EWGAE2002

J. Acoustic Emission, 20 (2002) 206 © 2002 Acoustic Emission Group

NON-DESTRUCTIVE TESTING FOR CORROSION MONITORINGIN CHEMICAL PLANTS

M. WINKELMANS1 and M. WEVERS2

1 BASF Antwerpen N.V., Haven 725, B 2040 Antwerpen 4, Belgium2 Departement MTM, KU Leuven, B 3001 Heverlee, Belgium

Abstract

An appropriate and reliable tool for corrosion monitoring is necessary to guarantee thestructural integrity of chemical production units. By combining the electrochemical noise meas-urements and the acoustic emission technique, the active corrosion processes are revealed. Bycombining acousto-ultrasonics and fibre optics a reliable quantification of the corrosion damagecan be obtained. Electrochemical noise measurements, the acoustic emission technique, acousto-ultrasonics and fibre optics are very complementary non-destructive testing methods for moni-toring the ongoing corrosion in chemical production units.

Keywords: Corrosion monitoring, electrochemical noise measurements, acoustic emission tech-nique, acousto-ultrasonics, optical fibres

Introduction

Every day the chemical industry has to face new failures due to corrosion. In fact the majorpart of failures in the chemical industry are due to corrosion. In this way corrosion costs a lot tothe chemical industry. Corroded parts of the production units have to be replaced and bringabout production losses due to the shutdown of facilities. Corrosion failures also imply a consid-erable hazard to people and the environment. An appropriate and reliable tool for corrosionmonitoring is necessary to guarantee the structural integrity of chemical production units.

Since 1999, BASF Antwerpen N.V. (Belgium) and the research group "Materials Perform-ance and Non-Destructive Evaluation" of the Department of Metallurgy and Materials Engi-neering, K.U. Leuven (Belgium) are partners in developing a sensor to identify the active corro-sion processes destructive to the equipment and to quantify the degree of corrosion damage in anindustrial environment.

The research project is focused on the most important types of corrosion for the chemical in-dustry. General corrosion of carbon steel, general corrosion of stainless steel, pitting of stainlesssteel, stress corrosion cracking of carbon steel and stress corrosion cracking of stainless steel arestudied by four non-destructive testing methods. Electrochemical noise measurements, theacoustic emission (AE) technique, acousto-ultrasonics (AU) and fibre optics are the non-destructive testing methods used in this research project.

Experimental set up

Two completely different measurement principles are possible for a corrosion sensor. On theone hand a sensor can be developed so that it has to be fixed on the outside of the structurewhose corrosion behaviour is of interest. The structure itself corrodes and the sensor reveals

EWGAE2002

207

information about this corrosion behaviour. On the other hand it is possible to develop a sen-sor,whose sensor body itself corrodes. Information can only be obtained about the corrosion be-haviour of the sensor body itself. In this case the sensor has to be built in such a way that its cor-rosion behaviour is representative for the ongoing corrosion in the equipment of interest.

The decision has been made to develop a sensor, whose sensor body itself corrodes and revealsinformation about the ongoing corrosion. Figure 1 illustrates this principle. The four measure-ment techniques are incorporated on the same sensor body. The two passive non-destructivetesting methods (electrochemical noise measurements, the AE technique) are activated during thewhole life-cycle of the sensor body. The two active non-destructive testing methods (AU andfibre optics) are periodically activated. The periodicity is adapted as a function of the informa-tion revealed by the two passive non-destructive testing methods.

Fig. 1. Sensor principle

The basic research was done in the laboratory. The corroding sensor body was placed in aFaraday cage to prevent electromagnetic interference. Temperature and flow rate were con-trolled. The conversion to industrial environment has been done in two steps. First a mini-plantwas build to study the influence of temperature fluctuations, fluctuations in flow rate and inter-ference by other measurement equipment, pumps, heating systems, etc. Finally there were someexperiments in a bypass of the chemical facilities themselves. Here the basic measurement con-cept was tested under the rough industrial conditions.

Carbon steel 1.0038 according to DIN EN 10025 and stainless steel 1.4541 according to DINEN 10088 part 1 are used in the experiments. Figure 2 gives an overview of the sample geome-try used for the experiments.

EWGAE2002

208

(a) Sample geometry for corrosion monitoring (general and pitting).

(b) General corrosion. (c) Pitting corrosion.

(d) Sample geometry for SCC monitoring. (e) Stress corrosion crack.

Fig. 2. Sample geometries used for the corrosion and SCC experiments.

EWGAE2002

209

Electrochemical Noise Measurements (EN)

Due to the oxidation and reduction reactions of the corrosion processes at the interface ofmetal and electrolyte, a potential difference arises across this interface. The short and long-termevolution of this potential reveals information about the active corrosion processes [1, 2].

A nanovoltmeter (Keithley 2182) with an input resistance > 10 GΩ and a resolution of 100nV is used for the potential noise measurements. The nanovoltmeter is connected to a computerby a GPIB-interface. The nanovoltmeter is controlled by the computer and the measured data arestored on the hard disk. The sample frequency is 10 Hz. The potential differences are measuredbetween the sensor body and an Ag/AgCl reference electrode (Fig. 3).

Fig. 3. Experimental set up electrochemical noise measurements.

Figure 4 shows typical EN signals. General corrosion of carbon steel, general corrosion ofstainless steel, pitting of stainless steel and stress corrosion cracking of stainless steel can be dis-tinguished by pattern recognition techniques in combination with statistical analysis in the timedomain. Any information about pit growth during pitting and about crack growth during stresscorrosion cracking cannot be obtained. Electrochemical noise measurements have a higher accu-racy than more conventional non-destructive testing methods like dye penetrant, magnetic inter-ference or ultrasonic testing.

During the scaling-up of the experiments the potential noise measurements have been foundto be very reliable. No additional difficulties arose. In fact no additional filtering of the meas-ured data was necessary. Apparently the developed measurement concept for the potential noisemeasurements can be used in an industrial environment.

Acoustic Emission Technique

Corrosion processes cause microstructural changes at the metal/electrolyte interface. The en-ergy released by these microstructural changes results in transient elastic waves. Sounds ofhigher frequency and lower intensity than audible sounds can be measured by the acoustic emis-sion technique [3, 4]. The Fracture Wave Detector (Digital Wave Co., U.S.) is used for the ex-periments. Figure 5 shows the basic components of this system. The typical sample frequencyfor the experiments is 20 MHz. Broadband sensors (Digital Wave B1025) with a frequencyrange from 50 kHz to 2 MHz are used.

EWGAE2002

210

no corrosionstainless steel - CaCl2 40% - 85°C

general corrosion carbon steel - H3PO4 10% - 20°C

pitting - stainless steel brackish water+FeCl3 1%- 45°C

Fig. 4. Electrochemical noise measurements - typical signals.

-0.255

-0.250

-0.245

2475.0 2477.5 2480.0 2482.5 2485.0 2487.5 2490.0

time [minutes]

V

-0.475

-0.470

360.0 362.5 365.0 367.5 370.0 372.5 375.0

time [minutes]

V

0.070

0.075

0.080

0.085

0.090

0.095

2160.0 2162.5 2165.0 2167.5 2170.0 2172.5 2175.0

time [minutes]

V

EWGAE2002

211

stress corrosion cracking stainless steel - CaCl2 40% - 85°C

Fig. 4 (continued). Electrochemical noise measurements - typical signals.

Fig. 5. Experimental set up for AE technique.

Figure 6 shows typical AE signals. General corrosion of carbon steel, general corrosion ofstainless steel, pitting of stainless steel, stress corrosion cracking of carbon steel and stress corro-sion cracking of stainless steel can be distinguished by pattern recognition techniques, statisticalanalysis and the energy content in the time domain in combination with frequency analysis. Thepit initiation or crack initiation as well as the pit growth or crack growth can be monitored by theacoustic emission technique. This is important additional information revealed by the acousticemission technique in comparison to the electrochemical noise measurements. Apart from thisadditional information, the accuracy of the acoustic emission technique is higher than that ofmore conventional non-destructive testing methods like dye penetrant, magnetic interference orultrasonic testing.

The scaling-up of the AE experiments turned out that the development of an additional andvery accurate filtering system is necessary to use the AE technique as a reliable measurementtool in an industrial environment.

EWGAE2002

212

general corrosioncarbon steel - H3PO4 10% - 20°C

pitting - stainless steelbrackish water + FeCl3 1% - 45°C

stress corrosion crackingstainless steel - CaCl2 40% - 85°C

Fig. 6. AE technique - typical results.

-0.5

-0.4

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

0 50 100 150 200 250 300 350

µsec

V

-0.5

-0.4

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

0 50 100 150 200 250 300 350

µsec

V

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0 50 100 150 200 250 300 350

time [µsec]

V

EWGAE2002

213

Acousto-Ultrasonics

A calibrated ultrasonic pulse with a frequency of 300 KHz (Fig. 7a) is introduced in themetal. This ultrasonic pulse is introduced by a function generator (PHILIPS PM 5138). Changesin the metal morphology, the metal thickness and the metal microstructure change the wavepropagation and therefore the signal measured after propagation through the metal (Fig. 7b) [5,6]. The propagated signal is measured by Fracture Wave Detector (Fig. 5). The typical samplefrequency is 20 MHz. The complete experimental set up for the acousto-ultrasonic technique isshown in Fig. 8.

a injected pulse b pulse after propagation through the metal

Fig. 7. Injected pulse and pulse after propagation through the metal; general corrosion, carbonsteel - H3PO4 10% - 20°C.

Fig. 8. Experimental set up acousto-ultrasonic technnique

Knowledge of the wave type and the interference between several wave types is necessary toselect the most interesting part of the signal for further calculation. An energy loss in this par-ticular part of the signal is measured during ongoing corrosion. (Fig. 9) Corrosion damage due togeneral corrosion, pitting or stress corrosion cracking can be quantified in this way. Figure 10illustrates this quantification for general corrosion. The correlation between the calculated en-ergy losses of the acousto-ultrasonic measurements and the calculated loss in thickness based onthe weight loss of the specimens during the tests has a very high accuracy. The highest accuracyis obtained for intermediate and high corrosion rates. For low corrosion rates the accuracy is notthat high.

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

0 100 200 300 400 500 600 700µsec

V

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0 50 100 150 200 250 300 350

[µsec]

V

computer

sensor B1025listening

preamplifiertriggering

&filtering

functiongenerator

sensor B1025sending

EWGAE2002

214

general corrosion carbon steel H3PO4 10% - 20°C

Fig. 9. Energy loss during corrosion attack.

general corrosion carbon steel H3PO4 10% - 20°C

Fig. 10. Correlation energy loss - thickness loss.

The highest regression coefficients (Fig. 10) could be obtained in the laboratory. Howeveralso for the experiments in the mini-plant and in the industrial environment high regression coef-ficients can be obtained when additional filters are applied. In fact an accurate band-pass filtercentred around 300 kHz turned out to be very efficient.

y = 23.036x-1.4612

R2 = 0.9749

0.00

0.10

0.20

0.30

0.40

0.50

10 20 30 40 50 60 70 80

thickness loss [%]

rms /rms initial [%]

0

4

8

12

16

20

0 50 100 150 200 250 300 350

time [h]

rms [V]

EWGAE2002

215

Optical Fibres (OF)

An optical fibre is attached on the surface of interest. A laser-diode forces a light pulsethrough the fibre. Changes in the stress conditions on the surface due to the corrosion activitychange also the stress conditions of the optical fibre. The changing stress conditions of the opti-cal fibre modulate the light pulse. Analysis of this modulation is very useful in quantifying thecorrosion damage [7, 8].

Figure 11 shows the experimental set up. The main components are a 30-mW laser, a trans-impedance photodiode preamplifier with a DC bandwidth of 2 MHz and a gain of 1-10 MV/Aand the silica/silica optical fibre. The photodiode is connected to the computer by a PCI Bushigh-speed A/D board (Keithley KPCI-3110). The typical sample rate is 110 Hz.

Fig. 11. Experimental set up of optical fibres.

The variation in intensity of the light pulse is an appropriate tool for the quantification of theongoing corrosion. Figure 12 illustrates the rise in intensity for several experiments. The fibre isfixed on the surface of interest under a certain prestress. During ongoing corrosion the prestressdiminishes and the intensity of the laser pulse through the fibre will rise. The linear curve fittingat the initial part of this curve correlates with the corrosion rate. Optical fibres are very interest-ing to monitor general corrosion with relative low corrosion rates. In this way the optical fibresare complementary to the acousto-ultrasonics. During the scaling-up of the experiments the op-tical fibres has been found to be very reliable. No additional difficulties arose.

Conclusion

Electrochemical noise measurements, the acoustic emission technique, acousto-ultrasonicsand optical fibres are very complementary non-destructive testing methods for the monitoring ofgeneral corrosion of carbon steel, general corrosion of stainless steel, pitting of stainless steel,stress corrosion cracking of carbon steel and stress corrosion cracking of stainless steel (Fig. 13).

By combining the electrochemical noise measurements and the acoustic emission results theactive corrosion processes are revealed. The electrochemical part of the corrosion processes isvisualised by the electrochemical noise measurements. The acoustic emission technique moni-

computer

photo diode

LASERbeam

optical fibre

EWGAE2002

216

general corrosioncarbon steel - H3PO4 10% - 20°C

Fig. 12. Intensity rise in optical fibres. (Io is the initial intensity).

tors the mechanical part of the corrosion processes like the breakdown of passive films, crackgrowth, pit growth.

By combining acousto-ultrasonics and fibre optics the corrosion damage can be quantified.Acousto-ultrasonics have the highest accuracy for intermediate and relative high corrosion rates.Optical fibres have the highest accuracy for relative low corrosion rates.

By combining electrochemical noise measurements, the acoustic emission technique,acousto-ultrasonics and fibre optics a sensor to identify the active corrosion processes and toquantify the corrosion damage in chemical production units has born.

This combination of non-destructive testing methods has accuracy higher than that of moreconventional non-destructive testing method like dye check, magnetic interference or ultrasonictesting.

Acknowledgements

This research has been realised with financial support from the Flemish Institute for the pro-motion of scientific-technological research in industry (IWT), project 010261/BASF Antwerpen.

0.90

1.00

1.10

1.20

1.30

1.40

1.50

1.60

1.70

0 25 50 75time [h]

I/Io

increasingcorrosion rate

EWGAE2002

217

Fig. 13 Four complementary non-destructive measurement methods.

References

[1] C. Gabrielli and M. Keddam, Review of Applications of Impedance and Noise Analysis toUniform and Localized Corrosion, Corrosion, National Association of Corrosion Engineers,October, pp. 794- 811 (1992).[2] J. Göllner und A. Burkert, Untersuchung der Korrosionsneigung mit Hilfe des elektro-chemischen Rauschens, Materials and Corrosion, 49, 614-622 (1998).[3] R.V. Williams, Acoustic emission, Adam Hilger Ltd, Bristol, 1980.[4] M. Wevers, I. Verpoest, P. De Meester, and E. Aernoudt, Acoustic emission: current practiceand future directions (ASTM STP 1077), eds. W. Sachse, J. Roget and K. Yamaguchi, AmericanSociety for Testing of Materials, 1991, pp. 416-423.[5] J.H. Williams Jr. and S.S. Lee, Promising quantitative nondestructive evaluation techniquesfor composite materials Materials Evaluation 43, April 1985, 561-565.[6] A. Vary and R.F. Lark, Correlation of fibre composite tensile strength with the ultrasonicstress wave factor, Journal of Testing and Evaluation, 7(4), 185-191.[7] A. Mendez, T.F. Morse, and F. Mendez, Applications of Embedded Optical Fiber Sensors inReinforced Concrete Buildings and Structures, Proc. SPIE Fiber Optic Smart Materials andStructures, Vol 1170, Sept. 1994.[8] Monitoring Corrosion On-Line With Fiber Optics, EPRI Journal, November/December 1995,p. 5.

AU, (OF)EN, AEpitting stainless steel

AU, (OF)EN, (AE)stress corrosion cracking stainless steel

AU, (OF)AEstress corrosion cracking carbon steel

AU, OFEN, (AE)general corrosion stainless steel

AU, OFEN, (AE)general corrosion carbon steel

EN, (AE)no corrosion

QuantificationIdentification

AU, (OF)EN, AEpitting stainless steel

AU, (OF)EN, (AE)stress corrosion cracking stainless steel

AU, (OF)AEstress corrosion cracking carbon steel

AU, OFEN, (AE)general corrosion stainless steel

AU, OFEN, (AE)general corrosion carbon steel

EN, (AE)no corrosion

QuantificationIdentification