ASR PREVENTION IN DOD

Transcription

1 ASR PREVENTION IN DOD L.J. Malvar, PhD, PE Naval Facilities Engineering Service Center 00 rd Avenue, Port Hueneme, CA , Fax 0-- luis.malvar@navy.mil 1 November 00 Word Count 1

2 ABSTRACT Supplementary cementitious materials (SCM) are used in concrete to prevent alkali silica reactivity (ASR) in Department of Defense (DOD) concrete construction. To insure that deleterious expansion does not take place, the DOD specifications rely primarily on ASTM C and C 1. For airfield pavements, the threshold expansion indicative of ASR potential was conservatively set to 0.0% after days of exposure. In this paper, the rationale for the reliance on these tests is explained, as well as the choice for the threshold expansion. The approach for ASR prevention, currently being expanded to all concrete construction, is also detailed. 1

3 INTRODUCTION The Department of Defense (DOD) has been promoting the use of Supplementary cementitious materials (SCM) as cement replacement in concrete, primarily for alkali-silica reaction (ASR) prevention, but also as means to reduce carbon dioxide emissions and increase the recycling of SCMs, which typically are waste products [1-]. The resulting concrete is generally cheaper and more durable, and results in lower maintenance costs. In addition to the use of SCMs, a methodology had to be devised to decide which tests to choose to determine the ASR potential of a concrete mix, both with and without SCMs. While several tests are available [1], lately ASTM C, C 1, and C 1 [-] have been recognized as more reliable for ASR potential assessment, and have become the backbone of current specifications [1-, ]. These are accelerated tests, which try to predict within a period of 1 week to years (depending on the test) whether deleterious expansion will occur. ACCELERATED TESTING ASR may take 1 to years, or more, to result in deleterious expansion which may affect, or at least cast doubt on, the structural or serviceability capacity of the structure. Therefore accelerated tests are required to determine the ASR potential of a mix. Of course, the tests should be able to accelerate the reaction without changing the reaction type, which puts a limit on the accelerating conditions (such as temperature), and the overall maximum acceleration achievable. Accelerated tests try to predict if a significant amount of deterioration (quantifiable, for example, by the amount of deleterious expansion) will take place at a given time in the future (e.g. the required structural life). Depending on the amount of acceleration, the test expansion will be limited to a certain threshold after a certain time of exposure which should be determined by correlation to actual field data. In ASTM C 1, for example, the expansion threshold is 0.0% (related to the onset of cracking), and the exposure time is 1 year ( years if SCMs are used). If in real life, if the same expansion threshold is used and the required life is say 0 years, then the expected test acceleration is 0. In other words, it would be expected that the relatively high test temperature and the addition of alkalis to the mix would be able to accelerate the reaction twenty-fold. Less accelerated tests are more realistic, for example C 1 is generally accepted to be more realistic than C, but less accelerated tests would take longer to run: a perfectly realistic test would have an acceleration of 1 and take 0 years (or the required life) to run. In any case, the key is to provide a correlation between the test threshold expansion at the chosen exposure time and the maximum allowed field expansion at the required structural life. Predictions from less accelerated tests may, however, be more susceptible to actual field conditions. For example C 1 relies in part on its relatively high temperature of C (0. F) to provide the reaction acceleration. This temperature will indeed result in significant acceleration when used to predict structural behavior in cold climates, but may not result in much acceleration in a tropical climate (e.g. Hawaii) with median high temperatures around 0 C ( F). Hence the predictions may be significantly affected by the local ambient temperature. In contrast, C is a less realistic test, but it is relatively less affected by the local climate since its temperature is so high (0 C or 1 F). In addition, tropical climates can present constant wet-dry cycles which could result in higher acceleration than the constant moisture provided by C 1 (e.g. as Fournier et al. state: wetting and drying cycles can significantly accelerate the deterioration of concrete suffering from internal expansion processes such as ASR []). Just because less accelerated tests are more realistic does not make them better predictors of field performance. For example if the expansion threshold for C 1 was much higher, it is obvious that the prediction would be significantly affected. Hence the key is to determine the threshold expansion and exposure time which provide the best field correlation. If the exposure time is set, then the optimum expansion threshold has to be determined. This expansion threshold is often reported in the appendix (non-mandatory information) of the standard test method, since it is more a criteria requirement than a test requirement. Some flexibility in the exposure time is often provided in the test method, for example ASTM C and C 1 indicate if readings are continued beyond the 1-day period, take at least one reading per week, and C 1 states additional readings beyond those required for the specific application are suggested at -month intervals. It is logical then to next assess the validity of the chosen threshold expansion and exposure time for the tests.

4 FALSE POSITIVES AND FALSE NEGATIVES To assess the validity of the chosen threshold expansion and exposure time for the tests, we must first define what is an optimum test. An optimum test is one that has no false predictions, but of course such a perfect test does not exist. There are two types of false predictions: - False Negative: if the ASTM threshold and exposure time chosen predict no failure (negative) but the field specimens or structure show failure, this is a false negative. False negatives will result in premature structural loss and can be very expensive. False negatives should really be avoided. - False Positive: If the ASTM threshold and exposure time chosen predict failure (positive) but the field specimens show no failure, this is a false positive. False positives can result in additional costs if the aggregate is unnecessarily rejected and another one needs to be imported, usually from further away, adding transportation costs to the aggregate cost. However, in the continental United States, SCMs such as fly ash and ground granulated blast furnace slag are usually available that can mitigate the reactivity, and result in a concrete mix that is generally slightly cheaper and more eco-friendly (due to the reduced cement content). Hence false positives are often not a significant concern. Therefore, when comparing the test predictions to the actual field observations, the optimum test threshold expansion and exposure time should be those that minimize first the false negatives and second the false positives. OPTIMUM THRESHOLD EXPANSION AND EXPOSURE TIME It should first be noted that ASTM C, C 1, and C 1 all allow for exposure times beyond standard practice, and the thresholds reported are included in the non-mandatory information, since the thresholds are criteria decisions, and not really part of the test method itself. Therefore optimizing threshold expansion and exposure time does not really affect the test procedure. When reporting predictions from accelerated test data, such as C expansions, it is usually not possible to compare those predictions to field data that are sufficiently old, as such field data usually do not exist. Hence when assessing C predictions, it is often assumed that C 1 is the standard to meet. This is not correct: C 1 is an imperfect test itself, and comparison for the purpose of validation should only be made to field data. In fact, if the origin of the C 1 threshold expansion of 0.0% at 1 year is examined, it is found that C 1 (under X1.) points to a single comparison to field data in its reference, which correctly states: A laboratory method of evaluating length change of concrete is only useful if it predicts with acceptable accuracy the expansion and cracking of concrete structures in the field. However, it also indicates: Measurements, however, have been made on very few concrete structures or large scale test samples under Canadian exposure conditions, so that correlation between laboratory test results and deterioration of similar concrete in structures is uncertain. Reference does not compare expansion limits to field data, rather it compares expansion limits between agencies (ASTM, CSA, USACOE). So where is the original comparison to field data? Reference (1) of C 1 points to CSA CAN-A.1-M (1) which includes several thresholds. However, in 1, CSA A.-1A (precursor to ASTM C 1) was changed significantly by increasing the alkali in the mix, but the threshold (derived from the 1 work) was not changed. Hence it appears that the current threshold expansion of 0.0% at 1 year may be based on an obsolete test used in 1. ASTM C 1 was first published in 1 (its equivalent Canadian precursor was developed in 1), and C was first published in 1, so for both tests comparisons can only be made to structures constructed since then, i.e. that are currently about 1 years old or less (except for any specimens or structures constructed during the standards development, e.g. constructed during the development of the 1 version of CSA A.-1A or the development of P-1, which are the precursors to C 1 and C, respectively). Comparisons have been reported with structures that are less than years old [,, ]. While such field data can be used to assess high to moderate reactivity aggregates, it is not very useful for low to moderate reactivity comparisons. One such set of field data, a CANMET-sponsored field and laboratory research program [], was used to assess C /C1 and C 1 thresholds and exposure times [, ]. Table 1, derived from this work [, ], compares field data to C, C 1, and C 1 predictions. In the comparison, field specimens were assumed to have cracked when they reached an expansion of 0.0% (which is generally accepted as the onset of cracking): - For C /C 1, a combination of a 1-day exposure time and 0.1% threshold expansion results in a significant percentage of false negatives, as does C 1 with 0.0% at years. - C /C 1 with -day exposure time and either 0.0% or 0.1% threshold expansions resulted in low percentages of false negatives. Although the false positives are higher, those are less important, as indicated

5 earlier. For this dataset, using 0.1% appears to minimize the total false predictions. For C 1, an expansion between 0.0% and 0.0% would similarly result in few false negatives and lower total falses. Since these comparisons were made to field data less than years old, it is likely that more reactivity will be found in the following or 0 years, and therefore the percentage of false negatives will further increase, while the percentage of false positives may decrease. This further emphasizes the need to minimize the false negatives by choosing a conservative duration/threshold expansion combination for the criteria. Finally, in some field applications, synergism can take place between the ASR mechanism and another mechanism [] (e.g. delayed ettringite formation in steam cured components, or applied traffic loads in pavements) which would accelerate the structural degradation in the actual field application compared to the controlled field data presented in Table 1. TABLE 1. Percentage of false negatives and false positives from comparison to field data Test C C C C C C 1 C 1 C 1 C 1 C 1 C 1 C 1 C 1 Exposure Time 1-day 1-day 1-day -day -day -year -year -year Threshold (%) False (%) 1 0 False + (%) Total Falses (%) 1 1 Samples (#) The 1-year-old field data set at the Kingston outdoor exposure site in Canada includes accelerated mortar bar test data (CSA A.-A, equivalent to ASTM C or C 1) []. A threshold anywhere from 0.0% to 0.% at 1 days would have properly predicted the cracking of mixes,,, and in unreinforced beams and pavement slabs (some of the cracking was faint or slight, but even those cracks may not be acceptable in airfield pavements, for example). Prisms were also prepared following CSA A.-1A. Unfortunately, the version of CSA A.-1A appears to have been used, which was substantially changed in 1 by including additional alkali to the mix [1]: the 1 version is the one close to the current ASTM C 1. Those prisms included 1 kg/m of cement in the mix, and the cement had an alkali content of 0.%, for a total alkali content of. kg/m, which does not meet the current required. kg/m. Therefore this data set cannot really be used directly to evaluate the current C 1. The preceding comparisons explain the current DOD choice of ASTM C (and C 1) with a threshold of 0.0% at days. Per Reference, this is approximately equivalent to an expansion of 0.0% at 1 days. Although a C 1 threshold of about 0.0% at -years (with SCMs) would be similar for this data set, C 1 has other shortcomings, such as the long test duration, which is impractical and can cast doubt on the reliability of a test prediction based on aggregates quarried 1 year (or years) ago, whose reactivity could have changed. For example, for Pacific Northwest aggregates, Shrimer reports: in one pit, evaluation showed reaction levels ranging from low (safe use) for one year s production to high (very reactive) in tests conducted for another year s production [1]. DOD SPECIFICATIONS The DOD specifications then rely primarily on ASTM C, by first assessing each aggregate against a threshold expansion of 0.0% at days (of exposure). If the aggregates fails, then cement replacement by an SCM is required until an expansion of less than 00% at days is obtained in C 1. This threshold is lower than the optimum (lowest false negatives and total falses) of 0.1% at days shown in Table 1. It was however chosen by the Tri-Service Pavement Team committee since (1) the data used in Table 1 is less than years old and more false negatives are possible, () it would provide safety against variability in aggregate reactivity and SCM composition, and () it would result in increased use of SCMs. In DOD applications, the use of SCMs is promoted for environmental benefits (recycling of waste products and reduction in CO emissions from reduced cement usage), increased durability (including protection against variability in aggregate reactivity), lower maintenance costs (due to increased durability), and potentially lower

6 upfront costs (if fly ash or slag are used). Hence, even if the aggregates are found to be innocuous, the use of minimum amounts of SCMs (as percentage of total cementitious) is encouraged (Army and Air Force), or required (Navy), due to the previous benefits. Those minimum SCM contents were chosen because lower amounts have been shown to provide little benefit, or even exhibit pessimums, i.e. worse expansion (e.g. at replacement levels from % to 1% for many Class F ashes, and up to % for slag [1, 1]). Table shows the minimum SCM contents per the guide specification on Marine Concrete (UFGS 0 1 ). Table also shows some maximum contents: those are rather recommendations, and are provided as protection against slow strength gain rate, and finishing and air content control difficulties (e.g. for fly ashes with high loss on ignition). Ternary mixes including two SCMs are also permitted, since, for example, mixes of SCMs of various particle sizes can be very efficient at preventing ASR and reducing permeability. These requirements are very similar to those recently proposed by the California Department of Transportation in its Section 0 [1], which also provides a simple way to address mixes with three or more SCMs. If either aggregate is reactive, the New Mexico Department of Transportation requires Class F fly ash that meets SiO + Al O + Fe O > % in its Section 0 [1]. In cold climates, some restrictions on SCM contents exist, to avoid constructability problems such excessive setting time, delays in stripping and joint-cutting, and slow strength gain rate. For example ACI 1 [1] limits the contents of fly ash, slag, and silica fume to %, 0%, and %, respectively, for Exposure Class F (very severe freezing and thawing, and in contact with moisture and exposed to deicing chemicals). In general this is not a problem: for slag and silica fume those limits match the maximums allowed in Table, and for fly ash, a % replacement is generally sufficient to mitigate low to medium reactivity. In some cases, typically outside of the continental United States, these SCMs may not be readily available. In these instances, information from other sources (such as C 1, petrographic analysis, and performance history) can be considered on a case by case basis to further pinpoint the reactivity potential and determine the best mitigation approach, if needed. TABLE. Supplementary cementitious material content. Supplementary Cementitious Material Minimum Content Maximum Content Fly Ash or Class N Pozzolan SiO + Al O + Fe O > % SiO + Al O + Fe O > 0% SiO + Al O + Fe O > 0% SiO + Al O + Fe O > 0% 0% % 0% 1% 0% 0% 0% 0% Ultra fine fly ash or pozzolan % 1% GGBF Slag 0% 0% Silica Fume % % Figure 1 provides a qualitative representation of the DOD approach compared to current practice, and applies to predictions based on either ASTM C or C 1. Many current specifications try to provide the most accurate prediction, which, as shown in Table 1, can still result in significant total false predictions for the current dataset (which is less than years old) of % for ASTM C (and 0.1% at 1 days) or % for ASTM C 1 (and 0.0% at 1 or years) per Table 1. In Figure 1, this is shown under current practice as a significant risk at an unnecessarily high life cycle cost, mainly because this cost includes structural damage and failures due to false negatives (see Table 1), and more cement than needed. A more conservative prediction (e.g. using DOD s practice with ASTM C and 0.0% at days) results in more false positives, increased SCM requirement, but lower risk and generally significantly lower cost, which comes from slightly lower upfront cost (from substitution of cement with fly ash or slag), and prevention of structural damage and failures by significant reduction of false negatives. Note that if the cement replacement is further increased this can start to significantly interfere with strength gain rate, ultimate strength, etc, somewhat increasing costs (although false negatives will be prevented). These guidelines have been implemented in most DOD unified facilities guide specifications (UFGS) dealing with concrete construction. These UFGS are available at and include: 1. UFGS 0 01 Concrete Rehabilitation For Civil Works. UFGS Mass Concrete. UFGS Cast-in-Place Concrete. UFGS 0 0 Miscellaneous Cast-in Place Concrete. UFGS Cast-in-place Structural Concrete

7 UFGS Cast-in-place Structural Concrete For Civil Works. UFGS 0 1 Marine Concrete (draft). UFGS 0 00 Preplaced-Aggregate Concrete. UFGS 0 1 Shotcrete. UFGS 0 Concrete for Cutoff Walls. UFGS 0 Roller-Compacted Concrete for Mass Concrete Construction 1. UFGS 0 Precast [Prestressed] Structural Concrete 1. UFGS 0 1 Tilt-Up Concrete 1. UFGS 0 00 Lightweight Concrete Roof Insulation 1. UFGS Lightweight Insulating Concrete Overlay 1. UFGS Cast-in-Place Concrete Piles 1. UFGS Precast/Prestressed Concrete Piles 1. UFGS 1 1. Prestressed Concrete Cylinder Piles 1. UFGS 1 1 Piling: Composite, Wood and Cast-in-Place Concrete 0. UFGS 1.1 Cast-in-Place Concrete Piles, Steel Casing 1. UFGS 1 Concrete Pavement for Airfields and other Heavy-Duty Pavements > 000 Cubic Yards. UFGS Airfields and Heavy-duty Concrete Pavement less than 000 Cubic Yards. UFGS [Pervious] Portland Cement Concrete Pavement For Roads And Site Facilities. UFGS Roller Compacted Concrete (RCC) Pavement. UFGS 1 1 Concrete Sidewalks and Curbs and Gutters. UFGS Wire-Wound Circular Prestressed Concrete Water Tank. UFGS 1 1 Heat Distribution Systems In Concrete Trenches. UFGS Articulating Concrete Block Revetments. UFGS 1.1 Prestressed Concrete Fender Piling The guidelines were slightly modified for each specification, and many still need further modifying (e.g. to change or remove the SCM maximum contents in mass concrete) and updating to reflect the latest changes. 0 Less or none Risk SCM REQUIREMENT Less or none Life cycle cost More Very high RISK (%) 0 Current practice (high) Minimum COST ($) 0 Current practice DOD practice Low Unconservative More false negatives Most accurate Less false predictions ASR PREDICTION 0 Figure 1. Comparison of current and DOD practices to prevent ASR. Conservative More false positives

8 CONCLUSIONS The background behind the approach currently followed in DOD specifications to prevent ASR in concrete was presented. The approach relies on a conservative threshold in C and C 1 to insure that false negatives are avoided, and with the knowledge that this will increase the use of SCMs. Further, minimum SCM contents are either encouraged or required, even if the aggregates are found to be innocuous. The increased use of SCMs such as fly ash and slag provides additional safety against variations in aggregate reactivity and SCM composition, has environmental benefits, and can often reduce both upfront and maintenance costs of the structure. ACKNOWLEDGMENTS This work is supported by the Office of Naval Research and the Naval Facilities Engineering Command, Pavements Design Technical Center of Expertise. REFERENCES 1. Malvar, L.J., Cline, G.D., Burke, D.F., Rollings, R., Sherman, T.W., Greene, J. Alkali Silica Reaction Mitigation: State-of-the-Art and Recommendations. ACI Materials Journal, V., No., September-October 00, pp Malvar, L.J., Cline, G.D., Burke, D.F., Rollings, R., Sherman, T.W., Greene, J. Closure to the Discussion of Alkali Silica Reaction Mitigation: State-of-the-Art and Recommendations. ACI Materials Journal, V. 0, No., July-August 00, pp Malvar, L.J., Lenke, L.R. Efficiency of Fly Ash in Mitigating Alkali Silica Reaction Based on Chemical Composition. ACI Materials Journal, Vol., No., September-October 00, pp Lenke, L., Malvar, L.J. Alkali Silica Reaction Criteria for Accelerated Mortar Bar Tests Based on Field Performance Data. World of Coal Ash Conference, Lexington, KY, 00.. ASTM C. Standard Test Method for Potential Alkali Reactivity of Aggregates (Mortar-Bar Method). ASTM International, West Conshohocken, PA.. ASTM C 1. Standard Test Method for Determining the Potential Alkali-Silica Reactivity of Combinations of Cementitious Materials and Aggregate (Accelerated Mortar-Bar Method). ASTM International, West Conshohocken, PA.. ASTM C 1. Standard Test Method for Determination of Length Change of Concrete Due to Alkali-Silica Reaction. ASTM International, West Conshohocken, PA.. Thomas, M.D.A., Fournier, B., Folliard, K., Shehata, M., Ideker, J., Rogers, C. Performance Limits for Evaluating Supplementary Cementing Materials using Accelerated Mortar Bar Test. ACI Materials Journal, V., No., March-April 00, pp Fournier, B., Nkinamubanzi, P.-C., Chevrier, R. Comparative Field and Laboratory Investigations on the Use of Supplementary Cementing Materials (SCM) to Control Expansion Due to Alkali-Silica Reaction (ASR) in Concrete. 1th International Conference on Alkali Aggregate Reaction in Concrete, Beijing, China, October 00, pp. - (also, ibid, CANMET Report MTL 00-1).. Stokes, D., Johnston, D., Surdahl, R. The Year Concrete Prism Test for ASR Is It Worth the Wait? Transportation Systems 00 Workshop, Phoenix, AZ, 00.. Hooton, D., Rogers, C., Ramlochan, T. Preventative Measures for Alkali-Silica Reaction: The Kingston Outdoor Exposure Site for ASR After 1 Years. Report MERO-01, Materials Engineering and Research Office, Ontario Ministry of Transportation, Downsview, Ontario, CA, June 00, pp.

9 1 1. Thomas, M.D.A., Hooton, R.D., Rogers, C.A. Prevention of Damage due to Alkali-Silica Reaction (ASR) in Concrete Construction - Canadian Approach. Cement, Concrete and Aggregates, Vol. 1, No. 1, 1, pp Shrimer, F. Progress in the Evaluation of Alkali-Aggregate Reaction in Concrete Construction in the Pacific Northwest, United States and Canada. Bulletin 0 K, U.S. Department of the Interior, U.S. Geological Survey, ACI.R. Use of Fly Ash in Concrete. American Concrete Institute, Farmington Hills, MI, California Department of Transportation. Section 0 Portland Cement Concrete. Issued New Mexico Department of Transportation. Section 0: Portland Cement Concrete Mix Designs, Standard Specifications For Highway And Bridge Construction ACI 1. Building Code Requirements for Structural Concrete and Commentary. American Concrete Institute, Farmington Hills, MI, 00.