»The future model for chloride ingress is based on observations from 20 years of natural exposure By Jens Mejer Frederiksen Managing Engineer B.Sc. Civ.Eng. (hon) »Chloride ingress in concrete, Problem • Too fast chloride ingress in concrete will cause too early corrosion on the steel reinforcement • Chloride can be transported in the moist parts of the concrete pores/defects in many different ways: • Diffusion • Due to convection (moisture movements) • Migration due to an electrical field • Chloride can react or interact with the binder in concrete • Therefore, to define a general model for chloride ingress into concrete is pretty complicated 2015-04-27 2 » Today’s agenda • Problem • Methodology o Experimental Ø Exposure Ø Specimens Ø Analysis, tendencies o Models Ø Fick’s 2nd Law Ø Solutions Ø Constitutive formulas Ø Misuse of models – on obtained data Ø Misuse of models – in predictions o Optimisation • Results o A tool for design/specification of concrete » Today’s agenda • Problem • Methodology o Experimental Ø Exposure Ø Specimens Ø Analysis, tendencies o Models Ø Fick’s 2nd Law Ø Solutions Ø Constitutive formulas Ø Misuse of models – on obtained data Ø Misuse of models – in predictions o Optimisation • Results o A tool for design/specification of concrete »Chloride ingress in concrete, Problem • The most important problem at today’s lecture is the short time available for this topic/presentation! 2015-04-27 5 »Chloride ingress in concrete, Problem 2015-04-27 6 Source: Betonhåndbogen, Chapter 19.3 (http://www.betonhaandbogen.dk/bogen-i-kapitler) » Prediction of the service life is in many cases closely related to Prediction of time to corrosion initiation Mosquée Hassan II, Casablanca The Great Belt Link Constructed (Constructed: 1987-1993) No requirements! Repaired (!!): 2005-2008 1987-1998 Requirement 100 years New Port Tanger Med. I Constructed 2004-2007 Requirement 100 years »Kyösti Tuutti’s service life phases 2015-04-27 8 »Kyösti Tuutti’s service life phases Structural integrity Initiation phase Acceptable level of deterioration Service life Time 2015-04-27 9 »Chloride ingress in concrete Protection against too fast chloride ingress was in focus for the durability design of the Great Belt Link, Denmark, 1987 2015-04-27 10 Source: Betonhåndbogen, Chapter 19.3 (http://www.betonhaandbogen.dk/bogen-i-kapitler) »Chloride ingress in concrete Protection against too fast chloride ingress was also in focus for the durability design of the breakwater structures of the new harbours in Tangier, Morocco, 2004 … and a lot of other major and minor projects around the world 2015-04-27 11 »Chloride ingress in concrete • Since approx. 1988 the way to measure chloride content of concrete has been agreed internationally (the common standards are older, but in practice a number of alternative methods were introduced and influenced the “de facto standard”). • Since approx. 1991 the way to measure calcium content of concrete has been agreed internationally. • Therefore, nowadays chloride ingress can be observed pretty easy and with a correct result. 2015-04-27 12 » Today’s agenda • Problem • Methodology o Experimental Ø Exposure Ø Specimens Ø Analysis, tendencies o Models Ø Fick’s 2nd Law Ø Solutions Ø Constitutive formulas Ø Misuse of models – on obtained data Ø Misuse of models – in predictions o Optimisation • Results o A tool for design/specification of concrete »Chloride ingress in concrete • It is well-known that we can learn a lot from laboratory experiments • But it is also well-known that data from the laboratory are difficult to use in natural exposure conditions • The main problem when we are trying to gather information about chloride ingress in various environments and for various types of concrete is the lack of a good homogenous set of data • This was foreseen by a group of swedes in the early 1990’s - the exposure station in Träslövsläge was formed 2015-04-27 14 »Träslövsläge Marine exposure station under the supervision of SP in Borås Specimens Concrete panels (1000 ´ 700 ´ 100 mm) Exposure: • Marine atmospheric zone • Marine splash zone • Marine submerged zone (14 ± 4 g chloride/l) 2015-04-27 15 Träslövsläge 2015-04-27 16 » Natural exposure in Träslövsläge •ATM •SPL •SUB » Natural exposure in Träslövsläge • 40 types of concrete • Profiles measured at 1 to 5 different exposure times • In total more than 200 chloride and calcium profiles • More than 1000 points on profiles •ATM •SPL •SUB Measured chloride profiles » Today’s agenda • Problem • Methodology o Experimental Ø Exposure Ø Specimens Ø Analysis, tendencies o Models Ø Fick’s 2nd Law Ø Solutions Ø Constitutive formulas Ø Misuse of models – on obtained data Ø Misuse of models – in predictions o Optimisation • Results o A tool for design/specification of concrete » Fick’s 2nd law – the error function solution 2 ì ¶C ¶ C = D 2 , x > 0, t > 0 ï ¶x ï ¶t ïC (0, t ) = C , t > 0, (constant ) s í ïC ( x ,0) = 0, x > 0 ï ï lim C ( x , t ) = 0, t > tex î x ®+¥ The simple error function solution Two parameters constant in time Time Cs = Csa = constant for all t Diffusivity (D) Boundary condition (Cs) æ x ö C ( x, t ) = Ci + (Cs - Ci )erfcç ÷ è 2 Dt ø constant in time Time D = Da = constant for all t Measured chloride profiles Da is decreasing Fitted parameters from chloride profiles - Da is decreasing Fitted parameters from chloride profiles - Da is decreasing Applied model – three parameters constant in time Diffusivity (D) Boundary condition (Cs) æ ö 0 . 5 x ÷ C ( x, t ) = Ci + (Csa - Ci )erfcç ç (t - t ) D (t - t ) ÷ ex a ex ø è Time decreasing in time Time D changed into Da An average over the exposure time æ tex ö Da (t ) = Daex ç ÷ è t ø a Measured chloride profiles Another problem: Csa is increasing! Fitted parameters from chloride profiles Csa is increasing Fitted parameters from chloride profiles Csa is increasing »Poor correlation for constant Da and Csa from 10 years of exposure to 20 years Plot of • K0,1 for 10 years of exposure vs. • K0,1 for 20 years of exposure K Cr -1 æç C - Ci = 2 × Da × erfc ç r è C sa - C i ö ÷× 1 ÷ ø 2015-04-27 30 » Fick’s 2nd law – Leif Mejlbro’s solution 2 ì ¶C ¶ C = D 2 , x > 0, t > 0 ï ¶x ï ¶t ï p íC (0, t ) = t , t > 0, p > -1 (constant ) ïC ( x ,0) = 0, x > 0 ï ï lim C ( x , t ) = 0, t > tex îx ® +¥ » Fick’s 2nd law – Leif Mejlbro’s solution C ( x , t ) = C i + (C sa (t - tex ) - C i ) ´ Yp ( 0.5 x (t - tex )Da (t - t ex ) Ψp is a new function – called “Mejlbro’s Psi function”, and is simply the error function complement (erfc) made general, i.e. Ψp=0(z) = erfc(z) = 1-erf(z) ) increasing in time Time Diffusivity (D) Boundary condition (Cs) The Mejlbro-Poulsen Model decreasing in time Time C ( x , t ) = C i + (C sa (t - tex ) - C i ) ´ Yp ( 0.5 x (t - tex ) Da (t - tex ) ) The Mejlbro-Poulsen Model has four parameters 0.5 x C ( x, t ) = Ci + (C sa (t - tex ) - Ci ) ´ Yp ( ) (t - t ex ) Da (t - tex ) a C sa æ ö t æ ö ex = C i + S ´ ç (t - t ex ) ´ D aex ´ ç ÷ ÷ ç t ø ÷ è è ø p S = S p /(tex Daex ) p æ t ex Da (t ) = Dach (t ) » Dav (t ) = Dav,ex çç è t a ö ÷÷ ø »Constitutive models are based on laboratory results – HETEK 1995 » Constitutive formulas – 17 coefficients! ( c) eqv wL D = W + k air ,D × L C + FA × k FA, D + MS × k MS , D » Constitutive formulas – 17 coefficients! ( c) eqv wL D = W + k air ,D × L C + FA × k FA, D + MS × k MS , D ( ( c) ) D1 = B × eqv wL D100 æ 1 ö = D1 ç ÷ è 100 ø N D × k D1, env[mm2/year] a [mm2/year] a = k a ,env » Constitutive formulas – 17 coefficients! ( c) eqv wL D = W + k air ,D × L C + FA × k FA, D + MS × k MS , D ( ( c) ) D1 = B × eqv wL D100 æ 1 ö = D1 ç ÷ è 100 ø N D × k D1, env[mm2/year] a [mm2/year] C1 = ( A × eqv (w / c )C + k air ,C × L ) × k C1 ,env C100 = C1 × kC100 ,env eqv( w / c ) C = a = k a ,env [% mass binder] [% mass binder] W C + FA × k FA,C + MS × k MS ,C » Today’s agenda • Problem • Methodology o Experimental Ø Exposure Ø Specimens Ø Analysis, tendencies o Models Ø Fick’s 2nd Law Ø Solutions Ø Constitutive formulas Ø Misuse of models – on obtained data Ø Misuse of models – in predictions o Optimisation • Results o A tool for design/specification of concrete » The optimised coefficient for the constitutive equations Coefficients in transport Coefficients in binding 1. B = 771 [mm²/year] 1. A = 5.3 [% mass binder] 2. N = 1.19 2. kair,C = 0.11 3. kair,D = 0.04 [kg/%] 3. kFA,C = -1.33 4. kFA,D = 5.3 4. kMS,C = -0.49 5. kMS,D = 10.7 5. kC1,ATM = 0.56 6. kD1,ATM = 0.15 6. kC1,SUB = 1.1 7. kD1,SUB = 0.54 7. kC100,ATM = 1.1 8. ka,ATM = 0.42 8. kC100,SUB = 2.1 9. ka,SUB = 0.44 » The resulting optimisation » How well can we estimate 20 years data with a model based on 10 years data? » How well can we estimate 20 years data with a model based on 10 years data? »Poor correlation for constant Da and Csa from 10 years of exposure to 20 years Plot of • K0,1 for 10 years of exposure vs. • K0,1 for 20 years of exposure K Cr -1 æç C - Ci = 2 × Da × erfc ç r è C sa - C i ö ÷× 1 ÷ ø 2015-04-27 44 » Today’s agenda • Problem • Methodology o Experimental Ø Exposure Ø Specimens Ø Analysis, tendencies o Models Ø Fick’s 2nd Law Ø Solutions Ø Constitutive formulas Ø Misuse of models – on obtained data Ø Misuse of models – in predictions o Optimisation • Results o A tool for design/specification of concrete » A tool for design/specification of concrete • With the found coefficients the chloride ingress into a variety of concretes made with CEM I + Silica Fume + Fly ash – with or without air entraining can be calculated. • We do however miss good and reliable data for chloride threshold values in order to make the service life design. • Also the expected requirements for acceptance testing of the concrete’s chloride ingress parameters can be an output – not included yet » Concluding remarks – and further studies • The set of data from Träslövsläge is probably the most comprehensive set of data we have • The data set do have limitations (e.g. small and thin specimens, concrete with fly ash has a limited presence, lack of corrosion data – though some do exist) • If we want to have a tool for desktop design of chloride resistant concrete the data set forms a good basis for further development • The optimisation procedure applied here is simplistic – much better, but more complicated tools are available, e.g. for covariance analysis – not included yet • Data from the road environment do exist and ought to be given the same type of attention – not included yet »Thanks for your attention Follow ALECTIA www.linkedin.com/company/alectia www.alectia.com