electromagnetic acoustic emission (EMAE) with pulsed current excitation can effectively generate and detect acoustic emissions linked to active corrosion defects. The identified Kaiser effect behavior and the specific 46.875–62.5 kHz frequency band serve as diagnostic indicators for corrosion detection. This study lays the groundwork for applying this EMAE methodology to corrosion detection in tank bottom plates.

Keywords

storage tank corrosion detection non-destructive testing electromagnetic acoustic emission wavelet packet analysis

1. Introduction

Electromagnetic acoustic emission (EMAE) is an emerging non-destructive testing technique. It enables dynamic inspection by applying localized electromagnetic loading to a metal component, thereby inducing internal acoustic emission phenomena. A key distinction from traditional acoustic emission (AE) lies in its loading mechanism: EMAE replaces conventional hydraulic or pneumatic loading with electromagnetic excitation. This localized approach offers significant advantages, including avoidance of potential structural damage from overall loading, concentration of energy on defect areas to enhance signal strength, and on-demand loading capability that significantly reduces the long-term loading time typically required by traditional AE.

Electromagnetic acoustic emission (EMAE) has been widely studied for detecting metal cracks. Early work by Finkel P. et al. in the U.S. induced acoustic emissions at cracks using localized dynamic electromagnetic loading, confirming the technique’s feasibility in thin plates.¹ In China, related research began around 2005. A group at Hebei University of Technology pioneered EMAE studies, achieving preliminary crack localization in aluminum plates and analyzing signals under varying excitation.^2–7 Subsequent work included studies on EMAE characteristics for different crack sizes, an improved transducer for ferromagnetic materials, and investigations into the crack response mechanism.^8,9 More recently, Chang, Y.Q. et al. proposed a rail crack detection method combining EMAE with the Optimal Local Mean Decomposition algorithm.¹⁰ The application of electromagnetic acoustic techniques is also expanding. Electromagnetic acoustic resonance (EMAR) has reliably measured thickness loss in acid-corroded steel plates, outperforming electrochemical methods.¹¹ Similarly, guided waves from electromagnetic acoustic transducers have shown promise for detecting corrosion in steel plates, supported by simulations and experiments.¹²

Beyond cracks, detecting other critical damage such as corrosion presents a distinct challenge for NDT. While methods like vibration analysis effectively monitor bearing defects^13,14 and advanced hybrid techniques (e.g., EMAT combined with eddy current testing) are being developed for various flaws,¹⁵ sensitivity to early-stage or smooth corrosion often remains limited.^16,17 In contrast, electromagnetic acoustic emission (EMAE) is well-established for crack detection, leveraging localized current concentration and Lorentz forces at sharp crack tips.^18–20 Its potential for corrosion, however, is less explored. Although corrosion lacks sharp features to focus current, its underlying electrochemical processes (e.g., metal dissolution) involve transient energy releases capable of generating acoustic emission (AE).²¹ Research on Kaiser and Felicity effects in composites and granular media underscores the value of AE memory effects in characterizing damage progression.^22,23 Moreover, correlated Kaiser effects observed in concurrent AE and electromagnetic radiation monitoring reveal a fundamental link between stress memory and electromagnetic-acoustic phenomena.²⁴ This suggests that the active excitation and response analysis intrinsic to EMAE could be adapted to probe corrosion activity.

To address this gap, this study employs a comparative experiment with intact and electrochemically corroded Q235 steel plates simulating storage tank conditions. This work introduces a novel strategy for detecting active corrosion by combining high-intensity pulsed current excitation with advanced time-frequency analyses like short-time Fourier transform and wavelet packet decomposition. Unlike conventional methods that focus on geometric flaws, our approach investigates the dynamic response of corrosion activity itself. We aim to develop new diagnostic criteria derived from unique acoustic emission signatures, specifically characteristic frequency bands and the Kaiser effect, generated during electromagnetic stimulation. This methodology offers a new means for active corrosion identification and establishes a foundation for a proactive non-destructive testing protocol for tank bottom plates.

2. Method

2.1. Experimental specimen preparation

The experimental material is Q235 steel, processed into a rectangular specimen with the size of 150 ×150 ×15 mm, the blank control group is set up as plate 1, and the experimental group is set up as plate 2. Plate 1 is not subjected to electrochemical corrosion treatment, only PVC pipe is bonded in the center of the steel plate, and 100 mL of deionized water is added to the PVC pipe to simulate the liquid filling situation of the storage tank. A circular area with a radius of 25 mm at the center of the Plate 2 specimen was subjected to constant-current electrochemical corrosion treatment to simulate uniform corrosion on the tank bottom. The specimen served as the anode and was connected to the positive terminal of the DC power supply. A graphite rod acted as the cathode, connected to the negative terminal. Both the specimen and the graphite rod were immersed in a 0.1 mol/L NaCl electrolyte solution contained within a PVC tube. A constant current of 0.001 A was applied for 2 hours to induce galvanic corrosion. A schematic of this electrochemical corrosion setup is shown in Figure 1, while schematic diagrams of the setups for the plate 1 and plate 2 are provided in Figure 2. The electromagnetic acoustic emission experimental system is illustrated schematically in Figure 3.

Figure 1.

Experimental setup for electrochemical corrosion.

Figure 2.

Schematic diagram of sample preparation device: (a) Diagram of the experimental setup for plate 1 and (b) Diagram of the experimental setup for plate 2.

Figure 3.

Schematic diagram of electromagnetic acoustic emission experimental system.

2.2. Experimental system setup

The electromagnetic acoustic emission test system contains a pulse current generator, a metal plate specimen, an acoustic emission signal sensor, a preamplifier and a PC. The short-time impulse current generated by the current generator is passed into the metal specimen by the magnetic suction electrode, and the acoustic emission signal generated by the specimen is collected and transmitted to the PC through the acoustic emission sensor and preamplifier,²⁵ and the acoustic emission signal collected by the PC is subjected to the analysis of waveform characteristics and time-frequency feature extraction, so as to obtain the characteristic information of the acoustic emission signal of the corrosion defects finally. The arrangement of the sensor and electrode is shown in Figure 2. The peak current of the pulse current generator used in the test can reach up to 3500 A, and the acoustic emission sensor is the WSα(The Alpha Series Wide-Frequency Sensor)of PAC Company, with an operating frequency range of 1∼1000 kHz. The acoustic emission acquisition system is the 8-channel Express-8 of PAC Company, and the acquisition parameters are shown in Table 1.

Table 1.

Acoustic emission acquisition parameter settings.

Acquisition threshold	Integrated amplification	Filter bandwidth	Sampling frequency	Impact resolution time
38 dB	40 dB	20-1000 kHz	1 MHz	200 μs

The electromagnetic loading strategy adopted in the experiment is to repeat the current of 1.00 kA, 1.24 kA, 1.44 kA, 1.78 kA and 2.56 kA to the specimens of plate 1 and plate 2 in sequence, and the number of loading times of a single group is four times, aiming at exploring the effects of different current sizes and repeated loading on the electromagnetic acoustic emission (EMAE) signal intensity. During the experimental process, the full waveform was captured by the acquisition system, and the acquired acoustic emission signals were analyzed in terms of waveform characteristics and time-frequency joint analysis, so as to extract the electromagnetic acoustic emission signal characteristics related to corrosion defects.

3. Experimental results and analysis

3.1. Feasibility verification of EMAE detection for corrosion defects based on positioning accuracy

To fundamentally verify that the electromagnetic acoustic emission (EMAE) signals originate from the preset corrosion defect and to evaluate the detection capability of this method under conditions close to actual service, an independent localization verification experiment was first conducted on a large-scale Q235 steel plate (450 mm × 500 mm × 5 mm). As shown in Figure 4, four acoustic emission sensors were arranged in a diamond-shaped array. A circular area with a radius of 25 mm at the center of the test plate was subjected to electrochemical corrosion treatment (parameters consistent with those described in Section 2.1) to simulate early-stage uniform corrosion. Its macroscopic morphology is shown in Figure 5, indicating a relatively mild corrosion degree without the formation of significant corrosion pits. To systematically explore the detection threshold, short-duration high-current pulses of 800 A, 1.34 kA, 1.64 kA, 2 kA, 2.16 kA, 2.32 kA, and 2.48 kA were applied four times each.

Figure 4.

Positioning experimental device diagram.

Figure 5.

Corrosion damage condition.

The results show that when the pulsed current was below 2 kA, the system could only capture sporadic hit signals but failed to form valid localized events. However, when the current reached or exceeded 2 kA, clear acoustic emission events could be consistently triggered and successfully located. The two-dimensional localization results, as illustrated in Figure 6, demonstrate that the event sources were highly concentrated within and along the edge of the preset corrosion zone. Quantitative analysis further revealed (as shown in Table 2) that for the vast majority of events, the distance between the located coordinates and the corrosion center was less than or equal to the radius of the corroded area (25 mm), indicating minimal localization error. This provides conclusive evidence for the spatial correlation between the EMAE signals and the corrosion defect. Furthermore, a significant attenuation in the amplitude of the acoustic emission signals was observed during the repeated loading experiments. This is a typical manifestation of the Kaiser effect, confirming from the perspective of the irreversibility of energy release that the signals originated from genuine microstructural changes within the material, rather than from external electromagnetic interference.

Figure 6.

Two-dimensional plane localization results of acoustic emission events.

Table 2.

Experimental data for locating corrosion defects.

Event	Load current (kA)	Event coordinates (x, y)	Source amplitude (dB)	R_x,y	Positioning error (m)
1	2.00	(0.1989,0.2404)	99	0.0278	0.0018
2	2.00	(0.1903,0.2436)	99	0.0352	0.0103
3	2.00	(0.1903,0.2436)	93	0.0352	0.0103
4	2.00	(0.1989,0.2404)	83	0.0278	0.0018
5	2.16	(0.2009,0.2434)	91	0.0250	0
6	2.16	(0.2009,0.2434)	82	0.0250	0
7	2.16	(0.2009,0.2434)	82	0.0250	0
8	2.16	(0.2009,0.2434)	85	0.0250	0
9	2.32	(0.2009,0.2434)	99	0.0250	0
10	2.32	(0.2009,0.2434)	83	0.0250	0
11	2.32	(0.2009,0.2434)	83	0.0250	0
12	2.32	(0.2009,0.2434)	82	0.0250	0
13	2.48	(0.2009,0.2434)	95	0.0250	0
14	2.48	(0.2009,0.2434)	88	0.0250	0
15	2.48	(0.2009,0.2434)	86	0.0250	0
16	2.48	(0.2009,0.2434)	88	0.0250	0

At the determined detection threshold (≥2 kA), EMAE technology can effectively identify and accurately localize early-stage uniform corrosion defects. It is noteworthy that even though no localized events were formed below the 2 kA threshold, the presence of hit signals itself suggests that the corrosion defect had already initiated activity under lower energy excitation.

3.2. Effect of current magnitude and repeated loading on acoustic emission signal amplitude

The magnitude of the pulsed high current directly determines the size of the current density at the defect.²⁶ Kaiser effect is a typical feature of acoustic emission signals, that is, when the same load is repeatedly loaded, the acoustic emission signal energy will decay, or even no longer occurs, based on this feature can be collected whether the signal is acoustic emission signals to discriminate. The electromagnetic acoustic emission signals obtained by repeated loading of plate 1 and plate 2 with different sizes of currents, and their signal amplitudes are shown in Figure 7.

Figure 7.

Effect of current magnitude and number of repetitions on signal amplitude: (a) Signal amplitude vs excitation current for Plate 1 and (b) Signal amplitude vs excitation current for Plate 2.

As can be seen from Figure 7, the current strength and signal amplitude are approximately linear, the signal amplitude increases with the increase of current strength, and the signal amplitude in plate 2 is significantly larger than that in plate 1. Repeated loading has almost no effect on the signal amplitude of plate 1, while the signal amplitude of plate 2 shows a significant attenuation during repeated loading, which is especially dramatic during the 2nd loading, in line with the Kaiser effect. This phenomenon reflects the nature of the Kaiser effect, i.e., the energy of the acoustic emission signal will show a tendency of attenuation and even eventually disappear under the repeated action of the same load.

Therefore, the amplitude of the electromagnetic acoustic emission signal in plate 1 is determined by the current magnitude, while the electromagnetic acoustic emission signal in plate 2 is generated by the active defects, and the corrosion defects with the tendency to expand produce acoustic emission phenomena consistent with the Kaiser effect under the excitation of the short-time pulsed current.

3.3. Characterization parameter correlation analysis

The generation of acoustic emission signals is closely related to the type and state of the acoustic emission source, so the acoustic emission signals contain information that can characterize the type of the source and the degree of damage. Acoustic emission signal characteristic parameter correlation analysis method is the most commonly used acoustic emission signal analysis method, which is the statistical relationship between any two signal parameters, this statistical relationship can be identified on the type of acoustic emission source, to a certain extent, to eliminate the interference of noise signals.^27–29 The most commonly used characteristic parameters are magnitude, Rising time, span, ringing counts, absolute energy, etc. The schematic diagram of the commonly used parameters is shown in Figure 8, and the meanings and uses are shown in Table 3.

Figure 8.

Schematic diagram of acoustic emission signal characteristic parameters.

Table 3.

Meanings and uses of acoustic emission characterization parameters.

Serial number	Parameters	Meaning	Usage
1	Magnitude	The maximum magnitude value of the signal waveform, 0 dB=1 μV.	Relates to the size of the event and characterizes the intensity of the corrosion tendency or crack propagation activity
2	Ringer Count	Number of signal oscillations crossing the threshold value.	Roughly reflects the intensity and frequency of acoustic signals and is widely used to evaluate acoustic emission activities.
3	Absolute Energy	the area under the signal detection envelope, 1 aJ=10^-18 J.	True reflection of acoustic emission impact signal energy.
4	Rising time	The time taken for the signal to first cross the threshold of the gating line to the maximum magnitude, unit is μs.	Rising time can be affected by signal propagation and is used for noise discrimination.
5	Span	The time interval that elapses between the first time the signal crosses the threshold and the final drop to the threshold, unit is μs.	Similar to ringer counting, common for noise identification and special wave source identification.
6	Signal amplitude	Another form of unit of measure for acoustic emission impact signal energy, 1picovolt-sec=10-12 voc-sec	Reflects the intensity of acoustic emission activity and is often used to comprehensively evaluate the strength of acoustic emission activity.

The correlation between the ringing count, rising time and span of the electromagnetic acoustic emission signals of the plate 1 and plate 2, respectively, and the magnitude of the signals is shown in Figure 9. From the magnitude correlation graph, it can be seen that the plate 1 board ringing count, span and magnitude show a good positive correlation, the plate 2 board of electromagnetic acoustic emission signal has a higher magnitude, ringing counts also rise significantly. The ratio of the Rising time to the magnitude characterizes the rise slope of the signal. The smaller slope in Plate 2 indicates that the electromagnetic acoustic emission signals generated by active corrosion defects exhibit a steeper waveform front and faster rise.

Figure 9.

Correlation of rising time, span, ringer counts and magnitude: (a) rising time vs magnitude, (b) span vs magnitude and (c) ringer counts vs magnitude.

Another typical characteristic parameter of acoustic emission signals is span, the correlation between ringer counts, absolute energy and span. As shown in Figure 10, the ringer counts and span of plate 1 show a significant linear correlation, and the absolute energy has a specific functional relationship with the span. In addition, the signal generated in plate 2 significantly outperforms plate 1 in terms of both ring count and span. Correspondingly, the absolute energy of the signals in plate 2 also reached a higher level.

Figure 10.

Correlation of span with absolute energy and ringing counts: (a) ringing counts vs span and (b) absolute energy vs span.

In summary, the electromagnetic acoustic emission signals of plate 1 and plate 2 have obvious differences in all kinds of typical parameter correlation analysis diagrams, which is the nature of the acoustic emission source is different, these signal characteristics can be used for the subsequent electromagnetic acoustic emission technology based on the corrosion defects in the bottom plate of the storage tanks corrosion defects detection and identification of the basis for discrimination.

3.4. Time-frequency analysis of signal based on short-time Fourier transform (STFT) and wavelet packet analysis

The waveform analysis is based on the time-domain waveform of the acoustic emission signal collected by the acoustic emission instrument, and the acoustic emission source information is obtained by time-frequency analysis of the transient waveform.^30,31 Since the characteristic parameters only extract some local features of the signal waveform, the time-domain waveform contains richer acoustic emission source information than the characteristic parameters.

Taking the analysis results of the electromagnetic acoustic emission signal under 1000 A current excitation as a demonstration, the signal spectrum after STFT transformation and the time-frequency relationship after STFT-CS processing are shown in Figures 11 and 12. As can be seen from the figure, the characteristic frequency range in plate 1 is narrower, mainly concentrated below 35 kHz, and the peak frequency changes with time, mainly appearing after 200 μs, while the signal spectrum in plate 2 is wider, with new peaks appearing at 40∼50 kHz and throughout.

Figure 11.

STFT transformed spectrum of electromagnetic acoustic emission signal: (a) Plate 1 frequency-amplitude curve and (b) Plate 2 frequency-amplitude curve.

Figure 12.

Time-frequency relationship spectrum after STFT-CS processing: (a) Plate 1 time-frequency diagram and (b) Plate 2 time-frequency diagram.

On the basis of the STFT analysis results, the frequency band layers are divided based on the wavelet packet decomposition principle. The selection of the wavelet packet decomposition level involves a critical trade-off between frequency resolution and computational complexity. In this study, the sampling frequency $f_{s}$ was 1 MHz, setting the effective analysis frequency range up to 500 kHz according to the Nyquist theorem. The bandwidth $Δ f$ of each sub-band after L-level decomposition is given by:

Δ f = \frac{f_{s} / 2}{2^{L}} = \frac{500 kHz}{2^{L}}

(1)

Preliminary STFT analysis (Figures 11 and 12) indicated that the signal information was predominantly concentrated within the 0-150 kHz range. To achieve sufficient frequency resolution for identifying subtle features within this range while avoiding excessive subdivision that could lead to computational inefficiency and potential signal fragmentation, a decomposition level of L=5 was selected. This choice results in a sub-band bandwidth of

Δ f

=15.625 kHz, effectively partitioning the 0-150 kHz range into 10 sub-bands (Table 4), which optimally balances discriminative capability and algorithmic efficiency. The subsequent analysis confirmed that this parameter clearly captures the unique energy response of the corroded specimen within the 46.875-62.5 kHz band (Figures 10 and 11).

Table 4.

Spectral range of wavelet packet component reconstructed signals.

Frequency band	Wavelet packet node component	Frequency range (kHz)
1	(5, 1)	0∼15.625
2	(5, 2)	15.625∼31.25
3	(5, 4)	31.25∼46.875
4	(5, 3)	46.875∼62.5
5	(5, 7)	62.5∼78.125
6	(5. 8)	78.125∼93.75
7	(5, 6)	93.75∼109.375
8	(5, 5)	109.275∼125
9	(5, 13)	125∼140.625
10	(5, 14)	140.625∼156.25

For the STFT analysis, a Hanning window with a length of 256 points (corresponding to 256 μs) and an overlap ratio of 50% was employed. This parameter set provides a frequency resolution of approximately 3.9 kHz, adequate for resolving spectral features in the tens of kHz range, while maintaining satisfactory continuity in the time dimension.

Figure 13 shows the energy distribution of the electromagnetic acoustic emission signals of the specimens before and after corrosion under 1000 A current loading in various frequency bands after wavelet packet analysis and processing, and the vast majority of the energy in the signals of plate 1 and plate 2 is concentrated in node (5, 1), node (5, 2) and node (5, 4), i.e., within the frequency band from 0 to 46.875 kHz. In contrast, a new component of the signal energy of the specimen signal of plate 2 appears at node (5, 3) from 46.875 to 62.5 kHz, which is a feature significantly different from that of plate 1 and consistent with the spectral characteristics of the waveform in Figure 11.

Figure 13.

Wavelet packet decomposition node energy spectrum coefficient entropy.

Electromagnetic acoustic emission (EMAE) signals were acquired from Plate 1 (intact) and Plate 2 (corroded) under four repeated loadings at a current of 1000 A, followed by wavelet packet decomposition and reconstruction. The energy proportions of each node are presented in Figure 14. As shown in the figure, the frequency-band energy distribution of Plate 1 remained nearly unaffected by repeated loading. In contrast, for Plate 2, the normalized energy proportion at node (5,3) (46.875–62.5 kHz) reached 6.77% during the first loading, which attenuated sharply after the second loading. The energy shares of the remaining nodes were almost unchanged under repeated loading.

Figure 14.

Repeated loading effect on electromagnetic acoustic emission signal band energy: (a) Plate 1 wave packet - normalized energy and (b) plate 2 wave packet - normalized energy.

A repeated experimental analysis of the signal energy distribution characteristics at wavelet packet node (5,3) was conducted under different peak current conditions (1.24 kA, 1.44 kA, 1.78 kA, and 2.56 kA). The results are summarized in Table 5. Further analysis of the data leads to the following conclusions:

(1) The signal energy at node (5,3) of the corroded plate (Plate 2) decreased rapidly after repeated loading under the same current level, demonstrating a typical acoustic emission Kaiser effect. Statistical analysis shows that, across all tested current levels, the average energy of Plate 2 during the first loading cycle (2.02%, 2.27%, 2.62%, and 1.83% for currents ranging from 1.24 kA to 2.56 kA, respectively) was significantly higher than that of the intact plate (Plate 1, with averages all below 0.07%). Moreover, starting from the second loading cycle, the average energy of Plate 2 consistently dropped substantially. Therefore, it can be concluded that the characteristic frequency band capable of characterizing uniform corrosion damage lies within 46.875–62.5 kHz.

(2) The energy proportion in this characteristic frequency band is not determined by the magnitude of the current but is significantly influenced by the loading history, with the highest proportion occurring during the first loading. Although the absolute energy values varied with increasing current, the relative pattern indicative of the Kaiser effect—namely, “significantly higher energy during the first loading compared to subsequent loadings”—was consistently observed under all current conditions. Furthermore, the standard deviations of each data set fell within a reasonable range, indicating that this phenomenon is stably reproducible.

Table 5.

Energy spectrum coefficients of (5, 3) wavelet packet under different experimental conditions.

Load condition (kA)	Specimen type	Loading cycle	Energy at node (5, 3) (%)	Mean ± SD (%)
1.24	Plate 1	1st	0.09	0.06 ± 0.03
		2nd	0.03
		3rd	0.08
		4th	0.04
	Plate 2	1st	4.21	2.02 ± 1.42
		2nd	1.42
		3rd	1.24
		4th	1.22
1.44	Plate 1	1st	0.05	0.04 ± 0.01
		2nd	0.05
		3rd	0.03
		4th	0.04
	Plate 2	1st	5.21	2.27 ± 1.84
		2nd	1.32
		3rd	1.34
		4th	1.22
1.78	Plate 1	1st	0.08	0.07 ± 0.02
		2nd	0.05
		3rd	0.07
		4th	0.09
	Plate 2	1st	5.05	2.62 ± 1.81
		2nd	3.17
		3rd	1.29
		4th	0.95
2.56	Plate 1	1st	0.05	0.05 ± 0.02
		2nd	0.02
		3rd	0.05
		4th	0.08
	Plate 2	1st	4.94	1.83 ± 1.93
		2nd	0.96
		3rd	0.70
		4th	0.72

3.5. Discussion on the source mechanisms of corrosion-induced EMAE signals

The EMAE response of the corroded specimen is fundamentally distinct from that of the intact specimen, primarily characterized by the emergence of acoustic emission (AE) activity within a characteristic frequency band (46.875–62.5 kHz) and the exhibition of a pronounced Kaiser effect. The spatial correlation of the AE event sources with the preset corrosion zone, as definitively shown by the localization results in Figure 6, confirms that these signals originate from the corrosion damage itself. The irreversible nature of the underlying mechanisms is conclusively demonstrated by the rapid attenuation of both signal amplitude and the specific energy proportion at node (5, 3) upon repeated loading (Figure 14; Table 5), effectively ruling out reversible electromagnetic noise.

The consistent concentration of the Kaiser effect within the 46.875–62.5 kHz band (Figure 14) points to a unique source mechanism associated with the corroded microstructure. The interaction of the high-intensity pulsed current with the compromised material is hypothesized to generate AE signals through the following mechanisms:

(1) Fracture of Corrosion Product Layers. The generation of AE signals is closely associated with the formation and rupture of corrosion products, as established in studies on corrosion rate correlation for mild steel in acidic environments.³² Specifically for Q235 steel, the brittle oxide films and corrosion products are susceptible to micro-fracture under the transient thermal stress or Lorentz forces induced by the pulsed current. This micro-fracture process, which involves the release of localized mechanical energy and can include the disintegration of the corrosion product layer, is recognized as a fundamental source of acoustic emission in corroding metals.³³ Being inherently irreversible, this process directly accounts for the observed Kaiser effect, as the damage caused during the initial loading event is not replicated upon reloading.

(2) Microplastic Deformation at Corrosion Sites. As demonstrated by prior research, the metal substrate beneath corrosion pits or within corroded areas becomes mechanically weakened due to anodic dissolution, creating sites of stress concentration that are prone to localized plastic deformation under external loading.³⁴ In the context of this study, the base metal beneath corrosion products is therefore susceptible to microplastic deformation induced by the localized electromagnetic forces of the pulsed current at these microscopic imperfections. The Kaiser effect manifests because the initial loading causes yielding and irreversible deformation, whereas subsequent loads below the new, higher yield stress of the pre-deformed material produce diminished acoustic activity.

(3) Release of Localized Residual Stresses. The corroded surface inherently contains microscopic irregularities that act as potent stress concentrators. The significance of stress in corrosion processes is well-established, residual stress is a critical factor in classic models of stress corrosion cracking, alongside susceptible material and a corrosive environment.³⁵ Corrosion itself, particularly when non-uniform, introduces a complex field of residual stresses in the near-surface region. Research on X80 steel under elastic and yield stresses in a microbiologically influenced corrosion environment further demonstrates that applied stress significantly alters corrosion morphology and rate, a phenomenon linked to stress-modulated electrochemical activity at the interface and its interplay with residual stresses.³⁶ The pulsed electromagnetic loading in this study can trigger the release of these localized residual stresses, generating an AE signal upon the first load. Crucially, the Kaiser effect is itself a classical method in rock and material mechanics for estimating internal stress levels, as its physical basis is the release of stored strain energy from internal defects (e.g., microcracks, dislocation pile-ups) when a material is stressed beyond its previous maximum.³⁷ Therefore, the observed subsequent signal attenuation aligns with a partially relaxed stress state following the initial energy release, directly linking the detected Kaiser effect to the relief of residual stresses around active corrosion sites.

In summary, the distinct time-frequency signature and the Kaiser effect provide a diagnostic basis for identifying active corrosion through EMAE. The evidence strongly suggests that the signals originate from irreversible micro-fracture and micro-deformation processes at corrosion sites.

4. Conclusion

Based on a comparative experimental study using intact and electrochemically corroded Q235 steel plates, this work leads to the following main conclusions.

(1) Under the experimental conditions of this study, the electromagnetic acoustic emission signals generated by uniform corrosion defects exhibit characteristic waveforms including a steep front, fast rise time, long duration, and high ringing count. Parametric correlation analysis demonstrates that these features effectively differentiate the signals from corrosion defects from those of intact specimens. This provides a basis for screening and identifying characteristic corrosion signals within a complex background.

(2) The results indicate the feasibility of using the high-intensity pulsed current excited electromagnetic acoustic emission technology for detecting corrosion defects. This study successfully distinguishes the characteristic signals produced by corrosion, which primarily occupy a frequency band of 40 to 70 kHz, from the low-frequency electromagnetic noise inherently generated during the detection process, which is mostly below 30 kHz. Effective identification of acoustic emission signals from defects was achieved by analyzing the distribution of signal energy within this characteristic frequency band.

(3) The clear Kaiser effect observed in corroded specimens, characterized by the rapid attenuation of signal energy in the characteristic frequency band (46.875–62.5 kHz) upon repeated loading, confirms that the EMAE signals originate from irreversible microstructural changes associated with active corrosion defects. These changes are likely attributed to mechanisms such as the rupture of corrosion products, microplastic deformation, or localized stress release, which are fundamentally different from the mechanisms governing crack-related acoustic emission.

It is important to note the limitations of this study. The aforementioned findings and diagnostic criteria, such as the characteristic frequency band of 46.875 to 62.5 kHz and its associated Kaiser effect, are derived from a single type of uniform corrosion model prepared in a laboratory environment using a limited number of samples. The generalizability and engineering applicability of these results, particularly for localized corrosion morphologies like pitting corrosion which is more common in actual tank bottom plates, and under complex service conditions involving coatings, deposits, or coupled stresses, require further validation through more extensive and systematic experimentation. Therefore, the present method is more accurately described as providing a promising new approach for identifying active uniform corrosion defects under controlled conditions, thereby laying a necessary theoretical foundation for subsequent applied research aimed at practical engineering applications.

Footnotes

ORCID iD

Shibin Jiang

Ethical considerations

This article does not contain any studies with human or animal participants.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the Scientific and Technological Project of PipeChina (Project No. 80915844).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The authors confirm that the data supporting the findings of this study are available from the corresponding author upon reasonable request.*

Disclaimer/publisher’s note

The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of SAGE and/or the editor(s). SAGE and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

References

Chen

Yang

. Research on electromagnetic acoustic emission technology. Popular Science and Technology 2006; 15(06): 23–24.

Tan

Dong

. Application of acoustic emission technology in corrosion detection of large storage tanks. Petrochemical Corrosion and Protection 2007; 24(3): 45–49.

Jiang

Han

, et al. Research status of on-line acoustic emission detection technology for storage tank bottom plates. Corrosion and Protection 2016; 37(5): 375–380.

Finkel

Godinez

. Electromagnetic simulation of the ultrasonic signal for nondestructive detection of ferromagnetic inclusions and flaws. IEEE Transactions on Magnetics 2004; 40(4): 2179–2181. https://doi.org/10.1109/tmag.2004.829313

Zhang

Liu

Yang

, et al. Electromagnetic acoustic emission signal processing based on FFT and wavelet packet transform. Journal of Electrotechnology 2010; 25(4): 24–28.

Zhang

. Electromagnetic Acoustic Emission Signal Detection and Processing of Cracks in Sheet Metal. Doctoral dissertation, Hebei University of Technology, 2010.

Liu

Yang

Jin

, et al. Research on the application of electromagnetic acoustic emission technology in nondestructive testing. Journal of Electrotechnology 2009; 24(1): 23–27.

Jin

Kou

Guo

, et al. Electromagnetic acoustic emission detection of ferromagnetic materials based on electromagnetic ultrasonic transducer method. Journal of Electrotechnology 2017; 32(18): 98–105.

Liu

Zhai

Zhang

, et al. Study of Wavelet Packet in the Electromagnetic Acoustic Emission Signal Processing. In: Proceeding of the 11th International Conference on Electrical Machines and Systems, Wuhan, 2008.

10.

Chang

Zhang

Shen

, et al. Rail crack detection using optimal local mean decomposition and cepstral information coefficient based on electromagnetic acoustic emission technology. IEEE Transactions on Instrumentation and Measurement 2024; 73: 1–12.

11.

Diguet

Miyauchi

Takeda

, et al. EMAR monitoring system applied to the thickness reduction of carbon steel in a corrosive environment. Materials and Corrosion 2022; 73(5): 658–668. https://doi.org/10.1002/maco.202112915

12.

Fang

, et al. Electromagnetic ultrasonic shear-horizontal wave to detect corrosion defect of flat steel for grounding device of transmission pole tower. Applied Sciences 2023; 13(12): 6947. https://doi.org/10.3390/app13126947

13.

Nabhan

. Vibration analysis of adding contaminants particles and carbon nanotubes to lithium grease of ball bearing. Vibroengineering PROCEDIA 2016; 8: 28–32.

14.

Nabhan

Nouby

Sami

, et al. Multiple Defects Detection in Outer Race of Gearbox Ball Bearing Using Time Domain Statistical Parameters. International Journal of Vehicle Structures & Systems (IJVSS) 2016; 8(3): 167–174.

15.

Song

Xiao

Wang

, et al. A composite approach of electromagnetic acoustic transducer and eddy current for inner and outer corrosion defects detection. IEEE Transactions on Instrumentation and Measurement 2023; 72: 1–11. https://doi.org/10.1109/tim.2023.3238689

16.

Zima

Roch

. Nondestructive global corrosion measurement using guided wavefield data. Measurement 2024; 235: 114949. https://doi.org/10.1016/j.measurement.2024.114949

17.

Lyu

Zhou

Ding

, et al. Application research of ultrasonic-guided wave technology in pipeline corrosion defect detection: A review. Coatings 2024; 14(3): 358. https://doi.org/10.3390/coatings14030358

18.

Zhang

Liu

Yang

, et al. Experimental Study on High-Current Pulse Generator for Electromagnetically Induced Acoustic Emission. In: Proceeding of the 11th International Conference on Electrical Machines and Systems, Wuhan, 2008.

19.

Chen

. Research on electromagnetic acoustic emission technology. Master’s thesis, Hebei University of Technology, 2006.

20.

Meng

. Research on acoustic response mechanism and modeling of cracked metal plate under electromagnetic loading. East China Jiaotong University, 2022.

21.

Wang

, et al. Uniform corrosion characterization of tank floor steel based on acoustic emission and electrochemistry. Corrosion Science and Protection Technology 2017; 29(04): 381–386.

22.

Jung

Lee

. Enhancing Reliability and Safety in Industrial Applications: Assessing the Applicability of Energy b-Value to Composites. Materials 2024; 17(2): 447. https://doi.org/10.3390/ma17020447

23.

Canel

Jia

Campillo

, et al. Acoustic monitoring of compaction in cohesive granular materials. Physical Review E 2024; 109(2): 024902. https://doi.org/10.1103/PhysRevE.109.024902

24.

Zhang

Chen

Mao

. Damage evolution, acoustic emission and electromagnetic radiation of rock under uniaxial cyclic loading. Journal of Applied Geophysics 2024; 228: 105470. https://doi.org/10.1016/j.jappgeo.2024.105470

25.

Christian

Masayasu

. Acoustic emission testing. Springer-Verlag, 2008, pp. 41–48.

26.

Zhang

Liu

Jin

, et al. Electromagnetic acoustic emission testing based on high-current direct excitation. Transactions of China Electrotechnical Society 2013; 28(01): 101–105. https://doi.org/10.19595/j.cnki.1000-6753.tces.2013.01.015

27.

Dunegan

Harris

Tatro

Fracture analysis by use of acoustic emission. Engineering Fracture Mechanics 1968; 1(1): 111–122. https://doi.org/10.1016/0013-7944(68)90018-0

28.

Riahi

Shamekh

Khosrowzadeh

. Differentiation of leakage and corrosion signals in acoustic emission testing of aboveground storage tank floors with artificial neural networks. Russian Journal of Nondestructive Testing 2008; 44(6): 436–441. https://doi.org/10.1134/s1061830908060107

29.

Yang

Geng

. Acoustic emission detection. Machinery Industry Press, 2005.

30.

Aggelis

Matikas

. Effect of plate wave dispersion on the acoustic emission parameters in metals. Computers and Structures 2012; 98-99: 17–22. https://doi.org/10.1016/j.compstruc.2012.01.014

31.

Yang

. Research on Finite Element Analysis and Signal Processing of Electromagnetically Induced Acoustic Emission. Chongqing Three Gorges University, 2023. https://doi.org/10.27883/d.cnki.gcqsx.2023.000053

32.

Kasai

Utatsu

Park

, et al. Correlation between corrosion rate and AE signal in an acidic environment for mild Steel. Corrosion Science 2009; 51(8): 1679–1684. https://doi.org/10.1016/j.corsci.2009.04.021

33.

Qiu

Bai

, et al. Quantitative characterization of Q235B steel electrochemical corrosion by acoustic Emission. International Journal of Pressure Vessels and Piping 2022; 199: 104686. https://doi.org/10.1016/j.ijpvp.2022.104686

34.

Zhang

, et al. Effects of chloride on electrochemical and stress corrosion cracking behavior of 9Cr ferritic–martensitic Steel. Journal of Iron and Steel Research International 2019; 27(6): 719–731. https://doi.org/10.1007/s42243-019-00345-4

35.

Mohtadi-Bonab

. Effects of Different Parameters on Initiation and Propagation of Stress Corrosion Cracks in Pipeline Steels: A Review. Metals 2019; 9(5): 590. https://doi.org/10.3390/met9050590

36.

Wei

, et al. Corrosion mechanism of Pseudomonas stutzeri on X80 steel subjected to Desulfovibrio desulfuricans under elastic stress and yield Stress. Corrosion Science 2023; 216: 111084. https://doi.org/10.1016/j.corsci.2023.111084

37.

Zhao

Yang

Huang

, et al. Investigations and new insights on the relation between the valid interval of the Kaiser effect and the characteristic Stress. Earth-Science Reviews 2024; 249: 104673. https://doi.org/10.1016/j.earscirev.2023.104673