RISE CBI SWEDISH
CEMENT AND CONCRETE
RESEARCH INSTITUTE
Chloride Ingress and Reinforcement Corrosion - After 20
years’ field exposure in a highway environment
Luping Tang
Dimitrios Boubitsas
Peter Utgenannt
Zareen Abbas
Chloride Ingress and Reinforcement Corrosion - After 20
years’ field exposure in a highway environment
Luping Tang
Dimitrios Boubitsas
Peter Utgenannt
Zareen Abbas
Abstract
Chloride Ingress and Reinforcement Corrosion - After 20
years’ field exposure in a highway environment
This report presents the results from a research project financed by Trafikverket, the Swedish Transport Administration, co-financed by Cementa AB.
In this part of the project work about 35 chloride profiles and ten moisture profiles have been measured from various types of concrete specimens exposed to a de-icing salt highway environment for about 20 years. The non-destructive technique, RapiCor, for corrosion measurement was used to assess the conditions of steel embedded in concrete beams with different types of binder and water-binder ratios. The ClinConc model were used to predict chloride ingress in concrete. Some laboratory test methods and numerical simulations were carried out to study the behaviour of concrete after long-term exposure. The results show that chloride ingress profiles measured after 20 years’ exposure under the de-icing salt highway environment are in general lower than those measured after the similar exposure duration under the marine splash environment.
Non-destructive corrosion measurement by RapiCor instrument is in general in reasonably good agreement with chloride ingress. The corrosion rust observed from the destructive examination verified again that the non-destructive technique RapiCor is a useful tool for detection of ongoing corrosion of steel in concrete.
The ClinConc model in general gives reasonably good prediction of chloride ingress front in the bulk concrete but not in the near surface zone due to the drying-wetting cycles in the highway environment. After a certain modification with the consideration of redistribution of bound chloride in the near surface zone, the model can reasonably well describe the chloride ingress profiles measured from the exposure site.
Key words: Concrete, chloride ingress, corrosion, durability
RISE Research Institutes of Sweden AB RISE Report 2018:66
ISBN: 978-91-88907-10-3 Borås 2018
Content
Abstract ... 1 Content ... 2 Preface ... 3 Summary ... 4 1 Introduction... 62 Concrete Specimens and Exposure Conditions ... 8
2.1 Concrete specimens ... 8
2.2 Exposure conditions ... 10
3 Measurements of Chloride and Moister Profiles ... 11
3.1 Sampling ... 11
3.2 Measurement of chloride profiles ... 12
3.2.1 Effect of binder type ... 12
3.2.2 Effect of water-binder ratio and entrained air ... 14
3.2.3 Effect of exposure environment ... 17
3.2.4 Effect of exposure environment ... 17
3.2.5 Curve-fitted parameters ... 19
3.3 Measurement of moisture profiles ... 21
3.4 Measurement of rapid chloride migration ... 23
4 Corrosion Measurement ...27
4.1 Specimens ... 27
4.2 Technique for corrosion measurement ... 28
4.2.1 Validation of corrosion measurement ... 29
4.2.2 Results from corrosion measurement ... 30
5 Modelling of Chloride Ingress ... 32
5.1 Main input data – RCM-values ... 32
5.2 Other input data... 33
5.3 Results from the previous model ... 33
5.4 Further modification of the model ... 37
6 Concluding remarks ... 44
7 References ... 45 Appendix 1 – Chloride profiles from the Highway 40 field site
Appendix 2 – Moister data from the Highway 40 field site Appendix 3 – Curve-fitted results from chloride profiles
Preface
The report is the second and final part of the research project “Field testing of concrete exposed to highway environment with de-icing salts for 20 years – Frost damage and chloride ingress after 20 years of exposure”. In part one (separate report) results from investigations regarding the frost and salt-frost resistance was presented. In part two (this report) results from investigations regarding chloride ingress and reinforcement corrosion is presented. The project is part of the Swedish Transport Administrations BBT-research program. The authors wish to acknowledge the financial support provided by the Swedish Transport Administration (Trafikverket) and the Cement producer Cementa AB.
The project was coordinated by Chalmers University of Technology (Chalmers) and was carried out in co-operation with RISE Research Institutes of Sweden (RISE), Lund University (LTH) and Gothenburg University (GU). For the work presented in this report the main part of the measurements on the specimens from the field exposure sites were carried out by RISE. Some laboratory tests as well as the modelling of the chloride ingress were carried out at Chalmers. GU mainly contributed to the modelling of chloride binding in concrete. LTH is the main responsible for modelling of frost resistance reported in part one.
Sammanfattning
Frostskador och kloridinducerad korrosion av armeringsjärn är de två främsta utmaningarna vid livslängdsdimensionering av betongkonstruktioner. Dessa skademekanismer är vanligt förekommande i miljöer där tösalter används, till exempel i vägmiljö.
Modeller för hur fukt och klorider transporteras i betong är av yttersta vikt. Detta för att kunna dimensionera betongkonstruktioner på ett ekonomiskt, hållbart och säkert sätt samt göra livslängdsbedömningar på redan befintliga konstruktioner.
Svensk forskning inom områdena frostskador och kloridinducerad korrosion av armeringsjärn ligger i den internationella fronten. Det finns dock fortfarande luckor mellan forskningsresultaten och den praktiska tillämpningen vid livslängds-dimensionering.
Sedan 1996 exponeras ett stort antal provkroppar tillverkade av betong med olika sammansättningar, med olika bindemedel och bindemedelskombinationer vid fältprovplatsen vid Riksväg 40 strax väster om Borås. Vid denna provplats exponeras betong för en fuktig, tidvis kall och tö-saltad miljö. Provkropparna har genom åren undersökt vid ett flertal tillfällen och i olika omfattning. Efter 20 års exponering (2016-2017) beslutades om en omfattande undersökning för att få information om betongens långsiktiga beständighetsegenskaper. Projektet finansierades av Trafikverket (inom forskningsprogrammet BBT), den svenska cementtillverkaren Cementa AB samt av forskningsutövarnas egna insatser.
Projektet samordnades av Chalmers Tekniska Högskola i samarbete med RISE CBI Betonginstitutet, Lunds Tekniska Högskola (LTH) och Göteborgs universitet (GU). Projektarbetet delades upp i två delar: Del 1, frostbeständighet hos betong (separat rapport) och Del 2, kloridinträngning och armeringskorrosion (rapporteras i denna rapport). Huvudarbetet för hantering och mätningar av provkropparna från fältprovplatsen utfördes av RISE CBI. Vissa labbprovningar och modellering för kloridinträngning i betong utfördes vid Chalmers. GU var huvudsakligen delaktig i modelleringen av kloridbindning i betong. LTH var huvudsakligen delaktig i modellering av frostbeständighet hos betong som rapporterats i del ett.
I detta delprojekt har 35 kloridprofiler och tio fuktprofiler uppmätts. Vidare har korrosionstillståndet bestämts för armeringsjärn ingjutna i betongbalkar med olika typer av bindemedel och vattenbindemedeltal. För korrosionsmätning användes den icke-förstörande tekniken, RapiCor. I ett par fall kompletterades RapiCor-mätningen med förstörande prov där armeringsjärn sågades ut och frilades från balkar för okulär bedömning.
ClinConc-modellen användes för att modellera kloridinträngning i betong. Olika labbprovningsmetoder och numeriska simuleringar utfördes för att undersöka egenskaper hos betong och förändringarna av dessa efter långtids fältexponering. Enligt resultaten från uppmätta kloridinträngningsprofiler kan följande slutsatser dras:
• Bland fem typer av kommersiella cementprodukter visade finskt snabbcement den lägsta kloridinträngningen i betong med vbt ≤ 0,4.
• Av de tre bindemedelstyperna med blandcement visade svenskt Anläggningscement med 10% kiselstoft och finskt snabbcement med 44 % slagg bäst motstånd mot kloridinträngning.
• Vid den aktuella tösaltade vägmiljön kan kloridinträngning i betong med vbt 0,5 nå tröskelnivån 0,4 % av bindemedelsvikt vid djupet 30-40 mm efter exponering under 20 år. För betong med vbt ≤ 0,4 nås tröskelnivån på djup som är mindre än 30 mm efter samma exponeringslängd.
• När det gäller kloridinträngning i betong är den svenska tösaltade vägmiljön inte lika aggressiv som skvalpzonen vid marin miljö.
En grov uppskattning av det kritiska kloridinnehållet för korrosionsinitiering är 0,3-0,5 % av bindemedelvikten för armeringsjärn med såväl osprucket som sprucket betongtäckskikt. Det finns indikationer på att det kritiska kloridinnehållet kan vara något lägre för armeringsjärn där betongtäckskiktet var sprucket. Detta behöver dock undersökas ytterligare för att några säkra slutsatser skall kunna dras.
Resultaten visar också att den icke-förstörande korrosionsmätning med RapiCor-instrumentet generellt överensstämmer med kloridinträngningen. Den okulära destruktiva undersökningen bekräftade att den icke-förstörande tekniken RapiCor är ett användbart verktyg för detektering av pågående korrosion hos ingjutna armeringsjärn. ClinConc-modellen ger generellt en bra förutsägelse av kloridinträngningsfronten i betong. Dock blir förutsägelsen av kloridinträngningen mer komplicerad närmast den exponerade ytan. Detta på grund av torknings- och vätningscykler i motorvägsmiljön vid betongytan. Efter en viss modifiering med hänsyn till omfördelning av bundna klorider i den närliggande ytzonen kan modellen till en högre grad beskriva kloridinträngnings-profilerna.
1 Introduction
Chloride-induced corrosion of reinforcing steel and frost damage are the two biggest challenges in service life design of concrete structures exposed to de-icing salt road environments. To be able to design concrete structures in an economical, sustainable and safe manner, as well as evaluate residual service life of existing structures, models for chloride transport in concrete are needed. The models are used to assess when any reinforcement corrosion, frost scaling or internal frost damage are expected to start. For these models to become robust and credible, they must be based on the physical/chemical degradation processes and verified with results from research on concrete structures exposed in fields, especially after long-term exposure.
Swedish research in the areas of chlorine-induced corrosion of reinforcement steel and frost resistance has been in the international front since the 1980’s. Some internationally-known research findings include Tutti’s conceptual model for service life [1], The Borås method for frost scaling [2] and for internal frost damage [3], The rapid chloride migration (RCM) method [4][5], the ClinConc model for chloride ingress [5][6], Fagerlund’s frost model [7][8] and Petersson’s frost model [9], etc. During the last decade, Utgenannt [10] evaluated the aging effect on the resistance of concrete to frost scaling. Fridh [11] conducted experimental investigations of degradation mechanisms for internal frost damage. Tang and Utgenannt [12] investigated chloride penetration into concrete with different binders after 10 years of exposure in the road environment by the R40 field exposure site. Li [13] conducted an experimental study of frost-induced chloride penetration into concrete. In the past years, Tang et al. [14] evaluated different models for concrete resistance to chloride ingress and tried to validate these models against field data. Boubitsas et al. [15] through a SBUF project evaluated chloride ingress into concrete with different binders after over 20 years exposure in the marine environment in the Träslövsläge field exposure site. The knowledge of chloride ingress in concrete under the marine environment, especially in the splash/atmospheric zone, may also be applicable to road environments with de-icing salt with respect to similar ageing processes (carbonation and leaching).
Although there are ongoing research activities in the areas, there are still gaps between the research results and the practical application of service life design in the production of new concrete structures within the traffic system.
Since the winter of 1996-1997 many reinforced concrete specimens with different qualities have been exposed at the field exposure site by Highway 40 (Rv40) between Borås and Gothenburg in the western part of Sweden, where de-icing salt was intensively used on the road due to the sever winter climate. Measurements of chloride and moisture profiles have been made after exposure 1-2, 5 and 10 years. In 2016-2017 the specimens were exposed for 20 years. To investigate these specimens and obtain the “first-hand” information about the long-term behaviour of concrete with respect to frost resistance, chloride ingress and reinforcement corrosion under the de-icing highway environment, the Swedish Transport Administration financed this project. Swedish cement manufacturer Cementa AB also partially financed this project. The project was coordinated by Chalmers University of Technology and co-worked by RISE CBI (Swedish Cement and Concrete Research Institute), Lund Institute of Technology (LTH) and University of Gothenburg (GU). The project work was divided into two parts: Part One
for frost resistance of concrete, and Part Two for chloride ingress and reinforcement corrosion. The main work for the measurements of specimens from the exposure sites was carried out by RISE CBI. Some laboratory tests and modelling for chloride ingress in concrete were carried out by Chalmers. LTH was mainly involved in the modelling for frost resistance of concrete and GU was mainly involved in the modelling for chloride binding behaviour in concrete.
This report presents the results from Part Two, dealing with chloride ingress and reinforcement corrosion. The results from Part One, dealing with frost resistance are presented in a separate report.
2 Concrete Specimens and Exposure
Conditions
2.1 Concrete specimens
The relevant mixture proportions of concrete are summarised in Table 2.1 and the detailed information about the raw materials and hardened properties of each mixture proportion was published elsewhere [16]. The main variations include water-binder ratio (0.3, 0.35, 0.4 and 0.5, one up to 0.75), binder type (eight types of binder with different additions of limestone, silica fume and blast-furnace slag), and air content (5% entrained air and non-AEA). Two types of concrete specimens, one plain concrete block of 400×300×300 mm and another reinforced concrete beam of 1200×300×300 mm, were cast at SP Technical Research Institute of Sweden (previously Swedish National Testing and Research Institute). The plain concrete blocks were designed for sampling of chloride penetration profiles and the reinforced concrete beams were designed for testing corrosion resistance under uncracked and pre-cracked conditions. A typical structure of the reinforced concrete beam is shown in Figure 2.1 and the detailed information about the reinforcement placement in each beam was published elsewhere [17]. The specimens were cured in the laboratory for 35 to 70 days before placed at the field site.
In this project all the concrete types listed in Table 2.1 were sampled for chloride ingress profiles, but due to the limited resource, only ten types of concrete (as in bold mix number in the table) were sampled for moisture profiles and further laboratory investigations of their transport properties including migration coefficient, capillary water absorption and critical degree of saturation. The latter two properties were used in the project work Part One for the input parameters to Fagerlund’s model for internal frost damage in concrete [7]. These selected types of concrete include different types of binder and water-binder ratios, such as Swedish SRPC for civil engineering (Anl) with w/b 0.4 (Concrete mix #201) and w/b 0.3 (#203), Anl + 5% silica fume with w/b 0.4 (#206), Anl + 10% silica fume with w/b 0.4 (#225), Swedish cement for buildings (KalkC) with w/b 0.4 (#228), Finnish standard cement (FinStd) with w/b 0.4 (#211) and w/b 0.3 (#213), and Finnish rapid cement blended with 44% GGBS with w/b 0.4 (#219) and w/b 0.3 (#221).
Table 2.1 Mixture proportions of concrete exposed at the Highway 40 field site. Mix
No.
Binder type Binder kg/m3 W/B1) Fine aggreg. 0-8 mm kg/m3 Coarse aggreg. 8-16 mm kg/m3 Sp2) % of binder AEA3) % of binder Air % of vol. 28d compr. strength4) MPa 201 Anl5) 420 0.40 886.4 851.6 0.97 0.028 4.8 65.4 202 380 0.50 890.2 821.8 0.012 4.5 50.8 203 500 0.30 833.5 978.5 3.25 1.3 100.6 204 450 0.35 880.3 953.7 2.27 0.9 91.3 205 380 0.50 938.6 866.4 1.3 54.8 236 260 0.75 1007.4 791.6 0.012 4.5 30 206 Anl+5%SF6) 420 0.40 860 860 1.2 0.04 4.7 78.1 207 380 0.50 865.5 831.5 0.022 4.4 58.2 208 500 0.30 806 985.1 1.7 0 1.2 119.7 209 450 0.35 846.5 954.5 1.5 0 1.1 103.5 210 380 0.50 905.8 870.2 0 0 0.9 62.6 211 FinStd7) 420 0.40 863 863 1.2 0.02 4.9 48.8 212 380 0.50 885 817 0 0.008 4.5 42.1 213 540 0.30 767.7 938.3 3.6 0 2.5 66.9 214 390 0.50 915.2 844.8 0 0 1.2 47.4 215 Slite8) 420 0.40 880.3 845.7 1.7 0.027 4.9 60.7 216 390 0.50 874.1 806.9 0 0.01 4.4 46 217 520 0.30 799.9 939.1 3.8 0 2.3 85.7 218 410 0.50 892.3 823.7 0 0 1.4 59.9 219 FinRpd9) +44%SL10) 420 0.40 858 858 1.15 0.027 4.5 55 220 370 0.50 891.8 823.2 0 0.01 4.7 41.5 221 540 0.30 761.9 931.2 3.8 0.01 1.6 82 222 FinRpd 420 0.40 863 863 2 0.027 4.7 57.6 223 380 0.50 885 817 0 0.013 4.9 43.4 224 540 0.30 767.7 938.3 4.15 0 2.4 66.5 225 Anl+10%SF 420 0.40 851.5 851.5 1.42 0.08 4.8 81.4 226 500 0.30 796.5 973.5 2.1 0 0.5 126.9 227 450 0.35 844.1 951.9 1.7 0 0.9 112.7 228 KalkC11) 420 0.40 880.3 845.7 1.7 0.03 4.8 68.9 229 390 0.50 874.1 806.9 0 0.014 4.7 51.8 230 530 0.30 781.2 954.8 3.6 0 2.2 98.5 231 470 0.35 851 922 2.62 0 1.8 86.3 232 400 0.50 903.8 834.2 0 0 1.2 60.9
1) Mass ratio of water to binder without consideration of the efficiency factor for silica fume or blast-furnace slag.
2) Sp – Super-plasticizer, Cementa Melcrete 3) AEA – Air entraining agent, Cementa L17 4) According to SS 13 72 10
5) Anl – Anläggningscement (Swedish SRPC, CEM I) 6) SF – Silica fume (Elkem. Norway)
7) FinStd – Finnish standard Portland cement with 15-18% ggbf slag (CEM II/A-S) 8) Slite – Swedish ordinary Portland cement, made in Slite (CEM I 52.5N)
9) FinRpd – Finnish rapid Portland cement with 10-15% limestone filler (CEM II/A-LL 42.5R) 10) SL – Finnish ground granulated blast-furnace (ggbf) slag
2.2 Exposure conditions
The field exposure site at highway 40, outside Borås in the western part of Sweden, was established in the autumn of 1996. It consists of a 200-meter-long and a couple of metres wide gravel area along the highway, with specimens mounted in steel frames at road level, as shown in Figure 2.1. A guard rail was installed to separate the exposure site from the traffic. It was placed in such a way as to ensure the traffic safety and to have the specimens fully exposed to the splash water from the traffic. The climate around the specimens is moist, and the specimens are exposed during the winter to low temperatures and de-icing salts, producing a climate corresponding to exposure class XD3/XF4 in EN 206-1.
The monthly air temperature registered at the climate station near the field exposure site showed that the average temperature was about 10 ºC, excluding the freezing period under which the transport of chloride ions would be negligible. The detailed description of the exposure site has been published elsewhere, e.g. in the previous report by Tang and Utgenannt [12].
Figure 2.1. Illustration of the field exposure site at Highway 40 and the placement of the specimens.
0. 7 6 m 0. 4 5 m Guard rail
Concrete blocks & beams
3 Measurements of Chloride and
Moister Profiles
3.1 Sampling
Totally 33 concrete blocks of size 400300300 mm were taken to the laboratory at RISE CBI in Borås for measurements of chloride profiles (C). Further, for ten of the concrete blocks (see Table 2.1, marked with bold mix number), moisture profiles (M) and rapid chloride migration (RCM) measurements were also performed. One cores of diameter 100 mm, from the vertical exposure surface was taken at the positions as shown in Figure 4.1 for chloride profile. The sampling positions for the ten concrete blocks where moisture profile (M) and RCM measurements also were performed are shown in Figure 4.2.
From these concrete blocks two cores of diameter 100 mm were drilled throughout each block (Fig. 4.2), these core specimens were cut in half and the side exposed to the road was used to measure chloride (C)or moisture profiles (M) and the other half was used to measure RCM. Additionally, a slab slice of size 50300300 mm was cut for a modified RCM measurement which will be further explained in chapter 3.4.
Figure 3.2. Illustration of sampling position at concrete blocks where chloride (C), moister (M) and rapid chloride migration (RCM) measurements were performed.
3.2 Measurement of chloride profiles
The same techniques as used in the previous investigations [18][19] were used in this project for measurement of chloride profiles. Powder samples were taken from each core by means of dry-grinding on a lathe with a diamond tool, successively from the exposed surface to a certain depth. The depth of each sample was measured from the lathe with an accuracy of 0.5 mm. After the grinding, the powder samples were immediately dried at 105 °C and then stored in a desiccator for later chloride and calcium analysis.
The acid soluble chloride content in each sample was determined principally in accordance with AASHTO T260 using potentiometric titration on an automatic titrator. The soluble calcium content in each powder sample was determined parallel to the determination of chloride content, using the technique reported in [18]. The results of chloride and calcium profiles in each core are given in detail in Appendix 1. Further, the result from the chloride profiles are summarised in Figs. 3.3 to 3.10.
3.2.1 Effect of binder type and mineral additions
Figure 3.3 shows the chloride ingress profiles in concrete with five types of commercial
cement. It can be observed that the chloride content at a specific depth differ very
irregularly between these types of cement. It is difficult to rank the resistance of cement
to chloride ingress, although it looks clear that Finnish rapid cement revealed the lowest
chloride ingress when w/b ≤ 0.4. However, a general tendency is that the chloride ingress
is deeper in concrete with higher water-binder ratio, as expected.
Figure 3.4 shows the chloride ingress profiles in concrete with the replacement of silica
fume and ggbf slag. It seems no markedly effect on chloride ingress when Swedish SRPC
was replaced with 5% silica fume, but the effect became clear when replaced with 10%
silica fume. For Finnish rapid cement replaced with 44% slag, the effect on chloride
ingress was not markedly in concrete with w/b 0.3, but clearer in concrete with w/b
0.4-0.5.
Figure 3.3. Chloride profiles from concrete with different types of commercial cement.
Figure 3.4. Chloride profiles from concrete with replacement of silica fume and ggbf slag. 0 0.2 0.4 0.6 0.8 1 0 20 40 60 80 C l [% b y w t o f b in d e r] Depth [mm] Anl Slite KalkC FinStd FinRpd w/b 0.3 no entrained air 0 0.2 0.4 0.6 0.8 1 0 20 40 60 80 C l [% b y w t o f b in d e r] Depth [mm] Anl Slite KalkC FinStd FinRpd w/b 0.4 5% entrained air 0 0.2 0.4 0.6 0.8 1 0 20 40 60 80 C l [% b y w t o f b in d e r] Depth [mm] Anl Slite KalkC FinStd w/b 0.5 no entrained air 0 0.2 0.4 0.6 0.8 1 0 20 40 60 80 C l [% b y w t o f b in d e r] Depth [mm] Anl Slite KalkC FinStd FinRpd w/b 0.5 5% entrained air 0 0.5 1 1.5 2 0 20 40 60 80 C l [% b y w t o f b in d e r] Depth [mm] Anl Anl+5%SF Anl+10%SF FinRpd FinRpd+44%SL w/b 0.3 no entrained air 0 0.5 1 1.5 2 0 20 40 60 80 C l [% b y w t o f b in d e r] Depth [mm] Anl Anl+5%SF Anl+10%SF FinRpd FinRpd+44%SL w/b 0.4 5% entrained air 0 0.5 1 1.5 2 0 20 40 60 80 C l [% b y w t o f b in d e r] Depth [mm] Anl Anl+5%SF Anl+10%SF w/b 0.35 no entrained air 0 0.5 1 1.5 2 0 20 40 60 80 C l [% b y w t o f b in d e r] Depth [mm] Anl Anl+5%SF FinRpd FinRpd+44%SL w/b 0.5 5% entrained air
3.2.2 Effect of water-binder ratio and entrained air
Figure 3.5 shows chloride profiles from concrete with different binders, with and without entrained air. Comparing the chloride ingress for similar mixes where only the w/b is altered, it’s clear that a lower w/b results in a higher chloride ingress resistance. This is more obvious for w/b ≥ 0.40, with exception for the mixes with “FinRpd+44%SL” as binder, in this case the chloride ingress was low and quite similar and for the mixes with w/b 0.40 and 0.50.
Figure 3.6 shows chloride profiles from concretes with w/b 0.5 with different binders, with and without entrained air (non-AEA). From these limited results it’s not clear how entrained air (~4.4-4.9%) influences the chloride ingress compared to natural occurring air content (~1.2-2.5%). For the mixes containing “Anl” and “Anl+5% SF” (Fig. 3.6, left-hand side) the mixes containing entrained air show clearly a higher chloride ingress from the surface and to approximative 60 mm in from the surface, thereafter, the chloride profiles level out. For the concrete mixes at the right-hand side (binders Slite and KalkC) no difference in the resistance to chloride ingress can be observed between the mixes with are without entrained air.
Figure 3.6. Chloride profiles in concrete with w/b 0.5, different binders, with and without AEA.
3.2.3 Effect of exposure durations
Figure 3.7 shows chloride ingress profiles in two types of concrete after exposure for different durations. For the concrete with Swedish SRPC, the chloride profiles after exposure for 1-10 years were similar, and markedly less than that after exposure for 21 years. For the concrete with Swedish SRPC with a replacement of 5% silica fume, the chloride ingress in the depth of 0-20 mm was unexpected. The ingress after exposure for 1-5 years was higher than that after exposure for 10-21 years, probably due to higher amount of salt spread under the first four winters (1996-2000).
Figure 3.8 shows the chloride ingress profiles of concrete with eight types of binder with w/b 0.3 after exposure for 10 and 21 years. It can be observed that the profiles after exposure for 21 years are in general somewhat deeper than those after exposure for 10 years. However, the difference is small, indicating that for these types of concrete with w/b 0.3 the ingress in the actual highway environment is very slow after 10 years exposure.
Figure 3.7. Chloride ingress profiles in two types of concrete after exposure for different years.
Figure 3.8. Chloride ingress profiles in different types of concrete after exposure for 10 and 20 years. 0 0.5 1 1.5 0 20 40 60 80 C l [% b y w t o f b in d e r] Depth [mm] Meas. 1 y Meas. 5 y Meas. 10 y Meas. 21 y Anl w/b 0.4 5% entrained air 0 0.5 1 1.5 0 20 40 60 80 C l [% b y w t o f b in d e r] Depth [mm] Meas. 1 y Meas. 5 y Meas. 10 y Meas. 21 y Anl+5%SF w/b 0.4 5% entrained air 0 0.5 1 1.5 0 20 40 60 80 C l [% b y w t o f b in d e r] Depth [mm] Meas. 10 y Meas. 21 y Anl w/b 0.3 0 0.5 1 1.5 0 20 40 60 80 C l [% by w t of bi nd e r] Depth [mm] Meas. 10 y Meas. 21 y Anl+5%SF w/b 0.3 0 0.5 1 1.5 0 20 40 60 80 C l [% b y w t o f b in d e r] Depth [mm] Meas. 10 y Meas. 21 y Anl+10%SF w/b 0.3 0 0.5 1 1.5 0 20 40 60 80 C l [% by w t of bi nd e r] Depth [mm] Meas. 10 y Meas. 21 y Slite w/b 0.3 0 0.5 1 1.5 0 20 40 60 80 C l [% b y w t o f b in d e r] Depth [mm] Meas. 10 y Meas. 21 y KalkC w/b 0.3 0 0.5 1 1.5 0 20 40 60 80 C l [% by w t of bi nd e r] Depth [mm] Meas. 10 y Meas. 21 y FinStd w/b 0.3 0 0.5 1 1.5 0 20 40 60 80 C l [% b y w t o f b in d e r] Depth [mm] Meas. 10 y Meas. 21 y FinRpt w/b 0.3 0 0.5 1 1.5 0 20 40 60 80 C l [% b y w t o f b in d e r] Depth [mm] Meas. 10 y Meas. 21 y FinRpt+44%SL w/b 0.3
3.2.4 Effect of exposure environment
In a previous study [15] many chloride ingress profiles in concrete with Swedish SRPC were collected from a marine exposure site along Swedish western coast after exposure for 20 years. Figure 3.9 shows the comparisons of chloride ingress in two similar types of concrete exposed to these two different environments. Clearly, the exposure in the marine environment led to significantly higher chloride ingress.
Figure 3.9. Chloride ingress profiles in two types of concrete after exposure for 20 years in a marine splash zone and a de-icing highway environment.
Figure 3.10 shows chloride profiles from similar concrete mixes exposed at the highway and at the ATM-zone (marine field site) for about 10 years (10y) and 20 years (20y). It should be noted that the concrete mixes with w/b 0.35 placed at the highway site were without any entrained air in contrast to the mixes placed at the marine site. However, the vast difference in chloride ingress between the two exposure sites is similar as for the mixes with w/b 0.50 where the only difference is the exposure site. Fig.3.10, show a much higher chloride ingress for the specimens exposed at the marine site after 20 years. It is also obvious that the development of the chloride ingress with time differs a lot between 10 and 20 years exposure when comparing the two sites. The results show a much higher chloride ingress between the two measuring occasions at the marine site.
0 1 2 3 4 5 0 10 20 30 40 50 C l [% by w t of bi nd e r] Depth [mm] Road marine (splash) Anl w/b 0.35 0 1 2 3 4 5 0 10 20 30 40 50 C l [% b y w t o f b in d e r] Depth [mm] Marine (splash) Road Anl w/b 0.4 0 1 2 3 4 5 0 10 20 30 40 50 C l [% b y w t o f b in d e r] Depth [mm] Road marine (splash) Anl+5%SF w/b 0.35 0 1 2 3 4 5 0 10 20 30 40 50 C l [% b y w t o f b in d e r] Depth [mm] Marine (splash) Road Anl+5%SF w/b 0.4
Figure 3.10. Chloride profiles in similar concrete exposed in different environments, highway vs marine atmospheric zone (above splash zone).
0.0 0.5 1.0 1.5 2.0 2.5 0 10 20 30 40 50 60 Cl -% o f b in d er Depth [mm] Anl. cem., w/b 0.35
AEA, Marine-ATM (20y) AEA, Marine-ATM (10y) non-AEA, Highway (20y) non-AEA, Highway (10y)
0.0 0.5 1.0 1.5 2.0 2.5 0 10 20 30 40 50 60 Cl -% o f b in d er Depth [mm] Anl. cem. + 5% SF, w/b 0.35
AEA, Marine-ATM (20y) AEA, Marine-ATM (10y) non-AEA, Highway (20y) non-AEA, Highway (10y)
0.0 0.5 1.0 1.5 2.0 2.5 0 10 20 30 40 50 60 Cl -% o f b in d er Depth [mm] Anl. cem. + 5% SF, w/b 0.50, with AEA
Marine-ATM (20y) Highway (20y) Highway (10y)
3.2.5 Curve-fitted parameters
According to Fick’s second law, the following error function solution can be used to describe diffusion under the semi-infinite boundary:
𝐶(𝑥, 𝑡) = 𝐶
𝑖𝑛𝑖+ (𝐶
𝑠− 𝐶
𝑖𝑛𝑖) ∙ [1 − 𝑒𝑟𝑓 (
𝑥2√𝐷𝐹2∙𝑡
)]
(3.1)where C (x, t) is the chloride concentration at depth x after exposure period t, Cini is the
initial chloride concentration in concrete, Cs is the chloride concentration at the exposure
surface, DF2 is the chloride diffusion coefficient, and erf is the error function.
Theoretically, only the gradient of free chloride concentration is contributed to the driving force for chlorides to penetrate concrete. However, the free chloride profiles can, up to now, hardly be measured. The reported chloride profiles are normally based on the determination of total chloride content. The total chloride content is not necessarily proportional to the free chloride concentration due to the non-linear behaviour of chloride binding. Therefore, using the gradient of total chloride content as a driving force in Fick’s law may not be theoretically correct, but just empirically convenient for describing the characteristics of penetration profiles. The models based on Eq. (3.1) may be regarded as empiric models. Nevertheless, curve-fitting the measured chloride profile to Eq. (3.1), one can obtain two parameters, DF2 and Cs. These two parameters are a
description of the chloride ingress under a specific exposure condition and after a specific exposure period. A MS Excel-based program was used for curve-fitting calculations. For each chloride profile, the first one or two points were omitted prior to the calculation, if the values are significantly out of the diffusion curve. Fig. 3.11 shows an example of the curve-fitting procedure.
Figure 3.11. Example of curve fitting chloride profile using Eq. 3.1
The curve-fitted results are listed in Appendix 3 and summarised in Figs. 3.12 and 3.13. From the curve-fitted results it can be found that:
• The curve-fitted apparent diffusion coefficient Da, reveals a certain degree of
observed, the mixes containing “FinRpd+44%SL” as binder reveals a more linear relationship of the Da to the water-binder ratio.
• There is no clear relationship between the curve-fitted surface chloride content Cs and water-binder ration. However, for most of the mixes with the same binder
Cs do not differ much for the lower w/b (≤0.40) and tend to attain a lower value
for w/b 0.50. Also in these results the mixes containing “FinRpd+44% SL” as binder differ from the rest of the mixes, reaching a higher value for the mixes with w/b ≥0.40.
Figure 3.12. Results of the curve-fitted apparent diffusion coefficient.
3.3 Measurement of moisture profiles
The cores taken from the selected concrete blocks (marked with “M” in Figure 3.2) were individually sealed in double thick plastic bags and stored in the laboratory at the room temperature not longer than a few days prior to sampling. A slice of about 20 mm thick was split from each concrete core at depths of about 0~20, 20~40, 40~60 and 60~80 mm starting from the exposure surface, with the help of a compression jack. Several small pieces of sample were immediately taken, using hammer and chisel, from the centre portion of the freshly split slice. The sample pieces were sealed in a glass test tube with a rubber stopper After the above sampling, another slice was successively split and samples were taken. The above sampling process was repeated until all the samples were taken from each core. The glass test tubes containing samples were transported to Chalmers and stored in the climate room at 202 °C for measurement of RH (Relative Humidity) using Vaisala humidity probes.
The measured results together with those measured after exposure for 10 years are shown in Figures 3.14-3.16. As a general tendency, the moisture profiles in concrete after 21 years are lower than those after 10 years, because of the further hydration (self-desiccation) and drying through the exposure surface. In the near surface zone, however, the profiles vary irregularly, probably due to the variable wetting/drying on the exposure surface, or weathering effects. The behaviour of concrete with KalkC is difficult to explain, probably due to the uncertainty in the measurement or poor sealing of the test tube.
Figure 3.14. Moisture profiles in concrete (21 resp. 10 years) with Swedish cement, w/b 0.40. 50 55 60 65 70 75 80 85 90 95 0 20 40 60 80 100
R
el
at
iv
e
hum
idi
ty
, %
Depth, mm
Anl, w/b 0.4 Ditto, 10yrs Anl+5%SF, w/b 0.4 Ditto, 10yrs Anl+10%SF, w/b 0.4 Ditto, 10trs KalkC, w/b 0.4 Ditto, 10yrsw/b 0.40
Figure 3.15. Moisture profiles in concrete (21 resp. 10 years) with Swedish cement, w/b 0.30-35.
Figure 3.16. Moisture profiles in concrete (21 resp. 10 years) with Finnish cement, w/b 0.30-0.40. 50 55 60 65 70 75 80 85 90 95 0 20 40 60 80 100
R
el
at
iv
e
hum
idi
ty
, %
Depth, mm
Anl, w/b 0.30 Ditto, 10yrs Anl+5%SF, w/b 0.35 Ditto, 10yrsw/b 0.30-0.35
50 55 60 65 70 75 80 85 90 95 0 20 40 60 80 100R
el
at
iv
e
hum
idi
ty
, %
Depth, mm
Fin Std, w/b 0.4 Ditto, 10yrs Fin Std, w/b 0.3 Ditto, 10yrs FR+44%SL, w/b 0.4 Ditto, 10yrs FR+44%SL, w/b 0.3 Ditto, 10yrsFinnish binder w/b 0.30-0.40
3.4 Measurement of rapid chloride migration
To evaluate the effect of concrete age on the chloride transport property the cores taken from the selected concrete blocks (marked with “RCM” in Figure 3.2) were used to determine the chloride migration coefficient according to the rapid chloride migration test (RCM) [4][5], as a standardized version, NT BUILD 492. A specimen with thickness of 50 mm was cut from each core, on the side close to the centre of the block. There were two cores per concrete block. Therefore, two duplicates were used for the test. The measured results are shown in Figure 3.17, where the migration coefficients measured for the similar types of concrete from the previous BMB project are listed for comparison. The results show that the addition of silica fume (in the form of slurry) and slag can markedly reduce the chloride migration coefficient. For the concrete with plain Portland cement (#201) with w/b 0.4, the migration coefficient after 20 years reduced to a half of that at the age of 0.5 years, greatly owing to the further hydration which reduced larger capillary pores. However, for the concrete with low w/b or addition of silica fume, the migration coefficient after 20 years was increased. This is probably because the pore structures for these types of concrete were already very fine when they were in their early age. The further hydration may change their fine pore structures to relatively coarse ones due to the formation of new hydrates. This is in the line with the findings from the marine exposure site in the Träslövsläge harbour [15].
Figure 3.17. Chloride migration coefficients in concrete after 20 years (in comparison some results after 0.5 years).
#1-40 #H3 #H4 #H1
Binder
type Anl Anl
Anl +5%SF Anl +5%SF Fin Std Fin Std FinRapid +44%SL FinRapid +44%SL Anl +10%SF KalkC w/b 0.40 0.30 0.40 0.30 0.40 0.30 0.40 0.30 0.40 0.40 BMB concrete 0 2 4 6 8 10 12 14 #201 #203 #206 #208 #211 #213 #219 #221 #225 #228
D
RCM, x
10
-12m
2/s
Measured after 20 yrs exposure under a highway environment Measured from the similar concrete after 0.5 yrs curing in the lab
Note: Microsilica was added in the form of slurry for all mixes except for one (in the form of powder) in comparison with #225
3.5 Measurement of rapid iodide migration
To evaluate the effect of exposure on the chloride transport property, a specimen with a size of 10010050 was taken from the slab slice cut from the centre of the concrete block, as shown in Figure 3.18. Because the concrete near the exposed surface has already been contaminated by chlorides, sodium chloride as specified in the RCM test could not be used for detecting chloride migration front. Therefore, sodium iodide was used in the migration test, called the rapid iodide migration test (RIM). This idea was initially proposed by Lay et al. [20] and later modified by Liu [21]. The migration principle of RIM and RCM is similar, but the colour display for detecting the migration front is completely different in these two methods. In the RCM, 0.1 M silver nitrate was used as colour display agent because the chloride ions directly react with the silver ions to form silver chloride white deposits. In the RIM, however, the procedures for colour display are more complicated. Firstly, a solution of acetic acid (HAc 30% by mass) was sprayed on the newly split concrete surface to reduce the alkalinity of concrete. When the surface was slightly dry, a solution of potassium iodate (KIO3 1% by mass) was sprayed on the
surface to react the iodide migrated in concrete to form molecular iodine, and then an aqueous suspension of starch (about 3% by mass) was sprayed. This iodine-starch complex would turn the concrete to a dark blue-black colour. An example of specimen preparation for the RIM test and the colour display after both the RCM (with 10% NaCl) and RIM (with 10% NaI) tests is shown in Figure 3.19, in which it can be observed that the chloride ions has contaminated the right side (near the exposure surface) of the specimen due to the exposure. From the iodide migration profile it becomes possible to evaluate the migration coefficient along the exposure depth, as shown in Figure 3.20.
Figure 3.18. Sampling of specimen for the RIM test.
50 100
100
…
Na,KOH NaI
Rapid Migration Test
Figure 3.19. Specimen preparation for the RIM test (left) and colour displays after the tests (right).
Figure 3.20. Specimen preparation for the RIM test (left) and colour displays after the tests (right).
Owing to the limited availability of specimens and also difficulties in carrying out such a test, only for concrete with Anl cement (#201) two duplicates tests were used. For the other types of concrete one specimen in the RIM test were used. The results are shown in Figure 3.21. It is obvious that the concretes blended with 44% slag and 10% silica fume reveal less change in transport property along the exposure depth, implying less deterioration due to exposure. It seems that the higher the base transport property (deeper part), the larger the change in the transport property near the surface zone.
Su
rf
ac
e
ex
p
o
sed
to
th
e
ro
ad
s
id
e
Chloride migration direction
Iodide migration direction
Su rf a ce e xp o se d to t he ro ad s id e
Iodide migration direction 0 10 20 30 40 50 60 70 80 90 100
0 3 6 9 12 D if fu si o n co e ff ici e n t/ -1 2m 2/S Depth/mm PC04 PC03 Anl w/b 0.40 Anl w/b 0.30 Io d id e mi gr ati on c o e ff ic ie nt , x 10 -12 m 2/s
Figure 3.21. Effect of exposure on the transport property in concrete. 0 2 4 6 8 10 12 14 0 20 40 60 80 100 Io di de m ig ra ti on c oef fi ci en t, x1 0 -12 m 2/s
Depth from the exposure surface, mm
Anl, w/b 0.4 Anl , w/b 0.3 Anl+5%SF, w/b 0.4 Anl+5%SF, w/b 0.3 Anl+10%SF, w/b 0.4 0 2 4 6 8 10 12 14 0 20 40 60 80 100 Io di de m ig ra ti on c oef fi ci en t, x1 0 -12 m 2/s
Depth from the exposure surface, mm
FinStd, w/b 0.4 FinStd, w/b 0.3
FinRpt+44%SL, w/b 0.4 FinRpt+44%SL, w/b 0.3 KalkC, w/b 0.4
4 Corrosion Measurement
4.1 Specimens
Totally 45 reinforced concrete beams of size 3003001200 mm were taken to the laboratory at RISE CBI in Borås to test corrosion resistance under uncracked and pre-cracked conditions.
A typical structure of a reinforced concrete beam, including four rebars (R1 to R4) and the three measure points (M1, M2 and M3) is shown in Figure 4.1. The size of the concrete beam is 3003001200 mm. Detailed information about the reinforcement placement in each beam is published elsewhere [17]. Most of the beams contained two similar sets of rebars one at the left side and one at the right side. Further, one (most often the left side), or both sides of the concrete beam were pre-cracked. The specimens were cured in the laboratory for 35 to 70 days before placed at the field site.
Ribbed rebars were used with 12 mm diameter, the rebars were bend at the ends in order to place them at the correct position. The bended ends were coated with epoxy and the exposed length of the rebars was about 480 mm. The bars were placed at two different levels with a cover depth of 15 mm and 30 mm. In some beams bars were also embedded with concrete cover 60 mm above the set of bars with 30 mm cover. Each steel bar was soldered with a copper wire. The other end of the copper wire was connected to a plinth in an electronic box outside the specimen. This wire supplied the connection to the rebar for electrochemical measurement.
4.2 Technique for corrosion measurement
All corrosion measurements were performed in the laboratory with the commercially available RapiCor instrument (see Fig. 4.2). The RapiCor instrument is based on galvanostatic pulse technique, this technique is further described by Tang [22]. A wet sponge is placed on the concrete surface over the steel to be measured, to improve the contact between the concrete and the electrodes unit. Then the rectangular shaped electrodes unit is placed on the wet sponge and the instrument measures the corrosion potential (Ecorr). A galvanostatic current is applied to the steel to be measured and the
potential response is recorded. From the recorded potential-time curve the polarisation resistance (Rp) is obtained. The Rp is used to calculate the corrosion current density (icorr)
using the Stern-Geary equation:
p corr
AR
B
i
=
(6.1)Where, B is a constant assumed to be 26 mV and A is the polarized area. The instrument gives the corrosion rate in µm/year assuming uniform corrosion. Further, the instrument gives the corrosion potential (versus the copper/copper sulphate reference electrode) and the concrete resistivity in kΩ·cm. It should be noted that the uncertainty in the corrosion rate is large. A factor of 2 (multiplying by 2 for upper and dividing by 2 for lower limits) has normally been adopted [22].
Figure 4.2. The handheld RapiCor instrument used for the corrosion measurements.
Because the instrument measures not only corrosion rate, but also the half-cell potential and ohmic resistivity of concrete, it would reduce the uncertainty if all these three parameters are utilised in the assessment. The corrosion rate is taken as the main parameter, and half-cell potential and resistivity as complementary parameters in the assessment of corrosion level. The corrosion level is expressed as a corrosion index, an example of the criteria for classification of corrosion index is shown in Table 4.1.
Table 4.1 Criteria for classification of corrosion index.
Corrosion Index
Corrosion rate
(m/yr)
Half-cell potential
(mV
(CSE))
Resistivity
(kcm)
1
Negligible
<1
-
-
1~5
> -200
> 100
2
Low
1~5
< -200
< 100
5~10
> -200
> 100
3
Moderate
5~10
< -200
< 100
>10
> -200
> 100
4
High
>10
< -200
< 100
According to Table 4.1, the parameter corrosion rate is used for primary classification. If the two complementary parameters are not lower than the criteria for possible corrosion, the corrosion condition will be classified as one level lower. For example, if the corrosion rate is >10 m/yr corresponding to “high”, but the half-cell potential is higher than -200 mV CSE or the resistivity is larger than 100 kcm, the corrosion level will be classified as “moderate”. In this way the assessment would be safer than that based on only one parameter, considering the complication of reinforcement corrosion in the real structures.
4.2.1 Validation of corrosion measurement
To verify the measurement using the non-destructive technique RapiCor, six concrete beams were selected, with mix numbers according to Table 2.1; 201 AB1, 202 AB2, 205 B1, 214 B1, 222 B1, and 228 AB2.
After the RapiCor measurements the rebars were released from each of these concrete beams by sawing and splitting and the corrosion condition of each rebar was visually examined and documented. Figures 4.3 to 4.5 show the results from the corrosion measurement and visual examination of beam 201 AB1, 214 B1 and 222 B1. As can be seen in Figs. 4.3 to 4.5 the non-destructive corrosion measurement with the RapiCor instrument is in reasonably good agreement with the visual examination.
Figure 4.4. Results from corrosion measurement and visual examination of beam 214 B1.
Figure 4.5. Results from corrosion measurement and visual examination of beam 222 B1.
4.2.2 Results from corrosion measurement
All the results from the corrosion measurements can be found in Appendix 4. Also, the assumed chloride content at the depth of the rebar (Cl%-binder at the rebar) is included. The chloride content at the depth of the rebar is assumed to correspond to the chloride profile ingress in the concrete block with the same concrete mix. However, for rebars where the concrete cover was pre-cracked this assumption is uncertain.
Table 4.3 shows the results from the corrosion measurements for rebars with 30 mm cover, where the cover was at one side uncracked and at the other side pre-cracked (see Fig. 4.1)
Table 4.3 Results from corrosion measurement, rebars with 30 mm (concrete cover) embedded in beams where one side was uncracked and the other side was pre-cracked.
Remark from the result in Table 4.3:
A rough estimation of the critical chloride content for corrosion initiation is 0.3-0.5% by weight of binder for rebars with uncracked as well as cracked concrete cover. A tendency towards somewhat lower critical chloride content may be seen for rebars in pre-cracked concrete cover. This need, however, to be investigated further to be able to draw clear conclusions. Binder Concrete beam Cl%-binder at the rebar Corrosion Index Crack [mm] Corrosion Index 201 BB2 0.3 Low No 203 CB1 0.1 Negligible No
204 AB1 0.1 Negligible 0.3 Negligible
205 B1 0.3 Moderate 0.3 Moderate
206 BB2 0.3 Negligible No
208 BB1 0 Negligible No
210 B1 0.3 Negligible 0.3 Low
212 B1 0.3 Negligible 0.3 Low
213 AB1 0 Negligible 0.3 Negligible
214 B1 0.5 High 0.4 Moderate
215 B1 0.1 Negligible 0.2 Low
216 B1 0.4 Negligible 0.4 Negligible
217 AB1 0 Negligible 0.4 Negligible
218 B1 0.4 Low 0.4 Low
219 B1 0.1 Low 0.4 Negligible
220 B1 0.2 Negligible 0.3 Negligible
221 AB1 0 Negligible 0.4 Negligible
222 B1 0.1 Negligible 0.4 Low
224 AB1 0 Negligible 0.3 Negligible
225 B1 0 Negligible 0.4 Negligible
226 AB1 0 Negligible 0.2 Negligible
227 AB1 0 Negligible 0.2 Low
230 AB1 0 Negligible 0.2 Moderate
231 AB1 0.1 Negligible 0.2 Negligible
232 B1 0.3 Low 0.3 Moderate FinRpd Anl+10%SF KalkC Uncracked Pre-cracked Anl. Cem Anl.+ 5% SF Fin Std Slite Std FinRpd+44%SL
5 Modelling of Chloride Ingress
The results from our previous publications [12][23][24] have shown that, among three different models (the simplest Fick’s law, DuraCrete and ClinConc), the ClinConc model revealed better prediction of chloride ingress in concrete under the de-icing road environment. Therefore, in this project the ClinConc model will be based for further modification.
5.1 Main input data – RCM-values
In the ClinConc model, the main input parameter is the chloride migration coefficient measured by the RCM test at the age of 6 months. These data were unfortunately not measured in the beginning of the exposure in 1997, when the RCM test was not standardized. Therefore, the RCM data from the previous studies were analysed to obtain some regression equations, as shown in Figure 5.1. It is obvious that the equations follow a power relationship. B w D A b = (5.1)
where constants A and B are dependent on the type of binder and air entraining. In the following modelling
Figure 5.1. Regression equations for estimating RCM-values. Data in a) to c) were based on [25] and data in d) were based on [26].
y = 266.72x3.6352 0.1 1 10 100 0.2 0.3 0.4 0.5 0.6 0.7 0.8 DR C M ,6 m , x 1 0 -12 m 2/s Water-binder ratio, w/b 4-6% air non-AEA Binder: 100% SRPC (Anl) DRCM ,6m= 267(w/b)3.64 DRCM ,6m= 86.8(w/b)2.85
a)
y = 163.99x3.4749 y = 92.125x2.6901 0.1 1 10 100 0.2 0.3 0.4 0.5 0.6 0.7 0.8 DR C M ,6 m , x 1 0 -12 m 2/s Water-binder ratio, w/b 4-6% air non-AEABinder: 100% OPC (Slite)
DRCM ,6m= 164(w/b)3.47 DRCM ,6m= 92.1(w/b)2.69
b)
y = 102.01x3.7524 0.1 1 10 100 0.2 0.3 0.4 0.5 0.6 0.7 0.8 DR C M ,6 m , x 1 0 -12 m 2/s Water-binder ratio, w/b 4-6% air non-AEA Binder: Anl+(5%-10%)SF DRCM ,6m= 130(w/b)3.86 DRCM ,6m= 102(w/b)3.75c)
0 0.2 0.4 0.6 0.8 1 1.2 0% 20% 40% 60% 80% R el at iv e m igr at ion co ef fi ci entGGBS content in the binder w/b 0.40 w/b 0.45
D6m/ D6m0= 1/(1+8S)
Binder: Anl+Slag (S)
d)
5.2 Other input data
Similar as in our previous modelling for the road environment [12][23][24], the chloride ionic concentration of 1.5 g/l and the annual mean air temperature of +10 °C were adopted as the environmental data. The initial chloride content was estimated based on the water content in each concrete mix, under the assumption that chloride content in the mixing water is 15 mg/l. The initial alkali content was based on the data supplied by the manufacturers of cement and slag. The time-dependent factor of chloride binding was considered as one-third that of the submerged environment. To facilitate the further modification, the expansion factor due to ageing in the field and the age factor due to drying under the road environment were not considered. All the other parameters relevant to the model were the same as described in [6] or [23][24].
5.3 Results from the previous model
Figure 5.1 shows some results from the previous ClinConc model with the input data as described in sections 5.1 and 5.2. It seems that the predicted ingress depths are in a fair agreement with the profiles from the exposure site. However, the shapes of the predicted profiles are different from the field ones, except for the concrete with low water-binder ratio (e.g. w/b 0.3). For most of the chloride ingress profiles in concrete with w/b >0.3, there appears a peak in the near surface zone, owing to the drying-wetting cycles in the road environment [27]. This peak phenomenon makes the prediction difficult when using models based on Fick’s law. Therefore, further modifications to the previous ClinConc model are needed in order to make the predicted profiles closer to the measured ones. One of the advantages in the ClinConc model is that the free chloride concentration is taken as a driving force for chloride ingress. The fair matched chloride ingress front between the predicted and the measured one indicates that the prediction for free chloride concentration in concrete, especially in its deeper zone, is reasonably correct. Therefore, the non-error-function shape of chloride profiles is probably due to the changes in chloride binding behaviour along the ingress depth. In our modification, we consider two points of depth, that is, xpeak and xdiff, the former indicates the depth with
maximum chloride content at the peak depth and the latter indicates the depth after which the profile follows diffusion mechanism, as shown in Figure 5.3, in which zone I indicates severe reduction of chloride binding due to the effects of dry-wet and/or carbonation, zone II indicates partial reduction of chloride binding due to partial carbonation and/or calcium leaching, and zone III indicates ordinary diffusion with normal chloride binding. The values of xpeak can be obtained from the measured chloride
profiles, whilst the values of xdiff can be obtained from of the predicted profiles with the
measured ones. The relationships between the values of xpeak / xdiff and water-binder ratio
Figure 5.2. Some results predicted by the previous ClinConc model for concrete with Anl and Slite cement. 0 0.5 1 1.5 2 0 20 40 60 80 100 C l [% b y w t o f b in d e r] Meas. 1 y Meas. 5 y Meas. 10 y Meas. 20 y Pred. 1 y Pred. 5 y Pred. 10 y Pred. 20 y Pred. 100 y #201 Anl w/b 0.4 0 0.5 1 1.5 2 0 20 40 60 80 100 C l [% b y w t o f b in d e r] Depth [mm] Meas. 10 y Meas. 20 y Pred. 10 y Pred. 20 y Pred. 100 y 0 0.5 1 1.5 2 0 20 40 60 80 100 C l [% b y w t o f b in d e r] Meas. 10 y Meas. 20 y Pred. 10 y Pred. 20 y Pred. 100 y #203 Anl w/b 0.3 non-AEA 0 0.5 1 1.5 2 0 20 40 60 80 100 Depth [mm] Meas. 10 y Meas. 20 y Pred. 10 y Pred. 20 y Pred. 100 y #204 Anl w/b 0.35 non-AEA 0 0.5 1 1.5 2 0 20 40 60 80 100 C l [% b y w t o f b in d e r] Meas. 10 y Meas. 20 y Pred. 10 y Pred. 20 y Pred. 100 y #205 Anl w/b 0.5 non-AEA 0 0.5 1 1.5 2 0 50 100 150 Depth [mm] Meas. 1 y Meas. 5 y Meas. 20 y Pred. 1 y Pred. 5 y Pred. 10 y Pred. 20 y Pred. 100 y 0 0.5 1 1.5 2 0 20 40 60 80 100 C l [% b y w t o f b in d e r] Depth [mm] Meas. 20 y Pred. 10 y Pred. 20 y Pred. 100 y #215 Slite w/b 0.4 0 0.5 1 1.5 2 0 20 40 60 80 100 Depth [mm] Meas. 20 y Pred. 10 y Pred. 20 y Pred. 100 y 0 0.5 1 1.5 2 0 20 40 60 80 100 C l [% b y w t o f b in d e r] Meas. 10 y Meas. 20 y Pred. 10 y Pred. 20 y Pred. 100 y #217 Slite w/b 0.3 non-AEA 0 0.5 1 1.5 2 0 20 40 60 80 100 C l [% b y w t o f b in d e r] Depth [mm] Meas. 10 y Meas. 20 y Pred. 10 y Pred. 20 y Pred. 100 y
Figure 5.3. Some results predicted by the previous ClinConc model for concrete with silica fume. 0 0.5 1 1.5 2 0 20 40 60 80 100 Depth [mm] Meas. 1 y Meas. 5 y Meas. 10 y Meas. 20 y Pred. 1 y Pred. 5 y Pred. 10 y Pred. 20 y Pred. 100 y #206 Anl+5%SF w/b 0.4 0 0.5 1 1.5 2 0 20 40 60 80 100 Depth [mm] Meas. 10 y Meas. 20 y Pred. 10 y Pred. 20 y Pred. 100 y 0 0.5 1 1.5 2 0 20 40 60 80 100 C l [% by w t of bi nd e r] Depth [mm] Meas. 10 y Meas. 20 y Pred. 10 y Pred. 20 y Pred. 100 y 0 0.5 1 1.5 2 0 20 40 60 80 100 C l [% b y w t o f b in d e r] Depth [mm] Meas. 10 y Meas. 20 y Pred. 10 y Pred. 20 y Pred. 100 y 0 0.5 1 1.5 2 0 20 40 60 80 100 C l [% by w t of bi nd e r] Meas. 10 y Meas. 20 y Pred. 10 y Pred. 20 y Pred. 100 y #225 Anl+10%SF w/b 0.4 0 0.5 1 1.5 2 0 20 40 60 80 100 C l [% by w t of bi nd e r] Meas. 10 y Meas. 20 y Pred. 10 y Pred. 20 y Pred. 100 y #226 Anl+10%SF w/b 0.3 non-AEA 0 0.5 1 1.5 2 0 20 40 60 80 100 C l [% by w t of bi nd e r] Meas. 10 y Meas. 20 y Pred. 10 y Pred. 20 y Pred. 100 y #227 Anl+10%SF w/b 0.35 non-AEA 0 0.5 1 1.5 2 0 20 40 60 80 100 Depth [mm] Meas. 10 y Meas. 20 y Pred. 10 y Pred. 20 y Pred. 100 y #210 Anl+5%SF w/b 0.5 non-AEA
Figure 5.3. Illustration of the chloride profile with different zones.
Figure 5.4. Relationships between xpeak and w/b.
x
peak
x
diff
III
II
I
0 2 4 6 8 10 12 14 16 18 0.2 0.4 0.6 0.8 xpe ak , m m -2 0 2 4 6 8 10 12 14 16 0.2 0.3 0.4 0.5 0.6 e a , KalkCFigure 5.5. Relationships between xdiff and w/b.
It can be seen from Figures 5.4 and 5.5 that It is reasonable that the relationship between xpeak and w/b follows a logarithmic equation whilst the relationship between xdiff and w/b
follows a linear equation.
peak α lnp βp w x b = + (5.2) diff αd βd w x b = − (5.3)
The values of constants and for different binders are listed in Table 5.1.
Table 5.1 Constants for the calculation of peak and diffusion depths in Eqs (5.2) and (5.3).
Binder p p d d
Anl, Anl+5%SF, Slite, FinStd, FinRpt 16.2 21 200 57.6
Anl+10%SF 17.4 21 70 20.5
FinRpt+44%SL 13.6 16 35 10.7
KalkC 25 32 208 57.6
5.4 Further modification of the model
In the literature the peak phenomenon was often attributed to the effect of drying-wetting cycles [27][28][29]. Besides this effect, we strongly believe that calcium leaching/redistribution plays also a vital role in the changes of chloride binding capacity. In this project the Monte Carlo simulations were carried out to study
the dissolution
behavior of Ca(OH)
2in various chloride solutions. From the simulation results, the
0 10 20 30 40 50 60 70 80 90 100 0.2 0.4 0.6 0.8 di , 0 10 20 30 40 50 60 0.2 0.3 0.4 0.5 0.6 xdi ff , m m
log Q SI K = (5.4)