Final Report for Early-Opening-to-Traffic Portland Cement Concrete for Pavement Rehabilitation (2005)

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.

45 CHAPTER 3 EXPERIMENTAL PROGRAM AND RESULTS The previous chapter explained that premature deterioration in EOT concrete mixtures result from the altered microstructure and/or microcracking in the hydrated cement paste caused by rapid hydration. The altered or cracked microstructure increases permeability, paste porosity, and solubility of paste constituents, thus negatively affecting the concrete durability. Also, the altered or cracked microstructure may result in paste expansion from delayed ettringite formation, as the temperatures achieved during hydration can exceed 70oC. The factors that have the greatest influence on EOT concrete durability are those that impact shrinkage, thermal stress, altered microstructure, and chemical attack. The selection of the constituent materials, mix design/proportioning, construction, curing, and age at opening-to-traffic will influence these properties and the durability of the resulting EOT concrete. Earlier research has shown that achieving early-age mechanical properties (strength, abrasion, etc.) with EOT concrete materials is not difficult. Therefore, this project focused on durability issues related to the interaction between the EOT concrete materials and the environment in which they serve and not issues related to the effect of traffic loading. Addressing durability issues is a difficult task because no universally accepted tests that “measure” durability are available. Durability is a function of the material properties and the environment in which the material serves. For example, a non–air-entrained concrete might make a durable indoor floor, but will likely not be durable if exposed to freezing and thawing in the

46 presence of deicer applications. Similarly, concrete with poor sulfate resistance might be extremely durable in a non-sulfate environment, but might perform poorly if exposed to an external source of sulfate ions. The research included a limited field evaluation and a laboratory evaluation of EOT concrete mixtures. The results of the laboratory research were used to develop guidelines to assist engineers and contractors in selecting appropriate materials, mixtures, and construction techniques for use in EOT concrete repairs. This chapter discusses the field and laboratory phase of this study. 3.1 FIELD EVALUATION It was not possible to develop a statistically valid experiment for the field evaluation because the data for many repair installations were sparse and of questionable quality and because researchers could not locate repairs that met all the requirements specified in a factorial experimental plan. For this reason, an alternative approach was followed for the field evaluations that recognized these limitations while yielding information useful for this study. Various SHAs were contacted to identify those agencies that had broad material and climatic representation. The following general conclusions were drawn from discussions with these agencies: • Most SHAs had both 6- to 8-hour and 20- to 24-hour EOT PCC mixtures. • Other than the experimental sites constructed as part of the SHRP studies (Whiting et al. 1994), none of the SHAs had any extant experimental sites that could be evaluated in the

47 course of this project. Further, although all SHAs had mixture design information available, few maintained detailed construction records on repair installations. • None of the SHAs tracked performance of their full-depth repairs, thus no performance data were available. However, the SHAs reported their general feelings about full-depth repairs. Most SHAs felt that they had both “good” performing and “poorly” performing 6- to 8-hour EOT concrete repairs, but felt that the 20- to 24-hour repairs were performing satisfactorily. • In all cases, SHAs offered to provide assistance to this research effort by providing traffic control and coring. On the basis of this information, it was decided to seek assistance from one SHA in each of the four long-term pavement performance (LTPP) climatic zones. A brief description of each test site is presented in the following section. Photographs from each site and the results of the visual assessment are presented in Appendix B. Table 2 lists the EOT concrete mixture design parameters for the mixtures used at the selected sites. 3.1.1 Experimental Sites Interstate 20, West of Augusta, Georgia Eastbound Interstate 20, a four-lane divided highway, contained full-depth EOT concrete repairs placed as part of the Strategic Highway Research Program studies (Whiting et al. 1994). Two mixtures placed during the SHRP study selected for inclusion as part of the field investigation in this experiment are known as the Fast Track or FT1 (4 to 8 hour) and GA DOT

48 mixtures (20 to 24 hour). Both mixtures were placed in July 1992, making the repairs 9 years old at the time of this investigation. Georgia Department of Transportation (GA DOT) personnel indicated that a number of patches had been removed and replaced because of poor performance, making it difficult to locate “poor” performing repairs for sampling operations for this study. In addition, the pavement had been diamond ground, making it difficult to visually assess surface defects although GA DOT officials indicated that surface cracking of the patches was present prior to the diamond grinding of the pavement. Of the four patches chosen for sampling in this project, only one patch was found to be performing poorly, containing a longitudinal crack across the full length of the patch. There was no staining evident at the crack location; however, the crack appeared to be beginning to spall at some locations. The remainder of the patches sampled appeared to be performing satisfactorily.

49 TABLE 2 Specifications for field mixtures included in study SHA Mixture Year Const. Cement Type Cement Factor w/c ratio Coarse Aggregate Fine Aggregate Accelerator Air Water Reducer FT1 1992 III 439 kg/m 3 (740 lb/yd3) 0.35 842 kg/m3 (1420 lb/yd3) 783 kg/m3 (1320 lb/yd3) None 5.6% Darex Type A GA GA DOT 1992 I 446 kg/m 3 (750 lb/yd3) 0.38 1071kg/m3 (1805 lb/yd3) 608 kg/m3 (1025 lb/yd3) CC2 (0.5%) 3.7% Darex No 18502 1993 III 490 kg/m 3 (825 lb/yd3) 0.39 849 kg/m3 (1430 lb/yd3) 758 kg/m3 (1280 lb/yd3) CC (2.5%) 4.5 to 6.0% Daravair No NY Modified Class F 2000 I 418 kg/m3 (705 lb/yd3) 0.39 1124 kg/m3 (1895 lb/yd3) 602 kg/m3 (1015 lb/yd3) NS 6.5% Type A Class FS 1992 III 534 kg/m 3 (900 lb/yd3) 0.41 843 kg/m3 (1420 lb/yd3) 593 kg/m3 (1000 lb/yd3) CC (2%) 4.5% Type D OH Class MS1 1993 I 456.5 kg/m 3 (770 lb/yd3) 0.30 754 kg/m3 (1270 lb/yd3) 848 kg/m3 (1430 lb/yd3) CC (0.5%) 7.5% AE-360 Type F & Type D Class K 1996 III 418 kg/m 3 (705 lb/yd3) 0.39 1144 kg/m3 (1930 lb/yd3) 636 kg/m3 (1070 lb/yd3) Type C NC 3.0 to 6.0% AE-90 Type A TX Class K Modified 1998 I 390 kg/m3 (658 lb/yd3) 0.40 1137 kg/m3 (1915 lb/yd3) 597 kg/m3 (1005 lb/yd3) Type C NC 3.0 to 6.0% Air 30 Type A 1 This mixture includes 415-kg cement and 41.5-kg AXIM microsilica per cubic meter (700 lb cement and 70 lb microsilica per cubic yard). 2 CC: calcium chloride; NC: non-chloride; NS: not specified. Long Island Expressway (US 454) in Suffolk County, Long Island, New York The New York Department of Transportation (NY DOT) performed full-depth EOT concrete repairs for 4- to 8-hour opening on a six-lane divided highway (Long Island Expressway [US 454] in Suffolk County). Repairs in this section were placed during the summer of 1993 using a concrete mix specified under the NY DOT specifications as Item 18502.6027. Information on the fresh concrete properties are not available; however, all mixtures placed were reported to have met specification tolerances. One unique aspect of this mixture is that the temperatures of the aggregate are taken prior to batching to determine the temperature of the heated mixing water to be added. The heated

50 mixing water is added to a tank on the truck, where it remains until the truck arrives on site. This water and the calcium chloride solution are then added just prior to placement in order to achieve a concrete temperature of 32 to 38oC (90 to 100oF) at time of placement. The opening criteria for these repairs was based on temperature, being opened to traffic when the surface temperature reaches 65oC (150oF), which has been found to correspond to a compressive strength of approximately 13.8 MPa (2,000 psi). Peak hydration temperatures for these mixes can reach 82oC (180oF). The visual assessment of the repairs and of the surrounding pavement prior to sampling operations found the repairs to be in good to excellent condition, with only one of the patches exhibiting a transverse crack at mid-panel. Almost all of the patches surveyed displayed slight map cracking on their surface. The surrounding pavement was in generally good condition, with some spalling occurring at joint locations in the pavement structure. Based on these observations, two repairs were selected for coring operations, one of which was performing satisfactorily, whereas the other exhibited mid-panel transverse cracking with some brown staining present. Interstate 390, Monroe County, South of Rochester, New York The NY DOT Modified Class F EOT concrete materials were obtained from a four-lane divided Interstate (Northbound Interstate 390). The repairs in this section were placed during August of 2000. Fresh and hardened concrete properties were recorded by representatives of the NY DOT.

51 Selection of repairs and assessment of patch condition were conducted by representatives of the NY DOT. Records indicate that one satisfactorily performing repair and one unsatisfactorily performing repair were sampled. The unsatisfactorily performing repair contained a transverse crack across the entire slab width. This crack is believed to be a thermal crack resulting from the overall patch length of 6 m (20 ft). State Route 2, West of Cleveland, Ohio The Ohio DOT Class FS test site was on Westbound State Route 2, a four-lane divided highway west of Cleveland, Ohio. This site was constructed as part of the SHRP studies during September of 1992. The Class MS site was constructed on the same section of State Route 2 in September of 1993. Although not part of the SHRP study, it was part of an Ohio Department of Transportation (ODOT) project; thus, detailed construction data were available. The opening strength for the Class FS mixtures ranged from 7.6 to 36.5 MPa (1,100 to 5,300 psi) as determined from both insulated and uninsulated cylinder specimens, and the peak measured hydration temperature for the ODOT mix was nearly 74οC (165oF) (Whiting et al. 1994). Ohio specifications stipulate that the MS mixtures may be opened to traffic after 24 hours, provided that a flexural strength of 2.76 MPa (400 psi) is met. Surveys of the Class FS repairs were conducted 2 months after placement. Results from the inspection revealed that nearly all the FS patches had cracked longitudinally at this time. SHRP studies found that the higher the peak curing temperature of a repair, the more likely for

52 that repair to exhibit longitudinal cracking in the 2-month survey (Whiting et al. 1994). Additional surveys of the Class FS repairs were conducted annually from 1994 to 1998. All of the repairs eventually contained longitudinal cracking, whereas two out of nine repairs had transverse cracking at the end of 1998. A visual assessment was made of the patches prior to core sampling; however, little information was available on the existing condition of the concrete because the pavement had been overlaid with asphalt. Despite these challenges, patches were located through the use of milepost stations and reflective cracking visible at the joints. Although surface deterioration of the concrete could not be directly assessed, in many cases manifestation of joint deterioration and cracking could be seen in the surface of the overlay. Northbound US Interstate 81, Wise County, Texas The 6- to 8-hour EOT concrete repairs on northbound US Interstate 81 in Wise County, Texas (a four-lane divided highway), were evaluated. The repairs, constructed in 1996, were made on the continuously reinforced concrete pavement with a Texas DOT Class K mix. Fresh concrete properties were documented by the Texas DOT; however, information for the individual patches included in this study was not available. A visual distress survey was conducted of the patches and surrounding pavement prior to selection of individual patches for core sampling operations. As is typical with continuously reinforced concrete pavements, the surrounding pavement surface contained hairline transverse

53 cracks at regularly spaced intervals. Several of the patches also contained transverse cracks at mid-panel, and many showed signs of slight joint faulting and moderate-severity joint spalling. In addition, one of the patches also contained a corner break. Frontage Road, Texas 2871, Tarrant County, Texas This 20- to 24-hour EOT concrete test site was a frontage road connecting US Interstate 20 to Texas 2871 in Tarrant County, Texas. The repairs on this continuously reinforced concrete pavement were constructed during April and May of 1998 using a Texas DOT Class K modified mix. Although the fresh concrete properties were documented for the project, information for the individual repairs was not available. A visual distress survey was conducted of the patches and surrounding pavement prior to selection of individual patches for core-sampling operations. As is typical with continuously reinforced concrete pavements, the surrounding pavement surface contained small transverse cracks at short, regularly spaced intervals. All patches surveyed appeared to be performing satisfactorily, which made it impossible to select a good and bad patch for this location. As a result, two satisfactorily performing patches were selected for coring operations. 3.1.2 Sampling of Test Sites Using the results from the visual survey, two repairs were selected from each test site, with few exceptions, one rated as “satisfactory” and the second as “unsatisfactory.” Four

54 150-mm (6-in.)-diameter cores were then obtained from each repair, for a total of eight cores for each mixture type being evaluated. Coring locations were based on whether the repair was rated as “satisfactory” or “unsatisfactory.” The only difference between the two sampling locations is that one core was taken in a cracked/deteriorated area in the “unsatisfactory” repair. Core Sample A was obtained at roughly the geometric center of the patch. This location was selected to evaluate the influence of hydration temperatures on the concrete, as this location is where the highest temperature within the repair would be expected to occur. Core Sample B was obtained on the inside edge of the repair at the transverse joint between the dowels. This location was selected to determine the effects of moisture and chemical ingress at the joints. Core Sample C was taken at the approximate location of the outside wheel path, which was determined in the field by examining the concrete surface for evidence of wear or by selecting a location 0.6 m (24 in.) in from the shoulder line. The final sample, Core Sample D, was obtained at an interior location. In the case of an “unsatisfactory” repair, the sample was taken in a cracked or deteriorated location, representing an area of the typical distress present for that repair. Details regarding the core locations for each repair are provided in Appendix B. 3.1.3 Approach to Laboratory Testing of Field Specimens The analysis of the field-collected specimens included pulse velocity testing, microstructural characterization, absorption/sorptivity testing, and the determination of the CTE, as shown in Table 3. Of the four cores collected from each repair, two were used for

55 microstructural characterization and two were used in the pulse velocity testing, in the absorption/sorptivity testing, and to determine the CTE in accordance with AASHTO TP 60-00. Pulse Velocity Testing Two specimens from each material/condition combination were tested using pulse velocity. The procedure for conducting pulse velocity testing required providing a smooth surface at each end of the core sample. Pulse velocity measurements were conducted on 100-mm (4-in.)-diameter specimens that were cored from the 150-mm (6-in.)-diameter specimens obtained from EOT concrete repairs. A petroleum-based jelly was applied to the ends of the cores to improve contact between the transmitter/receiver and the test specimen. Three tests were conducted on each sample using a Jones Instruments NDT 3000 apparatus that generated a compression wave (P wave) from a transmitter with an energizing pulse of 500 volts. To minimize error produced by vibration or movement of the testing apparatus transmitter and receiver, a bracket was developed to hold these instruments in place at the ends of the sample. Microstructural Characterization Two specimens from each material/condition combination were evaluated microstructurally using four separate test methods: staining, stereo microscopy, petrographic microscopy, and limited scanning electron microscopy (SEM). Staining techniques were performed to identify the depth of carbonation and the degree of voids/cracks infilling with ASR gel and/or sulfate minerals. Such observations help to assess the failure mechanisms. Stereo

56 optical microscopy was used to make a general assessment of the concrete’s condition and to determine the characteristics of the air-void system using ASTM C 457 procedures. An analysis of the amount of air void infilling with secondary deposits (ettringite, calcite, ASR gel, etc.) was also made to determine the contribution of paste freeze-thaw damage and other concrete durability problems. TABLE 3 Field evaluation tests (number of samples per material/condition combination) Test Attribute Test Name/ Equipment Measured Property No. of Specimens Specification and Laboratory1 Pulse Velocity Pulse Velocity Non-destructive estimation of relative stiffness 2 Non-Standard (MTU) Staining Techniques Staining of carbonated paste, ASR gel and sulfate minerals 2 No Standard Test (MTU) Stereo Optical Microscopy Air-void system parameters 22 ASTM C 457 (MTU) Petrographic Optical Microscopy with UV dye impregnation Microstructure analysis, reaction products, and evidence of deterioration 22 ASTM C 856 (MTU) Microstructural Characterization Scanning Electron Microscopy (SEM) Microstructure analysis, elemental mapping 22 No Standard Test (MTU) Volume Change CTE CTE 23 AASHTO TP 60-00 (MSU) Total Specimens per Treatment 4 1 The agency responsible for running each test is shown in parentheses. 2 Single cores were cut in such a way that specimens were obtained for staining, stereo optical microscope, petrographic optical microscope, and SEM. 3 Tests were performed on the same specimens used for pulse velocity testing. Petrographic microscopy was used to examine the concrete microstructure. Mineralogical identification, evidence of paste expansion, and microcracking were characterized to identify causes of deterioration. The petrographic analysis was also used to assess paste porosity. The

57 SEM was applied in limited cases to identify microstructural features that were not easily resolved using optical microscopy. Determination of CTE The CTE was determined on the field-obtained core specimens using the provisional AASHTO Standard Test Method for the Coefficient of Thermal Expansion of Hydraulic Cement Concrete (TP 60-00). The test was performed on the same 100-mm (4-in.)-diameter trimmed cores used for the pulse velocity testing. 3.1.4 Results and Discussion of Field Specimens This section presents and discusses the results of tests conducted on the field-obtained specimens, including the pulse velocity testing, microstructural evaluation, and CTE values. The discussion focuses on how the analysis of these results provides insight into typical EOT concrete mixtures as actually constructed. Pulse Velocity Testing Results Tests conducted on four core specimens from each mixture indicated a good consistency within each mixture for most repairs sampled, as evidenced by the low coefficients of variation values. Detailed results are presented in Appendix B.

58 Pulse velocity values are known to be highly correlated with density of the concrete, the amount of air present in the concrete (which is also related to the concrete density), and the length of the signal path. The relationship between density and pulse velocity for all specimens, plotted in Figure 1, shows a strong correlation exists. Statistical analysis using a t-test conducted on satisfactory and poorly performing samples found statistically significant differences in the density values for satisfactory and poorly performing NY Modified Class F repairs and velocity values of Texas Class K repairs; however, neither mixture contained statistically significant differences in both density and velocity data. Therefore, existence of trends between the density, velocity, and repair condition for mixtures studied has not been demonstrated. Depth-of-Carbonation Testing Depth of carbonation was assessed on two cores from each repair sampled, for a total of four samples per mixture. Carbonation measurements were taken at 10 points across the face of the sample at the surface of the repair and were then averaged to obtain an average depth of carbonation for each repair. Test results revealed no significant carbonation to be present in any of the repairs sampled. Nearly all repairs exhibited some depth of carbonation, with most samples carbonated to a depth of less than 1 mm (0.04 in.) uniformly across the surface. Many samples also revealed localized areas of significant carbonation depth. ODOT repairs were found to exhibit the greatest depth of carbonation, with carbonation extending to a depth in excess of 1 mm (0.04 in.) for at least part of the surface. Samples obtained from repairs that had been

59 diamond ground might not adequately reflect the depth of carbonation because part of the carbonated layer was removed. Detailed test results are presented in Appendix B. Results of t-tests conducted on satisfactory and poorly performing repairs indicate that a significant statistical difference was found only for the NY Modified Class F mixture, with the “poor” performing repairs having less carbonation than repairs performing in a “satisfactory” manner. These results indicate that the depth of carbonation is not a predictor of repair performance.

60 Figure 1. Density versus pulse velocity for all samples tested. Air-Void Analysis Polished samples that had been stained with barium chloride and potassium permanganate were used for air-void system characterization conducted in accordance with ASTM C 457 modified point count method. Samples were placed on an automated stage and viewed at a magnification of 70 times through a stereo optical microscope. Several air-void system parameters were determined and compared with accepted values for a satisfactory air- void system as presented in ASTM C 457. Further analysis of data obtained from these tests was conducted using t-testing to determine statistical significance of differences between key air-void system parameters for satisfactory and poorly performing repairs. y = 0.3196x + 886.83 R2 = 0.7608 2100 2150 2200 2250 2300 2350 2400 2450 2500 2550 3500 4000 4500 5000 5500 Pulse Velocity (m/s) D en si ty (k g/ m ^3 )

61 Over time, air voids in some specimens have filled with secondary deposits, ettringite being the most common. Through the use of the barium chloride/potassium permanganate (BCPP) staining technique, concentrations of sulfate-bearing phases such as ettringite can be identified in the course of the ASTM C 457 testing. With computation of the existing air content, the BCPP staining technique allows for both the estimation of the original air-void system parameters and the assessment of the degree of air-void infilling that may have occurred. Figures 2 through 5 present the impact of infilling on the air-void system in the field specimens. Significant infilling has occurred in the two mixtures from Georgia and in the Ohio Class FS mixture. Negligible infilling was observed in mixtures from New York and Texas.

62 Figure 2. Percent reduction in air content due to secondary ettringite infilling. 0 1 2 3 4 5 GA DOT GA FT1 ODOT Ohio MS NY 18502 NY 20-24 Class K Class K ModifiedP er ce nt R ed uc tio n in A ir C on te nt D ue to S ec on da ry E ttr in gi te

63 Figure 3. Original spacing factors for various field mixtures. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 GA DOT GA FT1 ODOT Ohio MS NY 18502 NY 20-24 Class K Class K Modified sp ac in g fa ct or (m m )

64 Figure 4. Existing spacing factor for various field mixtures. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 GA DOT GA FT1 ODOT Ohio MS NY 18502 NY 20-24 Class K Class K Modified sp ac in g fa ct or (m m )

65 Figure 5. Percent increase in spacing factor for field specimens due to infilling. 0 10 20 30 40 50 60 70 80 90 100 GA DOT GA FT1 ODOT Ohio MS NY 18502 NY 20-24 Class K Class K Modified % in cr ea se

66 Figure 3 presents the estimate of the as-constructed, original spacing factors for the various mixtures. The dashed horizontal lines illustrate the normally accepted range in values for this parameter, with 0.200 mm (0.008 in.) commonly considered as the upper limit for protecting the paste against freeze-thaw damage. As can be seen, the mixtures from the two southern states (Georgia and Texas) did not generally meet the recommended limit. This finding is not of great of concern for these sites, which are located in “non-freeze” zones, although these sites are subjected to a few freeze-thaw cycles during the year. Of greater concern are the relatively poor spacing factors for the Ohio Class FS and New York 18502 mixtures, which are located in severe freeze-thaw environments. As shown in Figures 4 and 5, the infilling of the air voids with sulfate-bearing phases such as ettringite increased the existing spacing factor for mixtures from Georgia and Ohio, but not from New York and Texas. The increase did not appear to compromise the air-void systems of mixtures that started with sufficient spacing factors. Statistical analysis of the air-void system parameters for satisfactory and poorly performing concrete found, in general, little difference for a given mixture, with two exceptions: the GA DOT and New York Modified Class F. Significant differences between satisfactory and poorly performing repairs were observed for the GA DOT mixtures, with poorly performing repairs having more infilling. It is not possible to determine from the test results whether the infilling of voids in GA DOT repairs with ettringite caused the poor performance or whether the infilling is the result of the poor performance. The New York Modified Class F mixtures had similar results, with the satisfactory repairs having more favorable air-void system parameters than poorly performing repairs. The observed distress (a single crack), however, is not consistent

67 with paste freeze-thaw deterioration; therefore, it is unlikely that the air-void system is responsible for the poor performance. Stereo Optical Microscope Observations In addition to characterizing the air-void system, stereo optical microscopy was used to make general observations of the concrete, particularly the existence of ASR. ASR was observed in the two mixtures from Ohio. It was isolated to the dark shale constituent of the fine aggregate and was most severe in the Ohio Class FS mixture, but was also identified in the Ohio Class MS mixture. It is believed that the high cement content aggravated the ASR in the FS mixture by increasing the total alkalinity of the mixture, whereas the use of silica fume in the MS mixture helped suppress ASR. This observation suggests the need for considering ASR potential when constructing EOT concrete repairs. Petrographic Microscope Observations Petrographic observations were made of thin sections prepared from the field concrete. To provide a means of comparison, each thin section was rated for paste homogeneity and degree of microcracking on a scale of 1 to 3, with 1 being best (i.e., homogenous and free of microcracking) and 3 being worst (i.e., highly inhomogeneous with significant microcracking). The nature of calcium hydroxide was also recorded. The results of this evaluation are presented in Table 4. Micrographs of each thin section are presented in Appendix B. No observable differences in microstructure were observed between satisfactorily and poorly rated repairs.

68 TABLE 4 Summary of petrographic evaluation of thin sections from field concrete Site Paste Homogeneity Microcracking Condition of Calcium Hydroxide GA FT1 1 1 Normal GA DOT 1 1 Normal NY 18502 2 2 Large and patchy NY Mod. Class F 1 1 Normal OH Class FS 2 3 Normal OH Class MS 1 2 Very small, less abundant TX Class K 1 1 Normal TX Class K Modified 1 2 Large and patchy The results from the petrographic analysis indicate that four of the repair materials (Georgia DOT, Georgia FT1, New York Modified Class F, and Texas Class K) had microstructure with good homogeneity, little microcracking, and normal calcium hydroxide. Two of the mixtures (Ohio Class MS and Texas Class K Modified) had good homogeneity, but moderate microcracking. The calcium hydroxide was very small and less abundant in the Ohio Class MS mixture, whereas it was large and patchy in the Texas Class K Modified mixture. The finding for the Ohio Class MS material is not surprising given that the mixture contained microsilica, which is a very active pozzolan that reduces calcium hydroxide in the concrete. The fast-setting materials (Ohio Class FS and New York 18502 mixtures) have moderate paste inhomogeneity and moderate microcracking, with the calcium hydroxide being large and patchy in the New York 18502 mixture. These observations suggest that on occasion, difficulties in achieving adequate dispersion of cement may occur in these high–cement-content mixtures (these two mixtures had the highest cement content of the field mixtures). Further, both these mixtures had marginal air-void systems, as indicated by the spacing factors. It is impossible to

69 determine whether the observed microcracking resulted from construction, freeze-thaw damage to the poorly protected paste, or ASR as observed in the case of the Ohio Class FS mixture. 3.1.5 Summary of Field Evaluation The following observations were made from the information collected in the field study: • No single cause of distress was observed in the field concrete. In general, the concrete was of good quality, and it was difficult to find “distressed” repairs for use in this study. This finding indicates that although some durability problems have been observed in EOT concrete repairs, it is clearly possible to construct durable, long-lasting repairs. • One problem observed in the field concrete was poorly formed air-void systems that may have been inadequate for protecting the hydrated cement paste against freeze-thaw damage. In some cases, the original spacing factors were inadequate even though the air content was sufficient. In other cases, infilling of the air voids with secondary deposits, such as ettringite, increased the spacing factor to unacceptable levels. • Although attempts were made to avoid EOT concrete with known alkali-aggregate reactivity problems, ASR was observed in both types of materials obtained from Ohio, but was far less prevalent in the Ohio Class MS mixture, which contained a microsilica supplementary cementitious material. • In general, less homogeneous paste, increased microcracking, and large patchy calcium hydroxide was observed in the faster-setting materials. These properties and the poor air-void

70 systems observed in many of these mixtures indicate that thorough blending of the constituents in these high–cement-content mixtures may not have occurred. 3.2 LABORATORY EVALUATION Because of the complexity of EOT concrete mixtures and the variability of the various constituent materials, information obtained during the initial phases of this study was used to select material combinations that broadly represent the type of EOT concrete mixtures being constructed nationwide. A factorial experimental design that allows for more than two levels to be set for each variable was then adopted. These levels were set based on knowledge of concrete technology applicable to EOT concrete mixtures, for variables of greatest relevance. Each of these material combinations is discussed in detail below. 3.2.1 Material Combinations for 6- to 8-hour EOT Concrete Materials The material combinations used for the 6- to 8-hour EOT concrete materials are presented in Table 5. A number designation (1 through 7) has been placed above each independent variable, and an alpha-numeric designation (A-1 through N-14) is used for each mixture. The initially proposed strength criterion at 6 hours was 13.8 MPa (2,000 psi) for compressive strength or 2.1 MPa (300 psi) for third-point flexural strength. Each of the seven independent variables considered in the study of the 6- to 8-hour EOT concrete mixtures has either two or three levels. This consideration results in a possible 288

71 material combinations, of which 14 material/curing combinations were evaluated. Two batches were made for each mixture; thus, 28 batches of were made for tests. A discussion of the seven independent variables follows. Independent Variables for Laboratory Experiment The seven independent variables were selected after considering the findings of the NCHRP Project 18-04A and information contained in published articles and specifications used by various SHAs in the construction of EOT concrete repairs. Each of the first six variables (cement type, cement factor, w/c ratio, coarse aggregate type, accelerator type, and water reducer type) is common in the mix design/proportioning process; they are included in most SHA specifications. Further, of these variables seems to directly affect the shrinkage and durability of concrete.

72 TABLE 5 Summary of 6- to 8-hour EOT concrete mixture designs used in laboratory study 1 2 3 4 5 6 7 Combination Cement Type Cement Factor w/c Ratio Coarse Aggregate Type Accelerator Type Water Reducer Type Curing Temperature A-1 Type I 425 kg/m 3 (716 lb/yd3) 0.40 Carbonate Non-Chloride No 23 oC (73oF) B-2 Type I 525 kg/m 3 (885 lb/yd3) 0.40 Carbonate Non-Chloride No 23 oC (73oF) C-3 Type I 525 kg/m 3 (885 lb/yd3) 0.40 Carbonate Non-Chloride No Heated blanket D-4 Type I 525 kg/m 3 (885 lb/yd3) 0.40 Siliceous Non-Chloride No Heated blanket E-5 Type I 425 kg/m 3 (716 lb/yd3) 0.40 Siliceous Non-Chloride No 23 oC (73oF) F-6 Type I 525 kg/m 3 (885 lb/yd3) 0.36 Carbonate Non-Chloride No 23 oC (73oF) G-7 Type I 525 kg/m 3 (885 lb/yd3) 0.40 Carbonate Non-Chloride Type F 23 oC (73oF) H-8 Type I 525 kg/m 3 (885 lb/yd3) 0.36 Carbonate No Type E 23 oC (73oF) I-9 Type I 425 kg/m 3 (716 lb/yd3) 0.40 Carbonate Calcium Chloride No 23 oC (73oF) J-10 Type I 525 kg/m 3 (885 lb/yd3) 0.40 Carbonate Calcium Chloride No 23 oC (73oF) K-11 Type III 425 kg/m 3 (716 lb/yd3) 0.40 Carbonate Non-Chloride Type F 23 oC (73oF) L-12 Type III 525 kg/m 3 (885 lb/yd3) 0.40 Carbonate Non-Chloride Type F 23 oC (73oF) M-13 Type III 525 kg/m 3 (885 lb/yd3) 0.40 Carbonate Non-Chloride Type F Heated blanket N-14 Type III 425 kg/m 3 (716 lb/yd3) 0.40 Carbonate Calcium Chloride Type F 23 oC (73oF) The seventh variable, curing temperature, was also thought to have a potentially large impact on the durability of EOT concrete repairs. Field and laboratory studies have shown that many of these 6- to 8-hour EOT concrete mixtures obtain relatively high temperatures (in excess of 65oC [150oF]) in the first few hours after placement because of the high heat of hydration that occurs. These high temperatures may negatively affect the durability of the EOT concrete through adverse alteration of the hydration products (e.g., DEF), increased thermal stresses, and increased rate of evaporation. It is very difficult to produce this phenomenon in the laboratory using heat generated through hydration alone, since the test specimens are of inadequate size to contain the heat. Since SHA laboratories commonly conduct only standard tests that stipulate ambient laboratory temperatures during curing, the effect of high curing temperatures may never

73 be observed in laboratory tests. For this reason, a limited study of the effect of increased curing temperature was included in this study. Below is a detailed description of each of the proposed variables. Cement Type. Based on the review of SHA specifications and relevant literature, only Type I and III portland cements were selected for evaluation. The Type I cement was produced by Lafarge in the Alpena Plant and the Type III by Lafarge at the Woodstock Plant. Chemical and physical properties of the cements are presented in Table 6. Sufficient quantities of both cements were purchased prior to making specimens so that the same lots were used throughout the project.

74 TABLE 6 Chemical and physical properties of the cements used in this study Property Type I Type III Silicon Dioxide 20.7% 20.5% Aluminum Oxide 4.2% 4.7% Ferric Oxide 2.3% 2.7% Calcium Oxide 65.3% 63.2% Magnesium Oxide 2.0% 2.1% Sulphur Trioxide 2.5% 4.1% Tricalcium Silicate 70.0% 55.0% Tricalcium Aluminate 7.0% 8.0% Alkalies (Na2O + 0.685 K2O) 0.5% 0.4% Loss on Ignition 2.0% 2.2% Fineness (Blaine m2/kg) 394 608 Compressive Strength (1 day) 15.1 MPa (2,190 psi) 27.2 MPa (3,945 psi) Cement Factor. Cement factor is a commonly specified variable in SHA specification. Mixtures with a higher cement factor have higher hydrated cement paste content and are thus more susceptible to shrinkage and paste deterioration problems. Two levels (425 and 525 kg/m3 [716 and 885 lb/yd3]) were selected for the cement factor variable reflecting the general range in values used by SHAs. w/c Ratio. The w/c ratio is commonly recognized to be the single most important mixture variable with regards to concrete strength and durability. Two levels of w/c (0.36 and 0.40) were evaluated. Although the difference between these two levels is somewhat small, the levels fall within the ranges specified by most SHAs. Coarse Aggregate Type. Recognizing the variability inherent in aggregates, it was decided to investigate two coarse aggregate types: a quarried carbonate and a quarried silicate

75 aggregate. The use of quarried (or fully crushed) materials allows for a direct comparison to be made based on aggregate type alone, since a gradation meeting AASHTO No. 57 requirements was used for both aggregate types. The carbonate aggregate selected was a high-quality limestone from the Presque Isle Corporation (MDOT Pit No. 71-47) located in northern Michigan. The siliceous aggregate was a high-quality quarried gabbro from Bruce Mines, Ontario, Canada (MDOT Pit No. 95-10). The fine aggregate used was from the Doctors Pit (MDOT No. 34-86), which is primarily a natural siliceous sand deposit known not to be alkali-silica reactive. Accelerator Type. Concrete mixtures made with no accelerator (although a Type E water-reducing and water-accelerating admixture [Degussa Lubricon NCA] was used) and mixtures made from different types of accelerator (calcium chloride [Dow Flake] and Grace PolarSet [a non-chloride accelerator meeting AASHTO M 194 Type C requirements]) were evaluated. Water Reducer Type. Concrete mixtures made with two types of levels of water reducer or with no water reducer, reflecting SHA specifications that in general do not specify a water reducer, were evaluated. Yet, it is known that water reducers can be advantageous, because they allow less water to be added while maintaining workability and also help disperse the cement grains more uniformly, which is important in high–cement-content mixtures. The types of water reducer selected were an AASHTO M 194 Type E (Degussa Lubricon NCA) and a Type F

76 (Grace ADVA Flow), respectively. Both Type E and Type F (a high-range water reducer) are specified in a limited number of SHA specifications. Curing Temperature. Two curing temperatures were evaluated in this study. The first temperature—the laboratory ambient temperature of 23oC (73oF)—was used for curing 11 different mixtures. For the second temperature, three mixtures (“C,” “D,” and “M”) were cured under electric heated blankets to achieve a temperature of approximately 65oC (150oF) to evaluate the effects that higher-temperature curing has on strength development, shrinkage, durability, microstructure, and sorptivity. Previous studies have found that 6- to 8-hour EOT concrete materials can be placed during summer months, and the New York Department of Transportation’s specifications even use a 65oC (150oF) surface temperature as an opening criterion. The higher-temperature curing was applied after the final set had occurred by grouping the relevant specimens together in a frame covered with an electric blanket for 6 hours after casting. Other Comments Regarding 6- to 8-hour EOT Concrete Mixture Designs It was obvious that dosage rates for the various admixtures would need to be adjusted from mixture to mixture once specific constituents were identified and experience was gained in the laboratory. For example, it was necessary to adjust the accelerator dosage rate depending on the cement type, cement content, and the w/c ratio. The volume of the aggregate also had to be varied as the cement factor changed, but the ratio of coarse to fine aggregate was held constant.

77 Further, certain fresh mixture properties were not specifically controlled and thus were allowed to vary within a specified range from mixture to mixture. For example, the consistency of the mixture, as measured by slump, varied from mixture to mixture, with a desired range of 50 to 150 mm (2 to 6 in.). All mixtures were also air-entrained with a desired fresh concrete air content of 6 ± 1.5 percent. Table 7 summarizes the 6- to 8-hour EOT concrete mixture design parameters as batched for this study. A detailed summary of the mixture design information is presented in Appendix C.

78 TABLE 7 Summary of 6- to 8-hour EOT concrete mixture design parameters as batched per cubic meter (cubic yard) Cement Accelerator Water Reducer Mix Number Batch kg (lb) Type w/c Ratio Coarse Agg. Type Amount Type Amount ml (oz) Cure Temp A 425 (716) I 0.40 71-47 NC 5,868 ml (198 oz) No 23oC (73oF)A-1 B 425 (716) I 0.40 71-47 NC 5,868 ml (198 oz) No 23oC (73oF) A 525 (885) I 0.40 71-47 NC 3,912 ml (132 oz) No 23oC (73oF)B-2 B 525 (885) I 0.40 71-47 NC 3,912 ml (132 oz) No 23oC (73oF) A 525 (885) I 0.40 71-47 NC 3,912 ml (132 oz) No blanket C-3 B 525 (885) I 0.40 71-47 NC 3,912 ml (132 oz) No blanket A 525 (885) I 0.40 95-10 NC 3,912 ml (132 oz) No blanket D-4 B 525 (885) I 0.40 95-10 NC 3,912 ml (132 oz) No blanket A 425 (716) I 0.40 95-10 NC 5,868 ml (198 oz) No 23oC (73oF)E-5 B 425 (716) I 0.40 95-10 NC 5,868 ml (198 oz) No 23oC (73oF) A 525 (885) I 0.36 71-47 NC 3,920 ml (132 oz) No 23oC (73oF)F-6 B 525 (885) I 0.36 71-47 NC 3,920 ml (132 oz) No 23oC (73oF) A 525 (885) I 0.40 71-47 NC 3,920 ml (132 oz) F 130 (4.4) 23oC (73oF)G-7 B 525 (885) I 0.40 71-47 NC 3,920 ml (132 oz) F 87 (2.9) 23oC (73oF) A 525 (885) I 0.36 71-47 No E 2,608 (88.2) 23 oC (73oF)H-8 B 525 (885) I 0.36 71-47 No E 1,630 (55.1) 23 oC (73oF) A 425 (716) I 0.40 71-47 CC 1.08 kg (2.38 lb) No 23 oC (73oF)I-9 B 425 (716) I 0.40 71-47 CC 1.08 kg (2.38 lb) No 23oC (73oF) A 525 (885) I 0.40 71-47 CC 0.67 kg (1.48 lb) No 23oC (73oF)J-10 B 525 (885) I 0.40 71-47 CC 0.67 kg (1.48 lb) No 23oC (73oF) A 425 (716) III 0.40 71-47 NC 5,868 ml (198 oz) F 211 (7.1) 23oC (73oF)K-11 B 425 (716) III 0.40 71-47 NC 5,868 ml (198 oz) F 157 (5.3) 23oC (73oF) A 525 (885) III 0.40 71-47 NC 3,912 ml (132 oz) F 130 (4.4) 23oC (73oF)L-12 B 525 (885) III 0.40 71-47 NC 3,912 ml (132 oz) F 130 (4.4) 23oC (73oF) A 525 (885) III 0.40 71-47 NC 3,912 ml (132 oz) F 130 (4.4) blanket M-13 B 525 (885) III 0.40 71-47 NC 3,912 ml (132 oz) F 154 (5.2) blanket A 425 (716) III 0.40 71-47 CC 1.08 kg (2.38 lb) F 157 (5.3) 23oC (73oF)N-14 B 425 (716) III 0.40 71-47 CC 1.08 kg (2.38 lb) F 211 (7.1) 23oC (73oF) NC: non-chloride. CC: calcium chloride.

79 3.2.2 Material Combinations for 20- to 24-hour EOT Concrete Mixtures The material combinations used for the 20- to 24-hour EOT concrete materials are presented in Table 8. A numerical designation (1 through 6) has been place above each independent variable and an alpha-numeric designation (A-15 through N-28) is alongside each combination.

80 TABLE 8 Summary of proposed 20- to 24-hour EOT concrete mixture designs for laboratory study 1 2 3 4 5 6 Combination Cement Type Cement Factor w/c Ratio Coarse Aggregate Type Accelerator Type Water Reducer Type A-15 Type I 400 kg/m3 (674 lb/yd3) 0.43 Carbonate Non-Chloride No B-16 Type I 400 kg/m3 (674 lb/yd3) 0.43 Carbonate Calcium Chloride No C-17 Type I 400 kg/m3 (674 lb/yd3) 0.43 Siliceous Non-Chloride No D-18 Type I 400 kg/m3 (674 lb/yd3) 0.43 Gravel Non-Chloride No E -19 Type I 400 kg/m3 (674 lb/yd3) 0.40 Carbonate Non-Chloride No F-20 Type I 400 kg/m3 (674 lb/yd3) 0.40 Carbonate Calcium Chloride No G-21 Type I 475 kg/m3 (800 lb/yd3) 0.43 Carbonate Non-Chloride No H-22 Type I 475 kg/m3 (800 lb/yd3) 0.43 Carbonate No Type E I-23 Type I 475 kg/m3 (800 lb/yd3) 0.43 Carbonate No Type A J-24 Type I 475 kg/m3 (800 lb/yd3) 0.43 Carbonate No No K-25 Type I 475 kg/m3 (800 lb/yd3) 0.43 Siliceous No No L-26 Type I 475 kg/m3 (800 lb/yd3) 0.43 Gravel No No M-27 Type III 400 kg/m3 (674 lb/yd3) 0.40 Carbonate Non-Chloride No N-28 Type III 400 kg/m3 (674 lb/yd3) 0.40 Carbonate Calcium Chloride No Each of the six independent variables considered in the study of 20- to 24-hour EOT concrete mixtures has either two or three levels. This consideration results in a possible 216 material combinations, of which 14 combinations were evaluated. Two batches of each material combination were produced and test specimens were cured at 23°C. High curing temperature was not considered. A discussion of the six independent variables follows.

81 Independent Variables for Laboratory Experiment The six variables considered in the 20- to 24-hour EOT concrete experiment are the same first six variables in the 6- to 8-hour EOT concrete experiment. However, the range of each variable and the selected levels differed from those used for the 6- to 8-hour EOT concrete experiment to consider the slower strength gain requirement of the 20- to 24-hour EOT concrete mixtures. Below is a description of each of these variables. Cement Type. The same Type I and III portland cements (Lafarge Alpena Type I and Lafarge Woodstock Type III) were used in both the 20- to 24-hour and the 6- to 8-hour EOT concrete experiments. (See Table 6 for the properties of the cement.) Both Type I and Type III cements have been used by SHAs and in studies on 20- to 24-hour EOT concrete materials. Cement Factor. Two levels (400 and 475 kg/m3 [674 and 800 lb/yd3]) were selected for evaluation. This range is lower than that considered for the 6- to 8-hour EOT concrete experiment to account for the slower rate of strength gain required for 20- to 24-hour EOT concrete materials and the range stipulated in SHA specifications. It is noted that the lowest cement factors were used only for Type III cement. w/c Ratio. Two levels of w/c ratio (0.40 and 0.43) were evaluated. This range is slightly higher than that proposed for the 6- to 8-hour EOT concrete mixtures to consider the slower rate of strength gain stipulated in SHA specifications.

82 Coarse Aggregate Type. In addition to the two quarried coarse aggregate types included in the 6- to 8-hour EOT concrete experiment, a processed gravel was included in the 20- to 24- hour EOT concrete experiment as a third coarse aggregate type because many SHAs use this type of material in their 20- to 24-hour EOT concrete repairs. The gravel aggregate was obtained from the same source of the natural sand being used as the fine aggregate (Doctors Pit [MDOT No. 34-86]). This coarse aggregate is primarily siliceous in nature and known not to be alkali- silica reactive. Accelerator Type. The accelerators used in the 20- to 24-hour EOT concrete experiment were the same as those used in the 6- to 8-hour experiment, although the addition rates were altered to reflect differences in mixture components and early strength gain requirements. These accelerators included calcium chloride (Dow Flake), a non-chloride accelerator (Grace PolarSet) specified under AASHTO M 194 as Type C, and an AASHTO M 194 Type E water-reducing and accelerating admixture (Degussa Lubricon NCA). Also, mixtures containing no accelerator were made. Water Reducer Type. In addition to mixtures containing no water reducer, mixtures were made with two types of water reducers: AASHTO M 194 Type A (Grace WRDA 20) and Type E (Degussa Lubricon NCA). The Type A water reducer is a low-range water reducer that is specified by a number of SHAs for 20- to 24-hour EOT concrete materials.

83 Notes on Mixture Proportioning As is true with the 6- to 8-hour EOT concrete mixtures, dosage rates for the various admixtures were varied once specific components were identified and experience was gained in the laboratory. Certain fresh mixture properties were not specifically controlled, but were allowed to change within a specified range from mixture to mixture. For example, the consistency of the mixture, as measured by slump, was targeted to vary from 50 to 100 mm (2 to 4 in.) as other mixture parameters were changed. All mixtures were air-entrained using a vinsol resin–based admixture (Axim Catexol VR), as is common in most SHAs, with a desired fresh concrete air content of 6 ± 1.5 percent. Table 9 summarizes the as-batched parameters for the 20- to 24-hour EOT concrete mixtures. A detailed summary is presented in Appendix C.

84 TABLE 9 Summary of 20- to 24-hour EOT mixture design parameters as batched per cubic meter (cubic yard) Cement Accelerator Water Reducer Mix Number Batch kg/m 3 (lb/yd3) Type w/c Ratio Coarse Agg. Type Type Amount Type ml (oz) Curing Temperature A 400 (674) I 0.43 71-47 NC 978 ml (33.1 oz) No 23oC (73oF) A-15 B 400 (674) I 0.43 71-47 NC 978 ml (33.1 oz) No 23oC (73oF) A 400 (674) I 0.43 71-47 CC 0.51kg (1.12 lb) No 23oC (73oF) B-16 B 400 (674) I 0.43 71-47 CC 0.51kg (1.12 lb) No 23oC (73oF) A 400 (674) I 0.43 95-10 NC 978 ml (33.1 oz) No 23oC (73oF) C-17 B 400 (674) I 0.43 95-10 NC 978 ml (33.1 oz) No 23oC (73oF) A 400 (674) I 0.43 34-86 NC 1,300 ml (44.0 oz) No 23oC (73oF) D-18 B 400 (674) I 0.43 34-86 NC 1,300 ml (44.0 oz) No 23oC (73oF) A 400 (674) I 0.40 71-47 NC 978 ml (33.1 oz) No 23oC (73oF) E-19 B 400 (674) I 0.40 71-47 NC 978 ml (33.1 oz) No 23oC (73oF) A 400 (674) I 0.40 71-47 CC 0.51kg (1.12 lb) No 23oC (73oF) F-20 B 400 (674) I 0.40 71-47 CC 0.51kg (1.12 lb) No 23oC (73oF) A 475 (800) I 0.43 71-47 NC 520 ml (17.6 oz) No 23oC (73oF) G-21 B 475 (800) I 0.43 71-47 NC 520 ml (17.6 oz) No 23oC (73oF) A 475 (800) I 0.43 71-47 None E 1,300 (44.0) 23oC (73oF) H-22 B 475 (800) I 0.43 71-47 None E 1,300 (44.0) 23oC (73oF) A 475 (800) I 0.43 71-47 None A 79 (2.7) 23oC (73oF) I-23 B 475 (800) I 0.43 71-47 None A 79 (2.7) 23oC (73oF) A 475 (800) I 0.43 71-47 None No 23oC (73oF) J-24 B 475 (800) I 0.43 71-47 None No 23oC (73oF) A 475 (800) I 0.43 95-10 None No 23oC (73oF) K-25 B 475 (800) I 0.43 95-10 None No 23oC (73oF) A 475 (800) I 0.43 34-86 None No 23oC (73oF) L-26 B 475 (800) I 0.43 34-86 None No 23oC (73oF) A 400 (674) III 0.40 71-47 NC 520 ml (17.6 oz) F 199 (6.7) 23oC (73oF) M-27 B 400 (674) III 0.40 71-47 NC 520 ml (17.6 oz) F 199 (6.7) 23oC (73oF) A 400 (674) III 0.40 71-47 CC 0.51kg (1.12 lb) F 199 (6.7) 23oC (73oF) N-28 B 400 (674) III 0.40 71-47 CC 0.51kg (1.12 lb) F 199 (6.7) 23oC (73oF) NC: non-chloride. CC: calcium chloride.

85 3.2.3 Specimen Preparation Laboratory specimens were prepared in strict accordance with AASHTO T 126 at the Michigan DOT’s Construction and Technology Laboratory in Lansing, Michigan. A 0.4-m3 (14- ft3) capacity Lancaster mixer was used to ensure that all specimens needed for a given mixture were produced in a single batch, reducing experimental variability. As noted, two batches were made of each material combination. Furthermore, the order in which batches were prepared was randomized to avoid systematic error. Figure 6 lists the specimens that were made from each batch, how they were cured, and how they were ultimately tested. As can be seen, 30 specimens were made from each batch, including cylinders, beams, prisms, and rings. All coarse aggregates were sieved into standard-size fractions and then combined to the desired grading. The aggregate volume fraction was changed for the various mixtures as cement factor and w/c ratio varied, but the proportion of coarse aggregate to fine aggregate was held constant. Details on the volume fraction of coarse and fine aggregate for all mixtures are provided in Appendix C. The aggregates were brought to SSD condition prior to mixing to ensure accuracy in the resulting w/c ratio. Molds were stripped from the specimens 6 and 20 hours after casting for the 6- to 8-hour and 20- to 24-hour EOT concrete mixtures, respectively. Specimens were cured as specified in the applicable test method.

86 3.2.4 Testing Laboratory-Prepared Specimens Tests were conducted on all the 6- to 8-hour and 20- to 24-hour EOT concrete materials using the same testing procedures. As presented in Table 10, 28 specimens were needed per batch for each 6- to 8-hour EOT concrete mixture, and 26 specimens were needed per batch for the 20- to 24-hour EOT concrete mixtures. In total, 1,512 specimens were made and tested in the course of this study. The tests dealt with the following five concrete attributes: • Properties of fresh concrete, • Volume change, • Freeze-thaw durability, • Microstructural characterization, and • Absorption/sorptivity.

87 Figure 6. Specimen production and disposition for each batch. Specimens 1 and 2: 100 mm diameter cylinders Coefficient of thermal expansion Cure overnight Pick up by MSUDay after casting Specimens 3 through 10: 100 mm diameter cylinders Compressive strength Cure as assigned Test at MDOT Appropriate time/date Specimens 11 and 12: 100 mm diameter cylinders Extra specimens Cure 7 days MDOT store and deliverto MTU Specimens 13 and 14: 100 mm diameter cylinders Density, absorption, and voids Cure 7 days Test at MDOTAppropriate time/date Specimens 15 and 16: 100 mm diameter cylinders Sorptivity Cure overnight Picked up by MSUDay after casting Transfer to MSU Specimens 17 and 18: 150 mm diameter cylinders Maturity Cure overnight Monitor with maturitymeter during curing Save specimens Specimens 19 and 20: 150 mm diameter cylinders Petrographic analysis Cure 7 days MDOT store and deliverto MTU Specimens 21 through 24: 100 mm by 100 mm beams Flexural strength Cure as assigned Test at MDOT Appropriate time/date Save beam ends Specimens 25 and 26: Freeze-thaw beams Freeze-thaw testing Cure as assigned Test at MDOT Appropriate time/date Send to MTU Specimens 27 and 28: Pan specimens Scaling test Cure overnight Picked up by MSUDay after casting Specimens 29 and 30: Ring specimens Shrinkage ring test Cure overnight Picked up by MSUDay after casting

88 TABLE 10 Details of the laboratory evaluation Concrete Attribute Test Name/Equipment Measured Property No. of Specimens per batch Specification Slump Concrete workability One per batch AASHTO T 119-93 MDOT1 Air Content Total air content of fresh concrete One per batch AASHTO T 152-93 MDOT Properties of Fresh Concrete Maturity and Time of Setting Time of curing and temperature of strength gain/time and depth 2 ASTM C 1074 AASHTO T131-93 MDOT Coefficient of Thermal Expansion Coefficient of Thermal Expansion 2 AASHTO TP 60-00 MSU Volume Change Restrained Drying Shrinkage Test Ring Susceptibility to drying shrinkage cracking 2 AASHTO PP 34-99 MSU/MDOT Resistance of Concrete to Freezing and Thawing Loss of concrete stiffness due to freeze- thaw damage 2 AASHTO T 161-93 MDOT Freeze-Thaw Durability Exposure to Deicers Scaling resistance under deicer application 2 ASTM C 672-92 MDOT Compressive Strength Concrete stiffness and strength 82 AASHTO T 22-92 MDOT Flexural Strength Concrete Flexural Strength 43 AASHTO T 97-86 MDOT Stereo Optical Microscopy Air-void system parameters 2 originals others4 ASTM C 457 and C 856 MTU Petrographic Optical Microscopy with UV Dye Impregnation Microstructure analysis, reaction products, and evidence of deterioration Same specimens as stereo microscopy ASTM C 856 MTU Scanning Electron Microscopy (SEM) Microstructure analysis, elemental analysis Same specimens as stereo microscopy ASTM C 856 MTU Microstructural Characterization X-Ray Microscopy (XRM) Chloride profiling and elemental analysis Deicer test specimens No standard test MTU Specific Gravity, Absorption, and Voids Determines specific gravity, percent adsorption, and voids 2 ASTM C 642-90 MDOT Absorption/Porosity Sorptivity Assessment of concrete permeability 2 Proposed ASTM Sorptivity Test MSU Total Specimens 28 1 The agency responsible for running each test is identified in Italics in the last column of the table. 2 For the 6- to 8-hour EOT concrete mixtures, two cores were tested at 6, 8, and 24 hours and two at 28 days. For the 20- to 24-hour EOT concrete mixtures, two cores were tested at 20 and 24 hours, and two at 28 days. 3 Tested either at 6 hours or 20 hours. 4 Microscopy was done on two original specimens, one from AASHTO T161 and one from ASTM C 672.

89 Properties of Fresh Concrete Testing of fresh concrete was conducted in accordance with the relevant test methods, and the results were used to verify workability and air content and to establish maturity trends. This testing was necessary to ensure that the mixtures produced and tested could be constructed and thus have practical application. The measured air contents were also compared with the air- void system parameters obtained from the hardened concrete as determined by ASTM C 457. The established maturity trends help provide a basis for an early opening criterion, as well as an understanding of the heat of hydration characteristics for the various mixtures. Volume Change Two tests were used to assess the volume change characteristics of the concrete mixtures: the CTE and the restrained drying shrinkage cracking tests. The CTE is an important parameter with regards to thermal stress development. It is largely influenced by aggregate type. Concrete made with siliceous aggregates typically has higher coefficients of thermal expansion. Variations due to other mixture parameters were also investigated, including paste volume, w/c ratio, and cement type. The results of the restrained drying shrinkage cracking test (AASHTO PP 34-99) provide a direct indication of the propensity for the various mixtures to undergo potentially damaging volume change due to drying.

90 Freeze-Thaw Durability Two freeze-thaw durability tests were conducted on specimens that were cured for 28 days. AASHTO T 161 Procedure A was used to assess the durability of concrete subjected to cyclic freezing and thawing. Specimens were tested to 300 cycles in accordance with the temperature cycling regime specified in the test method and the dilation measured (only dilation was measured in accordance with the MDOT standard operating protocol). ASTM C 672 was performed to evaluate the concrete’s resistance to scaling from deicer applications. Microstructure of the test specimens was also characterized after testing. The degree of microcracking in the specimens after the freezing and thawing tests (AASHTO T 161) and the scaling resistance tests (ASTM C 672) was assessed to determine if significant alterations resulted from the environmental conditioning. The microstructural characterization observations verified that the distress was indeed due to freezing and thawing and/or deicer application. Microstructural Characterization Seven separate test methods were conducted under the microstructural characterization test attribute. Two of these tests are standard compressive strength (AASHTO T 22) and flexural strength (AASHTO T 97) tests. Compressive strength tests were conducted at 6 hours, 8 hours, 24 hours, and 28 days for the 6- to 8-hour EOT concrete mixtures and at 20 hours, 24 hours, and 28 days for the 20- to 24-hour EOT concrete mixtures. Flexural strength tests were only

91 conducted at 6 hours and 8 hours for the 6- to 8-hour EOT concrete and at 20 hours and 24 hours for the 20- to 24-hour EOT concrete to provide a comparison between the two modes of testing. The other microstructural characterization tests include the same staining and microscopy techniques previously described under the field evaluation. Specimens from all 28 material combinations were evaluated with these techniques. The purpose of the intensive microstructural characterization was to gain a good understanding of how the microstructure varied among mixtures and to determine the impact of microstructural characteristics on EOT concrete performance. The results of the microstructural characterization were also correlated, to the degree possible, with the volume change, freeze-thaw durability, and sorptivity test data. ASTM C 457 was used to determine the air-void system parameters from polished slabs using the stereo optical microscope. Reported factors used in the statistical analysis included the air content, spacing factor, specific surface, and paste-to-air ratio. Petrographic evaluation in accordance with ASTM C 856 was also conducted on thin sections. Because of the subjective nature of petrographic analysis, a system was developed to rate the degree of paste inhomogeniety on a scale of 1 to 3 (with 1 being homogenous and 3 being highly inhomogenous) and the degree of microcracking also on a scale of 1 to 3 (with 1 being free of microcracks and 3 being a high degree of microcracking). Beams that had been subjected to AASHTO T 161, the freezing and thawing test, were sectioned. The degree of microcracking (reported as microns per square millimeter) was measured using a high-resolution flatbed scanner. The degree of chloride ion ingress into the ASTM C 672 specimens was assessed using the x-ray microscope.

92 Absorption/Sorptivity Two 100-mm (4-in.)-diameter specimens from each material/condition combination were evaluated for absorption and sorptivity in accordance with ASTM C 642 and the proposed ASTM sorptivity test to provide an indication of the porosity and permeability. 3.2.5 Results and Discussion of the Laboratory Evaluation This section presents the results of the laboratory experiment. It describes the individual results for the following test attributes: properties of fresh concrete, volume change, freeze-thaw durability, microstructural characterization, and adsorption/porosity. Summary plots, tables and detailed data for both the 6- to 8-hour and 20- to 24-hour EOT concrete mixtures are presented in Appendix C and are briefly discussed in this chapter. Statistical analysis of the results is also presented. Properties of Fresh Concrete The properties of fresh concrete measured for each mixture include the slump (AASHTO T 119), air content (AASHTO T 152), unit weight (AASHTO T 121), and maturity (ASTM C 1074). As would be expected, the results varied between the 6- to 8-hour and the 20- to 24-hour EOT concrete and also within each category of mixture. The average slump was slightly higher for the 20- to 24-hour mixtures than for the 6- to 8-hour mixtures, although one 6- to 8-hour mixture (G-7) exhibited slump values in excess of 250 mm (10 in.). This mixture had a high

93 cement content, a high w/c ratio, a calcium chloride accelerator, and a Type F HRWR, making it difficult to obtain the desired properties with a lower slump. The results of the laboratory-measured air content of the fresh concrete, shown in Figures 7 and 8, for the most part were within the desired range, with a few exceptions for the 6- to 8- hour EOT concrete mixtures category. One mixture, E-5, exhibited an air content slightly higher than desired. This mixture contained a low cement content and was made with a high w/c ratio and a non-chloride accelerator. Three of the eight mixtures made with Type III cement had low measured fresh air content. All mixtures containing Type III cement were made using a Type F HRWR to facilitate mixing. An overall higher variability was observed in the 6- to 8-hour mixtures than for the 20- to 24-hour mixtures, as reflected in the coefficient of variability. The maturity results followed expected trends, with the 6- to 8-hour mixtures having much higher maturity, especially at the 8-hour test time. These numbers are a little misleading because some of the mixtures (C-3, D-4, and M-13) from the 6- to 8-hour category were cured at higher ambient temperatures, contributing to a noticeable increase in maturity. But even accounting for these mixtures, the 8-hour maturity levels were higher for the 6- to 8-hour mixtures than for the 20- to 24-hour mixtures. Within a category, there is little noticeable difference due to different cement content, w/c ratio, accelerator type, or cement type.

94 Figure 7. Air content for 6- to 8-hour mixtures (dashed lines indicate desired target range). 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 A -1 -A A -1 -B B -2 -A B -2 -B C -3 -A C -3 -B D -4 -A D -4 -B E -5 -A E -5 -B F- 6- A F- 6- B G -7 -A G -7 -B H -8 -A H -8 -B I-9 -A I-9 -B J- 10 -A J- 10 -B K -1 1- A K -1 1- B L- 12 -A L- 12 -B M -1 3- A M -1 3- B N -1 4- A N -1 4- B Mixture La bo ra to ry M ea su re d A ir (p er ce nt )

95 Figure 8. Air content for 20- to 24-hour mixtures (dashed lines indicate desired target range). 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 A -1 5- A A -1 5- B B -1 6- A B -1 6- B C -1 7- A C -1 7- B D -1 8- A D -1 8- B E -1 9- A E -1 9- B F- 20 -A F- 20 -B G -2 1- A G -2 1- B H -2 2- A H -2 2- B I-2 3- A I-2 3- B J- 24 -A J- 24 -B K -2 5- A K -2 5- B L- 26 -A L- 26 -B M -2 7- A M -2 7- B N -2 8- A N -2 8- B Mixture La bo ra to ry M ea su re d A ir (p er ce nt )

96 Volume Change Assessment of volume change was made based on the CTE (AASHTO TP 60) and restrained drying shrinkage (AASHTO PP 34) values. The CTE values varied from mixture to mixture, with the type of coarse aggregate having the biggest influence. Consistent results were obtained for each mixture that exhibited higher-than-expected CTE values except for the second replicates for C-3-B and L-12-B. Overall, there was slightly more variability observed between replicates in the 6- to 8-hour mixtures than in the 20- to 24-hour mixtures, reflecting a higher inherent variability in the higher early strength mixtures, although all the reported values are within the expected range for the materials used. Over the range of variables studied, the cement content did not correlate with the CTE. The time to first crack in days and the total number of cracks that occurred over the duration of the test were determined from the restrained shrinkage ring tests. Evaluating the data for time to first crack revealed a statistically significant difference (α = 0.05) between the 6- to 8-hour and the 20- to 24-hour mixtures, indicating that the higher early strength materials will crack earlier. No such difference could be statistically shown for the total number of cracks, suggesting that higher early strength mixtures do not necessarily crack more often. It is also observed that although most mixtures cracked at some point, some did not. The most notable observation made was that curing temperature had a large influence on the shrinkage/cracking behavior of the concrete, with the high-temperature curing resulting in a reduced propensity to crack (of the 12 shrinkage ring specimens cured at the elevated

97 temperature, only 1 cracked). The diverse factors contributing to this observation (e.g., early strength development, shrinkage at elevated temperatures, and expansion/contraction of the steel ring) and the limited tests made it impossible to determine the role of higher-temperature curing on the propensity for cracking in the concrete. The restrained shrinkage test did not always provide repeatable results between batches or even between replicates made from the same batch. In some cases, excellent repeatability was observed, such as for mixture B-2. But as often, poor repeatability was observed, such as in the case of mixture I-9. The use of the restrained shrinkage test in this study indicates that the test might provide useful information regarding the shrinkage and cracking propensity of high early strength materials. Freeze-Thaw Durability The resistance of concrete to freezing and thawing test (AASHTO T161) results indicate that the dilation values varied greatly from mixture to mixture. This variation can be seen graphically in Figures 9 and 10. Statistically, a significant difference (α = 0.05) exists between the dilation values for the 6- to 8-hour and the 20- to 24-hour mixtures, with the 6- to 8-hour mixtures having higher dilation values. In general, none of the 20- to 24-hour EOT concrete mixtures had unacceptable dilation values, whereas approximately 20 percent of the 6- to 8-hour mixture specimens exceeded 0.01-mm/mm dilation. Most of these specimens were made with Type III cement and Type F HRWR, which would initially suggest that the cement type and/or

98 water reducer might play a role. Further, one of the 6- to 8-hour mixtures (N-14) made with Type III cement and calcium chloride accelerator did not show exceptionally high dilations.

99 Figure 9. Dilation for 6- to 8-hour mixtures. -0.0100 0.0000 0.0100 0.0200 0.0300 0.0400 0.0500 0.0600 0.0700 A -1 -A A -1 -B B -2 -A B -2 -B C -3 -A C -3 -B D -4 -A D -4 -B E -5 -A E -5 -B F- 6- A F- 6- B G -7 -A G -7 -B H -8 -A H -8 -B I-9 -A I-9 -B J- 10 -A J- 10 -B K -1 1- A K -1 1- B L- 12 -A L- 12 -B M -1 3- A M -1 3- B N -1 4- A N -1 4- B Mixture D ila tio n (m m /m m ) First Replicate Second Replicate

100 Figure 10. Dilation for 20- to 24-hour mixtures. -0.0100 0.0000 0.0100 0.0200 0.0300 0.0400 0.0500 0.0600 0.0700 A -1 5- A A -1 5- B B -1 6- A B -1 6- B C -1 7- A C -1 7- B D -1 8- A D -1 8- B E -1 9- A E -1 9- B F- 20 -A F- 20 -B G -2 1- A G -2 1- B H -2 2- A H -2 2- B I-2 3- A I-2 3- B J- 24 -A J- 24 -B K -2 5- A K -2 5- B L- 26 -A L- 26 -B M -2 7- A M -2 7- B N -2 8- A N -2 8- B Mixture D ila tio n (m m /m m ) First Replicate Second Replicate

101 Batch G-7-A, which is the only batch made with Type I cement that contained both the Type F HRWR and the non-chloride accelerator had high dilations. The dosage of the Type F admixture was higher for G-7-A than for G-7-B (the admixture dosage was reduced to correct the high slump), whereas the dosage of air entrainer was increased (both G-7-A and G-7-B had measured air contents of 5.2 percent). Similar results were observed for one of the Type III mixtures, where the first replicate (K-11-A) was made with a higher HRWR dosage and lower air entrainer dosage than the second replicate (K-11-B). These results suggest that a possible interaction between the various admixtures may negatively affect the air-void system in high– cement-content, low–w/c-ratio mixtures. The deicer scaling results were even more variable than the freeze-thaw results. Statistically, the amount of scaling present was higher for the 6- to 8-hour mixtures than for the 20- to 24-hour mixtures. Many of the 6- to 8-hour Type III mixtures (K-11, L-12, M-13, N-14) that had relatively high dilation values also had a high degree of scaling. But there is little direct correlation between scaling rating and dilation (e.g., M-13-A had the highest dilation, but the degree of scaling rating was less than 1, whereas K-11-B and N-14-A had the highest severity of scaling, but both had relatively low dilation values).

102 Microstructural Characterization The microstructural characterization of the mixtures included compressive and flexural strength tests, stereo optical microscopy, petrographic microscopy, electron microscopy, x-ray microscopy to assess chloride penetration into scaling specimens, and crack length in specimens that were subjected to freeze-thaw testing. A summary of the strength data for the 6- to 8- hour and 20- to 24-hour EOT concrete mixtures is presented in Tables 11 and 12, respectively. The data follow predicted trends, with the 6- to 8- hour mixtures gaining strength quickly, having higher average 24-hour and 28-day compressive strength than the 20- to 24-hour mixtures. The initial criteria were to achieve a compressive strength of 13.8 MPa (3,000 psi) and a flexural strength of 2.1 MPa (300 psi) within the opening time criterion. The gray shading in Tables 11 and 12 identifies the strength values that were below the desired values. Of the twenty-eight 6- to 8-hour EOT concrete mixtures, nineteen did not achieve the desired 6-hour compressive strength, and five of the remaining nine mixtures that achieved the desired early strength were cured at a high temperature, demonstrating the importance of increased temperature. The four batches cured at laboratory ambient temperatures that met the minimum compressive strength criterion were from two mixtures made with Type I cement: one mixture had a low cement content and high w/c ratio, and the other mixture had high cement content and low w/c ratio. Within 8 hours, only five batches from three mixtures (all made with Type I cement) did not meet the minimum compressive strength criterion. Of the two mixtures that failed to meet the minimum compressive strength criterion within 8 hours, one (G-

103 7) used a Type E admixture, and the other (H-8) used a Type F admixture. The fifth batch (J-10- B) had unexpectedly low strengths.

104 TABLE 11 Compressive and flexural strength for 6- to 8-hour EOT concrete mixtures Compressive Strength, MPa (psi) Flexural Strength, MPa (psi) Mixture 6 hour 8 hour 24 hour 28 day 6 hour 8 hour A-1-A 14.6 (2,120) 20.4 (2,960) 36.7 (5,325) 46.7 (6,775) 2.2 (320) 2.7 (390) A-1-B 14.5 (2,105) 19.9 (2,885) 33.0 (4,785) 47.9 (6,950) 2.6 (375) 2.8 (405) B-2-A 13.3 (1,930) 16.5 (2,395) 29.3 (4,250) 43.7 (6,340) 1.9 (275) 2.4 (350) B-2-B 11.0 (1,595) 16.3 (2,365) 28.7 (4,165) 44.2 (6,410) 2.0 (290) 2.4 (350) C-3-A 9.4 (1,365) 14.8 (2,145) 27.0 (3,915) 37.0 (5,365) 1.7 (245) 2.3 (335) C-3-B 14.4 (2,090) 20.2 (2,930) 31.1 (4,510) 38.1 (5,525) 2.1 (305) 2.5 (365) D-4-A 14.3 (2,075) 19.5 (2,830) 28.1 (4,075) 37.1 (5,380) 2.1 (305) 2.2 (320) D-4-B 16.9 (2,450) 21.3 (3,090) 28.5 (4,135) 38.9 (5,640) 2.6 (375) 2.6 (375) E-5-A 9.4 (1,365) 14.1 (2,045) 25.2 (3,655) 43.4 (6,295) 1.9 (275) 2.4 (350) E-5-B 8.4 (1,220) 14.0 (2,030) 24.3 (3,525) 41.5 (6,020) 2.0 (290) 2.5 (365) F-6-A 15.3 (2,220) 20.9 (3,030) 35.0 (5,075) 52.9 (7,670) 2.6 (375) 2.9 (420) F-6-B 13.8 (2,000) 19.8 (2,870) 35.1 (5,090) 54.3 (7,875) 2.5 (365) 3.0 (435) G-7-A 5.5 (800) 10.2 (1,480) 28.5 (4,135 46.0 (6,670) 1.4 (205) 2.1 (305) G-7-B 4.9 (710) 9.6 (1,390) 24.1 (3,495) 43.8 (6,350) 1.4 (205) 2.0 (290) H-8-A 1.1 (160) 4.3 (625) 26.2 (3,800) 47.3 (6,860) 0.3 (45) 0.9 (130) H-8-B 1.5 (220) 6.1 (885) 23.1 (3,350) 48.9 (7,090) 0.6 (90) 1.2 (175) I-9-A 10.8 (1,565) 17.9 (2,595) 33.4 (4,845) 52.6 (7,630) 1.7 (245) 2.4 (350) I-9-B 10.3 (1,495) 16.1 (2,335) 35.2 (5,105) 55.1 (7,990) 1.8 (260) 2.5 (365) J-10-A 7.8 (1,130) 15.2 (2,205) 31.0 (5,715) 47.0 (6,815) 1.0 (145) 2.2 (320) J-10-B 2.2 (320) 8.7 (1,260) 29.1 (4,220) 46.8 (6,790) 0.8 (115) 1.6 (230) K-11-A 12.4 (1,800) 20.8 (3,015) 39.0 (5,655) 54.7 (7,935) 2.3 (335) 3.1 (450) K-11-B 11.0 (1,595) 17.4 (2,525) 35.2 (5,105) 49.8 (7,225) 2.4 (350) 2.6 (375) L-12-A 11.5 (1,670) 23.9 (3,465) 39.4 (5,715) 53.0 (7,685) 2.2 (320) 3.0 (435) L-12-B 8.2 (1,190) 21.4 (3,105) 36.4 (5,280) 49.8 (7,225) 1.8 (260) 2.8 (405) M-13-A 22.9 (3,320) 29.5 (4,280) 39.9 (5,785) 51.0 (7,395) 2.4 (350) 3.1 (450) M-13-B 22.5 (3,265) 31.4 (4,555) 39.8 (5,770) 48.3 (7,005) 2.5 (365) 2.8 (405) N-14-A 4.8 (695) 19.4 (2,815) 47.4 (6,875) 62.7 (9,095) 1.5 (220) 2.9 (420) N-14-B 11.3 (1,640) 24.5 (3,555) 49.7 (7,210) 67.6 (9,805) 1.9 (275) 3.1 (450) Average 10.86 (1,575) 17.65 (2,560) 32.83 (4,760) 48.22 (6,995) 1.86 (270) 2.46 (355) St. Dev 5.415 (785) 6.240 (905) 6.709 (975) 7.107 (1,030) 0.612 (89) 0.543 (79) COV 49.9% 35.4% 20.4% 14.7% 32.8% 22.0% COV = coefficient of variation. St. Dev = standard deviation. Gray shading indicates that the strength values are below the desired criterion.

105 TABLE 12 Compressive and flexural strength for 20- to 24-hour EOT concrete mixtures Compressive Strength, MPa (psi) Flexural Strength, MPa (psi) Mixture 20 hour 24 hour 28 day 20 hour 24 hour A-15-A 20.4 (2,960) 23.2 (3,365) 37.5 (5,440) 3.3 (480) 3.6 (520) A-15-B 21.1 (3,060) 24.8 (3,590) 37.4 (5,425) 3.6 (520) 3.6 (520) B-16-A 25.1 (3,640) 27.5 (3,990) 44.8 (6,500) 3.3 (480) 3.6 (520) B-16-B 24.0 (3,480) 29.2 (4,235) 47.3 (6,860) 3.5 (510) 3.9 (565) C-17-A 19.5 (2,830) 21.1 (3,060) 38.6 (5,600) 3.8 (550) 3.8 (550) C-17-B 17.7 (2,565) 20.9 (3,030) 36.1 (5,235) 3.5 (510) 3.5 (510) D-18-A 18.1 (2,625) 21.1 (3,060) 35.7 (5,180) 3.3 (480) 3.3 (480) D-18-B 20.5 (2,975) 22.0 (3,190) 36.5 (5,295) 3.5 (510) 3.7 (535) E-19-A 20.8 (3,015) 23.0 (3,335) 39.9 (5,785) 3.8 (550) 4.1 (595) E-19-B 19.0 (2,755) 23.3 (3,380) 41.2 (5,975) 3.9 (565) 4.1 (595) F-20-A 32.2 (4,670) 35.1 (5,090) 52.0 (7,540) 3.5 (510) 4.3 (625) F-20-B 33.4 (4,545) 36.7 (5,325) 49.9 (7,235) 4.2 (610) 3.9 (565) G-21-A 21.9 (3,175) 23.6 (3,425) 37.5 (5,440) 3.5 (510) 3.3 (480) G-21-B 18.0 (2,610) 19.6 (2,845) 35.5 (5,150) 3.3 (480) 3.8 (550) H-22-A 14.2 (2,060) 16.1 (2,335) 38.7 (5,615) 2.9 (420) 3.0 (435) H-22-B 14.5 (2,105) 16.9 (2,450) 39.0 (5,655) 2.7 (390) 3.2 (465) I-23-A 13.9 (2,015) 16.1 (2,335) 37.9 (5,495) 2.3 (335) 2.8 (405) I-23-B 13.3 (1,930) 16.3 (2,365) 38.3 (5,555) 2.3 (335) 3.0 (435) J-24-A 20.9 (3,030) 24.2 (3,510) 33.8 (4,900) 3.6 (520) 3.6 (520) J-24-B 19.6 (2,845) 21.7 (3,145) 37.8 (5,480) 3.2 (465) 3.6 (520) K-25-A 19.4 (2,815) 21.2 (3,075) 39.8 (5,775) 3.8 (550) 3.7 (535) K-25-B 16.2 (2,350) 18.9 (2,740) 38.8 (5,625) 3.3 (480) 3.3 (480) L-26-A 15.5 (2,250) 20.6 (2,990) 40.4 (5,860) 2.8 (405) 3.1 (450) L-26-B 15.8 (2,290) 18.8 (2,725) 38.7 (5,615) 2.7 (390) 3.2 (465) M-27-A 34.1 (4,945) 38.0 (5,510) 54.6 (7,920) 4.8 (695) 4.7 (680) M-27-B 33.7 (4,890) 37.9 (5,495) 55.4 (8,035) 4.2 (610) 4.8 (695) N-28-A 43.1 (6,250) 47.7 (6,920) 64.2 (9,310) 4.8 (695) 5.0 (725) N-28-B 45.7 (6,630) 49.4 (7,165) 66.5 (9,645) 4.5 (655) 4.6 (665) Average 22.56 (3,270) 25.53 (3,705) 42.64 (6,185) 3.50 (510) 3.72 (540) St. Dev 8.580 (1,244) 9.066 (1,315) 8.623 (1,251) 0.641 (93) 0.568 (82) COV 38.0% 35.5% 20.2% 18.3% 15.3% COV = coefficient of variation. St. Dev = standard deviation. Gray shading indicates that the strength values are below the desired criterion.

106 The flexural strength results were slightly better, with 12 of the 28 batches meeting the 6- hour criterion of 2.1 MPa (300 psi). These batches included those from the same mixtures that met the compressive strength criterion and those from mixtures K-11 and L-12. The latter two mixtures were made with Type III cement. Only four batches did not meet the flexural strength criterion at 8 hours: three of these batches were from mixtures G-7 and H-8, and one was from mixture J-10-B, which is the same batch that had low compressive strengths. Overall, meeting the strength criterion for the 20- to 24-hour mixtures was not a problem. All batches met the compressive strength criterion at 24 hours and the flexural strength criterion at 20 hours. Only one batch (I-23-B) did not meet the compressive strength criterion at 20 hours. These results indicate a difficulty in meeting high early strength criterion for the fastest- setting mixtures, although the heat generated during hydration assists in this process. Also, admixtures can have a large effect on early strength gain, with the type of water reducer and accelerator playing a role. Difficulty in achieving the desired strength within 8 hours was encountered only for a few mixtures that were primarily made with Type I cement and a water reducer. Figure 11 shows the relationship between compressive and flexural strength for all mixtures (the data shown are for 6- and 8-hour strengths for the 6- to 8-hour EOT concrete and for the 20- and 24-hour strengths for the 20- to 24-hour EOT concrete). These data indicate no unique relationship between these two measures of strength; therefore, if compressive strength is to be used during construction to estimate flexural strength, the relationship should be established for the actual job mix.

107 Figure 11. Compressive versus flexural strength for all mixtures. The stereo optical microscope was used in accordance with ASTM C 457 to collect air- void system parameters. A detailed discussion is presented in Appendix C. The measured air content of the hardened concrete averaged 5.64 percent and 5.54 for the 6- to 8-hour and 20- to 24-hour mixtures, respectively (desired value was 6 ± 1.5 percent). Figures 12 and 13 show the measured values. In general, most mixtures had air contents within the desired range, with an obvious deviation for three of the four 6- to 8-hour mixtures made with Type III cement (L-12, M-13, and N-14), which had air contents significantly below the desired value. These mixtures also contained Type F HRWR, which in combination with fine cement can negatively influence the air-void system. Two of the 20- to 24-hour batches produced with Type III cement (K-25-A and N-28-B) had similarly low air contents. It is not known what factors may have contributed to the low air content in the other 20- to 24-hour mixture (F-20). Obviously, it is difficult to control 0 1 2 3 4 5 6 0 10 20 30 40 50 60 Compressive Strength (MPa) Fl ex ur al S tr en gt h (M Pa ) 6-Hour, 6 to 8 Hour Mixtures 8-Hour, 6 to 8 Hour Mixtures 20-Hour, 20 to 24 Hour Mixtures 24-Hour, 20 to 24 Hour Mixtures Linear (8-Hour, 6 to 8 Hour Mixtures) Linear (6-Hour, 6 to 8 Hour Mixtures) Linear (20-Hour, 20 to 24 Hour Mixtures) Linear (24-Hour, 20 to 24 Hour Mixtures)

108 air content in mixtures made with high cement contents, low w/c ratios, and multiple admixtures, especially if fine cement is used.

109 Figure 12. Air content measured using ASTM C 457 for 6- to 8-hour mixtures. 0.00 2.00 4.00 6.00 8.00 10.00 12.00 A -1 -A A -1 -B B -2 -A B -2 -B C -3 -A C -3 -B D -4 -A D -4 -B E -5 -A E -5 -B F- 6- A F- 6- B G -7 -A G -7 -B H -8 -A H -8 -B I-9 -A I-9 -B J- 10 -A J- 10 -B K -1 1- A K -1 1- B L- 12 -A L- 12 -B M -1 3- A M -1 3- B N -1 4- A N -1 4- B Mixture A ST M C 4 57 M ea su re d A ir (p er ce nt )

110 Figure 13. Air content measured using ASTM C 457 for 20- to 24-hour mixtures. In general, the total air content of a mixture is not thought to be nearly as important in protecting the paste against freeze-thaw damage as the size and spacing of the air bubbles. The most common parameter used to assess the spacing of air bubbles is the spacing factor. The average spacing factors are 0.1591 mm (0.006263 in.) and 0.1273 mm (0.005010 in.) for the 6- to 8-hour and 20- to 24-hour mixtures, respectively. Individual values are shown in Figures 14 and 15. Although both average values are below the criterion set for protecting the paste against damage (0.200 mm [0.008 in.], which is shown as a dashed line in the figures), the value for the 6- to 8-hour mixtures is significantly higher than that for the 20- to 24-hour mixtures. This point is illustrated in Figure 16. Further, it can be seen in Figures 14 and 15 that all of the spacing factors exceeding the limiting criterion are in mixtures made with a Type III cement, even though some batches made with Type III cement had acceptable spacing factors (K-11-B, M-27- 0.00 2.00 4.00 6.00 8.00 10.00 12.00 A -1 5- A A -1 5- B B -1 6- A B -1 6- B C -1 7- A C -1 7- B D -1 8- A D -1 8- B E -1 9- A E -1 9- B F- 20 -A F- 20 -B G -2 1- A G -2 1- B H -2 2- A H -2 2- B I-2 3- A I-2 3- B J- 24 -A J- 24 -B K -2 5- A K -2 5- B L- 26 -A L- 26 -B M -2 7- A M -2 7- B N -2 8- A N -2 8- B Mixture A ST M C 4 57 M ea su re d A ir (p er ce nt )

111 A, and M-27-B). These findings further illustrate that the combination of fine cement and multiple admixtures made it difficult to create the desired air-void system. Interestingly, most (although not all) of these mixtures had acceptable fresh air contents, suggesting the need to better assess the air-void system for such mixtures. After stereo optical microscope evaluation, thin sections were made to assess the microstructural characteristics of the concrete. Using the petrographic microscope, the paste homogeneity and microcracking were assessed on a scale of 1 to 3 for each batch. It was observed that the 6- to 8-hour mixtures had higher degrees of inhomogeneity and microcracking than the 20- to 24-hour mixtures, indicating variation in the density of the hydrated cement paste because of difficulties in uniformly dispersing the cement grains. The highest degree of inhomogeneity was found in mixtures made with Type III cement, although some batches made with Type III cement (e.g., N-14-A, M-27-A, and M-27-B) had good paste uniformity. In comparing 6- to 8-hour mixtures made with Type I cement, the effect of the Type F HRWR is observed in that mixture G-7 had greater paste uniformity than B-2. Similar benefits were not noted with the Type E water reducer, where moderate inhomogeneity was observed both in the 6- to 8-hour mixture (H-8) and in the 20- to 24-hour mixture (H-22). The degree of microcracking was considerably higher in the 6- to 8-hour mixtures than in the 20- to 24-hour mixtures, although by no means was it absent in the latter. A small degree of microcracking is a common feature in concrete and thus is not of great consequence. A rating of “1” indicated an absence of microcracking. Moderate microcracking (i.e., rating of “2”), although not desirable, does not indicate a problem with the mixture and should not be assumed

112 to have a great impact on the durability of the concrete. But the severe microcracking (i.e, rating of “3”) noted in many of the mixtures is out of the ordinary and may indicate potential problems, especially in the higher early strength materials.

113 Figure 14. Spacing factors measured using ASTM C 457 for 6- to 8-hour mixtures. 0.0000 0.0500 0.1000 0.1500 0.2000 0.2500 0.3000 0.3500 A -1 -A A -1 -B B -2 -A B -2 -B C -3 -A C -3 -B D -4 -A D -4 -B E -5 -A E -5 -B F- 6- A F- 6- B G -7 -A G -7 -B H -8 -A H -8 -B I-9 -A I-9 -B J- 10 -A J- 10 -B K -1 1- A K -1 1- B L- 12 -A L- 12 -B M -1 3- A M -1 3- B N -1 4- A N -1 4- B Mixture Sp ac in g Fa ct or (m m )

114 Figure 15. Spacing factors measured using ASTM C 457 for 20- to 24-hour mixtures. 0.0000 0.0500 0.1000 0.1500 0.2000 0.2500 0.3000 0.3500 A -1 5- A A -1 5- B B -1 6- A B -1 6- B C -1 7- A C -1 7- B D -1 8- A D -1 8- B E -1 9- A E -1 9- B F- 20 -A F- 20 -B G -2 1- A G -2 1- B H -2 2- A H -2 2- B I-2 3- A I-2 3- B J- 24 -A J- 24 -B K -2 5- A K -2 5- B L- 26 -A L- 26 -B M -2 7- A M -2 7- B N -2 8- A N -2 8- B Mixture Sp ac in g Fa ct or (m m )

115 Figure 16. Relationship between spacing factor and dilation due to freezing and thawing. These same thin sections were evaluated using the scanning electron microscope (SEM) in backscatter electron mode. As was seen using the stereo optical microscope, very different air- void systems were observed in the mixtures prepared using a Type I cement from those made with a Type III cement, with air bubbles being far more abundant in the Type I cement mixtures. This was true for both the 6- to 8-hour and 20- to 24-hour EOT concrete. In addition, in the 6- to 8-hour EOT concrete, the hydrated cement paste was more uniform in the Type III cement mixtures, primarily a result of the fine cement, which left fewer unhydrated cement grains. The SEM images also revealed that the hydrated cement paste in mixtures made with the non- chloride accelerator was distinctly more uniform than that in the mixtures made with calcium chloride. In the 20- to 24-hour EOT concrete, it was also observed that the hydrated cement paste -0.01 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.0000 0.0500 0.1000 0.1500 0.2000 0.2500 0.3000 0.3500 Spacing factor (mm) D ila tio n (m m /m m ) 6 to 8 hour mixtures 20 to 24 hour mixtures ASTM C457 Criterion

116 in non-chloride accelerator mixtures was more uniform than that produced when no accelerator was used (without the use of accelerators, patches of higher porosity were observed). Similar observations were not made using the petrographic optical microscope because of a limitation in resolution. Only the 20- to 24-hour mixtures with the Type E water reducer showed a more uniform paste than those with Type A water reducer or those containing no water reducer at all. No other differences in the mixtures were readily observed. The x-ray microscope was used to assess the degree of chloride ion ingress into the specimens that had undergone exposure to deicers in accordance with ASTM C 672. From the data collected, an effective diffusion coefficient and the area under the chloride ion profile were computed. Overall, the determination of the effective chloride ion diffusion coefficient was straightforward except when a calcium chloride accelerator was used in the mixture. Although steps were made to correct for this, in one case (batch N-14-B), the background chloride concentration was not zero, significantly affecting the computed coefficient. The diffusion coefficients varied widely between mixtures, and although the average effective diffusion coefficient for the 6- to 8-hour mixtures was less than for the 20- to 24-hour mixtures, the variability made it impossible to assess this difference. The area under the chloride profile also shows a great amount of variability, yet a statistically significant difference does exists, with less area observed for the 6- to 8-hour mixtures than for the 20- to 24-hour mixtures. This result would be expected because of the lower permeability of the lower–w/c-ratio, higher–cement- content, faster-setting mixtures. Figure 17 shows the effect of w/c ratio on the area under the chloride profile for all mixtures.

117 Another microstructural observation was the estimation of crack length per unit area of concrete subjected to freeze-thaw testing (AASHTO T 161). It is observed that significantly more microcracking was present in the 6- to 8-hour mixtures than in the 20- to 24-hour mixtures. This observation is similar to that observed using petrographic means. Microracking correlated in an imprecise way to the measured dilation values, but, as shown in Figure 18, more cracking is measured in higher–cement factor mixtures. This trend occurs because of the paste that is present in such mixtures or the high susceptiblity to microcracking of the high–cement-content mixtures. Absorption/Porosity The apparent density of the concrete varied slightly, with coefficients of variation of 1.52 and 1.97 percent for the 6- to 8-hour and 20- to 24-hour EOT concrete mixtures, respectively. As would be expected, mixtures with the highest apparent density (D-4, E-5, C-17, and K-25) were made using the siliceous coarse aggregate, which had the highest specific gravity (2.91) of the coarse aggregates used. The percent permeable void space varied from mix to mix, with no discernible difference between the 6- to 8-hour and the 20- to 24-hour EOT concrete. As shown in Figure 19, a clear trend exists between the cement factor and the percent permeable void space, with increasing cement content (and thus greater overall porosity) resulting in a greater percentage of permeable void space. Sorptivity measurement plots are characterized by an initial phase with a relatively steep slope and a later phase with a much less steep slope, reflecting the continued uptake of water. The sorptivity test can be run with water ponded on the surface (top) or with the bottom of the

118 sample barely suspended in water (bottom). The latter procedure provides more repeatable results. As with the percent permeable void spaces, sorptivity is related to the cement factor in 20- to 24-hour mixtures but not clearly for the 6- to 8-hour mixtures.

119 Figure 17. w/c ratio versus area under chloride profile for Type I cement. 0.000 0.500 1.000 1.500 2.000 2.500 3.000 3.500 4.000 4.500 5.000 0.35 0.36 0.37 0.38 0.39 0.40 0.41 0.42 0.43 0.44 Water-to-Cement Ratio A re a U nd er C hl or id e Pr of ile (% cl *c m ) 6 to 8 hour mixtures 20 to 24 hour mixtures

120 Figure 18. Cement factor versus crack length per unit area for all mixtures. 0 10 20 30 40 50 60 350 400 450 500 550 Cement Factor (kg/m^3) C ra ck L en gt h pe r U ni t A re a (m ic ro ns /m m ^2 ) 6 to 8 hour mixtures 20 to 24 hour mixtures

121 Figure 19. Cement factor versus percent permeable void space for all mixtures. 10.00 11.00 12.00 13.00 14.00 15.00 16.00 17.00 18.00 360 380 400 420 440 460 480 500 520 540 Cement Factor (kg/m^3) Pe rc en t P er m ea bl e Vo id S pa ce 6 to 8 hour 20 to 24 hour

122 3.2.6 Summary of Laboratory Study Test Results The following observations can be made from the information collected in the laboratory study: • Because of the nature of this study, the slump of the mixtures was allowed to vary in order to maintain other experimental mixture design parameters. Type F HRWR was used with all mixtures made with the Type III cement; otherwise, mixtures were unworkable. • Overall, the laboratory-measured air contents of the fresh concrete fell within the desired range, with a few exceptions in the 6- to 8-hour mixtures. Of the eight 6- to 8-hour Type III mixtures, three had fresh air contents lower than desired. • CTE results were highly repeatable, with only a couple of exceptions. The variability was slightly higher for the 6- to 8-hour mixtures than for the 20- to 24-hour mixtures. • Results from the restrained shrinkage ring test were highly variable, but a statistical difference in the time to first cracking was observed, with 6- to 8-hour mixtures cracking earlier than 20- to 24-hour mixtures. The curing temperature had a large impact on shrinkage cracking, with increased curing temperatures producing far less tendency to crack. In general, the repeatability of the shrinkage ring test as used in this study was not very good between batches or even between replicates from the same batch. • The 6- to 8-hour mixtures statistically performed more poorly in freeze-thaw testing than did the 20- to 24-hour mixtures. None of the 20- to 24-hour mixtures performed poorly in freeze- thaw testing, whereas a number of the 6- to 8-hour mixtures had unacceptably high dilation values. Many of the poorly performing mixtures were those made with Type III cement and

123 Type F HRWR mixtures that had low air contents. There appears to be some type of indefinable interaction between the various constituents—including the cement type and content, water reducer, accelerator, and air-entraining admixture—that have influenced the freeze-thaw performance of some mixtures. • The scaling results were variable with the degree of scaling being significantly higher in the 6- to 8-hour mixtures than in the 20- to 24-hour mixtures. The same Type III cement, Type F HRWR mixtures that had poor dilations also exhibited high degrees of deicer scaling, but there was little batch-to-batch correlation between the two. These results illustrate the importance of using multiple batches and replicates for durability testing. • Strength relationships followed expected trends, with the 6- to 8-hour mixtures having higher 24-hour strengths than the 20- to 24-hour mixtures. The majority of the 6- to 8-hour mixtures did not meet the compressive and flexural strength criterion at 6 hours, although most gained sufficient strength by 8 hours. Almost all of the 20- to 24-hour mixtures met the strength criterion by 20 hours. The 6- to 8-hour EOT concrete mixtures with the lowest early strength were those made with Type I cement, Type E water reducer, and Type F HRWR. In particular, mixtures with the Type E water reducer had extremely low strengths both at 6 and 8 hours. The relationship between compressive strength and flexural strength was variable, indicating that this relationship should be determined on a mix-by-mix basis. • Most mixtures had hardened air contents (as measured by ASTM C 457) that fell within the desired range, although there were some exceptions. Three of the four 6- to 8-hour mixtures (six of the eight batches) made with Type III cement and Type F HRWR and two of the 20- to 24-hour batches had insufficient air contents. Air content was difficult to control in

124 mixtures made with high cement contents, low w/c ratios, and multiple admixtures, especially when Type III cement was used with a Type F HRWR. • The ASTM C 457 spacing factors were generally considered adequate to protect the paste against freeze-thaw damage, except for some of the mixtures made with Type III cement and Type F HRWR. Insufficient spacing factors were measured for some of the mixtures that had satisfactory air contents, as measured on the fresh concrete. It was also observed that the spacing factor criterion appears to be adequate to prevent freeze-thaw damage, as excessive dilation occurred in only a single batch meeting the spacing factor criterion. • Paste homogeneity, as assessed using the relatively low magnification of the petrographic optical microscope, was considerably better in the 20- to 24-hour mixtures than in the 6- to 8- hour mixtures, indicating a better-blended, more uniform cement paste. Mixtures made with Type III cement were slightly less homogenous than those made with Type I cement, although better cement grain dispersion was observed when the Type F HRWR was used with Type I cement. • Severe microcracking of the paste was observed in all of the 6- to 8-hour mixtures, but less severe microcracking was observed in the 20- to 24-hour mixtures, indicating a lower stress in the paste. • A difference in the air-void system was observed in backscatter electron images from the SEM, with a far greater abundance of air in mixtures made with Type I cement than with Type III cement for both the 6- to 8-hour and 20- to 24-hour mixtures. Under the higher magnification of the SEM (1000x), the hydrated cement paste in the Type III mixtures appeared more uniform than that in the Type I mixtures because of more complete hydration

125 of the cement grains. The paste was also more uniform in mixtures made with the non- chloride accelerator than in mixtures made with calcium chloride or without any accelerator. • Based on analysis using the x-ray microscope, there was significantly less overall penetration of chloride ions into the 6- to 8-hour mixtures than into the 20- to 24-hour mixtures, resulting from the reduced w/c ratio. • The crack length per unit area measurements indicated that significantly more microcracking was present after freeze-thaw testing in the 6- to 8-hour mixtures than in the 20- to 24-hour mixtures. In general, the higher the cement content, the more microcracking was observed. It is unclear whether this observation means that the paste has more microcracking in the high- cement mixtures or whether the measurement simply reflects that there is more paste in the high-cement mixtures. • The cement factor appeared to have the largest impact on the percent volume of permeable voids measured in ASTM C 642, with void volume increasing with increasing cement content. Sorptivity results were more variable and did not provide as clear a trend as the results from ASTM C 642. More repeatable results were obtained when the bottom of the specimen was barely immersed in water as opposed to water being ponded on the surface. 3.3 STATISTICAL ANALYSIS OF TEST RESULTS This section presents a statistical analysis of the data. The data set and the output of the statistical analysis are provided in Appendix D. The results of the multiple single-factor experiments are presented to assess the significance of the various independent variables. Results of the statistical analyses were used to draw generalized conclusions on how mixture design

126 parameters affect the properties of fresh concrete, strength gain, and potential durability as assessed through the test methods used in the experiment. Based on an initial analysis of the test results, results from the 47 individual tests presented in Table 13 were analyzed for the EOT concrete mixtures. A summary of the statistical analysis and the derived conclusions are presented in this section. 3.3.1 Six- to 8-hour EOT Concrete Forty-two of the 47 tests (tests 1–5, 8–22, 24–27, and 30–47) listed in Table 13 were included in the analysis of the 6- to 8-hour EOT concrete. Both single-factor and two-factor experiments were analyzed. Single-Factor Effects Table 14 lists the mixture pairs used to study the single-factor effect of each mixture variable (cement type, cement factor, etc.). For example, the effect of cement type was studied using mixture pairs A-1 and B-2, I-9 and J-10, and K-11 and L-12. Each pair has similar composition and curing regime, differing only in the factor under study. For example, mixtures A-1 and B-2 both use Type I cement with no water reducer, while mixtures K-11 and L-12 both use Type III cement with a Type F HRWR.

127 TABLE 13 Tests used in the statistical analysis 01 – CTE 25 – Compressive Strength (28 Day) 02 – Shrinkage Ring Test (Days to 1st Crack) 26 – Flexural Strength (6 Hour) 03 – Shrinkage Ring Test (Initial Slope of 1st Crack) 27 – Flexural Strength (8 Hour) 04 – Shrinkage Ring Test (Days to 2nd Crack) 28 – Flexural Strength (20 Hour) 05 – Shrinkage Ring Test (Initial Slope of 2nd Crack) 29 – Flexural Strength (24 Hour) 06 – Shrinkage Ring Test (Days to 3rd Crack) 30 – Density (% Absorption after Immersion) 07 – Shrinkage Ring Test (Initial Slope of 3rd Crack) 31 – Density (% Absorption after Immersion and Boiling) 08 – Sorptivity Testing - Top of Sample (Initial Slope) 32 – Density (Bulk Dry Density) 09 – Sorptivity Testing - Top of Sample (Final Slope) 33 – Density (Bulk Density after Immersion) 10 – Sorptivity Testing - Bottom of Sample (Initial Slope) 34 – Density (Bulk Density after Immersion and Boiling) 11 – Sorptivity Testing - Bottom of Sample (Final Slope) 35 – Density (Apparent Density) 12 – ASTM C 672: Scaling Test Rating 36 – Density (Volume of Permeable Pore Spaces) 13 – ASTM C 457: Point Count; Air Volume % 37 – Air Measured by Field Method 14 – ASTM C 457: Point Count; Air Void Specific Surface 38 – Shrinkage Ring Test - Total Number of Cracks 15 – ASTM C 457: Point Count; Paste-to-air Ratio 39 – Initial Concentration (wt % Cl) at Surface 16 – ASTM C 457: Point Count; Air Void Spacing Factor 40 – Effective Diffusion Coefficient 17 – Maturity (8-Hour Average) 41 – R-Squared Value between Fick’s Law and Data 18 – Maturity (24-Hour Average) 42 – Area under Curve (as Measure of Absorbed Cl) 19 – Freeze-Thaw Dilation 43 – Specific Gravity from Unit Weight Bucket 20 – Slump 44 – Specific Gravity from Point Count Data 21 – Compressive Strength (6 Hour) 45 – Homogeniety Rating 22 – Compressive Strength (8 Hour) 46 – Microcracking Rating 23 – Compressive Strength (20 Hour) 47 – Crack Length per Unit Area 24 – Compressive Strength (24 Hour) For each test conducted, the mean, variance, and 95-percent confidence intervals for each factor was calculated. The results for the 42 tests are presented in Appendix D. To determine if a mixture variable affects a certain test result, p-values were calculated. A p-value of less than 0.05 indicates that the mixture variable has a significant impact on that test result. The p-values for the tests on the 6- to 8-hour mixes are also presented in Appendix D. The significant test results for each of the seven mixture variables are summarized in the following text. Cement Type. The statistical analysis showed that varying the cement type from Type I to Type III cement increased scaling (test 12), maturity (tests 17 and 18), and some of the early-

128 age compressive and flexural strengths (tests 22, 24, and 27), but reduced air content that was determined from ASTM C 457 (test 13) tests. It is noted that only limited direct comparison between Type I and III cements could be conducted, since all mixtures made with Type III cement also contained Type F HRWR in order to achieve satisfactory consistency during mixing. This one comparison shows that although early strength was enhanced through the use of the Type III cement Type F HRWR, the use also resulted in increased scaling and a decrease in the total air content.

129 TABLE 14 Mixture pairs used in the 6- to 8-hour EOT concrete analysis Factor Mixes Used Cement Type Cement Factor kg/m3 (lb/yd3) w/c Ratio Coarse Aggregate Type Accelerator Type Water Reducer Type Curing Temperature oC (oF) Cement Type G-7 and L-12 I/III 525 (885) 0.40 Carbonate NC F 23 (73) Cement Factor A-1 and B-2 I-9 and J-10 K-11 and L-12 I I III 425/525 (716/885) 0.40 Carbonate NC CC NC No No F 23 (73) w/c ratio B-2 and F-6 I 525 (885) 0.40/ 0.36 Carbonate NC No 23 (73) Coarse Aggregate Type A-1 and E-5 C-3 and D-4 I 425 (716) 525 (885) 0.40 Carbonate/ Siliceous NC No 23 (73) 65 (150) Accelerator Type A-1 and I-9 B-2 and J-10 K-11 and N-14 I I III 425 (716) 525 (885) 425 (716) 0.40 Carbonate NC/CC No No F 23 (73) Water Reducer Type B-2 and G-7 I 525 (885) 0.40 Carbonate NC No/F 23 (73) Curing Temperature B-2 and C-3 L-12 and M-13 I III 525 (885) 0.40 Carbonate NC No F 23/65 (73/150) Cement Factor. In general, it appears that increasing the cement factor increased absorption, volume of permeable void space, and shrinkage potential, while not significantly improving strength (in some cases, strength was detrimentally affected). The only durability measure that was significantly improved through the use of higher cement contents was scaling. w/c Ratio. Reducing the w/c ratio from 0.40 to 0.36 resulted in a significant increase in the CTE (test 1), various measures of strength (tests 22, 24, 25, 26, and 27), and density (tests 33 and 34), while reducing absorption (tests 30 and 31) and paste inhomogeniety (test 45). There was no significant adverse effect observed in reducing the w/c ratio. Coarse Aggregate Type. Coarse aggregate type had a big effect on density, with the mixtures made with siliceous aggregate having higher concrete density than mixtures made with the carbonate aggregate. Yet the CTE was not significant, although the average was higher for the mixtures made with the siliceous coarse aggregate. The use of the siliceous aggregate

130 produced mixtures that had slightly better durability, as measured by a lower effective chloride diffusion coefficient and decreased crack length per unit area after freeze-thaw testing. Accelerator Type. In general, the non-chloride accelerator was very effective (at times more effective than calcium chloride) at increasing early strength, but long-term strength was either similar to or less than that achieved with calcium chloride. Other parameters were not notably affected by the change in accelerator. Water Reducer Type. Using a Type F water reducer versus not using a water-reducing admixture decreased the air content as measured by ASTM C 457 (test 13), the air-void specific surface (test 14), the maturity (tests 17 and 18), and 6- to 8-hour strength (tests 21, 22, 26, and 27). It also increased the paste-to-air ratio (test 15). The mixture made without water-reducing admixture had a better-entrained air-void system, higher maturity, and significantly higher early strength values than the mixture made with the Type F HRWR. One of the four batches made with the Type F HRWR (G-7-A) had very high dilation values in cyclic freeze-thaw testing as a result of the poor air-void system. However, mixtures made with the Type F HRWR had better paste homogeneity (test 45). Curing Temperature. Raising the curing temperature from 23oC (73oF) to 65oC (150oF) led to less shrinkage cracking, as assessed through the restrained shrinkage ring test. It was also observed that the air-void system parameters and freeze-thaw durability were improved by high- temperature curing, although the long-term compressive strength was reduced for the mixtures made with Type I cement.

131 Two-Factor Analysis There was only one two-factor analysis possible with the data collected in the 6- to 8- hour experiment because of the need to use the Type F HRWR with the Type III cement mixtures. The mixtures used in the two-factor analysis are A-1, B-2, I-9, and J-10. These mixtures allowed for the interaction between cement factor and accelerator type. It was found, in general, that the cement factor of a mixture is a more important factor then accelerator type. The only meaningful two-way interaction that existed was for the apparent density (test 35). In this test, both of the two factors are significant and the p-value of the interaction term is 0.0048. For mixtures with a cement factor of 425 kg/m3 (716 lb/yd3), a mix without accelerator has a higher apparent density, but for mixtures with a cement factor of 525 kg/m3 (885 lb/yd3), the mixture containing the non-chloride accelerator has a higher apparent density. Relationships Among Tests The relationships among the 47 individual tests were studied to see if the results from some of the complex, costly tests could be obtained from simpler, less costly tests. Details regarding this analysis are provided in Appendix D. As expected, a number of strong correlations existed for different measurements made within the same test (e.g., the various measures of density and absorption reported under ASTM

132 C642 and the two reported maturity values made at different times) and between tests measuring similar properties (such as specific gravity measurements and density measurements and compressive and flexural strength at similar ages). Very few correlations were observed between simple, easy-to-run tests and more complex, expensive tests. For example, no correlations exceeding 0.50 were observed for sorptivity (tests 8–11) to some of the more complex durability tests, such as scaling (test 12) or freeze-thaw dilation (test 19). Only one sorptivity test (test 10) had correlations exceeding 0.50 with any other important test, having some correlation with some elements (tests 30, 32, 33, and 34) of density (ASTM C 642) and specific gravity (test 43), as well as for the measure of cracking per unit area (test 47). It was also observed that various measurements collected for ASTM C 642 (tests 30, 32, 33, 34, and 36) correlated well with the crack length per unit area (test 47). Dilation due to freezing and thawing correlated mildly with the spacing factor, as is shown in Figure 20. It appears the maximum spacing factor criterion of 0.200 mm (0.008 in.) is acceptable to minimize damage due to freezing and thawing. There is some correlation between the air content measured on fresh concrete and that measured using ASTM C 457 (see Figure 21). A similar trend is observed with the spacing factor, as shown in Figure 22. Although, in some instances, the air content was “acceptable,” the spacing factor was not achieved. The only correlation related to paste homogeneity (test 45) is with freeze-thaw dilation, whereas decreasing paste homogeneity resulted in increased dilation (see Figure 23).

133 3.3.2 Twenty- to 24-hour EOT Concrete Forty-one of the 47 tests (tests 1–5, 8–20, 23–25, and 28–47) listed in Table 13 were included in the analysis of the 20- to 24-hour EOT concrete mixtures. Both single-factor and two-factor experiments were analyzed. Single-Factor Effects Table 15 lists the mixture pairs used to study the single-factor effect of each mixture variables (cement type, cement factor, etc.). For each test conducted, the mean, variance, and 95- percent confidence intervals for each factor were calculated. The p-value was calculated to determine if a mixture variable affects a certain test result, with a p-value of less than 0.05 indicating that the mixture variable has a significant impact on that test result. Details of the statistical analysis are presented in Appendix D. A summary follows. Cement Type. It was found that changing the cement type from Type I to Type III increased the compressive strength (tests 23–25) regardless of the accelerator type used. When a non-chloride accelerator was used, changing the cement type from Type I to Type III increased the time to first cracking (test 2) and 24-hour flexural strength (test 29) and decreased microcracking (test 46). When calcium chloride accelerator was used, the air-void system (tests 14 and 16) was negatively affected by changing cement type from Type I to Type III. Changing

134 to Type III also decreased maturity (tests 17 and 18) and absorption and percent permeable voids (tests 30, 31, and 36). Cement Factor. Varying the cement factor from 400 kg/m3 (674 lb/yd3) to 475 kg/m3 (800 lb/yd3) increased the percent absorption (test 31), total number of shrinkage cracks (test 38), microcracking rating (test 46), and crack length per unit area (test 47). Further, the mixture with the lower cement factor had a higher bulk density (test 32).

135 Figure 20. Spacing factor (test 16) versus freeze-thaw dilation (test 19). -0.01 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.0000 0.0500 0.1000 0.1500 0.2000 0.2500 0.3000 0.3500 Spacing factor (mm) D ila tio n (m m /m m )

136 Figure 21. Air content measured using ASTM C 457 (test 13) versus air content of fresh concrete (test 37). 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 ASTM C457 Air Content (Percent) A ir C on te nt o f F re sh C on cr et e (P er ce nt )

137 Figure 22. Spacing factor (test 16) versus air content of fresh concrete (test 37). 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 0.0000 0.0500 0.1000 0.1500 0.2000 0.2500 0.3000 0.3500 Spacing factor (mm) A ir C on te nt o f F re sh C on cr et e (P er ce nt )

138 0 0.5 1 1.5 2 2.5 3 3.5 -0.01 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 Dilation (mm/mm) Pa st e H om og en ei ty R at in g Figure 23. Paste homogeneity rating (test 45) versus freeze-thaw dilation (test 19).

139 TABLE 15 Mixture pairs used in 20- to 24-hour EOT concrete analysis Factor Mixes Used Cement Type Cement Factor kg/m3 (lb/yd3) w/c Ratio Coarse Aggregate Type Accelerator Type Water Reducer Type Cement Type E-19 and M-27 F-20 and N-28 I/III 400 (674) 0.40 Carbonate NC CC No Cement Factor A-15 and G-21 I 400/475 (674/800) 0.43 Carbonate NC No w/c Ratio A-15 and E-19 B-16 and F-20 I 400 (674) 0.43/0.40 Carbonate NC CC No Coarse Aggregate Type A-15 and C-17 A-15 and D-18 C-17 and D-18 J-24 and K-25 J-24 and L-26 K-25 and L-26 I 400 (674) 400 (674) 400 (674) 475 (800) 475 (800) 475 (800) 0.43 Carb/Silic Carb/Grav Silic/Grav Carb/Silic Carb/Grav Silic/Grav NC NC NC No No No No Accelerator Type A-15 and B-16 E-19 and F-20 G-21 and J-24 M-27 and N-28 I I I III 400 (674) 400 (674) 475 (800) 400 (674) 0.43 0.40 0.43 0.40 Carbonate NC/CC NC/CC NC/No NC/CC No Water Reducer Type H-22 and I-23 I-23 and J-24 I 475 (800) 0.43 Carbonate No E/A A/No w/c Ratio. In the case where a non-chloride–based accelerator was used, it was found that decreasing the w/c ratio from 0.43 to 0.40 increased the 28-day compressive strength (test 25) and 24-hour flexural strength (test 29). Similarly, the 20-hour and 24-hour compressive strength (tests 23 and 24) increased when the w/c ratio was decreased from 0.43 to 0.40 in. mixtures using calcium chloride accelerator. Also, the CTE (test 1) and maturity (test 18) both increased with a decrease in the w/c ratio. Coarse Aggregate Type. Aggregate type had a significant impact on the measured density (tests 30 through 35), with mixtures made with the carbonate coarse aggregate having the lowest density and those made with the siliceous coarse aggregate having the highest density. The CTE (test 1) was higher for the mixture made with the gravel coarse aggregate. Mixtures with gravel had lower 20-hour compressive strength (test 23) and 24-hour flexural strength (test 29) than the other mixtures. Also, mixtures made with a cement factor of 400 kg/m3 (674 lb/yd3)

140 and siliceous coarse aggregate had significantly higher scaling than the other mixtures. Using a higher cement factor (475 kg/m3 [800 lb/yd3]) and no accelerator changed the results slightly, with the gravel mixtures having some minor scaling (rating of 1), the siliceous mixtures having even less scaling (rating of 0.50), and the carbonate mixtures having no scaling. Accelerator Type. Although many results were influenced by the specific mixture parameters, it was generally observed that the use of calcium chloride increased compressive strength at various ages. In some cases, the use of calcium chloride increased scaling, decreased paste homogeneity, and produced poorer air-void system parameters. Water Reducer Type. The maturity test results (tests 17 and 18) were highest when no water reducer was used, followed by the mixture with the Type E water reducer. The mixture containing the Type A water reducer had the lowest maturity values. Similarly, the mixture without water reducer also had the highest 20-hour flexural strength (test 28) of the three, followed by the mixture made with the Type E water reducer. The mixture made with Type A water reducer had the lowest 20-hour flexural strength. The mixture made with the Type E water reducer had less paste homogeneity (test 45) and more microcracking (test 46) than either of the other two mixtures. The percent volume of permeable voids was highest with the mixture made with Type E water reducer, followed by the mixture made with the Type A water reducer. The mixture made without water reducer had the lowest volume of permeable voids. The mixture made with the Type A water reducer also had the smallest level of chloride penetration after the salt ponding test (test 42).

141 Two-Factor Models A couple of different two-factor analyses were possible for the 20- to 24-hour mixtures. One analysis was conducted using mixtures A-15, B-16, E-19, and F-20, in which the w/c ratio (0.40 to 0.43) and the type of accelerator used (non-chloride versus calcium chloride) were varied. Another analysis was conducted using mixtures E-19, F-20, M-27, and N-28, in which cement type (Type I versus Type III) and accelerator type (non-chloride versus calcium chloride) were varied. In the first analysis, the w/c ratio was generally more important than the accelerator type in affecting mixture properties, and meaningful two-way interactions existed for 20- and 24-hour compressive strength (tests 23 and 24) and chloride penetration (test 42). For the compressive strength tests, both w/c ratio and accelerator type were highly significant by themselves, and when analyzed together, the results were highly significant as well, with a lower w/c ratio and a calcium chloride accelerator resulting in the highest compressive strength values, whereas a high w/c ratio with a non-chloride–based accelerator resulted in the lowest compressive strength values. Chloride penetration was also significantly affected by the w/c ratio and accelerator type alone or when analyzed together. Chloride penetration was the lowest for mixtures with a low w/c ratio and a non-chloride accelerator and highest for mixtures with a high w/c ratio and a calcium chloride accelerator. In the analysis, the type of cement used in a mixture was a more important factor than the accelerator type. A meaningful two-way interaction existed for paste homogeneity (test 45),

142 where both the cement type and accelerator type were extremely significant factors individually and when combined. A combination of Type III cement and calcium chloride accelerator resulted in the lowest paste homogeneity, whereas mixtures made with a non–chloride-based accelerator and/or Type I cement had good paste homogeneity. Relationships Among Tests The relationships among the 47 individual tests were studied to see if the results from some of the complex, costly tests could be obtained from simpler, less costly tests. Details regarding this analysis are provided in Appendix D. Similar to the 6- to 8-hour mixtures, a number of strong correlations existed for different measurements made within the same test (e.g., the various measures of density and absorption reported under ASTM C 642 and the two reported maturity values made at different times) and between tests measuring similar properties (such as specific gravity measurements and density measurements and compressive and flexural strength at similar ages). The same trends observed in the 6- to 8-hour mixtures were also observed for the 20- to 24-hour mixtures in that very few correlations were found between simple, easy-to-run tests and more complex, expensive tests. For example, only a single sorptivity test (test 10) had some correlations to other tests, mainly measurements of voids made by ASTM C 642 (tests 30 and 36). Various measurements collected for ASTM C 642 had far more correlations with other tests. For example, measurements of density (tests 32–35) correlated with scaling (test 12), but this correlation most likely reflects simply the density of coarse aggregate used and is not indicative

143 of a predictive relationship. As with the 6- to 8-hour mixtures, the volume of permeable voids (test 36) correlated to some measures of strength (tests 30, 32, 33, 34, and 36) as well as to the measure of crack length per unit area (test 47). No strong correlations existed for dilation due to freezing and thawing for the 20- to 24- hour mixtures, as none of these mixtures suffered significant dilation. There was some correlation between the air content measured on fresh concrete and that measured using ASTM C 457 (see Figure 24). The air content measured in the fresh concrete was commonly higher than that measured using ASTM C 457. 3.3.3 Summary of Statistical Analysis The statistical analysis provided some insights into both the influence of mixture parameters on mixture behavior and the correlations between various tests used to assess concrete mixtures. Based on this analysis, the following findings were obtained regarding the influence of mixture parameters on concrete test behavior: • The cement type had a notable impact on various measures of early concrete strength, with Type III cements producing higher strengths at a given age than Type I cements. The increased rate of hydration was evidenced in the maturity readings for the 6- to 8-hour mixtures, which were also significantly higher. In many cases, the change from Type I to Type III cement negatively affected the air-void system parameters, most acutely when a

144 calcium chloride accelerator was used. The resistance to deicing was also compromised in the 6- to 8-hour mixtures made with Type III cement. • The cement factor had a large effect on measures of paste porosity, increased sorptivity, absorption, and percent volume of permeable void space was observed under many situations as cement content increased. Increases in paste content also increased the number of shrinkage cracks. Increasing the cement content in 6- to 8-hour mixtures decreased scaling, but also led to a reduction in some measures of early strength. • As expected, decreasing the w/c ratio increased various measures of strength. It also increased the CTE. In some cases, absorption was reduced and paste homogeneity improved when the w/c ratio was lowered.

145 Figure 24. Air content measured using ASTM C 457 (test 13) versus air content of fresh concrete for 20- to 24-hour mixtures (test 37). • As expected, the coarse aggregate type had a very significant impact on measures of density, with denser aggregate resulting in denser concrete. Mixtures made with the gravel coarse aggregate had the highest CTE, and mixtures made with the carbonate aggregate had the least scaling. Aggregate type also affected some of the strength properties of the mixtures, most notably compressive strength. • In some instances, mixtures made with a calcium chloride accelerator as opposed to a non- chloride accelerator had lower early strengths, but increased long-term strengths. In some cases, using a calcium chloride accelerator significantly increased the CTE, although the increase was small. Results on the scaling test were variable. With the use of calcium chloride, the scaling resistance was improved in some cases but deteriorated in others. 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 ASTM C457 Air Content (Percent) A ir C on te nt o f F re sh C on cr et e (P er ce nt )

146 Similar conflicting results were observed for some of the air-void system parameters and paste homogeneity. • The use of the Type F HRWR in the 6- to 8-hour mixtures reduced the air content and negatively impacted a number of the air-void system parameters, as well as the early strength characteristics of the concrete. Further, a number of the specimens made with the Type F HRWR had high dilation values in cyclic freeze-thaw testing. The use of a Type F HRWR appeared to improve paste homogeneity. The 20- to 24-hour mixtures made without water reducer had the highest maturity values, followed by those made with a Type E water reducer. Mixtures made with a Type A water reducer had the lowest maturity values. Similar results were obtained for the 20-hour flexural strength test results. Mixtures made with the Type E water reducer had the lowest level of paste homogeneity and the highest volume of permeable voids. • Raising the curing temperature had many unexpected positive results, including reduced shrinkage, improved air-void system parameters, reduced dilation due to cyclic freezing and thawing, and reduced scaling. Although some early strength measures were increased, 28-day strength was decreased in some cases. • In the limited study of interactions between mixture parameters, it was found that the w/c ratio, cement factor, and cement type were more important parameters than the type of accelerator used. In one case, a strong interaction between cement type and accelerator type was observed, where mixtures made with Type III cement and calcium chloride accelerator produced a less homogenous paste than that observed in mixtures made with Type I cement and a non-chloride accelerator.

147 Based on the statistical analysis, the following findings were obtained regarding the relationships between the various concrete tests: • As would be expected, certain tests are highly correlated one with another, such as the various measures of compressive strength, flexural strength, maturity, and density. However, few strong correlations were observed between tests that are relatively simple to conduct and more complex material characterization tests. • One measure of sorptivity correlated mildly with some elements of ASTM C 642, but overall, the sorptivity test did not correlate well with other tests. Results from ASTM C 642, however, had good correlation with a number of other tests and therefore might be worth further evaluation as a possible routine test for high early strength materials. • In cases where significant dilation occurred, dilation due to cyclic freezing and thawing correlated mildly with the spacing factor, as determined by ASTM C 457. Further, there was mild correlation between the air content measured with ASTM C 457 and that measured on the fresh concrete, but less correlation between the air content of fresh concrete and the spacing factor, as measured by ASTM C 457. • There was fairly strong correlation between spacing factor and various measures of concrete compressive strength, with increasing the spacing factor resulting in higher strength. This correlation is not simply a reflection of higher air content, but an indication that larger air voids spaced further apart produce stronger concrete.