Guidelines for Concrete Mixtures Containing Supplementary Cementitious Materials to Enhance Durability of Bridge Decks (2007)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

9Introduction The first task in determining the optimum concrete mix- ture for a particular application is to define what properties of the concrete are significant. This procedure requires dif- ferentiating between what properties are not relevant, what properties must meet but not necessarily exceed a minimum level of performance, and what properties are to be maxi- mized (e.g., durability) or minimized (e.g., shrinkage). The definition of the properties (the performance under a given set of conditions) must be in both general terms (e.g., the concrete must be resistant to freezing and thawing) and in specific terms (e.g., the concrete must be capable of with- standing 500 cycles of cyclic freezing and thawing as defined by ASTM C 666, Standard Test Method for Resistance of Concrete to Rapid Freezing and Thawing, Procedure A, with a durability factor of no less than 85% at the end of testing). The general definition requires local knowledge of the serv- ice environment to which the structure will be exposed and the potential mechanisms that may lead to deterioration of the structure. The specific definition includes both a test method and desired test responses. Because the test methods used to evaluate durability are accelerated tests, developed to predict field performance over a time scale greater than actu- ally tested, determining the appropriate desired response requires a good understanding of what is actually tested and the accuracy of the simulation to real performance for a given method. While specification of minimum performance require- ments will not be unfamiliar to most engineers, this methodology requires taking the definition of desirable performance one step further. To provide a basis for com- paring the concrete mixtures to be evaluated, the minimum acceptable and desired responses are interpreted in terms of a single desirability function that summarizes the desir- ability of the performance over the full range of possible test outcomes. The objectives of Step 1 of the methodology are the gener- ation of the following: • A list of general performance objectives for the concrete mixture based on the service environment and design objectives • The selection of test methods that will provide a basis for evaluating the concrete mixtures • The definition of desirability functions to be used to numerically compare the measured responses and generate an overall desirability for each mixture. Performance Definition Process To aid in the achievement of the objectives of Step 1, users can consider a series of questions regarding the service envi- ronment of the bridge deck that is to be constructed with the goal of defining the performance characteristics required for the concrete in their specific application. This process is formalized as the flowchart in Figure S1.1. Each potential service environ- ment relates to a concrete property, a test method, and a target test response for that property. For each entry in Figure S1.1, there is a corresponding narrative description in this section of the Guidelines giving background to the entry, recommended performance requirements for each test, a general discussion of the likely influence of SCMs on the concrete with respect to each property, and some general recommendations for other mixture proportion variables, such as w/cm. The identification number given in square brackets (e.g., [D1]) in each decision box of the flowchart refers to the code placed after the heading of the subsection that discusses the background relevant to that entry. Worksheet S1.1 is provided to aid users in summarizing the information required for the experimental design process including (1) concrete properties selected as important in the chosen service environments; (2) target values for specific test methods related to those properties; (3) guidance for S T E P 1 Define Concrete Performance Requirements

10 SCM types, ranges for use, and effect on each property; and (4) other relevant mixture issues. Typical SCM ranges for use for each property are summarized in Table S1.1. The first step in using Figure S1.1 is to define the concrete parameters that are universal performance requirements for bridge decks. These parameters usually include concrete strength, workability, and finishability—properties typically required for the concrete irrespective of the particular service environment. To help define service environments and concrete charac- teristics, answers to six questions about categories of service environment/concerns are needed to determine if the bridge will be built in (1) a freezing and thawing environment, (2) a location where chemical deicers are used, (3) a coastal environment, (4) an abrasive environment, (5) an area of concern for alkali-silica reactivity (where, if so, the user is for- warded to Step 2), and (6) an area of concern for cracking. If the answer to any of these questions is “yes,” the concrete properties that are required for durability in that environ- ment and the tests to be used to assess the concrete’s per- formance relative to that type of environment are described for consideration. The user should progress through the serv- ice environment categories, considering each individually and recognizing that more than one service environment category probably exists for most bridge decks. The background information associated with each deci- sion box in Figure S1.1, found in the subsection titled “Guid- ance on Concrete Design Requirements and Appropriate Test Methods,” provides guidance on the test method selection and target values for the test results. The user should sum- marize the target values relevant to the structure to be built in the first column of Worksheet S1.1 to provide a basis for developing desirability functions. The desirability functions are an important part of the analysis to be performed in Step 4. Example desirability functions are provided in the “Guidance” subsection. These functions should be reviewed and adjusted as needed to match the target values identified for the specific project in question. The background information also discusses the influence of SCMs on each property. Table S1.1 summarizes typical usage of SCMs to produce desirable performance for each property. Also, background information and suggestions are provided in Table S1.1 regarding each SCM and its influence on the proper- ties considered. The recommendations relevant to the experiment being conducted should be recorded in Worksheet S1.1, which lists each property and provides columns for recording recom- mended ranges for each mixture parameter, such as the contents of the SCMs. A few other requirements for properties are included in Tables S1.2 through S1.7. This information should be con- sidered and also included in the relevant rows and columns of Worksheet S1.1. Any tests not deemed relevant can be crossed out of Work- sheet S1.1, leaving the tests that must be performed on each mixture in the developed experimental design matrix (as de- scribed in Step 3). After completion of Worksheet S1.1 (except for the row on ASR if applicable), the user can then proceed to Step 2. If ASR is not to be considered, the user can proceed by summarizing the values of each of the columns, which list the mixture parameters recommended to achieve the desired per- formance. The objective is to provide a basis for determining the ranges over which the experiment will be conducted by identifying the ranges of SCM contents and other mixture parameters consistent with most, if not all, of the recommen- dations. If many properties are being evaluated, it is possible that a small or even non-existent range remains after all the recommendations have been considered; the user is then free to broaden the range for testing as necessary. Because the Guidelines consider each property separately, recommendations for improving one property may conflict with the recommendations for improving another. These con- flicts may be reconciled either by including both recommenda- tions in the scope of the experiment so that the experimental process will identify the best balance or by choosing to follow one of the recommendations and maximize performance in terms of the property known to be the most important at the expense of the other. While a set of performance measures is discussed in the Guidelines based on meeting the demands of the specific service environment categories, the experimental analysis procedure is flexible and capable of differentiating and mod- eling any characteristic of the concrete if that characteristic can be evaluated with a desirability function. For example, the cost of the concrete mixture could be included as a response in the analysis process and a desirability function defined so that less expensive mixtures are more desirable. However, cost was not included as a response in the default experimental program or the hypothetical case study because the in-place cost is diffi- cult to predict and is often a secondary concern compared to durability-related performance. Example from Hypothetical Case Study The hypothetical case study (Appendix A of NCHRP Web- Only Document 110) was based on a bridge deck application in a northern climate. For this hypothetical environment, the steps outlined by Figure S1.1 were used to characterize the universal design requirements and evaluate issues relevant to a freezing climate subjected to chemical deicers, where crack- ing was a concern. This environment was assumed to be nei- ther coastal nor abrasive. Table S1.8 presents an example of Worksheet S1.1 com- pleted for the hypothetical case study according to the

guidance provided in this chapter. The recommended test- ing program based on the service environment of the hypo- thetical case study is summarized on this worksheet, which lists the properties of interest, the test methods to measure each property, and optimum target values that will be used to develop the desirability functions. Categories that were not applicable to the hypothetical case study environment were struck out. The recommended ranges of SCM contents expected to produce desirable performance were collected for each property and the columns were summarized in the row at the bottom of the worksheet. This summary will serve as a reference point for selecting the ranges for testing over which each material may be optimized. Guidance on Concrete Design Requirements and Appropriate Test Methods The following subsections discuss concrete performance requirements that are mentioned in Figure S1.1. Universal Performance Requirements Nearly all concrete construction projects involve perform- ance requirements that characterize the strength of the mate- rial and evaluate the influence of the concrete on the ease of construction. Compressive Strength [D1] Compressive strength is almost always specified by the designers of the bridge. It is measured by AASHTO T 22 (ASTM C 39), Standard Test Method for Compressive Strength of Cylindrical Concrete Specimens. In this method, a compressive axial load is applied at a specified rate to molded concrete cylinders. The peak load applied divided by the cross-sectional area gives the compressive strength. Desirability Function for Compressive Strength. The desired compressive strength will depend on the design spec- ifications for the structure, which will vary depending on the project. The compressive strength must meet the minimum design criteria. However, it is often disadvantageous for the compressive strength to be much higher than required because the accompanying higher elastic modulus may ren- der the concrete more susceptible to cracking and higher strength is usually associated with increased cost. Consider- ing a maximum strength when defining the compressive strength desirability function to minimize cracking potential is particularly important if appropriate limits are not set for the elastic modulus or cracking tendency. The cracking ten- dency can be measured with ASTM C 1581, Standard Test Method for Determining Age of Cracking and Induced Tensile Stress Characteristics of Mortar and Concrete Under Restrained Shrinkage (also known as the restrained ring shrinkage test). Figure S1.2 gives a suggested 7-day desirability function for a concrete expected to meet a specified 28-day strength of 5000 psi (34.5 MPa). The strength of the mixture is consid- ered to need no improvement if the strength is between 3500 and 5500 psi (24.1 and 37.9 MPa) and is therefore assigned a desirability of 1 over that range. This function penalizes mix- tures if the strength is greater than 5500 psi (37.9 MPa) at this age. Penalizing mixtures with high early strengths is a strat- egy for minimizing cracking because the stresses that cause most early thermal and drying shrinkage–related deck crack- ing are related to the stiffness (elastic modulus) of the concrete at these early ages. A reduction in the desirability of mixtures with high early strength is especially important for bridge deck designs that have high restraint conditions if other tests to evaluate early-age cracking, such as elastic mod- ulus or cracking tendency, are not included in the experi- mental program. Avoiding very high-strength mixtures also will make the concrete more economical, a factor not directly considered in this example. Figure S1.3 shows an example where the target average 28-day compressive strength is at least 5500 psi (37.9 MPa) and desirability of 1 would be assigned to the mixture for this response if the strength is between 5500 and 8500 psi (37.9 and 58.6 MPa). This target average may be selected for a spec- ified design strength of 5000 psi (34.5 MPa), because it will ensure that even with some variation, the design strength will be exceeded. Because mixtures with high strengths at later ages will also likely have high early strengths, which may be related to early cracking, a penalty was assigned to high- strength mixtures. Figure S1.4 gives a similar desirability function for 56-day strength. Evaluating the concrete per- formance at this later age, if consistent with construction schedules, may be more appropriate than 28 days because of the decreased rate of hydration and strength gain typical with many SCM-based mixtures. Effects of SCMs on Compressive Strength Fly Ash. The interactions between cement and fly ash are complex, and their effect on compressive strength is not always predictable. Typically, mixtures with Class F fly ashes develop strength more slowly than comparable mixtures con- taining only portland cement. Although 28-day strengths may be lower for concrete containing fly ash, particularly Class F fly ash, fly ash continues to hydrate over time, and the long-term strength of concrete with fly ash typically exceeds that of a comparable portland cement concrete. Silica fume can be used in combination with fly ash to increase the rate of strength gain at early ages. 11

Class C (high calcium) fly ashes show a higher rate of reaction at early ages than Class F fly ashes and typically do not result in a significant difference in compressive strength from pure portland cement concrete at replacements up to 30%. Therefore, they can be proportioned on a one-to-one replacement basis for portland cement without a negative effect on early-age strength (3) and they can achieve 28-day strengths that are comparable to portland cement concrete. At later ages (beyond 28 days), some Class C fly ashes may not show the strength gain expected from Class F fly ashes (4). GGBFS. Fernandez and Malhotra (5) found that lower strengths were obtained through 91 days when GGBFS was used, with a possible lesser effect at lower (25% vs. 50%) dosages. However, slag is categorized in AASHTO M 302 (ASTM C 989) by grades, which is a measure of the reactiv- ity, defined as the percentage of strength achieved in mortar cubes made with 50-50 portland cement–slag combinations versus only portland cement. American Concrete Institute (ACI) Committee 233 (6) reports that Grade 120 slag can have reduced strength at early ages (1 to 3 days) and increased strength at 7 days and beyond. A Grade 100 slag will typically have lower strength until 21 days, and equal or greater strength thereafter. A trend in recent years has been to grind slag to sufficiently high fineness that little reduction in the rate of hydration occurs when compared with cement. Cement manufacturers sometimes blend portland cement and slag or fly ash at low levels while conforming to ASTM C 150 cement requirements. Silica Fume. The effect of silica fume is most pro- nounced on strength between 3 and 28 days, after which its influence on strength is minimal. At conventional dosage rates, silica fume–containing concrete always has higher strength than ordinary portland cement concrete at a com- parable w/cm (7). Class N Pozzolans (Metakaolin). Concrete with an addi- tion of 7% of metakaolin was found to have higher early-age strengths than portland cement concrete and concrete with 7% silica fume addition. At 28 days, metakaolin-containing concrete was found to be 10% stronger than portland cement concrete and 5% to 10% weaker than an equivalent mixture with silica fume. In one study, between 90 and 365 days, metakaolin-containing concrete gained strength, whereas the strength of silica fume–containing concrete decreased slightly (8). Ding and Li (9) found compressive strength of concrete with 5% to 15% replacement of metakaolin at 3 to 65 days increased from 4% to 53% over plain portland cement concrete. Calcined clay additions were found to exhibit slower early-age strength gain but higher 28-day strength than portland cement concrete (10). Tensile/Flexural Strength [D2] The tensile strength of concrete is theoretically defined as the peak tensile load divided by the cross-sectional area. However, because of the difficulty in applying direct tension, it is almost never tested in this way. Instead, the tensile per- formance is approximated using splitting tension or flexural testing. Flexural strength is the peak tensile stress developed in a beam assuming elastic beam theory. Flexural strength is usually specified for highway or airfield pavements and occa- sionally for bridge decks. Test Methods. Three types of tests are related to tensile strength of the concrete: (1) direct tension, (2) flexure, and (3) splitting tension. Only flexure and splitting tension will be discussed here. Flexural strength is measured by two methods. The most common method is AASHTO T 97 (ASTM C 78), Test Method for Flexural Strength of Concrete (Using Simple Beam with Third-Point Loading). In this method, a concrete beam, 6×6×21 in. (150×150×535 mm) in size, is supported at the ends and loaded at the third points until failure. The modulus of rupture is calculated as the stress at the extreme fiber. The second method is AASHTO T 177 (ASTM C 293), Test Methods for Flexural Strength of Concrete (Using Sim- ple Beam with Center-Point Loading), which loads the beam at its center point. This method is rarely used because its re- sults are more variable (11). Because the specimens are quite large and heavy for both tests, neither test method for flexural strength is convenient. The test results may be influenced by the concrete moisture content at the time of testing. Also, the modulus-of-rupture calculations are based on elastic beam theory, which is not completely accurate for these conditions and overestimates the tensile strength of concrete. Neville (11) cites a reference by Raphael (12) that states the “correct” tensile strength is three-quarters of the calculated modulus of rupture. Tensile strength can be measured indirectly using the split- ting tensile test, AASHTO T 198 (ASTM C 496), Test Method for Splitting Tensile Strength of Cylindrical Concrete Speci- mens. In this test, a concrete cylinder or core is compressed parallel to its axis, resulting in a splitting failure. The tensile strength is calculated from the peak compressive load. Neville (11) reports that the results are believed to be closer to the actual tensile strength of concrete than those measured in flexural strength testing. Also, the tensile splitting test is less variable and easier to conduct. However, if splitting tensile strength results are to be used to show compliance with flex- ural strength specifications, a correlation should be estab- lished for the mixture in question. Desirability Function for Tensile/Flexural Strength. For 5000 psi (34.5 MPa) compressive strength concrete, the 12

tensile strength measured with a splitting tensile test is ap- proximately 500 psi (3.5 MPa), while for 6000 psi (41.4 MPa) concrete, the tensile strength is close to 600 psi (4.1 MPa) (11). The expected corresponding moduli of rupture for mix- tures with these compressive strengths are 530 and 580 psi (3.7 and 4.0 MPa), respectively. An example of a desirability function for modulus of rupture is shown in Figure S1.5 based on a specified minimum modulus of rupture of 530 psi (3.7 MPa) and an assumed target average modulus of rupture of 580 psi (4.0 MPa) or greater. Effect of SCMs on Tensile/Flexural Strength. Because tensile and flexural strength are typically linked to compres- sive strength, the effects of SCMs on these properties are generally similar. Fly Ash. If fly ash is substituted on a one-to-one basis by weight or volume for portland cement, flexural strengths are usually lower until about 3 months of age but may be higher beyond this age (13). GGBFS. GGBFS typically increases modulus of rupture because of increased denseness of paste and improved bond between aggregates and paste (6). Silica Fume. Silica fume can result in higher flexural strength, compared with plain portland cement concrete (7) likely due to the increase in interfacial bond between paste and aggregates. Class N Pozzolans (Metakaolin). Taylor and Burg (8) report little difference in splitting tensile strength results between portland cement, metakaolin-containing, and silica fume–containing concrete. However, Caldarone et al. (14) report 28% to 36% increases in flexural strength over port- land cement concrete at 7 to 90 days when 10% metakaolin was added. Workability and Finishability [D3] “Workability” is a qualitative or subjective term that has been defined by various sources, according to Neville (11), as (1) “the amount of useful internal work necessary to produce full compaction,” (2) “property determining the effort re- quired to manipulate a freshly mixed quantity of concrete with minimal loss of homogeneity,” and (3) “that property of freshly mixed concrete or mortar which determines the ease and homogeneity with which it can be mixed, placed, con- solidated, and finished.” No test measures workability directly, but there are tests that measure properties related to workability. Some of these include the slump test (AASHTO T 119/ASTM C 143), Test Method for Slump of Hydraulic-Cement Concrete, which measures the tendency of the concrete to flow under its own weight and is often used as a measure of consistency; a slump loss test, which is simply the difference in two measurements of slump taken over an interval of time; and time of setting as measured by AASHTO T 197 (ASTM C 403), Test Method for Time of Setting of Concrete Mixtures by Penetration Resistance, which measures the setting time of a mortar sieved from concrete. No direct measurement of finishability exists, but a relative, qualitative assessment (non-standard test) can be useful in comparing concrete mixtures. Slump. During the slump test, concrete is placed in three layers into a conical mold held against a flat surface. Each concrete layer is rodded before the introduction of the next layer. The concrete is struck off, and the cone is slowly lifted, allowing the concrete to drop or slump when unsupported. The decrease in height is measured and is called the slump. When slump is too low, workability and the ability to con- solidate the concrete are likely to be poor. When slump is too high, segregation of the paste from the aggregate is possible. Too high of a slump can also indicate that the w/cm is too high, or if superplasticizer is used, that the dosage is high. Desirability Functions for Slump. The target value for slump at placement varies depending on the placing require- ments and procedures and on mixture proportions of the concrete, especially the presence of chemical admixtures, such as superplasticizer or high-range water reducers (HRWRs). For cast-in-place bridge decks, the target value for slump is typically 5 in. (125 mm) at placement. If HRWRs are used, the slump at the start of placement can be as high as 8 in. (200 mm) without problems; however, if an appropriate amount of the HRWR is not used, an 8-in. (200-mm) slump will likely be achieved through the use of excess water, result- ing in low strength or segregation. A desirability function for slump, based on concrete con- taining HRWR, is given in Figure S1.6. This function suggests any slump in the range of 4 to 8 in. (100 to 200 mm) is ac- ceptable, but slumps higher or lower than this range are penalized severely to ensure that concrete is workable enough to allow easy consolidation but prevent segregation. HRWRs are typically used for water reduction and higher workability. The admixture coats the surfaces of the cementi- tious particles and, by electrostatic repulsion and hindrance, helps the particles stay dispersed in water. Decreasing the need for additional water for flow and workability generally increases the quality of the concrete. The permeability of the concrete is decreased and strength is increased because the pore space occupied by water is decreased. When silica fume is used, HRWRs are needed to decrease water demand and increase silica fume dispersion. If the slump is outside the desired range, the mixture can be modified to achieve the 13

desired slump by increasing or decreasing the amount of HRWR, usually without adverse effects. Slump could be highly weighted (important) in a test program if the method- ology is used for evaluating the effect of admixtures on slump and other plastic properties. Effect of SCMs on Workability and Slump Fly Ash. Because fly ash particles are spherical in shape, they have generally been shown to permit a reduction in water con- tent for a given workability (4). Also, when fly ash is substituted for cement by weight, the volume of the paste relative to total concrete volume is increased because the density of fly ash is less than that of cement and results in an increase in plasticity and cohesiveness (4). Other studies summarized in Malhotra and Ramezanianpour (3) indicate that fly ash particles that are less than 45 μm can reduce water requirements. In general, fly ash substitutions increase slump; however, there are data to the contrary as well. Apparently, fly ashes with higher carbon contents and coarser particle sizes can increase water and air- entraining agent requirements. Because fly ash behavior in con- crete is complex, the effect of the particular fly ash can be determined only by testing the specific concrete mixture. GGBFS. Several studies cited in ACI Committee 233 (6) indicate slump increases with increasing GGBFS content. Silica Fume. ACI Committee 234 (7) summarized the effect of silica fume on water demand and workability and suggests that the water demand of the concrete increases when silica fume content increases. An HRWR is generally used to optimize the properties of silica fume concrete. Silica fume–containing concrete exhibits increased cohesiveness and decreased tendency to segregate compared with ordinary portland cement concrete. ACI Committee 234 (7) recom- mends an initial slump of silica fume concretes 2 in. (50 mm) above that required for ordinary portland cement concretes to maintain the same apparent workability. Class N Pozzolans (Metakaolin). Less HRWR is required to achieve a similar workability for metakaolin-containing con- crete than for silica fume–containing concrete at equivalent dosages (8, 9). Caldarone et al. (14) quantified the decrease in HRWR required for equivalent slump as 25% to 35%. Slump Loss. Concrete loses slump over time because of many factors including absorption of mix water by aggre- gates, loss of water by evaporation, initial chemical reactions of the cementitious materials, and interactions with chemical admixtures and temperature. SCMs can influence the rate of slump loss. An additional measurement of the slump after a designated amount of time is of value, as it gives an indica- tion of slump at the time of placement, for example, after the concrete has been batched and transported to the job site. Desirability Function for Slump Loss. Slump loss is measured as a difference in slump between an initial test and after a specified time chosen to represent job conditions. The target value for slump 45 minutes after cement and water are mixed might be 4 in. (100 mm). While a high slump after 45 minutes is generally desirable, it may also indicate the con- crete will experience high setting times. If setting time tests are not conducted as part of the experimental program, decreasing the upper limit of acceptable slump at 45 minutes should be considered. The desirability function for slump loss is shown in Figure S1.7. This function can be interpreted to suggest that a lower slump loss is always best, such that slump losses less than 2 in. (50 mm) lead to a slightly reduced desirability. Slump losses more than 2 in. (50 mm) are penalized more severely and mixtures with a slump loss more than 6 in. (150 mm) would be considered unacceptable for a bridge deck application. Effect of SCMs on Slump Loss. Slump loss is typically governed by temperature effects, cement chemistry, and the water-reducing admixtures. Fly Ash. A study by Ravina (15) indicates that at least some Class F fly ashes reduce slump loss. Slump retention is greater when the cement replacement percentage is increased (up to 40%). This phenomenon may reflect a decreased re- liance on HRWR admixtures in fly ash mixtures. Some Class C fly ashes react quickly with water and may have a negative effect on slump retention (16). GGBFS. There is little information on the effect of GGBFS on slump loss (6). Some studies reported a reduction in slump loss, whereas other studies showed no difference in slump loss with GGBFS. Silica Fume. Experience suggests silica fume does not usually affect slump loss (7). However, silica fume is usually used with chemical admixtures, resulting in changes in slump loss characteristics. Class N Pozzolans (Metakaolin). Taylor and Burg (8) found slightly higher slump losses associated with concrete containing metakaolin or silica fume compared with plain portland cement concrete. Time of Setting. Time of setting is measured by AASHTO T 197 (ASTM C 403), Standard Method of Test for Time of Setting of Concrete Mixtures by Penetration Resis- tance. In this method, the mortar fraction is sieved from the concrete test sample, and penetration resistance is measured as a function of time. As defined by the test method, initial setting is attained when the penetration resistance reaches 14

500 psi (3.5 MPa), and final setting is attained at 4000 psi (27.6 MPa). Desirability Function for Time of Setting. The time of setting indicates the working time of the concrete. Because time of setting is affected by temperature, it may be desirable to evaluate the time of setting at different temperatures. An example of a desirability function for time of initial set- ting, developed for cast-in-place bridge decks, is shown in Figure S1.8. According to this function, concrete that reaches initial setting in less than 2 hours (regardless of other perform- ance) is not acceptable and a time of initial setting between 3 and 8 hours cannot be improved. The desirability then decreases for mixtures with increasing times of initial setting up to 24 hours because finishing operations will be affected and the risk of plastic and settlement cracking will increase. Effect of SCMs on Time of Setting Fly Ash. In general, both Class F and Class C fly ashes have been found to extend the time of setting of concrete (4). Detwiler et al. (17) state that fly ash can retard setting time by approximately 20 minutes per 10% addition of fly ash. GGBFS. In general, GGBFS increases the setting time of portland cement concrete mixtures. The amount of GGBFS, w/cm, and concrete temperature all influence the setting time. Typical extensions of setting time are reported to be 0.5 to 1 hour at 73°F, with little change at temperatures above 85°F (6). At colder temperatures, the increase in setting time may be significant. Hooton (18) found that with GGBFS replacement levels of 50%, setting time can be extended 1 to 2 hours at low temperatures (less than 15°C). This extension would require delayed finishing and extended curing periods. Fernandez and Malhotra (5) found that, at a w/cm of 0.55, final setting was more affected than initial setting when GGBFS was used in the mixture. Detwiler et al. (17) state that delays of 10 to 20 minutes per 10% addition of GGBFS- blended cement can be expected. Silica Fume. ACI Committee 234 (7) indicates that use of silica fume by itself should not affect setting time. However, setting time may be affected if chemical admixtures are used in conjunction with silica fume. Class N Pozzolans (Metakaolin). Caldarone et al. (14) found the initial time of setting of metakaolin-containing concrete was slightly less than portland cement concrete and comparable to concrete with an equivalent dosage of silica fume. However, Taylor and Burg (8) report comparable time of setting for all mixtures. Finishability. “Finishability” is a qualitative term that describes the ease of being able to screed and finish concrete. Concrete bridge decks are typically screeded to the desired thickness and finished with an Astroturf or burlap drag. Areas near edges where the mechanical screeds cannot reach are usually floated by hand. Finishability can include characteristics such as (1) stickiness, whereby the concrete sticks to finishing tools (usually because of the presence of silica fume), or creaminess, whereby the con- crete flows well and finishes easily; (2) segregation, whereby the mortar fraction separates from the aggregates; (3) harshness, whereby there is not enough paste or the aggregate is poorly shaped or graded; and (4) proneness to tearing, whereby the surface of the concrete tears when manipulated. There is no standard test method to assess finishability. During the hypothetical case study, a non-standard qualita- tive assessment was performed in which a concrete sample was screeded, floated, and broomed by three experienced per- sonnel. A numerical rating of 1 to 5 was then assigned to the concrete by each of the three individuals with respect to the stickiness, segregation, harshness, and tear resistance of the concrete. The average ratings in each category were added together for a total score. Lower scores indicate worse per- formance and higher scores indicate better performance. This test was conducted with consistent concrete temperature, air temperature, and relative humidity, and each mixture was evaluated by the same personnel. Desirability Function for Finishability. The example shown in Figure S1.9 is based on the finishability test method defined above. Four categories were assessed, so the finisha- bility scores ranged from 4 to 20. Because this test is qualita- tive in nature and somewhat subjective, the desirability function was designed to range from 0.5 to 1, and a relatively generous gradient was chosen so that, while higher finishabil- ity ratings receive a higher desirability, all mixtures with a fin- ishability rating of 12 or higher are assigned a desirability greater than 0.95. Mixtures with a finishability rating of 12 or less were considered to be less desirable; therefore, the nega- tive slope of the function increases below this value. This test was not deemed definitive enough to eliminate any mixture from consideration, and therefore the lowest desirability assigned is 0.5 and not 0. If another method for testing finishability is used, a new desirability function will be needed. Obviously, the rating depends on the number of finishing characteristics that are chosen for rating. For example, if three categories are chosen, the rating scale could range from 3 to 15. Effect of SCMs on Finishability Fly Ash. Several issues are relevant to finishing concrete containing fly ash. Because some fly ash–containing concretes can take longer to set, such concrete should be finished at a later time to avoid trapping bleed water under the top surface 15

and causing a plane of weakness (4). Also, very light fly ash particles can float to the surface and cause an unacceptable appearance. Typically this occurrence is not a problem on bridge decks that are not hard-trowel finished. Also, sticki- ness may result if the fly ash contains too much fine material. GGBFS. The influence of GGBFS on finishability is un- clear. Hooton (18) states that the finer particle size of GGBFS makes concrete more cohesive and easier to place, finish, and compact. Balogh (19) reports that improved pumping and placing properties are because the smooth and dense surfaces of slag particles result in a more fluid paste with more work- ability. Concrete containing slag has also been described as chalky. Silica Fume. Silica fume, particularly at dosages in the range of 10% to 20%, has been shown to greatly decrease bleed water and increase the stickiness of the concrete (7). Class N Pozzolans. Ding and Li (9) suggest that metakaolin-containing concrete may have better finishing characteristics than silica fume–containing concrete because of its lower requirement for HRWR. Caldarone et al. (14) describe metakaolin-containing concrete as less sticky than silica fume–containing concrete. The time of finishing of metakaolin-containing concrete may “require more care,” because of the very small amount of bleeding (8). Barger et al. (10) suggest the mass replacements of lower density (than portland cement) calcined clay increases paste volume and results in better flow characteristics and workability. Metakaolin mixtures are typically creamier than similar silica fume mixtures. Freezing and Thawing Climates [F1] Freezing and thawing environments adversely influence the durability of concrete bridge decks. Freezing water within critically saturated concrete generates internal pres- sures that may produce damage in either paste or aggregate phases within the concrete. For the paste, resistance to the damaging mechanisms resulting from freezing water is achieved through the use of air-entraining admixtures that stabilize and help distribute air bubbles within the concrete. Because the air void system is critical to performance, it is often independently characterized microscopically. Direct testing of the freezing and thawing response of the concrete is often also conducted to confirm the effectiveness of the air void system. The resistance of aggregate to freezing and thawing damage depends on the aggregate’s resistance to sat- uration by moisture and subsequent expansion, which is a function of pore structure and aggregate strength. This resistance is typically evaluated directly in concrete speci- mens exposed to freezing. In addition to providing a quality air void system, the gen- eral recommendations listed in Table S1.2 can be taken to maximize the freezing and thawing resistance of concrete (24). The importance of designing a concrete mixture to be resist- ant to freezing and thawing depends on the environmental conditions to which the concrete will be exposed. For this pro- gram, moderate exposure is defined as 3 to 50 freezing and thawing cycles per year, and severe exposure is defined as more than 50 cycles per year (20). Local weather would dictate the emphasis placed on this property of the concrete. Air Void System Parameters [F2] Air voids in concrete are typically classified as either en- trapped or entrained air. Entrapped air occurs unintentionally in concrete as a by-product of the mixing and placing processes. Entrained air, on the other hand, is intentionally introduced in the concrete to provide resistance to freezing and thawing and to salt scaling. Entrained air is achieved through the use of chemical admixtures that stabilize air bub- bles introduced into the concrete during mixing. The air- entraining admixtures (AEA) are electrically charged and thus produce very small, spherical bubbles that are prevented from coalescing and breaking. These bubbles are small, with the majority of the bubbles between 0.0004 and 0.0039 in. (10 and 100 µm) in diameter, and have a reduced likelihood of rising to the concrete surface. By comparison, entrapped air voids are larger and are typically larger than 0.0394 in. (1 mm) in diameter (21). Entrained air increases concrete resistance to freezing and thawing damage by providing local release for the hydraulic and osmotic pressure produced when water within the con- crete expands before and during freezing (water expands about 9% just before freezing). Water moves through the concrete into these voids, and the air voids must be closely spaced so that the pressure can be released before freezing occurs. The critical distance that water can be expected to move without causing freezing damage in typical paste is approximately 0.010 in. (0.254 mm) (22). Therefore, the effectiveness of the entrained air at providing freezing and thawing resistance to the concrete is governed by the volume and spacing of the air bubbles within the concrete. While air is entrained in concrete primarily to resist damage by freezing and thawing, it also has beneficial ramifications for water demand, bleeding reduction, increased slump and work- ability, and some resistance to internally driven expansion mechanisms such as ASR. Entrained air has negative ramifica- tions for strength (compressive strength is typically reduced 2% to 6% for each percentage of air) and elastic modulus (21). Test Methods. The quantity of air voids can be measured when the concrete is plastic, while the characteristics of the 16

air void system are assessed after the concrete has hardened. The two primary methods for determining the plastic air con- tent of concrete are AASHTO T 152 (ASTM C 231), Standard Method for Test of Air Content of Freshly Mixed Concrete by the Pressure Method, and AASHTO T 196 (ASTM C 173), Standard Method for Test of Air Content of Freshly Mixed Concrete by the Volumetric Method. Method T 152 is based on the principle that the volume of air is inversely related to pressure. A known pressure is applied to concrete, and the amount of volume reduction is measured and used to deter- mine the air content. Method T 196 is typically used with lightweight aggregate where the pressure method does not apply. A known volume of consolidated concrete and water is agitated, and the final combined volume is used to deter- mine the air content. The method for characterizing the air voids in hardened concrete is ASTM C 457, Standard Test Method for Micro- scopical Determination of Parameters of the Air-Void System in Hardened Concrete. In this method, a sample of concrete is cut and prepared to expose a smooth plane of concrete for examination under a microscope. The specimen is moved in a predefined pattern, either along parallel lines or to points in a grid under the microscope, and the operator identifies the air voids, paste, and aggregate constituents that are observed. Based on limited observations, two parameters are deter- mined statistically to characterize the full air void system. One parameter is the specific surface area that is defined as the ratio of the surface area of the air voids to the total volume of air within the sample. This parameter does not describe the size distribution of all the air voids, which can range from 0.0004 to 1 in. (10 to 25,400 µm), but is analogous to an av- erage for all the air voids. The other parameter is the spacing factor, which is an index related to the maximum distance from a point in the cement paste to the nearest air void. This parameter is calculated based on the observations made on the sample surface and assumptions about the relative geom- etry of the air voids (23). Desirability Functions for Air Content. The air content of the concrete should be balanced such that sufficient air is present to provide freezing and thawing resistance without unduly reducing concrete strength. Because the air voids pro- vide relief for freezing water primarily from the cement paste, the amount of air needed in the concrete to provide freezing damage protection will be higher in mixtures containing a larger volume of paste. The suggested desirability function is based on ACI recommendations for 0.75 to 1 in. (19 to 25 mm) maximum nominal aggregate size (24). The recom- mendations for air parameters in hardened concrete are based on correlations with laboratory and field performance (23). To limit the total number of desirability functions (and the weight or importance of the air void system measurement on the total response), it is recommended that only one function be included to characterize the air void parameters. The spac- ing factor is commonly believed to be the most important value in determining freezing and thawing performance be- cause there is a critical distance that water can move as it freezes without causing damage. The spacing factor is analo- gous to the average distance between any point in the concrete and the nearest entrained air bubble. Therefore, concretes with spacing factors greater than the critical dis- tance are susceptible to freezing and thawing deterioration. Spacing factor, however, depends on the total air content and is largely determined by the amount of air-entraining admix- ture added to the mixture, which is adjusted to accommodate the influence of SCMs or other changes in the mixture on the air void system. Specific surface, on the other hand, is nor- malized to the total volume of air observed in the mixture and, therefore, it may be more indicative of influences of the mixture components on the ability of the mixture to entrain air. It is generally true that spacing factors less than 0.008 in. and specific surfaces of greater than 600 in.2/in.3 will provide good freezing and thawing resistance. Strength also may play a role in freezing and thawing resistance. Desirability func- tions for air content in plastic and hardened concrete subject to severe exposures are shown in Figure S1.10. These func- tions are designed such that mixtures with air contents between 5.5% and 10% in severe exposures and between 4.5% and 10% in moderate exposures can not be improved with respect to this response. Air contents below these ranges may result in inadequate freezing and thawing protection, while air contents above this range may have a significant detrimental effect on strength or surface quality and, there- fore, are assigned lower desirabilities. When the w/cm is less than 0.38 and HRWR is used, slightly higher spacing factors will not negatively impact freeze/thaw durability. Desirability functions for spacing factor for con- cretes with and without HRWR are shown in Figure S1.11. These functions suggest that it is very important to have a spacing factor less than 0.010 and 0.008 in. for mixtures with and without HRWR, respectively. After a steep decline that gives some credit to mixtures with spacing factors only slightly higher than these thresholds, mixtures with larger spacing fac- tors are considered unusable. Because there is no advantage to having entrained air bubbles closer together than the thresh- old value, all mixtures above the threshold are assigned the same desirability value of 1. A desirability function for specific surface area is given in Figure S1.12. This function assigns higher desirability to concrete mixtures with increasing specific surface area, up to 800 in.2/in3. The slope of the increase in desirability with specific surface area is lower as the specific surface area in- creases because the benefits of those increases become less significant. 17

Effect of SCMs on Air Content. The composition of the SCMs may have a significant influence on the effectiveness of air-entraining admixtures and the stability of the air voids in the plastic concrete. Therefore, trial batches are essential for selecting mixtures with acceptable air void systems. In gen- eral, materials with smaller particle sizes tend to increase the amount of AEA required for a given air content. Fly Ash. It is difficult to generalize the effect of fly ash on air entrainment. However, air content for a given dosage of air-entraining agent typically decreases with an increase of the loss on ignition of the fly ash, which is a measure of its carbon content (4). GGBFS. Slag, if finely ground, may reduce air content for a given dosage of air-entraining agent (21). Silica Fume. Silica fume can be expected to reduce air content for a given dosage of air-entraining agent (7). Class N Pozzolans (Metakaolin). Taylor and Burg (8) report that more AEA was required with a 7% addition of metakaolin than for plain portland cement concrete. Resistance to Freezing and Thawing [F3] While freezing of moisture in the cementitious paste leads to deterioration, another potential source of freezing damage is moisture within the aggregates. Aggregates absorb water, and, when contained in freezing concrete, some aggregates are prone to internal cracking and expansion that damages surrounding paste. This damage may take the form of popouts, which are small areas spalled out of the concrete when the aggregates are near the concrete surface, or overall expansion if the aggregates are deeper within the concrete. Evaluation of the influence of aggregate on freezing and thawing resistance is typically performed directly on concrete exposed to freezing and thawing conditions. More details on the selection of aggregates are included in Step 2. Test Methods. The total effect of all the factors on the concrete freezing and thawing resistance may be evaluated by directly simulating cyclic freezing in laboratory conditions using one of two methods. The first and most common method is AASHTO T 161 (ASTM C 666), Standard Method for Test of Resistance of Concrete to Rapid Freezing and Thawing, in which concrete specimens are subjected to cyclic freezing using a controlled test chamber. The second method, ASTM C 671, Test Method for Critical Dilation of Concrete Specimens Subjected to Freezing, determines the magnitude of the critical dilation of a continuously wet concrete speci- men sample during a single freezing event. When the focus of interest is specifically the aggregates, this method may be used as laid out in ASTM C 682, Practice for Evaluation of Coarse Aggregate in Air-Entrained Concrete by Critical Dilation Procedures. (Both ASTM C 671 and C 682 were withdrawn in 2003.) AASHTO T 161 is the most common method. In this test, concrete specimens between 3 and 5 in. (75 and 125 mm) high and wide and 11 to 16 in. (279 to 406 mm) long are subjected to 300 cycles between 40°F and 0°F (4°C and −18°C), each of which takes between 2 and 5 hours (testing can also be extended to 500 or more cycles if a more rigorous test is desired). AASHTO T 161 includes two procedures: Procedure A, where freezing and thawing take place with the specimens submerged in water, and Procedure B, where the freezing is achieved in air, and the thawing occurs in water. The deterioration of the concrete during the cycling is measured at least every 36 cycles on thawed specimens according to ASTM C 215, Standard Test Method for Funda- mental Transverse, Longitudinal and Torsional Resonant Frequencies of Concrete Specimens. The fundamental trans- verse frequency is used to calculate the relative dynamic mod- ulus of elasticity of the specimen, Pc, after c cycles, which is given by where n = frequency at 0 cycles and n1 = frequency after c cycles. Internal damage occurring in the specimen will reduce the dynamic modulus. The durability factor at the end of the test is equal to Pc unless the test was stopped prematurely. Two additional characterizations of the specimens also are made to provide more information about the effect of the freezing and thawing cycles on the concrete. Tracking the weight loss during the test provides information about the sur- face damage such as would be caused by surface popouts that may occur without reducing the relative dynamic modulus. In addition, measuring the length of the specimens gives an indi- cation of the influence of freezing aggregates and internal microcracking (21). For highly durable mixtures containing quality air void systems, 500 cycles or more may be necessary to differentiate concretes of varying composition. Some criticisms of the AASHTO T 161 test include (1) the rate at which freezing occurs is more severe than typical en- vironmental conditions, (2) the rigidity of the container in which the specimens are held during freezing may increase or relieve pressure resulting from freezing depending on its stiffness, and (3) the concretes exhibiting intermediate per- formance show greater variability than those performing well or poorly (25). However, despite the theoretical consid- erations, this accelerated test gives a direct measure of the performance of concrete under freezing conditions and is useful for making relative comparisons between concrete mixtures. P n n c = × 1 2 2 100 18

Desirability Functions for Freezing and Thawing Resis- tance. A durability factor of 100% corresponds to no decrease in relative dynamic modulus, suggesting no damage has occurred; performance that is as close to this level as possi- ble is desired. This performance level is reflected in the sets of example desirability functions that are given for the durability factor after 300 and 500 cycles in Figures S1.13 and S1.14 for severe and moderate environments, respectively. For severe environments (Figure S1.13), a durability factor of 100% is assigned a value of 1, with a gentle decrease in desirability to 0.9 for lower durability factors of 90% after 300 cycles and 80% after 500 cycles. Below these values the desirability decreases more steeply with decreasing durability factor. For durability factors lower than 50% after 300 cycles and 40% after 500 cycles, the performance is considered unacceptable and assigned a desirability of 0. A similar but somewhat more forgiving shape of the desirability function is assumed for moderate environments (Figure S1.14) because of the less severe exposure environment. Also, the AASHTO T 161 method provides for only 14 days of moist curing before cycling is initiated. Therefore, con- cretes containing SCMs may compare unfavorably to those containing portland cement only because the strength gain of the SCM mixtures is likely to occur more slowly. As a result, at least 28 days of curing prior to testing is recommended. Effect of SCMs on Freezing and Thawing Resistance. The ability of concrete to resist freezing and thawing is largely governed by the amount and quality of the air void system within the concrete and by the aggregate behavior during freezing. For comparable air void systems, age, paste maturity, moisture content, and aggregate soundness, concretes con- taining SCMs will perform similarly to those without SCMs (4, 6, 7, 8, 10). However, SCMs can adversely affect the air void parameters; therefore, testing of SCM mixtures is suggested. Scaling Resistance [S1] The use of chemical deicers on bridges and pavements in- creases the occurrence of scaling due to freezing and thawing. The primary mechanism causing scaling is physical (24); it involves osmotic and hydraulic pressures in the paste portion of the concrete. Osmotic pressure develops in the surface re- gion of the concrete as water moves to the concrete surface layers to equalize the concentration gradient of salt. When freezing takes place at the surface, hydraulic pressure is also generated. There is also the possibility that in some areas de- icing agents increase the number of freeze-thaw cycles (11). The deterioration mechanism is strongly dependent on the saturation of the concrete, the type of deicer, and the concen- tration gradient of the deicer (24, 26). Other mixture consid- erations for salt scaling resistance are listed in Table S1.4. Test Method. ASTM C 672, Standard Test Method for Scaling Resistance of Concrete Surfaces Exposed to Deicing Chemicals, was first introduced in 1971 (26). The method consists of fabricating concrete of a desired proportion into specimens of at least 72 in.2 (465 cm2) in area and 3 in. (75 mm) deep. The concrete must be finished with three passes of a wood strike-off board, with a final surface treat- ment of a medium-stiff brush or other finish if desired. The specimens are removed from molds at 20 to 24 hours, cured in moist storage for 14 days, and then stored in air at 45% to 55% relative humidity for an additional 14 days. The method states that if concretes of differing rates of strength gain are compared, the specimens should be maintained in moist storage until a desired strength level is obtained. Dikes are ad- hered to the surfaces of the specimens during the drying time. After completion of the moist and air curing, the surfaces of the specimens are covered with approximately 0.25 in. (6 mm) of 4% calcium chloride solution. Other deicer solu- tions may be used to simulate actual exposure conditions. The specimens are then placed in a freezing environment for 16 to 18 hours, followed by conditioning in laboratory air (73.5 ± 3.5°F [23.0°C ± 2.0°C], 45% to 55% RH) for 6 to 8 hours. Cycling continues generally for 50 cycles, but may continue beyond if differences between comparative tests have not developed. The solution is flushed every five cycles, a visual examination made, and the number of popouts may be quan- tified to assess the aggregate performance. The method relies on a visual examination of the test specimens after the proce- dure is completed. Performance may also be evaluated by quantifying the re- moved debris when flushing the slabs (28). In this method, detached particles and debris are collected when the surface is rinsed with fresh saline solution. The particles are washed in water, sieved using an 80 μm sieve, dried at 60°C (140°F) until a constant mass is obtained, and weighed. The mass is expressed in g/m2 (oz/ft2) of surface. The limit on mass loss after 56 cycles is cited as 500 g/m2 (1.6 oz/ft2) and 1000 g/m2 (3.3 oz/ft2) in Quebec and Sweden, respectively (28). Desirability Functions for Visual Inspection. ASTM C 672 describes a visual rating system of scaling that ranges from 0 to 5, 0 being best and 5 being worst. Table S1.3 describes the meaning of these ratings. Previous work included scaling resistance as one of the cri- teria for defining high-performance concrete (HPC) based on long-term performance (20). This evaluation recommended that the grade or performance required be tied to the volume of deicing salt application (in tons/lane-mile/year) used (use of more salt requires better performance), although a realistic value for applied salt is not always easily obtained. Because records of amount of salt applied over a given area are often not available, use may be categorized according to potential 19

concern of deicer salt scaling: a large concern, a moderate con- cern, or no concern. For severe and moderate deicer salt scaling environments, visual ratings of 0 to 1 and 2 to 3 at 50 cycles, respectively, are recommended. However, this test does not need to be run if deicer salt scaling is of no concern. Example desirability functions for the two conditions where testing is warranted are given in Figure S1.15. For severe exposure, lower ratings are assigned higher desirabilities, with average scaling ratings of 1 and 2.5 assigned desirability values of 0.9 and 0.4, respectively. For moderate exposure, the desir- ability function allows higher ratings; an average scaling rating of 1 or less is assigned a desirability of 1, and an average rating of 3 is assigned a desirability of 0.4. For both exposures, a scal- ing rating of 5 means that the entire mixture is unacceptable and thus a desirability of 0 is given. Desirability Function for Mass Loss. Some researchers use a mass loss measurement to quantify the material that has scaled off a specimen (28). This method generates non- subjective metrics of the concrete performance that are not based on, at least, a half number scale; it, therefore, can dif- ferentiate between mixtures that display good performance and would ordinarily receive the same visual rating. One problem with mass loss testing is that in addition to collect- ing scaled material, aggregates also can be collected that have failed because of other mechanisms such as popouts and popoffs. Popouts include mechanisms whereby porous, sub- standard aggregates absorb water and fail when exposed to freezing and thawing cycles. They are discerned by a conical failure near the surface of the concrete, with part of the ag- gregate that has failed left at the bottom of the cone. Popoffs are related to non-substandard aggregates (29). In this case, a thin mortar layer above a near-surface coarse aggregate pops off. These failures are described as “a very shallow trun- cated cone centered over or near the offending coarse aggre- gate particle.” Although all these failures are related, there are some differences as to the causes of the loss of surface material. If a mass loss value is desired, the following desirability functions apply to the environmental cases described above. The Quebec specification limits on mass loss are described in Saric-Coric and Aïtcin (28). Example desirability functions are shown in Figure S1.16. For both severe and moderate exposures, a lower mass loss is rated higher. A change in the function was defined at a mass loss of 500 g/m2 below which the desirability decreases at a greater rate with increasing mass loss than above this value. These functions do not assign great importance to variations in response over this range and assign no desirability lower than 0.5 for mass losses less than 1200 g/m2. If higher mass loss results are found during testing, the function should be extended so lower desirabilities are assigned. Effect of SCMs on Scaling Resistance. SCMs are gener- ally known to decrease scaling resistance of concrete (30). SCMs usually hydrate more slowly than portland cement, re- sulting in a lower early-age strength (less than 28 days) and a higher initial permeability; these trends are often reversed at later ages. However, in comparative early-age tests between concrete containing various combinations of SCMs, a mini- mum strength, instead of moist curing for 14 days, could be specified. Conflicting information exists in the literature as to the impact of additional curing time on scaling resistance. For example, Talbot et al. (30) states that the negative influence of the SCMs on scaling resistance was not noticeably reduced when the curing time for mixtures containing fly ash and slag was increased from 14 to 28 days. However, Saric-Coric and Aïtcin (28) report that increasing moist curing time from 13 to 27 days allowed slag-blended cements (up to 80% slag) to fully mature and pass the Quebec Ministry of Transportation limits on mass loss during ASTM C 672 testing. The method of curing used when comparing mixtures including SCMs should be selected carefully. Fly Ash. ACI Committee 232 (4) recommends that con- crete exposed to deicing salts be air entrained and allowed to reach a specified strength prior to exposure to the salts; no guidance is given as to the minimum strength. However, a minimum strength of 3500 psi (24 MPa) is given for freez- ing and thawing resistance. A mention is made that concrete containing 40% fly ash may be more susceptible to scaling. Talbot et al. (30) show that Class C fly ashes tested at 20% replacement increase the mass lost on a trowelled surface by approximately two times or more (from 1 kg/m2 to 1.8–2.8 kg/m2 with various Class C fly ashes at a w/cm of 0.4). Afrani and Rogers (31) state that, for the Ontario Ministry of Transportation, the maximum amount of fly ash that can be substituted for portland cement is 10% when the concrete is exposed to freezing and thawing and deicers are used. Addi- tionally, if fly ash and GGBFS are used together, the fly ash content is also limited to 10%, and the total fly ash plus GGBFS content is limited to 25%. ACI Committee 318 (32) limits fly ash and natural pozzolans to 25% by mass for concrete exposed to deicing chemicals. GGBFS. ACI Committee 233 (6) states that some labora- tory tests indicate less resistance to deicer salt scaling when GGBFS is used; although, little difference is observed in serv- ice. More scaling is found when concrete has higher w/cm and higher percentages of GGBFS. No quantities are mentioned. Experience cited in Saric-Coric and Aïtcin (28) indicates that GGBFS quantities greater than 20% reduced scaling resistance. Hooton (18) states that the Ontario Ministry of Transportation limits GGBFS quantities to 25% based on poor results in ASTM C 672 but suggests that up to 35% will not impact salt 20

scaling performance whereas 50% might be a concern. ACI Committee 318 (32) limits GGBFS to 50% by mass when con- crete is exposed to deicing chemicals. Silica Fume. ACI Committee 234 (7) states that with proper air-entrainment, silica fume should have no detrimen- tal effect on scaling resistance. However, a study by Pigeon et al. (33) showed reduced resistance to salt scaling when silica fume replacement exceeded 5%, while another study by Sørensen (34) showed similar results when the silica fume content ex- ceeded 10%. ACI Committee 318 (32) limits silica fume to 10% when concrete is exposed to deicing chemicals. Class N Pozzolans. Barger et al. (10) found good scaling resistance (visual ratings 0 to 1) for concrete containing 18% to 20% calcined clay by replacement. Combinations of SCMs. ACI Committee 318 (32) limits the total of fly ash or other pozzolans, GGBFS, and silica fume to 50% by mass of concrete exposed to deicing chemicals, with the fly ash or other pozzolans limited to 25% and silica fume limited to 10% of the total cementitious materials. If fly ash or other pozzolans are used with silica fume, the total is limited to 35% by mass, with the fly ash or other pozzolans limited to 25% and the silica fume limited to 10%. Corrosion Concerns for Concrete [CL1] The most frequent cause of deterioration in concrete bridge decks in northern climates is corrosion of the reinforcing steel. The products of the corrosion process occupy a larger volume than the original steel and thus generate internal forces that lead to cracking and spalling. In freshly placed concrete, rein- forcing steel is protected from corrosion by the formation of a passive oxide layer on the surface of the steel. This oxide layer is stabilized by the highly alkaline environment produced by the combination of portland cement and the mixture water. However, the oxide layer can eventually break down through the process of carbonation, the penetration of carbon dioxide into the concrete which causes a reduction in the pH, or by the presence of chloride ions. This latter mechanism, which typi- cally occurs when the concentration of chloride at the steel surface is approximately 0.2% by mass of cement, is especially significant for bridge decks because chloride-based deicing salts are often used to improve vehicle traction during winter. Chloride Penetration Resistance [CL2] Chloride ions permeate through concrete to the reinforcing steel initially by capillary action, especially if the concrete is dry, and then by other mechanisms, the most significant of which is diffusion. The rate of permeation is largely deter- mined by the pore structure of the concrete matrix, which can be modified significantly through the use of SCMs. Concrete with low permeability is more resistant to corrosion-related damage and to other deterioration mechanisms that require the ingress of water or aggressive water-borne agents into the concrete such as ASR, sulfate attack, and freezing and thawing distress. While water permeability, which directly influences these deterioration mechanisms, can be measured, testing for chloride penetration is the most common test for corrosion- related concerns. Test Methods. Resistance to chloride ion penetration can be measured directly using AASHTO T 259, Standard Method of Test for Resistance of Concrete to Chloride Ion Penetration, or ASTM C 1556, Standard Test Method for Determining the Apparent Chloride Diffusion Coefficient of Cementitious Mixtures by Bulk Diffusion, and ASTM C 1543, Standard Test Method for Determining the Penetration of Chloride Ion into Concrete by Ponding. It can also be as- sessed indirectly by accelerated testing conducted according to AASHTO T 277 (ASTM C 1202), Standard Method of Test for Electrical Indication of Concrete’s Ability to Resist Chlo- ride Ion Penetration, also known as the rapid chloride per- meability (RCP) test or the electrical conductivity test; this test is referred to as the electrical conductivity test in these Guidelines. Diffusion Testing. In the procedure laid out by AASHTO T 259, four concrete slabs not less than 3 in. (75 mm) thick are cast, moist cured for 14 days, and dried for 14 more days in a 50% RH environment. At that time, the concrete surface may be abraded to simulate the wearing effects of vehicle traffic. Dams that will hold a sodium chlo- ride solution on top of the slabs are then attached to all but one control slab. The slabs are then subjected to an additional 14 days of drying, after which a ponding solution of 3% sodium chloride is applied to a depth of 0.5 in. (13 mm) and allowed to permeate the concrete for 90 days. ASTM C 1543 suggests a very similar exposure routine except that the spec- imens are to be cured as laid out by ASTM C 672. For AASHTO T 259, two samples are taken from the slabs at 0.0625 to 0.5000 in. (1.6 to 13.0 mm) and 0.5 to 1.0 in. (13 to 25 mm) from the surface. For ASTM C 1543, samples are taken at four depths between 10 and 65 mm. These samples can be either cores, which are ground, or powder obtained after drilling with a rotary hammer-drill. The water or acid- soluble chloride content of the samples is measured according to AASHTO T 260, Standard Method of Test for Sampling and Testing for Chloride Ion in Concrete and Concrete Raw Materials, or ASTM C 1152, Test Method for Acid-Soluble Chloride in Mortar and Concrete. Testing for acid-soluble chloride gives a measure closer to the total chloride content in the sample and includes the chloride that is chemically 21

combined. Measurements of water-soluble chloride approxi- mate the amount of chloride that has permeated the concrete without including chemically bound, currently unavailable chloride. Either of these methods may be used for comparing relative mixture performance because the chloride content in the control slab is used as the baseline to estimate how much chloride penetrated the test slabs. Many variations of chloride ponding tests can be found in the literature. Comparisons between studies are difficult because not only is the permeability affected by the concrete microstructure, but also the test results are strongly influ- enced by the moisture content of the concrete when exposure is initiated, the temperature of the concrete, the extent of hydration of the cement, the curing regimen, and the surface preparation. When conducting tests to judge relative per- formance, it is important to keep these factors constant. Some of the limitations of the standard AASHTO T 259 test include the 14-day wet curing regime, which may not permit the complete hydration of the SCMs which typically occurs more slowly than the hydration of portland cement. Also, the wet curing regimen has been viewed by Sherman et al. (35) as unrealistically long in comparison with standard practice of 7 days of curing. Because of the low permeability of high- performance concrete, a 90-day test duration using 3% sodium chloride solution may not be sufficient to differenti- ate between proposed mixtures. Therefore, using 15% sodium chloride solution and longer ponding times, such as 6 months or 1 year, may be necessary. Precise sampling is best achieved with cores where the depth of penetration can be more accu- rately measured and there is less likelihood of contamination than when drilling powder samples from the test slabs. ASTM C 1556 addresses some of these issues. In this test, finished cylinders that had been cured 28 days are saturated in lime solution and submerged in 16.5% sodium chloride solution for at least 45 days. After the exposure period, near surface samples are taken using a machine lathe or end mill. Because of the precision of these instruments, concrete sam- ples can be taken at 1 mm intervals, and for low permeability concretes, as many as eight samples may be taken within the first 12 mm. Common practice during chloride ponding testing is to es- timate the chloride concentration profiles and determine the “effective” diffusion coefficient as defined by Fick’s second law of diffusion (36), using the procedure provided in ASTM C 1556. The diffusion coefficient is “effective” because other mechanisms besides diffusion also play a role in the transport of chloride through the concrete and are lumped into this sin- gle coefficient. AASHTO T 259 provides for determining the chloride concentration at only two sample depths, which is not sufficient to accurately determine diffusion coefficients. When the diffusion coefficient is to be determined, the chloride concentrations of six or more thin concrete slices at various depths anticipated to span the range of chloride content within the samples should be measured. The diffusion coefficient is calculated using an iterative least squares fitting procedure or other methods (37). Test Method for Electrical Indication of Concrete’s Abil- ity to Resist Chloride Ion Penetration. In the AASHTO T 277 test (referred to here as the electrical conductivity test), 2-in. (50-mm) thick concrete specimens are cut from nomi- nally 4-in. (100-mm) diameter cylinders or cores, saturated with water, and placed between solutions of 3% sodium chlo- ride and 0.3 N sodium hydroxide. A 60 V DC potential is applied across each specimen, and the total electrical charge passing through the specimen in 6 hours is measured. The relative simplicity and short duration of this electrical conductivity test has resulted in its widespread adoption for measuring chloride permeability in lieu of diffusion testing. This test is also commonly used for quality control where dura- bility is a primary objective. Many researchers and engineers have viewed the use of the electrical conductance value as an oversimplified approximation of permeability and inappro- priate for certain materials, particularly when SCMs are used. This view is held especially for mixtures containing silica fume. Electrical conductivity depends on conductivity of pore fluid and the nature of the porosity in the paste. Despite the appar- ent relationship between permeability and the charge passed presented in tabular form in AASHTO T 277, correlations between the two test methods should be established for the specific material combinations before concretes containing different cementitious materials are compared (38). Also, the measured charge passed overestimates the actual chloride permeability because it includes the movement of hydroxyl, chloride ions, and other ions (39). SCMs may react with hydroxides preventing their movement and making concretes containing these materials appear to have a lower permeability than is actually the case. The electrical conductivity test also has a higher variability than the diffusion tests. All of the chloride penetration tests discussed here are strongly influenced by the age and curing conditions of the specimen. The microstructure of the concrete continues to develop over time and significant reductions in penetration can be observed when testing is initiated at later ages, espe- cially for concretes containing SCMs. Electrical conductivity test results for concretes containing SCMs obtained after 56 or 90 days of wet curing may be expected to be significantly lower than those measured after only 28 days. Therefore, dur- ing concrete mixture selection, testing both at 28 days and at 56 and/or 90 days of age is recommended. Desirability Functions for Chloride Penetration and Electrical Conductivity. To minimize the rate of penetra- tion of harmful water-borne agents into the concrete, the 22

permeability should be minimized. ASTM C 1556 or the long- term AASHTO T 259 test method more realistically simulate the penetration mechanisms of chlorides in concrete that occur in bridge decks. A desirability function for the apparent chloride ion diffusion coefficient test is shown in Figure S1.17. This function rewards lower apparent chloride diffusion coef- ficients, only assigning a desirability of 1 for a coefficient of 0. Because this response is likely to be important when this test is included in the test program, a steep slope of the function is assumed: a coefficient of 2 × 10−12 m2/s is assigned a desirabil- ity of 0.85 and a coefficient of 6 × 10−12 m2/s is assigned a desirability of 0.15. Because of this slope, the response will strongly influence the overall desirability and optimization analysis. Electrical conductivity test results are a less precise pre- dictor of actual concrete performance and generally more prone to variability. A sample desirability function for the electrical conductivity test is shown in Figure S1.18. This function rewards lower amounts of passed charge by assigning a desirability of 1 for 0 Coulombs passed and increasingly lower desirabilities for higher amounts of passed charge up to 5000 Coulombs passed. This function provides less steep slope than that for the desirability func- tion for chloride diffusion coefficient because a concrete mixture that may have a passed charge as high as 2000 Coulombs could be highly resistant to chloride penetration as measured using the more accurate ponding- or exposure- based test methods. Therefore, a passed charge of 2000 Coulombs or less is assigned a moderately high desirability of 0.8 or higher. Effect of SCMs on Chloride Penetration Resistance. The chloride penetration resistance of uncracked concrete is determined at the microstructural level by the porosity and extent of interconnected voids in the bulk cementitious paste and in the interfacial zone between paste and aggregates. Lower water–cementitious materials ratios result in increases in the density of the paste microstructure (reduced porosity) and reduced permeability. SCMs also densify the paste microstructure by virtue of their size distributions, which typically are smaller than portland cement, and enable the SCMs to fill the spaces between the cement particles. Blends of two or more SCMs may be more effective at minimizing penetration than use of just one, because the particle size dis- tributions of the various materials may complement each other. In addition, SCMs may react with calcium hydroxide to form additional hydration products that also help fill pore space in the concrete. Fly Ash. Class F and Class C fly ashes are typically used at rates of 15% to 25% and 15% to 40% by mass of total cemen- titious materials, respectively (21), although some researchers recommend 25% to 35% replacement (40). Because the reac- tivity of fly ashes is variable, testing is essential to determine the optimum content required for minimizing penetration. Some concretes containing fly ash, especially as replacement of cement, can have low chloride penetration resistance. Some fly ash increases penetration resistance by refining the pore structure as a result of its reaction with calcium hydroxide (4). This reaction occurs slowly and thus requires adequate mois- ture and prolonged curing before exposure. GGBFS. GGBFS has been shown to be effective at reduc- ing the permeability of concretes both when used alone and when used in combination with other SCMs (6). Recom- mended quantities for bridge deck concretes range from 15% to 30% by mass of total cementitious materials. Silica Fume. Silica fume is finer than the other typical SCMs, and it is therefore most effective in reducing concrete permeability because it fills voids in the pore structure and along the interfacial transition zone. For projects where min- imizing chloride penetration is essential, silica fume (or metakaolin) should be included. Silica fume additions of more than 5% by weight have been shown to significantly reduce electrical conductivity and additions as high as 7.5% to 10% have been specified where corrosion resistance is con- sidered essential (7). However, high dosages (greater than 7%) of silica fume can make the concrete more susceptible to early cracking and can defeat the low permeability within the paste if special precautions are not taken to prevent early cracking. Class N Pozzolans. The small size of the metakaolin par- ticles has been shown to significantly reduce chloride perme- ability at contents of 8% to 12% by weight of cementitious materials when tested using both the electrical conductivity and diffusion approaches (41). Metakaolin is marketed as an alternative to silica fume and reportedly achieves similar reductions in permeability (9). Detwiler et al. (17) report an apparent diffusion coefficient of 1.7 to 1.9 × 10−12 m2/s for concrete containing 20% to 30% calcined clay, respectively, compared to 5.4 × 10−12 m2/s for a Type I portland cement concrete made with a water-cement ratio (w/c) of 0.4. Other Mixture Considerations for Minimizing Chloride Penetration. HRWR admixtures have been shown to reduce the chloride penetration of concretes containing SCMs (42) presumably because the combination of HRWRs and SCMs improves dispersion synergistically and reduces water demand for a given workability thereby minimizing penetra- tion. For this reason, especially when silica fume is used, HRWR admixture should be used. The general recommendations listed in Table S1.5 can be followed to minimize chloride penetration of concrete. 23

Coastal Environments [CO1] Concerns about durability in coastal environments also in- clude airborne-chloride exposure that can lead to corrosion problems in a manner similar to exposure to chloride from deicing salts as discussed previously in “Chloride Penetration Resistance.” Direct contact with sea water in tidal zones is the most aggressive exposure; airborne-chloride exposure is less severe than tidal or deicer exposure. Two methods are suggested to decrease the potential of corrosion due to airborne sources of chloride. In one method, the depth of cover is increased in a manner related to the dis- tance from the coast. Tanaka et al. (43) describe a study per- formed in Japan in which the level of airborne chloride was measured in different regions of the country and at different distances from the coastline and provide a method for deter- mining a recommended minimum depth of cover based on the location of the structure and the design life. The other method involves restricting the upper limit of the w/cm to decrease concrete permeability. SCMs can be used to decrease the permeability of the concrete and decrease the dif- fusion coefficient. Table S1.6 lists mixture considerations for chloride penetration resistance in coastal environments. Con- crete exposed to airborne chloride only is considered to be subject to moderate exposure while concrete in sea water or within 4 m of sea water is considered to be subject to severe ex- posure. Abrasive Environments [AB1] Abrasion on highway bridges causes deterioration of the concrete surface through two mechanisms: (1) wearing, as foreign particles such as sand or other grit are ground across the surface by vehicular traffic, and (2) impact of steel chains or studs attached to the tires of automobiles and trucks to provide additional traction during winter. The latter mecha- nism is the more severe. Chains are required in a limited number of areas, mostly where mountainous roads demand better traction. Because the surface of the concrete is also affected by many other deterioration mechanisms, such as salt scaling, abrasion resistance may strongly influence the lifespan of the bridge deck. The duration of winter season and the frequency of traffic with chains or studded tires indicate the relevance of this property to the overall durability of the bridge deck. Table S1.7 lists the mixture considerations other than SCM contents that are relevant to abrasion resistance. Abrasion Resistance [AB2] The ability of concrete to resist abrasion is governed by ag- gregate properties, concrete strength, and surface treatment (44). Because aggregate is generally the hardest component of concrete, it is usually the determining factor in abrasion per- formance. Aggregates with greater density and hardness have been shown to produce concretes that wear more slowly than aggregates made of softer materials. The shape of the aggre- gates is also a factor; more angular aggregates produce a bet- ter bond with the cementitious paste and a more uniform traffic-bearing surface. Concrete strength has also been linked to abrasion resist- ance, probably because strength is related to paste density and hardness (e.g., concretes with low w/cm containing silica fume have high strengths and high abrasion resistance). Properties that reduce concrete density such as entrained air have been shown to adversely affect abrasion resistance. When soft aggregates are used, steps to increase strength can be taken. The third factor in determining abrasion resistance is sur- face treatment, i.e., finishing and curing. A smoother finish provides a more uniform load-carrying surface that will dis- tribute abrasion-causing wear and impact. However, the smoother surface will reduce skid resistance. Curing proce- dures influence the strength of the surface because the amount of hydration at the surface depends on the presence of moisture. Test Methods [AB3]. While several methods exist to as- sess abrasion resistance, two test methods best simulate the loading that would be experienced by bridge deck concrete: ASTM C 779, Standard Test Method for Abrasion Resistance of Horizontal Concrete Surfaces Procedure B, and ASTM C 944, Standard Test Method for Abrasion Resistance of Con- crete or Mortar Surfaces by the Rotating-Cutter Method. Both methods provide for abrading the concrete surface with dressing wheels (steel wheels with sharp points) that roll over the concrete surface under a specified load in a manner that is intended to simulate chain impact. The ASTM C 779 method requires a more elaborate setup and a machine specifically designed for this test. The ASTM C 944 test uses a modified drill press and is easier to run. However, it typically results in poorer reproducibility than the ASTM C 779 method, which uses three instead of just one dressing wheel assembly. The ASTM C 944 method measures abrasion in terms of the depth of wear or the mass loss from a concrete specimen after it has been abraded using the dressing wheel head under 22 lbs (10 kg) of load for three 2-minute periods, between which the specimen is cleaned. The abrasion is con- ducted on three separate areas on the concrete, and the aver- age depth of wear or mass loss is reported. Desirability Function for Abrasion Resistance. In spite of its poor reproducibility, the ASTM C 944 test method has value in distinguishing relative abrasion performance. There- fore, after the testing is completed, the desirability function 24

for this property may have to be modified to bracket the actual test results. However, Figures S1.19 and S1.20 give examples of desirability functions for mass loss and depth of wear as a starting point. Either mass loss or depth of wear can be used to assess performance for either test method. Both of the functions for abrasion resistance are based on thresholds (mass loss of 2 g after 6 minutes and depth of wear of less than 0.5 mm after 6 minutes) below which the performance is con- sidered as good as necessary and assigned a desirability of 1. Above these thresholds, the desirability function decreases with increasing wear. Effect of SCMs on Abrasion Resistance. The aggregate has the largest influence on abrasion resistance, but mixture modifications that will increase concrete strength will generally increase abrasion resistance. Fly Ash. Replacement of cement with fly ash has little influence on abrasion resistance (45). GGBFS. A study conducted to assess the influence of slag on abrasion showed that slag increased the loss due to abra- sion when used at more than 25% (5). Therefore, if abrasion resistance is a prime objective, the recommendation is to limit the content of slag to 25% replacement. Silica Fume. Silica fume is known to increase the strength of concrete and can be used to increase abrasion resistance (46). To have a significant effect on abrasion resistance, the recommended dose for silica fume content would be greater than 5% addition. Alkali-Silica Reactivity Potential [ASR1] ASR is a concern in some areas of the United States. Prob- lems occur when aggregates of certain compositions react with the hydroxides of cement pastes. This reaction is mani- fested by the formation of gel and subsequent cracking of concrete. This form of cracking is inherently a materials issue; it will be dealt with in Step 2, which describes the selection of durable raw materials. Cracking Resistance [CR1] Concrete cracking is a nationwide problem on bridge decks. Cracking can occur while the concrete is still plastic and after the concrete is hard. Krauss and Rogalla (47) reported on extensive work performed to identify causes of early-age cracking. The cracking usually occurs within the first 6 to 12 months after construction. The cracks often can become quite large (0.010 to 0.015 in. [0.25 to 0.38 mm] wide). Iden- tifying the causes and preventing cracking in these structures is difficult and complex. Concrete develops cracks when local tensile stresses exceed the local tensile strength of the concrete. Tensile stresses in bridge decks are caused by temperature changes in the bridge, concrete shrinkage, and sometimes bending from dead or live loads. A combination of shrinkage and thermal stresses causes most early-age cracks. Shrinkage and thermal stresses develop in all composite decks, because the girders and decking restrain the natural thermal and shrinkage movement of the concrete. When the deck and girders consist of different materials (steel and concrete, or different concretes) with different thermal expansion rates, even a constant temperature change will cause stresses because the different materials expand differ- ently and cannot expand freely where they are attached. Temperatures are rarely uniform or linearly distributed, and shrinkage is also not linearly distributed. Nonlinear shrink- age and temperature changes cause internal stresses, even without external restraint. Because considerable friction exists between the concrete and its supporting girders or a composite metal deck, these elements restrain concrete movement. To a lesser extent, embedded reinforcement in the deck also restrains the deck against shrinkage and against thermal movements. A decision should be made on whether deck cracking is a concern. Many aspects of high performance concrete and modern cement characteristics increase the potential for deck cracking (47). Historically, conventional concrete mixtures cast on typical bridge supports have experienced cracking. Therefore, cracking resistance is an important consideration unless the deck is prestressed, will not experience shrinkage, or will receive a membrane, or somehow it is known that cracking will occur. If cracking is not considered an important concern or if there is no time to evaluate this characteristic, general design and construction procedures are provided to assist engineers with design and construction decisions. Causes of Cracking Deck cracking in hardened concrete (not plastic cracking) is usually either restrained concrete shrinkage cracking and/or thermal cracking due to temperature differentials within the concrete or relative to the supporting structure. Neither of these factors is usually measured directly. How- ever, several tests are available to help compare different mixtures to estimate if one mixture is more prone to crack- ing than another mixture. Many concrete properties affect its cracking tendency. Such properties include heat of hydration, coefficient of thermal expansion, tensile strength, paste-aggregate bond strength, drying shrinkage, modulus of elasticity, and creep. The time dependency of each of these properties is important. The 25

following tests are used to evaluate some of these individual properties: • Drying Shrinkage: AASHTO T 160, ASTM C 157 • Tensile Strength: AASHTO T 198 • Flexural Strength: AASHTO T 97, AASHTO T 177 • Modulus of Elasticity: ASTM C 469 • Creep: ASTM C 512 • Heat of Hydration: CRD-C38 • Coefficient of Thermal Expansion: CRD-C39 or AASHTO TP 60 Restrained Shrinkage [CR2] Restrained shrinkage cracking is evaluated using AASHTO PP 34-99, Standard Practice for Estimating the Cracking Tendency of Concrete, or ASTM C 1581, Standard Test Method for Determining Age of Cracking and Induced Tensile Stress Characteristics of Mortar and Concrete Under Restrained Shrinkage. Although free shrinkage—as measured in AASHTO T 160 (see next subsection)— provides some indication of the driving force behind drying shrinkage cracking, the test for it does not offer sufficient information alone to predict cracking behavior of concrete structures because almost all structures are restrained in some fashion, and stiffness and creep of the concrete inter- act with shrinkage to determine whether cracking will occur. The restrained ring test was found to correlate better to deck cracking than standard shrinkage tests because most of the important time-dependent factors are part of the test (drying shrinkage, rate of tensile strength gain, elastic mod- ulus, creep, etc.) Only a limited amount of work has been done in evaluation of SCMs and restrained shrinkage. For evaluation of SCM mixtures, only restrained ring and dry- ing shrinkage testing is proposed. However, for projects with strict cracking concerns, tests for determining tensile strength gain, elastic modulus, and creep at early ages may provide useful data. Drying Shrinkage [CR3] Drying shrinkage is typically measured using AASHTO T 160, Standard Test Method for Length Change of Hardened Hydraulic Cement, Mortar, and Concrete. Specimens (3×3×11.25 in. [75×7×281 mm] for maximum aggregate size less than 1 in. [25 mm]) of concrete or mortar are cast and moist cured in lime water for 28 days. They are then dried at 73.4 ± 3.0°F (23.0 ± 1.7°C) and 50% RH, and the drying shrinkage is measured. The test may be modified to closer simulate field curing conditions (e.g., 7- or 14-day curing may be more representative of the age when drying of the in-place bulk concrete begins). Desirability Function for Drying Shrinkage. Usually minimization of concrete shrinkage due to drying is desired because the lower shrinkage results in lower stress and thus less potential for cracking. Figure S1.21 shows a sample desirability function for drying shrinkage. Free drying shrink- age of the concrete mixture is only one of the factors that determine whether shrinkage-related cracking will occur. A measurement of drying shrinkage provides some indica- tion of whether cracking is likely but does not measure the cracking tendency directly. Therefore, this desirability func- tion was defined to avoid rating any mixture with a very low desirability. It assigns a desirability of 1 to mixtures with shrinkage strains at 90 days below a magnitude of –0.04%; the desirability decreases gently with increasing (more negative) shrinkage strain between –0.04% and –0.06%, and then decreases more rapidly with greater shrinkage increases. Effect of SCMs on Drying Shrinkage. In general, use of SCMs in concrete mixtures increases the potential for shrink- age because a higher paste volume results when SCMs are used as replacement of cement. However, Klieger and Perenchio (48) found that Class F fly ash had little or no effect on drying shrinkage, but Brooks and Neville (49) found that fly ash can increase shrinkage by as much as 20%. Most of the research on silica fume and GGBFS shows an increase in shrinkage with the replacement of cement. Some of the increased shrinkage associated with the use of SCMs can be mitigated by proper curing procedures. The fine pore struc- ture in SCM concretes is also thought to contribute to the increase in drying shrinkage. Testing is recommended because many chemical admixtures used in concrete mix- tures can increase shrinkage. Cracking Tendency [CR4] Cracking tendency is most directly measured using AASHTO PP 34-99, Standard Practice for Estimating the Cracking Tendency of Concrete, or ASTM C 1581, Standard Test Method for Determining Age of Cracking and Induced Tensile Stress Characteristics of Mortar and Concrete Under Restrained Shrinkage. These methods call for concrete to be cast around steel rings and subjected to a drying environment (50% RH). Strain gauges measure the stress in the steel, and the time when a sudden drop in stress is observed is recorded as the time to first cracking. The procedure is intended for use in de- termining the effects of variations in the properties of concrete on the time to cracking of the concrete when restrained. The procedure provides comparative data and is not intended to determine the time of initial cracking of concrete cast in a specific type of structure. Cracking in service is influenced by many factors including degree of restraint, hydration effects, and environmental conditions. The method is useful to 26

determine the relative likelihood of early concrete cracking and to aid in selection of concrete mixtures that are less likely to crack. The test method may also be modified to evaluate other factors that may affect cracking such as curing time, method, or temperatures. The rate of stress increase can also be com- pared to evaluate concrete mixtures that do not crack; the mix- ture exhibiting the lowest strain rate would be preferred. Desirability Function for Restrained Shrinkage. Mix- tures that do not crack or crack at later ages during cracking tendency testing are preferred. The standard test methods call for only 24 hours of wet curing while bridge decks are gener- ally cured for 7 days. Depending on the likely curing regimen for the actual bridge deck, this period may be varied and the de- sirability function must be adjusted accordingly. Figure S1.22 shows an example of a desirability function for time to first crack after 7 days of curing. This function rewards longer times to cracking; if the concrete mixture can last up to 6 weeks before cracking, it is rated a 0.95 or higher. Mixtures cracking at less than 6 weeks receive lower desirabilities with a desirabil- ity of 0 assigned to mixtures that crack before 1 week. Twenty- four hours of curing appears to be the more rigorous test and thus the desirability function would have to be shifted to assign higher desirabilities at lower ages of cracking if such curing reg- imen was applied. Effect of SCMs on Cracking Resistance Fly Ash and GGBFS. The effect of SCMs on the cracking resistance of concrete under restrained shrinkage is not well known. Slowing the early-age (less than 7-day) hydration of the concrete may help reduce the risk of cracking by slowing the rate at which the modulus of elasticity increases and by in- creasing early-age creep. Fly ash and slag typically reduce the early-age strength of concretes when used as a replacement of cement. Replacement of cement with Class F fly ash is typically more effective at slowing strength gain than replacement with Class C fly ash or GGBFS. Generally, the addition of fly ash or GGBFS to the concrete does not affect the cracking tendency of the concrete greatly if the total cementitious volume is not changed. Cracking (drying shrinkage) may be reduced if the improved workability of the mixture containing the SCM con- tributes to reduced water demand and reduced paste volume. Silica Fume. Silica fume has been shown to increase the cracking tendency of concretes in the restrained shrinkage test. Li et al. (50) found that silica fume replacement of cement not only increases the cracking tendency but also increases the crack width in the restrained shrinkage test. Class N Pozzolans (Metakaolin). Ding and Li (9) report that metakaolin shrinks less than portland cement concrete and less than silica fume–containing concrete at equivalent dosages. For metakaolin they also found that the rate of shrink- age was higher than for plain portland cement concrete up to 5 days and lower after that. With respect to restrained shrink- age, crack widths of metakaolin-containing concrete were less than those for portland cement concrete but were initiated at earlier ages. Taylor and Burg (8) report similar shrinkage for metakaolin- and silica fume–containing concretes. Thermal Concerns for Cracking Resistance [CR5] Heat evolution is usually considered only for more massive concrete applications. However, thermal gradients can cause large early-age stresses in concrete bridge decks and result in cracking. Concretes containing large quantities of cement with rapid hydration characteristics are more prone to ther- mal cracking and, therefore, large thermal gradients within the deck should be avoided. Modifying construction practices is one approach to minimize the likelihood of cracking. How- ever, testing concrete mixtures to identify mixtures with low rate of heat evolution, low total heat of hydration, and low coefficient of thermal expansion and thus to reduce the risk of thermal cracking is another approach. If a review of the bridge structural analysis and proposed mixture indicates only small or moderate risk for cracking, consideration should be given to curing practices that min- imize thermal gradients in the concrete (see “Special Con- struction Practices [CR10]”). An effective means to reduce thermal cracking may involve temperature monitoring and appropriate adjustment of curing procedures. If the deck is thick or integral with piers or beams such that high concrete hydration temperatures may occur, or if the con- crete has large amounts of rapid hardening cementitious ma- terials, it may be prudent to test concrete mixtures and select SCM mixtures that reduce the heat of hydration. This property can be used to model the three-dimensional temperature gra- dients within the structure. Aggregates also can be selected to minimize the coefficient of thermal expansion of the concrete. Heat of Hydration [CR6] The heat of hydration and the rate of the heat evolution are not normally measured for bridge deck concrete. However, the rate of heat evolution may influence the development of thermal cracks because the rapid heat evolution can result in high temperatures during initial hardening and develop more severe thermal gradients when the concrete cools rapidly. The heat of hydration, or total energy released during the hydra- tion process, is a major concern for mass concrete place- ments. However, for bridge decks, the rate of heat evolution is more important than the heat of hydration because peak temperature differentials in decks typically occur within a few days of construction and are determined by the interaction between hydration-related heat evolution and cooling. 27

Another temperature-related concern for concrete is the phenomenon of delayed ettringite formation (DEF), in which the formation of early-age ettringite is suppressed by high curing temperatures but ettringite reforms under moist conditions at some time later in the concrete’s life, causing deleterious expansion and cracking. For precast concrete, the Canadian Standards Association (51) currently recommends that ordinary portland cement concrete in a moist environ- ment should not reach a concrete temperature above 140°F (60°C) during curing. This concrete temperature curing limit is also applicable to concrete where no heat is applied. Use of SCMs may reduce the risk of DEF. Test Methods. There are no direct test methods for meas- uring the heat of hydration of concrete. The U.S. Army Corps of Engineers published a method in 1973, CRD-C38, which measures the temperature increase of freshly mixed concrete over time in adiabatic conditions (i.e., in a controlled envi- ronment where no heat is allowed to enter or escape the sam- ple). This test is difficult to perform and few laboratories are equipped to perform it. Instead, it is more practical to evalu- ate the relative performance of mixtures by monitoring tem- perature rise in an insulated container. Some heat loss is inevitable with this type of testing configuration and this loss may approximate the cooling experienced by freshly placed concrete decks with peak temperatures typically occurring within 8 to 24 hours. Concrete mixtures that have a higher rate of heat evolution will produce a higher peak temperature. Therefore, performance of mixtures relative to thermal stresses in concrete bridge decks can be evaluated based on the peak temperature in the insulated container. The peak tem- perature in this type of testing is different from the likely significantly higher peak temperature that might be achieved in adiabatic conditions. It is also possible to track temperature evolution over time and calculate the maximum rate of tem- perature increase, which may be used as an evaluation crite- rion. The insulated containers and ambient conditions during this testing must be similar to provide a basis for relative performance. For the case study conducted as part of the development of this methodology, 6×12-in. (150×300-mm) diameter cylinders were cast for each mixture and cured inside blocks of insulat- ing foam. The cylinders were cured inside containers that meet ASTM C 684, Standard Test Method for Making, Accelerated Curing, and Testing Concrete Compression Test Specimens, Procedure C. These containers were kept in a controlled envi- ronmental chamber held at a temperature of 73°F (23°C) before and during hydration. Before batching, all the raw materials were stored in a climate-controlled room and stabi- lized at room temperature. The temperature of the cylinders was monitored with thermocouples that were embedded in the center of the concrete, and the temperature rise resulting from the heat evolution during hydration for each mixture was recorded. Desirability Function for Temperature Rise. Because the test for temperature rise is not standardized, each labora- tory will have to develop a specific method of test and inter- pretation of the results. A sample desirability function for heat of hydration (temperature rise) is shown in Figure S1.23. This function stipulates that a lower temperature rise is preferred. However, because temperature rise, like free drying shrinkage, is only one of the many factors that determine whether a bridge deck cracks, this function was defined to downgrade mixtures only slightly for adequate performance or better. It assigns a desirability of 1 to temperature rises of 20°F (11°C) or less and slightly lower desirabilities to temperature rises of up to 50°F (28°C). If the temperature rise is more than 80°F (44°C) above the initial 73°F (23°C), the maximum allowable temperature of 150°F (66°C) chosen for this example based on potential susceptibility to DEF will be exceeded and thus the mixture would be considered unacceptable. This function should, however, be adjusted based on the specific application and the test procedure used. Effect of SCMs on Heat of Hydration. The heat of hydra- tion is influenced by the amount and type of cement, the type and amount of SCMs used, and curing conditions. Modern portland cements tend to have high tricalcium silicate (C3S) contents, and high fineness that result in more rapid heat gen- eration during hydration compared with cements used in past decades. Therefore, the heat of hydration may be a more important parameter than it has historically been. Fly Ash and GGBFS. Replacement of cement with fly ash and GGBFS will reduce the rate of heat evolution in concrete structures. The greater the replacement, the lower the early heat of hydration. The hydration processes of fly ash and to some extent GGBFS are slow because of their reaction with the calcium hydroxide products of the cement hydration process. In the case of fly ash, the pozzolanic reaction may not actively occur until a week after mixing. Use of Class F fly ashes to replace cement is typically most effective at reducing the rate of hydration. Some Class C fly ashes may actually increase the rate of heat production. Cement replacement of up to 75% with GGBFS or 40% with Class F ash is sometimes used for mass concrete applications. Silica Fume. Researchers do not agree on the effect of sil- ica fume on the development of heat of hydration. Kadri and Duval (52) found that, at early ages, silica fume accelerates the hydration process, and that at 10% cement replacement, the cumulative heat of hydration is greater than portland cement concrete. However, Alshamsi (53) found that a 10% cement replacement of silica fume decreases the heat of hydration. 28

This decrease may result from the reduction in hydroxide ions available to react with the other SCMs. Low dosages of silica fume (i.e., 3% to 5%) are sometimes used to achieve cohesiveness while minimizing heat generation. Coefficient of Thermal Expansion [CR7] The coefficient of thermal expansion of the concrete is important with respect to early-age cracking because it de- scribes the free strain in the material for a given temperature change. Coefficient of thermal expansion is measured using the U.S. Army Corps of Engineers test CRD-C39, Test Method for Coefficient of Linear Thermal Expansion of Concrete, or AASHTO TP 60, Standard Test Method for Coefficient of Thermal Expansion of Hydraulic Cement Concrete. The coefficient of thermal expansion is mainly a function of the aggregate; the effect of SCMs on this value is expected to be very small. Therefore, testing for this coefficient is not suggested as part of the screening for SCM mixtures. Modulus of Elasticity [CR7] The modulus of elasticity (E) is the ratio of stress to strain in the linear portion of the stress-strain curve before the onset of significant microcracking; it can range from 2 to 6 × 106 psi (13.8 to 41.4 GPa). The modulus of elasticity is determined according to ASTM C 469, Standard Test Method for Static Modulus of Elasticity and Poisson’s Ratio of Concrete in Com- pression. The deformation of the concrete under compression loading is measured with strain gauges adhered to the concrete specimens or using deflection gauges mounted in a frame known as a compressometer. The E value relates the stress in the material for a given strain; high values of E may indicate susceptibility to cracking resulting from external strains such as drying shrinkage or thermal deformation since, for a given strain, greater stress is developed. The modulus of elasticity is a function of w/cm, volume of paste, and the aggregate modu- lus of elasticity and is typically related to strength. Desirability Function for Modulus of Elasticity. The modulus of elasticity may be specified by the designer. Low elastic modulus values are undesirable because they may result in large deformations; however, too high a value may increase the potential for cracking due to restraint. A sample desirabil- ity function for elastic modulus at an early age (7 days) is shown in Figure S1.24. This function suggests that the ideal range for elastic modulus is between 3 and 5 × 106 psi (20.7 to 34.5 GPa) at 7 days. Mixtures with elastic moduli greater than 1 × 106 psi (6.9 GPa) outside this range are considered unac- ceptable and assigned a desirability of 0, while those with elastic moduli less than 1 × 106 psi (6.9 GPa) outside this range are given an intermediate desirability. Effect of SCMs on Modulus of Elasticity Fly Ash. The modulus of elasticity of Class F fly ash– containing concrete is somewhat lower at early ages and a little higher at later ages than ordinary portland cement concrete (4). GGBFS. Fernandez and Malhotra (5) found that the elas- tic moduli were comparable between mixtures containing GGBFS and ordinary portland cement. Silica Fume. ACI Committee 234 (7) reports that the modulus of elasticity of concrete containing silica fume is similar to that of ordinary portland cement concrete of sim- ilar strength. Class N Pozzolans (Metakaolin). Taylor and Burg (8) report a slightly higher modulus of elasticity for metakaolin- containing concrete than for silica fume–containing con- crete. Caldarone et al. (14) found modulus of elasticity to be 15% to 18% higher with metakaolin at 5% to 10% addition than portland cement concrete. Plastic Shrinkage [CR8] Plastic shrinkage cracks occur after the concrete has been placed but before the concrete has set. They usually appear on exposed horizontal surfaces and can occur anytime when the ambient conditions (temperature, humidity, wind velocity) are conducive to rapid evaporation. Plastic shrinkage cracking generally occurs when the rate of evaporation exceeds the rate of bleeding of the concrete. The width of cracks at the surface may be as much as 0.25 in. (6.3 mm); however, the cracks are usually no more than 2 or 3 ft (0.6 or 0.9 m) long and are rarely more than 2 to 3 in. (50 to 75 mm) deep. Such cracks are sel- dom significant structurally, and once the crack forms, the stresses rapidly dissipate. Plastic shrinkage cracks occur very early after casting and are not associated with long-term aging. However, these early-age cracks can allow the penetration of deicer salts, water, and gases into the concrete, which may adversely influence the durability of the structure. The likelihood of plastic shrinkage cracking may be reduced by reducing the evaporation rate or increasing the bleeding capacity of the concrete. The former can be accomplished to various degrees by sunscreens, windbreaks, fog mist, or monomolecular curing films. Increasing the bleeding capac- ity of the concrete is usually not practical and may cause settlement cracking; however, use of a water-reducing admix- ture containing hydroxylated carboxylic acid may increase the bleeding rate. The most effective means of avoiding the loss of bleed water (reducing evaporation) is by using impermeable curing covers such as polyethylene sheeting. During placing and fin- ishing operations, the use of a fog mist applied just above the 29

concrete surface from the upwind side of the work is also very effective. Use of a commercial grade fog nozzle is required to provide broad coverage and produce a fine mist that does not damage the concrete by depositing drops of water on the sur- face. Dry cotton blankets can be applied to the plastic con- crete and wetted immediately after concrete finishing (even before the concrete can support the weight of workers). The wetted blankets can then be covered with soaker hoses and plastic sheeting. This method provides immediate curing and should eliminate most plastic cracking problems. Plastic shrinkage cracks sometimes form in winter condi- tions when concrete is cast outside in heated forms and covered with a tarpaulin. The forms and the warm concrete heat the air between the concrete and the tarpaulin, greatly increasing the air’s capacity to hold moisture. This warm, moist air leaks out and is replaced by cold, dry air that quickly warms up and absorbs moisture from the concrete, possibly leading to plastic shrinkage cracking. Test Methods for Plastic Shrinkage Cracking [CR9]. Testing of concrete for plastic cracking is uncommon. One test method has been developed for evaluating the effect of fibers on plastic shrinkage cracking. This method is described in ICC Evaluation Service (ICC-ES) report “Acceptance Criteria for Concrete with Synthetic Fibers” (ICC AC32 Annex A). While testing for plastic cracking is not normally suggested as part of mixture development programs, if prob- lems occur, comparative testing may be helpful in identifying better performing mixtures; a desirability function curve can be proposed. Effect of SCMs on Plastic Shrinkage Cracking. Plastic shrinkage cracking has become a significant problem in recent years because of the wide usage of silica fume, latex modifiers, superplasticizers (HRWRs), and air-entraining admixtures. HRWRs greatly reduce the water content and therefore the bleeding capacity of concrete. As a result, the rate of evapora- tion can more easily exceed the rate of bleeding. Silica fume intensifies the problem because HRWRs must be used to com- pensate for the extreme fineness of the silica fume material. The HRWR reduces the amount of bleed water available while the high fineness of the silica fume reduces the rate at which the water can move through the concrete. Air entrainment also reduces the bleeding rate. Special Construction Practices [CR10] Krauss and Rogalla (47) identify practices that will reduce the tendency for deck cracking: (1) reduce concrete restraint (structural design), (2) reduce rapid heat of hydration (mix- ture), (3) reduce the rate of concrete cooling (construction procedures), (4) reduce the concrete coefficient of thermal expansion (aggregate selection), (5) reduce drying shrinkage (materials), (6) reduce the rate of drying (construction prac- tice), (7) increase concrete creep (mixture), and (8) reduce the concrete modulus (materials). Krauss and Rogalla provide additional information con- cerning early-age deck cracking (47). Also, the software pack- age HIPERPAV (HIgh PERformance PAVing) that was recently developed through the FHWA (54) identifies the risk of early-age problems based on portland cement concrete pavement design, concrete mixtures, construction methods, and environmental conditions. Although the concrete mixture proportions and specific construction practices can affect cracking, key elements of the design also can greatly influence cracking. Design factors such as the geometry of the deck and the size, spacing, and type of supporting girders can have a major effect on the amount of cracking. In some cases, these design features can lead to cracking despite the construction practices. Restrained ring tests can be used to measure the cracking tendency of concrete mixtures, evaluate proposed mixtures, and select mixtures having a low cracking tendency. Instrumenting concrete placements to monitor concrete strains and concrete temper- atures (at the top and middle of the slab) can also help in iden- tifying causes of and reducing cracking in future applications. Restraint can cause large shrinkage and thermal stresses, but creep (stress-relaxation) serves to offset this stress devel- opment. This relationship is illustrated in the following sim- plified one-dimensional example. If the concrete has a free shrinkage of 500 microstrain (με), but it is restrained and allowed to shorten only 250 με, the restraint is 50%. Concrete with a modulus of elasticity of 4 × 106 psi (28 GPa) might have an effective modulus of only 2 × 106 psi (14 GPa), because of creep. The resultant tensile stress would be the product of the strain (500 με) times the restraint (50%) times the effective modulus of elasticity [2 × 106 psi (14 GPa)] for a resultant ten- sile stress of 500 psi (3.4 MPa). If the tensile strength of the concrete is greater than 500 psi (3.4 MPa), cracking will not occur. However, additional tensile stresses that may result from thermal gradients or loading could crack such a con- crete. Therefore, the effects of shrinkage and temperature changes, effective concrete modulus, restraint conditions, tensile strength, and loading conditions must be considered during the process to reduce deck cracking. Concrete Mixture Influences on Cracking. The mixture and raw materials of concrete significantly affect cracking. Generally, high-strength concrete is more prone to cracking. These concretes are stiffer (higher elastic modulus) and de- velop higher stresses for a given temperature change or amount of shrinkage, and, most important, these concretes creep (relax) little to relieve these stresses. However, these concretes also develop greater tensile strengths, and the 30

interaction between these properties is complex. Typically, high-early-strength concretes are particularly prone to crack- ing because little shrinkage has dissipated before the concrete has developed a high modulus and low creep properties. Also, because high-strength concretes typically contain more cement, they may shrink more and develop higher tempera- tures during early hydration. The risk of cracking may be reduced by selecting a concrete mixture that does not excessively exceed the required com- pressive strength yet is still durable in service. Cement content, fineness, and chemical composition also affect the rate of hydration, early strength gain, and the heat generated initially by the concrete. Modern cements are more apt to cause cracking because they are finer and have higher sulfate and alkali contents. In general, concretes with higher aggregate contents and lower cement paste contents are less likely to develop cracks. Because the hydrated cement paste is the component of the concrete that shrinks, reducing cement paste volume reduces shrinkage. Leaner mixtures are also thermally less expansive and develop smaller thermal stresses. The concrete mixture should contain the largest possible aggregate size consistent with placement and consolidation requirements. Larger aggregates permit a leaner mixture, help maintain workabil- ity, and reduce thermal and shrinkage stresses. The maximum aggregate size should be either one-third the deck thickness or three-fourths the minimum clear spacing between reinforcing bars, whichever is smaller. Concrete Placement. Mixtures having a high cracking tendency may be selected based on good performance in other tests. If this is the case, measures should be taken to reduce the risk of cracking during construction. The first large stresses in a new concrete slab usually develop during the first 12 to 24 hours, when the concrete temperatures change rapidly during early hydration. Reducing the concrete temperatures during this process will reduce early stresses. Concrete temperatures can be reduced by placing concrete during cooler periods (such as during the evening or at night), placing cool concrete, misting the concrete during placement, wet curing, and shading the deck. Plastic shrinkage cracking can occur when SCMs are used and the concrete has little bleed water. When wind speeds are higher than 5 mph (light breeze) during placement, moisture evaporation rates from the concrete should be measured and special precautions to reduce drying should be taken if evap- oration rates are high. For normal concreting, an evaporation rate of 0.2 pounds per square foot (psf) per hour (1.0 kg/m2 per hour) is considered excessive, while, for modern concretes with high cement contents, silica fume, HRWRs, or other ingredients that reduce bleed water, an evaporation rate of 0.1 to 0.15 psf per hour (0.5 to 0.75 kg/m2 per hour) is considered excessive. In summary, the ideal concrete mixture should incorpo- rate the largest practical aggregate size, lowest paste volume consistent with other performance requirements, and mini- mum strength required to meet project requirements; gain strength slowly; and contain the highest w/cm that will pro- vide the required strength and durability. A cement with a low rate of heat evolution and the use of SCMs should be considered to achieve a concrete that gains strength slowly but steadily. 31

Worksheet for Step 1 Worksheet S1.1. Desired concrete performance and associated considerations for mixture proportioning. Environment Property/Test Method Target Value for Test Method Range of Class C Fly Ash Range of Class F Fly Ash Range of GGBFS Range of silica fume Range of other SCM w/cm Aggregate restrictions Specified aggregate top size Specified cement content Other requirements Compressive strength: AASHTO T 22, ASTM C 39 Flexural strength: AASHTO T 177, T 97, or T 198, ASTM C 293, C 78, or C 496 Slump and slump loss: AASHTO T 119, ASTM C 143 Time of setting: AASHTO T 197, ASTM C 403 Universal performance requirements Finishability Chloride penetration: AASHTO T 259, ASTM C 1566 Electrical conductivity: AASHTO T 277, ASTM C 1202 Freezing and thawing with chemical deicers Scaling Resistance: ASTM C 672 Air content, %: ASTM C 457 Spacing factor: ASTM C 457 Freezing and thawing without chemical deicers Freezing and thawing resistance: AASHTO T 161 A, ASTM C 666 A Chloride penetration: AASHTO T 259, ASTM C 1566 Coastal Electrical conductivity: AASHTO T 277, ASTM C 1202 Abrasive Abrasion: ASTM C 944 or C 779 Procedure B Cracking resistance: ASR Go to Raw Materials Flowchart Restrained ring cracking: AASHTO PP 34-99, ASTM C 1581 Cracking resistance: restrained shrinkage Free drying shrinkage: AASHTO T 160, ASTM C 157 Heat of hydration Cracking resistance: thermal concerns Modulus of elasticity, ASTM C 469 Cracking resistance: plastic shrinkage Plastic shrinkage cracking: ICC AC32 Annex A Other design requirements SUMMARY

Figures for Step 1 Concrete Service Environment In a Freezing Climate? [F1] Subject to Chemical Deicers? [CL1] In a Coastal Environment? [CO1] In an Abrasive Environment? [AB1] Is Cracking Resistance a Concern? [CR1] Is there Restrained Shrinkage? [CR2] Thermal Concerns? [CR5] Go to Raw Materials Flow Chart in Step 2 Cracking Tendency [CR4]: AASTHO PP34-99 (ASTM C 1581) Drying Shrinkage: [CR3]: AASHTO T 160 (ASTM C 157) Plastic Shrinkage Cracking? [CR8] Best Practice [CR10] Heat of Hydration [CR6] Related Factors [CR7]: Coefficient of Thermal Expansion Modulus of Elasticity ASTM C 469 ICC AC32 Annex A [CR9] Universal Performance Requirements Freeze/Thaw Resistance [F3]: AASHTO T 161 Procedure A (ASTM C 666 A) Air Entrainment [F2]: Air Analysis in Hardened Concrete ASTM C 457 Chloride Penetration Resistance [CL2]: AASHTO T 259, ASTM C 1566 and AASHTO T 277 (ASTM C 1202) Chloride Penetration Resistance [CL2]: AASHTO T 259, ASTM C 1566 and AASHTO T 277 (ASTM C 1202) Scaling Resistance [S1]: ASTM C 672 Resistance to Abrasion [AB2]: ASTM C 779 Procedure B [AB3] Or ASTM C 944 [AB3] Is ASR a Concern? [ASR1] Yes No No No No No No No No No Yes Yes Yes Yes Yes Yes Yes AN D /O R Workability and Finishability [D3]: Slump: AASHTO T 119 (ASTM C 143) Slump Loss Setting Time: AASHTO T 197 (ASTM C 403) Finishability: Qualitative Comparison AN D AN D /O R Concrete Strength [D1, D2]: Compressive: AASHTO T 22 (ASTM C 39) Flexural: AASHTO T 177 (ASTM C 293) or AASHTO T 97 (ASTM C 78) Splitting Tensile: AASHTO T 198 (ASTM C 496) Yes Figure S1.1. Selecting concrete service environment and properties.

34 0 0.2 0.4 0.6 0.8 1 0 1000 2000 3000 4000 5000 6000 7000 8000 Compressive Strength (psi) De s ira bi lit y 0 0.2 0.4 0.6 0.8 1 0 2000 4000 6000 8000 10000 12000 Compressive Strength (psi) De s ira bi lit y 0 0.2 0.4 0.6 0.8 1 0 2000 4000 6000 8000 10000 12000 Compressive Strength (psi) De s ira bi lit y Figure S1.2. Desirability function for average compressive strength at 7 days. Figure S1.3. Desirability function for average compressive strength at 28 days. Figure S1.4. Desirability function for average compressive strength at 56 days.

35 0 0.2 0.4 0.6 0.8 1 400 500 600 700 800 900 Modulus of Rupture (psi) De s ira bi lit y 0 0.2 0.4 0.6 0.8 1 0 2 4 6 8 10 12 Slump (in.) De s ira bi lit y 0 0.2 0.4 0.6 0.8 1 0 2 4 6 8 10 Slump loss (in.) D e s ira bi lit y Figure S1.5. Desirability function for modulus of rupture (flexural strength). Figure S1.6. Desirability function for slump (HRWR is used). Figure S1.7. Desirability function for slump loss.

36 0 0.2 0.4 0.6 0.8 1 0 5 10 15 20 25 30 Initial Setting (hrs.) De s ira bi lit y 0 0.2 0.4 0.6 0.8 1 0 5 10 15 20 25 Finishability Rating De s ira bi lit y 0 0.2 0.4 0.6 0.8 1 0 2 4 6 8 10 12 14 Air Content (%) De s ira bi lit y Severe Exposure Moderate Exposure Figure S1.8. Desirability function for time of initial setting. Figure S1.9. Desirability function for finishability. Figure S1.10. Desirability functions for air content.

37 0 0.2 0.4 0.6 0.8 1 0.004 0.008 0.012 0.016 0.02 Spacing Factor De s ira bi lit y With HRWR, w/cm < 0.38 No HRWR 0 0.2 0.4 0.6 0.8 1 0 200 400 600 800 1000 Specific Surface Area (in2/in3) De s ira bi lit y 0 0.2 0.4 0.6 0.8 1 40 50 60 70 80 90 100 110 Freeze-thaw Durability Factor (%) D es ira bi lit y After 300 cycles After 500 cycles Figure S1.11. Desirability functions for spacing factor with and without HRWR. Figure S1.12. Desirability function for specific surface area. Figure S1.13. Desirability functions for durability factor in a severe environment after 300 cycles and after 500 cycles.

38 0 0.2 0.4 0.6 0.8 1 40 50 60 70 80 90 100 110 Freeze-thaw Durability Factor (%) D e s ira bi lit y After 300 cycles After 500 cycles 0 0.2 0.4 0.6 0.8 1 0 1 2 3 4 5 Scaling Rating D e s ira bi lit y Severe Exposure Moderate Exposure 0 0.2 0.4 0.6 0.8 1 0 200 400 600 800 1000 1200 Scaling mass loss (g/m2) D e s ira bi lit y Severe Exposure Moderate Exposure Figure S1.14. Desirability function for durability factor in moderate environment after 300 and 500 cycles. Figure S1.15. Desirability function for visual rating of scaling for severe exposure. Figure S1.16. Desirability function for mass loss for scaling for severe exposure.

39 0 0.2 0.4 0.6 0.8 1 0 2 4 6 8 10 12 14 Apparent Chloride Diffusion Coefficient (x 10-12 m2/s) De s ira bi lit y 0 0.2 0.4 0.6 0.8 1 0 1000 2000 3000 4000 5000 6000 Charge Passed (Coulombs) De s ira bi lit y 0 0.2 0.4 0.6 0.8 1 0 2 4 6 8 10 12 14 16 Mass loss after six mintues (g) De s ira bi lit y Figure S1.17. Desirability function for chloride ion penetration test. Figure S1.18. Desirability function for electrical conductivity test. Figure S1.19. Desirability function for mass loss after abrasion (ASTM C 944).

40 0 0.2 0.4 0.6 0.8 1 0 0.5 1 1.5 2 2.5 Depth of wear after six mintues (mm) De s ira bi lit y 0 0.2 0.4 0.6 0.8 1 -0.12 -0.1 -0.08 -0.06 -0.04 -0.02 0 Shrinkage (% strain) De s ira bi lit y 0 0.2 0.4 0.6 0.8 1 0 5 10 15 20 25 30 Age at cracking (wks.) De s ira bi lit y Figure S1.20. Desirability function for depth of wear after abrasion (ASTM C 944). Figure S1.21. Desirability function for drying shrinkage after 90 days. Figure S1.22. Desirability function for time to first crack in restrained-ring shrinkage cracking test.

41 0 0.2 0.4 0.6 0.8 1 0 20 40 60 80 100 Temperature Rise (°F) De s ira bi lit y 0 0.2 0.4 0.6 0.8 1 0 1 2 3 4 5 6 7 8 9 Modulus of Elasticity (x106 psi) De s ira bi lit y Figure S1.23. Desirability function for temperature rise due to heat of hydration. Figure S1.24. Desirability function for modulus of elasticity at 7 days.

42 Tables for Step 1 Environment Property/Test Method Range of Class C Fly Ash (%) Range of Class F Fly Ash (%) Range of GGBFS (%) Range of Silica Fume (%) Range of Metakaolin (%) Compressive strength: AASHTO T 22, ASTM C 39 0-30 0-30 15-50 5-8 5-15 Flexural strength: AASHTO T 177, T 97, or T 198, ASTM C 293, C 78, or C 496 0-30 0-30 15-50 5-8 5-15 Slump and slump loss: AASHTO T 119, ASTM C 143 10-30 10-40 15-40 5-8 5-10 Time of setting: AASHTO T 197, ASTM C 403 0-30 0-25 15-40 5-8 5-15 Universal performance requirements Finishability 0-25 0-25 10-30 0-8 5-15 Chloride penetration: AASHTO T 259, ASTM C 1566 15-40 15-25 15-30 5-8 8-12 Electrical conductivity: AASHTO T 277, ASTM C 1202 15-40 15-25 15-30 5-8 8-12 Freezing and thawing with chemical deicers Scaling resistance: ASTM C 672 0-25 0-25 0-40 5-8 0-10 Air content, %: ASTM C 457 0-25 0-25 0-40 0-8 0-10 Spacing factor: ASTM C 457 0-25 0-25 0-40 0-8 0-10 Freezing and thawing without chemical deicers Freezing and thawing resistance: AASHTO T 161 A, ASTM C 666 A 0-25 0-25 0-40 5-8 5-10 Chloride penetration: AASHTO T 259, ASTM C 1566 15-40 15-25 15-30 5-8 8-12 Coastal Electrical conductivity: AASHTO T 277, ASTM C 1202 15-40 15-25 15-30 5-8 8-12 Abrasive Abrasion: ASTM C 944 or C 779 Procedure B 0-25 0-25 0-25 5-8 Unknown Cracking resistance: ASR Go to Raw Materials Flowchart Not recommended >25* >40* 5-8** 10-15 Restrained ring cracking: AASHTO PP 34-99, ASTM C 1581 10-25 10-25 15-35 0-5 0-10 Cracking resistance: restrained shrinkage Free drying shrinkage: AASHTO T 160, ASTM C 157 0-25 0-25 0-35 0-5 0-10 Heat of hydration 0-25 25-40 30-75 0-8 Unknown Cracking resistance: thermal concerns Modulus of elasticity, ASTM C 469 0-30 10-30 15-35 0-5 0-5 Cracking resistance: plastic shrinkage Plastic shrinkage cracking: ICC AC32 Annex A 0-25 0-25 0-25 0-5 Unknown *If ASR is an issue, the minimum quantity of SCM for ASR shall prevail over quantities for other properties **Combinations of silica fume and other SCMs have been found to be effective for controlling ASR Table S1.1. Summary of typical SCM ranges of use for each concrete property.

43 Property Recommendation Aggregate Good quality w/cm < 0.45 Minimum cement content 564 lbs/yd3 (335 kg/m3) Compressive strength when exposed to freezing Minimum 2,500 psi (17 MPa), minimum 4000 psi (28 MPa) if critically saturated Curing Minimum 7 days controlled wet curing Air entrainment See Kosmatka et al. (21), Tables 9-5 and 9-15 Rating Condition of Surface 0 No scaling 1 Very slight scaling (3 mm [1/8 in.] depth max., no coarse aggregate visible) 2 Slight to moderate scaling 3 Moderate scaling (some coarse aggregate visible) 4 Moderate to severe scaling 5 Severe scaling (coarse aggregate visible over entire surface) Table S1.2. Mixture considerations for freezing and thawing resistance. Table S1.3. ASTM C 672 surface rating conditions. Property Recommendation Aggregates Minimal popouts (low % of porous particles) Minimum cement content 564 lb/yd3 (335 kg/m3) Air void system Adequate or better w/cm < 0.45 Finishing No hard trowel Minimum compressive strength 3500 psi (24 MPa) Table S1.4. Other mixture considerations for salt scaling resistance. Property Recommendation Aggregates Hard Compressive strength, 56 days Maximize woL mc/w Table S1.7. Other mixture considerations for abrasion resistance. Property Recommendation w/cm < 0.40 Slump > 3 in. (75 mm) Property Recommendation Cover, minimum 2 in. (50 mm) w/cm 0.37-0.40 Table S1.5. Other mixture considerations for reducing chloride penetration. Table S1.6. Mixture considerations for reducing chloride penetration in coastal environments.

Environment Property/Test Method Target Value for Test Method Range of Class C Fly Ash Range of Class F Fly Ash Range of GGBFS Range of silica fume Range of other SCM w/cm Aggregate restrictions Specified aggregate top size Specified cement content Other requirements Compressive strength: AASHTO T 22, ASTM C 39 Flexural strength: AASHTO T 177, T 97, or T 198, ASTM C 293, C 78, or C 496 Slump and slump loss: AASHTO T 119, ASTM C 143 Time of setting: AASHTO T 197, ASTM C 403 Universal performance requirements Finishability Chloride penetration: AASHTO T 259, ASTM C 1566 Electrical conductivity: AASHTO T 277, ASTM C 1202 Freezing and thawing with chemical deicers Scaling resistance: ASTM C 672 Air content, %: ASTM C 457 Spacing factor: ASTM C 457 Freezing and thawing without chemical deicers Freezing and thawing resistance: AASHTO T 161 A, ASTM C 666 A Chloride penetration: AASHTO T 259, ASTM C 1566 Coastal Electrical Conductivity: AASHTO T 277, ASTM C 1202 Abrasive Abrasion: ASTM C 944 or C 779 Procedure B Cracking resistance: ASR Go to Raw Materials Flowchart Restrained ring cracking: AASHTO PP 34-99, ASTM C 1581 Cracking resistance: restrained shrinkage Free drying shrinkage: AASHTO T 160, ASTM C 157 Heat of hydration Cracking resistance: thermal concerns Modulus of elasticity, ASTM C 469 Cracking resistance: plastic shrinkage Plastic shrinkage cracking: ICC AC32 Annex A Other design requirements SUMMARY Table S1.8. Completed Worksheet S1.1 for the Hypothetical Case Study.