Alkali-aggregate reaction causes serious problems in concrete hydraulic structures, such as dams and powerhouses. A thorough understanding of the current knowledge and practice in managing this reaction can help owners of existing dams and companies planning to build new dams.
|This article has been evaluated and edited in accordance with reviews conducted by two or more professionals who have relevant expertise. These peer reviewers judge manuscripts for technical accuracy, usefulness, and overall importance within the hydroelectric industry.|
By Chongjiang Du
Alkali-aggregate reaction (AAR) is a chemical reaction between the alkali in Portland cement and reactive minerals in aggregate and additives that takes place when moisture is present. This reaction results in the formation of a hygroscopic gel that absorbs water and expands, causing significant expansion and characteristic cracking of the concrete and ultimately failure of the concrete in worst cases.
AAR often occurs in old concrete hydraulic structures, where the problem was not detected or not properly treated before and during construction. Many concrete dams and other hydraulic structures worldwide have suffered from AAR.1,2,3,4,5,6,7
Today, the reactive mechanism and consequences of AAR are recognized by concrete and dam engineers, and high attention is paid to the problem. The potential for AAR in a new large hydraulic structure should be thoroughly explored before and during its construction. Necessary measures should be taken to prevent or suppress the potential expansion within a tolerable limit.
It is important to understand the current knowledge and practice in managing AAR in the concrete of hydraulic structures.
Types and prerequisites of AAR
There are three subsets of AAR:
— Alkali-silica, a reaction between the alkali hydroxides in concrete and reactive forms of silica in aggregate (e.g., chert, quartzite, opal, strained quartz crystals);
— Alkali-silicate, a reaction between the alkali hydroxides in concrete and reactive forms of silica present in the combined form of phyllosilicates (e.g., chlorite, vermiculite, micas); and
— Alkali-carbonate, a reaction between the alkali hydroxides in concrete and the dolomite crystals present in aggregate.
Alkali-silica is the most common form of AAR. Alkali-silicate and alkali-carbonate reactions are relatively rare. AAR problems have been reported throughout the world, and intensive research has been performed. Concrete dams are particularly monitored for AAR because of their size, large investment, important role, and severe consequences in case of failure.
Research and practice indicate that the following prerequisites must be simultaneously fulfilled for AAR expansion to occur (see Figure 1):
— High content of alkali in concrete (2.4 to 3 kilograms per cubic meter or more);
— Presence of reactive minerals in aggregate; and
— Sufficient moisture supply (at least 80 percent).
If any one of these three conditions is not met, expansion due to AAR cannot occur. Other conditions, such as higher temperature, accelerate the speed of reaction. In addition, well-compacted concrete exhibits a greater volume expansion than poorly-compacted concrete, although well-compacted concrete reduces the contribution of moisture from outside.4,7 Chloride salts, which may be added to mixing water or penetrate mortars after hardening, may aggravate the expansion,. Both sodium chloride and calcium chloride accelerate reactions with reactive silica at elevated temperatures.7,8
In principle, all ingredients of concrete may contribute to its total alkali content. However, the majority of alkali is from Portland cement. The total alkali content in cement, known as total mass of “equivalent alkalies,” is Na2O + 0.658 K2O.
Pozzolans, which are commonly used as a part of cementitious materials in concrete for hydraulic structures, also contribute alkali to concrete. These pozzolans can include fly ash, silica fume, natural pozzolans, and ground granulated blast furnace slag. However, pozzolans consume alkali when they react with lime. Thus, when considering the alkali contribution of pozzolan to concrete, a reduction should be allowed to the alkali content of the pozzolans.
Aggregate containing plagioclase (feldspars), some biotites/micas, glassy rock, and glass may release alkali to concrete. And some admixtures containing sodium and potassium compounds may contribute to the alkali content of concrete. In addition, mixing water may also contain a certain amount of alkali.
In regions of cold weather, de-icing salt containing sodium compounds may increase alkali content on the surface layer of concrete. Soils containing alkali also may increase alkali content on the surface of concrete.
It should be emphasized that the total alkali content in concrete is the more significant index than the alkali content in Portland cement alone.9
Depending on their mineral content, alkali-reactive aggregates are divided into two types: normally and slowly reactive. Normally reactive aggregates are those porous and occasionally hydrated silica minerals (such as opal or chalcedony), as well as a variety of heterogenic rock types (including chert, flint, or certain types of volcano glass). Slowly reactive aggregates include well-crystallized, higher-density quartz rocks (such as greywackes, sandstones, clay-mica slates, or metamorphic rocks). Also, micro-crystalline or imperfectly crystallized quartz (stressed quartz) and granite may cause a similar slow/late reaction. In cement that contains aggregates of this type, the reaction leads to delayed concrete expansion and destruction.
Characteristics of mass concrete of hydraulic structures
Mass concrete for hydraulic structures — such as dams, weirs, locks, and canals — is significantly different from structural concrete. Because of these differences, mass concrete of hydraulic structures exhibits different behaviors as a result of AAR than does structural concrete or mass concrete for other structures.
These differences include:
— Moisture condition. Obviously, hydraulic structures work in an environment with water. Many components work underwater or underground with surrounding groundwater — directly in contact with water and under water pressure. Moreover, it is difficult for water in the concrete interior to evaporate out of the large mass, even for the parts above water. The water consumed during the hydration process of cement is only a fraction of the mix water. On the other hand, water can penetrate up into the upper parts of the concrete. As a result, there is generally sufficient moisture available to the mass concrete over the whole service period. It is well-known that water has a threefold function in an AAR. First, it is the basis of ionization of alkali. Second, it is the carrier of alkali to other parts. Third, it is the source of expansion of the gel produced by AAR.
— Larger aggregate sizes. The maximum size of aggregate used to produce the mass concrete of hydraulic structures usually is 76 to 150 millimeters, whereas the maximum aggregate size for structural concrete usually is 19 to 38 millimeters. Research indicates that expansions of large, small, and fine aggregate are incompatible over the course of AAR, so that a stress differential exists between large aggregates and small and fine aggregates, leading to stress concentration around large aggregates.10 Concrete is vulnerable to cracking as a result of local stress concentrations. Moreover, the reaction of large aggregates with alkali is slower. Consequently, AAR in hydraulic structures usually occurs later.
— Low strength. Mass concrete of hydraulic structures usually is low in strength. The compressive strength of mass concrete for dams normally is 10 to 20 megapascals (MPa) at 90 days, whereas for structural concrete it usually is 25 to 40 MPa at 28 days. Because the tensile strength of concrete is proportional to its compressive strength, the resistance of mass concrete of hydraulic structures to AAR-induced cracking is correspondingly lower than that of structural concrete.
— Low content of cementitious material. The cementitious material content of mass concrete for hydraulic structures is quite low as a result of the low strength requirement. Mass concrete for dams usually contains 100 to 200 kilograms per cubic meter of cementitious materials. From the point of view of AAR, cement provides the alkali. It is true that low content of cementitious material means a low content of alkali in concrete. However, this frequently leads to a misunderstanding that AAR is less serious. In fact, engineering practice and research indicate that AAR of concrete with low content of cementitious material will exhibit different behaviors than that with high content of cementitious material, when the two concretes have the same alkali content.10 The cementitious material not only provides the alkali but also restrains the expansion of AAR. Due to low content of cementitious material in mass concrete of hydraulic structures, this restraint is correspondingly poorer than that of structural concrete. For the same reason, the tensile strength of hardened cement paste of the mass concrete is lower than that of structural concrete. Therefore, the mass concrete of hydraulic structures is more susceptible to cracking than structural concrete as AAR occurs.
— Long life-span. Concrete hydraulic structures are usually expected to be operational for a long time, owing to their importance, extent, high costs, and long construction periods. It is generally expected that a concrete dam should work for at least 80 to 100 years, whereas for a concrete building its design service life-span usually is 50 years.
— Thermal behavior. Because the reaction between cement and water is exothermic, the temperature rise within a large concrete mass (where the heat of hydration of cement is not quickly dissipated) can be quite high. This rise usually is higher than that in structural concrete. As mentioned above, higher temperatures accelerate the speed of AAR.
Therefore, it can be concluded that AAR problems in mass concrete of hydraulic structures may be more serious than in structural concrete and therefore should be subject to more emphasis in design and construction.
Typical destructive effects of AAR
AAR can significantly damage concrete structures. The gel produced by the reaction increases in volume by taking up water. This exerts an expansive pressure, causing unrestrained concrete to expand and restrained concrete to develop large compressive forces. The rate of expansion caused by AAR typically has been found to be 20 to 200 x 10-6 millimeters per millimeter per year, depending on the severity of the reaction and the degree of restraint.1 The volumetric increase can reach a maximum value of 0.2 to 0.5 percent.7 The expansion becomes detectable in about five to ten years after construction, and the most noticeable expansion may be detected in about 15 to 25 years. The expansion may continue for more than 50 years, although in some instances the expansion may cease after 20 to 30 years. The compressive stresses directly caused by AAR are usually within 3 to 4 MPa.1
In most areas of a dam or other hydraulic structure, the volume expansion may result in serious surface cracks, whereas structural cracks may occur in areas of stress concentration or structural discontinuities (such as spillway piers). In the worst cases, the expansion may lead to failure or loss of function of the concrete structures. For example, the spillway gates cannot be opened. Structural geometry is an important factor in AAR damage. Generally speaking, the more complex the geometry of the structure, the more severe AAR damages may occur, although some structures with a simple geometry also can be severely affected by AAR expansion. The cracks can weaken or degrade structures and allow water to enter, which may promote other destructive action, such as the freeze-thaw action. Cracking can reduce the service life of a concrete structure and require expensive repair or even replacement of the structure.
|Cracking caused by AAR at the downstream end of a pier on a concrete weir.|
Strength loss of mass concrete is another frequent effect of AAR. Test results indicate that tensile and shear strengths tend to decrease first, while compressive strengths frequently remain close to normal levels.1 In most cases, strength loss is not a problem. However, the reduction in tensile and shear strength at horizontal joints will reduce the dam structure’s stability against static and especially seismic loads. This occurs because the horizontal lift joints are planes of weakness that are exacerbated by AAR-induced movements. In turn, large cracks may introduce new planes of weakness.
|Alkali-aggregate reaction caused cracking of the concrete in the spillway pier at Matala Dam in Angola. The 1,000-meter-long dam was constructed in the 1950s.|
In powerhouses, the most destructive effects of AAR are associated with misalignment, malfunction of the units, loss of clearance of the runners, and deformation of the stator base into an oval shape. Stress concentrations will result in local overstressing in substructure concrete elements. Diagonal cracking in draft tube piers has been observed. Because of the overall expansion, distortions of powerhouse superstructures have occurred in some cases.1,13
In arch dams, substantial vertical and upstream deformations have been observed. In gravity dams, the areas of most concern are at changes of geometries where large shear stresses may occur, causing serious cracking. Differential expansions caused by different rates of AAR within the mass concrete structures may cause loosening of the horizontal lift joints and thus reduce the stability of the dam. The greater upward deformation in the central portion of a dam relative to the abutments may cause inclined cracks adjacent to the abutments in the upper portion of the dam.3
For spillways, concrete movements may frequently cause loss of spillway gate clearances that will hinder the gate operation and flood discharging capacity, or even block the opening of the gates. Deformations in spillways and piers frequently are accompanied by inclined cracking.
Testing methods for AAR
Researchers have developed several test methods to identify potential reactivity of aggregate. These testing methods may be classified into three types: petrographic examinations, expansion tests, and chemical analyses. The following explanations use ASTM International standards.
ASTM C295, Standard Guide for Petrographic Examination of Aggregate for Concrete, describes the process of identifying the types and properties of minerals in aggregate or concrete by observation, using a microscope or other aids. This method depends on the competence and reliability of the person performing the tests. It can qualitatively determine the possibility that AAR may occur, but it cannot quantitatively predict whether the AAR will be deleterious. Because of these uncertainties, this method generally is used as one part of an investigation.
In hydraulic engineering, the method most commonly used to quantitatively determine the potential of AAR is the expansion test. In an expansion test, mortar bars or concrete prisms are made using the aggregate to be investigated. These specimens are placed in a specified condition, and expansion is measured. Because AAR takes years to complete under ambient conditions, measures are taken to accelerate the reaction. In general, two types of testing methods are applied in practice.
ASTM C1260, Standard Test Method for Potential Alkali Reactivity of Aggregate, is a short-term accelerated test using mortar bars that is probably the most widely used test method. The mortar bars are immersed in a 1N solution of NaOH at 80 degrees Celsius (C). The solution is supposed to provide a sufficient external source of alkali to complete any reaction, and the alkali content of the cement is supposed to have little or no influence. The accepted maximum expansion for innocuous aggregates is 0.1 percent at 14 days after the zero reading or at 16 days after casting. Between 0.1 and 0.2 percent is inconclusive, and higher than 0.2 percent is deleterious. Experience shows that this test method is best suited to normally reactive aggregates, while some slowly reactive aggregates (for example, some gneiss, quartzite, metabasalt and granite) have been found to be deleteriously expansive in field performance even though their expansion in this test is less than 0.1 percent at 16 days.
ASTM C1293, Standard Test Method for Determination of Length Change of Concrete Due to Alkali-Silica Reaction, is a longer-term testing method using concrete prisms. NaOH is added to the concrete mixing water to increase the alkali content of the mixture, expressed as Na2O equivalent, to 1.25 percent by mass of cement. This value is chosen to accelerate the process of expansion, rather than to reproduce field conditions. The concrete prism samples are placed in an environment of 100 percent relative humidity at a temperature of 38 C for up to one year. If the average expansion of three samples is equal to or greater than 0.04 percent at one year, the aggregate will be classified as potentially deleteriously reactive.
Theoretically speaking, ASTM C1293 is a more realistic method than ASTM C1260. Experiences show that this method is also more suitable for slowly reactive aggregates. However, a one-year testing duration often is deemed too long, especially when a dam is under construction. It may be difficult to ascertain that the aggregates used one year ago are still representative of those currently being used. In addition, previous work indicates that this method is in some cases less conservative than ASTM C1260 and more likely to allow some potentially deleterious aggregates.6 It should be noted that the title of ASTM C1293 is not fully correct. In fact, the method also can be used for tests of alkali-silicate and alkali-carbonate reaction.
ASTM C289, Standard Test for Potential Alkali-Silica Reactivity of Aggregates (Chemical Method), uses chemical analyses to identify potential reactivity of aggregates that may be deleterious or innocuous when used with high-alkali cement. Samples of crushed and sieved aggregates are reacted with an alkaline solution at 80 C. Aggregate reactivity is evaluated by measuring the amount of dissolved silica and the reduction of alkalinity in the reaction alkali solution. Similar to ASTM C1260, the tests require a short duration (24 hours), which is one primary advantage. However, this test method seems not to be reliable and fails to properly identify slowly reactive aggregate.6
For an existing structure, extracted samples can be examined according to ASTM C856, Standard Practice for Petrographic Examination of Hardened Concrete. The procedures include:
— Detailed visual examination of the samples;
— Examination of freshly fractured surfaces and lapped sections of samples using a stereomicroscope;
— Examination of oil immersion mounts and thin-sections of the samples using a petrographic microscope;
— Examination of polished sections and powder mounts in a scanning electron microscope with energy-dispersive X-ray microanalysis; and
— X-ray diffraction analysis of mortar fractions of concrete.
In addition, the application of uranium acetate and subsequent examination by ultraviolet (UV) light can be used to diagnose the presence of AAR. This non-destructive test relies on the principle that AAR gel will readily exchange alkali metal ions with those of uranium, which is very fluorescent under UV light.
Moreover, monitoring the existing structures with embedded instruments (such as plumb lines, joint meters, strain meters, and extensometers) and geodetic survey systems are effective methods to detect and measure evolution of concrete swelling in dams.1
There is no universally valid standard testing method for all cases of AAR. In particular, the existing methods are not specially developed for mass concrete of hydraulic structures. Methods to evaluate potential for expansive AAR of hydraulic structures are based on the methods developed for general structural concrete, as described above. The author’s experience shows that the test methods should be jointly used for hydraulic structures as an expedient substitute. In any case, concrete should be tested according to ASTM C1260. In doubtful cases, especially for slowly reactive aggregates, ASTM C1293 and other methods should be applied. As the example below shows, in some cases, the test results of ASTM C1293 may be more indicative and suggestive.
Analysis of the concrete in Matala Dam in Angola is an example of the use of various testing methods to diagnose AAR. The 1,000-meter-long dam was built in the 1950s.11 Serious cracking has occurred in mass concrete in all areas of the dam below and in the vicinities above the maximum wetted perimeters. Petrographic analysis indicated the aggregate used to produce the dam concrete is biotitic granite with high undulatory extinction and high content of K-feldspar. X-ray analysis showed the presence of various minerals, such as ettringites, quartz, calcium carbonates, calcium hydroxide, and other materials with a high content of alkalis, chlorides, and clay minerals. Microscopic examination revealed intensive cracking and gel formation through the mortar and aggregate matrix. Tests according to ASTM C1260 resulted in a relatively low expansion rate, with a value of 0.08 percent at 16 days — no clear evidence of AAR. However, one-year test results performed according to ASTM C1293 showed a high level of expansion with values of 0.22 and 0.36 percent for concrete in surface and deep samples, much higher than the threshold value of 0.04 percent and with a tendency to expand further.11 This clearly indicates the presence of AAR at Matala Dam.
It should be noted that AAR with granite aggregate has rarely been reported, and AAR tests for this type of aggregate have had mixed success. One report provides the results of testing over 17 years of the concrete at the Three Gorges Project.12 Results similar to the case at Matala Dam were obtained, which confirmed that the granite will react with alkali very slowly, especially in the first decade.
Preventing and mitigating AAR for new structures
Because repair and rehabilitation of concrete hydraulic structures is costly — primarily in relation to emptying the reservoir and interrupting operation — AAR should be primarily prevented and mitigated for a new hydraulic structure. Repair and rehabilitation may only be regarded as “the solution of last resort” for existing structures for which proper precautions were not taken during construction.
Measures to prevent and mitigate AAR involve eliminating one of the three prerequisites and/or changing the nature of the reaction by introducing admixtures. Because it is difficult to fully avoid the water ingress into mass concrete due to the nature of hydraulic structures, efforts are focused on the other two aspects.
— Limiting the alkali content in concrete. Because a link has been found between the use of Portland cement with alkali content greater than 0.6 percent Na2O equivalent in concrete and a more severe incidence of AAR, cement with less than 0.6 percent alkali content should be used if available. For potentially reactive aggregates, a maximum alkali content of 0.4 percent in cement is recommended. However, by itself, this is not an absolutely reliable method to control AAR, and other measures should also be taken.
— Using non-reactive aggregate. This may be aggregate that has historically performed well or aggregate shown to be non-reactive by tests. However, it is difficult to demonstrate the historical performance because the noticeable AAR deterioration may occur 15 years or more after construction. It should be noted that non-reactive aggregate frequently is not available for many hydraulic structures from the economical point of view.
— Adding pozzolans or slags to concrete. Research indicates and practice has confirmed that adding ground granulated blast furnace slag and pozzolanic materials (raw or calcined natural pozzolans, fly ash, rice husk ash, silica fume, and metakaolin) to the concrete mix can suppress and mitigate AAR.8 These materials, having a high content of reactive silica and low levels of calcium and alkali, tend to be efficient in controlling AAR. They may be named as mitigative materials. The mechanism by which a pozzolanic material or slag inhibits the potential AAR distress is well-documented.8 The effects of a pozzolan or slag will depend on the particular material, reactivity of the aggregate, and alkali content of the Portland cement.
In general, testing should verify the effectiveness of the pozzolan or slag in reducing the expansion potential. As a rule of thumb, the minimum replacement of 25 percent cement with Class F fly ash or Class N pozzolan should be used, while the minimum replacement of 30 percent cement with ground granulated blast furnace slag could be recommended. For example, limestone aggregate is being used to produce the roller-compacted concrete for Gomal Zam Dam in Pakistan. Tests, according to ASTM C1260, showed that expansion of the concrete specimen without using pozzolan at 16 days was 0.262 percent, indicating potentially deleterious expansion. However, when 30 percent of the Portland cement was replaced with fly ash, expansion at 16 days was reduced to 0.088 percent, indicating innocuous behaviour. Similarly, if 40 percent of the cement was replaced with ground granulated blast furnace slag, expansion of the concrete specimen at 16 days was reduced to below 0.1 percent.
— Using lithium compounds: As an electrochemical method, lithium compounds, especially lithium nitrate (LiNO3), can be added to the concrete mix to counter and mitigate AAR.7,8 However, special precaution should be taken, because lithium hydroxide (LiOH) and lithium carbonate (Li2CO3) have been found to increase the expansion of alkali-carbonate reactive aggregate. Further, some lithium compounds, in insufficient quantities, can actually increase the expansion. This is known as the pessimum effect: At a certain lithium level the concrete will expand significantly, whereas at other levels expansion may be negligible. The lithium nitrate does not exhibit a pessimum effect. Although lithium compounds have been used to mitigate AAR expansion in some structural concrete, research is needed to evaluate the effect on mass concrete.
Remedying AAR damage in existing structures
Prior to any remedies, the structural effects of AAR should be thoroughly investigated to determine the extent of AAR deterioration and necessity of repair, as well as repair techniques. The hydraulic structures should be rehabilitated, provided that stability or serviceability is of direct concern. In some severe cases, the dam may be completely replaced, as was the case at Maentwrog Dam in the United Kingdom,1 or a new spillway may be constructed, as at Chambon Dam in France. Due to the complexity of the hydraulic structures and the extent and nature of AAR, remedial programs for an existing hydraulic structure must be individually established.
The measures explained below are extracted from repair and rehabilitation of several concrete hydraulic structures. Most remedial measures fall into short-term solutions, because the AAR problem cannot be defined in the long term and it is difficult to predict the ultimate, maximum value of expansion.
— Lowering the pool level. If dam stability is compromised, lowering the reservoir water level may be considered the first step to ensure dam stability before any remedial work begins. This is regarded as a temporary measure.
— Installing anchors. Vertical anchors may be installed in dams to enhance the shear capacity of the horizontal lift joints. This measure is suitable for dams where the horizontal lift joints are weakened due to AAR effect. In addition, horizontal and inclined anchors can be installed in piers and powerhouses. At Hiwassee Dam in the United States, 140 seven-wire-strand tendons post-tensioned to stress levels between 25 and 35 percent of the ultimate load were installed.13
— Cutting slots. Diamond wire saw cutting is frequently used at dams affected by AAR to release excessive stresses and restore clearances. Thin slots of 10 to 15 millimeters may be cut through the dam, spillway, intake, or powerhouse. If the AAR expansion continues, the saw cutting may be repeated. For instance, at Mactaquac Generating Station in Canada, diamond wire saws have been used to cut slots in the spillway, power intake, and powerhouse in 1988, 1989, 1992, and 1995, as well as to re-cut some slots in 1999 and 2000.14 In the powerhouse, slots were cut between each of the six units, with the objective of relieving the effects of the distortions caused by longitudinal movements/thrust, and seven longitudinal slots were cut. This is a continuing operation, with cuts made on a yearly basis.
— Grouting the concrete. To control AAR-induced leakage, grouting the mass concrete may be considered. In several dams, grouting has been performed with cement grout for cracks wider than 0.5 millimeter and with chemical agents for smaller cracks.1,7 Practice demonstrates that grouting can alleviate the leakage but cannot stop the AAR process.
— Installing a membrane. Covering the dam with a geomembrane can be used to prevent water ingress into concrete and to control leakage. At Pracana Dam in Portugal, a 2.5-millimeter-thick polyvinyl chloride (PVC) membrane and a 1.5-millimeter-thick geotextile were placed on the upstream face to provide waterproofing.3 Similar to grouting, this measure can effectively alleviate the leakage but cannot fully stop the AAR process.
— Treating concrete surfaces. In some cases, concrete surfaces are treated, including removing calcite formation of concrete surfaces to a depth of 3 to 5 centimeters into sound concrete and then applying reinforced concrete or epoxy grout or similar coatings. The surface treatments can caulk cracks and help protect embedded reinforcement and reinstate the integrity of the cracked concrete. However, it will not significantly retard the rate of reaction and expansion. New cracks will inevitably form as the reaction continues.
1Charlwood, R.G., “A Review of Alkali-Aggregate Reaction in Hydro Plants and Dams,” International Journal on Hydropower and Dams, May 1994, pages 73-80.
2Alkali-Aggregat-Reaktion (AAR) in der Schweiz, Verband der Schweizerischen Zementindustrie (in German), Bern, Switzerland, 2005.
3Curtis, D.D., “A Review and Analysis of AAR-Effects in Arch Dams,” Proceedings of the 11th International Conference on Alkali-Aggregate Reaction, Centre de Recherche Interuniversitaire sur le Beton, Universite Laval, Quebec, Canada, 2000.
4Hunkeler, F., C. Merz, and P. Kronenberg, Alkali-Aggregat-Reaktion (AAR) – Grundlagen und Maßnahmen bei neuen und bestehenden Kunstbauten (in German), Bundesamt für Strassen ASTRA, Bern, Switzerland, 2007.
5Li, J.Y., “Alkali-Aggregate Reaction in Dam Concrete of China,” Water Power (in Chinese), Volume 31, No. 1, 2005, pages 34-37.
6Malvar, L.J., et. al., Alkali-silica Reaction Mitigation – State-of-the-art, Technical Report TR-2195-SHR, Naval Facilities Engineering Service Centre, Port Hueneme, Calif., United States, 2001.
7Saouma, V., and Y.P. Xi, Literature Review of Alkali Aggregate Reactions in Concrete Dams, Report CU/SA-XI-2004/001, Department of Civil, Environmental, & Architectural Engineering, University of Colorado, Boulder, Colo., United States, 2004.
8ACI Committee 221, State-of-the-art Report on Alkali-Aggregate Reactivity, Report ACI 221.1R, American Concrete Institute, Farmington Hills, Michigan, United States, 1998.
9Farny, J.A., and B. Kerkhoff, “Diagnosis and Control of Alkali-Aggregate Reaction in Concrete,” Technical Code IS413, Portland Cement Association, American Concrete Pavement Association, and National Ready Mixed Concrete Association, 2007.
10Wang, A.Q., and C.Z. Zhang, “Alkali-Aggregate Reactivity of Concrete Hydraulic Structures,” Journal of Hydraulic Engineering (in Chinese), No. 2, 2003, pages 117-121.
11Matos, D.S., and A. Camelo, “Rehabilitation of the Matala Dam,” Proceedings of HYDRO 2004, AquaMedia International, Surrey, United Kingdom, 2004.
12Wang, Z.Q., L. Xiao, and Z. Li, “Alkali Activity of Granite Aggregates and Control in Concrete of the Three Gorges Project,” Proceedings of the 2nd International Conference on Long Term Behaviours of Dams, Graz University of Technology, Graz, Austria, 2009.
13Singhal A.C., and L.K. Nuss, “Cable Anchoring of Deteriorated Arch Dam,” Journal of Performance of Constructed Facilities, Volume 5, No. 1, February 1991, pages 19-36.
14Curtis, D.D., “Analysis and Structure Response to Recent Slot Cutting at Mactaquac Generating Station,” Proceedings of the 11th International Conference on Alkali-Aggregate Reaction, Quebec, Canada, 2000.
Chongjiang Du is a concrete and dams expert with Lahmeyer International GmbH.