EFFECT OF CONCRETE COMPOSITION ON RESISTANCE OF CONCRETE TO ACID ATTACK

Transcription

1 EFFECT OF CONCRETE COMPOSITION ON RESISTANCE OF CONCRETE TO ACID ATTACK Robin E. Beddoe and Karl Schmidt Centre for Building Materials, Technische Universität München, Germany Abstract The effect of Portland cement concrete composition on the resistance to static and dynamic acid attack was investigated using a computer model (SATIR) and laboratory experiments. Regular concretes made with quartz aggregate, cement contents 28 to 2 kg/m and w/c ratios.45 to.6, were exposed to a sodium acetate buffer solution at ph 4.5 in a tank test and to flowing hydrochloric acid at ph 2. An ultra high performance concrete (UHPC) and a concrete designed for high acid resistance were also investigated. The model was able to predict corrosion for the wide range of experimental conditions and concrete compositions chosen. Simulations were also performed for acid attack encountered in agriculture and industry; cement content and w/c ratio were varied. The static attack of organic acids was represented by acetic acid with 28 mg/l acetate and ph 4. The attack of flowing mineral acids was simulated with sulphuric acid at ph 2 and 4. The corrosion resistance of concrete continuously exposed to acid of constant strength was found to be improved by low cement contents and w/c ratios providing abrasion is not excessive. The use of acid-resistant supplementary cementitious materials such as fly ash or silica fume to increase the denseness of the microstructure is advantageous.. INTRODUCTION The service life of unprotected concrete structural components can be significantly reduced when exposed to the dissolving attack of, for example, mineral acids, organic acids or ammonium, see [, 2]. The attack can take place under static (e.g. containers for liquid manure, silage or deicing agent run-off at airports) or dynamic (e.g. cooling towers, sewer pipes) conditions. Owing to the dissolution of the binder matrix and aggregate particles, a porous corroded layer of low mechanical strength forms which becomes thicker as time passes. Abrasion due to the mechanical action of flowing liquids, cleaning processes, etc. or degradation caused by precipitation in the near-surface concrete can result in removal of weakened surface concrete thus speeding up the rate of corrosion. Conventional concrete technological measures such as higher cement contents, low w/c ratios do not lead per se to an improvement in the acid resistance of concrete structural components. Although, for example, higher cement contents increase the neutralisation capacity of concrete with respect to the 87

2 acid, the corroded layer becomes more porous causing easier access of the acid to the undamaged concrete at the corrosion front. In the present investigations, tank tests with sodium acetate buffer solutions and flow tests with hydrochloric acid were carried out to simulate attack under static and dynamic conditions in the laboratory. A wide range of composition and performance was covered by the concretes used. The experimental results are compared with corresponding calculations using a computer model for acid attack. Finally, the model is used to investigate the effect of concrete composition on acid resistance for different conditions of attack. 2. CONCRETE COMPOSITIONS AND SPECIMEN PREPARATION Four concretes of regular composition fulfilling the requirements of DIN 45-2 for concrete exposed to chemical attack (XA to XA according to DIN EN 26-) were mixed using Portland cement at contents between 28 and 2 kg/m, w/c ratios between.45 and.6, and A/B 6 mm graded acid-resistant quartz aggregate, Table. Two special concretes with very different compositions were included in the investigations. A UHPC designed for high strength without heat-treatment was produced with silica fume and an optimized aggregate / mm grading. A concrete designed for high acid resistance according to [] was used in which the particles (/6 mm) were packed according to the Fuller-Thompson grading curve [4] so that the binder content was only 2.6 wt.%. The contents of cement (75%), fly ash (2%) and silica fume (5%) in the binder were optimized according to Puntke [5] to obtain a low water requirement. Consequently, the Portland cement content was at 24.2 kg/m much lower than for the other mixes. All the mixes were produced with the same Portland cement CEM I 42.5 R (64.7 wt.% CaO, 9.7 wt.% SiO 2, 5.9 wt.% Al 2 O, 2.8 wt.% Fe 2 O ) and acid resistant quartz aggregate. A superplasticizer based on polycarboxylate ether containing 4 wt.% solids was used for the mixes UHPC and SRB. Table : Composition and properties of the concretes Concrete Cement c Water w Fly ash f Silica fume s Superplasticizer w/c eff * Density f c,28d kg/m kg/m kg/m kg/m kg/m - kg/m MPa Aa Ab A A UHPC ** SRB * w/c eff = w/(c+.4 f + s), ** cylinders (5Ø 5 mm), else 5 mm cubes The fresh concrete was poured into moulds for specimens in the form of prisms (5 mm ) for the tank tests and thick-walled hollow cylinders (outer diameter 5 mm, inner diameter mm, length 245 mm) for the flow tests. The specimens were demoulded after 24 h and stored for at least 28 d submerged in water before testing. 88

3 . EXPERIMENTAL INVESTIGATIONS In the tank test, each concrete prism was stored at 2 C for a total of 2 days in litre sodium acetate buffer solution with an initial ph value of 4.5,.8 mol/l CH COOH and.2 mol/l NaCH COO. Afterwards the specimens were split and the fracture surfaces sprayed with a phenolphthalein indicator solution to determine the depth of corrosion, Figure. Figure : From left to right, corroded layer for concretes A2, UHPC and SRB after exposure to acetic acid in the tank test In the flow test, hydrochloric acid (ph 2) was pumped at ml/h just once through the hollow cylinders and collected at the outflow side. In this manner, fresh acid was continuously supplied to the inner cylinder surface. The collecting vessel was emptied after periods of 7 days and the contents chemically analysed by ICP OES. The ph of the solution at the outflow was also monitored. After 64 days the cylinders were split to determine the depth of corrosion, Figure 2. Figure 2: From left to right, corroded layer for concretes A2, UHPC and SRB after exposure to hydrochloric acid in the flow test 4. COMPUTER MODELLING A numerical model (SATIR) for the prediction of concrete corrosion due to acid attack has been developed at the Centre for Building Materials [6, 7]. The corrosion process can be simulated for concrete of known composition exposed to organic or mineral acids attacking concrete surfaces under static or flowing conditions. The concrete may contain dissolvable (calcite, dolomite) or acid resistant (quartz) aggregate of a given grading. The conditions of attack are defined by parameters which may be determined at the location in question and by chemical analysis of the attacking medium. The main simulation result is the time needed to reach a given - in practice permissible - depth of corrosion. It is also possible to simulate the effect of abrasion. Figure illustrates the mechanisms taken into account by the model. The diffusion of species in the pore solution of concrete is calculated from their concentration distribution in 89

4 the pore solution c i (x) using the well-known extended Nernst-Planck equation for one dimensional diffusion. eff ci Fzi ψ lnγ i J i = Di PC + ci + ci () x RT x x The effective diffusion coefficients for the individual species D eff i are related to the specific values D i for diffusion in water and the pore space P C in which diffusion takes place. As corrosion proceeds, the pore space increases due to dissolution of the binder matrix and, if dissolvable, the aggregate. In measurements of diffusion coefficients for corroded mortar disks [6, 8], it was found that the diffusion flux density is determined primarily by the porosity of the corroded material; the effect of diffusion in the uncorroded material being much smaller. Dissolution of aggregate: Reaction kinetics + i) Consumption of protons H ii) Dissolution of Ca, (Mg ) 2- Neutralisation capacity of aggregate (CO ) Dissolution of binder matrix: Reaction kinetics + i) Consumption of protons H ii) Dissolution of Ca, Fe, Al Neutralisation capacity of binder matrix (C-S-H, Ca(OH) ) 2 Corroded layer Uncorroded concrete Diffusion: Volume fraction and tortuosity of aggregate i) Near-surface concrete Water-filled corrosion space in binder matrix Water-filled corrosion space left by aggregate particles Pore solution: Acid dissociation Solublity of salts i) Precipitation ii) Dissolution Figure : Factors affecting acid attack on concrete Local values of the electric field ψ/ x in () are calculated from the concentration distribution of ions and their charge. The activity coefficients γ i are determined using the wateq Debye-Hückel equation. Now, the dissolution of the binder matrix results in the transfer of Ca 2+ ions from the binder matrix into the pore solution which is described by reaction kinetics. cca = k Ca sca c (2) H t Here, s Ca is the amount of potentially soluble calcium in concrete with respect to pore solution volume and c H the proton concentration of the pore solution. At the beginning of corrosion, s Ca is essentially the neutralisation capacity of the concrete given by the total quantity of calcium calculated from the CaO content of the cement and the cement content of the concrete. The increase in concrete porosity due to removal of calcium from the binder matrix is calculated from the increase in calcium concentration of the pore solution. A constant for the pore volume per mol dissolved Ca 2+ was determined by storing mortar disks (2.5 mm thick) in sodium acetate buffer solutions and comparing the increase in porosity with 9

5 the amount of dissolved calcium [6, 8]. The rate constant k Ca in (2) was found by monitoring the increase in calcium concentration of the storage solution. In addition, the dissolution behaviour of finely ground mortar was also investigated. The dissolution of Fe + and Al + was treated similarly. In the case of dissolvable aggregate, surface dissolution is assumed by the model. Since modelling the mechanical processes which cause loss of corroded surface material during abrasion or surface precipitation is difficult, an upper limit is set on the thickness of a residual corroded layer x RCL defined to exist between the abraded surface of the corroded concrete and a depth at which a ph of, for example, 8.5 is reached. In practice, the depth of the residual corroded layer could be determined by spraying a phenolphthalein indicator solution onto fractured concrete surfaces. In the model, the distance of the abraded surface x S from the original surface of the uncorroded concrete (x=) is given by the conditions xs = xph x ; 8. 5 RCL x ph > x and 8.5 RCL x S = ; xph x. () 8.5 RCL Here x ph8.5 is the depth at which the proton concentration in the pore solution c H (x) first exceeds ph 8.5. As abrasion proceeds, the external boundary is moved to x S, i.e. its position is time dependent. The model considers static attack, where a certain volume of acid of given initial composition is in contact with a given concrete surface area and dynamic attack, where the concrete surface is continuously exposed to fresh acid, the amount depending on the rate of flow. Owing to its reaction with the concrete, the composition of the acid changes in the course of the corrosion process. The boundary condition at the acid/concrete interface is defined such that the concentrations of each individual species in the external acid and in the surface pore solution are equivalent at all times. Charge conservation and the equilibrium of the acid species and water are observed for the external acid and the pore solution throughout the concrete. Figure 4 compares the depths of corrosion measured for the different concrete compositions and experimental conditions with the corresponding values calculated from the model. Depth of corrosion, model [mm] Flow test UHPC Ab SRB UHPC A2 Tank test Ab Depth of corrosion, experiment [mm] Figure 4: Experimental and calculated thickness of corroded layer for the concretes investigated in tank and flow tests In view of the wide range of concrete compositions as well as the very different conditions of attack and specimen geometry, the agreement between the experimental and simulation results is acceptable. At first sight it appears that UHPC offers the best acid resistance because Aa SRB A2 Aa A 9

6 a small depth of corrosion was both measured and calculated for this material. However, the high cement content of UHPC caused the ph of the acid to increase (in the tank test from 4.5 to 4.7) thus reducing the severity of attack, i.e. the testing conditions were not constant. 5. EFFECT OF CONCRETE COMPOSITION 5. Static Acetic Acid Attack The static attack of organic acids, as encountered in agriculture [9], was simulated by the attack of sodium acetate buffer solutions with a total acetate concentration of 28 mg/l and ph values of 4., 4.5 and 5.5 on Portland cement concrete containing quartz aggregate. Although aggressiveness is defined in terms of ph according to DIN EN 26 (XA, XA2, XA, respectively), the concentration of acetic acid c A (.84,.9,. mol/l, respectively) also affects the severity of attack. The acid composition at the concrete surface was, as opposed to the experiments, kept constant thus simulating the worst case situation, i.e. the severity of attack is not diminished by the reaction with the concrete. The resistance of the concrete to attack was characterized by the time needed to reach a corrosion depth of mm. Simulations were performed for w/c ratios between.4 and.6 and Portland cement contents ranging from 24 to 8 kg/m, Figure 5. 8 ph 4. 8 ph 4. with abrasion 7 7 Time to corrosion depth of mm [years] w/c Time to corrosion depth of mm [years] w/c Time to corrosion depth. of mm [years] Cement content [kg/m ] ph SRB 7 A 6 A2.4 w/c 5 Aa UHPC Ab Cement content [kg/m ] Time to corrosion depth. of mm [Jahre] Cement content [kg/m ] 7 ph w/c Cement content [kg/m ] Figure 5: Effect of concrete composition on resistance to static acetic acid attack at constant ph 4. (with and without abrasion), ph 4.5 (including tested concretes) and ph 5.5. Additional simulations were performed for ph 4 with abrasion for a residual corroded layer thickness x RCL of 4 mm, Figure 5. Simulation results for the laboratory concretes at 92

7 constant ph 4.5 are included in the figure. The results indicate that the corrosion resistance of concrete can be significantly improved by the use of low cement contents and low w/c ratios for the attack scenario simulated. As expected, the time needed for the corrosion depth to reach mm depends on the concentration of acid molecules in the attacking medium. The time required increases by % between ph 4. (c A =.84 mol/l) and ph 4.5 (c A =.9 mol/l) and then by 45% between ph 4.5 and ph 5.5 (c A =. mol/l). To understand the effect of w/c ratio and cement content on acid resistance, it should be borne in mind that the rate of corrosion depends on the transport of the acid through the corroded layer to the undamaged concrete. Thus factors which affect the porosity of the corroded layer also affect the rate of corrosion. In general, a reduction in cement content has two possible effects on resistance to acid attack. a) Acid resistance is improved because the volume of dissolvable binder matrix is lower and thus the porosity of the corroded layer is lower as well. b) The neutralisation capacity of the concrete with respect to the acid is lower and thus, in some attack scenarios, the acid resistance too. If concrete surfaces are exposed to large amounts of strong buffer solutions with a high potential concentration of protons in non-dissociated acid molecules, the neutralisation capacity is of secondary importance which means that low cement contents are advantageous. Moreover, low w/c ratios improve acid resistance too because the contribution of the capillary porosity of the original undamaged concrete to the porosity of the corroded layer is less. The dissolution process destroys the tortuosity of the pore system opening pores for faster transport [8]. Water/cement ratios below about.4 are not expected to improve acid resistance much because the capillary porosity is not significantly lower. Thus the acid resistance of the UHPC in Figure 5 (bottom, left) is lower than for the normal concrete mixes owing to its high cement content. In fact, the best acid resistance was calculated for the concrete SRB specially designed for high acid resistance. The corrosion resistance of this concrete is also enhanced by the use of silica fume and fly ash which reduce the actual w/c ratio from.56 to a calculated effective value of.48, see Table, and lower the capillary porosity by about 6%. Providing the conditions of attack do not vary, the depth of corrosion d may be described approximately by a root time law. d = a t (4) c A The constant a depends on the composition of the concrete and the conditions of attack. In this case it is assumed that the effect of any possible precipitation processes as well as abrasion is negligible. The effect of abrasion on corrosion resistance is demonstrated in Figure 5. The time needed to reach a corrosion depth of mm, in this case comprising 6 mm abraded material and 4 mm residual corroded layer, is shortened. Figure 5 (top right) represents an intermediate case between complete removal and retention of corroded concrete. Although low w/c ratios and cement contents are still advantageous, their effect on corrosion resistance is less pronounced, cf. Figure 5 (top left). If the corroded material is completely and continuously removed higher concrete neutralisation capacities are expected to slow down the rate of corrosion because the change in corrosion depth produced by a given amount of acid is 9

8 smaller. This is illustrated by the exposure of concrete pavement to acid rain where the thin corroded layer produced by each rainfall is immediately removed by the mechanical action of vehicle types [] and thus the supply of acid determines the rate of corrosion. 5.2 Flowing Sulphuric Acid Attack The dissolution of concrete in industrial plants exposed to flowing mineral acids was considered using sulphuric acid. The corrosion was simulated for Portland cement concrete made with quartz aggregate (A/B 6 mm) exposed to ph values of 2 and 4, i.e. exposure class >XA and XA, respectively. The ph was kept constant at the concrete surface, i.e. the flow rate was high enough to maintain the severity of attack. Gypsum precipitation and loss of surface material were not considered. Figure 6 shows some simulation results for w/c=.45 and cement contents between 24 and 8 kg/m. w/c =.45 Time to corrosion depth. of mm [years] ph 4 ph 2 Cement content [kg/m ] Figure 6: Effect of cement content on resistance of concrete to flowing sulphuric acid attack In general, the effect of cement content, w/c ratio and ph on corrosion resistance were found to be similar to static acetic acid attack. However, at ph 4 the rate of corrosion due to dissolving attack is negligible for all the concrete compositions considered. Even at ph 2 the time needed to reach a corrosion depth of mm is over years. As opposed to organic acids whose severity is determined by the (non-dissociated) acid concentration c A, the severity of attack by mineral acids is determined by their proton concentration c H. In the present case, the corrosion depth increases with the square root of time and proton concentration. d = a t. (5) c H This formula yields both the depth of corrosion measured in the laboratory after 64 days and the calculated values in Figure 6. Now, corrosion rates of 5 mm per year and more have been reported for cooling tower concrete exposed to ph 4 which are much faster than the calculated values in Figure 6. This difference may be explained by the loss of corroded surface material due to mechanical action and precipitation. By reducing the residual corroded layer thickness in the simulations it was found that the time-dependence of the corrosion depth is given approximately by t p where p changes continuously from.5 (no abrasion ) to. (complete removal of corroded concrete). In the latter case, the corrosion rate is determined by the supply of acid to the concrete surface. If, for example, concrete made with 2 kg/m Portland cement concrete is attacked 94

9 at ph 4, the acid volume supplied per square metre concrete must, in theory, amount to about 72 m for complete corrosion of mm concrete. 6. CONCLUSIONS Based on the results of tank and flow tests as well as computer simulations it is concluded that the use of low w/c ratios and low cement contents in combination with quartz aggregate promotes the acid resistance of Portland cement concrete continuously exposed to acids of constant strength, providing the loss of corroded surface material is not excessive. The use of acid-resistant supplementary cementitious materials such as fly ash or silica fume to increase the denseness of the microstructure is advantageous. If concrete is attacked by buffered organic acids, the rate of corrosion depends on the concentration of the acid rather than just the ph. The rate of corrosion of concrete exposed to mineral acids at ph 4 is slow, but can be significantly accelerated by loss of surface material. The supply of acid to the concrete surface determines the rate of concrete corrosion. ACKNOWLEDGEMENTS The authors express their thanks to the German Research Foundation (DFG) for supporting this project financially. REFERENCES [] Zivica, V., Bajza, A.: 'Acidic attack of cement based materials a review. Part. Principle of acidic attack', Constr. Build. Mater. 5 (2) -4 [2] Bertron, A., Duchesne, J.: 'Attack of cementitious materials by organic acids in agricultural and agrofood effluents - a review'. Proc. workshop on performance of cement-based materials in aggressive aqueous environments, Ghent, Belgium, Sept 27, Ed. Nele De Belie, 5-28 [] Hüttl, R., Hillemeier, B.: Hochleistungsbeton Beispiel Säureresistenz, Betonwerk + Fertigteil- Technik (2) 52-6 [4] Fuller, W.B., Thompson, S.E.: 'The laws of proportioning concrete, Am. Soc. of Civ. Eng., Transactions, Paper No. 5 [5] Puntke, W.: Wasseranspruch von feinen Kornhaufwerken, beton 5 (22) [6] Beddoe, R. E., Dorner, H.W.: 'Modelling acid attack on concrete: Part I. The essential mechanisms', Cem. Concr. Res. 5 (25) 2-29 [7] Beddoe, R.E. und Hilbig, H.: 'Acid attack on concrete - a new computer model', Proceedings 5th International Workshop, Transport in Concrete - Nanostructure and Macrostructure, TRANSCON 7, Essen, June 27 [8] Beddoe, R.E., Dorner, H.W.: 'Prognose der Betonkorrosion infolge Säureangriff', 5. ibausil, Tagungsbericht, Band 2, S , Weimar, Sept. 2 [9] Bertron, A., Duchesne, J., Escadeillas, G.: 'Accelerated tests of hardened cement pastes alteration by organic acids: analysis of the ph effect', Cem. Concr. Res. 5 (25) [] Schießl, P., Wenzl, P.: 'Measurement of changes in texture of existing concrete pavement surfaces with contact-free surface measurement devices and laboratory calibration', Forschung Straßenbau und Straßenverkehrstechnik 995 (28) 95