Category The Alkali-Silica Reaction. in Concrete

Whenever alkali-reactive aggregates are encountered, it has been customary to use cement of low alkali content-the limit being 0.6% Na2O eq. As elsewhere in the world, availability of low-alkali ordinary Portlant cement in India has been somewhat limited. Figure 11.18 indicates that the average level of alkalis in cements in India has increased during the last two decades26. This, to a large extent, is due to the manufacturing technologies adopted for conservation of energy and requirements of environmental protection, in that modern dry – process cement plants require hot exit gases containing the volatiles as well as the kiln dust to be recirculated in the process stream and not vented to the atmosphere. Use of blended cements like Portland pozzolana cement and Portland slag cements would also tend to increase the level of total alkalis in the cement, because pozzolanas and slags in general contain alkalis higher than in the cement clinker. In India, the proportion of blended cements is 70% of the total production and dry-process cement plants now constitute 72% of the total installed capacity.

A comprehensive investigation has been carried out on the role of indigenous blended cements as well as pozzolana and slags used in commercial production of such blended cements in India in alleviating ASR27. The results show that, in general, blended cements are helpful, and optimum results are obtained when the amount of substitution of pozzolana or slag is relatively high, i. e. of the order of 25-30% in the case of pozzolana and more than 50% in case of slags. Although Indian Standard specifications permit pozzolana contents in blended cements to vary between 10 and 25%, the average in practice is of the order of 11-15%. Similarly, the slag content in Portland slag cement is less than 50% whereas the maximum permitted is 65%. In the manufacture of blended cements with a prefixed quantity of pozzolana or with slags interground in the cement to meet the other requirements of specifications, the flexibility to add larger doses of cement substitutes becomes somewhat restricted.

One question that has often worried engineers is the safe limit of total alkalis in blended cements when the additives (pozzolana or slag) are not separately available for analysis. Depending upon the hydraulic activity of the slag or pozzolana, part of the alkalis contributed by them becomes available in the pore solutions28. In certain specifications, a limit of 0.9% total alkalis in the case of Portland slag cements in which the slag content is greater than 60% has been suggested10. No such limit has, however, been established for commercially produced Portland pozzolana cements. The investigation cited showed that a limit of 0.6% total alkalis in ordinary Portland cements corresponded to a limit in the case of Portland pozzolana cements of the order of 0.8-0.9% (Figure 11.19). Nevertheless, the safe values depend upon a host of factors such as the chemical composition of the cement clinker and the pozzolana, the reactivity of the pozzolana to lime-water systems and the reactivity of the aggregates, thereby making any generalisation hazardous. It is prudent, therefore, to establish a safe aggregate-cement-pozzolana (or slag) combination by prior trials. For many of the new constructions reported in 11.3, use of active pozzolana as part replacement of cement has been envisaged.

In addition to lowering the total soluble alkali content in the concrete to the extent that cement is replaced by active silicious pozzolana and hydraulic slags, they also combine with the CH liberated during the hydration of cement. It has been reported that if the CH liberated can be fully consumed by large proportions of slags, ASR would not occur29. In the context of the foregoing, modified cement compositions having no C3S phase or a lower C3S phase merit consideration12. In these cement systems, the amounts of CH liberated upon hydration are considerably lower. Use of such cements with known reactive aggregates is presently under investigation30.

Instances of ASR in concrete structures in India have mainly been due to the presence of silicious aggregates such as quartzites, granites, granodiorites, granite porphyry and diorites etc containing strained quartz. The potential reactivity of these slowly reactive aggregates could not be detected by the test procedures and evaluation norms existing at the time of construction. This has led to modifications in the test methods and adoption of revised threshold values for mortar bar expansion tests, according to which aggregates proposed for many new constructions are now judged as being potentially reactive. The use of low-alkali cements, along with relatively large dosages of active pozzolana, is contemplated in such situations. Although the availability of low-alkali cements is somewhat restricted as yet, it has been possible to meet the demand through indigenous sources.

*Petrographically similar to alkali-silica reactive granite in 70-year-old structure. f Alkali reaction identified in part of one 30-year-old structure.

^Alkali reaction identified in many structures.

§Alkali reaction identified in small part of one 20-year-old dam (but the use of fly ash has prevented reaction in most of this structure).

to raise the Na2O eq. to 3%; then, at 24 hours they are exposed to saturated steam at 125°C (0.15 MPa pressure) for 4 hours. Based on their preliminary tests Hooton and Rogers25 modified this test method using ASTM C227 bars with the Na2O eq. raised to 4.0%.

[2] Chinese autoclave test32. Mortar bars are cured at 125°C in steam prior to autoclaving in a 10% KOH solution for 6 hours at 150°C. Hooton and Roberts25 had used this test with ASTM 25×25 mm mortar bars and ASTM mix proportions. None of the aggregates expanded to 0.10%, and the test was abandoned. However, they report that the test worked well with Canadian aggregates when the originally proposed proportions of bars and mix were used.

The results of this study, shown in Tables 3.2 and 3.3, led to the following conclusions. It was found that to obtain expansion > 0.10%, the ASTM C227 method needed more than 18 months to indicate potentially deleterious late – expanding alkali-silicate aggregates. Bars sealed in polyethylene bags showed

[3]Microstrains.

Copyright 1992 Blackie and Son Ltd

[4]Additionally, alkali-silicate reaction might be involved.

Copyright 1992 Blackie and Son Ltd

Figure 5.25 Calculation of the ‘reactive alkali content’ of concrete in accordance with the Concrete Society TR30 guidance, one of several equation options given in the report. After

Ref. 13.

[6] Long-time storage tests show that addition of fly ash to concrete mixes has no adverse effect even if its alkali content is as high as 2.34% Na2O eq.

[7] Accelerated tests according to ASTM show that the addition of fly ash, even of high alkali content, to Portland cements reduces expansions due to ASR.

[8] Accelerated tests with unlimited supplies of alkali salt show that the addition of fly ash to Portland cement reduces, at least, the rate of

[9] Kristjansson, Rfkhardur (1979) Steypuskemmdir-Astandskonnun (an outline in Icelandic). The Icelandic Building Research Institute.

2. Gudmundsson, G. and Asgeirsson, H. (1975) Some Investigation on Alkali Aggregate Reaction, Cement and Concrete Research, Vol. 5. New York, pp. 211-220.

3. Gudmundsson, G. (1975) Investigation on Icelandic Pozzolans, Symposium on AAR – Preventive Measures. The Icelandic Building Research Institute.

4. Samundsson, K. (1975) Geological Prospecting for Pozzolanic Materials in Iceland, Symposium on AAR-Preventive Measures. The Icelandic Building Research Institute.

5. Gudmundsson, G. (1971) Alkali Efnabreytingar i Steinsteypu. The Icelandic Building Research Institute.

6. Thaulow, N. (1976) Undersogelse af Beton Borek^rne fra Reykjavik. Aalborg Portland.

7. Asgeirsson, H. (1986) Silica fume in cement and silane for counteracting of alkali – silica reactions in Iceland. Cement Concr. Res. 16, 423-428.

8. Olafsson, H. and Helgason, Th. (1983) Alkalivirkni Steypuefna a Islandi og Ahrif Salts og Possolana a Alkalivirkni i Steinsteypu. The Icelandic Building Research Institute.

9. Kristjansson, R., Olafsson, H., ^ordarson, B., Sveinbjornsson, S. and Gestsson, J. (1979-1987). Field Surveys of Houses. The Icelandic Building Research Institute.

[10] Waterproofing type of coating is not so effective since the piers in Table

10.2 and 10.3 expanded greatly after coating and cracked again.

(2) The effect of polybutadiene cannot be concluded at this stage since the

[11] Effects of shape and dimensions of the test specimen.

[12] Effects of storage conditions.

[13] Effects of the unit cement content and the total alkali content.

[14] Comparisons of the test results obtained from concrete specimens and mortar bars.

[15] Fine-grained, glassy to microcrystalline basalts containing more acidic glassy phases and occurring in the Deccan Plateau, west coast, Maharashtra, Madhya Pradesh, Gujarat, Andhra Pradesh, Jammu and Kashmir, West Bengal and Bihar.

[16]Denotes two generations of quartz.

Qz, quartz; Ir, iron oxide; Bio, biotite; Chl, chlorites; .Fels, feldspar; Aug, augite; Gm, groundmass; Or, orthoclase; P1, plagioclase, M, muscovite; Acc, accessory minerals.

results are shown in Figure 11.15. Such an alkali-dependent expansion of the aggregates would, prima facie, classify them as ‘alkali-reactive’.

For quantifying the effects of ‘strained quartz’, various parameters, namely grain size, proportion of quartz in the modal composition, percentage of quartz showing strain effect and UE angle, were considered. Except for very fine-grained (smaller than 0.1 mm) rocks, average grain size did not exhibit any discernible influence on the resultant expansion. The amount of expansion was more dependent upon the precentage of quartz grains showing strain effect23.

From Table 11.3, it can be seen that, except for aggregates 1 and 4 (Table 11.2), the maximum expansion with 1.00% alkali cement at 38°C was generally within the limit of 0.05% at 3 months that is stipulated for reactive aggregates. The rapid chemical tests also failed to indicate the potential reactivity of these aggregates (Figure 11.14). Yet the dependence of expansion in mortar bars on the level of alkalis in cements would suggest that the aggregates were indeed responsive to the alkali in the system (Figure 11.15). Field experiences with similar aggregates, as reported in 11.3, would confirm the potential reactivity of such aggregates. As such, revised criteria for these types of aggregates containing strained quartz are necessary.

A summary of the recommendations made in relation to reactive aggregates, including those containing strained quartz, is reproduced in Table 11.4. Recognising the relatively slower kinetics of expansion reaction of aggregates containing strained quartz, Buck suggested that either the limits of expansion at 38°C (ASTM C227) be applied at later ages too or that different limits be fixed for the 60°C regime24. Gogte had suggested a lower limit of expansion of 0.05% at 6 months when tested at 50°C2. The Canadian (CSA) specifications CAN3-A23.1 and A23.2-M77 (1986) have also recognised the relatively slow expansion in concrete prism tests with aggregates whose reactivity is due to strained quartz, and have suggested a limit of 0.04% expansion at 38°C after 1 year.

Presently Indian Standard specification IS:383 (1970) does not specify any limit of expansion in mortar bar tests, perhaps because until now little concern has been expressed. In view of recent experiences, a revised criterion is being proposed, for which the reasoning is as follows23. In the current investigations, expansion at 60°C with low-alkali (0.57%) cements seldom exceeds 0.05% at 90 days (Table 11.3). Since the well-accepted remedy for use with reactive aggregates is such low-alkali cements, expansions recorded with such cements can be considered as acceptable. On the other hand, aggregates similar to those used in hydraulic structures which have exhibited distress due to ASR (e. g. 11, 12, 15 of Table 11.2) resulted in expansion exceeding 0.05% at 3 months with

I. 00% alkali cement at 60°C. In many cases, formation of ASR gel was noticed on the broken surfaces of the mortar bars with high-alkali cement (Figure

II. 17). A limit of acceptable expansion of the order of 0.05% at 3 months or 0.06% at 6 months, when tested with higher alkali (about 1.00%) cement at 60°C is, therefore, proposed23.

Attempts have been made to set ‘prescription’-type limits of UE angle or percentage of quartz grains exhibiting strain effect24. ‘Performance’-type specifications of the nature of limiting expansion in mortar-bar tests is to be

preferred. Note that the suggested limits were reached by quartzite aggregates, in which a minimum of 25% quartz grains showed strain effect with UE angle of 25° or above (Table 11.1). The limits in the case of granitic aggregates would be somewhat lower, because of the role of alkali feldspars and micabearing phases.

It is proposed that these limits are incorporated in the national specification (IS:383:1970). The relevant test procedures (IS:2386:Part VII) would stipulate mortar bar tests at 60°C for aggregates containing strained quartz and petrographic examination to quantify strain effect in terms of UE angle would also be recommended in IS:2386:Part VIII.

From the foregoing, it would transpire that most of the reactive aggregates in India are those containing strained quartz as the reactive component. In addition, reactivity of granitic rock aggregates is also partly due to the presence of alkali feldspar, which can undergo alterations as a result of the action of hydrothermic solution and the normal process of weathering12. Laboratory experiments by Van Aardt and Visser had shown that alkali feldspars can release alkali in the presence of calcium hydroxide and water21, in which case they can supplement the alkali derived from cement. While detection of potential alkali reactivity of aggregates containing such secondary silica minerals is quite straightforward, e. g. by rapid chemical test or mortar bar test, detection of reactivity due to the presence of strained quartz poses some problems. In such cases, rapid chemical test may not detect the potential reactivity, and the threshold values for mortar bar expansion may have to be somewhat lowered122. As a result of this, attention is being focused on the effects of strained quartz in evaluating concrete aggregates for use in a number of concrete dams to be constructed in India. Indeed a large volume of data on aggregates from different sources has now become available and is summarised below.

A large number of such aggregate samples were of the quartzite type, while others were granitic rocks containing feldspars and mica-bearing phases in substantial quantities as well as other varieties23. In addition, composite samples containing rocks of more than one type were also involved. A summary of the petrographic details of the aggregates is shown in Table 11.2. The strain effect in quartz grains was measured in terms of UE angle9. Alkali feldspars in the granitic aggregates were found to have altered to clay minerals or sericite. Metastable secondary silica minerals were not detected in any of these aggregate samples. As such, any potential alkali-silica reactivity of these aggregates could be ascribed mainly to the presence of strained quartz.

11.3.2.1 Test procedure. Knowing that the existing standardised test procedures ASTM C227, C289, C586 and IS:2386:Part VII may not detect the potential reactivity of such aggregates, some improvisation in the testing regimes was necessary. Accordingly, the aggregate samples were subjected to rapid chemical test for a total duration of 24 hours, 3 days and 7 days. Similarly, the mortar bar expansion test was carried out at 38°C as per IS:2386:Part VII, as well as at 60°C, as suggested by Buck24. Three samples of ordinary Portland cement containing a total of 1.00, 0.57 and 0.25% alkalis (Na2O eq.) were used in mortar bar tests. The composition of the mortar bars for both 38 and 60°C regimes was identical, which permitted direct comparison of the results. Exploratory tests were also carried out on aggregate samples immersed in KOH solution (1 N) at 60°C and changes observed by SEM and IR spectroscopy25.

11.3.2.2 Test results and discussions. As anticipated, the rapid chemical test ASTM C289 for 24 hours showed most of the aggregate samples to be innocuous. When tested after prolonged storage for 3 and 7 days, the results shifted somewhat to the ‘right’ of the demarcation line in the case of quartzite aggregates (Figure 11.14). Many of the other aggregates continued to be in the ‘innocuous’ zone. It is not surprising that the limiting curve of Figure 11.14, developed in relation to mortar bar tests on aggregates containing secondary silica minerals, should not be valid for aggregates containing strained quartz, which are relatively slowly reactive.

A summary of the mortar bar expansion tests is given in Table 11.3. All the aggregate samples exhibited increased expansion with increasing alkali content in the cements, when tested in either of the temperature regimes. A few

Strain effect

SI Quartz grain UE IJE

no. Type size (mm) (degrees) (%) Modal composition (%)

Copyright 1992 Blackie and Son Ltd

Since the amount of quartz grains was relatively greater in quartzite rocks than in other aggregates, these were considered separately. For both the sets of aggregates, the amount of expansion increased with the UE angle in the quartz grains (Figure 11.16).

After continuous immersion in 1 N KOH solution at 60°C, quartzite aggregates revealed differences in IR spectra in the region above 850 cm-1. A broad and intense peak in the 1600 cm-1 region, due possibly to OH bending mode, was identified. Under SEM, typical gel formation and microcrystalline growth on the aggregates could be seen. However, the trend was not very clear. Fuller details are given in Ref. 25.

The presence of secondary silica minerals had been known to be responsible for the reactivity of common aggregates for 40 years or more, and the first comprehensive assessment of certain Indian aggregates was made from this chemical point of view19. Accordingly, the following common rock types were identified as potentially reactive on the basis of their composition as well as results of the rapid chemical test (ASTM C289) and mortar bar expansion test (ASTM C227): 1

□ CHARNOCKITE

GRANITE

GREYWACKE

LIMESTONE

QUARTZITE

SANDSTONE

T RAP ROCK

Figure 11.13 Types of natural aggregates commonly used in India (schematic map-not to

scale).

(2) Sandstones containing secondary silica minerals such as chalcedony, crypto – to microcrystalline quartz, opal and quartzites having a reactive binding matrix and occurring in Madhya Pradesh, West Bengal, Bihar and Delhi.

(3) Granites and pigmetites containing opal, rhyolites and glasses and occurring in south India, notably Tamil Nadu and Karnataka.

(4) Trap aggregate containing reactive constituents occurring in Jammu and Kashmir and Deccan Plateau.

Since reactivity of aggregates due to the presence of secondary silica minerals was known at the time some of the major concrete dams were built in India in the early 1950s, such aggregates were used with due caution. Bhakhra Dam in Punjab, which was the highest concrete gravity dam in the world when completed, used river gravels composed predominantly of quartzites,
metasandstones, greywackes and sandstones, which contained cherts, chalcedonic sandstones and glassy andesites—2.3% of the total mass on an average-as well as limestone and dolomites. Similarly, the natural sand contained less than 2% cherts. These aggregates were proven to be innocuous after exhaustive laboratory evaluation and were used without any detrimental effect till now20. In the case of Hirakud Dam spillway, the quartzite river shingles containing cherts and chalcedony, which proved to be potentially reactive in laboratory evaluation, were used inadvertently.

Gogte in 1973 postulated that the reactivity of common aggregates could also be due to mineralogical and textural features of the crystal rocks, in which the commonly accepted susceptible forms of silica, such as opal and chalcedony, were absent2. He ascribed the reactivity of such aggregates to the presence of strained quartz. As a modification of the then existing test procedures of the Indian Standards (IS) and ASTM, Gogte recommended that mortar bar tests be carried out at a temperature of 50°C and suggested a criterion of mortar bar expansion above 0.05% in 6 months as indicative of reactivity. Accordingly, Gogte identified a number of granites, charnockites and quartzites and schistose rocks, mostly from Andhra Pradesh, Karnataka and Tamil Nadu in south India, as well as basalts from Maharashtra and Gujarat in western India, as being potentially reactive because of the presence of strained quartz. In most of these rocks, samples showed strongly undulatory, fractured and granulated quartz, nearly 35-40% of quartz grains showing a UE angle between 18 and 20°. On the other hand, rocks which contained less than 20% strained quartz, or in which most of the quartz showed uniform or faint undulatory extinction, were considered ‘innocuous’. Sandstones from Andhra Pradesh, Rajasthan, Himachal Pradesh and Madhya Pradesh owed their reactivity to presence of cherts as a detrital constituent and sometimes as binding matrix. Sandstones devoid of cherts but containing a few grains of strongly undulatory quartz showed expansion in mortar bar tests within tolerable limits2.

In view of the vast size of the country and the wide variations in the geomorphological characteristics of natural rocks from one location to another, any generalisation as to the reactivity of Indian aggregates could be misleading. A broad classification of the common rock types used as natural aggregates for concrete in different parts of India is shown in Figure 11.13. The common rock types so identified have been found to be reactive to varying degrees in different parts of the world, depending upon their modal composition, essentially the presence of secondary silica minerals such as cherts, chalcedony, opal, etc.9

Long-term observations on concrete core samples immersed in KOH solution at different temperatures indicated that the expansion potential of the concretes was not yet exhausted14. The repair techniques, therefore, had to take into account the present state of distress as well as possible aggravation in future. Analyses of stability against overturning as well as sliding failure, assuming cracked sections, indicated the margins of safety to be satisfactory3,4. The remedial measures were accordingly aimed at minimising the rate of reaction, restoring structural integrity and accommodating possible future expansion of concrete.

Cracks at the upstream faces were sealed with epoxy grout, followed by epoxy painting to minimise ingress of water. It was recommended that chemical grouting be carried out in the mass of the structures effectively to fill up the cracks and restore their monolithic behaviour. In the case of the spillway, at locations where cracking was extensive, stitching across the cracks by providing anchor rods to act as shear pins was suggested3. The drainage holes were cleaned to make them function effectively to reduce the uplift pressure.

In the reinforced concrete columns of the powerhouse in which reinforcement had snapped, additional reinforcement was provided and the columns were than jacketed with steel plates. To accommodate future expansion, the fixed joints in the penstock gallery frame were released (Figure 11.9) and the expansion joint at the toe of the dam was made functional. For operation of gates, gantries, cranes, etc., portions of concrete were chipped off to provide easy movement. Extensive instrumentation for periodic monitoring of the structures was recommended34.

The general description of the microstructure of reaction products as observed by SEM has been mentioned already. Since these were the first reported cases of ASR in India, detailed examination of the reaction products was undertaken in order to compare them with the features reported in the literature6715.

A composite gel sample was made in both cases by carefully scooping out the gel from various locations; this was used for the chemical analyses, X-ray diffraction and optical microscopy. The chemical analysis of the gels is presented in Table 11.1. The alkali contents were determined by flame photometry. The compositions were similar to the ranges indicated by others as representative of alkali-silica gel15.

11.2.3.1 Petrography. The composite gel was petrographically examined under a polarising microscope in immersion liquids. In the case of quartzite aggregates containing secondary silica minerals, the material showed the following distinct composition: (i) amorphous gel-type matter of irregular shape with a refractive

index of 1.48-1.50 which compared favourably with 1.455-1.502 reported by Mather15; (ii) distinct small grains of chert or chalcedony with a refractive index of 1.50-1.52 (Figure 11.10); (iii) crystalline material with patches of opaque mineral; and (iv) crystals with slight anisotropy and no birefringence, presumably of the crystalline white deposits with a refractive index of 1.42-1.4816. The white material also showed occasional grains of aragonite and calcite with a refractive index of 1.65-1.66. In the case of granite aggregates containing strained quartz, similar features were noted, except for the presence of secondary silica such as chert or chalcedony.

11.2.3.2 X-ray diffraction analysis. Typical X-ray diffractograms of the alkali – silica gel obtained in both cases are given in Figure 11.11. The nature of the gel was predominantly amorphous, and in addition to the typical cement hydration products, some new crystalline products, believed to be due to ASR, were identified16. In the case of quartzite pebbles, the peaks at 2? degrees = 5.5, 16.7 and 25.7 (Cu-Ka) were assigned to a crystalline alkali-silicate hydrate of composition NaSi17O13(OH)3’3H2O, and other prominent peaks at 2? degrees = 31, 29.8 and 29.4 to a composition K2Ca(SO4)2- H2O. In contrast, in the case of granitic aggregates, the prominent peaks at 2? degrees = 6.9, 13.6, 30.5 and

53.3 were ascribed to a composition of crystalline potassium-sodium-calcium – silicate hydrate (K2Na2Ca)16Si32O80- 2H2O. These compositions are different from these reported earlier17.

11.2.3.3 Infra-red spectroscopy (IR). The use of IR to study ASR has been reported before, when certain absorption bands in the 650-1600 wave number region were taken as characteristic of silica gel and calcium carbonates18. Composite samples of the reaction products in both the cases discussed in this chapter were studied by recording the IR spectra in the range 4000-200/cm, and compared with a synthetic silica gel. The samples were prepared by grinding and passing through a 45-p m sieve before drying in an oven at about 110°C for a few hours. The fine powder was made into a pellet with KBr.

The results are presented in Figure 11.12. In the case of the reaction products (alkali-silica gel) obtained in the two cases, the broad band observed in the region 3000-3600/cm is due to the O-H stretching vibrations. The

bands at 1400 and 1140/cm are assigned to carbonates and feldspars respectively. While the band at 1100/cm is due to monosulphates, the bands at 680 and 590/cm are assigned to S04 group. The band at 1000/cm indicates the presence of hydrated calcium silicates. Lastly, the peaks at 1030, 960, 860, 760, 460 and 440/cm are assigned to different modes of SiO4 vibrations.

Similarly, the IR spectrum from the synthetic silica gel contains a broad band in the range 3000-3600/cm. Most of the bands characteristic of different modes of Si04 vibrations, such as 1230, 1150, 1050, 940, 790 and 450/cm are observed. The occurrence of a ‘shoulder’ at 590/cm, which is assigned to SO| , is also observed. The absence of bands at 1400, 1100, 1000 and 680 described earlier is quite understandable as the alkali-silica gel was extracted from the concrete cores. In all other respects, the IR spectra of the two materials were similar.

In summary, the microstructure of the ASR products was predominantly of the amorphous gel type, with occasional crystals. In the case of metastable silica minerals, distinct reaction products seemed to be formed, whereas in the case of strained quartz the result of ASR was alteration in and of the aggregates16.

The concrete gravity dam and adjacent powerhouse of this hydroelectric project, 25 years after their construction, showed extensive distress, which was attributed to ASR4. External manifestations included cracking of concrete, misalignment of machinery and difficulties in the operation of gates, cranes and passenger lifts as a result of movements in concrete. In the powerhouse, the rotor assembly had sunk in relation to the stator, leading to fouling of rotor blades, and high spill current resulting in frequency tripping of the machines was reported. The rotor runner assembly had risen in relation to the speed ring. The horizontal labyrinth clearance at both top and bottom was progressively reduced in a longitudinal direction and increased in the transverse direction. There was horizontal displacement of about 3 cm between the powerhouse crane girders in different bays, and intake gates did not seal properly. The powerhouse seemed to have tilted upstream4.

Examination of the concrete samples with a hand magnifying glass and visual examination on the broken surface of the concrete revealed typical white deposits associated with ASR in the voids in concrete and on aggregates, but they were not frequent. On the other hand, reaction rims around aggregates, another manifestation of ASR10, were quite conspicuous. In some cases, the broken surface of the concrete showed the formation of a thin white rim around the aggregates. In most cases, however, a uniform dark band in the peripheral zone of the aggregate was observed11.

A complete scan of the sample morphology with SEM showed that the aggregates contained a reaction rim altering their borders, sometimes with microcracks either in the aggregate or in the mortar phase, similar to Figures

11.2 and 11.311. The presence of a fluffy gel-type formation with occasional

crystalline deposits was observed, in which potassium was predominant. The fact that the needle-like crystal structures were not ettringite was verified with the help of EDAX, which showed the absence of sulphur. The aggregates were observed to form white fluffy gels, giving an impression that these were oozing out from the main aggregate12. The aggregate sample itself contained alkalis, originating from alkali feldspars. However, EDAX analysis of the reaction products showed a much larger amount of potassium.

11.2.2.1 Aggregate types. Petrographic examination of aggregates extracted from concrete samples indicated these to be mainly biotite granite, muscovite granite and mica granite12. In each of these rocks, the quartz content varied from 32 to 45% and alkali (sodium-potassium or sodium-calcium) feldspars such as orthoclase, microcline and plagioclase from 35 to 45%; varying amounts of biotite, muscovite and other accessories, including iron ore, chlorite and apatite, were present. The average grain size of quartz varied from 0.10 to 0.20 mm, with small grains up to 0.03 mm in some cases and large grains up to 0.45 mm in others. Nearly 50-80% of the quartz exhibited strain effect with a UE angle varying from 25 to 30° (Figure 11.6).

Potash feldspar in most of the cases was orthoclase which was found to have altered to sericite. Plagioclase feldspars were found to have altered to clay minerals (Figure 11.7). Biotite occurred in the form of laths and sometimes as sheath-like structures. These and muscovite showed bending effect. The normal granitic texture of the aggregate was considerably disrupted because of the high degree of alteration of minerals. Laboratory

evaluation according to the criteria existing at the commencement of the project had indicated the aggregates to be ‘innocuous’ and there was felt to be no need to obtain low-alkali cement or even regularly to monitor the alkali content of the cement used4.

11.2.2.2 Structural interaction. The penstock gallery structure located on the toe of the concrete gravity dam comprised six blocks, corresponding to the six overflow bays. Each block had four reinforced concrete frames constructed integrally with the dam body interconnected with reinforced concrete beams running parallel to the axis of the dam (Figure 11.8). The major structural distress noticed in the penstock gallery and in the adjacent powerhouse structure after nearly 25 years of operation included multiple horizontal cracks on the face of columns marked 9-10-11 in Figure 11.8 (close to penstock) and extending to a height of 1 m or so from the foot of the columns, wide horizontal cracks near mid-height and snapping of main longitudinal reinforcement in column 9-10-11, spalling of concrete at the end section of beam 5-6, closing of 25-mm expansion joints between gallery and scroll concrete, and relative horizontal shifts in the beams supporting the generator floor13.

For such distress in concrete structures, temperature effects, deleterious chemical reactions and relative settlement of foundation merit prima facie consideration as probable causes. From the records, it could be ascertained that the concrete was precooled and the temperatures recorded were within the limits4. This, along with the fact that the movements were still continuing,

eliminated temperature effects as a cause. The dam was founded on granite rock which was considered as nearly ideal. Finite element analysis of the dam section, including a portion of the foundation rock, and allowing differential modulus of elasticity of the rock to result in differential settlement revealed that this could not have caused the distress noticed in the frames13.

Further analysis was carried out by imposing various levels of horizontal and vertical displacements of nodes 1, 4, 7 and 8 (Figure 11.8) representing the expansion in the dam body due to ASR. Because the deformation characteristics of all the frames were identical and all the frames had the same configuration and mechanical properties, a two-dimensional analysis was considered adequate. A systematic search was made to correlate the distress observed with the possible combination of vertical and horizontal movements transmitted to the joints of the frames due to volume change in the main dam.

From the analysis, it was concluded that the relative displacements due to ASR expansion could cause distress in the members of the penstock gallery frame, as observed13. In addition, the reaction in the turbine block, which could be either passive or active, could aggravate the distress in column 9-10-11. Horizontal and vertical displacements of 12.5 mm and 3 mm respectively, imposed relative to node 13 in Figure 11.8, produced the bending moment diagram shown in Figure 11.9. Visual observations of the shifts in the beams,

closing of 25-mm expansion joints and further long-term observations14 established that displacements of such magnitude were entirely feasible.

The first investigation relates to a concrete spillway in one of the longest earth dams in the world. The structure at the time of investigation was nearly 27 years old. It had suffered extensive cracking, mostly in the walls of openings like galleries, shafts and adits. Typical ‘map’ cracking was superimposed with longitudinal horizontal cracks. The extent of cracking had been increasing with time. In addition, malfunctioning of radial crest gates, snapping of bolts which fix the sluice gate roller tracks and guide-rails to the concrete, and deflection of side walls of the adit gallery were noticed3.

Samples of concrete from such locations where typical signs of ASR were present showed the unmistakeable presence of ASR, as evidenced by off-white, translucent to opaque agglomeration of fluffy gel-type deposits in voids bordering the aggregates, or on the aggregates; the aggregates had a dark reaction rim around their edges, visible to the naked eye (Figure 11.1).

Scanning electron microscopy (SEM) of representative samples obtained from concrete cores showed the aggregate boundaries to contain a reaction rim altering their edges, sometimes with microfractures in the aggregate pieces, with a considerable amount of white reaction products (Figure 11.2). In many instances, such cracks in the aggregates were apparently caused by the formation of the reaction products inside the aggregates5.

The reaction products thus formed were found to be essentially non­crystalline, gel-type (Figure 11.3). An EDAX point scan on the aggregate and the surrounding rim showed the presence of a considerable amount of silicon, less calcium and small amounts of potassium (Figure 11.4). The presence of sodium could not be detected by EDAX. In view of the gel-type nature of the reaction products and the elemental composition, these can be described as lime-alkali-silica gel, identified and described by Regourd6 as well as Thaulow and Knudsen7. In addition, the reaction products were occasionally found to be crystalline in nature (Figure 11.5) and their composition as detected by

ED AX was similar to that in Figure 11.4. Away from the reaction zone, the mortar phase was generally found to be cracked and gel formation was associated with such cracking5.

11.2.1.1 Aggregate types. On petrographic examination the following three types of reactive coarse aggregates were identified8:

(a) Quartzite river shingles, consisting predominantly of crystalline quartz (P form) with grains of varying dimensions, cemented by either crystalline silica or ferruginous matter. In a number of cases, the quartz grains were cemented by near-opaque or semi-opaque to translucent cryptocrystalline silica. The refractive index of around 1.53 indicated the material to be chert or chalcedony9. Quartz grains very often showed wavy (undulatory) extinction; more than 20% of the grains showed wavy extinction, the undulatory

extinction (UE) angle ranging from 19° to 28°, determined by the procedure suggested by Dolar-Mantuani9. Occasionally the river shingle consisted of dark fine-grained plagioclase, horn-biotite and hornblende which showed alteration to chlorite.

(b) Granitic rocks. These comprised granite, granodiorite or granite porphyry. The rock was porphyroblastic in texture and showed evidence of action of direct pressure, manifested by wavy or undulatory extinction of quartz. Large subidioblastic to xenoblastic plates of feldspar constituted both orthoclase and perthitic microcline as well as plagioclase, in which the polysynthetic twin lamellae were often deformed. The mineral analysis showed 60% alkali feldspar, 10% plagioclase feldspar, 22-25% quartz, 5-7% hornblende and biotite and 3-5% accessory minerals. Fracturing and wavy extinction were common in quartz grains. The extinction of individual grains ranged from 12° to 25°. In the patchy (mottled) variety the orthoclase, feldspar, occurred as large plates or laths (up to 8 mm), and this rock can be termed granite porphyry. A third type consisted of a smaller amount of quartz (15%), predominantly feldspar, and a comparatively large amount of biotite and hornblende. This type is termed granodiorite. The gradation between the three types was slight, such that they should not be considered separately. The rocks showed common and uniform alteration to chlorite (from biotite and hornblende) and sericite (from feldspar).

(c) Diorites. The rock showed a predominance of plagioclase and perthitic microcline. Biotite, hornblende and quartz occurred in varying amounts. The accessories were sphene, epidote and titaniferous magnetite. A modal analysis of typical rock showed 20-30% plagioclase, 11-14% hornblende, 6-12% biotite, 10-18% perthitic or pure microcline/orthoclase and 6-10% quartz. The quartz grains very often showed wavy extinction and cracking, the feldspars showed bending of twin lamellae and cracks along cleavage planes. More than 25% of quartz grains showed wavy extinction with a UE angle 15­30°. The rock showed conspicuous secondary alteration, which was manifested by conversion of hornblende to biotite and chlorite, feldspars to sericite and of perthitic feldspar to kaolin and other clay minerals8.

Laboratory evaluation had earlier revealed the river shingles to be potentially reactive, and these were inadvertently used in some locations during the peak construction period. However, the other two types comprising crushed aggregates were considered to be ‘innocuous’ according to the criterion prevalent in the 1950s. As a result, no effort was made to obtain ‘low-alkali’ cement, and cement used from two sources probably contained 0.8-1.0 total alkalis (Na2O eq.).

Some of the earliest references to the occurrence of ASR in concrete structures in India were made in 19622; however, few details were documented. The first reported cases of distress in a concrete spillway and a concrete gravity dam and the powerhouse structure can be found in Refs. 3 and 4 along with details of comprehensive investigations carried out, which are summarised below.