European Federation

European Federation of Corrosion Publications

NUMBER 31

Corrosion of Reinforcement in Concrete

Corrosion Mechanisms and Corrosion Protection

Papersfrom EUROCORR '99

Edited by J. MIETZ, R. POLDER AND B. ELSENER

Published for the European Federation of Corrosion by IOM Communications

www.iran-mavad.com مرجع علمى مهندسى مواد

Book Number 746 Published in 2000 by IOM Communications Ltd

1 Carlton House Terrace, London SWlY 5DB

IOM Communications Ltd is a wholly-owned subsidiary of

The Institute of Materials

0 2000 IOM Communications Ltd

All rights reserved

ISBN 1-86125-133-5

Neither the EFC nor The Institute of Materials is responsible for any views expressed

in this publication

Typesetting by spiresdesign

Made and printed in Great Britain


Reinforced concrete is such a widely used construction material that in most of the world the cost of damage and repair is an important issue despite the fact that the majority of structures are undamaged and have a long service life. In fact, only a small fraction of the total stock is responsible for most of the problems with reinforcement corrosion as the main cause. Owners, engineers and researchers are trying to find solutions to the problems that are involved. Although the most important factors in the corrosion of reinforcing steel are well known necessary work continues in the understanding of rate controlling mechanisms, in developing test methods to assess the severity of existing corrosion and in producing computer based diagnostic systems.

Various methods to protect structures have been applied for many years and new ideas have been introduced in recent decades. Cathodic protection of reinforcement is a successful method to stop ongoing corrosion and this application is being extended to new types of structures, to the use of new materials and to links with conventional repairs. Other new technologies in the field are the use of stainless steel, inhibitors and water repellent treatments for improved service life.

This volume in the EFC series brings together the full papers presented in the successful session ”Corrosion of Steel in Concrete’’ at E UROCORR ’99 held at Aachen, Germany. Thirteen papers were accepted after peer review, and included contributions from The Czech Republic, Denmark, Germany, Italy, The Netherlands, Norway, Poland and Switzerland.

The papers are grouped under two headings:

Corrosion Mechanisms and Corrosion Measurements Corrosion Protection of Reinforced Concrete Structures

We thank all authors who were willing to share their valuable ,.nowlecge with others and who participated in discussions. The editors hope these papers will encourage readers to apply the ideas and results that were presented to their own problems. The work reported in this volume can be read in conjunction with other EFC publications from this Working Party, for example:

EFC 18 EFC 24

EFC 25

Stainless Steels in Concrete: State of the Art Report Electrochemical Rehabilitation Methods for Reinforced Concrete Structures: State of the Art Report Corrosion of Reinforcement in Concrete: Monitoring, Prevention and Rehabilitation

and in conjunction with further volumes in preparation on the subjects of inhibitors for use in concrete, and embeddable reference electrodes for concrete.

J. MIETZ R. POLDER B. ELSENER

Chairman of the EFC Working Party on Corrosion in Concrete


European Federation of Corrosion Publications Series Introduction

The EFC, incorporated in Belgium, was founded in 1955 with the purpose of promoting European co-operation in the fields of research into corrosion and corrosion prevention.

Membership is based upon participation by corrosion societies and committees in technical Working Parties. Member societies appoint delegates to Working Parties, whose membership is expanded by personal corresponding membership.

The activities of the Working Parties cover corrosion topics associated with inhibition, education, reinforcement in concrete, microbial effects, hot gases and combustion products, environment sensitive fracture, marine environments, surface science, physico-chemical methods of measurement, the nuclear industry, computer based information systems, the oil and gas industry, the petrochemical industry, coatings, automotive engineering and cathodic protection. Working Parties on other topics are established as required.

The Working Parties function invarious ways, e.g. by preparing reports, organising symposia, conducting intensive courses and producing instructional material, including films. The activities of the Working Parties are co-ordinated, through a Science and Technology Advisory Committee, by the Scientific Secretary.

The administration of the EFC is handled by three Secretariats: DECHEMA e. V. in Germany, the Sociktk de Chimie Industrielle in France, and The Institute of Materials in the United Kingdom. These three Secretariats meet at the Board of Administrators of the EFC. There is an annual General Assembly at which delegates from all member societies meet to determine and approve EFC policy. News of EFC activities, forthcoming conferences, courses etc. is published in a range of accredited corrosion and certain other journals throughout Europe. More detailed descriptions of activities are given in a Newsletter prepared by the Scientific Secretary.

The output of the EFC takes various forms. Papers on particular topics, for example, reviews or results of experimental work, may be published in scientific and technical journals in one or more countries in Europe. Conference proceedings are often published by the organisation responsible for the conference.

In 1987 the, then, Institute of Metals was appointed as the official EFC publisher. Although the arrangement is non-exclusive and other routes for publication are still available, it is expected that the Working Parties of the EFC will use The Institute of Materials for publication of reports, proceedings etc. wherever possible.

The name of The Institute of Metals was changed to The Institute of Materials with effect from 1 January 1992.

The EFC Series is now published by the wholly-owned subsidiary of The Institute of Materials, IOM Communications Ltd.

A. D. Mercer EFC Series Editor, The Institute of Materials, London, UK


. I .

Vl l l Series Introduction

EFC Secretariats are located at:

Dr B A Rickinson European Federation of Corrosion, The Institute of Materials, 1 Carlton House Terrace, London, SWlY 5DB, UK

Mr P Berge Federation Europeene de la Corrosion, Societe de Chimie Industrielle, 28 rue Saint- Dominique, F-75007 Paris, FRANCE

Professor Dr G Kreysa Europaische Foderation Korrosion, DECHEMA e. V., Theodor-Heuss-Allee 25, D-60486, Frankfurt, GERMANY


Contents

Series Introduction

Preface

Foreword

uii

ix

X

Part I - Corrosion Mechanisms and Corrosion Measurements 1

1. Oxygen Reduction on Mild Steel and Stainless Steel in Alkaline Solutions S. JAGGI, B . E L S E N E R A N D H . BOHNI

3

2. Investigations on Cathodic Control of Chloride-Induced Reinforcement Corrosion M. RAUPACH A N D J . GULIKERS

13

3. Critical Factors for the Initiation of Rebar Corrosion L. ZIMMERMANN, B. ELSENER AND H. BOHNI

4. Field Tests of Chloride Penetration into Concrete with Microsilica 0. VENNESLAND AND J. HAVDAHL

5. Comparison of Electrochemical Data and Mass Loss Corrosion Rate Measurements for Steel Reinforcement in Concrete P. NOVAK AND R. MALA

25

35

41

Part 2 - Corrosion Protection of Reinforced Concrete Structures 49

6. Corrosion and Protection in Reinforced Concrete: A Computerised System for Studying its Phenomenology, Causes, Diagnosis and Remedies P. PEDEFERRI

51


vi Contents

7. Organic Corrosion Inhibitors for Steel in Concrete B. ELSENER, M. BUCHLER AND H. BOHNI

8. Corrosion Protection of Reinforcement by Hydrophobic Treatment of Concrete R. B. POLDER, H . B O R S ~ E AND J , DE VRIES

9. Cathodic Protection of Concrete Ground Floor Elements with Mixed-in Chloride G. SCHUTEN, J . LEGGEDOOR AND R. B. POLDER

10. Sacrificial Anodes for Cathodic Prevention of Reinforcing Steel Around Patch Repairs Applied to Chloride-Contaminated Concrete G. SERGI A N D C . L. PAGE

11. Layer Zinc Anodes in Cathodic Protection of Steel Reinforcement W. BOHDANOWICZ

12. Lifetime Extension of Thermally Sprayed Zinc Anodes for Corrosion Protection of Reinforced Concrete Structures by Using Organic Top-coatings I . SPRIESTERSBACH, A. MELZER, 1. W I S N I E W S K I , A. WINKELS A N D M. KNEPPER

13. Practical and Economical Aspects of Application of Austenitic Stainless Steel, AIS1 316, as Reinforcement in Concrete

A N D T. SKOVSGAARD 0. KLINGHOFFER, T. FRaLUND, B. KOFOED, A. KNUDSEN, F. M. J E N S E N

61

73

85

93

101

109

121

List of Abbreviations

Index

135

137


Part 1

Corrosion Mechanisms

and Lorrosion

Measurements


1 Oxygen Reduction on Mild Steel and Stainless

Steel in Alkaline Solutions

S. JAGGI, B. ELSENER and H. BOHNI Institute of Materials Chemistry and Corrosion, Department of Civil Engineering Swiss Federal Institute of

Technology, ETH Honggerberg, CH-8093 Zurich, Switzerland

ABSTRACT The cathodic polarisation curve of steel in alkaline solutions always shows three regions: (1) oxygen reduction with a Tafel behaviour at potentials cathodic to the open circuit potential followed by (2) a diffusion limited current of oxygen reduction at more negative potentials and (3) hydrogen evolution at very negative potentials. The diffusion limited region of the cathodic current density is controlled both by the oxygen concentration in solution and the flow rate whereas in the Tafel region (charge transfer) the temperature and the pretreatment of the sample determine the intensity of the current density and the slope of the Tafel line. On stainless steels the cathodic reduction currents are lower then on mild steel. It can be concluded that under usual corrosion conditions for steel in concrete the cathodic oxygen reduction is not diffusion limited but charge transfer controlled.

1. Introduction

The corrosion reaction, i.e. the anodic dissolution of steel in concrete, has to be sustained by a corresponding cathodic reaction: in general, this is the reaction of oxygen with water producing hydroxyl ions:

0, + 2H,O + 4e- -+ 40H- (1)

The availability of oxygen at the steel surface and the reaction kinetics of oxygen reduction are thus key factors in the corrosion of steel in concrete. Quite often the reduction of oxygen at the steel surface in alkaline environments is called diffusion limited although only a few papers report results on kinetics and mechanism of oxygen reduction on passive iron or steel in alkaline solutions [1,2]. The influence of oxygen on corrosion of steel in concrete has been studied [3-51. In this work, the influence of oxygen content, temperature and ageing of the passive film on the oxygen reduction reaction on normal and stainless steel in alkaline solutions has been studied. The results are discussed with respect to the mechanism of oxygen reduction and the importance for corrosion of steel in concrete.


4 Corrosion of Reinforcement in Concrete: Corrosion Mechanisms and Corrosion Protection

2. Experimental

Potentiodynamic polarisation curves (scan rate 1 mVs-l) were recorded in an electrochemical flow cell (Fig.1) at defined temperatures and hydrodynamic conditions. The flow velocity was regulated by the flux of solution, 1 mLs' corresponds to a flow velocity of ca. 1.4 mms-I. The counter electrode was a platinum wire spiral and the reference electrode was a saturated calomel electrode. The materials tested (working electrode) were mild steel and DIN 1.4301 stainless steel cylinders with a diameter of 8 mm, embedded in resin and mounted in the flow cell. For each experiment the samples were freshly ground with 180 grit emery paper in water, cleaned with ethanol in an ultrasonic bath, rinsed with deionised water and immersed for 24 h in the alkaline test solution open to air to form the passive film. As electrolytes 0 . 1 ~ NaOH and synthetic pore solution (Table 1) were used. The temperature (5-47OC), oxygen concentration (open to air or saturated) and flow velocity (stagnant or 1 mLs-l) were varied; in addition, experiments with prolonged immersion times (ageing of the passive film up to 4 months) of the samples were performed.

Fig. 2 Schematic representation of the electrochemical pow-cell that allows the registration of cathodic polarisation curves under controlled temperature, potential and oxygen content.


Oxygen Reduction on Mild Steel and Stainless Steel in Alkaline Solutions 5

Table 1. Composition of the Synthetic concrete pore solution

Concrete pore solution

mgL-'

Ca(OH), KOH Na,SO, NaOH

9.6 13967.8 3121.6 616.8

3. Results

The cathodic polarisation curves, starting from the open circuit potential, in solutions of 0 . 1 ~ NaOH with different oxygen contents (open to air, saturated) are shown in Fig. 2(a). The diffusion limited current density increases by a factor of about two in 0, saturated conditions. The slope of the Tafel region of the curve remains constant at ca. 250 mV per decade, although the higher oxygen content results in a slightly higher cathodic current density in the Tafel region. The influence of flow velocity at constant oxygen content (solution open to air) is shown in Fig. 2(b). As expected, the diffusion limited current density, io Dl increases with higher flow velocity, the current densities of oxygen reduction in tAe Tafel region of the polarisation curve are not influenced by the flow velocity.

The influence of temperature on the cathodic polarisation curves of oxygen reduction was studied in more detail in synthetic pore solution, with solutions open to air and flow velocity 0.1 mLs-l. As is shown in Fig. 3, the diffusion limited current



-1400 -1200 -1000 -800 -600 -400 -200 0 Potential (mV, SCE)

Fig, 2(b)Typical catkodicpolarisation curves of mild steel in 0 . 1 ~ NaOHsolu f ions as a function of solutionflow rate. Solutions open to air.

Potential (mV, SCE) Fig. 3 Catkodicpolarisation curves of mild steel in synthetic pore solution (pH 13.4) a t different temperatures, flow rate 1 rnLs-l, open to air.



density, ioZD, remains practically constant. The Tafel region of the polarisation curve is slightly shifted to more negative potentials, the slope of the Tafel region increases slightly with increasing temperature (Table 2).

Prolonged exposure to the test solution (ageing of the passive film) showed the most pronounced effect on the cathodic polarisation curves (Fig. 4). An increase of this pre-passivation time results in a marked increase of the slope of the Tafel region, the diffusion limited current density remaining unchanged. Finally, as can be seen from Fig. 5, the electrode material influences the cathodic polarisation curve: with the same surface preparation the cathodic current density of DIN 1.4301 stainless steel is about 4 times lower compared to mild steel. Platinum shows practically uninhibited oxygen reduction with a Tafel slope of -60 mV.

4. Discussion

4.1. Kinetics of the Cathodic Oxygen Reduction

As it is well known from literature, cathodic oxygen reduction (e.g. Fig. 2) shows a Tafel line at low overpotentials and a diffusion limiting current density at high overpotentials (more negative potentials). The thermodynamic equilibrium potential E corresponding to eqn (1) depends, according to the Nernst equation,

0 2

(2 ) 0 2

E = +1.27-2.3RT/F pH+2.3RT/4F logp 0 2

on pH, oxygen content and temperature of the solution and E = +0.23 V SCE would result theoretically. On platinum, a potential E of -0.05 V SCE was

0 2 measured. On passive steel in alkaline solutions, the equilibrium potential E cannot be measured; since due to the interaction with the anodic dissolution of the passive film mixed open circuit potentials around -0.2 V SCE are found which influenced by the magnitude of the cathodic current densities and the passive film dissolution rates (e.g. Fig. 5).

Table 2. Cathodic Tafel slopes b,, charge transfer coefficient a and exchange current densities for the oxygen reduction on passive mild steel i n synthetic pore solutions at different temperatures

0 2

0 2



0 -1400 -1200 -1000 -800 -600 -400 -200

Potential (mV, SCE) Fig. 4 Influence of the ageing time of the passivefilm in synthetic pore solution on cathodic polarisation curves (flow rate 1 mLs-I, open to air).

1 l000.00

100.00

10.00

1 .oo

0.10

0.01 0 -1400 -1200 -1000 -800 -600 -400 -200

Potential (mV, SCE) Fig. 5 Cathodic polarisation curves ofmild steel, 1.4301 stainless steel and platinum in synthetic pore solution atflow velocity of 1 mLs-I compared to results of Pedeferri 161.



On increasing the overpotential in the cathodic direction a Tafel behaviour is found corresponding to the charge transfer controlled region of oxygen reduction. In the experiments on passive metals, this Tafel line could be observed only at E << Eocp (Figs 4,5). The current densities in this charge transfer controlled region do not change with flow velocity (Fig. 2b) as observed also in studies with rotating disk electrodes [1,2]. Assuming that in this potential region oxygen transport does not influence the overall reaction rate, the cathodic polarisation curve can be written as

io2 = io,o2 exp [- ( E - Eo2) / b,] (3)

with E , = equilibrium potential, io,, = exchange current density at E and the cathodi?Tafel slope b, = RT/F (1-a), a heing the charge transfer coefficient and R, T, and F have their usual meanings. Curve fitting of the experimentally measured cathodic polarisation curves (Fig. 3) with eqn ( 3 ) allows these parameters to be determined (Table 2).

Whereas on platinum a steep Tafel line (slope 60 mV/decade) is observed, on passive metals (both mild steel and stainless steel) the cathodic reduction of oxygen is inhibited. This inhibition might cease once the steels start to corrode. The lower cathodic current densities observed on 1.4301 stainless steel, reported also by Bertolini e t al . [6], correspond to a lower exchange current density iOro2. Thus stainless steel in alkaline solutions is a less effective cathode compared to mild steel (Fig. 5), this can be important when considering repair of reinforced concrete structures.

The observed influence of temperature (Fig. 3) on the charge transfer reaction rate (eqn (3)) can be explained by the factor RT/F in the Tafel slope, the charge transfer coefficient aremaining constant (Table 2). A marked increase in the exchange current density (factor of 3 between 5 and 47°C) can be observed, the calculated activation energy is AE = 16 kJmol-l.

At very high overpotentials the overall reaction rate is limited by oxygen transport to the electrode surface. In this potential region a diffusion limited current density, io,.o, is observed. According to

0 2

iOpD = 4 F D Co2 16 (4 ) 0 2

the oxygen concentration, C and the thickness of the diffusion layer, 6, determine the limiting current density. Despite increasing temperatures iOpD remains practically constant (Fig. 3), thus the higher diffusion coefficient is compensated by the lower oxygen solubility at higher temperatures.

The kinetics of oxygen reduction as discussed above can be considered valid also for the cathodic reaction on steel in mortar or concrete - with two additional features: the oxygen content and the water content in the porous cementitious materials will vary strongly with humidity. According to eqn (1) both the oxygen content and free water may influence the reaction kinetics of oxygen reduction.

the oxygen diffusion coefficient D 02, 0 2


10 Corrosion of Reinforcement i n Concrete: Corrosion Mechanisms and Corrosion Protection

4.2. Corrosion of Steel in Concrete

The results of this laboratory study on the cathodic reduction of oxygen in alkaline solutions have several implications in the field of corrosion of steel in concrete. First, the progressive inhibition of the oxygen reduction (Fig. 4) with prolonged ageing of the passive steel (long time in the alkaline pore solution, mortar or concrete) will lead -for otherwise constant conditions -to lower corrosion rates of steel in concrete because the cathode efficiency decreases. This means that in constant environmental conditions - as frequently established in laboratory experiments - the corrosion rate of steel in concrete will decrease with time. Similar results have been reported by Yalcyn et al. [9]. Regarding the rate controlling reaction of corrosion of steel in concrete, these results from solutions indicate that a true diffusion limited corrosion current is found only for fully immersed conditions which is in agreement with the experiments of Raupach [3,4]. Under more usual environmental conditions, i.e. not completely water-saturated, the cathodic reaction is oxygen reduction under charge transfer control as has been confirmed by numerical simulations [lo].

Not only the cathodic but also the anodic reaction (pitting corrosion of steel in chloride-containing media) is strongly influenced by the ageing of the passive film: as is shown in Fig. 6, prolonged immersion in alkaline solution shifts the pitting potential to much more positive values. Similar results have been reported in a surface analytical study on the ageing of passive films on stainless steels [ll]. This fact has to be considered in studying pitting corrosion of steel in concrete.

4.3. Measuring Oxygen Content

Cathodic polarisation at a constant potential and recording the resulting reduction current is used to determine the oxygen content in situ (on structures or on laboratory samples) [5,12]. The results of this work indicate that the choice of the electrode material (mild steel, stainless steel, platinum) and of the cathodic polarisation potential may strongly affect the results: with increasing duration of the experiment, i.e. with ageing of the passive film, lower oxygen reduction currents will be measured at potentials in the charge transfer region (e.g. -0.7 V SCE (Fig. 4)) even if the oxygen content effectively remains constant. Only at potentials in the limiting current density region has this ageing effect not been observed [5]. The limiting diffusion current densities of steel embedded in mortar or concrete have not been determined in this work but it is expected that they will be lower than in solution [5,8]. Further results will be reported [13].

4.4. Beneficial Influence of Stainless Steel

The DIN 1.4301 stainless steel shows much lower oxygen reduction currents compared to normal mild steel (Fig. 5) and this has also been reported by Pedeferri et al. [6,7]. This will reduce the corrosion of an actively corroding steel part in a macrocell and thus it is beneficial to use stainless steels in new and repaired structures not only because of a much higher pitting potential but also because of the reduced macrocell corrosion activity [6,7,13]. Care has to be taken because this effect can get lost when the stainless steel has been cold-deformed or welded [6,7].



c c

Potential [mVscE ]

Fig. 6 Influence of the ageing of the passivefilm on mild steel in synthetic pore solution on the anodic polarisation curves (positive currents) and on the pit t ing potential E, (chloride concentration 1 molL-l).

5. Conclusions

From this laboratory study on the reduction of oxygen on mild steel and stainless steel in alkaline solutions it can be concluded that

the oxygen reduction on passive mild steel and even more on stainless steel in alkaline solutions is strongly inhibited compared to that on platinum. This inhibition increases with prolonged ageing of the passive film;

the oxygen reduction current density increases with temperature, this is mainly the result of higher exchange current densities; and

experiments to monitor the oxygen content in mortar or concrete should be performed only in the limiting current density region.

Further experiments in mortar and concrete are ongoing to establish the potential region of the diffusion controlled limiting current density region.



6. Acknowledgements

The authors acknowledge the financial support of this research by the Swiss Federal Highway Authorities within the research program 'Briickenunterhaltsforschung'.

References

1. S. LJ. Gojkovic, S. K. Zecevic and D. M. Drazic, Electrackim. Acta, 1994, 39, 975-982. 2. H. S. Wroblowa and S. B. Qaderi, J. Electroanal. Ckem., 1990, 279,231-243. 3. M. Raupach, Mater. Struct., 1996, 29, 174-184. 4. M. Raupach, Mater. Struct., 1996, 29, 226-232. 5. C. Alonso, C. Andrade and A. M. Garcia, The role of oxygen in the kinetics of the corrosion of reinforcements, in Proc. EUROCORR '98, Utrecht, 1998. Publ. Netherlands Corrosion Centre, P.O. Box 120,3720 AC Bilthoven, The Netherlands. 6. L. Bertolini, F. Bolzoni, T. Pastore and P. Pedeferri, Brit. Corros. I., 1996, 31,218-222. 7. L. Bertolini, M. Gastaldi, T. Pastore, M. P. Pedeferri and P. Pedeferri, Experiences on Stainless Steel Behaviour in Reinforced Concrete, in Proc. EUROCORR '98, Utrecht, 1998. Publ. Netherlands Corrosion Centre, P.O. Box 120,3720 AC Bilthoven, The Netherlands. 8. J. M. Sykes, Mater. Sci. Forum, 1995,192-194,833-842. 9. H. Yalcyn and M. Ergun, Cern. Concr. Res., 26, 1593. 10. M. Raupach and J. Gulikers, Investigations on Cathodic Control of Chloride-Induced Reinforcement Corrosion, this volume, pp.13-23. 11. A. Rossi and B. Elsener, Mater. Sci. Forum, 1995,185-188,337-346. 12. Vennesland, Nordic Concrete Res., 1991,10, 139. 13. S. Jaggi, B. Elsener and H. Bohni, to be published in Cement and Concrete Res.


2 Investigations on Cathodic Control of

Chloride-Induced Reinforcement Corrosion

M. RAUPACH and J. GULIKERS* Institute for Building Materials Research of the Technical University of Aachen, ibac, Germany

*Ministry of Transport, Utrecht, The Netherlands

ABSTRACT There are still contradictions regarding the mechanisms of reinforcement corrosion with respect to the controlling rate-determining factors of the corrosion process. It is often stated that the electrolytic resistance of the concrete is the controlling factor and that the corrosion rates can subsequently be calculated from concrete resistivity. However, extensive research carried out by the authors has clearly demonstrated that instead of concrete resistivity the resistance to cathodic polarisation is normally the controlling factor in the case of chloride-induced macrocell corrosion.

Generally, cathodic control can be related to restricted oxygen diffusion or activation control. In the present paper these relationships are discussed in detail based on results of numerous tests on the cathodic polarisation behaviour of passive reinforcement. For simple defined geometrical conditions simulating practical cases it is shown by a numerical analysis that the resistance to activation is normally the controlling factor for the corrosion rate and that oxygen diffusion has only to be taken into account when the concrete is permanently water saturated or extremely dense.

To verify whether it is correct to estimate corrosion rates from concrete resistivity data, tests should be carried out to check the influencing parameters on concrete resistivity and cathodic activation of passive steel surface areas.

1. Introduction

To evaluate the durability of existing structures and the design for durability of new reinforced concrete structures there is a need to estimate or calculate corrosion rates for specific environmental conditions. Besides complex models based on the electrochemical nature of the corrosion process (see e.g. [l]) simplified models have also been developed which assume that the resistivity of the concrete is the dominating and controlling factor for the corrosion rate of steel in concrete (see e.g. [2]).

However, extensive research on the mechanism of chloride-induced macrocell corrosion of steel in concrete carried out by both authors of this paper has shown that for conditions relevant to actual practice the dominating factor is not the resistivity of the concrete, but the resistance of the passive reinforcement to cathodic polarisation.

In this chapter the basic relationships regarding the mechanism of chloride-induced corrosion are briefly explained to allow the further discussion of the controlling factors, especially the resistance of passive steel to polarisation for combined oxygen diffusion and charge transfer (activation).



2. Electrochemical Background of Macrocell Corrosion

Based on a simplified electrical circuit model [ 11 the electrical macrocell current, Igu,, which is proportional to the corrosion rate, can be calculated from the ratio between the driving voltage and the resistances present in the corroding system, using the following basic eqn (1):

galvanic current between anode and cathode rest potential of the cathode rest potential of the anode specific anodic polarisation resistance of the anode anodically acting steel surface area of the anode specific cathodic polarisation resistance of the cathode cathodically acting steel surface area of the cathode specific resistance of the concrete (resistivity) cell constant (geometry)

The resistance of the steel related to the transport of electrons is negligibly small compared to the other resistances in the circuit. The resistances of anode, cathode and electrolyte can again be sub-divided into different partial resistances (see e.g. [3]). The influence of the corrosion cell geometry on the overall concrete resistance is taken into account by the introduction of a cell constant, k .

It is evident that a prediction of the corrosion rate is difficult due to the large number of influencing factors. In addition, the relationships between these factors and the three main resistances are different: water saturation leads e.g. to a low concrete resistivity but a high cathodic resistance (oxygen diffusion control).

The relationships of eqn (1) can be illustrated by a so-called Evans diagram as shown in Fig. 1.

To quantify the control of the resistances, the voltage drops induced by the individual resistances can be calculated and related to the overall driving voltage, U,,, e.g. for the cathodic control part CTRL,:

CTRL, = AU,/ U,, X 100 [%] ( 2 )

where AU, = voltage drop induced by the cathodic reaction = Ux,, - U, and U,, = open circuit voltage (driving force) = UR,, - UR,a.

The controlling parts of the other two resistances can be calculated analogous to eqn (2).


lnvestigations on Cathodic Control of Chloride-Induced Reinforcement Corrosion 15

potential U

1 uA I U R,A

I I

lgal i current?

Fig. 1 Evans diagram showing schematically the relationship between voltage and current of a corrosion cell.

Before these controlling factors are calculated for typical macrocells, the influence of oxygen diffusion on the cathodic resistance to polarisation is discussed in the following section.

3. Influence of Oxygen Diffusion and Charge Transfer on the Cathodic Resistance to Polarisation

The voltage drop associated with the cathodic reaction is composed of a charge transfer and a diffusion component, i.e. AUc = AUc,cf + AUc,,. Hence, the overall cathodic resistance to polarisation, rc, can be divided into two sub-resistances:

rc = rc,ct + r,,d (nm2) (3)

where rc,cf = charge transfer resistance (activation control) and rc,d = diffusion resistance.

The polarisation behaviour of a passive steel surface in concrete can be calculated using the modified equation from the authors of this paper as described in [l] which allows for a discrimination between AUc,ct and AUC,, as follows:



For passive steel values of l.1015 mV/dec and 176 mV/decade for the anodic and cathodic Tafel slopes, b, and b, respectively, are introduced which results in a Stern- Geary constant B = 76.6 mV. The corrosion current density of passive steel at the rest potential, i,,,,, is set at 0.1 rnArn-, (0.01 yAcm-2) [1,3,4]. The influence of oxygen diffusion through the concrete cover on the polarisation behaviour is taken into account by introducing a limiting diffusion current density, ilim [3].

For steady state conditions, the limiting current densities for the cathodic reaction can be determined from the oxygen diffusion coefficients based on Fick's first law of diffusion assuming an oxygen concentration of 9.4 .lo6 mol 0, / cm3 at the concrete surface and 0 at the steel surface:

where Do2 = oxygen diffusion coefficient (m2s-I);

c, = concrete cover in (mm);

Acon/Ast = surface area ratio between concrete and steel acting cathodically; - < 1 (reduction factor if the concrete surface is the critical path).

For reinforced concrete structures the surface ratio Aco,z/Ast is usually > 1. For this case the oxygen diffusion near the steel surface is the critical path and the concrete surface area does not play the dominating role with respect to oxygen flow. On the contrary, if the reinforcement density is high, Acon/Ast may be 5 1 and the concrete surface area becomes the critical path resulting in a reduction of the potential limiting cathodic current density as shown in eqn (5).

Figure 2 shows the results from calculations on cathodic behaviour of passive steel using eqn (5) for different oxygen diffusion coefficients, a concrete cover of 80 mm and a surface area ratio of Acon/Ast = 0.67 as an example for a concrete cover with a high resistance to oxygen diffusion. The same results as given in Fig. 2 are shown in Fig. 3 in a modified form by plotting the control part induced by oxygen diffusion with respect to the overall cathodic resistance against polarisation in %.

It is clearly shown, that under these conditions an influence of oxygen diffusion is significant only for oxygen diffusion coefficients below about m2s-I, A considerable influence of concrete cover of the corrosion rate should be expected only for oxygen diffusion coefficients below about m2s-I. According to [SI, such low diffusion coefficients in the range of m2s-I occur mainly in water-saturated concretes.

These calculations show, that in situations where the concrete is not permanently


Investigations on Cathodic Control of Ckloride-lnduced Reinforcement Corrosion 17

h

k U E

E a

Y

m E c

L

0 0 ‘CI 0

.- 5 8

100

10

1

0.1

0.01

00 1 o-8 10-9

1 O-l1

0 -100 -200 -300 -400 -500 Cathodic potential shift, AU, (mV) .

Fig. 2 Cathodic behaviour of passive steel i n concrete calculated for different oxygen diffusion coeficients.

100 90 80 70 60 50 40 30 20 10 0

DO, (mzs-1)

lo-”

1 o-’O

I 0-9

1 o4 00

-1 00 -200 -300 -400 -500 Cathodic potential shift, AUc (mV)

Fig. 3 Influence of the oxygen diffusion coeficients on the cathodic resistance to polarisation of passive steel in concrete (see also Fig. 2).

water-saturated, oxygen diffusion is not a decisive factor influencing the corrosion rate of the reinforcement directly. On the other hand, the cathodic charge transfer resistance is not negligible.



4. Numerical Simulation of Macrocells with Planparallel Arrangement

The control exerted by the anodic, electrolytic and cathodic processes on the electrochemical behaviour of a macrocell was investigated by using a numerical simulation model. This mathematical analysis is based on the electrochemical relationships between current and potential as outlined in the foregoing. For simplicity, a planparallel arrangement of two separate steel plate electrodes representing a variable number of steel bars is considered placed on the top and bottom surface of a concrete specimen, respectively (see Fig. 4).

The steel/concrete interface of the top electrode is considered to be actively corroding whereas the interface of the bottom electrode remains essentially passive. The potential difference between corroding and passive steel serves to provoke a so- called active /passive macrocell with the corroding steel acting as the overall anode and the passive steel as the overall cathode. The resulting electrochemical interaction may have a pronounced accelerating effect on the corrosion rate of the anodic component of the macrocell couple. This situation is designed to simulate actual conditions of galvanic corrosion, e.g. the cross section of a bridge or parking deck suffering from chloride penetration from the top concrete surface. The ensuing corrosion process is stimulated by the passive steel rebars at the bottom surface which remain essentially passive.

The concrete specimen is given a cross section, measuring 400 x 200 mm2, and a unit depth of 1000 mm. The width of the passive and corroding steel electrodes (depth of 1000 mm) can be adjusted to vary the surface area ratio PIA. In this numerical simulation the ratio is increased in a stepwise fashion from 0.04 to 25.

As a starting point for the numerical simulation, the initial corrosion current density of the corroding steel surface, denoted A, due to microcell action is arbitrarily set at i,,,, = 10 mAm-2 (1.0 yAcm-2) (self-corrosion rate). The corresponding polarisation

Anode

CONCRETE

Fig. 4 Cross-section ofa reinforced concrete compound and modelling using a macrocell specimen with planparallel steel plate electrode arrangement (unit depth: 1000 mm).


lnvestigations on Cathodic Control of Ckloride-lnduced Reinforcement Corrosion 19

behaviour is characterised by anodic and cathodic Tafel constants, b, = 91 mV/ decade and b, = 176 mV/decade, respectively, resulting in a Stern-Geary constant B = 26 mV

For the passive component, denoted P, the initial corrosion current density is set at i,,,, = 0.1 mAm-2 (0.01 pA cm-2), The polarisation behaviour is described by b, = 1 mV / decade and b, = 176 mV / decade, reflecting ideal passive behaviour (corrosion current density remains at a constant negligible value).

For both electrodes the influence of limited oxygen diffusion on the cathodic processes is not taken into account since this situation is considered to be of practical relevance only for prolonged submerged conditions [3]. Throughout the numerical simulations the driving voltage between the cathodic and anodic component is fixed at U,, = 500 mV. The influence of the electrolytic resistivity of the concrete body of the specimen, p,,,, on macrocell action was investigated for p,,, = 50, 100 and 500 Qm (see Fig. 6).

The electrochemical behaviour of the steel / concrete interface is modelled by discrete equidistant interface elements of 2 mm width (1000 mm depth). The electrolytic properties of the concrete material are modelled by pure resistive elements each representing a square cross section of 2 x 2 mm2 (1000 mm depth). For practical reasons the numerical analysis is used to calculate the electrochemical potential and current distribution in a two-dimensional situation under steady state conditions. The macrocell current, I , is measured by a so-called Zero Resistance Ammeter (ZRA), inserted in the externarbart of the equivalent electrical circuit. Since the method is capable of calculating the anodic and cathodic currents, the accelerating effect of macrocell action on the corrosion rate can be quantified.

The control exerted by the anodic, electrolytic (concrete) and cathodic processes on the electrochemical behaviour of the macrocell is defined according to eqn (2). The voltages AU, and AU, correspond to the shift in potential of the anodic and cathodic component, respectively, relative to their respective initial free corrosion potential. The voltage U,,, refers to the voltage drop over the concrete electrolyte induced by the flow of the macrocell current between the anodic and cathodic component.

In Fig. 5(a) and 5(b) the development of the control is depicted as a function of surface area ratio PIA for p,,, = 50 and 100 Qm, respectively. As the ratio PIA increases from 0.04 to 25 the cathodic control decreases gradually from 93% to 61% (p,,, = 50 S2m) and from 90 to 56% (100 Qm). Simultaneously, the anodic and electrolytic control increase from 1 and 6% to 20 and 19% respectively (p,,, = 50 Qm), and from 1 and 10% to 18 and 26%, respectively (p,,, = 100 Qm).

This demonstrates that the influence of the cathodic process on macrocell behaviour can be very pronounced. Especially for small ratio values the cathodic process forms the major portion of the total macrocell circuit resistance. However, it should be borne in mind that the electrolyte resistance is influenced by the PIA ratio and the geometrical constraints of the selected specimen.

The accelerating effect of macrocell action on the corrosion rate of the anodic component is demonstrated in Fig. 6 for p,,, = 50, 100 and 500 Qm. The corrosion current density, i,, including macrocell effects, is given as a function of surface area ratio PIA. As the ratio PIA increases from 0.04 to 25 the corrosion rate is increased considerably by 11.8 to 1241% for p,,, = 50 Qm, by 9.3 to 864% for p,,, = 100 Qm and

121 *



W -

Surface area ratio PIA (-)

Fig. 5(a) Development of control exerted by macrocell components as afunction of surface area ratio P / A: p,,, = 50 am.

Fig. 5(b ) Development of control exerted by macrocell components as afunction of surface area ratio P / A: p,,, = 100 Rm.

by 4.2 to 291% for p,,, = 500 Qm. Generally, the galvanic current, lgar underestimates the actual corrosion rate, but for high values of P/A this may give reliable information on the magnitude of the corrosion process of the anodic component.

In addition, active-passive macrocells normally result in a strong cathodic polarisation of the passive steel. In Fig. 7 this effect is illustrated for P/A = 25 (L , = 100 mm and LA = 4 mm) by the equicontour plot of potentials in the concrete between anode and cathode.


Investigations on Cathodic Control of Chloride-Induced Reinforcement Corrosion 21

1 50

c s 125

Ea 75 c- 50 8 25 ii

.- u)

UT I00

2;

O O

o m .- .-

0 5 10 15 20 25 Surface area ratio P/A (-)

Fig. 6 Development of corrosion current density of anodic component as a function of surface area ratio P 1 A.

5. Discussion and Concluding Remarks

It is generally accepted that macrocell action plays an important role in the corrosion process of reinforcement steel embedded in concrete. Normally, oxygen diffusion and concrete resistance are considered to dominate the magnitude of macrocell action under most practical conditions. However, this view is based on implicit assumptions rather than objective quantitative information. Hence, there is a potential need for research addressing this subject.

200

E .w I 00 s e

50

0

Surface area ratio PIA (-)

Fig. 7 Equipotential plot for PIA = 25 (for explanations, see text).



Based on the electrochemical nature of reinforcement corrosion, the behaviour of corroding and passive steel can be described by relationships between current and potential. These mathematical expressions allow a distinction to be made between activation (charge transfer) and oxygen diffusion controlled processes. For most practical conditions, oxygen diffusion is not considered to be of importance except when exposed to continuous or prolonged water immersion. This finding is supported by the results presented in [6].

For a macrocell the total circuit resistance can be regarded as a connection of three resistances in series, viz. a resistance associated with the anodic, electrolytic, and cathodic components, respectively. However, most practical situations will not allow a quantitative distinction between these resistances to be made. Hence, the relative importance of these components on the overall macrocell behaviour cannot be easily evaluated. This problem can be overcome by a numerical simulation of the relevant electrode and electrolytic processes.

The current-potential relationship for the cathodic and anodic processes can be simulated by introducing discrete non-linear resistive elements for the steel I concrete interface. These elements represent the resistance to polarisation of passive and corroding steel, respectively. The ionic pathway between the anodic and the cathodic component of a planparallel macrocell arrangement can be modelled by using pure resistive elements. The numerical method is applied to simulate a situation known to be occurring in practice with distinct spatial separation between corroding and passivated reinforcement steel.

The surface area ratio between passive and corroding steel, PIA, is varied in a stepwise fashion intended to cover the range of practical importance for two values of concrete resistivity (50 and 100 Qm).

The results of the numerical simulation clearly demonstrate that for small values of PIA the resistance to polarisation of the passive steel is the main contributor to the overall macrocell resistance. This can be explained by the fact that in contrast to actively corroding steel passive steel exhibits a high resistance to polarisation. Considering equal surface area, passive and corroding steel correspond to resistances of 765 and 2.6 Qm2 respectively (at the free corrosion potential). These values only apply to the specific values of the electrochemical factors used in this stimulation. The resistance values decrease as the change in potential increases and as the surface area increases. Therefore the three resistances involved are very much dependent on geometrical factors. As the surface area ratio PIA increases, the resistance to polarisation of the passive steel component decreases and hence it’s relative contribution to the overall macrocell resistance. Nevertheless, for PIA = 25 approximately 60% of the circuit resistance is attributable to the passive steel. However, the contribution of the corroding steel component always remains limited to less than 20%. This situation is largely due to the high initial corrosion current, I,,,,. In general, the resistance to polarisation is inversely proportional to Ice,,.

The results of the numerical simulation also clearly demonstrate that corrosion can be seriously aggravated by the presence of macrocells. According to [7] the corrosion rate is hardly affected by the presence of macrocells. However, this conclusion is only based on laboratory experiments with PIA =1.0. From the present numerical simulation it is clear that the corrosion current density, i,, i.e. the rate of attack, is strongly influenced by the surface area ratio PIA.


lnvesfigafions on Cathodic Control of Chloride-Induced Reinforcement Corrosion

6. Outlook

23

In view of the practical importance of reinforcement corrosion, further experimental research is urgently needed to validate the results obtained by a numerical analysis. Especially, investigations on the effect of cathodic charge transfer at the steel/ concrete interface under different exposure conditions and for a range of concrete qualities may improve the model and its practical use.

References

1. M. Raupach and J. Gulikers, Determination of corrosion rates based on macrocell- and microcell-models - General principles and influencing parameters, in EUROCORR '98, Utrecht, The Netherlands, WP 11,6 pp. 2. C. Alonso, C. Andrade and J. A. Gonzalez, Relation between resistivity and corrosion rate of reinforcement in carbonated mortar made with several cement types, Cem. Concr. Res., 1988,

3. M. Raupach, Investigations on the influence of oxygen on the corrosion of steel in concrete, Mater. Struct., 1996,415,174-184 and 226-232. 4. J. Gulikers and A. De Boer, Numerical simulation of corrosion processes in reinforced concrete, finite elements in engineering and science, in Proc. 2nd Int. Conf., Amsterdam, 4-6 June 1997, pp. 63-72 (M. Hendriks, H. Jongedijk, J. Rots and W. van Spanje, eds). 5. H. Hurling, Oxygen Permeability of Concrete, in Proc. RILEM Seminar on the Durability of Concrete Structures Under Normal Outdoor Exposure, Hannover, 26th-29th March 1984, pp. 91- 101. RILEM; Institut fur Baustoffe und Materialprufing, 1984. 6. S. Jaggi, B. Elsener and H. Bohni, Oxygen reduction on passive steel in alkaline solutions, in EUROCORR '99, European Federation of Corrosion, Aachen, Germany, 7 pp. 7. J. A. Gonzalez, S. Feliu, M. L. Escudero, C. Andrade and A. Macias, Relative influence of galvanic macrocells and local microcells in the corrosion of reinforced concrete structures, in Int. Conj on Measurements and Testing in Civil Engineering, Lyon-Villeurbanne, France, 13-16 September 1988, pp. 237-249.

18,5,687-698.


3 Critical Factors for the Initiation of Rebar

Corrosion

L. ZIMMERMANN, B. ELSENER and H. BOHNI Institute of Materials Chemistry and Corrosion, Swiss Federal Institute of Technology, ETH Honggerberg,

CH-8093 Zurich, Switzerland

ABSTRACT The critical chloride content for the initiation of rebar corrosion has been determined in a laboratory study using a multiple macrocell sample arrangement in order to allow a statistical treatment of the results. In synthetic pore solution (pH 13.4) the probability of corrosion initiation is zero at chloride concentrations < 0.2 mole L-* and reaches 100% at chloride concentrations > 0.6 mole L-I (T = 22°C) at open circuit condition. At lower temperatures (5 and 14°C) the initiation probability is less distributed and the critical chloride content for initiation increases. Experiments using steel embedded in mortar blocks showed that the free chloride content (measured with chloride sensors) necessary for depassivation of the rebars was similar to that found in solution experiments. The calculated Cl-/OH ratio threshold value is ca. 0.6 + 0.1. Differences between open circuit conditions and anodic polarisation are highlighted. The main parameters and their interaction that influence the critical chloride content for the initiation of rebar corrosion and the consequences for the durability of reinforced concrete structures are discussed.

1. Introduction

Corrosion of rebars resulting from the action of chloride ions from de-icing salts or sea water is the main cause of damage and early failure of reinforced concrete structures. The corrosion process can be divided into an initiation period and a propagation period; the initiation period will last as long as the time necessary to reach the critical chloride content for depassivation (localised disruption of the passive film) of the reinforcing steel. This ’critical chloride content’, as is well known from practical experience and the literature [1,2], is not a constant value but depends on the electrochemical conditions (potential, pH and oxygen content) at the steel surface, on the concrete humidity and on the chloride binding in the cement paste. Several difficulties have to be considered when addressing the threshold value:

Only chlorides dissolved in the pore solutions (’free chlorides’) can act as depassivating ions -in experiments performed in mortar or concrete or from field results usually only the total chloride content is known.

Initiation of chloride-induced localised corrosion is a stochastic process and thus a great number of experiments have to be performed in order to get statistically significant results.



Depassivation of the steel is a necessary but not the only condition for corrosion to occur.

Concrete pore solution

rngL-'

The threshold value for initiation of chloride-induced localised corrosion can be determined in solutions [3,4 and literature cited therein] and a critical Cl-/OH- ratio can be calculated. Surface roughness and preparation as well as the influence of prepassivation time of the steel electrodes prior to chloride addition [5] can explain part of the considerable variations in the Cl-/OH- ratio reported in the literature. In mortar or concrete the free chloride content at the steel surface in the pore solution has to be known. Pore solution expression needs a certain mortar volume and the free chloride contents determined are average values that do not take into account chloride profiles. Chloride-sensitive sensor elements [6] are able to measure the free chloride content in s i t u at the same depth as the rebars providing the calibration curve is known.

In this work, results of a laboratory study using a multiple macrocell arrangement in order to allow a statistical treatment of the initiation of localised corrosion are reported for different temperatures and chloride concentrations. The solution experiments are verified by experiments using steel embedded in mortar blocks with, chloride solutions penetrating from the outside. The results are compared with literature data and the main parameters and their interactions that influence the critical chloride content for the initiation of rebar corrosion are discussed.

Ca(OH), KOH Na,SO, NaOH

9.6 13967.8 3121.6 616.8

2. Experimental

Each solution experiment (Fig. 1) was performed with five pairs of anode cathode macrocells (area ratio 1:l). The rebars were sand blasted and rinsed in ethanol. Synthetic pore solution [6] with a pH value of 13.4 was used as electrolyte (Table 1). In the cathode compartment, oxygen was bubbled continuously. The mortar blocks (Fig. 2) were prepared from ordinary Portland cement (OPC) with a water / cement (w/c) ratio of 0.6, a sand/cement ratio of 3 [6]. After casting and demoulding the blocks were cured for 28 days in the humidity cabinet and afterwards dried for 10 days at 50°C. The total porosity determined from mercury intrusion porosimetry (MIP) was 17 k 1%. Each mortar block contained five rebars (cover 30 mm) and 4 chloride-sensitive sensor elements at a depth of 30 mm [6]. During the chloride uptake experiments (600 h) the mortar blocks were immersed 3 mm in a NaCl solution. The potential of the chloride sensors and of the rebars was measured continuously versus a calomel electrode (SCE) put into the NaCl solution. Potentials of the steel bars, macrocell currents and the readings of the chloride sensors were registered with a programmable high impedance voltmeter (Keithley 2001, switch system 7001) equipped with zero-resistance ammeter.


Critical Factorsfor the lnifiation of Rebar Corrosion 27

Fig. 1 Experimental setup of the macrocell experiments with f ive rebar pairs short circuited by zero resistance ammeters in solution. Anode and cathode chamber are connected by porous glass diaphragms.

230 mm 1

rebar sensor 1 epoxy-coating

Fig. 2 Experimental setup o f tke initiation experiments in mortar blocks.



3. Results

-0.1 *

E -0.2- 0

5 -0.3-1

' -0.4-

m E Q)

.- c c

An example of half cell potential and macrocell current versus time of five rebar pairs (A-E) in solution is given in Fig. 3. In a first phase (two hours) the steels were passivated in chloride-free, oxygenated pore solution and the open circuit potential reached cu. -0.15 V SCE. Chlorides were then added to the solution in the anode compartment and after variable initiation times the half cell potential dropped and simultaneously the macrocell current increased (Fig. 3). The probability of corrosion initiation, P, is calculated from the percentage of samples that showed a drop in potential within 24 h; in Fig. 3 two of five samples became active thus P = 40%.

I I

Initiated

1 Chloride addition

-0.5 j j 0 5 10 15 20

0.30

Z 0.25 - E v

a 0.15- -

0 5 10 15 20 Time (h)

Fig. 3 Potential us time and macrocell current vs time for five rebars (A-E) of a macrocell experiment in solution. Chloride concentration 0.3 mol L-I, pH 13.4.


Critical Factors for the Initiation of Rebar Corrosion 29

Looking at the distribution of initiation times (Fig. 4) only a very small fraction of samples will initiate at t > 24 h, thus the percentages of initiation are representative.

The probability of corrosion initiation in solution for different chloride concentrations and temperatures is summarised in Fig. 5. At room temperature (22°C) no initiation occurs at chloride concentrations < 0.2, above 0.55 molL-l the initiation probability is 100%. At lower temperatures the minimum chloride concentration to get initiation increases to 0.3 molL-l(l4"C) and 0.45 molL-l at 5"C, the value of 100% initiation remaining practically constant.

Half cell potentials of the embedded rebars and the free chloride content of the embedded chloride sensors in mortar (w/c 0.6) during the experiment of chloride penetration from the surface are shown in Fig. 6. Potentials of the steel bars embedded in mortar prior to depassivation were -0.12 V SCE, slightly more positive then in the solution experiments. The probability for initiation of chloride-induced corrosion of steel embedded in mortar was calculated in the same way as for the solution experiments, the results are shown in Fig. 7. At the time of depassivation (determined from the drop of the half cell potential of the steel), the corresponding free chloride content was measured by the chloride sensors embedded at the same depth as the steel [6]. The minimum chloride concentration for corrosion initiation of steel in mortar is about two times lower than that found in solutions (Fig. 7).

1 00

80

60

40

20

0 0 5 10 15 20 25 30 35 40

Time (h)

Fig. 4 Cumulative frequency distribution of initiation times of the macrocell experiments in solution at different temperatures. Chloride concentration 0.5 mol L-l.



100

80

60

40

20

0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Chloride concentration (mol L-l)

Fig. 5 Probability of corrosion initiation for different temperatures and chloride contents in synthetic pore solutions (pH 23.4).

0.00

-0.05 h n w

-0.15 W - 2 -0.20 c Q c 8 -0.25 2 d -0.30 a 0)

-0.35

-0.40

.......................... ..........................

................... ....................

0.5

0.4

0.3

0.2

0.1

0 0 100 200 300 400 500

Time (h)

Fig. 6 Potential of the embedded steels andfree chloride concentration determined with chloride sensors vs time in a chloride uptake experiment in mortar (wlc 0.6).

n r

i

I u


100-

8 0 n

5 * 0

3 CT

a 0

60

2

> 40 * .- CI - z

20

0

4. Discussion

i A i

4 ...: ................................... +. ................................

’ ’ ’ I * ‘ ‘ ’ ‘

!

‘ ’ * 1 ‘ * ’ -

m : I

i i

j

_ ................................ 4 ........................... d..,....q I

I : , : I

1 ; m : I _ ................................ ; .............................. + ................................... ; ................................. A i

; l + l f I i

! I _ ........................... ..... 4.: ....... !.# ............... 4 ................................... ; ................................... t ................................ -

i M 1

I A : I

........... ........... ................................... ............... .......... + A + ;

j - * - 1

i f -.. .............................. & ............................... t”””””””””” ....... .................................... ............................

On the basis of the concept of concurrent adsorption of protective hydroxide and aggressive chloride ions [7] involved in pit initiation, the ratio Cl-/OH- can be used as a characteristic value to describe the probability of pit initiation. A plot of the probability of corrosion initiation under open circuit condition versus the Cl-/ OH- ratio (Fig. 8) for solutions with different pH and chloride concentration (T = 22 T) shows that all experiments can be grouped along a single line: at Cl-lOH- ratios < 0.7 no initiation is possible, at Cl-/OH- ratios > 1.7 initiation of localised corrosion always occurs. The results are in good agreement with the early data reported by Hausmann [B]; for solutions with pH 12.4 he reported a minimum value of Cl-/OH- = 0.6, with increasing initiation probability at higher chloride contents. It is important to note that both sets of results were obtained under open circuit conditions. Based on measurements of the polarisation resistance at open circuit conditions a C1-/ OH- ratio in the range from 0.3-0.8 was found in the pH range from 11.6-13.4 [9 ] .

The possibility to measure experimentally the free chloride concentration in the pore solution at the same depth as the rebars and at the time of depassivation (Fig. 6) [6] allows pit initiation probabilities for steel embedded in mortar to be determined. The concentration of free chlorides that can initiate pitting of steel in mortar is about two times lower than in alkaline solutions.



100

0

0 0.5 1 1.5 2 2.5 [CI-] [OH-] ratio

Fig. 8 Probability of corrosion initiation in pore solutions (pH 13.2-13.5) with different chloride contents us CI-IOH- ratio.

Frequently, experiments are not performed at open circuit conditions but with potentiodynamic anodic polarisation. The pitting potential, Epit, determined in this way as a function of chloride content characterises the susceptibility of metals and alloys to pitting. These results allow a ranking of the resistance to pitting (higher pitting potentials at the same chloride content indicate higher resistance against pitting) but provide no direct information on the occurrence of pitting either in laboratory experiments or in the field. The condition for the occurrence of pit initiation can be written as

A 0 this work. of Hausmann 151, pH 12.4.

E c o r r > ‘ p i t

i.e. the open circuit potential of the passive steel has to be more positive then the pitting potential. Thus, both the pitting Epii and potential the open circuit or corrosion potential, Ecorr influence the occurrence of pitting: lower pitting potentials and more positive open circuit potentials favour the probability of pit initiation.

The open circuit or corrosion potential of passive steel in alkaline solutions or in cement based materials decreases with increasing pH of the pore solution and with lowering of the oxygen content. More positive corrosion potentials, as can occur when silica fume or flyash is added to the cement or on aerated structures, induce, according to eqn (l), more severe conditions for pit initiation. As a consequence the chloride content necessary to initiate pitting decreases [ 101. Potentiostatic polarisation at +0.25 V SCE [4] or potentiodynamic anodic polarisation increases the potential of


Critical Factors for the Initiation of Xebar Corrosion 33

the steel far beyond the region usually observed in open circuit conditions or on structures (-0.2-0 V SCE). The 'critical chloride concentrations' determined in this way will be much lower than those resulting from experiments at open circuit conditions.

5. Conclusions

From the results of this electrochemical study in alkaline solutions or mortar blocks it can be concluded

the minimum value of free chlorides necessary for depassivation of steel in alkaline solutions or mortar can be estimated from the ratio Cl-/OH- to be cu. 0.6-0.8 in the pH range of 13-13.8;

pit initiation is a stochastic phenomenon, the probability of pit initiation increases with increasing concentration of free chlorides; and

the condition for the occurrence of pitting in real situations is E,,,, > Epit thus both parameters, corrosion potential and pitting potential, influence the occurrence of pitting.

6. Acknowledgements

The authors acknowledge financial support of this study by the Research Foundation of the Swiss Cement and Lime Producers (VSZKG).

References

1. L. Zimmermann, Korrosionsinitiierender Chloridgehalt von Stahl in Beton, PhD thesis ETH Zurich, submitted. 2. W. Lukas, Betonwerk und Fertigteil-Technik, 1985, 51, 730-734. 3. G. K. Glass and N. R. Buenfeld, Chloride threshold levels for corrosion induced deterioration of steel in concrete, in Chloride Penetration into Concrete, Proc. RILEM Int. Workshop (L.O. Nilsson and J.P. Olivier, eds). RILEM, 1997, pp.429441. 4. W. Breit, Mater. Corros., 1998,49, 539-550. 5. S. Jaggi, B. Elsener and H. Bohni, this volume, pp.3-12. 6. B. Elsener, L. Zimmermann, D. Fliickiger, D. Burchler and H. Bohni, Chloride penetration - Non-destructive determination of the free chloride content in mortar and concrete, in Chloride Penetration into Concrete, Proc. RILEM Int. Workshop (L.O. Nilsson and J.P. Olivier, eds). RILEM,

7 . S. Matsuda and H.H. Uhlig, J. Electrochem. SOC. ,1964,111, 156. 8. D.A. Hausmann, Mater. Prot., 1967, 6, 19-23. 9. S. Goni and C. Andrade, Cem. Concr. Res., 1990,20,525-539. 10. K. Byfors, Cem. Concr. Res., 1987,17, 115.

1997, pp.17-26.


4 Field Tests of Chloride Penetration into Concrete

with Microsilica

0. VENNESLAND and J. HAVDAHL* Department of Structural Engineering, Norwegian University of Science and Technology. Rich,

Birkelands vei la, N-7034, Norway *SINTEF Building and Environmental Technology

ABSTRACT Ten concrete bocks 150 x 150 x 150 cm of five different concrete qualities were placed in the tidal zone in 0stmarkneset in the Trondheimsfjord, Norway, in March 1983. The concretes were strength classes C35, C65 with 10 and 20% microsilica. All blocks had reinforcing nets with 30 mm and 50 mm cover.

In addition to visual inspection the following measurements were made: surface potentials, concrete resistivity, chloride profiles and accelerated chloride penetration test.

From one of the blocks of each concrete quality the chloride profiles were determined both from the side facing land (south) and from the side facing the sea (north).

The amount of penetrated chlorides is relatively identical for all types of concrete. The C65 concrete with 10% microsilica contains most chlorides, 380 gm-2 surface, while the C35 concrete with 20% microsilica shows the least chloride penetration, 217.8 gm-2 surface.

The penetrated amount is about the same from the land side as from the sea side (exception for C35 concrete with 20% microsilica shows twice as much penetration from the sea side).

The shapes of the profiles are very different for the types of concrete. Dense concretes show steep profiles with a high chloride content at the surface and limited penetration while more open concrete shows flat profiles with less chloride in the surface but deeper Penetration.

In the accelerated testing C35 shows a marked effect of the microsilica content on the chloride penetration, as expected. This is reasonable and reflected also in the field result for one of the specimens without microsilica, while the other specimen without microsilica shows a remarkably low chloride penetration compared to the specimens containing microsilica.

1. Introduction

In 1982 Elkem Materials initiated construction of 10 concrete blocks 150 x 150 x 50 cm of five different concrete qualities to be placed (March 1983) in the tidal zone at Ostmarkneset in the Trondheimsqord.

The concretes were strength class C35 with 0, 10 and 20% microsilica and C65 with 10 and 20 % microsilica. At the time of construction there was a lot of discussion in Norway about the effect of microsilica on the corrosion protective properties of



concrete, especially on the chloride penetration into concrete. All blocks (two blocks of each quality) had reinforcing nets with 30 mm and 50 mm cover. The nets had external electrical connections. The set-up and material data have been reported earlier. The investigation presented in this report was made September 1997. In the evaluation of the condition much weight has been given to the developments since the earlier investigations.

2. Test Programme

In addition to visual inspection the following measurements have been made: chloride profiles and accelerated chloride penetration, surface potentials and concrete resistivity. The chloride profiles was determined at drilled cores by 2 mm slices. The accelerated test (submerged test according to NT Build 443) where made at the inner part of the core. The surface potentials (corrosion potential of the embedded nets) were measured using a copper / copper sulfate reference electrode. Resistivity was measured by the Wenner method with four drilled-in bolts at 100 mm separations and a 50 mm depth.

3. Results and Discussion

Algae and shells covered all blocks from top to bottom but little physical damage were observed, except for some frost damages at the corners. Most of the electrical connections were destroyed, however, and were renewed.

From one of the blocks of each concrete quality the chloride profiles were determined both from the side facing land (south) and from the side facing sea (north). All profiles are presented in Fig. 1. The amounts of chlorides penetrated are also presented as grams of chlorides per m2 concrete area (the area of the chloride profile) in Table 1.

As shown in the Figure the shapes of the profiles are very different for the types of concretes. Dense concretes show steep profiles with a high chloride content at the surface and limited penetration while more open concrete shows flat profiles with less chloride in the surface but deeper penetration. The extremes are - naturally - C35 without microsilica as the less dense concrete and C65 with 20% microsilica as the densest concretes.

The penetrated amounts of chloride are, however, relatively identical for all types of concrete. In fact, the C65 concrete with 10% microsilica contains most chloride, 380 gm-2 surface, while the C35 concrete with 20% microsilica shows least chloride penetration, 217.8 gm-2 surface. The penetrated amount is about the same from the land side as from sea side (exception for C35 without microsilica, which shows twice as much penetration from the sea side).

Two cores of each concrete quality were subjected to immersion testing of chloride penetration according to NT Build 443. The results are shown in Table 2.

At two points an important difference between the field results and accelerated laboratory testing is observed. In the accelerated testing the C35 shows a marked effect of the microsilica content on the chloride penetration - as expected - and as


Tests of Concrete with Microsilica 37

Chloride penetration 1

0.9

0.8 +C35 10% MS-Ian E u C 3 5 10% MS-se 0.7 s *C35 20% MS-Ian -c C35 20% MS-se 5 0.6

8 + C65 10% MS-Ian ,- O S

Q -C35 10% MS-se E 0.4 -c C65 20% MS-Ian -4C65 20% MS-se Q) 0.3

E

h Q

0

v

C

s - b 0.2

6 0.1

0 0 2 4 6 8 10 12

Penetration depth (cm)

Fig. 1 Chloride profiles (MS = microsilica).

Concrete quality

C35 no microsilica

C35 10% microsilica

C35 20% microsilica

Table 1. Penetrated amount of chlorides

Direction Penetrated amount (g C1-/m-2) Individual Mean

Facing land 192.9 Facing sea 386.2 289.6



C65 10% microsilica

C65 20% microsilica



shown in Fig. 2. This is reflected also in the field result for one of the specimens without microsilica, while the other specimen without microsilica shows a remarkably low chloride penetration compared to the specimens containing microsilica.


38

Table 2. Resultsfrom immersion chloride penetration test

Corrosion of Reinforcement in Concrete: Corrosion Mechanisms and Corrosion Protection

Concrete quality

C35 no microsilica

C35 10% microsilica

C35 20% microsilica

C65 10% microsilica

C65 20% microsilica

Surface concentration Diffusion constant Penetrated amount (% of concrete) m2 s-l) (g m-?

0.90 28.85 240.35

1.05 6.02 144.2

0.85 7.07 139.7

0.948 4.04 113.3

0.998 2.48 95.7

The second deviation between field and laboratory data is for the C65 concrete with 10% microsilica. While the field data show high chloride contents in the surface area, this is not observed in the laboratory testing.

The surface potential values are compared to corresponding values from the 1992 survey. The potential values of 1997 are definitely more negative than the values of 1992. While the most negative mean value in 1992 was -349 mV and the least negative mean value 4 0 mV the corresponding results in 1997 were -666 mV as the most negative and -109 mV as the least negative referred to the Cu / CuSO, electrode.

It is worth noting that the most negative result in 1992 was measured for the presumably densest concrete (C65 with 20% microsilica) and the least negative result was measured for the presumably least dense concrete (C35 without microsilica). This illustrates how difficult it is to draw conclusions with respect to state of corrosion based only on potential measurements. The potential values might be very low but

n 1.4 Q, 4- 6 1.2 c 8 1.0

8 0.8

+ 0

W

CI C 0.6 Q, c 0 0.4 0 Q,

4-

E 0.2

- 5 0.0

c I I I

9 I I

I i I t C 3 5 no MS-2

-a- C35 no MS-1 I

I I

, I tC3510%MS-l 1 +C35 10% MS-2

~

t 5 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Distance from surface (cm) Fig. 2 Accelerated chloride penetration into C35 class concrete without and with 10% microsilica.


Tests of Concrete with Microsilica 39

the chloride content is almost negligible and corrosion cannot have been initiated. In other cases the chloride content is high without causing a negative potential. The resistivity values measured in 1992 and 1997 are presented in Table 4. For all

concretes except C35 without microsilica there is a considerable increase in the resistivity from 1992 to 1997 with a 30% increase as typical.

Table 3. Mean values f o r surface potentials (mV vs Cu/CuSO,) in 1992 and 1997

Table 4. Resistivity measured in 1992 and 1997 (.Q m)

Land side (South) 30 mm cover

Sea side (North) 50 mm cover



4. Conclusions

The field data and laboratory data on chloride penetration do not correspond fully. While the laboratory data show a definite effect of microsilica in reducing the chloride penetration the field data are not so distinctive. They show a small effect of MS on the amount of penetrated chlorides but a very clear effect on mechanism of penetration. The MS containing concretes have a steeper profile with higher chloride content in the surface and smaller penetration depth. This effect was most marked with 20 % MS.

References

1. M. Magne and M. Sandvik, Concrete blocks for long time exposure. SINTEF Report STF65 A84030, ISBN 82-595-3605-6, Trondheim, 1984 (in Norwegian). 2. T. A. Hammer and J. Havdahl, Concrete blocks for long time exposure - 1.5 years exposure time. SINTEF report STF65 A86003 1986-07-10), ISBN 82-5954073-8, Trondheim, 1986 (in Norwegian). 3. T. A. Hammer, J. Havdahl and I. Meland, Concrete blocks for long time exposure - 5 years exposure time. SINTEF report STF65 A90010, ISBN 82-595-58084, Trondheim, 1991(in Norwegian). 4 .0 . Gautefall, Experiences from 9 years exposure of concrete in the tidal zone. SINTEF report STF70 A92190, ISBN 82-595-7517-5, Trondheim, 1992 (in Norwegian). 5. ASTM C 876-87, Standard Test Method for Half-Cell Potentials of Uncoated Reinforcing Steel in Concrete. 6. 0. Vennesland, Potential measurements on normal density and light weight concrete. Report 3.2 in the Lightcon project, 1997 (in Norwegian). 7. P. Langford and J. Broomfield, Monitoring the corrosion of reinforcing steel, Construction Repair, 1987, 5, 32-36.


5 Comparison of Electrochemical Data and Mass Loss Corrosion Rate Measurements for Steel

Reinforcement in Concrete

I?. NOVAK and R. MALA Institute of Chemical Technology, Department of Metals and Corrosion Engineering, CZ-16628 Prague,

Czech Republic

ABSTRACT The average corrosion rates of steel reinforcement with different initial surface conditions (u,,,,) calculated from mass loss after four-year exposure in_ concrete were evaluated in relation to the average free corrosion potential ( E m , , ) and polarisation resistance ( xp). The corrosion of pre-rusted steel in concrete co_uld not be evaluated reliably based on the E,,,, value. Qualitative correlation of Emrv and u,,,, was found only for machined and scaled surfaces. xp values were in good qualitative agreement with u,,,, for all the tested surface conditions. At polarisation resistances above 30 Qm2, ucorr values were below 2 pm/year.

1. Introduction

The basic non-destructive qualitative information on the corrosion activity of steel reinforcement in concrete is usually gained by measurements of free corrosion potentials [ 1/21. Another frequently applied method is the measurement of polarisation resistance, which theoretically provides quantitative data on the corrosion rate. However, steel reinforcement in concrete does not allow for a clear quantitative relation between the corrosion rate and polarisation resistance of steel. In spite of objections [ 3 ] , various modifications of the polarisation resistance measurement are applied for reinforced concrete both on the laboratory scale and in practice [4-81. To calculate the corrosion current of steel from the polarisation resistance value, an inverse proportionality constant B is usually applied. Andrade and Gonzales [9,10] determined this constant empirically after short-term exposure of steel with a practically unrepresentative clean metal surface in model solutions or in aggressive concrete as 26 or 52 mV respectively.

The objective of our study was to evaluate the free corrosion potential and polarisation resistance of steel reinforcement with different initial surface conditions in relation to the average corrosion rate determined from mass losses following a long-term exposure in concrete.



2. Experimental

Four steel rods 10 mm in diameter and 100 mm in length were embedded in concrete blocks (230 x 120 x 50 mm) with a 20 mm covering layer [ll]. The exposed surface area of each steel specimen was approximately 30 cm2. The mortar was made with 0.58 water to cement (OPC) ratio and sand-to-cement ratio was 3.8. Various chloride (NaC1) contents (0.01, 0.4 and 2.5 mass% /cement) were introduced in concrete blocks, by immersion, after the curing time. The blocks were then dried to an equilibrium water content corresponding to the chosen humidity level. Each concrete block contained two rods of scaled and machined carbon steel CSN (Czech standard) 10335 (high C - 0.26 mass%) and two rods of pre-rusted and machined carbon steel CSN 10216 (low C - 0.06 mass%). The scaled surface was in the state as received from the supplier. The pre-rusted surface was that obtained after a half-year exposure in an outdoor atmosphere (C4 according to IS0 9223). The average thickness of the rust on the pre-rusted bars was approximately 50 pm, and that of scale (Fe,O,) on the scaled bars was between 15 and 20 pm. From the analysis of rust it was found that there were no chlorides, although sulfates were present. Scale was regular and adherent over the whole surface. The pre- rusted surface was covered with a partially non-adherent brown layer on the black base. The machined surfaces were prepared by machining on a lathe. The concrete blocks were placed in glass cabinets in an atmosphere of constant relative humidity (60, 70, SO, 90,100%). The equilibrium content of water in the concrete (checked by weighing) was reached after four months of exposure. Two parallel identical concrete blocks were exposed for every humidity level 70, 80, 90, 100% and chloride content (0.4 and 2.5 mass % /cement) although only one set was exposed for RH 60% and 0.0 mass % C1-/ cement. Altogether, 23 concrete blocks were exposed with 92 steel specimens (of which 2 x 36 were in parallel sets). Temperature during exposure was in the range 12 to 26°C. The linear polarisation technique was used to measure polarisation resistance (Rp) and the free corrosion potential (E,,,,) of all steel specimens during exposure. Corrosion measurement systems (CMS 352 /273 EG&G or CMS 100 / PC3 Gamry) were used for the electrochemical measurement and data analysis. Platinum wire was used as a counter electrode, placed parallel to the measured steel specimen and on the opposite side of the concrete block, which was in contact with a saturated calomel electrode (SCE). The potential was changed +lo mV vs E,,, at a scan rate of 0.1 mV s-l.

The exposure of 3 of the concrete blocks (4 x 14 steel specimens) was terminated in the 42nd and the remaining 11 in the 48th month. The blocks were then broken and the steel specimens taken out. The corrosion was evaluated visually and further evaluation was based on the mass loss. In order to determine mass changes, the specimens were pickled according to ASTM G1 (C.3.3). In order to determine the original metal mass of the scaled or pre-rusted specimens, mass corrections to the original oxide layer were made on a series of non-exposed original reinforcement specimens. Given the variance of values, these corrections lowered the accuracy of the corrosion rate determination in pre-rusted specimens to ~ 0 . 5 pm/year. In the case of machined and scaled specimens, however, mass changes allowed us to determine the average corrosion rate after four- year exposure ( ~ c o r r ) with an accuracy of <0.1 pm/year. After breaking the concrete blocks the depth of carbonation using phenolphthalein was also determined. The depth of carbonation was less than one half of the cover layer thickness (20 mm) over the four years of exposure.


Comparison ofElectrochemica1 and Mass Loss Data for Steel Reinforcement in Concrete

3. Results and discussion

43

70 80 90 100

70 80 90 100

70 80 90 100

Mass loss results, visual observations and long-term electrochemical measurements for different initial surface conditions of steel reinforcement in concrete were given in a previous paper [12]. For the purpose of comparison of electrochemical data and mass loss corrosion rate measurement averages ( xP, Ecorr ) were calculated from the RP and EfOrr values for the whole period of the exposition of steel specimens in concrete blocks. The typical time course is evident from Fig. 1 for RH 80%; similar time courses were obtained for all humidities tested and both parallel sets of specimens. Reproducibility of xP and Ecorr values for parallel specimens was high (Table 1). The time dependence of E,,,, showed a moderate increase for all humidities at chloride contents of 0.0 and 0.4% C1-/cement, regardless of the initial state of the surface. The effect of the chloride content of the concrete on E,,,, was seen on specimens with all initial conditions, with the free corrosion potential becoming more negative with increasing chloride content. At lower chloride contents (0.0-0.40/0 / cement), the free corrosion potential reached values which corresponded to the redox potential of an inert (Pt) electrode. The comparison of EcOrr for machined and scaled specimens for the whole period of the exposure with the corresponding corrosion rates calculated from mass loss uc0,, is shown in Fig. 2(a). This makes clear that the definition of the corrosion probability according to the ASTM C 876 standard holds for the machined and scaled steel surface [l], i.e. that the potentials between -125 mV(SCE) and -275 mV(SCE) partition the range of a high and a low corrosion rate. The value of 4 ym/year was set as the limit of corrosion rate acceptability in

1st set 2nd set 1st set

6 - -

4 - +0.011 2 - +0.013 2 5 -0.030

5 11 +0.005 4 7 -0.008 5 4 -0.155 2 3 -0.187

5 4 -0.293 4 3 -0.259 2 - -0.499 1 3 -0.262

Table 1. Average values of polarisation resistance E, and free corrosion potential Ecorr for steel specimens with pre-rusted surface after four-year exposure in concrete at RH > 60%. The 1st set was used for mass-loss measurement, the 2nd set is still exposed

c1- (%hem.)

< 0.01

0.4

2.5

RH (% ) I 2nd set

+0.051 - -

-0.046

+0.009 -0.019 -0.086 -0.125

-0.133 -0.322 -

-0.318


1000

100

T si rr"

10

I

Lmachined 1

I

2oo

0 z w 0 8 >-200 E i I

w -400

[ K z q -600

0 10 20 30 40 50 months

0 10 20 30 40 50 m o n t h s 0 10 20 30 40 50 months

Fig. 1 Time dependence of polarisation resistance R andfree corrosion potential Ecorr for steel wi th different surface states in concrete (0.0; 0.4; 2.5% Cl-/cement) during atmospheric exposure at RH 88%.


Comparison of Electrochemical and Mass Loss Data for Steel Reinforcement in Concrete 45

terms of technical requirements. In cases where the corrosion rate did not exceed 1 pm/ year, free corrosion potentials were always more positive than -275 mV(SCE) for both the scaled and machined steel surfaces.

However, this conclusion did not apply to the pre-rusted steel surface (Fig. 2b) because even at corrosion rates above 1 pm/ year, Ecorr values were often found in the range of low probability of corrosion (according to ASTM). The free corrosion potential of the pre-rusted steel reinforcement reached values, which did not correspond to the extent of the corrosion. With low aggressiveness of the pore solution in concrete (0.0 or 0.4% / C1-/ cement), the measured potential seemed to reflect the oxidation conditions on the rust surface rather than the corrosion process at the metal-rust interphase itself. The corrosion process probably proceeds under rust in occluded solution formed during atmospheric exposure before embedding of the reinforcement (sulfates, low pH), without any significant effect of the high alkalinity of the pore solution in concrete [12].

The typical time dependence of R values (for RH 80%) is shown in Fig.1. In contrast to the Ecorr time dependence, a significant effect of the initial superficial state of steel was evident in this case. Polarisation resistance of machined specimens in concrete with a low chloride content (0.0 and 0.4% /cement) was high for all humidities during the whole period of exposure. Polarisation resistance values of scaled specimens were a little lower under the same conditions. For pre-rusted specimens, R, values were always low, regardless of the chloride content (Table 1).

Values of E , were related to vcorr values in a similar way as were free corrosion potentials. Figure 3 shows for all initial surface conditions of steel the corrosion rate calculated from mass loss was lower than 1-2 pm/year at E , > 30 Rm2 and higher than 1-2 pm/year at E p -= 30 am2.

The B constants were evaluated from a set of 56 E , values for different initial surface conditions of steel and for different chloride contents and humidities in concrete. The empirically evaluated values were in a broad interval from 3 to 115 mV (for Fe2+), or 5 to 173 mV (for Fe3+). No dependence of B constant values on the concrete humidity was found; 89.3% of B values were found in the interval

P

0.1 1 2 10 100 Vcorr [ p d ~ e a rl

Fig. 2 (a, b ) Correlation between average p e e corrosion potential Ecorr and mass loss corrosion rate vcorr of machined and scaled steel (a), and of pre-rusted steel (b) in concrete.



1

1000

100 N-

E c - 30

I ar" 10

I

. . p= 0 "

0 I \

I- - ? .

0 scaled

\ A O \ \

0.1

:0,01V \\ 100

Fig. 3 Correlation between average polarisation resistance E, and mass loss corrosion rate v,,,, ofmachined, scaled and pre-rusted steel in concrete (z is oxidation state ofiron).

5 to 65 mV (for Fe2+). Average values of the B constant for different groups of behaviour are given in Table 2.

For the machined surface, 82% of the B constant values were 2 10 mV. Low values of the B constant (2 10 mV) prevailed (in 83%) of surfaces with an oxide layer (scale, rust) in concrete with a low chloride content (0.0 and 0.4 mass % / cement), while B values above 25 mV prevailed (in 80%) in the case of concrete with a high chloride content for all surface conditions.

4. Conclusions

Free corrosion potential values cannot be used for the evaluation of the corrosion activity of steel reinforcement with a pre-rusted surface in non-carbonated concrete (with various water and chloride contents). An acceptable qualitative agreement of the free corrosion potential with the corrosion rate calculated from mass loss was found after long-term exposure only for machined surfaces and scaled surfaces. In those cases, the free corrosion potential was always more positive than -275 mV(SCE) at an average corrosion rate lower than 1 ymlyear.

The average polarisation resistance values for all the tested surface conditions (machined, scaled and pre-rusted) were in good qualitative agreement with the corrosion rates determined from mass loss. Our results show that for polarisation resistances higher than 30 Q m2, the average corrosion rate of steel was lower


Comparison of Electrochemical and Mass Loss Data for Steel Reinforcement in Concrete 47

Table 2. Average values of the B constant for steel in concrete at RH > 60%

c1- (%/cement)

2.5

Initial surface condition

- State B

(mV)

Active 44 Machined, scaled, pre-rusted

Machined

Scaled

0.0; 0.4 Passive 25 15

0.0; 0.4 Passive 9 4

Standard deviation (mV)

29

Pre-rusted 1 0.0; 0.4 I Active I 9 I 3

than 2 pm/year. The values of the B constant were found to lie in a broad interval, with low values prevailing in the case of surfaces with an oxide layer (scale, rust) and exposure in concrete with a low chloride content, and high values prevailing in the case of all surface conditions in concrete with a high chloride content. The empirically evaluated B values contradict the commonly used values. Our results support the opinion that for corrosion monitoring in concrete it is not necessary to recalculate polarisation resistance values to corrosion current densities by a problematic constant B. For a decision on steel reinforcement corrosion acceptability it is sufficient to use original polarisation resistance values.

5. Acknowledgement

The authors are grateful to the Grant Agency of Czech Republic who have supported this research as grant No.103/98/1576, which is part of the project MSM 223100002.

References

1. ASTM C 876-91 Standard, Standard test method for half-cell potentials of uncoated reinforcing steel in concrete, ASTM (1991). 2. B. Elsener, H. Wojtas and H. Bohni, in Proc. 12th Int. Corros. Congr., 1993, NACE International, Houston, Tx, 1993, pp.3260-3270. 3. K.Videm and R. Myrdal, Paper No. 180, in Proc. 13th Int. Corros. Congr., Melbourne (1996),

4. B. Elsener, Mater. Sci. Forum, 1997,247, 127-138. 5. C. Cigna, E. Proverbio and G. Rocchini, Corros. Sci., 1993,35 (5-8), 1579-1584. 6. K. R. Gowers and S. G. Millard, Corros. Sci., 1993, 35, 1593-1600. 7. J. Flis, S. Sabol, H. W. Pickering, A. Sehgal, K. Osseo-Asare and P. D. Cady, Corrosion, 1993,

pp.1-8.

49 (7), 601-613.



8. N. Berke, V. Chaker and D. Whiting (Eds), Corrosion rates of steel in concrete, in STP 1065, ASTM (1990). 9. C. Andrade and J. A. Gonzales, Werkst. Korros., 1978,29, 515-519. 10. J. A. Gonzales, S. Algaba and C. Andrade, Brit. Corros. J , , 1980,15, 135-139. 11. P. NovBk, D. Zhang and L. Joska, Koroze Ockr. Mater., 1996,40,2-7. 12. P. NovBk, R. Mala and L. Joska, Cem. Concr. Res., in print.


Part 2

Corrosion Protection of

Reinforced Concrete

Structures


6 Corrosion and Protection in Reinforced Concrete:

A Computerised System for Studying its Phenomenology, Causes, Diagnosis

and Remedies

P. PEDEFERRI Department of Applied Physical Chemistry - Politecnico di Milano, Piazza Leonard0 da Vinci, 32

1-20133 Milano, Italy

ABSTRACT The Department of Applied Physical Chemistry and METID (Center of Innovative Methods and Technologies for Teaching) of Politecnico di Milano has produced a computerised system for studying the corrosion of reinforced concrete structures*. It can be used as a support system for engineers and architects who must deal with problems of durability in the phases of planning, construction, management and repair of their works.

1. Introduction

Buildings of reinforced concrete are nowhere near as eternal as they were thought to be up until the 1970s, but have a limited life span precisely because of the corrosion of their steel reinforcement [l].

The phenomenon involves not only bridges or highway and marine infrastructures, still the most seriously affected, but also public and private structures, churches, stadiums, monuments and others.

The corrosion is often visible to the casual observer. In other instances, corrosion occurs in the least exposed and least visible parts of the structure, though it is nonetheless just as dangerous.

The Department of Applied Physical Chemistry of the Politecnico di Milano in cooperation with the Center of Innovative Methods and Technologies for Teaching (METID) has produced a computerised system for studying the corrosion of reinforcements in structures of reinforced concrete.

It can be used as a support system for engineers and architects who must deal with problems of durability in the phases of planning, construction, management and repair of their works. It is obvious, however, that such a system does not in itself solve the problem. The solution lies in the hands of the experts. However, the

*The system has been created with the cooperation of L. Bertolini, E Bolzoni, L. Lazzari, M. Ormellese and P. Pedeferri of the Department of Applied Physical Chemistry, and A. Belluscio, M. Pillan, P. Sassaroli of the Center of Innovative Methods and Technologies for Teaching (METID) of Politecnico di Milano.


52

CD-rom can be used also without a specific background.


The complete structure of the CD-rom is shown in Fig. 1. Four different methods may be used to consult the hypertext: the entire pathway

may be read from the beginning; alternatively, one of the three guided pathways dealing with a specific issue ('corrosion and protection', 'corrosion and repair' and 'repair') may be followed. It is also possible to reach a specific area (e.g. 'exercises')

b

"Prevention"

I HOME PAGE

4-

"Corrosion, inspection protection and repair''

First guided pathway

Third guided pathways

"Repair"

" Regulations and Legislation "

b

4-

I '#Exercises I

Fig. 1 Diagram of the CD-rom menu.


Corrosion and Protection in Reinforced Concrete 53

from the main CD-rom menu. Four worksheets are reported that illustrate the choice of a protection method (prevention), a technical-economical way to evaluate the system of corrosion control and methods of repair (cost) and the definition of the cement mixture (mix design). The fourth worksheet (repair) illustrates techniques of repair that can be used for structures subjected to corrosion induced by carbonation or chlorides. Lastly, one area, entitled ‘Regulations and Legislation’, lists the laws and the regulations referring to the situation in Italy.

At the moment the CD-rom is available only in an Italian version. An advanced professional version is under development primarily aimed to evaluate expected or residual or extended life of concrete structures when corrosion is likely.

2. Hypertext

The hypertext illustrates the phenomena of corrosion in reinforced concrete structures and techniques of prevention and repair. Contents of the hypertext are taken up in detail in [l].

The hypertext is divided into eleven chapters that deal with different topics related

Corrosion Concrete constituents I in concrete I

+onCrete properties

4TransDort phenomena I

9Reinforcin.g bars I *Electrochemical aspects

+Corrosion mechanisms I

?Concrete degradation

?Prevention and urotection I ADiagnosis and monitoring

Repair

9Electrochemical techniaues I Fig. 2 Overall structure of the hypertext.


54

to concrete, corrosion, protection and repair (Fig. 2). There are three levels of detail in this section: chapters, paragraphs and pages,

according to a father-son logic. Within the hypertext the user can move from one screen to another simply by clicking on key words (hotwords) that make up the hypertext links between two different subjects.

An understanding of the topics presented in the hypertext is facilitated by a variety of features: pictures, some of which are interactive; tables; animated drawings; photographs; film clips on specific issues as well as on experimental measures (potential, corrosion rate, carbonation); exercises.


The following tools of navigation facilitate consultation of the hypertext:

a map which provides a list of paragraphs for every chapter and a list of pages for every paragraph;

an index which lists, for any given word, all the pages in the hypertext dealing with a specific topic; and

a bookmark which can tag various pages during consultation, and can be recalled at the end of navigation.

3. Guided Pathways

The hypertext can be consulted following one of the three guided pathways dealing with a specific issue: ’corrosion and protection’, ’corrosion and repair’ and ’repair’.

In the first guided pathway, ‘corrosion and protection’, the electrochemical aspects underlying corrosion are initially analysed. Subsequently, the various causes of corrosion (carbonation, chlorides, hydrogen embrittlement and macrocouples) are presented as can be seen in Fig. 3.

In regard to corrosion by carbonation, for example, an analysis is made of the environmental factors involved (relative humidity, temperature, CO, level) and of the properties of concrete (w / c ratio, type of cement) that influence the penetration of carbonation and corrosion rate once it has begun.

For corrosion by chlorides, their content is analysed (surface concentration, diffusion coefficient and critical concentration of chlorides) as well as their influence on the period of initiation and propagation of corrosion.

The guided pathways then continue with an illustration of phenomena of concrete degradation: freeze-thawing attacks, alkali-aggregate reactions, acid attacks, washing away, attacks by sulfates and by sea water (Fig. 4).

Various techniques of prevention are then described, beginning with mix design and planning, and continuing with additional protection techniques, i.e. corrosion inhibitors, surface treatments, corrosion resistance reinforcements and cathodic prevention.

The guided pathway ends with a chapter dedicated to techniques of inspection and monitoring. In this, a description is given of the principal measurement techniques: potential, corrosion rate, concrete resistivity, carbonation depth and chloride profile.


Corrosion and Protection in Reinforced Concrete

nvironmental

+hloride corrosion I-'A -

55

ttack mechanism -Chloride surface content 'Chloride diffusion coefficient

Critical chloride content Initiation time

-/Corrosion rate

I mechanism I

Macrocouple 'Macrocouple in aerial structures corrosion -Macrocouple in buried or

xposure class

aximum allowable penetration

*arbonation I influence of relative humidity borrosion

b k a r b o n a t i o n coefficient p b n i t i a t i o n time I

d t a t i o n a r v interference I 1 I

bmmersed structures J

ydrogen embrittlement

Fig. 3 Corrosion.



-orrosion inhibitors I-.A

ix Design ix design procedure Prevention

ddition method -):nhibitors efficiency -).Mechanisms

-CFjurface treatments/-+ urface treatments efficiency

*tainless steel rebars 1-9 ritical chloride content

h a l v a n i z e d rebars 1 I I

alvanized rebars in chloride

*oated rebars

+esign I dditional protection I

athodic-I perating conditions I Fig. 4 Prevention.

The third guided pathway provides a detailed analysis of repair methods (Fig. 5), according to the RILEM recommendations [3], both for carbonated structures and for structures containing chlorides. These include:

repassivation with alkaline concrete;

localised repair;

control of humidity in concrete; and

coated rebar.

Tables and figures illustrate applications of the measures listed. One section deals with repair techniques for structures affected by interference: limiting dispersion of current, impermeabilisation, electrical sectioning of the reinforcements.

Lastly, electrochemical techniques of repair are described, such as cathodic protection, electrochemical removal of chlorides and electrochemical realkalisation (Fig. 6).



- +kepair of structures subjected kimiting dispersion of current

to stray current corrosion Impermeabilization

Repair 1

carbonated concrete -+ Operating conditions -)Required current density

O Decision level

'protection criteria

?Repair o f c a r b o n a t e d l I 'b estoration of alkalinity using I lconcrete structures lkaline concrete or mortar

epair of chloride kestoration of alkalinity using alkaline concrete or mortar containing structures

A L o c a l patching with lkaline concrete or mortar imiting m'ater content of concrete

W o a t i n g reinforcement I

Fig. 5 Different methods of repair,

CP of buried or immersed

metallic structures CP of rebars in chloride contaminated concrete CP of rebars in

Removal mechanism I

Realkalisation mechanism I realkalisation

Fig. 6 Electrochemical methods of repair.



4. Worksheets

Worksheets may be consulted starting from the main menu, as previously described (Fig. l), or from the hypertext, in correspondence with the chapters ‘prevention’ and ’repair’. Some of the methodologies and problems dealt with in the worksheets are given below.

5. Prevention

The worksheet provides a means of evaluating solutions proposed by current regulations [2] and of adopting an alternative solution to prevent corrosion of reinforcements.

First, data on the service life of the structure is introduced together with the class of environmental exposure to which it will be subjected. The program then provides the solutions advised by the guidelines of the Italian Department of Public Works. Details are given of the maximum w / c ratio, minimum cement content, minimum class of mechanical resistance and minimum concrete cover. A chloride profile, i.e. penetration of carbonation, is estimated from data introduced by the user, and is represented graphically.

The user can evaluate the advisability of adopting an alternative solution, choosing to use a particular type of cement, varying the w / c ratio and concrete cover, using galvanised or stainless steel reinforcements.

The two solutions are displayed on the same graph, and the more suitable one highlighted.

6. Mix Design

This worksheet allows a first approximation of the cement mixture to be chosen. The user inserts data on the class of environmental exposure, type of cement and class of resistance desired.

The program furnishes the w / c ratio able to satisfy requirements of both mechanical resistance and durability. Water and cement contents are both evaluated on the basis of the class of workability required, the type of aggregates used and the presence of plasticiser (information inserted by the user). A final display summarises the components of the mixture so that mixtures obtained with different initial data can be compared.

6.1. Costs

The worksheet based on a technical-economical approach allows one to compare possible alternative corrosion control methods of repair for structures such as, for example, highway viaducts corroded by chlorides present in de-icing salts.

The program evaluates the most economic alternative among the use of reinforcements in carbon steel, in galvanised steel, in stainless steel and cathodic prevention. It calculates the relative costs involved. Results are expressed as life cycle



cost,* and indicate the different contributions made to the total cost of construction, operation and repair.

The construction costs are evaluated on the basis of available data in literature (related to 1999), and they are expressed in costs per m2. The operating and monitoring costs (related to corrosion control) are based on authors’ experience and they are fixed for every year period. Repair costs for each considered construction techniques are input at period of corrosion initiation and they are evaluated as 10% of construction costs. Social costs, valued on the basis of number of days of service interruption, are also a fixed percentage of construction costs. A table reports the LCC (present value) for each techniques.

For instance, let us consider a 5000 m2 surface bridgedeck, with a design life of 50 years, chlorides surface content 3% with respect to cement weight, Portland cement, 0.5 w / c ratio and a 35 mm concrete cover. The program evaluates the corrosion initiation as follows: 2 years for carbon steel, 6 years for galvanised steel, no corrosion initiation time in case of stainless steel and cathodic prevention. The program calculates as the more economic construction technique the use of stainless steel, with present value equal to 7,325,000 Euro (6,825,000 for construction costs and 500,000 for operating costs). No repair costs are considered since no corrosion is expected. Increasing the surface content of chlorides to 5%, and keeping the same other conditions, the most economic technique is then cathodic prevention. In this case stainless steel is expected to corrode after 43 years. The use of reinforcement in carbon steel is economic when a ground granulated blast furnace slag (GGBS) cement or higher concrete cover are used or when the design life of the structure is low (less than 10 years), since the chlorides do not have time to diffuse to the rebar.

6.2. Repair

This worksheet illustrates techniques of repair that can be used for structures subject to carbonation or chloride corrosion, according to RILEM regulations [3].

The user must define the extent of delamination, the cause of corrosion and, according to the type, the extension of carbonated areas or areas containing chlorides.

Possible methods of repair are illustrated, and the most suitable is pointed out.

References

1. I? Pedeferri and L. Bertolini, La Corrosione nel Calcestruzzo Arrnato. McGrawHill Libri Italia, Milano, 2000 (in Italian). 2 . Linee guida sul calcestruzzo strutturale, della Presidenza del Consiglio dei lavori Pubblici, Dicembre 1996. 3. RILEM, Technical Recommendation 124 SRC: ’Guidelines to Repair Strategies for Concrete Structures Damaged by Reinforcement Corrosion’, 1993.

*The LCC method (Life Cycle Cost) is based on the definition, for every alternative, of costs referring to the entire life cycle of the structure. These can then be calculated if the inflation rate and rate of interest are known. Using this approach, the optimal solution is the one which guarantees the lowest LCC.


7 Organic Corrosion Inhibitors for Steel in

Concrete

B. ELSENER, M. BUCHLER and H. BOHNI Swiss Federal Institute of Technology, Institute of Material Chemistry and Corrosion, ETH - Honggerberg,

CH-8093 Zurich, Switzerland

ABSTRACT The efficiency of an organic corrosion inhibitor blend in preventing and stopping ongoing chloride-induced corrosion of mild steel has been investigated in saturated. Ca(OH), solutions and in ordinary Portland cement (OPC) mortar. The results show that only high concentrations of the inhibitor (=lo%) allow the inhibition of pit initiation in solution. However, the inhibiting properties can be lost either by evaporation of the volatile constituent of the inhibitor or by the precipitation of the non-volatile fraction of the inhibitor in presence of calcium ions. The addition of the inhibitor blend to mortar yielded no inhibiting effect except the retardation of the corrosion initiation in the case of chloride- induced corrosion. Once corrosion had started the polarisation resistance values were similar for samples with and without inhibitor. On already corroding steel samples in chloride-containing saturated Ca(OH), solutions a slight increase (ca. 3-4x) of the polarisation resistance was found after adding 10% inhibitor to the solutions. No significant increase of the polarisation resistance was observed after applying the inhibitor on chloride-containing mortar samples with corroding steel. Field tests on chloride-contaminated structures also showed that virtually no effect of the surface applied organic inhibitor blend on half cell potentials or a reduction of macrocell currents were found.

1. Introduction

Corrosion of reinforcing steel due to chloride ingress from de-icing salts or carbonation represents the most widespread form of deterioration of concrete structures. Despite the huge demand, a simple, cheap, and reliable technique which either protects the steel from corrosion or at least lowers its corrosion rate is still lacking. Over the past decade, however, the concrete repair industry has developed novel techniques that are claimed to prevent or at least reduce corrosion of steel in concrete. These 'corrosion inhibitors' can be used in reinforced concrete as preventive measure for new structures (as addition to the mixing water) or as surface applied inhibitors for repair purpose. The application from the concrete surface could be an especially promising technique to protect already existing structures or increase lifetime of structures that are already showing corrosion. The published knowledge is summarised in a recent state of the art report on corrosion inhibitors for steel in concrete [l]. Regarding organic corrosion inhibitors, a broad variety of compounds can inhibit corrosion of steel in neutral and alkaline environments [2,3]. The published results for organic inhibitors in concrete



repair systems mostly concern commercially available systems and are conflicting: hydroxy alkylamine based blends are reported to exhibit inhibiting properties against chloride-induced pitting corrosion in laboratory and field tests [4-61 but on the other hand insignificant inhibiting properties of a dimethylethanolamine in concrete have been reported [7]. This might in be due to the rather limited information regarding the detailed composition of these products and to a shortage of publications from independent research in this field [8].

The aim of this investigation is the evaluation of the efficiency of various organic inhibitor blends [4-61. The inhibiting properties were determined (a) in solution, in order to determine the critical concentration of the inhibitor, and (b) with mortar samples, to characterise the behaviour under more realistic conditions for preventive and repair application.

2. Experimental

The cell used for immersion experiments is shown in Fig. 1. Three rebars (sand blasted and degreased in acetone and ethanol) were mounted in the cell. Saturated Ca(OH), solution (pH 12.5) was used to simulate the alkaline pore solution of concrete and 0.1, l and 10% of an organic inhibitor blend was added. The cell was closed with a plug in order to avoid evaporation of the volatile part of the inhibitor and only opened

Fig. 1 Experimental set-up for immersion experiments with simultaneous monitoring of the corrosion potential and the polarisation resistance: 1. PMMA-cover; 2. O-ring; 3. glass vessel; 4. solution; 5. rebar; 6. stainless steel screw; 7. electrical contact; 8. PVC-pipe; 9. plug.


Organic Corrosion Inhibitors for Steel in Concrete 63

for short time intervals to provide oxygen in the gas phase. After prepassivation for 7 days, allowing the formation of a stable passive film, the solution was replaced by with saturated Ca(OH), containing 1~ NaC1.

The corrosion potential (SCE) and the polarisation resistance (measured at 0.1 mV-l in the range of + 10 mV relative to the corrosion potential with a Zahner IM6) were recorded over time.

Experiments in mortar were performed with cylindrical lollipop type samples containing a sand blasted rebar (Fig. 2). Mortar with a w / c ratio of 0.75, cement content of 350 kg m-3 and cement/sand ratio of 1:3 was used. After casting and demoulding the samples were cured for 70 days in 100% relative humidity. One series of samples was prepared with admixed inhibitor blend in concentrations (0, 0.015, 0.075 and 0.375% by weight of cement); after curing and drying in the laboratory the samples were subjected to cyclic treatment with 2.5 days drying in air and 1 day immersion in 6% NaCl solution. A second series without inhibitor was subjected to cyclic treatment as above. After the onset of corrosion the samples were dried for 24 h at 30°C and 40% RH and then soaked once with pure solution of the inhibitor blend for 24 h. The corrosion potential, the ohmic resistance and the polarisation resistance of all the embedded steels was determined after each cycle by rapid galvanostatic pulse measurement [9].

3. Results

The formation of a passive film in chloride-free solution in the first seven days results in increases of the corrosion potential and the polarisation resistance. After the change to chloride-containing solutions, a sharp drop of the corrosion potential and the polarisation resistance was observed, indicating the onset of pitting corrosion. As previously reported, concentrations below 10% of the inhibitor blend cannot avoid pitting corrosion in saturated Ca(OH), with IM NaCl[2]. In solutions with 1% inhibitor an increase of the polarisation resistance by a factor of about 3 compared to the non-

Fig. 2 Sample geometry for the investigation of the inhibiting of the inhibitor in mortar. The diameter of the mortar cylinder is 42 mm; 1. PVC pipe; 2. stainless steel screw; 3. epoxy resin; 4. rebar; 5. mortar.



inhibited solution was observed [lo]. Opening of the cell after 50 days of exposure of the rebars to a saturated Ca(OH), + 1~ NaCl solution with 10% of inhibitor blend resulted in a drop of the open circuit potential and the polarisation resistance within a few days (Fig. 3), indicating the initiation of pitting.

Chemical analysis of the inhibitor blend revealed that it consists of more than 90 % of a volatile hydroxy alkylamine (mainly dimethylethanolamine), while the other non-volatile part was composed of carboxylic acids (mainly benzoic acid) that were separated by distillation at 3OoC at 40 mbar. The two fractions were added separately to the saturated Ca(OH), solution in the relative concentration corresponding to a 10% inhibitor content in the solution (0.5% and 9.5% respectively). As reported already [lo], the potential drop after the change of the solution shows that neither the volatile nor the non-volatile fraction of the inhibitor alone can prevent the initiation of pitting corrosion, but addition of both fractions result in an increase of the polarisation resistance by a factor of 2-3 compared to the non-inhibited solution.

The time to corrosion initiation (drop in corrosion potential and in polarisation resistance) was determined for all mortar samples of the series with admixed inhibitor subjected to cyclic chloride ponding. The first activation occurred in the series without inhibitor, after about 60 days. The first activation in the series with high concentration of the inhibitor was observed 30 days later, thus the initiation of pitting corrosion when chlorides penetrate from outside is delayed. Additionally, the time period to activate all six samples of a series is much longer for the inhibited samples than for

-100

ic 0 -200 v) > E u -300

-400 0 a

-500

-600 0 10 20 30 40 50 60 70 80

t [days]

1 o2

10'

1 oo

n 0

Fig. 3 Repeated experiments of corrosion potential and polaristion resistance of rebars in satd Ca(OH), containing 10% inhibitor. After 50 days the cell was opened by removing the plug in the couer (MCI 2000 is the commercial amine-based inhibitors).


Organic Corrosion Inhibitors f o r Steel in Concrete 65

100

80

60

40

-+- Series 1 --B- Series 2 + Series 3 * Series 4

20

0

0 50 100 150 200 250 300 350 t [days1

Fig. 4 Percentage of corroding samples plotted against the time of cyclic wetldry treatment. Inhibitor content: series 1, 0; series 2, 0.35 kgm”; series 3, 1.75 kgm” (recommended dosage); series 4, 8.75 kgm”.

the inhibitor-free-samples (Fig. 4). No significant differences between the different inhibitor contents can be observed. This result from electrochemical measurements was confirmed by measuring the mass loss of all the samples studied [ lo] .

Samples with active chloride-induced corrosion in saturated. Ca(OH), + IM NaCl solutions showed an increase of the polarisation resistance by a factor 3-4 after addition of 10% of inhibitor blend (Fig. 5). The effect of adding inhibitor to chloride- containing mortar samples with corroding steel bars (Fig. 6), simulating a restoration treatment with surface applied inhibitor was to show a slight increase of the open circuit potential but no significant change in polarisation resistance.

4. Discussion

4.1. Corrosion Initiation

When sufficiently high concentrations are present at the steel surface, the organic inhibitor blend demonstrates a marked effect on pit initiation in the solutions containing 1~ NaC1: no pitting corrosion occurs at an inhibitor content of 10% but at


--* - / - -

- (d

C a 0

.- c c

a

-200 iii

-0.2

-0.3

-0.4

-0.5

-0.6

-0.7

2 1 o4

9- 1

: P 1 -

' I

1 o5

4 ' -300 > E a '= -400 - C a, 0 a

Y

: - -

-500 -

-600 ~ ' " ' " " ' " " ' " " ' " " ' ' ~ ' ~ ' '

P

3 U CI

i o 4 9 dJ

a 3 0 -

103 Y

Id

1 o3 0

Fig. 6 Polarisation resistance and corrosion potential of steel in mortar samples. Time period 1: cyclic treatment with I M NaCl solution; Time period 2: 80% humidity. After 12 days the samples were dried and soaked for 24 h in pure inhibitor.


Organic Corrosion Inhibitorsfor Steel in Concrete 67

concentrations of 1% pitting corrosion is initiated similar to the non-inhibited solution. For the inhibition of pit initiation, both constituents of the inhibitor blend have to be present on the steel surface, as has been shown by the separate investigation of the two fractions of the inhibitor [ lo] . Despite the comparatively high concentration of 9.576, the volatile fraction (mainly dimethylethanolamine) cannot prevent the onset of pitting corrosion. This result is in agreement with the relatively poor inhibiting properties reported for the pure dimethylethanolamine [7,11] and surface analytical studies, where no specific adsorption of dimethylethanolamine on passive steel in alkaline solution was found [12,13]. The presence of the non-volatile fraction is thus crucial for the observed inhibiting effect of the inhibitor blend, but this fraction alone also cannot prevent steel from the initiation of pitting corrosion. Evaporation of the volatile fraction (after opening of the cell) results in a decrease of the dimethylethanolamine concentration in solution and chloride-induced corrosion that starts (Fig. 3). Hence, for the inhibition of the corrosion initiation in chloride- containing solutions a sufficiently high concentration and the presence of both inhibitor fractions are required.

Experiments in mortar with inhibitor demonstrated clearly (Fig. 4) that the corrosion initiation is delayed. However, unlike the experiments in solutions, initiation of pitting corrosion cannot be prevented. The evaluation of the chloride content yielded comparable values for all sample series after 380 days of cyclic treatment in chloride solution. Thus, the inhibitor blend does not influence the chloride transport and the observed delaying effect on the initiation of pitting corrosion is caused by its inhibiting properties.

4.2. Corrosion Propagation

Besides the inhibition of the corrosion initiation, corrosion inhibitors can also influence the corrosion rate after the onset of pitting. In the present work the corrosion rate was obtained by determination of the polarisation resistance. The use of these values for absolute comparison can be critical, as the area of the active corrosion site is not known. The polarisation resistance is normalised with the whole sample surface, which is of course mainly passive. Therefore, differences in the polarisation resistance could be exclusively due to a different pit size instead of a different propagation rate of the pit. Nevertheless, the obtained polarisation resistances are useful for a qualitative comparison of the effect of the corrosion inhibitor.

In chloride-containing solutions the inhibitor indeed has an influence on the corrosion rate after the initiation of pitting. According to Table 1 a concentration of 1% inhibitor cannot prevent the initiation of pitting corrosion, but the polarisation resistance is about four times higher, indicating a lower corrosion rate.

Contrary to the experiments in solution, no effect of the inhibitor on the corrosion rate was found in mortar experiments. This result was obtained by polarisation resistance measurements and was confirmed by determination of the mass loss of the rebar [lo]. Hence, the inhibitor added to the mortar mix does not affect the corrosion rate in mortar after initiation. For application on new structures it can thus be concluded that the inhibiting effect is limited only to some retardation of the corrosion initiation.



Table 1. Polarisation resistance of the samples immersed in Ca(OH), solution with different content of inhibitor after the addition of chlorides

Mass% inhibitor R, (kRcm2)

10 490 + EO

0.1 2 + 1

10 3 + 1

5. Pre-corroded Samples

No reduction of the corrosion rate was found when applying the inhibitor blend on chloride-containing mortar samples where pitting corrosion of the rebars was ongoing (Fig. 6), despite a significant reduction found in solution (Fig. 5) and a high diffusion rate reported for the inhibitor [14,17]. A similar result was reported in recent work of Page and Ngala [a] which studied another proprietary, alkanolamine-based blended inhibitor known to contain ethanolamine and an inorganic phosphate. Repeated ponding and drying according to the manufacturer’s dosage caused only a modest reduction in the corrosion rates of pre-corroded steel bars embedded at 12 mm depth in concrete with 0.65 w / c and with low to modest levels of chloride contamination and was apparently ineffective in cases of high chloride content (2.4% by mass of cement). A possible explanation of this discrepancy may be that the blended inhibitor studied in this work may have fractionated and only the volatile part of the inhibitor (hydroxyalkylamine) shows a high diffusion rate and reaches the steel surface [lo]. The same inhibitor as studied in [a] was included in a field test with surface applied inhibitors on chloride-contaminated structures and no reduction in the corrosion rate (macrocell current) was found [la].

6. Inhibitor Concentration

The inhibition or retardation of the initiation of pitting corrosion requires a comparatively high concentration of 10%. Typical amine concentrations for the inhibition of uniform corrosion in acidic solutions are in the region of lo4 mol L-*. This contrast is a result of the completely different corrosion mechanisms which are taking place for example, in acidic solutions where the bare metal surface is in contact with the electrolyte and the adsorption of specific inhibitor molecules on the metal can result in a strong decrease of the corrosion rate. In alkaline solutions iron is protected against uniform corrosion by the passive film. Corrosion is initiated locally in the presence of chlorides leading to a heterogeneous system with separated anodes and cathodes, adverse mass transport conditions in the pit, and migration of ions. It is beyond the scope of the present work to discuss the detailed mechanisms, but it is


Organic Corrosion Inhibitors for Steel in Concrete 69

a generally observed phenomenon that inhibitors of pitting corrosion require high concentrations. For nitrite [ 151 and monofluorophosphate (MFP) [16] the ratio between inhibitor and chloride concentration is reported to be in the order of 1. Considering the molar weight of the dimethylethanolamine of 89.14 gmol-* and the chloride concentration of 1 mol L-I used in this work, a ratio of inhibitor/Cl- of about 1 is also obtained.

7. Stability and Long-term Efficiency

The efficiency of the inhibitor blend was investigated in accelerated tests to obtain information within a reasonable time scale. The acceleration was obtained by a strongly enhanced penetration of chlorides due to the cyclic treatment. However, the obtained results are only correct when no other relevant time-dependent processes are taking place or when they are accelerated in the same way. Surface application of organic inhibitor on mortar samples with corroding steel did not show any reduction in the corrosion rate (Fig. 6) in agreement with long term laboratory studies [8] and with field tests on chloride-contaminated structures [ 181.

On considering the loss of inhibiting effect observed in experiments performed in solutions (Fig. 3) when evaporation of the inhibitor occurs, the question arises whether this process might have an influence on the long-term performance of the commercial inhibitor blend when admixed in mortar or concrete structures. The fast diffusion of the hydroxyalkylamine through the concrete has been demonstrated [ 14,171. The diffusion direction follows the concentration gradient of the hydroxyalkylamine. As the hydroxyalkylamine concentration can be expected to be very low on the concrete surface, it can be postulated that the substance should evaporate from concrete structures, which must cause a decrease of concentration over time. Measuring the amine concentration in airtight compartments containing a mortar sample of the series 2 and 3 respectively after 350 days cyclic treatment showed that the amine concentration (determined qualitatively with an amine electrode (Orion; model 9512)) in the air surrounding the samples is increasing with time (Fig. 7). Additionally the evaporation rate is higher for samples with higher inhibitor concentrations which is in agreement with the expected steeper concentration gradient. Hence, the main component of the inhibitor is leaving the concrete structure over time and the question arises whether the observed retardation of the corrosion initiation will still occur under realistic conditions (after several years of service).

8. Conclusions

Experiments in solutions are useful for the characterisation of the inhibitor efficiency and the determination of influencing parameters. However, experiments in mortar are more severe and thus necessary to obtain results relevant for practical applications.

In chloride-containing alkaline solution, the inhibitor can prevent steel from corrosion initiation at sufficiently high concentrations (10%). At lower



0 . 7 ~ 1 1 1 ' 1 I I 8 I I I I t 1 , - 1

0.6

0.5

0.4

0.3

0.2

0.1

0 0 5 10 15

t ways1 Fig. 7 Relative amine concentration of the air in a closed compartment containing mortar samples with admixed inhibitor.

concentrations pitting corrosion initiated, but the corrosion rate is lowered by the presence of the inhibitor.

Contrary to the experiments in solution, in mortar neither complete inhibition of the initiation of pitting corrosion nor an influence on the corrosion rate is obtained. Nevertheless, the inhibitor can delay the corrosion initiation.

The inhibitor reduces the corrosion rate of pre-corroded samples in chloride- containing alkaline solutions, no reduction in corrosion rate was found when the inhibitor was surface-applied on mortar samples. Recent field tests on chloride-contaminated structures gave the same result.

The volatile constituent of the inhibitor was found to evaporate from solutions and from mortar with a consequent loss in inhibiting properties. The long term efficiency of the admixed inhibitor in field application is questionable.

8. Acknowledgements

The authors are pleased to acknowledge the financial support of Holderchem for this project and Prof. Chris Page for making available his paper [8] in advance.


Organic Corrosion Inhibitors for Steel in Concrete

References

71

1. B. Elsener, Corrosion Inhibitors for Steel in Concrete - European Federation of Corrosion State of the Art report, to appear in the EFC series published by The Institute of Materials, London. 2. Y. I. Kusnetsov, Organic Inhibitors of Corrosion of Metals. Plenum, New York, 1996. 3. B. Elsener, M. Buchler and H. Bohni, Corrosion of Reinforcement in Concrete - Monitoring, Prevention and Rehabilitation (J. Mietz, B. Elsener and R. Polder, eds). Publication No. 25 in European Federation of Corrosion series published by The Institute of Materials, London,

4. A. Eydelnant, B. Miksic and L. Gelner, ConChem-I., 1993, 2,38. 5 . U. Mader, in Proc. Conf. on Corrosion and Corrosion Protection of Steel i n Concrete, Sheffield UK, 1994, Vol. 2, p.851. 6. Concrete Bridge Protection and Rehabilitation: Corrosion Inhibitors and Polymers, SHRP Report S-666, National Research Council Washington DC, 1993. 7. N. S. Berke, M. C. Hicks and P. G. Tourney, in 12th Int. Corros. Congr., Houston, Tx, USA, 1993,5, p. 3271. 8. C. L. Page and V. T. Ngala, Corrosion inhibitors in concrete repair systems, Mag. Concr. Res.,

9. B. Elsener, H. Wojtas and H. Bohni, in Corrosion and Corrosion Protection of Steel in Concrete, (R.N. Swamy, Ed.). Sheffield Academic Press, 1994, 1, p.236. 10. B. Elsener, M. Buchler, F. Stalder and H. Bohni, A migrating corrosion inhibitor blend for reinforced concrete: Part I, Prevention of corrosion, Corrosion, 1999, 55, 1155. 11. A. Phanasgaonkar, B. Cherry and M. Forsyth, in Con$ on Understanding Corrosion Mechanisms of Metals in Concrete - A Key to Improving Infrastructure Durability, MIT Boston, USA, 1997. 12. A. Rossi, B. Elsener, M. Textor and N. D. Spencer, Analusis, 1997,25, M30. 13. A. Rossi, B. Elsener, M. Textor and N. D. Spencer, XPS study of the absorption of inhibitor on iron in alkaline solutions, Proc. EUROCORR '96, Nice, 1996, Vol. 1, paper 11. 14. D. Bjegovic, L. Sipos, V. Ukrainczyk and B. Milksic, in Corrosion and Corrosion Protection of Steel in Concrete (R. N. Swamy, Ed.). Sheffield Academic Press, 1994,2 p. 865. 15. B. B. Hope and A. K. C. Ip, ACIM I., 1989,86,602. 16. C. Andrade, C. Alsono, M. Acha and B. Malric, Cem. Concr. Res., 1992, 22, 869. 17. B. Elsener, M. Buchler, F. Stadler and H. Bohni, A migrating corrosion inhibitor blend for reinforced concrete - Part 11, Inhibitor as repair strategy, Corrosion, 2000,56, 727-732. 18. Y. Schiegg, F. Hunkeler and H. Ungricht, The Effectiveness of Corrosion Inhibitors - A Field Study, submitted to IABSE Congress 'Structural Engineering for Meeting Urban Transportation Challenges', Lucerne, 18- 21 September, 2000.

1998, pp.154-69.

2000,52,25-37.


8 Corrosion Protection of Reinforcement by

Hydrophobic Treatment of Concrete

R. B. POLDER, H. BORSJE and J. de VRIES* TNO Building and Construction Research, P.O. Box 49 AL - 2600 AA, Delft, The Netherlands

*Ministry of Transport, Civil Engineering Division, Utrecht, The Netherlands

ABSTRACT Penetration of de-icing salts into concrete bridge decks may cause corrosion of reinforcement, even with an asphalt overlay. Hydrophobic treatment of concrete was studied in the laboratory as additional protection. It was shown that hydrophobic treatment strongly reduces chloride ingress, both during semi-permanent contact and in wetting/drying situations. The protection remains effective for at least five years under full exposure to outside conditions. Carbonation of concrete is not significantly accelerated. Hydrophobic treatment does not stop corrosion if initiation by chlorides has already taken place before the treatment. Methods and criteria for testing hydrophobic products are available. Hydrophobic treatment is an effective, low cost preventative measure against corrosion of reinforcement in chloride-contaminated environment. It has become standard for all new concrete bridge decks in The Netherlands.

1. Introduction

Exposure to de-icing salts of concrete bridge decks is a potential cause of damage because chloride ions may promote corrosion of steel reinforcement. Bare concrete bridge decks under de-icing salt load may show severe corrosion damage, as has occurred on a large scale in the United States. In Europe, most decks are provided with a layer of dense asphalt. In many countries, some type of additional protection is applied, like a membrane between concrete and asphalt. In The Netherlands, this is not the usual practice: the asphalt is applied directly on the concrete. In the 1990s, the favourable road surface properties of open asphalt (better visibility in wet conditions, lower noise production) led to the decision to use open asphalt on all highways and bridges in these highways. Because the open asphalt requires a higher amount of de-icing salts and is more permeable to chlorides, the need was felt to introduce some form of additional protection of those bridges. Hydrophobic treatment of the concrete with silicone compounds seemed promising. Such treatment makes concrete water repellent, so theoretically water and dissolved chlorides would no longer be absorbed. It was expected that corrosion initiation would be prevented or at least postponed. Hydrophobic treatment is relatively cheap (5 to 10 Euro/m2), so it seemed an economically attractive way to improve the corrosion protection of bridge decks. A research programme was carried out to investigate the effectiveness and to select suitable hydrophobic products. First, a test procedure and criteria were



drawn up. Several commercial products were tested [1,2]. Some products complied with the requirements, others failed. Only products specifically developed for use on concrete appeared suitable; products for brick masonry did not comply with the test requirements. The test set-up and the requirements were evaluated, slightly modified and subsequently laid down in a Rijkswaterstaat (Ministry of Transport) Recommendation. Application of hydrophobic treatment with an approved product has become standard practice on all new bridge decks.

2. Theoretical Background

When water comes into contact with a porous material such as concrete, it is transported rapidly into the pores by capillary action. The rate of absorption depends on the surface tension, the viscosity, and the density of the liquid, on the angle of contact between the liquid and the pore walls and on the radius of the pores. Since the characteristics of the liquid (water) and of the concrete are given constants, the most important factor is the contact angle (theta). In normal concrete, the contact angle is small (<90") due to the presence of molecular attraction between water and cement paste (hydrophilic behaviour). Under these conditions, a drop of water will spread on a flat surface, the level inside a capillary pore will rise above the surrounding liquid and the concrete will absorb the water. The opposite may occur when concrete is made hydrophobic: forces of attraction exerted on the liquid are greatly reduced and the contact angle is >90°. A drop takes the form of a sphere, while capillary rise is negative, so the level of liquid in the pore is lower than the surrounding liquid. Both cases are shown in Fig. 1.

Fig. 1 Interaction between wter and either non-hydrophobic (top) or hydrophobic (bottom) concrete surface.


Corrosion Protection of Reinforcement by Hydrophobic Treatment of Concrete 75

The molecular attraction between water and concrete can be weakened by impregnating the concrete with hydrophobic agents, such as silicones. From the silicone group of substances, silanes and siloxanes are most important for concrete. Silanes are small molecules having one silicon atom; siloxanes are short chains of a few silicon atoms. Their molecules contain (organic) alkoxy groups linked to the silicon atoms, which can react with the silicates in the concrete to form a stable bond. In addition, silanes and siloxanes contain organic alkyl groups which have a fatty character. After reaction of the alkoxy-groups, the alkyl groups protrude from the pore surfaces. As a result, water molecules will be repelled, the contact angle is greater than 90" and ideally water is no longer absorbed by capillary suction. In reality, capillary absorption is reduced to 10-20oJo of non-treated concrete. Because the pores are left open, silicone treatment does not block, transport of single water (vapour) molecules.

3. Objectives of Research

Hydrophobic treatment is expected to reduce the penetration of water and dissolved ions (chloride). However, the water-repellent pores may not resist water under pressure; imperfections in the treated layer will allow some liquid absorption; the hydrophobic effect may be lost gradually due to ultra-violet light and alkaline attack of the hydrophobic molecules. Hydrophobised concrete exposed to rain will be drier than non-hydrophobised concrete, because the former does not absorb water during rain and loses water in dry periods by evaporation. This might slow down ongoing corrosion. On the other hand, drying out might increase carbonation and potentially promote corrosion.

Research was carried out into various types of performance and durability of the water-repellent effect. The work reported here includes:

absorption of salt water during semi-permanent contact ;

absorption of salt water during intermittent contact;

water absorption as a function of exposure time to outside conditions;

effect on corrosion of bars in chloride-contaminated concrete; and

carbonation depth after exposure.

4. Materials: Concrete

For bridge decks (exposed to de-icing salts), Dutch standards prescribe a maximum water-to-cement ratio (w / c) of 0.45. The top layers of such decks may have a slightly higher w / c due to bleeding of the fresh concrete mix. It was decided to simulate such top-of-deck concrete with a w / c of 0.50. Traditionally, in our national practice two cement types are used, Portland cement ( O K ) and Blast furnace slag cement


76 Corrosion of Reinforcement in Concrete: Corrosion Meckaiiisms and Corrosion Protection

(BFSC with a high amount of slag,). For the experiments, two standard concrete compositions were designed, one with OPC (CEM I) and one with BFSC (CEM III/B LH HS, cu. 70% slag). Both mixes contain 340 kgm” cement, have a w / c 0.50 and river sand and gravel with maximum grain size 32 mm [1,2]. The compressive cube strength of these mixes at 28 days is about 43 MPa.

From these two ’standard’ mixes, slabs of 0.5 x 0.5 m2 with a thickness of 150 mm were cast and the top surfaces were finished with a steel trowel. The moulds were covered for 24 h with plastic foil. Then the slabs were cured for 48 h in a fog room. Specimens were sawn of 100 x 100 mm2 (finished or bottom) surface with a depth of about 75 mm. They were stored in air of 20°C and 65% RH until (at least) 28 days age before treatment. Hydrophobic agents were applied by dipping specimens twice in the liquid for 5 s with 10 min intervals. After application, specimens were stored for (at least) four weeks in air of 20°C and 65% RH before testing.

5. Materials: Hydrophobic Agents

Commercial hydrophobic products for concrete consist of silanes and/or siloxanes. Silanes contain 100% active substance or they are dissolved in alcohol or hydrocarbons (with 1040% active substance). Siloxanes are dissolved in alcohol or hydrocarbon solvents (cu. 10-2072 active substance). Silanes and mixtures of silanes and siloxanes are also available as water-borne systems (with 10-20% active substance). For environmental reasons, in the Netherlands the product is not allowed to contain any volatile organic solvents. Three commercially available hydrophobic products which comply to all test criteria, are used:

product A, 99% silane (no solvent);

product 8, 100% silane (no solvent); and

product E, 20% silane/ siloxane dispersed in water.

In standard concrete, these products penetrate to such depth and amount, that the hydrophobic zone is at least 2 mm and typically up to 5 mm deep.

6. Salt Water Absorption during Prolonged Exposure

In previous work, absorption of pure water was tested up to 24 h only. Water absorption was reduced by a factor of five to ten [1,2]. However, locally on a bridge, water containing de-icing salt may be in contact with the concrete for longer periods. Tests were carried out to determine the influence of hydrophobic treatment on absorption of chloride solution for longer periods. Standard specimens (controls and those treated with hydrophobic agents B (100% silane) and E (20% silane dispersed in water)) were brought in permanent contact with pure water or 10% NaCl (by mass) solution for 28 days. At any time during these 28 days, salt water absorption was much lower for hydrophobised concrete than for untreated concrete, roughly



1500

1000

500

--- I

- e------------ 0 I

0 10 20 30 Absorption time (day)

150

n sr € 100 P) U

0 10 20 30 Absorption time (day)

Fig. 2 Salt water absorption of OPC concrete (top): same with tenfold vertical magnification showing hydrophobised specimens only (bottom).



by a factor of ten (Fig. 2); similarly for OPC and BFSC concrete. The total amount of liquid absorbed was slightly lower for salt solution than for pure water. It appears that the favourable effect of hydrophobic treatment is maintained during long periods of exposure to high chloride concentrations.

7. Chloride Penetration during Simulated De-icing Salt Exposure

The effect of hydrophobic treatment on the penetration of chloride has been investigated by exposing specimens to wetting and drying cycles with a salt solution, simulating de-icing salt application. The specimens (only finished surfaces) were made with OPC and BFSC (standard composition). They were treated with product B or product E according to the standard application procedure. Per cycle, the treated surfaces were allowed to absorb a 10% NaCl (by mass) solution for 24 h and then allowed to dry in air of 20°C and 50% RH for 6 days. The chloride penetration profiles were determined after 12 months by dry grinding off layers of 4 mm thick and analysing the dust for total (acid soluble) chloride. After 52 weekly cycles the chloride content was between 0.2% and 0.5% by mass of cement at a depth of 16 to 20 mm in hydrophobised concrete; in non-treated OPC concrete this was 2.8%; in BFSC non- treated concrete 2.4%, as shown in Figs 3 and 4. This means that the hydrophobic treatment (with products B and E) has reduced the chloride penetration by a factor of 5 to 10. This corresponds rather well to the reduction of the (pure) water absorption. The chloride content of the outermost layer is reduced by at least a factor of 3. It may be concluded that hydrophobic treatment of concrete, made with either Portland

OPC I

- + Control

-+ Product 6

I +- Product E - 1.5

- 5 a

5 0.5 .. * 2 - 3

0-4 h a 8-12 12-16 16-20 0,

Sample depth (mm)

Fig. 3 Chloride penetration in control and hydrophobic Portland cement concrete after 12 months weekly de-icing salt cycles.


Corvosion Protection of Reinforcement by Hydrophobic Treatment of Concrete 79

n

c C +,- Control

0-4 4-8 a-12 12-16 16-20

Sample depth (mm)

Fig. 4 Chloride penetration in control and hydrophobic blast furnace slag cement concrete after 12 months weekly de-icing salt.

cement or blast furnace slag cement, strongly slows down the penetration of chloride under salt application drying cycles.

8. Durability of Hydrophobic Effect

The durability of the water-repellent effect was studied by exposing treated and control samples to outside climate and repeated testing for water absorption. Specimens made of standard concrete with OPC and BFSC were hydrophobised. Three approved hydrophobic products (codes A, B, E) were applied on the finished surface. They were exposed horizontally for a total of 62 months with the treated sides upward, on the roof of a building with free access of wind and rain. For each measurement, the specimens were taken inside, stored in 20°C and 65% RH for four weeks and tested for water absorption in 24 h, after which the exposure was continued. After 5 years exposure, additional testing of one OPC and one BFSC control sample was carried out using polarising and fluorescence microscopy (PFM) and nuclear magnetic resonance (NMR).

In Fig. 5 the water absorption coefficient (WAC) of samples is plotted as a function of exposure time. This WAC is the slope of water absorption in 24 h against the square root of time. Three different trends can be seen: (i) the WAC of OPC controls show a strong reduction, (ii) the WAC of BFSC controls showed only a small reduction, and (iii) the WAC for all hydrophobised specimens was fairly constant at a low level.

Microscopy showed that the control OPC sample was carbonated over a depth of about 3 mm. The microstructure of the carbonated zone had a slightly lower capillary porosity than deeper, uncarbonated parts. The control BFSC sample had carbonated


80 Corrosion of Reinforcement in Concrete: Corrosion Mechanisms and Corrosion Protection ~~ ~

+ Control OPC + OPC Hy 6

k +BFS Control --~f. ,BFS Hy 9 5

:P 4 5 8 3 .- 5 E 2

4- $ 1 s

n

4- S Q)

4-

P tu

0 0 10 20 30 40 50 60 70

Sample depth (mm) ~~~~~~~

Fig.. 5 Water absorption coefficient (24 h) after exposure outside.

about 7 mm and had become more porous as compared to uncarbonated parts. This is normal for BFSC concrete and is due to the lower calcium hydroxide content of the cement. From NMR testing it was found that in 24 h the water had penetrated gradually but deeply into the OPC sample, corresponding to a rather homogeneous capillary structure (see Fig. 6). In the BFSC sample, the water penetration followed a steep profile. The first few millimetres had absorbed a high amount of water, while the deeper parts were hardly penetrated at all. Qualitatively this steep water penetration profile in the BFSC sample corresponds to the abrupt change in microstructure at the carbonation front. The overall absorption by the OPC control concrete is reduced by the densification due to carbonation. This reduction does not occur significantly in BFSC concrete.

It is clear that the water repellence of concrete treated with each of the three products had not significantly deteriorated during 62 months exposure to outdoor climate. The reduction of the water absorption of untreated OPC concrete is probably caused by further hydration of the cement and densification due to carbonation of the surface layer. For non-treated BFSC concrete, the water absorption of control specimens decreased only slightly. BFSC concrete does not show densification upon carbonation, so the water absorption decreases only due to further hydration of the cement, mainly the slag particles.

The experiments show that under full exposure to climatic weathering conditions, hydrophobised concrete has retained its water repellence for five years and the expectation is that the water repellent effect will be present for many more years. With a layer of asphalt on top of treated concrete, i.e. in the absence of UV-radiation, the durability of the hydrophobic effect is expected to be at least the same or better.



20000 , after 24 hour absorption

- OPC 15000

10000 4 I

5000

1'

I I , I

0 2 4 6 8 10 Depth (mm)

Fig. 6Moisture profiles obtained by Nuclear Magnetic Resonance ( N M R ) in non-kydropkobised samples of OPC and B F S C after absorption of water for 24 k .

9. Effect on Chloride-Induced Corrosion

To study the corrosion rate of bars in chloride-contaminated concrete, twelve macrocell specimens were made of standard OPC and BFSC concrete. They were beams of 300 x 150 x 150 mm3 with two reinforcing bars (12 mm x 200 mm length) at 25 mm depth from the upper surface (intended anodes) and two bars at 110 mm depth (intended cathodes) (see Fig. 7). Specimens were subjected to cycles of wetting with 10% NaCl solution for one day and 13 days drying for about one year. In order

NaCl I I I 1

anodes cathodes I

Fig. 7 Macrocell specimen.



to promote corrosion initiation, the upper bars were polarised anodically with a cathode in the chloride solution on the top surface for one week at 1 Am-2, 59 weeks after the start of exposure. Subsequently, strongly negative steel potentials indicated that corrosion had initiated. Seventy weeks after casting six specimens were hydrophobised with product E. Cyclical salt application was continued for another year. Subsequently the specimens were exposed as follows:

outside for about a year;

in air of 20°C and 80% RH for half a year;

for a short time in a fog room; and

outside for another half year.

The total exposure after hydrophobic treatment lasted four years. Corrosion rates were determined as macrocell currents between top and bottom bars and as linear polarisation resistance of top bars using current confinement (GECOR6). Steel potentials were measured using external reference electrodes. Finally, specimens were destructively analysed. Chloride penetration profiles were determined and steel bars were etched in inhibited acid and visually inspected for number and volume of corrosion pits.

In OPC concrete, with about 1% to 1.6% chloride by mass of cement near the bars, a significant corrosion rate was found of the order of 1 pAcm-2. Macrocell currents and linear polarisation results corresponded well. Steel potentials were quite negative (-250 to -400 mV vs Ag/AgCl). In BFSC specimens, the chloride content was lower (0.4% to 1.1%) and low corrosion rates were observed (below 0.2 pAcm-2); potentials were much less negative (-100 to -200 mV vs Ag/AgCl). Apparently the chloride content near the bars (and possibly other cement type related factors) dominates the corrosion rate. Corrosion rates in hydrophobised and untreated OPC concrete were similar (Fig. 8). During exposure in the fog room, the corrosion rate increased, indicating that hydrophobic treatment did not prevent moisture from entering the concrete and accelerating the corrosion process. Three years after hydrophobic treatment, fine corrosion cracks were visible (in OPC specimens). In half a year, they increased in width and number.

Destructive analysis showed that the amount and volume of pits in the anodic bars strongly varied, from 500 to 1000 mm3 of material loss per bar. The amount of corrosion was related to the appearance of cracking. From 100 mm3 of corroded material, fine cracks were found on the concrete surface; from 500 mm3 corrosion, wide cracks occurred. Converted to average loss of cross section, these figures suggest that from 16 pm loss of steel diameter, fine cracking took place. From 80 pm material loss heavy cracking occurred. The observed material loss corresponded roughly with the amount of corrosion calculated from integration of the measured corrosion rates over time.

From these tests, it is clear that in cases where hydrophobic treatment is applied after chloride penetration has initiated corrosion, this does not stop the corrosion if the concrete is exposed to regular wetting.



2.0 T + OPC control OPC hydrophobised 1 .+ BFSC control - * - BFSC hydrophobised

1 .o

0.5

0.0 70 120 170 220 270

Sample depth (mm)

Fig, 8 Corrosion rates in chloride-contaminated OPC and BFSC concrete, control (untreated) and kydropkobised (in week 70 after casting).

10. Carbonation

There is some concern that drying out due to hydrophobic treatment causes deeper carbonation and may promote corrosion of reinforcement. It could be argued, that the corrosion rate (for carbonation induced corrosion) would be low in such relatively dry concrete. Carbonation depths were determined in the hydrophobised and non- treated macrocell specimens (see previous section) after 5 years exposure to various conditions, including wetting / drying cycles and outside climate.

Carbonation depths in OPC control specimens were 1-2 mm; in OPC hydrophobised 1-3 mm. In BFSC controls, carbonation depths were 1-7 mm, in BFSC hydrophobised 1-5 mm. Obviously, results from each group show a large variation. Carbonation depths of hydrophobised concrete did not significantly differ from untreated concrete. It appears that the cement type has a larger influence than hydrophobic treatment.

11. Conclusions

Hydrophobic treatment of concrete is an effective, low cost preventative measure against corrosion. It has become the standard practice in The Netherlands for all new bridge decks built for the Ministry of Transport. The beneficial effect is mainly a strong reduction of chloride ingress, both in semi-permanent contact and in wetting /



drying situations. The protective effect remains intact for many years. Carbonation of concrete is not significantly accelerated. Hydrophobic treatment does not stop corrosion that has already been initiated by chloride penetration before the treatment.

12. Acknowledgements

Bouwdienst Rijkswaterstaat (Ministry of Transport, Civil Engineering Division) is gratefully acknowledged for supporting this research; Dr Leo Pel of Eindhoven Technical University is acknowledged for carrying out the NMR experiments.

References

1. J. de Vries and R. B. Polder, Hydrophobic treatment of concrete, Constr. Build. Mater., 1997,

2. R. B. Polder, H. Borsje and J. de Vries, Hydrophobic treatment of concrete against chloride penetration, in Proc. 4th Int. Syrnp. on Corrosion ofReinforcernent i n Concrete Construction, (C. L. Page, P. B. Bamforth and J. W. Figg, eds), Cambridge, UK, 1 4 July, 1996, pp.546-565. Publ. Society of Chemical Industry, London.

11 (4), 259-265.


9 Cathodic Protection of Concrete Ground Floor

Elements with Mixed-in Chloride

G. SCHUTEN, J. LEGGEDOOR and R. B. POLDER* Leggedoor Concrete Repair, P.O. Box 3,9514 ZG Gasselternijveen, The Netherlands, *TNO Building and Construction Research, P.O. Box 49,2600 Delft, The Netherlands

ABSTRACT

Corrosion of reinforcement in precast concrete ground floor elements with mixed-in chloride is a major problem for many privately owned houses in the Netherlands. Conventional concrete repair does not provide durable corrosion protection. Cathodic protection has been installed to protect these ground floors for their remaining service life. The method involves installing sufficient new reinforcement, activated titanium strip anodes and cementitious grouting and a transformer /rectifier in each house. Monitoring procedures have been adjusted to suit the needs of a large number of small CP installations. The materials and equipment used will provide a service life of at least 25 years. The cost of CP compares favourably with other solutions.

1. The Problem

Corrosion of reinforcement in precast concrete ground floor elements containing mixed-in chloride has become a major problem inThe Netherlands (Fig. l), potentially concerning about 100,000 privately owned houses. Concrete damage due to corrosion in some structures had been discovered over 20 years ago. Due to the particular circumstances of these structures it was not then appreciated that it would occur on such a large scale in houses. Specific solutions for privately owned houses have now become necessary.

Normally, carbon steel reinforcement in concrete is protected against corrosion by passivation due to the high alkalinity of concrete. Loss of passivation can only occur due to carbonation of concrete or the presence of chloride ions. In the houses discussed here, a crawling space of about 0.5 m height is present below the ground floor elements. The ground floor elements are supported by the foundation beams, with spans between 3 and 4 m. The soil below the crawling space is usually damp sand or clay, so the concrete is relatively wet. In such conditions, carbonation of concrete is slow. Penetration of chloride from external sources is not relevant. During the sixties and seventies, calcium chloride was mixed into precast concrete in many cases as a set accelerator at typical contents of 0.5 to 2% (total) chloride ion by mass of cement. Passivation is however compromised at some point in time, possibly because of some carbonation or seasonal movements of the evaporation front have increased the chloride content at the steel and active corrosion is initiated. Due to the



Fig. 1 Heavy corrosion damage in precast groundfloor elements.

usually high humidity, subsequent corrosion rates may be quite high. The expanding corrosion products crack the concrete cover until it spalls. Because occupiers do not inspect their crawling space regularly, the problem is not discovered before it has caused a large amount of damage and by that time the steel cross section may have been reduced significantly. In typical cases, structural safety has become impaired after about 20 years service life. In a typical example that will be mentioned more often in this paper, 90% of the elements show corrosion-related damage of the concrete after less than 20 years.

2. Re-installing Structural Capacity and Conventional Repair

An option is to reinstate the strength of the floor elements by an alternative structural system: steel profiles or prestressing cables in plastic ducts (unbonded tendons), supporting the floor fields. The poor accessibility makes practical application of steel beams difficult and consequently expensive. Applying prestressing cables is only economically attractive if at least a number of houses in a row use this method. Another possible option to be considered is conventional concrete repair. Conventional repair requires a high quality of steel cleaning before the application of new chloride-free concrete. This cannot be achieved in practice with sufficient reliability due to the limited working space. Some chloride ions may migrate from the old concrete into the new repair material, so there will always be the risk of new corrosion of rebars. Besides, the rebars in the old concrete are still suffering from


Cathodic Protection of Concrete Ground Floor Elements with Mixed-in Chloride 87

corrosion, although this may not yet be visible. Thus, it may be concluded that conventional repair only fights the symptoms and does not solve the real problem. The Dutch Building Decree requires durable safety for the full remaining service life, typically 25 years, which cannot be guaranteed with conventional repairs.

3. Cathodic Protection

The only reliable method to stop the corrosion is provided by electrochemical means. As part of a Eureka research project, cathodic protection (CP) for ground floors was developed and for about 20 years CP of concrete structures has been a practical and effective method. Many apartment buildings with mixed-in chloride in The Netherlands have been successfully protected using CP [l]. However, specific procedures are required for this completely new application to private homes. In comparison to 'normal' concrete CP installations, these systems are small (50 to 100 m2 concrete surface). Of course, CP cannot bring back the steel which has corroded away so that, where necessary, new reinforcement must be installed.

4. Applying Cathodic Protection to Ground Floor Elements

In the precast ground floors, each element consists of two ribs and a web with a total width of 500 mm. Each rib contains one main rebar of about 12 mm dia. The elements are repaired and CP is applied as follows. First, loose concrete is removed and where heavily damaged, new bars are fixed by welding, so that the amount of steel will meet the original design requirement. The old concrete and the steel are cleaned superficially by means of a needle hammer. Then 20 mm wide activated titanium strip anodes are inserted between and below adjacent ribs as shown schematically in Fig. 2. Anode strip connections are made by spot-welding bare titanium wires. Steel continuity is provided by welding steel wires to all bars in the ribs. Semi-tubular

Fig. 2 Schematic cross-section of cathodic crotection for precast ground floor elements.



plastic forms with such a length as to provide easy handling are put together below the ribs to shape a continuous form. The form is filled with flowable cementitious grout from a pump placed outside, completely embedding the anode and any damaged rib parts. The new mortar tightly adheres to the old concrete. Wires connecting anode strips and reinforcement from the floor fields (parts separated by foundation beams) are connected at the transformer /rectifier, which is placed in a cabinet near the front door (also containing the gas and electricity meters).

For economical reasons, the existing regulations for design of CP installations could not be met on all issues. This was dealt with as follows. According to the Dutch recommendation for CP of concrete, CUR 45 [2], the current should be fed into the anode network at a minimum of two points to provide redundancy in case of cable failure. However, connecting the anode strips by spot-welding titanium wire was found to be very reliable in other installations. Moreover, the possibility of damage to the wires is very small due to the closed nature of the crawling space. It is therefore considered acceptable to use just one connection to each anode field. Similarly, CUR 45 prescribes a double connection from the steel cage to the power supply. Again, because of the small possibility of damage to the cables, in these cases it is acceptable to use only one connection per floor field.

5. Life Time Considerations

The service life of this CP system has been estimated as follows [3 ] . For a design current density of 20 mAm-2 steel surface (2 bars of 4 2 mm dia., see Fig. 2), the titanium anode strip has to deliver 1.5 mA per running metre. The maximum anode current density is 110 mAm-2 anode surface, corresponding to 5.5 mArn-l. This leaves a ’safety factor’ of over 3, so the lifetime of the anode material itself will be much more than 25 years, probably 100 years or more. The current exchange at the anode will produce acid. In the long term acid attack of the grout may occur, potentially causing failure of the system. The amount of acid generated can be calculated according to Faraday’s law. The amount of alkaline substance can be calculated from the cement content of the mortar and the calcium oxide content of the cement. The design current density of 20 mAm-2 steel surface would cause dissolution of the cement paste to a maximum of 25% of the grout volume surrounding the anode in 25 years. This is considered to be acceptable as a worst case scenario. Practical experience shows that two effects will mitigate the amount of acid attack that occurs in reality. Firstly, the real current density in the long term is probably 2 to 5 times lower than the design value. Secondly, the net acid production is 5 to 10 times lower than that calculated theoretically, due to migration to the anode of hydroxide ions produced at the cathode [4]. Based on the transport numbers of the ions involved, it can be argued that about 80% of the current is carried by hydroxide ions (produced at the steel), neutralising the acid produced at the anode. Consequently only about 20% of the current will actually cause acid to be formed. The combination of both effects reduces acid production in the anode region at least by a factor of 10, so only about 2.5% of the grout around the anode may be dissolved and consequently acid attack is no threat to a service life of well over 25 years.

The power supply is a specially designed type with a MTBF (Mean Time Between


Cathodic Protection of Concrete Ground Floor Elements with Mixed-in Chloride 89

Failures) of at least 100,000 h at full capacity of 3 A. This means that after an average period of 11.4 years the power supply may fail. The lifetime of this type of electronics is inversely proportional to the temperature, corresponding to the current delivered. In practice the average current is 1 A or less, so the expected lifetime is much longer than 25 years.

6. Test Programme

The CP system described above was installed and tested in one house (the example mentioned above) in November 1998. Subsequently similar systems were installed in 30 houses of the same type during 1999. An intensive test programme is being carried out in some houses, in order to establish safe operating criteria for the whole group. Activated titanium (Ti*) reference electrodes have been installed in these houses. The objective is to determine the average and the maximum (worst-case) current necessary to obtain between 100 mV and 250 mV depolarisation in 24 h. In the activation stage, each installation is tested for anode to cathode resistance and steel polarisation using external silver-silver chloride reference electrodes. Usually, over 100 mV polarisation is obtained in a short time (15 min) after energising. Table 1 reports decay measurements for the first house after three months of CP at 1.4 V driving voltage, resulting in 0.4 Aprotection current. At three locations there is already a good depolarisation after just 4 h. After the measurements, the voltage was increased to 2.0 V. Table 2 gives average data of three houses measured six to twelve months after energising. Future measurements will include depolarisation over 24 h.

7. Control and Maintenance

The CP system is guaranteed for 10 years full corrosion protection. The guarantee contract requires the owner to check the front panel displays of the power supply once a month and to warn the contractor if they are outside specified 'normal' voltage and current ranges. Once a year full control measurements will be made in order to check the quality of protection and the complete system. The criterion will be:

Table 1. Depolarisation in 4 h; test house 1 after three months operation at 1.4 V , March 1999; E steel potentials (vs activated Ti, similar to AglAgCl), measured 'on', 'instant of, '5 min'and '4 hour'; D depolarisation

rLocation E (on) E (inst off) E (5 m i d E(4h) D ( i n 4 h ) 1 I 1 front door 1 -550 4 9 5 4 9 5 4 5 0 45

2 front door 2 -683 -625 -600 4 7 0 155

3 garage 1 -244 -1 78 -168 -71 107

4 garage 2 -368 -312 -244 -78 230


90

Table 2. Average data of three houses (all 76 rn2 concretefloor surface), November 1999


House Voltage (V) Current (A) Average D in 3 h (mV)

1 (Me) 2.0 0.37 95

LET- ~ 2.0 0.55 112

depolarisation in 24 h within 100 and 250 mV. Regarding the frequency of measurements, the existing procedures had to be adapted. According to CUR 45, testing should take place at least twice a year. Because of the closed character of the system and the obligation for the owner to carry out monthly checks it is considered to be safe to use a lower testing frequency.

8. Economic Considerations

Because the CP system described here can be installed very efficiently, this method is about 30% more economic than the alternative structural systems mentioned above (steel beams, prestressing cables). Conventional repair is not acceptable because the problem (reinforcement corrosion) is not solved in a reliable and durable way. This view is based on many cases of failing conventional repair of structures with mixed- in chloride. In such cases, conventional repair will only cover up the damage. In fact, many CP systems in The Netherlands had to be installed because previous conventional repairs had failed after some 5 to 10 years [l]. The owner of the structure remains responsible for possible re-activation of corrosion and subsequent damage in the future. This reduces the economic value of the house. With CP the owner has a fully reliable solution which is acceptable under the existing regulations (Dutch Building Decree), for a lower price than other solutions.

9. Conclusions

From previous experience and research, cathodic protection was developed into a complete solution for corrosion damaged ground floor elements. For economic reasons, some design principles and test procedures had to be simplified. In each case, it was shown that such modifications were acceptable. The life time of the system is estimated to be more than 25 years, matching the legally required remaining service life of the houses. Testing in a few selected houses has shown CP to work well over a period of one year. In this project, CP systems are being installed (1999) in a total of some 30 houses.


Cathodic Protection of Concrete Ground Floor Elements with Mixed-in Chloride

References

91

1. R. B. Polder, Cathodic Protection of Reinforced Concrete Structures in The Netherlands - Experience and Developments, in Proc. EUROCORX '97, Trondheim, 1997, Voi.1, pp.547-552; R. B. Polder, 1998, ditto, in Corrosion of Reinforcement in Concrete - Monitoring, Prevention and Rehabilitation, Papers from EUROCORR '97 (J. Mietz, B. Elsener and R. Polder, eds), pp.172-184. European Federation of Corrosion Publication Number 25, The Institute of Materials, London, ISBN 1-86125-083-5. 2. CUR 45,1996, Kathodische bescherming van wapening in betonconstructies, Aanbeveling 45, Technical recommendation for cathodic protection of reinforced concrete (in Dutch). 3. R. B. Polder, Durability of a system for CP of precast ground floor elements, 1999, TNO Building and Construction Research report 1999-BT-MK-ROO36 (in Dutch). 4. G. Mussinelli, P. Pedeferri and M. Tettamanti, The effect of current density on anode behaviour and on concrete in the anode region, in 2nd Int. Conf. on Deterioration and Repair of Reinforced Concrete in the Arabian Gulf, Bahrain, 1987.


10 Sacrificial Anodes for Cathodic Prevention of

Reinforcing Steel Around Patch Repairs Applied to Chloride-Contaminated Concrete

G. SERGI and C. L. PAGE" Building Research Establishment, Garston, Watford, WD2 7JR, UK

*University of Leeds, School of Civil Engineering, Leeds, LS2 SJT, UK

ABSTRACT When steel reinforcement suffers localised corrosion in chloride-contaminated concrete, the most anodic regions of the metal effectively provide cathodic protection to the bars in adjacent cathodic areas. If conventional patch repair is applied to the structure, this form of adventitious cathodic protection is removed and bars which were previously behaving as cathodes in moderately contaminated areas may be transformed to anodes of cells, coupled to cathodic steel in the repaired areas. This paper demonstrates the effectiveness of combining patch repair with embedded sacrificial anodes as a means of providing continuing protection to the surrounding reinforcing bars. The sacrificial anodes designed for this purpose consist of metallic zinc in a specially formulated mortar saturated with lithium hydroxide. The high pH pore electrolyte serves to maintain the activity of the zinc, whilst the presence of lithium ions avoids the risks of alkali-silica reaction that would be incurred if other forms of alkali metal hydroxide were used.

1. Introduction

Dense concrete normally contains an alkaline pore liquid phase which protects embedded steel from corrosion by allowing a passive oxide film to develop on its surface under conditions where oxygen is available. This form of protection can be undermined, however, if the concrete undergoes either carbonation or chloride contamination in the vicinity of the steel. Such effects can lead to cracking and spalling of the cover concrete. In cases of chloride-induced corrosion, pitting develops at localised sites, whilst the remainder of the surface remains passivated providing a large cathodic area for oxygen reduction.

For alkaline concrete, corrosion of steel may be expected to vary with potential and chloride content of the concrete, as illustrated by Bertolini et al . [l] and summarised in Fig. 1. 'Pitting' represents conditions that can lead to the initiation and propagation of pits on initially passive steel. 'Imperfect Passivity' signifies conditions that allow pre-existing pits to propagate but do not favour the initiation of new pits on initially passive steel. 'Pitting Unstable' indicates conditions that do not allow the initiation or propagation of pits, so that pre-existing pits tend to repassivate. Finally, 'Hydrogen Discharge' represents conditions that lead to highly


94 Corrosion of Reinforcement in Concrete: Corrosion Mechanisms and Corrosion Prevention

+400

n

0 W 0 v) v) > > v E -400

-1200 0 0.5 1.0 1.5 2.0

Chloride content (% by wt cement)

Fig. 2 Approximate domains of electrochemical behauiour of steel in concretes with different levels of chloride contamination (After Pedeferri).

negative potentials and are sufficiently reducing to render the passive film thermodynamically unstable. In such cases, hydrogen is formed cathodically.

These principles underpin the application of cathodic protection (CP) to reinforcing steel in concrete. Thus to achieve adequate CP in a reinforced concrete structure in which chloride-induced corrosion has been occurring for some time, it is only necessary to polarise the steel from its condition of pitting to the domain where pitting is unstable so that repassivation can be achieved. Furthermore, in cases where pits have not yet developed, a small current holding the potential of the steel in the 'Imperfect Passivity' domain will be adequate to prevent pit initiation. This was termed 'Cathodic Prevention' by Bertolini et al. [l].

Whilst most applications of CP to reinforced concrete structures have involved the use of impressed current systems with various forms of extended anode [2] there is now much interest in the use of sacrificial anodes to protect susceptible regions of the reinforcement following a conventional patch repair in chloride-contaminated concrete [ 3 ] . This paper will describe the development of an anode system specifically for such use.

2. Sacrificial Anode Systems in Patch Repairs

Conventionally patch repair involves the removal of chloride-contaminated concrete from around the reinforcing steel in areas where corrosion is detectable, cleaning of the exposed metal and reinstatement with fresh alkaline concrete or mortar. The major problem with such an approach is that unless stringent measures are taken to remove all material containing significant chloride concentration from around the corroding areas, the likelihood of corrosion reappearing and cracking the concrete


Sacrificial Anodes f o r Cathodic Prevention of Reinforcing Steel Around Patch Repairs 95

adjacent to the repairs is high [4]. This is because replacement of the most intensely anodic regions of the reinforcement with passive steel in the repaired zones effectively removes the adventitious form of sacrificial anode CP that was formerly being applied to the steel in the neighbouring regions (Fig. 2a). Hence the potential of the metal in these less severely contaminated areas can rise to a value at which pitting is liable to be initiated (Fig. 2b). To avoid triggering this problem of corrosion at incipient pits around the repair zones, it is desirable to reinstate some form of intentional ‘cathodic prevention’ and this can be accomplished with sacrificial anodes of an appropriate design, which are embedded near the periphery of the repair patches (Fig. 2c).

A form of sacrificial anode, which has been found to be successful under laboratory conditions, consists of zinc encased in a high alkalinity mortar. The ability of zinc to maintain an appropriate level of current when coupled to steel reinforcement was shown to be related to the pH of the surrounding mortar (Fig. 3 ) and the use of mortars saturated with LiOH (pH > 14.5) provides a suitable environment (Fig. 4). The reservoir of excess LiOH maintains a constantly high pH value in the pore solution, which promotes and sustains the anodic activity of the zinc, whilst the presence of lithium ions is intended to inhibit problems of alkali silica reaction in the surrounding concrete if susceptible aggregates happen to be present [2].

CI.conc., x m

- cathodc

I sites I I I I I

)p in

I I

icor

I

I

t cond.. X

, I

Carnoaic

PP in :r conc. X

3

Distance along steel log current

Fig. 2 (a) Localised corrosion of steel i n chloride-contaminated concrete; (b) Incipient anode formation on steel in ‘patch-repaired concrete‘; and ( c ) Protection of steel in ‘patch-repair‘from zinc sacrificial anode.



250

n

200 a E - 150 E

100

z - 50 d 0

W

C

J

0 C

L

.-

13 13.5 14 14.5 15 Pore solution (pH)

Fig. 3 Galvanic current between zinc anodes embedded in mortars of controlled pore-solution pH and steel cathodes embedded in mortars containing 1% chloride by zueight of cement after 28 days of connection.

Fig. 4 Zinc sacrificial anode contained in a specially formulated mortar wi th wire allowing connection to the steel reinforcement in a patch repair.

3. Experimental Evaluation

To evaluate the performance of sacrificial anodes of the type illustrated in Fig. 4, a number of reinforced concrete slabs, 1000 mm x 500 mm x 100 mm, were made with two different amounts of sodium chloride (0.8% and 4% chloride by weight of cement) cast into them at specific locations, as shown in Fig. 5.

Four steel bars were embedded at equal intervals along the whole length of each slab and electrically connected externally (Fig. 5a). Some had each of the four bars replaced by four shorter lengths of bar, giving a total of sixteen individual bars as shown in Fig. 5(b). The middle section of the top two bars (or lines of bars) lay in the zone of high chloride concentration. It was this area that showed considerable cracking of the concrete caused by reinforcement corrosion after exposure of the slabs either


Sacrificial Anodes for Cathodic Prevention of Reinforcing Steel Around Patch Repairs 97

T - T

- ---- S t t & ! w , 9 l e c t n c a l l v

connected externally 8

(a) External electrical connection

Fig. 5 Concrete slabs with chloride concentrations and arrangement of steel reinforcing bars as shown.

outdoors or in a constant 80% RH environment at room temperature for a period of 350 days. The damaged concrete was removed exposing the corroded reinforcement and, after cleaning of the bars, fresh chloride-free repair concrete was cast, restoring the original shape of each slab. In some slabs, a single disc of zinc metal weighing approximately 80 g embedded in a specifically formulated mortar containing an excess of LiOH over the amount required to saturate the mix water was buried and electrically connected to the steel reinforcement externally. The slabs were then repositioned either outdoors or in the constant RH environment in the laboratory. Potential mapping surveys were carried out at appropriate times and the current passing between the zinc anode and the steel reinforcement was monitored regularly.

Figure 6 shows a sequence of potential maps for two slabs exposed to the constant 80% RH environment. Prior to repair, both showed anodic regions within the areas of high chloride concentration, signifying corrosion of the steel (Figs 6a & 6c). One month after repair, Slab-2, which contained the zinc anode, maintained the most anodic region within the repaired patch (Fig. 6d), whilst Slab-1, without a zinc anode, developed a new cathodic region within this freshly repaired area. Consequently, new anodic sites, represented by the more negative potentials, appeared outside the patch after only two months (Fig. 6b). After 18 months, although the overall potentials became less negative, the anodic region remained within the repaired area of Slab2 (Fig. 6e). Disconnection of the anode for 24 h, confirmed that the steel had been cathodically polarised by the zinc anode as the potentials of the steel showed substantial depolarisation shifts (Fig. 6f).

The current, measured between the anode and steel reinforcement of Slab-2 for about 2 years remained above 60 pA when the concrete was exposed for extended periods to an external RH, maintained at 80% or higher (Fig. 7). This represents a current density of about 1 mAm-* of steel, a level believed to be adequate for cathodic prevention [l].

As a result of cathodic polarisation of the steel in concrete, there is a tendency for the material in the vicinity of the steel cathode to become enriched with respect to OH-, Li+, Na+ and K+ whilst being depleted of C1- and 0,. Correspondingly, at the anode the surrounding material becomes progressively acidified and enriched with C1-. The enhancement of the OH- concentration and reduction of the C1- concentration near the steel surface are both beneficial in reducing the risk of corrosion. The



Repaire: Area

( a ) Slab 1, before repair. ( d ) Slab 2, one month after application of repair system.

Repaiyd Area

(b) Slab 1, two months after repair, showing formation of incipient anodes.

(c) Slab 2, before repair.

-350mV \,,,,I (e ) Slab 2, 18 months after application of repair system. Anodic region is maintained within patch.

-250mV

-2OOmV (fl Slab 2, 24-h depolarisation potentials (mV), 18 months after 6 applications of repair system.

Fig. 6 Slabs 1 and 2 before and after repair.

movement of chloride was illustrated in mortar composite specimens half of which contained 2% C1- by weight of cement and a steel electrode whilst the other half contained excess LiOH and a zinc electrode. On coupling the electrodes, a current flowed between them, gradually diminishing from an early level of 12 mAm-2 of steel area to about 2 mAm-2 over a period of 50 days. Determination of the chloride concentration profile at termination by a dry profiling technique and analysis showed that the normal chloride migration by diffusion observed in the control samples was enhanced by polarisation (Fig. 8).


Sacrificial Anodes for Cathodic Prevention of Reinforcing Steel Around Patch Repairs 99

160

140

5 120 Y

E 100 E 5 80

5 60

40

L

0

>

.- -

UnderWet 80% 30% RH Hessian

2o 0 1 0 100 200 300 400 500 600 700 80

Time (days)

Fig, 7 Variation of galvanic current between the sacr$cial zinc anode and the steel reinforcement of Slab-2.

0.30

0.25 c u 2 =0.20 o * ‘0 5 0.15

-08

S O n .- gEi

Q)

? 20.10 C W

0.05

0.00

0

0 1 2 3 4 5 6 7 Distance from steel (cm)

Fig. 8 Chloride concentration profiles between the zinc and steel electrodes in cementitious composite mortars.



Field trials involving the use of sacrificial anodes of the type shown in Fig. 4 in patch repairs applied to reinforced concrete structures are now in progress and early results have been encouraging.

4. Conclusions

Zinc sacrificial anodes encased in a specially formulated mortar containing a pore electrolyte saturated with excess LiOH were shown experimentally to provide adequate cathodic prevention of the steel in repaired areas of concrete which were originally contaminated with high levels of chloride. Polarisation of the reinforcement was maintained for a period of at least two years.

The current densities obtained were sufficient to cause significant migration of chlorides into the repair from the surrounding material.

References

1. L. Bertolini, F. Bolzoni, A. Cigada, T. Pastore and P. Pedeferri, Cathodic protection of new and old reinforced concrete structures, Corros. Sci., 1993,35, 1633-1639. 2. C. L. Page, Cathodic protection of reinforced concrete - Principles and applications, in Proc. lnt . Con5 on Repair of Concrete Structures - From Theory to Practice in a Marine Environment, Svolvaer (A. Blanckvoll, ed.). Norwegian Road Research Laboratory, Oslo, (1997), 123-131. 3. C. L. Page and G. Sergi, Developments in cathodic protection applied to reinforced concrete, ASCE 1. Materials in Civil Engineering (In Press). 4. C. L. Page, Control of the corrosion of steel in concrete, in Cathodic Protection: Theory and Practice (V. Ashworth and C. Googan, eds). Ellis Horwood, Chichester, 1993, pp.264-278.


11 Layer Zinc Anodes in Cathodic Protection of

Steel Reinforcement

W. BOHDANOWICZ Technical University of Gdansk, Poland

ABSTRACT The continuing increase in the number of reinforced concrete structures being built, their cost and strategic importance, require that good attention is given to their durability. In the case of reinforced concrete road structures, contaminated with chlorides during winter from de-icing salts, cathodic protection provides a modem method of corrosion protection. Inert anodes as used in impressed current schemes in industrialised countries may pose a threat to concrete durability due to acidification of the environment close to the anodes. This work presents the preliminary examination results of the cathodic polarisation of concrete reinforcement either by the use of an external, layer-like zinc anode as a galvanic (sacrificial) anode or the use impressed current anodes. The results indicate that each solution could be very useful for structures of national highways.

1. Introduction

Reinforced concrete is a construction material that is now very widely used, especially in housing, industrial buildings (e.g. cooling towers), road and harbour engineering. Increasing environmental pollution in recent years has resulted in a drastic decrease in reinforced concrete corrosion resistance. Specialist literature more and more reports premature corrosion failures of structures designed for many years service [l]. Poland lacks precise data concerning the cost of corrosion losses caused by reinforced concrete corrosion. Nevertheless, with fairly high probability, losses can be assumed to reach some few per cent of GNP. There are also concerns with reinforced concrete road sttuctures - the majority of them being in a bad (or extremely bad) condition. According to recent data of the US Federal Highway Agency (FHWA) 39% of American bridges (226 000) are damaged, and 23% (134 000) have been classified as structurally damaged. Deterioration of the majority of them is caused by steel reinforcement corrosion [2].

2. Corrosion of Concrete and Steel Reinforcement

Chlorides are the most corrosion aggressive influences acting on reinforced concrete [3]. Their destructive action is assisted by mechanical factors, by carbonation and by acid rain. The combined action of these factors leads to concrete and steel reinforcement corrosion, resulting in corrosion damage.

In the case of reinforced concrete buildings, structures and road surfaces the



greatest damage is observed after the winter period. In our climate road structures require application of de-icing chemicals for a few months a year. A mixture of sand with sodium or calcium chloride is most commonly used. The consumption of chlorides for de-icing purposes is not officially known, but the amounts are significant. It is known that in the last few years in the USA approximately 18 million tonnes of salts have been used annually for this purpose [4].

The high aggressivity of chloride ions results from their capability to destroy the passive state and facilitate the corrosion process, and from the autocatalytic character of the reaction [3]:

Fe + 2C1- + FeC1, + 2e- (1)

FeCl, + 2H,O + Fe(OH), + 2H' + 2C1- (2)

It can be seen that iron chloride hydrolysis leads to acidification of the environment and liberation of chloride ions. Acidification of concrete results in acidic corrosion and in the transition from the passive to the active state this is accompanied by a change of corrosion mechanism. Liberation of chloride ions allows their return into the cycle and intensification of the reinforcement steel corrosion. The increased volume of corrosion products, compared to that of the base metal, may also lead to mechanical bursting of concrete. The corrosion processes can be further accelerated by physical factors such as changes of temperature and mechanical factors (e.g. vibrations, stresses).

3. Cathodic Protection of Concrete Reinforcement

Application of relevant corrosion protection is indispensable to prevent catastrophic damages of reinforced concrete structures contaminated with chlorides. Cathodic protection (CP) of reinforcement ensures the most efficient protection. This modern protection method has not however yet been implemented in Poland.

The main goal of reinforcement CP is to lower the potential of steel to the level at which a 'concrete friendly' reaction of cathodic oxygen reduction occurs at the surface of reinforcement (3):

0, + 2H,O + 4e- + 40H- (3)

Cathodic polarisation of concrete reinforcement can be accomplished by the use of sacrificial anodes or impressed current anodes.

The principle of concrete reinforcement CP is analogous to that in other electric current conducting environments, with the main difference with buried structures being connected with the placement of the auxiliary anodes [5]. For these types of structures the location of anodes is limited by the space that is available, the distance between reinforcement and anode being small and in the range of a few centimetres. This results in particular effects arising from electrochemical reactions occurring at the anodes.

The implementation of concrete reinforcement CP is neither easy, nor cheap - but it is a requirement of the moment, as documented by Miller [6] for


layer Zinc Anodes in Cathodic Protection ofsteel Reinforcement 103

chloride-contaminated reinforced concrete cathodic protection is four times cheaper than repairs.

4. Concrete Acidification at Anodes

In the case of above ground reinforced concrete structures anodes should be located on or within the concrete. In modern CP systems inert titanium anodes covered with cobalt, platinum or ruthenium oxides are used. In the case of a locally non-uniform concrete structure, such anodes may, in areas of lower resistance, lead to the danger of acidification of regions near the anodes according to the reactions:

2H,O + 0, + 4H' + 4e-

20H- + H,O + 1 / 2 0 , + 2e-

(4)

(5)

2C1- + C1, + 2e- (6)

C1, + OH- + C10- + H+ + C1- (7)

The most important effect is the formation of hydrogen cations and liberation of chloride ions, resulting in acidic corrosion of the concrete. Limitation or elimination of the above mentioned processes is possible through reduction of the anodic current density, i.e. by increasing the effective surface area of the anode, or by changing the anode material. In this work attempts were made to use a zinc layer applied to the concrete surface as an anode.

5. External Zinc Anode

Zinc, due to its high hydrogen liberation overpotential and passivation capability, is characterised by a low corrosion rate. When the zinc surface is covered with a E-Zn(OH), layer - it can undergo passivation at pH 9-11, and when carbon dioxide is present in water - with the formation of soluble ZnCO, or ZnCO, . 3Zn(OH), - passivation is possible in the pH range of 6-12. Apart from the E-Zn(OH), which is very stable at pH 9-11, the less stable amorphous Zn(OH), can be formed.The formation of protective layers on the zinc surface results in a low corrosion rate of zinc in aqueos environments [7]. Conditions ensuring high corrosion resistance of zinc can thus occur in a wet concrete environment.

A zinc layer applied to concrete by the flame metallisation method has a highly developed surface. This leads to the additional lowering of anodic current density and decreased consumption of zinc anode in the anodic reaction:

A zinc layer applied to the concrete surface can be used both as a sacrificial anode [8], and as an impressed current anode.



6. Results of Cathodic Polarisation Measurements

The laboratory experiments were conducted using reinforced concrete specimens (Fig. l), having a cylindrical shape of 36 x 440 mm and a surface area equal to 500 cm 2.

Specimens were made of concrete having water to cement ratio equal to 0.6 (w/c = 0.6), and a centrally placed carbon steel reinforcement of 6 mm dia. and surface area equal to 100 cm2. Sets of specimens were produced with additions of 0.1,0.5 and 1.0% NaCl (by cement weight) respectively. After sand blasting and de- dusting of specimens their surfaces were dried and covered with a layer of pure zinc (by a flame metallisation method, without pre-heating of concrete surface) to an average thickness of 300 pm. For ordinary service conditions, however, the zinc layer has to be much thicker - about 500 pm [9]. The thickness in our tests was reduced in order to accelerate examination. Due to the extreme porosity of the zinc layer applied by the flame metallisation method its adhesion to concrete is good.

Reinforced specimens used in the tests were produced at the Department of Construction Materials at the Technical University of Gdansk, and the zinc layer was applied by TERMOKOR Inc., Department in Gdansk.

salt bridge, Fl

/

\

/

concrete wic = 0.6 4 36 mm S = 500 cmz + 0.1 , 0.5 and

1.0 % NaCl

rebar

S = 100 cm2

Fig. 2 Scheme of the reinforced concrete specimens used in experiments.


Layev Zinc Anodes in Cathodic Protection of Steel Reinforcement 105

Examinations were conducted with specimens completely immersed in aqueous sodium chloride solutions (of concentration 0.1,0.5 and l.O%NaCl). The solutions were exchanged once a week and the temperature maintained constant at 20 f 2°C. In tests with impressed current the protective current densities were 2 and 10 mA m-2 of steel surface ( I = 20 and 100 PA), which corresponded to anodic current densities of around 0.4 and 2.0 mA m-2 (calculated on the geometrical surface area of the specimen).

A stabilised power supply type 9634M was used as a polarisation current source. Potential measurements were conducted with the use of a multimeter Testmate Model 335 (of internal resistance 10 MR) and saturated calomel electrode (SCE) with a special salt-bridge. Potential-time results are presented in Fig. 2.

For higher NaCl concentrations (0.5 and 1.0% NaC1) the E = f ( t ) curves are similar, but the cathodic polarisation increases with increasing concentration.

Examinations of reinforced concrete CP when the zinc anode was applied as a sacrificial anode were conducted in exactly the same conditions. Current intensities were measured by a zero resistance ammeter. Results of the examinations are presented in Figs 3 and 4.

The results obtained (means from three samples) indicate a gradual increase of reinforcement cathodic polarisation. Complete CP ( E < -0.85 V vs Cu/CuSO,) is achieved after approximately two months of polarisation. During the whole experiment no significant consumption, delamination or blistering of the zinc layer anode was observed. The evaluation of the adhesion of Zn layer to concrete were carried out using the adapted 'Pull-off test for adhesion' (IS0 4624). The

-

-

-

-

-

-

-0.3

-0.4 u1

% -0.5 v) > > -0.6 i a -0.7 F Z w -0.8 I- O a -0.9

1

+ 0.1 YO NaCl 2 mArnm2 A 10 mAm-*

- l . o ~ l l l l I I ' I I I I I I I I I ' I I I

0 50 100 150 200 TIME, days

Fig. 2 Cathodic polarisation of concrete reinforcement with a surface zinc anode under impressed current, = 2 and 10 mA w2 of steel suflace, concrete with addition of 0.1% NaCl under immersion.



5 A + 1.0% NaCl

+ 0.5% NaCl + O , l % NaCl

- 4 -

- 3 -

- 2 -

- 1 -

---La 0

0 1 0 50 100 150 200

TIME, days Fig. 3 Changes of sacrificial anode current density (of steel surface) i n time for variable NaCl content i n immersed concrete.

* i- 1.0% NaCl -v*u I -0.4

W 0 V) -0.5 to > > -0.6 i 5 -0.7 I- i w -0.8 c 0 a -0.9

-1 .o

-I . + 0.5% NaCl i- 0,1% NaCl

0 50 100 150 200 TIME, days

Fig. 4 Cathodic polarisation of concrete reinforcement for variable NaCl contents in immersed concrete using sacrificial anodes.


Layer Zinc Anodes in Cathodic Protection of Steel Reinforcement 107

measurements performed with an ERICHSEN Model 513 indicated 100% adhesion to the substrate; in all cases breaking of concrete took place (at a tensile stress of 1.6- 2.1 MPa). The good adhesion of Zn layer follows indirectly from its low consumption. The assessment of blistering was made visually.

7. Conclusions

A zinc coating on concrete can serve both as a sacrificial anode and as impressed current anode. Experiments indicate that the level of cathodic polarisation sufficient for CP of concrete reinforcement may be achieved at a low protection current density of approximately 2 mA m-,.

The use of zinc as an anode material is favoured by its low cost, ease of application and significant resistance in alkaline environments. Porous liquid in fresh concrete has a pH about 12.5 with a tendency to decrease (due to carbonation), and zinc shows minimal corrosion in the pH range 7-12 in presence of CO,. Additional advantages of making zinc coatings on concrete surfaces are an increase in leak tightness and improvement of aesthetic appearance. Due to the porosity and high development (surface area) of a concrete surface, a sprayed zinc layer shows very good adhesion to the substrate.

The tests in this work were conducted with the reinforced concrete specimens completely and constantly immersed in aqueous sodium chloride solutions. It is recommended that the performance of zinc anodes should be in field conditions in which alternating wetting and drying of concrete resulting in increased concentration of chlorides can occur.

The work was conducted in the framework of our own research Contract No. 0 12 846lT 105.

References

1. A. Pourbaix and S. Cargo, Management of corrosion of reinforced concrete in the Channel Tunnel, Proc. 12th Int. Corros. Congr., VolSA., pp.3314-3331, Houston, Tx, USA, Sept. 1993. 2. J. Bennet and C. Firlotte, Zinc/ hydrogel system for cathodic protection of reinforced concrete structures, Mater. Perform., 1997, 36 (3), 14-20. 3. J. Bennet, Corrosion of reinforcing steel in concrete and its prevention by cathodic protection, Anti-Corrosion, 1986, 33 (ll), 12-15. 4. R.Baboian, Environmental conditions affecting transportation infrastructure, Mater. Perform.,

5. B. S. Wyatt, Cathodic protection of steel in concrete, Corros. Sci., 1993,35 (5-8), 1601-1635. 6. D. Miller, Economic impact of corrosion of steel in concrete, Proc. 3rd Biennial Corrosion Seminar, The Atlanta Section of NACE, Atlanta, Ga, USA, May 1986. 7. M. Pourbaix, Atlas of Electrochemical Equilibria i n Aqueous Solutions, pp.406413. Pergamon Press, Oxford, 1966. 8. R. Brousseau, M. Ainot and B. Baldock, Laboratory performance of zinc anodes for impressed current cathodic protection of reinforcing concrete, Corrosion, 1995,51(8), 639-644. 9. B. S. Covino, Jr., S. J. Bullard, G. R. Holcomb, S. D. Cramer, G. E. McGill et al., Corrosion, 1997, 53 (5), 399411.

1995,34 (9), 48-52.


12 Lifetime Extension of Thermally Sprayed Zinc Anodes for Corrosion Protection of Reinforced

Concrete Structures by Using Organic Top - c o a t ing s

J. SPRIESTERSBACH, A. MELZER, J. WISNIEWSKI, A. WINKELS and M. KNEPPER*

Grillo-Werke AG, Duisburg, Germany *OSU Maschinenbau GmbH, Duisburg, Germany

ABSTRACT Structural damage in concrete structures caused by corrosion is widespread and demands comprehensive repair work. The additional installation of an active corrosion protective systems for reinforced concrete structures that are located in unfavourable conditions is imperative. Thermally sprayed coatings serving as anodes have been adapted from cathodic protection of steel. These systems have gained attention as they offer advantages in efficiency and lower cost. Thermally sprayed zinc coatings are applied to new steel reinforced concrete structures or those which are subject to rehabilitation. In this contribution, the capability of various systems is examined in field tests in a marine structure in the Arabian Gulf and in different laboratory tests under natural and accelerated conditions.

1. Introduction

Significant advances have been made in the development of sacrificial anode cathodic protection systems to mitigate steel reinforcement corrosion in the past two decades. The corrosion protection of the reinforcing steel in reinforced concrete structures is usually given by the alkalinity of the electrolyte accommodated in the pores of the concrete (pH > 12.5). This pH level leads to passivation of the steel surface which suppresses the corrosion of the steel. Unfavourable environmental conditions or deficiencies in the execution are often the cause for corrosion damage which in many cases leads to extensive reconstruction and repair work worldwide.

Spalling of the concrete cover in steel reinforced concrete structures results from the fact that the corrosion products of steel take up five times the volume of the steel. Various environmental parameters (temperature, humidity), but also concrete technology parameters (type of cement, water-cement ratio, additives) have a decisive bearing on the durability of the steel passivation [l-31. Corrosion mechanisms in concrete are carbonation and chloride contamination (exceeding the critical chloride content). Corrosion of steel reinforcement in concrete due to the penetration of chloride ions from sea water affects many structures located in marine environment.



This is also the case for inland structures effected by seasonal chloride loading (e.g. de-icing salt application during winter).

2. Active Corrosion Protection

In the case of active corrosion protection, like cathodic protection, the surface to be protected remains completely or partially exposed to the aggressive medium; there is, however, active intervention in the corrosion process. Electrons originating from a base metal - the galvanic sacrificial anode - are offered to the steel to be protected. This is realised by connecting the sacrificial anode and the reinforcement bars through an electron conducting material. The electric circuit is then closed through the electrolyte within the concrete. Thus a current flow is realised: the negative ions, e.g. chloride and hydroxide, migrate to the anode and consequently away from the reinforcement. The application of galvanic anode systems requires a minimum current flow between reinforcement steel and anode.

In cathodic protection systems working with impressed current the electrons to protect the steel reinforcement are taken from the negative pole of a direct-current supply. Titanium meshes are mostly used as the anode which has to be embedded in a fresh layer of sprayed concrete. High cost of material (titanium) and the need for embedding the titanium mesh led to the development of zinc anodes sprayed on top of the concrete surface. The zinc anode is not only able to serve as an anode in impressed current system but also as an anode in galvanic systems.

3. Thermally Sprayed Zinc Coatings as Anodes

Galvanic cathodic protection (GCP) systems which use sacrificial anodes, have recognised advantages of simplicity and reliability, and have recently become available as a viable alternative to impressed current CP systems (ICCP) [4]. Unlike ICCP, galvanic GCP systems require no extensive wiring or conduit, and no power supplies.

Thermally sprayed zinc anodes were first investigated for use on concrete in the USA in 1983 [5]. In principle, sprayed zinc coating can be applied in three different cathodic corrosion protection systems:

Galvanic cathodic protection (GCP) by sprayed zinc coatings for repair work without re-profiling. In this system concrete excavations in which the reinforcing steel is partially uncovered are not filled with repair mortar and the initial concrete surface is not restored. With this method the zinc layer is directly sprayed on to the exposed steel and on to the concrete surfaces.

Galvanic cathodic protection (GCP) system by sprayed zinc coatings with reprofiling. By this system the application of the zinc coating is possible when either no spalling of the concrete has occurred or spalls have been repaired. This variant allows measurement of the current between the electrodes. A conditional current regulation is also possible through the arrangement of


Lifetime Extension o fz inc Anodesfor Protection of Concrete Structures Using Organic Top-coatings 111

resistances. This variant offers the option of a retrofit with an additional current supply so that installations of this type can also be operated as impressed current systems.

3. Impressed current CP-system (ICCP) with sprayed zinc coatings. In this system the concrete reprofiling is required as a first step. The cathodic protection effect does not arise from the potential difference between the zinc anode and the protected reinforcing steel but from the feeding of an electric potential by means of an appropriately arranged power supply.

The substantial differences of the three above-mentioned systems are compared in Table 1. Figure 1 compares the application of the sprayed zinc anode operating in a galvanic system and in an impressed current system.

In principle, an investigation has to be made for every application to determine whether an external power supply is required. Optimal conditions for the operation without impressed current are in regions without long dry periods, such as coastal areas or tropical regions.

4. Thermal Spraying of Zinc on Concrete as Anode

Thermally sprayed coatings of zinc or zinc alloys can be produced by either of the techniques widely known as flame (acetylene-oxygen or propane-oxygen) spraying

Table 1. Characteristics ofdifferent cathodic carvosion protection concepts [61



galvanic sacrificial zinc anode in impressed zinc anode current system ,++

Fig. 1 Active corrosion protection systems: galvanic impressed current.

and electric arc spraying. In the wire arc spraying process (Fig. 2) two wires which serve as electrodes are fed together in a gun. An arc forms between the two wires which causes their ends to melt. The melted droplets are subsequently accelerated with an atomising gas flow in the direction of the substrate. On impact at the substrate surface the droplets solidify and adhere both to the substrate and to each other forming a coating [7] .

An acetylene-oxygen flame is mostly used for wire flame spraying (Fig. 3) in order to melt off the end of a wire which is introduced into this flame. The individual melted droplets are subsequently accelerated with an atomising gas in the direction of the substrate surface.

Today, arc spraying has gained greater attention for processing zinc as it offers advantages in efficiency and lower costs. In general, equipment for arc spraying is more expensive than it is for flame spraying. However, higher investment costs can rapidly be absorbed by the higher efficiency and lower operating cost of the arc spray system [S].

The concrete surface must be pre-treated before a zinc anode can be applied. The surface has also to be cleaned and roughened in order to support mechanical adhesion of the layers. Additionally the surface is heated immediately prior to zinc spraying in order to remove any existing residual humidity from the surface layer zone which would weaken the adhesion. The application of the zinc sprayed coating is accomplished in several layers. A mechanically and electrically connected coating is generated with sufficient adhesion to the concrete surface. The coating thickness of


Lifetime Extension of Zinc Anodesfor Protection of Concrete Structures Using Organic Top-coatings 113

Fig. 2 W i r e arc spraying according to EN 657,

the zinc anode is variable and usually ranges between 300 and 500 pm. The adhesion of the zinc coating normally ranges between 1.5 and 3.5 MPa. Special appliances and methods of measurement also permit the inspection of the mechanical stability of the zinc coating on concrete.

Studies carried out in the USA have shown that the service life of a thermally sprayed zinc anode can last up to 20 years if the parameters are correctly chosen and after a correct analysis of the structure and environmental conditions [9].

Fig. 3 Wiref lame spraying according to EN 657.



5. Lifetime Extension by using Organic Top-coatings

The lifetime of the zinc coating can be enhanced if an organic top-coating is applied to the arc sprayed zinc coating (Fig. 4). This system is applicable for concrete structures and was developed and modified by the GRILL0 company in 1997 for aggressive environmental conditions.

The advantage of the organic top-coating is that the zinc coating is not in direct contact with the atmosphere and thus is no longer subject to self corrosion. Therefore, zinc consumption only takes place at the interface between the zinc coating and the concrete. It is calculated that the zinc consumption can be reduced to 50% (Fig. 5). The life time of the sprayed zinc anode can be calculated according to Faraday’s 2nd law.

In laboratory tests several combinations of sprayed zinc coatings and organic top- coatings were evaluated to determine the best system configuration of anode material. Much of this work was focused on the adhesion of the system to the concrete, and on the behaviour as anodes. All systems were tested under accelerated conditions e.g. in the salt spray test. It was observed that the right choice of organic top-coating is of high importance for the adhesion strength of the system and the behaviour as anode material. Following this work, several concrete specimens were coated and exposed to the atmosphere (Fig. 6). To simulate marine conditions these concrete blocks were contaminated with NaC1. Since the installation permanent current and static potential measurements have been carried out and the depolarisation curves measured. Figure 7

Fig. 4 Arc sprayed zinc coating with organic top-coating.



u witboutorganictopcoating I 1 /

with organic topcoating A i I ife time

Fig. 5 Scheme of lifetime extension by organic top-coating.

Fig. 6 Sprayed zinc coated concrete specimen wi th organic top-coating.



shows the depolarisation curve of a sprayed zinc coated concrete specimen with organic top-coating. The instant-off measurement was carried out six months after installation and shows the potential decay after disconnecting re-bar and zinc coating. The potential decay is a criterion for the quality of the cathodic protection. Usually the ‘100 mV criterion’ has to be fulfilled (potential decay > 100 mV - the first 4 h after disconnecting re-bar and zinc anode). As shown in Fig. 7 the 100 mV criterion was already fulfilled after 1 h, which indicates the proper functioning of the GCP system.

6. Thermally Sprayed Zinc CP-system in the Arabian Gulf

A galvanic cathodic protection system using are-sprayed zinc anodes was installed by GRILL0 on steel reinforced concrete structures in the marine atmosphere of a harbour location in the Arabian gulf. The structures showed signs of severe corrosion. The structures were repaired at the end of 1997 and protected from corrosion by the installation of an arc sprayed galvanic zinc anode system. Additionally, an organic top-coating was applied to the arc sprayed zinc coating. The following repair work was carried out on each structure:

removal of loose concrete;

grit blasting of the corroded steel reinforcement;

installation of reference cells and electrical contacts;

Sprayed zinc coating with organic top-coating 700 -

11111 I I II I 111. I I 111 I I

600 ...................................................................................................................................................................................................

.................................................................................................... - zinc coating potential .....

- re-bar potential .....

.......................................................................................................

.....................................................................................................

...................................................................................................... I

300 ..........................................................................................................................................................................................................

250 +

time [12 h]

Fig. 7 Depolarisation curve ofa sprayed zinc coated concrete specimen with organic top-coating (instant-off measurement).



reprofiling of the concrete structure;

arc spraying of the zinc anode; and

application of the organic top-coating.

In order to guarantee and control the functioning of the installed corrosion protection system a monitoring system was installed. Manganese dioxide reference electrodes were installed for potential measurements. Static potentials were determined by operating the monitoring system. Figure 8 shows one of the galvanic protected structures one and a half year after installation of the CP-system.

Over one and a half year after installation of the corrosion protection system no sign of rebar-corrosion could be observed. The collected monitoring data since the installation suggest that the steel rebars are adequately protected from corrosion by the installed corrosion protection system. The obtained values of the static potential measurements indicate that there is no sign of corrosion. All measured potentials in April 1999 are more positive than -520 mV (Fig. 9). Possible corrosion can be expected at the earliest for potentials more negative than -800 mV measured with respect to a manganese dioxide reference electrode. Insignificant deviations of the static potentials with time are influenced by the climate and should not be considered as critical. By considering the curves of the static potentials measured at six different locations of the structure a stabilisation of the potential values can be observed.

Fig. 8 Concrete structure i n the Arabian G u l f o n e and a halfyears after repair (1999).



Static potentials (1 6 month period)

month

Fig. 9 Static potentials measured on six different locations of the structurt wi th respect to a manganese dioxide reference electrode.

7. Conclusions

Sprayed zinc coatings do not present any significant limitations with regard to their applicability in comparison with other cathodic protection variants for concrete, i.e. they are practically always suitable whenever structural elements are to be protected by cathodic protection. Galvanic corrosion protection with sprayed zinc coatings has the significant advantage in comparison with the other cathodic protection variants for concrete, that a reprofiling of the concrete surfaces is not absolutely necessary and that no electrical installations have to be carried out, except for monitoring areas.

Thermally sprayed zinc anodes can be renewed very easily after being consumed. Even after the complete removal of the zinc coating, the anode can be easily replaced by spraying a new zinc coating on the concrete surface.

The service fife of a thermally sprayed zinc anode can last up to 20 years or more if the parameters are correctly chosen and after a correct analysis of structure and environmental conditions. By applying organic top-coatings to the sprayed zinc coating, the lifetime can be enhanced considerably.


Lifetime Extension of Zinc Anodes for Protection of Concrete Structures Using Organic Top-coatings 119

References

1. P. Schiell, RILEM; Technical Committee 60-CSC: Corrosion of Steel in Concrete (RILEM Report). Chapman and Hall, New York, 1988. 2. U. Niirnberger, Korrosion und Korrosionsschutz in Bauwesen. Bauverlag (1995), Wiesbanden; Berlin, ISBN 3-7625-3199-4. 3. M. Raupach, Zur chloridinduzierten Makroelementkorrosion von Stahl in Beton, in Sckriftenreike des Deutscken Aussekusses fur Staklbeton (1992), Nr. 433, Beuth-Verlag, Berlin. 4. J. Bennett, Galvanic Cathodic Protection of Reinforced Concrete Using Surface Applied Zinc Anodes, in Proc. ICCRRCS, 1998, Orlando, F1. Publication No. FHWA-SA-99-014. 5 . L. S. Tinnea, Field Performance of Sprayed Zinc Cathodic Protection Anodes, in Proc. 15th Int. Thermal Spray Conference, 1998 (C. Coddet, ed.). 6. J. Spriestersbach, M. Knepper and M. Gamroth, Zinkschichten zum Korrosionsschutz von Stahlbetonbauwerken, Erzmetall, 1998, 51 (4), 291-298. 7. M. Knepper, J. Spriestersbach, How to successfully battle corrosion: Thermally sprayed coatings of zinc and zinc alloys, in Proc. ITSC, Shanghai, Nov. 18-21,1997: Surface Engineering Towards The 21st Century (X. Binishi et al., Eds). China Machine Press, Beijing 1997, pp. 384- 391. 8. R. Porter, Corrosion Resistance of Zinc and Zinc Alloys. Marcel Dekker, Inc., New York, 1994. 9. A. Sagues and R. Powers, SHRP-S-337: Sprayed Zinc Galvanic Anodes for Concrete Marine Bridge Substructures, 1993 (Strategic Highway Research Program, Washington).


13 Practical and Economic Aspects of Application of

Austenitic Stainless Steel, AISI 316, as Reinforcement in Concrete

0. KLINGHOFFER, T. FR0LUND, B. KOFOED, A. KNUDSEN", F. M. JENSEN* and T. SKOVSGAARDt

FORCE Institute, Park All6 345, DK-2605, Brandby, Denmark *RAMBOLL, Virum, Denmark +Arminox, Viborg, Denmark

ABSTRACT

Reinforced concrete used for housing and industrial construction is often damaged due to corrosion of the reinforcement. The total cost in the EC for repair of damages caused by corrosion may be estimated from the cost in the UK on highways alone to be around 50M ECU per year.

A way to lengthen the lifetime of a structure is to use corrosion resistant reinforcing materials, e.g. stainless steel. The intelligent use of stainless steel, which means combining with traditional carbon steel in locations exposed to very corrosive environments, can be a very cost-effective option when considering different rehabilitation methods.

However, most civil engineers have an unfounded fear of using stainless steel and carbon steel together in the same concrete structure. For this reason, the behaviour of the austenitic stainless steel, AISI 316, in connection with carbon steel has been evaluated in order to study the consequences of galvanic coupling for corrosion reinforced concrete structures. The experimental study includes results from different concrete samples, in which AISI 316, stainless steel, has been combined with carbon steel in the proportions that are foreseen for on-site applications. These results include measurements of the macrocouple current between stainless steel and carbon steel during exposure to accelerated ingress of chloride. Additionally, measurements of electrochemical potentials and corrosion rate of the macrocouple were made.

The obtained results show that galvanic coupling with stainless steel results in an enhanced corrosion rate of the active carbon steel in a chloride-contaminated solution. A Life Cycle Cost calculation, based on practical cases of repaired bridge columns, has confirmed that the intelligent use of stainless steel in combination with carbon steel is very cost-effective.

1. Introduction

Stainless steel derives its corrosion resistance from a naturally occurring chromium- rich oxide film that is present on its surface. This invisible film is inert, tightly adherent to the metal, and -most importantly - in an environment where oxygen is present



even at relatively low levels, the film reforms instantly if the surface is damaged [l]. There are, however, aggressive environments (e.g. those with carbonation or ingress of chlorides) that can give rise to breakdown of this passive layer, resulting in corrosion of the unprotected surface. When deterioration has developed to a given point, rehabilitation measures are required. Among the various rehabilitation options, modern stainless steel has become an attractive alternative compared to traditional methods with carbon (unalloyed) steel, epoxy coatings, corrosion inhibitors, cathodic protection, etc. [2].

Stainless steel is becoming cheaper, although still 5-8 times more expensive than uncoated carbon steel.

Therefore, an economical and technically attractive approach may be to substitute carbon steel with stainless steel in critical areas, such as the lower section of a column on a highway bridge exposed to de-icing salt, the splash zone for coastal structures, or an edge beam on a highway bridge. This is called 'intelligent use' (Fig. 1).

2. Practical Aspects

The manageability of stainless steel on-site is comparable to normal carbon steel. Therefore, no special precautions need to be taken when using stainless steel. However, due to the high cold-working properties of stainless steel, somewhat higher bending forces are necessary. For repairs comprising selective replacement of carbon steel with stainless steel in a limited area, three methods can be used to connect the stainless steel and carbon steel reinforcement: traditional unwelded laps, welded laps and mechanical couplers.

The diameter of the main reinforcement is typically in the range of 15 to 40 mm, requiring a minimum grip length (- lap length) of more than 50 cm at both ends. Therefore, unwelded lap joints are not a very competitive option, since an additional 1-1.5 m of concrete is to be removed.

Stainless steel bars are weldable on-site, whereas the weldability of the existing carbon steel bars is often questionable - and in some cases unknown. Therefore, welding on-site may not always be possible.

Bridge cross section

Potential areas where stainless steel may be used

Bridge longitudinal section

Fig. 2 Potential areas where stainless steel m a y be used intelligently for repair and in new structures as well.


Practical and Economic Applications of A l S I 326 as Reinforcement i n Concrete 123

The corrosion resistance of stainless steel is lowered by welding and by contamination with iron deposits from tools used in the handling [3]. However, problems may be avoided by careful post-treatment, e.g. sandblasting and pickling.

The use of mechanical stainless steel couplers between carbon steel bars and the stainless steel reinforcement is an alternative to welding. Some of these require that a thread be made on the existing carbon steel, which may be both difficult and time- consuming on-site. Another option is to use couplers that mechanically lock the bars to the coupler, thereby achieving strengths higher than the yield strength of the rebar itself. By using mechanical steel couplers, no additional lap length is required. The mechanical couplers can be made from stainless steel. The example in this paper assumes the use of the mechanical couplers described above.

3. Corrosion Aspects

Stainless steel that is freely exposed to sea water may, if in galvanic contact with a less noble metal such as carbon steel, initiate a galvanic type of corrosion of the latter. The corrosion rate will depend on the area ratio between the carbon steel and the stainless steel. The otherwise slow, cathodic oxygen reduction at the stainless steel surface is a catalyst for bacterial slime, which forms after a few weeks in sea water.

When stainless steel is cast into concrete, however, the cathodic reaction is a very slow process, since no such catalytic activity takes place on a stainless steel surface 141. A research project conducted at the FORCE Institute [5] has indicated that the cathodic reaction is inhibited on stainless steel embedded in concrete, as compared to the cathodic reaction on carbon steel reinforcement in galvanic contact with corroding carbon steel.

Later publications by Pedeferri et nl. [6] and Jaggi e t ul. [7] also provide results which confirmed the above findings.

Consequently, the connection between stainless steel and carbon steel should not promote significant galvanic corrosion. As long as both metals are in the passive condition, their potentials will be more or less the same when embedded in concrete. Even if there should be minor differences in potential, both carbon and stainless steel can be polarised significantly without any serious risk of corrosion, as their potentials will approach a common value without passage of significant current. Therefore, assuming the correct use of stainless steel, the two metals can be coupled without any problem in all positions where chloride ingress and subsequent corrosion might occur.

This behaviour, and the fact that stainless steel is a far less effective cathode in concrete than carbon steel, makes stainless steel a useful reinforcement material for application in repair projects. When part of the corroded reinforcement, e.g. close to the concrete cover, is to be replaced, it could be advantageous to use stainless steel instead of carbon steel. Since it is a poor cathode, the stainless steel should minimise any possible problems that may occur in neighbouring corroding and passive areas after repair.

At the same time, it is very important for the intelligent use of stainless steel that it be combined with carbon steel in proportions that guarantee both an optimal


124 Corrosion of Reinforcement in Concrete: Corrosion Mechanisms and Corrosion Protecfion

performance and cost-effective solution. For this reason, tests including probable volume combinations of between stainless steel and carbon steel aimed for repair of damaged highway and coastal bridges have been carried out.

4. Experimental Tests of Corrosion Resistance

The aim of the experiments described in this paper is to define objectives for use of stainless steel in the repair of corroding reinforcement. The galvanic couple formed between the passive stainless steel and the existing carbon steel, which in some cases is passive and in some cases corroding, will be studied in order to prove that the use of stainless steel for this purpose might even have a beneficial effect.

All test samples have the dimensions 300 x 170 x 70 mm and are cast from an ordinary Portland cement concrete of water / cement (w / c) ratio = 0.5 and without addition of fly ash and microsilica. All samples contain 5 reinforcement bar pieces in full sample length. These bars are either made of carbon steel or austenitic stainless steel (AIS1 316). Additionally, the test samples contain two small pieces of the austenitic stainless steel or carbon steel, corresponding to 5-10% of the total steel volume. The 5% and 10% chosen for the samples represent the percentage of stainless steel foreseen for use in application on-site, All bars have a diameter of 6 mm. A reference electrode of the MnO, type is embedded in each sample. A total of 10 concrete samples divided into four groups was cast and later used for measurements of galvanic current, electrochemical potentials and corrosion rate.

Figures 2 and 3 show the samples and the principles of measurement. One month after casting, all samples were exposed in a concentrated solution of

NaCl(l65 gL-' NaC1) with addition of Ca(OH),. In order to accelerate the chloride ingress, the exposure was a cycle of two days wetting in the NaCI solution and five days drying in a laboratory atmosphere [8]. The following measurements were conducted:

Macrocouple current between stainless steel connected to carbon steel. The macrocouple current is measured by means of a specially constructed Zero Ohm Ammeter.

Electrochemical potential of the abovementioned macrocouple against an embedded MnO, reference electrode.

Corrosion rate of all rebars by means of a galvanostatic pulse method.

5. Results of Corrosion Experiments

5.1. Macrocouple Current

The rapid potential drop in the corroding metal causes a significant increase in macrocouple current when the corrosion process starts. Thus, an electromotive force between two metals with different electrochemical potentials is created, resulting in


Practical and Economic Applications of AlSI 316 as Reinforcement in Concrete 125

Fig. 2 Sketch of test samples.

(Group 1 and 2)

Exposed surface (Group 3 and 4)

Fig. 3 Principle of exposure of test samples.



25.0

E 15.0 - 2 2 10.0 - L

a, Q -

Exposure time (days)

Fig. 4 Macrocouple cuvrent as function of exposure time.

the electrical current (corrosion current) flowing between them. Figure 4 shows an example of macrocouple current measurements made on one of the test samples represented in group 3. This sample had been cast from ordinary Portland cement with w / c ratio = 0.5 and without addition of fly ash and microsilica. Austenitic stainless steel, AIS1 316, represents 10% of the total steel volume in the sample. During the exposure and measurements of the macrocouple current stainless steel was electrically connected to the carbon steel. At the beginning of the experiment the measured current was very low due to passivity of both stainless steel and carbon steel. After 28 days of exposure in the concentrated NaCl solution this current rapidly increased due to pitting corrosion occurring on the carbon steel.

The increase in macrocouple current after initiation of corrosion depends on the type of passive material (cathode). The current will be much lower when corroding carbon steel is connected to a passive stainless steel, compared to the current registered between active and passive bars of carbon steel. For this reason, the increase in corrosion rate in carbon steel due to the galvanic coupling with stainless steel will be significantly lower than in the case of carbon steel.

The experimental results from measurements performed on sample 3 representing group 1 (Fig. 5) confirm this behaviour. When the current was measured between the carbon steel rebar starting to corrode and a small rebar (5%) of carbon steel that was still passive, a current density value of approx. 4.3 pAcm-2 was registered. If the same corroding carbon steel rebar was connected to the small rebar (5%) of stainless steel, the measured current density value was reduced to only 0.27 pAcm-*. This means a reduction in current density by a factor of approximately 15, which will result in a smaller decrease in corrosion rate.

The above numbers are typical of values measured on the remaining 9 samples included in these experiments.

The high cathodic overvoltage on stainless steel means that when stainless steel is polarised to a negative potential as a result of galvanic coupling with corroding carbon steel, it can produce a current density several times lower than the passive carbon


Practical and Economic Applications ofAlSI316 as Reinforcement in Concrete 127

5 4.5

4 3.5

3 2.5

2 1.5

1 0.5 0

Stainless steel Passive carbon steel

Fig. 5 Macrocouple current for stainless steel and passive carbon steel.

steel can generate [9]. Thus, the consequence of coupling with stainless steel is generally negligible, since passive areas of carbon steel always surround the area where corrosion takes place.

This behaviour has been proved in the present investigations. In one of the samples with small bars of stainless steel (sample No. 5 representing group 2 where stainless steel represents 5% of the total steel volume), the remaining bars of carbon steel, which had started to corrode, were coupled to the still passive small bars of carbon steel from another sample. This resulted in a remarkable increase in macrocouple current. This current started to decrease when the primary connection between carbon steel and stainless steel was re-established. The results of this test are shown in Fig. 6. This procedure has been repeated on two more samples (sample No. 1 and sample No. 7) with similar results.

As a consequence of these findings, stainless steel is considered to be an even better reinforcement material than the usual carbon steel for use in repair projects where part of the corroded reinforcement is to be replaced. Because it is a poor cathode, the stainless steel will minimise eventual problems that could occur in neighbouring corroding and passive areas after repair.

6. Determination of Corrosion Rate by Means of Galvanostatic Pulse Method

The galvanostatic pulse method is a transient polarisation technique working in the time domain. A short-time anodic current pulse is imposed galvanostatically on the reinforcement from the counter electrode placed on the concrete surface [lo-121. The applied current is usually in the range of 10 to 200 yA and the typical pulse duration



Macrocouple current - Sample 10

23 43 1w 123 im P Time (days)

Fig. 6 Influence of cathode material on the macrocouple current (see p.127).

is up to 10 s. The reinforcement is polarised in the anodic direction, i.e. compared to its free corrosion potential.

The resulting change in potential is dependent on the state of corrosion in the reinforcement, and can be expressed by means of polarisation resistance, ohmic resistance, double layer capacitance and the impressed current. Thus, it is possible to calculate the polarisation resistance and, moreover, the corrosion current. When the area of polarised reinforcement is known, it is also possible to calculate the instantaneous corrosion rate from the values of the corrosion current.

In the case of the present investigation, the area of polarised reinforcement was known exactly. However, the small size of the rebar in the investigated samples caused another problem. Even the smallest current that could be applied by means of the galvanostatic pulse device was found to be too big to achieve the optimal polarisation conditions (reinforcement should only be polarised to a maximum 20 mV from the free corrosion potential when the ohmic resistance is subtracted).

Therefore, the rather high current applied for poIarisation influences the obtained results. This high current has a special effect on the values of the corrosion rate determined for passive rebars (mostly stainless steel). These values are higher than could be expected for steel in the passive condition, but, nevertheless, the calculated corrosion rate values are much lower for passive stainless steel than for actively corroding carbon steel.

Experimental data from on-site measurements has shown that the average corrosion rates determined by means of the galvanostatic pulse equipment underestimates the real corrosion rate by a factor of 5-10, or even more, in the case of chloride-induced localised corrosion (’pitting’), where the active corroding area is much smaller than the confined area of the reinforcement used for the calculation.

Table 1 shows values of instantaneous corrosion rate calculated from galvanostatic


Practical and Economic Applications of AIS1 316 as Reinforcement in Concrete 129

Table 1. Average corrosion rate values calculated by means of galvanostatic pulse measurements and actual corrosion rate values obtained by correction of the corroding surface area

Sample

3

6

9

Reinforcement material (l)

1- Stainless Steel (5%) Stainless Steel (10%) 2- Carbon Steel 3- Carbon Steel 4- Carbon Steel 5- Carbon Steel 6- Carbon Steel

1- Carbon Steel (5%) Carbon Steel (10%) 2- Stainless Steel 3- Stainless Steel 4- Stainless Steel 5- Stainless Steel 6- Stainless Steel

1- Stainless Steel (5%) Stainless Steel (10%) 2- Carbon Steel 3- Carbon Steel 4- Carbon Steel 5- Carbon Steel 6- Carbon Steel

Free Corrosion Potential (mV vs MnO,)

-263 -223 -1 75 -162 -237 -185 -180

-575 -553 -280 -278 -276 -270 -274

-199 -262 -554 -570 -643 -586 -625

Average corrosion rate (pm/year) (*)

2.0 1.1 7.7 7.6 3.3 3.9 2.9

78 19 1.5 1.4 1.4 1.6 1.4

7.9 0.7 18 56 42 34 80

Actual corrosion rate (ym/year) (3)

2.0 1.1 385 23 9.8 1.9 2.9

233 75 1.5 1.4 1.4 1.6 1.4

7.9 0.7 359 845 422 508 161

(')Numbers before the reinforcement material indicates which bars have been galvanically connected. The bar with no number has not been connected.

(2)Values of corrosion rate calculated by means of galvanostatic pulse measurements without correction for the actually corroding surface area determined by means of the visual inspection.

(3)Values of corrosion rate after correction for the actually corroding surface area determined by means of the visual inspection.

pulse measurements without correction for the actually corroding surface area, as determined by means of the visual inspection. Based on the visual inspection the surface area of the corroding reinforcement is determined and then used for calculation of the actual corrosion rate. As expected, from experimental experience from on-site measurements, the values for the corroding carbon steel are much higher than values calculated by means of the galvanostatic pulse measurements. Laboratory experiments are therefore in good agreement with on-site measurements.



7. Total Cost and Updated Life-cycle Cost Analysis (ULCCA)

The intelligent use of stainless steel is evaluated by analysing an example of deteriorated, standard reinforced concrete (RC) columns of a coastal bridge in the splash zone.

The rehabilitation of a RC-column can be analysed separately from the rest of the bridge. This is possible as the administration, inspection, maintenance, rehabilitation, etc. are normally carried out independently of the rest of the bridge.

The two types of repair will be analysed using the net present value method, taking all costs into account (direct and indirect) from the time of repair and onwards. This updated life cycle cost analysis (ULCCA) will consider all relevant financial and technical aspects. The 'U' for 'updated' is added to 'LCCA', since the life cycle cost analysis starts when the structure is e.g. 30 or 40 years old and showing signs of serious deterioration.

The comparison of different strategies - with and without the use of stainless steel - using the net present value method is carried out in order to determine the repair strategy that is economically optimum for society as a whole, given the premises at the time of decision. This includes taking all costs into consideration; repair, maintenance, administration and indirect cost to society (traffic alterations). This is a generally accepted method approved in Denmark and several other countries [13].

8. Example of Repair of Reinforced Concrete Columns on a Coastal Bridge (Splash Zone)

An ongoing research project financed by the Danish Road Directorate shows that some fairly new coastal bridges need repair in the splash zone due to chloride-induced corrosion [ 141.

In this example, a 30-year-old coastal bridge with 12 columns with severe corrosion of the reinforcement in the splash zone is considered. The repair requires replacement of the outer layers of the steel reinforcement that had experienced heavy corrosion. The replacement is in the splash zone, i.e. from 0.5-1 m below to 2 m above normal sea level. The repair requires that approximately 50% of the outer layer of reinforcement be replaced in this area. The amount of steel to be replaced is approximately 5% of the total reinforcement steel in the column and foundation.

The three strategies proposed for rehabilitation of the coastal bridge are shown in Table 2 and the corresponding net present value analysis is shown on Fig. 7.

The life cycle cost analysis shows that for discount rates between 5% and 7% the three analysed strategies have comparable net present values. Based on this, it seems that a postponed repair strategy using stainless steel will be cost-optimal. It must be noted here that from experience - and the available data for the extent of deterioration and associated repair - the repair cost for coastal bridges is less than for the repair of highway bridges 1151. This is due to the higher number of repairs performed on highway bridges compared to the number of repairs performed on coastal bridges.


Practical and Economic Applications of AISI 316 as Reinforcement in Concrete 131

Table 2. Description of three repair strategies for coastal bridges

Strategy - Description of repair strategy

1 Repair of all columns using carbon steel after 1 year. The repair is done over 2.5-3 m of each column involving the breaking up of the concrete to behind the reinforcement and replacement of 50% of the reinforcement. The column in this example has two layers of reinforcement and only the outer layer is likely to corrode. At 20 and 40 years minor repair is required in the columns.

2 Repair of all columns using carbon steel after 10 years. The repair is done over 2.5-3 m of each column involving the breaking up of the concrete behind the first layer of reinforcement and replacement of 80 % of the reinforcement. At 25 and 45 years minor repair is required on the columns.

Repair of all columns using stainless steel after 1 year. Same repair as strategy 1, i.e. only the outer layer of old carbon steel reinforcement is replaced with stainless steel. At 20 and 40 years minor repair is required in the columns.

3

9. Case from Mexican Gulf - The Most Convincing Argument for Summary

The 2 100 metre long concrete pier in the Port of Progresso, Yucatan, Mexico was constructed between 1937 and 1941. The concrete pier has 175 spans of 12-m lengths and consists of massive columns and arches. Due to the harsh environmental exposure of the pier (hot and humid marine environment), it was decided to use stainless steel reinforcement (AISI 304) in selected areas of the pier.

Now, almost 60 years after construction, the pier has been investigated by means of visual inspection and both non-destructive and destructive techniques [ 161,

R 1900 v) 3 1700 -.- Standard concrete 0 repair (strategy 1) 8 1500 r Y * Postponed concrete 3 1300 repair (strategy 2) m > 1100 +- Concrete repair with

e! Q 700 a, = 500

- stainless steel (strategy 3) c

C 8 900

4-

0 1 2 3 4 5 6 7 8 9 10 discount rate [%]

Fig. 7 Example of the coastal bridge. Net present values (50 years remaining lifetime) for differenf discount rates.



No serious sign of corrosion of the stainless steel reinforcement embedded in the concrete was found. However, corrosion was detected on the freely exposed reinforcement (no cover), as could be expected for this grade of stainless steel in a marine environment. For reinforcement with a cover larger than approx. 20 mm, there was no significant corrosion on the bars, despite the extremely high chloride contents of up to 1.9% C1- of dry concrete weight. This is at least 10 times of that normally regarded as a critical chloride concentration for the initiation of corrosion of ordinary carbon steel.

For a reinforced concrete structure in marine environment with ordinary carbon steel, the lack of routine maintenance for a 60-year period would in many cases result in serious chloride- or / and carbonation-induced corrosion problems. This is clearly shown by the deterioration of the neighbouring pier located to the west of the inspected pier.

The unambiguous conclusion is, therefore, that the use of AISI 304-grade stainless steel as reinforcement has contributed significantly to the good durability of the Progresso pier.

10. Conclusions

The following conclusions can be drawn based on the experience gained from this work:

The coupling of corroding carbon steel with austenitic stainless steel, AISI 316, is without risk and provides lower corrosion current (corrosion rate) compared to the coupling to passive carbon steel, which always surrounds the corroding areas.

Stainless steel has a higher overvoltage for cathodic reaction of oxygen reduction with respect to carbon steel. Therefore, the increase in corrosion rate on carbon steel embedded in chloride-contaminated concrete due to galvanic coupling with stainless steel is significantly lower than the increase brought about with passive carbon steel.

Welding, which also decreases the chloride threshold value for initiation of corrosion, can destroy the low cathodic activity of stainless steel. For this reason the influence of welding will be further investigated in the future. The influence of cold working processes on the corrosion properties of stainless steel will also be investigated.

However, the evidence obtained so far, shows that carbon steel and stainless steel can be coupled with beneficial results regarding corrosion protection in chloride-contaminated concrete.


Practical and Economic Applications of AlSl316 as Reinforcement in Concrete

11. Acknowledgement

133

Part of this paper was presented at the International Conference on Corrosion and Rehabilitation of Reinforced Concrete Structures, held a t Orlando, Florida, USA, 7-11 December 1998.

References

1. S. R. Kilworth and J. Fallon, Stainless steels for reinforcement, development of the paper, ’Fusion Bonded Coated Reinforcement in the Gulf‘, in Int. Conf on Corrosion and Protection of Reinforced Concrete, Dubai, 1994. 2. British Concrete Society, ‘Guidance on the Use of Stainless Steel Reinforcement’, Technical report No. 51, 1998. 3. U. Nurnberger, W. Beul and G. Onuseit, Corrosion behaviour of welded stainless reinforced steel in concrete, Otto-Graf-Journal, 1993. 4. Stainless Steel in Concrete -State of the Art Report (U. Nurnberger, ed.). Publication No. 18, in the European Federation of Corrosion Series. Published by The Institute of Materials, 1996. 5. Materials for Corrosion Cell Cathodes’, Internal Report by the Danish Corrosion Centre as a part of a report regarding Brite/Euram Contract 102 D. 1990. 6. L. Bertolini, M. Gastaldi, T. Pastore, M. P. Pedeferri and P. Pedeferri, Experiences on stainless steel behaviour in reinforced concrete. Book of Abstracts, EUROCORR’98, The Dutch Corrosion Centre, Utrecht, The Netherlands. Published 1998. 7. S. Jaggi, B. Elsener and H. Bohni, Oxygen reduction on passive steel in alkaline solutions, Book of Abstracts, EUROCORR ’99, GFK, Germany. Published 1999. 8. Nordtest standard, ’Concrete Testing, Hardened Concrete. Chloride Penetration’. NT BUILD 443,1996. 9. L. Bertolini, M. Gastaldi, M. P. Pastore and P. Pedeferri, Experiences on stainless steel behaviour in reinforced concrete, in EUROCORR’ 98, The Dutch corrosion Centre Utrecht, 1998. 10. B. Elsener, 0. Klinghoffer, Frolund, T. Rislund, E. Schiegg and Y. H. Bohni, Assessment of reinforcement corrosion by means of galvanostatic pulse method, in Proc. Int. Conf on Repair of Concrete Structures, Svolvzx, Norway, 1997, pp. 391400. 11. J. Mietz and B. Isecke, in Electrochemical Potential Monitoring on Reinforced Concrete Structures using Anodic Pulse Technique, Non-destructive Testing in Civil Engineering (H. Bungey, ed.). The British Institute of NDT, 1993, 2, pp. 567. 12. K. R. Gowers and S. G. Millard, Corrosion and Corrosion Protection of Steel in Concrete (W. Swamy, ed.). Sheffield Academic Press, 1994, pp. 186. 13. Life-cycle Cost Case Study. River Crossing Highway Bridge (Schafjhausen Bridge, Switzerland). Published by Euro Inox, 1997. 14. E. Stoltzner, A. Knudsen and B. Buhr, Durability of marine structures in Denmark, in Proc. Int. Con. on Repair of Concrete Structures, Svolvm-, Norway, 1997, pp. 59-68. 15. A. Knudsen, F. M. Jensen, 0. Klinghoffer and T. Skovsgaard, Cost effective enhancement of durability of concrete structures by intelligent use of stainless steel reinforcement, in Proc. Int. Con& on Corrosion and Rehabilitation of Reinforced Concrete Structures, Florida, USA, December 1998. 16. Arminox Aps, Internal Report, ’Evaluation of the Stainless Steel Reinforcement of Pier of Progresso, Mexico ‘, March 1999.


List of Abbreviations

The following abbreviations occur in the text and in the Index of contents.

ASTM American Society for MIP Mercury intrusion Testing and Materials porosimetry

MS Microsilica MTBF Mean time between

BFSC Blast furnace slag cement failures (in power supply)

Cathodic protection Czechoslovak National NMR Nuclear magnetic

CP CSN

resonance Standard

OPC Ordinary Portland cement DIN German National

Standard

PVC Polyvinyl Chloride es . Electrochemical

FHWA Federal Highways RC Agency (US) RH

RILEM

GCP

GGBS

GNP

ICCP

IS0

Galvanic Cathodic

Ground Granulated Blast

Gross National Product

Protection

furnace Slag

Impressed current cathodic protection

International Standards Organisation SCE

Reinforced concrete Relative humidity Reunion Internationale

des Laboratoires d’Essais et de Recherches sur les Materiaux et les Constructions (International Union of Testing and Research Laboratories for Materials and Structures)

Saturated (KC1) calomel electrode

(U)LCC(A) (Updated) Life cycle costs WAC Water absorption (Analysis) coefficient

w / c Water / cement ratio


This page has been reformatted by Knovel to provide easier navigation.

INDEX

Index Terms Links

A Acidification of cement

at anodes 103

Anodes

external Zn 103

layered Zn 101

Sacrificial, in patch repairs 94

thermally deposited Zn 109

C Carbonation (of concrete)

of ground floor elements 85

in hydrophobised concrete 83

Cathodic control (see also Oxygen reduction)

of rebar corrosion 13

Cathodic prevention (of corrosion)

around patch repairs 94

experimental evaluation of 95

Cathodic protection (see also Sacrificial anodes)

of concrete reinforcement, principles of 102

of ground floor elements 85

life time consideration of 88

using layer Zn anodes 101

Chloride

induced corrosion of rebar 13 25

hydrophobic treatment and 81

to OH– ratio and pit initiation 31



Index Terms Links

Chloride (Cont.):

penetration into microsilica concrete 35

profiles in exposed cement 36

Corrosion prevention (see also Cathodic

protection, Hydrophobic treatment, Inhibitors)

of rebars

by cathodic protection 85

computer system for 51

by hydrophobic treatment 73

by organic corrosion inhibitors 73

Corrosion protection potential

value in rebar corrosion 46

Corrosion of steel (rebar) in concrete

computerised system for 51

critical factors for inhibition of 25

reasons of 10 13 101

Corrosion rate determination

by e.c. methods 41

by galvanostatic pulse method 127

by mass loss method 41

D De-icing salts (see Chlorides)

F Free chloride

definition of 25

H Hydrophobic treatment

agents for 76

and carbonation 83

chloride penetration and 78



Index Terms Links

Hydrophobic treatment (Cont.):

of concrete for rebar protection 73

durability of 79

I Inhibitors

experiments with 62

use in concrete 61

M Macrocell corrosion

e.c. background to 14

in experimental tests 26

numerical simulation of 18

role of 21

Mass loss data (for rebar corrosion)

vs e.c. data 41

Macrosilica

in concrete 35

O Organic corrosion inhibitors (see Inhibitors)

Organic top coats

on thermally sprayed zinc 109

Oxygen reduction (see also Cathodic control)

in corrosion of steel 3 15

kinetics of 7

on platinum 9

on stainless steel 10

Tafel behaviour of 9



Index Terms Links

P Patch repairs

application of 93

Potentiodynamic polarisation of steel

in NaOH and in synthetic pore solution 4

effect of flow rate on 7

effect of temperature on 5

R Repair procedures

options for 86

patch, with CP 93

S Sacrificial anodes (see Anodes)

Stainless steel rebars

case history 130

corrosion aspects of 123

life cycle cost analysis of 130

oxygen reduction on 10

practical and economic aspects of 121

in repair systems 130

results of corrosion tests with 124

Synthetic pore solutions

composition of 5 26

polarisation tests in 4

Z Zinc anodes (see Anodes)


Documents

European Federation