- Email: [email protected]
Alkali-aggregate reactivity in Canada
Cement & Concrete Composites 15 (1993) 13-19
Alkali-Aggregate Reactivity in Canada C. A. Rogers Ministry of Transportation, 1201 Wilson Avenue, Central Building, Downsview, Toronto, Ontario, Canada M3M 1J8
Abstract In Canada, three types of alkali-aggregate reaction in Portland cement concrete are recognized. Each type is evaluated using different tests. Corrective measures such as the use of low-alkali cement, lower cement contents, or pozzolans are seldom used with reactive aggregates. Beneficiation or selective extraction is used with some reactive aggregates. Work is being conducted on multilaboratory study of existing tests and new, rapid tests. Keywords: Concrete, aggregate, alkali reaction, Canada, test, cement, durability.
The wide range in rock types (Table 1) capable of causing damage as a result of A A R gives some idea of the difficulty of selecting appropriate testing procedures and corrective measures. For instance, the mortar bar expansion test used for alkali-silica reactive rocks gives only small expan-
I
~
. -
O U EBEC
~,~.~ ~ '
o
073 089 o
o
098
~ o
, .
! t
INTRODUCTION The oldest Canadian concrete structure known to have been affected by alkali-aggregate reactivity (AAR) was the Hurdman Bridge built in Ottawa in 1906 and demolished in 1987. It was not until the 1950s, however, that A A R was first documented in Canada. ~ Over the past 30 years, cases of A A R have been found in many parts of Canada (Figs 1 and 2) with a variety of different rock types (Table 1). There are now over 160 reports and papers documenting Canadian work on AAR. 2 In eastern Canada there are probably over 1000 structures affected by A A R . For many years, Canadian engineers had been aware of the cracking typically caused by A A R but generally had not connected the cracking with internal expansive reactions. Much of the cracking was attributed to frost damage, particularly in older structures built before the almost universal use of air entrainment. Nearly all Canadian concrete affected by A A R has also been exposed to freezing and thawing and it is sometimes difficult, in forensic studies, to judge the principal agent of damage.
Fig. 1. Locationof general areas of known cases of AAR, and alkalicontents of Type 10 cements in Eastern Canada.
•
ALKALI C(~iTENT$ OF TYPE tO CEMENTS
]
0.85
Fig. 2.
Location of general areas of known cases of A A R ,
and alkalicontents of Type 10 cements in WesternCanada.
13 Cement & Concrete Composites 0958-9465/93/$6.00 © 1993 Elsevier Science Publishers Ltd, England. Printed in Great Britain
14
C A. Rogers
Table 1. Types of alkali-reactive rock found in Canada
Phanerozoic Andesites and associated volcanic rocks in British Columbia Quartzite in Rocky Mountains of British Columbia and Alberta Opaline shale/siliceous shale in Saskatchewan and Manitoba Upper Silurian subgreywacke on Ellesmere Island Devonian and Silurian chalcedony-bearing chert in glacial drift in northern Ontario and in quarries and glacial drift in southern Ontario Middle Ordovician siliceous and sfighfly sificeous limestone in southern and eastern Ontario and the St Lawrence Lowlands of Quebec Middle Ordovician dolomitie limestone and calcitic dolostone in Hudson Bay Lowlands and southern Ontario Cambrian sandstones and orthoquartzites in eastern Ontario and southwestern Quebec Rhyolitic tuff, southeast of Quebec City Phyllite in southeastern Quebec Phylhtes, greywackes, siltstones, argillites, schists, silieified tuff, devitrified glass, and quartzites of New Brunswick, Nova Scotia, and Newfoundland
289) is also used, but gives misleading results when carbonates are present. (3) Late/slow alkali-silica/silicate reaction with sandstones and granites containing strained quartz, and metamorphosed sediments such as phyllite, ar~llite and greywacke, where exfoliation of phyllosilicates causes swelling of aggregate particles. Rocks susceptible to this type of reaction usually perform well in conventional alkali-aggregate tests. A concrete prism expansion test at 38"C and high cement and/or high alkali contents is the most useful procedure. Rock cylinder or miniature rock prism expansion tests are also useful. The procedures necessary to determine the type or types of reaction to which an aggregate may be susceptible are not simple. Flow charts used for determining the type of reaction and the necessary tests are described in the current Canadian Standard for Concrete?
Rhyolite in Nova Scotia
Precambrian Granite and granite gneiss with strained quartz Sandstone, quartzwacke, arkose, greywacke, and arglUite of the Huronian Supergroup in central Ontario and western Quebec Chert in banded iron formations (taconite)
sions with expansive dolomitic limestone susceptible to the alkali-carbonate reaction (ACR) or greywackes and argillites susceptible to the slower alkali-silicate reaction. For convenience, alkalisilicate reactions in Canada have been grouped into three types, each having different tests or criteria used in evaluation: (1) Alkali-carbonate reaction with dolomitic limestones and calcitic dolostones of Ordovician age. Chemical analysis for CaO :MgO and A1203, 3 the rock cylinder expansion test (ASTM C 586), and the concrete prism expansion test (CAN3-A23.214A) are useful for identifying these rocks. (2) Alkali-silica reaction with chalcedony, opal, cristobalite, glass, etc. is usually evaluated by the mortar bar expansion test (ASTM C 227); concrete prism expansion tests at 38"C with high cement contents, and rock cylinder expansion tests, are also used. The quick chemical test (ASTM C
CANADIAN CEMENTS Figures 1 and 2 show that the majority of areas where A A R occurs are in central and eastern Canada, where low-alkali cements are seldom used. West of Ontario, most of the normal Portland cements (CSA Type 10) have alkali levels of less than 0.8% Na20 equivalent) and, as a result, there are very few documented cases of A A R : In central and eastern Ontario, the majority of normal cements have alkali levels of more than 0.9%. For the 19 normal cements made g,a Canada in 1988, the Na20/K20 ratio ranged from 1.09 to 0.07, with a mean of 0.38 (median 0.31). In a 1977 survey of the alkali contents of Canadian Type 10 cements, six of 23 sources had alkali contents of more than 1"0%. 6 In 1988, only two of 19 producers had mean alkalis of more than 1.0% (Table 2). This reduction in alkali content can generally be attributed to the closing of obsolete plants that used high-alkali raw materials. At present, two normal Portland cements are made in Canada which contain a pozzolanic mineral admixture (Type 10P with about 6 and 8% silica fume). The remaining normal Portland cements usually contain up to 5% limestone interground with the clinker. Six plants (three in eastern and three in western Canada) which norreally make high-alkali cements also make a lower-alkali cement, usually without 5% limestone
Table
2. Alkali contents of Canadian portland cements (latest monthly or yearly average, January 1989)
Cement co.
Location
Capacity (tonnes x 10-6)
Cement type
At~ti (%) Na:O
K zO
Eqmv. NazO
Tilbury Cement
Delta, British Columbia
1-25
10
0.28
0.29
0.47
Lafarge Canada
Richmond, British Columbia
0"33
10 3 50
0"35 0"30 0"30
0.32 0-20 0-32
0.55 0.49 0.50
Lafarge Canada
Kamloops, British Columbia
0-19
10 15
0-20 0"20
0.44 0"50
0-49 0.53
Lafarge Canada
Exshaw, Alberta
1.10
10 3 50
0"15 0"15 0-15
0"79 0"80 " 0"68
0.67 0.68 0'60
Inland Cement
Edmonton, Alberta
0-73
10 50
0"22 0"24 0-20
0-84 0"76 0'47
0.77 0.74 0'51
0-20
10 50
0-32 0-35 No data
0-80 0"63
0"85 0.76
0"33
10 50 1
0-30 0-32 0.27
0"62 0-42 0-40
0.71 0.59 0-53
Low alkali to special order Inland Cement
Regina, Saskatchewan Low alkali to special order
Inland Cement
Winnipeg, Manitoba Low alkali to special order
St Marys
St Marys, Ontario
0-70
10 50
0.48 0-42
0'99 0-89
1.13 1.01
Lafarge Canada
Woodstock, Ontario
0.50
10/20 50 2 3
0.12 0-10 0-12 0-09
0-47 0.43 0-46 0-53
0-43 0.38 0.42 0-44
St Lawrence
Mississauga, Ontario
1-60
10 1 30 2
0"21 1-18 As above approx. As above approx. N o data
0-99
0.99 0-60
Type 2 to NY DOT specifications St Marys
Bowmanville, Ontario Type 2 to NY DOT specifications
0.60
10 2
0-27 0.24
Lafarge Canada
Bath, Ontano
1-00
10 30
0.27 0-92 As above approx.
0.88
Lake Ontario
Picton, Ontario
1.3
10 30 2 5
0.23 0-20 0"16 0.14
1.15 1"15 0.72 0-76
0.98 0.94 0-63 0.65
--
10
0-29
0.45
0'59
Type 2 to NY DOT specifications
1.10 0-55
Miron
Montreal, Quebec (supply imported cements from Spain and Greece)
10
0.13
0-62
0.54
St Lawrence
Jolliette, Quebec
1.10
10 1 20 50
0.37 0.37 0.27 0"31
0.92 0.94 0.88 0-91
0.98 0"99 0"85 0.91
St Lawrence
Beauport, Quebec
0.75
10 30 2
0.30 0-65 As above approx. 0.35 0.40
0.61
Ciment Quebec
St Basile, Quebec
0-55
10 20 30
0-19 0-19 0"18
1.07 1.02 1-11
0-89 0-86 0.91
LafargcCanada
St Constant, Quebec
0.94
10 20 3 10P
0-28 0-31 0-37 0-20
0.91 0.93 0-95 0.85
0.87 0.92 0.94 0.76
I0 50 3 10
0.09 0"08 0.04 0"35
1.24 0-41 1'20 1.10
0.90 0"35 0.83 1-07
10P contains 7.5% silica fume Lafarge Canada
Brookfield, Nova Scotia
0.45
North Star Cement
Comerbook, Newfoundland
0.16
0-73
16
C A. Rogers
addition, to meet US requirements. Many US highway authorities adjacent to the Canadian border specify either a low-alkali cement ( < 0-6%) or a reduced-alkali cement (<0.7%). These lower-alkali cements are usually reserved for the export market to the USA, and are not normally sold in Canada. The cement companies regard the universal adoption of the specification of low-alkali cement to control A A R with concern. The older, wetprocess plants can often make a low- or reducedalkali cement by not recycling alkali-rich dust from the precipitators. The newer, dry-process plants normally can achieve a low-alkali cement only by selecting low-alkali raw material sources. In either case, the cost of production is usually increased. There is a natural apprehension that purchasers will begin to specify the universal use of low-alkali cement to reduce the likelihood of AAR. However, with increased concern about long-term durability there appear to be renewed efforts by the cement producers to find ways of making low- or reduced-alkali cements.
CANADIAN RESEARCH In an effort to prevent the universal adoption of restrictions on maximum alkali levels, the Canadian Standards Association (CSA)(A5) Committee on Cement formed a sub-committee in 1980 to look for and recommend the adoption of rapid test procedures to evaluate concrete aggregates for AAR. At present, many of the current procedures require several months or years of testing, with the result that the structure is usually under construction or even completed by the time the aggregate testing is finished. The sub-committee has a membership of about 40 people. Meetings are usually held twice yearly, and one of these meetings is combined with a field trip to review interesting cases of AAR. A secondary purpose of the sub-committee is to provide a means of communication for those people in Canada involved in work on AAR. A number of authorities are sponsoring research into AAR. The Canadian government, through CANMET, has funded investigations of the incidence of A A R in many of the Provinces, and its control using Canadian pozzolans. This work is hampered to some extent by the lack of reliable, rapid test procedures. The Canadian government also funds work by university
researchers through operating grants, and finances individual projects. 7 Many Provinces also support A A R research; this work is either performed or funded by provincial authorities such as electrical utilities or transportation departments. Over the past 5 years, the total amount of money spent in Canada on investigations and testing of A A R probably has not exceeded an average of about S1 000000 yearly. The amount currently spent on research is small compared with the direct and indirect costs of damage caused by AAR.
EVALUATION OF AGGREGATES The first step recommended by CSA4 in evaluating the suitability of concrete aggregates is a thorough petrographic examination and investigation of prior field performance. This will determine what tests, if any, are needed. Failure to classify the aggregate properly or to recognize minor deleterious components will upset the remainder of the testing program. Petrographic examination should, if possible, be based on knowledge of the prior field performance of similar aggregate. A simple classification of the rock types and minerals present is rarely satisfactory. In Canada, the acquisition of field performance is often impossible because of construction in remote and undeveloped areas. Many of the tests used in Canada have been found to suffer from such problems as lack of correlation with field performance or poor multilaboratory precision. The concrete prism expansion test has been subjected to a multi-laboratory study to find less variable storage conditions than the conventional humid curing room. s The study found that the multi-laboratory coefficient of variation was 23%. Although different storage conditions resulted in significantly different expansion, the multi-laboratory variation was not substantially reduced. The mortar bar expansion test has been found to be susceptible to alkali leaching caused by the presence of wicks in the storage container specified by the American Society for Testing and Materials (ASTM); 9 this resulted in substantial reductions in expansion of alkali-silica reactive mortar. Alkali-carbonate expansive concrete in outdoor exposure has been found to expand about twice as much as concrete prisms stored in the laboratory under moist conditions at 23"C after 5 years. ~° Further studies are
Alkali-aggregate reactivity in Canada
under way to investigate the relationship between field and laboratory expansion of alkali-silica reactive concrete. The CSA subcommittee has established a draft procedure for the accelerated mortar bar test (1M NaOH at 80"C for 14 days) of Oberholster and Davies. ~ This test appears reasonably precise and capable of recognizing the majority of reactive aggregates found in Canada. A multilaboratory study has been conducted to finalize some details and investigate precision. ~2 It was found that the multi-laboratory coefficient of variation for experienced laboratories was about 10%, except with very small expansions. It is likely that only aggregate giving more than 0-15% expansion at 14 days (in solution) will be classed as potentially reactive. Recent work proposes the use of additional limits of 0"33% at 28 days and 0.48% at 56 days. ~3If the 14-day limit is exceeded then the material may still be satisfactory if the later limits are met. The reason for this approach is the satisfactory field performance of slightly cherty sands that caused continued expansion of mortars in this test. Aggregates giving expansions exceeding these limits can be classed as reactive or potentially reactive, and require further investigation and the possible use of corrective measures. The quick chemical test (ASTM C289) is generally unsatisfactory with alkali-silica reactive carbonate aggregates because much of the alkalisoluble silica released is re-precipitated as calcium silicate hydrates. Researchers in Quebec t4.~5 have had success with modifying the test so that it is done on the insoluble residue of quarried carbonate rocks. This modification of the test has been investigated in a multi-laboratory study, which found it to be rather variable, depending on the aggregate. ~6The variability and complex nature of the test prevent its adoption as a standard procedure. A rapid procedure of expansion testing of concrete cores proposed by Scott and Duggan 17 has been studied by researchers at the Universities of Calgary and Manitoba to determine the mechanism of expansion. This procedure has similarities to the Conrow test (ASTM C342), and at present does not appear to be satisfactory for separating reactive from non-reactive aggregate because A A R is not a major cause of expansion in the test. ts To assist in the calibration of existing and new procedures, the Ontario Ministry of Transportation has established three 110-tonne stockpiles of
17
known reactive aggregates. Aggregates from these stockpiles are available free of charge for calibrating test procedures or research into AAR. CORRECTIVE IVlFASURES The use of low-alkali cement to inhibit A A R is not a universally satisfactory solution because of its lack of widespread availability in eastern Canada. The specification of low-alkali cement is unusual in eastern Canada. In northern Ontario, the Ministry of Transportation used low-alkali cement on four bridges in 1970, but has seldom specified it since. A notable exception was the use in 1990 of a lower-alkali cement (<0.7%) for concrete paving with a slightly reactive aggregate near Peterborough, Ontario. Recently, however, Canadian National Railways has started to specify low-alkali cement for all concrete, irrespective of the nature of the aggregate. The reason for this was the costly and tedious deterioration of prestressed concrete railroad ties as a result of A A R . t9 In Alberta, an engineer recently proposed the use of a maximum cement alkali content of 0.3% for dam construction. Such a cement is not available in Canada, but such caution is understandable, given the high cost of remedial work on large concrete structures affected by AAR. Reducing cement contents and, hence, alkali levels is often not practical because of the severe Canadian climate, which generally requires low water/cement (w/c) ratio, air-entrained concrete for all exposed work. In practical terms, minimum cement contents of 325-450 kg/m 3 are required if durability is to be ensured. The use of blast furnace slag cement has generally not been adopted to control AAR, despite its apparent efficiency at high levels of substitution ( > 50%). A notable exception was the use of 50% ground granulated blast furnace slag cement on a recent hydro-electric development in northern Ontario. 2° The aggregate in this case was a rectavolcanic-rich gravel. The metavolcanic material was found to contain a high proportion of microcrystalline quartz, which caused accelerated mortar bar test expansions of about 0.2% at 14 days. It was judged to be marginally reactive and worthy of taking corrective measures. In the highway environment, where de-icer salts are used, concrete made with significant amounts of blast furnace slag cement, and other supplementary cementing materials, has been found to be more
18
C A. Rogers
sensitive to freeze-thaw damage. This has also been confirmed in laboratory salt-scaling tests (ASTM C672) which have shown that, at 50% substitution levels, the scaling loss from the surface may become unacceptable. For this reason, the Ontario Ministry of Transportation permits a maximum of only 25% blast furnace slag substitution. The harsh Canadian winter climate causes practical problems in the use of supplementary cementing materials to control AAR. Other pozzolans such as fly ash and silica fume are used in Canada, but rarely for the purpose of controlling AAR. A notable exception was the successful use of 20% fly ash by Ontario Hydro on the Lower Notch Dam to prevent an alkalisilica reaction with an ar~llite. 2~ Pozzolans are more commonly used for their other advantages such as reduced cost, increased strength-producing properties, or reduced heat of hydration. Condensed silica fume certainly seems effective in slowing or preventing expansion caused by alkali-silica reaction in the laboratory. 22 Longterm outdoor performance studies have to be conducted to see if this initial promise is confirmed in field exposure. The most commonly used control measure is the beneficiation or selective quarrying of aggregate. A technique used in areas of Ontario that contain significant amounts of chert in glacial drift is selectively to crush oversize gravel ( > 40 mm), which normally contains much less chert. Another technique used to remove shale and chert is the use of heavy media separation. There are currently two operations in Ontario. Jigs have also been used in the past to remove shale and chert from gravel, but with limited success. 23 In carbonate bedrock quarries, reserving a specific level or bench of superior quality is a common practice. The underlying or overlying beds may be alkalireactive, but careful, conscientious extraction and stockpiling can ensure an adequate supply of nonreactive rock. The testing of aggregate suitability is complex and time consuming. There is seldom sufficient time during construction to do more than simple tests such as gradation, petrographic examination, and organic content determination. In Ontario, the Ministry of Transportation has compiled lists of acceptable concrete aggregates. All sources shown in these lists have been extensively tested or have a satisfactory record of past performance in concrete. On all contracts with concrete items, a 'Concrete Aggregate Sources List' is issued. The
concrete aggregates used by the contractor must be from the approved sources shown on the list. THE NEXT 10 YEARS New cases of A A R will be discovered in Canada. A group of reliable, rapid test procedures will become available for each type of reaction. Each of these tests will have known multi-laboratory precision, and will be calibrated with known reactive aggregates. Corrective measures will generally only be used where non-reactive aggregates are more expensive. There will be more frequent specification of low- or reduced-alkali cements as more people become aware of the potential problem. There will also be a more widespread adoption of lists of prequalified, non-reactive aggregate. A document will be produced to provide guidance to engineers and owners on recognition of A A R and the management of structures affected by these reactions. ACKNOWLEDGEMENTS The author is indebted to all his Canadian colleagues. Without their observations, research, and assistance, this paper could not have been written. REFERENCES 1. Swenson, E. G., Cement-aggregate reaction in concrete of a Canadian bridge. ASTM Proc., 57 (1957) 1043-56. 2. Rogers, C. A. & Worton, S., Alkali-Aggregate Reactions in Canada -- A Bibliography, 3rd edn. Ontario Ministry of Transportation, Engineering Materials Office, Rep. EM-83, 1988, 26 pp. 3. Rogers, C. A., Evaluation of the potenaai for expansion and cracking of concrete caused by the alkali-carbonate reaction. Cement, Concrete Aggregates, CCAGEP, $ (1)
(1986) 13-23. 4. Canadian Standards Association, Appendix B -Alkali-aggregate reaction. Standards CAN3-A23.1M90 and CAN3-A23.2-M90, Canadian Standards Association, Rexdale, Ont., 1990, pp189-100. 5. Morgan, D. R., Examples of alkali-aggregate reactivity in concrete structures in western Canada. Canadian developments m testing concrete ~ggregates for alkali-aggregate reactivity. Ministry of lransportauon, Ontario, Engineering Materials Rep. 92, 1991, pp. 35-49. 6. Grattan-BeUew, P. E., Sereda, E J. & Dolax-Mantuam, L., The aggregate shortage and high alkali cement in a changing energy situation. Can. J. Civil Eng., 5 (1978) 250-61. 7. Chert, H. & Grattan-Bellew, P. E., Effect of cement composition on expansion of mortar bars due to alkali-silica reaction. Proc. 8th Int. Conf. AIkali-Agsregate Reacnon in Concrete, Kyoto, 1989, Society of Materials Science of Japan, Kyoto, pp. 635-40.
Alkali-aggregate reactivity in Canada 8. Rogers, C. A., Interlaboratory study of the concrete prism expansion test for the alkali-carbonate reaction. Canadian developments in testing concrete aggregates for alkali-aggregate reactivity. Ministry of Transportation, Ontario, Engineering Materials Rep. 92, 1991, pp. 136-49. 9. Rogers, C. A. & Hooton, IL D., Reduction in mortar and concrete expansion with reactive aggregates due to alkali leaching. Cement, Concrete Aggregates, CCAGDP, 13 (1)(1991)42-9. 10. Williams, D. A. & Rogers, C. A., Field trip guide to alkali-carbonate reacnons in Kingston, Ontario. Ministry of Transportation, Ontario, Engineering Materials Office, Information Rep. 145, 1991, 26 pp. 11. Oberholster, 1L E. & Davies, G., An accelerated method for testing the potential alkali-reactivity of siliceous aggregates. Cement Concrete Res., 16 (1986) 181-9. 12. Hooton, R. D., Interlaboratory study of the NBRI rapid test method and CSA standardization status. Canadian developments in testing concrete aggregates for alkali-aggregate reactivity. Ministry of Transportation, Ontario, Engineering Materials Rep. 92, 1991, pp. 225-40. 13. Hooton, R. D., New aggregate alkali-reactivity test methods. Ministry of Transportation, Ontario, Research and Development Branch Pep. MAT 91-14, 1991, 83 pp. 14. Berard, J. & Roux, R., La viabilit6 des brtons du Qudbec: le r61e des granulats. Can. J. Ctvil Eng., 13 (1) (1986) 12-24. 15. Fournier, B., Brrubr, M. A. & Vezina, D., Investigation of the alkali-reactivity potential of limestone aggregates from the Quebec City Area (Canada). Proc. 7th Int. Conf. Alkali-A~,regate Reaction in Concrete, Ottawa, 1986. Noyes, Park Ridge, NJ, 1987, pp. 23-9.
19
16. Fournier, B. & Bdrubd, M. A., Evaluation of a modified chemical method to determine the alkali-reactivity potential of siliceous carbonate aggregates. Canadian developments in testing concrete aggregates for alkali-aggregate reactivity. Ministry of Transportation, Ontario, Engineering Materials Rep. 92, 1991, pp. 118-35. 17. Scott, J. F. & Duggan, C. tL, Potential new test for alkali-aggregate reactivity. Proc. 7th Int. Conf. Alkah-Aggregate Reaction m Concrete, Ottawa, 1986. Noyes, Park Ridge, NJ, 1987, pp. 319-23. 18. Jones, T. N., Grabowski, E., Quinn, T., Gillott, J. E., Duggan, C. R. & Scott, J. F., Mechanism of expansion in the Duggan test for alkali-aggregate reaction. Canadian developments in testing concrete aggregates for alkali-aggregate reactivity. Ministry of Transportation, Ontario, Engineering Materials Rep. 92, 1991, pp. 70-82. 19. Rogers, C. A. & Tharmabala, T., Prestressed concrete members affected by alkali-silica reaction. Concrete International: Design and Construction. American Concrete Institute, Detroit, MI, 1990, pp. 35-9. 20. DonneUy, C. R., The use of slag cement to counteract alkali-aggregate reactivity at the Magpie River development. Proc. 43rd Canadtan Geotechnical Conf., Vol. 2, Universit6 Laval, Quebec, 1990, pp. 499-507. 21. Sturrup, V. R., Hooton, IL D. & Clendenning, T. G., Durabdity of Fly Ash Concrete, SP-79, Vol. 1. American Concrete Institute, Detroit, M/, 1983, pp. 71-86. 22. Durand, B., Berard, J., Roux, R. & Soles, J., Alkali-stlica reaction: the relation between pore solution characterisucs and expansion test results. Cement Concrete Res., 20 (1990) 419-28. 23. Ingham, K. W., The beneficiation of mineral aggregates. Can. Pit Quarry, Oct. (1969).
Alkali-Aggregate Reactivity in Canada C. A. Rogers Ministry of Transportation, 1201 Wilson Avenue, Central Building, Downsview, Toronto, Ontario, Canada M3M 1J8
Abstract In Canada, three types of alkali-aggregate reaction in Portland cement concrete are recognized. Each type is evaluated using different tests. Corrective measures such as the use of low-alkali cement, lower cement contents, or pozzolans are seldom used with reactive aggregates. Beneficiation or selective extraction is used with some reactive aggregates. Work is being conducted on multilaboratory study of existing tests and new, rapid tests. Keywords: Concrete, aggregate, alkali reaction, Canada, test, cement, durability.
The wide range in rock types (Table 1) capable of causing damage as a result of A A R gives some idea of the difficulty of selecting appropriate testing procedures and corrective measures. For instance, the mortar bar expansion test used for alkali-silica reactive rocks gives only small expan-
I
~
. -
O U EBEC
~,~.~ ~ '
o
073 089 o
o
098
~ o
, .
! t
INTRODUCTION The oldest Canadian concrete structure known to have been affected by alkali-aggregate reactivity (AAR) was the Hurdman Bridge built in Ottawa in 1906 and demolished in 1987. It was not until the 1950s, however, that A A R was first documented in Canada. ~ Over the past 30 years, cases of A A R have been found in many parts of Canada (Figs 1 and 2) with a variety of different rock types (Table 1). There are now over 160 reports and papers documenting Canadian work on AAR. 2 In eastern Canada there are probably over 1000 structures affected by A A R . For many years, Canadian engineers had been aware of the cracking typically caused by A A R but generally had not connected the cracking with internal expansive reactions. Much of the cracking was attributed to frost damage, particularly in older structures built before the almost universal use of air entrainment. Nearly all Canadian concrete affected by A A R has also been exposed to freezing and thawing and it is sometimes difficult, in forensic studies, to judge the principal agent of damage.
Fig. 1. Locationof general areas of known cases of AAR, and alkalicontents of Type 10 cements in Eastern Canada.
•
ALKALI C(~iTENT$ OF TYPE tO CEMENTS
]
0.85
Fig. 2.
Location of general areas of known cases of A A R ,
and alkalicontents of Type 10 cements in WesternCanada.
13 Cement & Concrete Composites 0958-9465/93/$6.00 © 1993 Elsevier Science Publishers Ltd, England. Printed in Great Britain
14
C A. Rogers
Table 1. Types of alkali-reactive rock found in Canada
Phanerozoic Andesites and associated volcanic rocks in British Columbia Quartzite in Rocky Mountains of British Columbia and Alberta Opaline shale/siliceous shale in Saskatchewan and Manitoba Upper Silurian subgreywacke on Ellesmere Island Devonian and Silurian chalcedony-bearing chert in glacial drift in northern Ontario and in quarries and glacial drift in southern Ontario Middle Ordovician siliceous and sfighfly sificeous limestone in southern and eastern Ontario and the St Lawrence Lowlands of Quebec Middle Ordovician dolomitie limestone and calcitic dolostone in Hudson Bay Lowlands and southern Ontario Cambrian sandstones and orthoquartzites in eastern Ontario and southwestern Quebec Rhyolitic tuff, southeast of Quebec City Phyllite in southeastern Quebec Phylhtes, greywackes, siltstones, argillites, schists, silieified tuff, devitrified glass, and quartzites of New Brunswick, Nova Scotia, and Newfoundland
289) is also used, but gives misleading results when carbonates are present. (3) Late/slow alkali-silica/silicate reaction with sandstones and granites containing strained quartz, and metamorphosed sediments such as phyllite, ar~llite and greywacke, where exfoliation of phyllosilicates causes swelling of aggregate particles. Rocks susceptible to this type of reaction usually perform well in conventional alkali-aggregate tests. A concrete prism expansion test at 38"C and high cement and/or high alkali contents is the most useful procedure. Rock cylinder or miniature rock prism expansion tests are also useful. The procedures necessary to determine the type or types of reaction to which an aggregate may be susceptible are not simple. Flow charts used for determining the type of reaction and the necessary tests are described in the current Canadian Standard for Concrete?
Rhyolite in Nova Scotia
Precambrian Granite and granite gneiss with strained quartz Sandstone, quartzwacke, arkose, greywacke, and arglUite of the Huronian Supergroup in central Ontario and western Quebec Chert in banded iron formations (taconite)
sions with expansive dolomitic limestone susceptible to the alkali-carbonate reaction (ACR) or greywackes and argillites susceptible to the slower alkali-silicate reaction. For convenience, alkalisilicate reactions in Canada have been grouped into three types, each having different tests or criteria used in evaluation: (1) Alkali-carbonate reaction with dolomitic limestones and calcitic dolostones of Ordovician age. Chemical analysis for CaO :MgO and A1203, 3 the rock cylinder expansion test (ASTM C 586), and the concrete prism expansion test (CAN3-A23.214A) are useful for identifying these rocks. (2) Alkali-silica reaction with chalcedony, opal, cristobalite, glass, etc. is usually evaluated by the mortar bar expansion test (ASTM C 227); concrete prism expansion tests at 38"C with high cement contents, and rock cylinder expansion tests, are also used. The quick chemical test (ASTM C
CANADIAN CEMENTS Figures 1 and 2 show that the majority of areas where A A R occurs are in central and eastern Canada, where low-alkali cements are seldom used. West of Ontario, most of the normal Portland cements (CSA Type 10) have alkali levels of less than 0.8% Na20 equivalent) and, as a result, there are very few documented cases of A A R : In central and eastern Ontario, the majority of normal cements have alkali levels of more than 0.9%. For the 19 normal cements made g,a Canada in 1988, the Na20/K20 ratio ranged from 1.09 to 0.07, with a mean of 0.38 (median 0.31). In a 1977 survey of the alkali contents of Canadian Type 10 cements, six of 23 sources had alkali contents of more than 1"0%. 6 In 1988, only two of 19 producers had mean alkalis of more than 1.0% (Table 2). This reduction in alkali content can generally be attributed to the closing of obsolete plants that used high-alkali raw materials. At present, two normal Portland cements are made in Canada which contain a pozzolanic mineral admixture (Type 10P with about 6 and 8% silica fume). The remaining normal Portland cements usually contain up to 5% limestone interground with the clinker. Six plants (three in eastern and three in western Canada) which norreally make high-alkali cements also make a lower-alkali cement, usually without 5% limestone
Table
2. Alkali contents of Canadian portland cements (latest monthly or yearly average, January 1989)
Cement co.
Location
Capacity (tonnes x 10-6)
Cement type
At~ti (%) Na:O
K zO
Eqmv. NazO
Tilbury Cement
Delta, British Columbia
1-25
10
0.28
0.29
0.47
Lafarge Canada
Richmond, British Columbia
0"33
10 3 50
0"35 0"30 0"30
0.32 0-20 0-32
0.55 0.49 0.50
Lafarge Canada
Kamloops, British Columbia
0-19
10 15
0-20 0"20
0.44 0"50
0-49 0.53
Lafarge Canada
Exshaw, Alberta
1.10
10 3 50
0"15 0"15 0-15
0"79 0"80 " 0"68
0.67 0.68 0'60
Inland Cement
Edmonton, Alberta
0-73
10 50
0"22 0"24 0-20
0-84 0"76 0'47
0.77 0.74 0'51
0-20
10 50
0-32 0-35 No data
0-80 0"63
0"85 0.76
0"33
10 50 1
0-30 0-32 0.27
0"62 0-42 0-40
0.71 0.59 0-53
Low alkali to special order Inland Cement
Regina, Saskatchewan Low alkali to special order
Inland Cement
Winnipeg, Manitoba Low alkali to special order
St Marys
St Marys, Ontario
0-70
10 50
0.48 0-42
0'99 0-89
1.13 1.01
Lafarge Canada
Woodstock, Ontario
0.50
10/20 50 2 3
0.12 0-10 0-12 0-09
0-47 0.43 0-46 0-53
0-43 0.38 0.42 0-44
St Lawrence
Mississauga, Ontario
1-60
10 1 30 2
0"21 1-18 As above approx. As above approx. N o data
0-99
0.99 0-60
Type 2 to NY DOT specifications St Marys
Bowmanville, Ontario Type 2 to NY DOT specifications
0.60
10 2
0-27 0.24
Lafarge Canada
Bath, Ontano
1-00
10 30
0.27 0-92 As above approx.
0.88
Lake Ontario
Picton, Ontario
1.3
10 30 2 5
0.23 0-20 0"16 0.14
1.15 1"15 0.72 0-76
0.98 0.94 0-63 0.65
--
10
0-29
0.45
0'59
Type 2 to NY DOT specifications
1.10 0-55
Miron
Montreal, Quebec (supply imported cements from Spain and Greece)
10
0.13
0-62
0.54
St Lawrence
Jolliette, Quebec
1.10
10 1 20 50
0.37 0.37 0.27 0"31
0.92 0.94 0.88 0-91
0.98 0"99 0"85 0.91
St Lawrence
Beauport, Quebec
0.75
10 30 2
0.30 0-65 As above approx. 0.35 0.40
0.61
Ciment Quebec
St Basile, Quebec
0-55
10 20 30
0-19 0-19 0"18
1.07 1.02 1-11
0-89 0-86 0.91
LafargcCanada
St Constant, Quebec
0.94
10 20 3 10P
0-28 0-31 0-37 0-20
0.91 0.93 0-95 0.85
0.87 0.92 0.94 0.76
I0 50 3 10
0.09 0"08 0.04 0"35
1.24 0-41 1'20 1.10
0.90 0"35 0.83 1-07
10P contains 7.5% silica fume Lafarge Canada
Brookfield, Nova Scotia
0.45
North Star Cement
Comerbook, Newfoundland
0.16
0-73
16
C A. Rogers
addition, to meet US requirements. Many US highway authorities adjacent to the Canadian border specify either a low-alkali cement ( < 0-6%) or a reduced-alkali cement (<0.7%). These lower-alkali cements are usually reserved for the export market to the USA, and are not normally sold in Canada. The cement companies regard the universal adoption of the specification of low-alkali cement to control A A R with concern. The older, wetprocess plants can often make a low- or reducedalkali cement by not recycling alkali-rich dust from the precipitators. The newer, dry-process plants normally can achieve a low-alkali cement only by selecting low-alkali raw material sources. In either case, the cost of production is usually increased. There is a natural apprehension that purchasers will begin to specify the universal use of low-alkali cement to reduce the likelihood of AAR. However, with increased concern about long-term durability there appear to be renewed efforts by the cement producers to find ways of making low- or reduced-alkali cements.
CANADIAN RESEARCH In an effort to prevent the universal adoption of restrictions on maximum alkali levels, the Canadian Standards Association (CSA)(A5) Committee on Cement formed a sub-committee in 1980 to look for and recommend the adoption of rapid test procedures to evaluate concrete aggregates for AAR. At present, many of the current procedures require several months or years of testing, with the result that the structure is usually under construction or even completed by the time the aggregate testing is finished. The sub-committee has a membership of about 40 people. Meetings are usually held twice yearly, and one of these meetings is combined with a field trip to review interesting cases of AAR. A secondary purpose of the sub-committee is to provide a means of communication for those people in Canada involved in work on AAR. A number of authorities are sponsoring research into AAR. The Canadian government, through CANMET, has funded investigations of the incidence of A A R in many of the Provinces, and its control using Canadian pozzolans. This work is hampered to some extent by the lack of reliable, rapid test procedures. The Canadian government also funds work by university
researchers through operating grants, and finances individual projects. 7 Many Provinces also support A A R research; this work is either performed or funded by provincial authorities such as electrical utilities or transportation departments. Over the past 5 years, the total amount of money spent in Canada on investigations and testing of A A R probably has not exceeded an average of about S1 000000 yearly. The amount currently spent on research is small compared with the direct and indirect costs of damage caused by AAR.
EVALUATION OF AGGREGATES The first step recommended by CSA4 in evaluating the suitability of concrete aggregates is a thorough petrographic examination and investigation of prior field performance. This will determine what tests, if any, are needed. Failure to classify the aggregate properly or to recognize minor deleterious components will upset the remainder of the testing program. Petrographic examination should, if possible, be based on knowledge of the prior field performance of similar aggregate. A simple classification of the rock types and minerals present is rarely satisfactory. In Canada, the acquisition of field performance is often impossible because of construction in remote and undeveloped areas. Many of the tests used in Canada have been found to suffer from such problems as lack of correlation with field performance or poor multilaboratory precision. The concrete prism expansion test has been subjected to a multi-laboratory study to find less variable storage conditions than the conventional humid curing room. s The study found that the multi-laboratory coefficient of variation was 23%. Although different storage conditions resulted in significantly different expansion, the multi-laboratory variation was not substantially reduced. The mortar bar expansion test has been found to be susceptible to alkali leaching caused by the presence of wicks in the storage container specified by the American Society for Testing and Materials (ASTM); 9 this resulted in substantial reductions in expansion of alkali-silica reactive mortar. Alkali-carbonate expansive concrete in outdoor exposure has been found to expand about twice as much as concrete prisms stored in the laboratory under moist conditions at 23"C after 5 years. ~° Further studies are
Alkali-aggregate reactivity in Canada
under way to investigate the relationship between field and laboratory expansion of alkali-silica reactive concrete. The CSA subcommittee has established a draft procedure for the accelerated mortar bar test (1M NaOH at 80"C for 14 days) of Oberholster and Davies. ~ This test appears reasonably precise and capable of recognizing the majority of reactive aggregates found in Canada. A multilaboratory study has been conducted to finalize some details and investigate precision. ~2 It was found that the multi-laboratory coefficient of variation for experienced laboratories was about 10%, except with very small expansions. It is likely that only aggregate giving more than 0-15% expansion at 14 days (in solution) will be classed as potentially reactive. Recent work proposes the use of additional limits of 0"33% at 28 days and 0.48% at 56 days. ~3If the 14-day limit is exceeded then the material may still be satisfactory if the later limits are met. The reason for this approach is the satisfactory field performance of slightly cherty sands that caused continued expansion of mortars in this test. Aggregates giving expansions exceeding these limits can be classed as reactive or potentially reactive, and require further investigation and the possible use of corrective measures. The quick chemical test (ASTM C289) is generally unsatisfactory with alkali-silica reactive carbonate aggregates because much of the alkalisoluble silica released is re-precipitated as calcium silicate hydrates. Researchers in Quebec t4.~5 have had success with modifying the test so that it is done on the insoluble residue of quarried carbonate rocks. This modification of the test has been investigated in a multi-laboratory study, which found it to be rather variable, depending on the aggregate. ~6The variability and complex nature of the test prevent its adoption as a standard procedure. A rapid procedure of expansion testing of concrete cores proposed by Scott and Duggan 17 has been studied by researchers at the Universities of Calgary and Manitoba to determine the mechanism of expansion. This procedure has similarities to the Conrow test (ASTM C342), and at present does not appear to be satisfactory for separating reactive from non-reactive aggregate because A A R is not a major cause of expansion in the test. ts To assist in the calibration of existing and new procedures, the Ontario Ministry of Transportation has established three 110-tonne stockpiles of
17
known reactive aggregates. Aggregates from these stockpiles are available free of charge for calibrating test procedures or research into AAR. CORRECTIVE IVlFASURES The use of low-alkali cement to inhibit A A R is not a universally satisfactory solution because of its lack of widespread availability in eastern Canada. The specification of low-alkali cement is unusual in eastern Canada. In northern Ontario, the Ministry of Transportation used low-alkali cement on four bridges in 1970, but has seldom specified it since. A notable exception was the use in 1990 of a lower-alkali cement (<0.7%) for concrete paving with a slightly reactive aggregate near Peterborough, Ontario. Recently, however, Canadian National Railways has started to specify low-alkali cement for all concrete, irrespective of the nature of the aggregate. The reason for this was the costly and tedious deterioration of prestressed concrete railroad ties as a result of A A R . t9 In Alberta, an engineer recently proposed the use of a maximum cement alkali content of 0.3% for dam construction. Such a cement is not available in Canada, but such caution is understandable, given the high cost of remedial work on large concrete structures affected by AAR. Reducing cement contents and, hence, alkali levels is often not practical because of the severe Canadian climate, which generally requires low water/cement (w/c) ratio, air-entrained concrete for all exposed work. In practical terms, minimum cement contents of 325-450 kg/m 3 are required if durability is to be ensured. The use of blast furnace slag cement has generally not been adopted to control AAR, despite its apparent efficiency at high levels of substitution ( > 50%). A notable exception was the use of 50% ground granulated blast furnace slag cement on a recent hydro-electric development in northern Ontario. 2° The aggregate in this case was a rectavolcanic-rich gravel. The metavolcanic material was found to contain a high proportion of microcrystalline quartz, which caused accelerated mortar bar test expansions of about 0.2% at 14 days. It was judged to be marginally reactive and worthy of taking corrective measures. In the highway environment, where de-icer salts are used, concrete made with significant amounts of blast furnace slag cement, and other supplementary cementing materials, has been found to be more
18
C A. Rogers
sensitive to freeze-thaw damage. This has also been confirmed in laboratory salt-scaling tests (ASTM C672) which have shown that, at 50% substitution levels, the scaling loss from the surface may become unacceptable. For this reason, the Ontario Ministry of Transportation permits a maximum of only 25% blast furnace slag substitution. The harsh Canadian winter climate causes practical problems in the use of supplementary cementing materials to control AAR. Other pozzolans such as fly ash and silica fume are used in Canada, but rarely for the purpose of controlling AAR. A notable exception was the successful use of 20% fly ash by Ontario Hydro on the Lower Notch Dam to prevent an alkalisilica reaction with an ar~llite. 2~ Pozzolans are more commonly used for their other advantages such as reduced cost, increased strength-producing properties, or reduced heat of hydration. Condensed silica fume certainly seems effective in slowing or preventing expansion caused by alkali-silica reaction in the laboratory. 22 Longterm outdoor performance studies have to be conducted to see if this initial promise is confirmed in field exposure. The most commonly used control measure is the beneficiation or selective quarrying of aggregate. A technique used in areas of Ontario that contain significant amounts of chert in glacial drift is selectively to crush oversize gravel ( > 40 mm), which normally contains much less chert. Another technique used to remove shale and chert is the use of heavy media separation. There are currently two operations in Ontario. Jigs have also been used in the past to remove shale and chert from gravel, but with limited success. 23 In carbonate bedrock quarries, reserving a specific level or bench of superior quality is a common practice. The underlying or overlying beds may be alkalireactive, but careful, conscientious extraction and stockpiling can ensure an adequate supply of nonreactive rock. The testing of aggregate suitability is complex and time consuming. There is seldom sufficient time during construction to do more than simple tests such as gradation, petrographic examination, and organic content determination. In Ontario, the Ministry of Transportation has compiled lists of acceptable concrete aggregates. All sources shown in these lists have been extensively tested or have a satisfactory record of past performance in concrete. On all contracts with concrete items, a 'Concrete Aggregate Sources List' is issued. The
concrete aggregates used by the contractor must be from the approved sources shown on the list. THE NEXT 10 YEARS New cases of A A R will be discovered in Canada. A group of reliable, rapid test procedures will become available for each type of reaction. Each of these tests will have known multi-laboratory precision, and will be calibrated with known reactive aggregates. Corrective measures will generally only be used where non-reactive aggregates are more expensive. There will be more frequent specification of low- or reduced-alkali cements as more people become aware of the potential problem. There will also be a more widespread adoption of lists of prequalified, non-reactive aggregate. A document will be produced to provide guidance to engineers and owners on recognition of A A R and the management of structures affected by these reactions. ACKNOWLEDGEMENTS The author is indebted to all his Canadian colleagues. Without their observations, research, and assistance, this paper could not have been written. REFERENCES 1. Swenson, E. G., Cement-aggregate reaction in concrete of a Canadian bridge. ASTM Proc., 57 (1957) 1043-56. 2. Rogers, C. A. & Worton, S., Alkali-Aggregate Reactions in Canada -- A Bibliography, 3rd edn. Ontario Ministry of Transportation, Engineering Materials Office, Rep. EM-83, 1988, 26 pp. 3. Rogers, C. A., Evaluation of the potenaai for expansion and cracking of concrete caused by the alkali-carbonate reaction. Cement, Concrete Aggregates, CCAGEP, $ (1)
(1986) 13-23. 4. Canadian Standards Association, Appendix B -Alkali-aggregate reaction. Standards CAN3-A23.1M90 and CAN3-A23.2-M90, Canadian Standards Association, Rexdale, Ont., 1990, pp189-100. 5. Morgan, D. R., Examples of alkali-aggregate reactivity in concrete structures in western Canada. Canadian developments m testing concrete ~ggregates for alkali-aggregate reactivity. Ministry of lransportauon, Ontario, Engineering Materials Rep. 92, 1991, pp. 35-49. 6. Grattan-BeUew, P. E., Sereda, E J. & Dolax-Mantuam, L., The aggregate shortage and high alkali cement in a changing energy situation. Can. J. Civil Eng., 5 (1978) 250-61. 7. Chert, H. & Grattan-Bellew, P. E., Effect of cement composition on expansion of mortar bars due to alkali-silica reaction. Proc. 8th Int. Conf. AIkali-Agsregate Reacnon in Concrete, Kyoto, 1989, Society of Materials Science of Japan, Kyoto, pp. 635-40.
Alkali-aggregate reactivity in Canada 8. Rogers, C. A., Interlaboratory study of the concrete prism expansion test for the alkali-carbonate reaction. Canadian developments in testing concrete aggregates for alkali-aggregate reactivity. Ministry of Transportation, Ontario, Engineering Materials Rep. 92, 1991, pp. 136-49. 9. Rogers, C. A. & Hooton, IL D., Reduction in mortar and concrete expansion with reactive aggregates due to alkali leaching. Cement, Concrete Aggregates, CCAGDP, 13 (1)(1991)42-9. 10. Williams, D. A. & Rogers, C. A., Field trip guide to alkali-carbonate reacnons in Kingston, Ontario. Ministry of Transportation, Ontario, Engineering Materials Office, Information Rep. 145, 1991, 26 pp. 11. Oberholster, 1L E. & Davies, G., An accelerated method for testing the potential alkali-reactivity of siliceous aggregates. Cement Concrete Res., 16 (1986) 181-9. 12. Hooton, R. D., Interlaboratory study of the NBRI rapid test method and CSA standardization status. Canadian developments in testing concrete aggregates for alkali-aggregate reactivity. Ministry of Transportation, Ontario, Engineering Materials Rep. 92, 1991, pp. 225-40. 13. Hooton, R. D., New aggregate alkali-reactivity test methods. Ministry of Transportation, Ontario, Research and Development Branch Pep. MAT 91-14, 1991, 83 pp. 14. Berard, J. & Roux, R., La viabilit6 des brtons du Qudbec: le r61e des granulats. Can. J. Ctvil Eng., 13 (1) (1986) 12-24. 15. Fournier, B., Brrubr, M. A. & Vezina, D., Investigation of the alkali-reactivity potential of limestone aggregates from the Quebec City Area (Canada). Proc. 7th Int. Conf. Alkali-A~,regate Reaction in Concrete, Ottawa, 1986. Noyes, Park Ridge, NJ, 1987, pp. 23-9.
19
16. Fournier, B. & Bdrubd, M. A., Evaluation of a modified chemical method to determine the alkali-reactivity potential of siliceous carbonate aggregates. Canadian developments in testing concrete aggregates for alkali-aggregate reactivity. Ministry of Transportation, Ontario, Engineering Materials Rep. 92, 1991, pp. 118-35. 17. Scott, J. F. & Duggan, C. tL, Potential new test for alkali-aggregate reactivity. Proc. 7th Int. Conf. Alkah-Aggregate Reaction m Concrete, Ottawa, 1986. Noyes, Park Ridge, NJ, 1987, pp. 319-23. 18. Jones, T. N., Grabowski, E., Quinn, T., Gillott, J. E., Duggan, C. R. & Scott, J. F., Mechanism of expansion in the Duggan test for alkali-aggregate reaction. Canadian developments in testing concrete aggregates for alkali-aggregate reactivity. Ministry of Transportation, Ontario, Engineering Materials Rep. 92, 1991, pp. 70-82. 19. Rogers, C. A. & Tharmabala, T., Prestressed concrete members affected by alkali-silica reaction. Concrete International: Design and Construction. American Concrete Institute, Detroit, MI, 1990, pp. 35-9. 20. DonneUy, C. R., The use of slag cement to counteract alkali-aggregate reactivity at the Magpie River development. Proc. 43rd Canadtan Geotechnical Conf., Vol. 2, Universit6 Laval, Quebec, 1990, pp. 499-507. 21. Sturrup, V. R., Hooton, IL D. & Clendenning, T. G., Durabdity of Fly Ash Concrete, SP-79, Vol. 1. American Concrete Institute, Detroit, M/, 1983, pp. 71-86. 22. Durand, B., Berard, J., Roux, R. & Soles, J., Alkali-stlica reaction: the relation between pore solution characterisucs and expansion test results. Cement Concrete Res., 20 (1990) 419-28. 23. Ingham, K. W., The beneficiation of mineral aggregates. Can. Pit Quarry, Oct. (1969).