Entrained Air-Void Systems for Durable Highway Concrete (2021)

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.

11 Experimental Program Research Scope and Approach To achieve the objectives of the project, the experimental program included the following steps: • Pavement sections with F-T related distress were identified across the U.S. and Canada. These sections were sampled for hardened air-void system analysis and investigation of potential factors contributing to the distress. • A comprehensive test matrix was designed to cover the potential scenarios for interaction between components of the cementitious system, as well as the probable interactions between the chemical admixtures. • A matrix of 144 concrete mixtures was formulated; these mixtures were tested and investi- gated for their air-void systems in the fresh and hardened states. Correlations were established between different test methods and analysis scenarios. Moreover, 54 of these concrete mixtures were reproduced in the laboratory to investigate clustering caused by retempering and variations in temperature. • Results obtained from the laboratory tests were combined with field observations to investi- gate the criteria considered necessary for securing F-T durability of highway concrete. • Various scenarios were considered for testing concrete specimens in accelerated F-T conditions. A summary of the conducted tests and measured properties is provided in Table 3. Laboratory Evaluations Materials The research included materials that represent commercially available products that may influence production and performance of the air-void system. The AEA, water-reducing admixture (WRA), and aggregates were selected to evaluate the factors reported to affect both the stability of bubbles in a fresh mixture and the risk of clustering. A limestone coarse aggregate was used for mixtures investigating stability, but a siliceous material was used for the clustering- related mixtures. Cements Three cement types were used to represent the range of products in use with chemistries that may affect the air-void system: • Type I, high-alkali (HA) content (alkali > 0.6%) • Type I, low-alkali (LA) content (alkali < 0.6%) • Type IL portland-limestone cement (TIL) The cements were analyzed for chemical composition; results are summarized in Table 4. C H A P T E R 3

12 Entrained Air-Void Systems for Durable Highway Concrete Cement Fly Ash Natural PozzolanLA HA TIL FA 4 FA 5 SiO2 21.08 19.82 20.07 40.72 38.49 56.99 Al2O3 4.19 5.43 4.90 20.45 21.92 38.06 Fe2O3 3.27 2.07 3.43 6.08 6.04 1.73 SO3 2.95 3.80 3.28 1.02 1.04 0.02 CaO 64.27 63.36 65.50 21.61 22.43 0.05 MgO 2.42 3.00 0.94 4.35 4.17 0.11 K2O 0.53 1.05 0.69 0.63 0.60 0.25 Na2O 0.13 0.36 0.13 1.36 1.53 0.00 P2O5 0.06 0.28 0.16 0.83 1.10 0.19 TiO2 0.25 0.24 0.25 1.51 1.57 1.49 BaO 0.58 0.58 0.07 SrO 0.09 0.33 0.19 0.31 0.32 0.02 Mn2O3 0.07 0.10 0.28 0.04 0.05 0.00 LOI (%) 2.00 0.21 1.54 Moisture (%) 0.07 0.03 0.46 Note: LA, HA, and TIL = low-alkali, high-alkali, and Type IL portland- limestone, respectively; LOI = losson ignition. Table 4. Chemical composition of the cementitious materials. Test SampleSource Number Specimen Air-Void Properties F-T Testing To ta l V ol um e Si ze Sp ac in g Fa ct or Si ze D is tr ib ut io n C lu st er in g C ur in g C on d. F lu id Fr ee zi ng F lu id F- T C yc le s R D M M as sL os s Su rf ac e Sc al in g Field Investigation to Verify Entrained Air-Void System Characteristics Field concrete collection and testing F-T durable (collection) Field 60 4 in. core F-T deteriorated (collection) 60 4 in. core ASTM C457 10 4 in. core X X X X Flatbed scanner 60 4 in. core X X X X X Petrography 10 4 in. core X X X X Measuring Air-Void System Characteristics—Laboratory Investigation Fresh concrete SAM Lab (duplicates) 164 0.25 cubic ft X X Hardened concrete Flatbed scanner 164 4×8 in. cylinder X X X X X ASTM C457 10 4×8 in. cylinder X X X X Fixed-focus optical microscope 10 4×8 in. cylinder X X X X Petrography 10 4×8 in. cylinder X X X X Test Methods for Evaluating F-T Durability F-T durability of concrete mixtures AASHTO T 161 “A” Lab (duplicates) 27 3×4×16 in. X X X X AASHTO T 161 CDF-A 27 3×4×16 in. X X X X X AASHTO T 161 CDF-B 27 3×4×16 in. X X X X AASHTO T 161 CDF-C 27 3×4×16 in. X X ASTM C672 27 72 in2×3 in. X X Note: Cond. Fluid = conditioning fluid, RDM = relative dynamic modulus. Table 3. Test methods and evaluated characteristics.

Experimental Program 13 Supplementary Cementitious Materials Two sources of Class C fly ash were selected from the same supplier, one with a more elevated loss on ignition (LOI) than the other. A natural pozzolan (metakaolin) was included to investigate its potential effects. Dosage rates were 20% for both fly ashes and 10% for the natural pozzolan. Air-Entraining Admixtures Because AEAs affect both stability and clustering, two products that were previously evaluated (Wang et al. 2019)—one stable and one unstable product that impacted clustering—were used: • Stable: vinsol-rosin based, anionic, with pH of 10.4–13.5 and specific gravity of 1.03 • Unstable: vinsol+amine+fatty acid, anionic, with pH of 11.0 and specific gravity of 1.03 Water-Reducing Admixtures Two polycarboxylate based WRA products that were previously evaluated (Wang et al. 2019)— one that affects stability and one that does not—were used. Aggregate Combinations Two types of coarse aggregate and two types of fine aggregate were used. Combinations were selected to ensure the combined gradation fell within the envelope recommended using Ley’s Tarantula curve for SAM numbers (Ley et al. 2012): • Natural fine. River sand with 1.49% water absorption and 2.65 specific gravity in a saturated surface dry (SSD) condition. • Manufactured fine. Crushed limestone with 2% water absorption and 2.65 specific gravity in an SSD condition. • Crushed limestone. Coarse with 1.0% water absorption and 2.68 specific gravity in an SSD condition. • Gravel. River-bed gravel with 1.67% water absorption and 2.68 specific gravity in an SSD condition. The following aggregate combinations were used: • Natural fine + limestone coarse • Natural fine + manufactured fine + limestone coarse • Natural fine + siliceous coarse (used for the clustering activity) Figure 5 shows the particle size distribution of the individual aggregate sources. Aggregates were air dried at room temperature for 24 hours before use to ensure a uniform temperature and moisture condition. The moisture contents were determined afterward and compared with the water absorption capacities in the SSD condition to ensure proper adjust- ments in the mixing water required for a uniform w/cm ratio. Test Matrix The main test matrix involved preparation of 144 concrete mixtures with different binary combinations of cements and supplementary cementitious materials (SCMs), different combi- nations of AEAs and WRAs, and two types of fine aggregate. Each combination was used to produce concrete mixtures with acceptable and marginal air content. The mixture proportions were based on the following parameters. • W/cm ratio = 0.42 • Target air content = 4% (marginally unacceptable) or 6% (acceptable)

14 Entrained Air-Void Systems for Durable Highway Concrete • Binder content = approximately 565 lb/yd3 • Temperature = 70°F Table 5 shows the test matrix used for the laboratory mixtures. The test matrix for the clustering study involved the preparation of 61 concrete mixtures prepared with different binary combinations of cements and fly ashes, different combinations of AEAs and WRAs, and two types of coarse aggregate. Selection of the parameters was based on the literature, and the mixture proportions were based on the following parameters. • W/cm ratio = Start with 0.42 and increase to 0.45 with retempering or 0.45 for control mixtures Figure 5. Particle size distribution of the fine and coarse aggregate sources. Factor Variant # of Variants Total Combinations Cement Type Low-alkali 3 144 combinations for full parametric study (details are provided in Appendix A) High-alkali Type IL portland-limestone SCM Type Higher-LOI fly ash 3 Lower-LOI fly ash Natural pozzolan AEA Stable 2 Unstable WRA Compatible 2 Incompatible Design Fresh Air Low 2 High Fine Aggregate River sand 2 Manufactured sand Coarse Aggregate Crushed limestone 1 Table 5. Test matrix for laboratory mixtures.

Experimental Program 15 • Target air content = 6% • Binder content = approximately 549 lb/yd3 • Mixing water temperature = 70°F or 90°F Table 6 shows the test matrix used for clustering mixtures. Test Procedures and Sample Preparation Investigating Air-Void Systems of Concrete in the Fresh State Using the Super Air Meter Data for fresh concrete were obtained by two operators using two different SAMs. Both opera- tors measured the fresh air content and the SAM number. Testing was performed immediately after the end of mixing. The same consolidation procedure was used by the operators and the test duration was similar for both. The SAM test was performed on fresh concrete according to AASHTO TP 118. Other tests conducted included the following: • Compressive strength according to AASHTO T 22, and • Petrographic analysis on selected specimens in accordance with ASTM C856 (see Appendix B). Automated System (Fixed-Focus Optical Microscope) An automated stereoscopic microscope was also employed for measuring the properties of the air-void system of hardened concrete. The system complies with requirements of the linear traverse method, as described in ASTM C457. The device generates two series of air-void data for hardened concrete extracted based on measurements of (a) chords longer than 30 microns and (b) all chords. Previous data obtained by the research team suggest that the “all chords” data obtained from this automated system overestimates the number of voids in a hardened concrete sample (Appendix C). Therefore, the research team modified the methodology intro- duced in ASTM C457 by excluding the chords shorter than 30 microns. Factor Variant # of Variants Total Combinations Cement type Low-alkali 3 61 selected combinations for partial parametric study (details are provided in Appendix A) High-alkali Type IL portland-limestone SCM type Low-LOI fly ash 3 High-LOI fly ash Natural pozzolan AEA Stable 2 Unstable WRA Compatible 2 Incompatible Fresh air content Design for 6% 1 Fine aggregate River sand 2 Manufactured sand Coarse aggregate Crushed limestone 2 River gravel Water content—w/cm ratio Before retempering—0.42 3 After retempering—0.45 No retempering—0.45 Water temperature (°F) Low (70) 2 High (90) Concrete temperature (°F) Low (70) 2 High (90) Proportions and procedures used to prepare the laboratory mixtures are provided in Appendix A. Table 6. Test matrix for the clustering mixtures.

16 Entrained Air-Void Systems for Durable Highway Concrete Flatbed Scanner The procedure for scanning hardened concrete samples was reported by Appropedia (2017), Peterson (2010, 2015), and Peterson et al. (2010) and is described in Appendix A. It is extremely important to use the scanned image in a raw format without any image enhancing features and to optimize the air content threshold and void frequency threshold values to ensure accurate outputs. The threshold values can vary for different laboratories. The optimization outputs are discussed in Appendix D. Clustering The clustering rating system proposed by Kozikowski et al. (2005) was adopted. This method involves inspection of each coarse aggregate particle available in a polished concrete sample under an optical microscope. Thirty-five coarse aggregate particles larger than 1⁄4 in. in diameter were selected for each sample. An example of a hardened sample used for determining the clustering rate is shown in Figure 6. Clustering of air voids around each aggregate particle is rated from 0 for no clustering to 3 for severe clustering, as shown in Figure 7. The average clustering rate of the sample is then calculated based on the rates obtained for the 35 individually inspected coarse aggregate particles using the following equation: Average Clustering Rate rate for individual inspected aggregate ii 35 1 35∑ )(= = Freeze–Thaw Testing Methods A CDF test method that represents a substantial procedural and functional modification to the AASHTO T 161 “A” test method is proposed. For each test variant (CDF-A, CDF-B, and CDF-C), three versions (FT-1 through FT-9) were investigated to replicate different sets of F-T conditions, as listed in Table 7. Figure 6. Marking the coarse aggregate particles for determining individual clustering rates.

Experimental Program 17 Source: Kozikowski et al. 2005. (a) (b) (c) (d) Rating 0—No clustering present Rating 1—Minor clustering Rating 2—Moderate clustering Rating 3—Severe clustering Figure 7. Rating system for clustering: (a) 0—No clustering present, (b) 1—Minor clustering, (c) 2—Moderate clustering, and (d) 3—Severe clustering. Test Variant Test Curing Time Conditioning Fluid Freezing Fluid Number of Cycles Total Test Duration CDF-A FT-1 7 days Limewater Deionized 56 63 days FT-2 28 days Limewater Deionized 56 84 days FT-3 56 days Limewater Deionized 56 112 days CDF-B FT-4 7 days Limewater NaCl (3%) 56 63 days FT-5 7 days Limewater CaCl2 (3%) 56 63 days FT-6 7 days Limewater MgCl2 (3%) 56 63 days CDF-C FT-7 7 days NaCl (35%) Deionized 56 84 days FT-8 7 days CaCl2 (74%) Deionized 56 84 days FT-9 7 days MgCl2 (54%) Deionized 56 84 days Note: Shading indicates varied parameters. Table 7. Freeze–thaw testing evaluation parameters.

18 Entrained Air-Void Systems for Durable Highway Concrete The CDF-A variant investigated the impact of sample curing time on F-T performance in water. The CDF-B variant replaced the freezing fluid with standard deicer salts similar to ASTM C672 or RILEM Technical Committee (TC) 117 (Setzer et al. 1996) and RILEM TC 176 (Setzer et al. 2004), albeit with the inclusion of CaCl2 and MgCl2. The CDF-C variant was the greatest departure from current testing methods, where dry specimens were conditioned in concentrated deicer salt solutions prior to F-T cycling. CDF-C was designed to simulate the current practice of brine pretreatment on dry pavements before winter storm events in cold climates. In all variants of the proposed CDF test, freezing fluid was introduced one-dimensionally by capillary action through the bottom of the specimen. The sides of the specimens were sealed using butyl rubber tape with aluminum backing meeting RILEM TC 117 recommendations (Setzer et al. 1996). Other functional changes included separation of the specimen container from the freezing table using an ethylene glycol coupling fluid. Both the concrete specimen and the stainless steel container containing the specimen and freezing fluid were placed on three-dimensional (3D)-printed acrylonitrile butadiene styrene (ABS) stands, which are schematically shown in Figure 8. CDF-A: Variation in Curing Time. For the CDF-A testing variants, the standard AASHTO T 161 concrete beam samples (3 in. × 4 in. × 16 in.) were placed and allowed to cure in molds for 24 hours under plastic before demolding. After the initial 24-hour cure, specimens were demolded and placed in limewater for the remainder of the 7-day curing period. After curing, samples were placed in a drying environment (73°F and 50% relative humidity) for 21 additional days. Twenty-six days after mixing, all lateral sides were covered with aluminum- backed butyl rubber tape and placed back into the drying environment. Twenty-eight days after mixing, the exposed faces were placed on the ABS support stand in the testing pan, and Figure 8. Cross-section of beam testing arrangement for CDF test method.

Experimental Program 19 the pan was filled with limewater so that the level was 0.20 in. (5 mm) above the top of the stand for the duration of conditioning, with refills as needed. When space was not immediately available in an F-T cabinet after the 21-day drying period, samples were placed in a freezer at –4°F until testing. When needed for testing, frozen samples were removed from the freezer and immediately transferred to the drying environment to thaw for 24 hours, after which conditioning was performed. Conditioning: The samples were stored in lime water for 7 days; sample mass and relative dynamic modulus (RDM) were then measured to establish the baseline values. Samples then underwent F-T cycling in contact with deionized water, with 4 hours each for heating or cooling between to 68°F and –4°F, respectively, and then reheated to 68°F. Samples were maintained at –4°F for 3 hours and at 68°F for 1 hour, for a total cycle time of 12 hours. A tolerance of ±30 minutes and ±5°F were achieved in any of the three tanks (used as modified) during this research. Mass, RDM, and visual surface scaling were evaluated after 4, 8, 14, 28, 35, 42, and 56 F-T cycles. For CDF-A variants, initial curing time was also extended to 28 and 56 days, with the additional curing time more representative of strength gain and microstructure densification of a wider variety of concrete mixtures. CDF-B: Variation in Freezing Fluid. The AASHTO T 161 test method was viewed to lack severity when freezing and thawing in water. In ASTM C672, concrete specimens are exposed one-dimensionally to a 4% solution of CaCl2 at a slow, 1-day cycle to evaluate surface durability. However, RILEM TC 176 (Setzer et al. 2004) allows either demineralized water or 3% sodium chloride (NaCl) for scaling evaluation, while the RILEM TC 117-CDF test (Setzer et al. 1996) requires 3% NaCl. The CDF-B test variant was designed to simulate more realistic F-T conditions when deicing salts are applied during a winter weather event. CDF-B samples were conditioned using the same procedure as CDF-A:FT1 but evaluated F-T performance during freezing and thawing for a range of deicing fluids including 3% combinations of NaCl, calcium chloride (CaCl2), and magnesium chloride (MgCl2). While 2% to 4% NaCl is common for both ASTM and RILEM scaling evaluations, it is not commonly used to compare the effect on performance of deicing salts more commonly used by state DOTs on roadways. Although NaCl is the most common salt used for deicer testing (Xiao et al. 2019), the investigative parameters were designed to require testing in more severe or less harsh conditions, as seems appropriate for each region. CDF-C: Variation to Include Brine Pretreatment. The CDF-C test variant represented the greatest departure from current test methods and potentially the most significant update to F-T testing in the U.S. Saturated brine solutions are applied to dry pavements prior to winter weather events in many locations. Significant anecdotal evidence and research suggests that this practice may be responsible for the seemingly increased rate of F-T deterioration experienced in the field (Sutter et al. 2006, Taylor et al. 2012). Samples for the CDF-C method were cured and dried as previously outlined; but instead of conditioning in limewater (as used in RILEM TC 117/176, CDF-A, and CDF-B), samples were exposed to highly concentrated deicer brine solutions for 7 days. Afterward, the samples were dried for 21 days and then conditioned using deionized (DI) water for 7 days prior to F-T cycling. Applying high deicer salt concentrations to the dry samples, drying the sample, and resaturating with DI water was intended to simulate field application and produce distinctly different performance than the lower salt concentrations used for deicer scaling testing.

20 Entrained Air-Void Systems for Durable Highway Concrete Deicer Scaling. In addition to the standard and modified rapid F-T tests, deicer scaling was performed according to ASTM C672. After the prerequisite drying period, a dam was placed around the edges of the samples using a polymer-modified grout, as shown in Figure 9. After curing for 7 days, a bead of 100% silicone caulk was applied to the inner junction between the dam and each sample to minimize leaking. The sides and bottoms of the specimens were open and exposed to ambient freezer conditions. The initial scaling samples had a surface area of 110 in.2 with 80 in.2 exposed to the deicer solution after installation of the dam. Performance of Modified Freeze–Thaw Equipment and Solutions. The proposed method incorporated components of capillary damage during freezing using water or deicing solu- tions that were fundamentally different from those used in AASHTO T 161, with significant adaptation from the RILEM TC 117 (Setzer et al. 1996), RILEM TC 176 (Setzer et al. 2004), and ASTM C1585 methods. The equipment currently used for AASHTO T 161 and RILEM TC 117/176 are both designed to cycle samples through freezing and thawing conditions using different methods. The AASHTO T 161 Procedure A (AASHTO T 161 “A”) cabinet consists of a stainless steel freezing table that has embedded coolant coils. Concrete beams are placed in stainless steel pans that are filled with water and then set on the freezing table such that freezing occurs from the bottom of the sample upward. Samples are thawed using a series of resistive heaters placed between the pans; thawing occurs from the sides inward. The cabinet control and reported temperatures are measured at the center of each concrete beam. In contrast, samples tested according to RILEM TC 117/176 (Setzer et al. 1996, 2004) are placed in a testing fluid (water or deicers) in a stainless steel pan placed in a coupling fluid bath. The coupling fluid (ethylene glycol) transfers the heating and cooling load to the sample containers. Temperatures in the TC 117/176 system are determined from a probe mounted on the bottom of one of the test containers in contact with the coupling fluid. While the RILEM tests expose the samples to a much wider range of temperatures (68°F to -4°F) compared with the Figure 9. Deicer sample with polymer-modified grout around the edges after the drying period.

Experimental Program 21 AASHTO T 161 test (40°F to 0°F), the objective of both tests is to ensure complete freezing and complete thawing during each cycle. Before commencing testing on the research concrete samples, equipment control programs were evaluated in three locations. The control locations were tested in the center of the specimen (consistent with AASHTO T 161), in the coupling fluid in contact with the bottom of the F-T sample pan (similar to RILEM TC 117 from Setzer et al. 1996), and in the freezing fluid. Because both heating and cooling in RILEM TC 117 (Setzer et al. 1996) occurs through the coupling fluid, controlling the test by monitoring the coupling fluid is appropriate and was the procedure used in this study. When using the AASHTO T 161 “A” equipment, freezing occurs from the bottom with heating from the sides, and all three conditions were evaluated to focus attention on freezing behavior at the bottom of the concrete specimens. A ± 9°F (5°C) window was used to refine the ramping programs with the goal of matching the desired profile without an excessive number of heating and cooling steps, to minimize unnecessary equipment cycling and wear and tear on the equipment. After tuning, the equipment was able to maintain the desired temperature profile. Figure 10 shows a refined temperature versus time profile where a single step was used for cooling. For subsequent testing, the reference temperature was controlled from the freezing fluid with two 12-hour cycles performed per day. CDF Modified Freeze–Thaw Procedure. The F-T test methods blended the existing capabilities from the AASHTO T 161 “A” equipment with the proposed modifications to condi- tioning and curing, the introduction and type of freezing fluid, the cycle time, and the testing duration. Since the AASHTO T 161 “A” equipment is designed to freeze samples relatively quickly, ethylene glycol coupling fluid was used to buffer thermal transfer from the freezing table and the samples. Figure 11 shows the tray before installing it on the freezing table. The stainless steel tray was 67.5 in. × 20 in. × 1 in. Moistened felt was used between the freez- ing table and the stainless steel tray for the thermal connection, consistent with AASHTO T 161. -40 -20 0 20 40 60 80 100 0 1 2 3 4 5 6 7 8 9 10 11 12 Te m pe ra tu re (° F) Time (hrs) Temp Target Temp Envelope Actual Temp Figure 10. Refined temperature versus time profile.

22 Entrained Air-Void Systems for Durable Highway Concrete Figure 11. Stainless steel tray used to hold coupling fluid and deicer pans. (a) (b) Figure 12. Relative positioning of stands, coupling fluid, and resistive heaters: (a) side view and (b) top view. The 1-in. side-height was designed to provide sufficient contact between the coupling fluid and the bottom of the sample containers, while also functioning as the stand for the resistive heaters and to clear the existing distance between the freezing table and the lid (see Figure 12). ABS plastic stands were 3D-printed to provide space for the coupling fluid and space for the freezing fluid. Figure 13 shows a stand with six support points to provide the required 25/64-in. (10-mm) offset. Figure 14 shows the stand supporting an F-T pan placed in the stainless steel tray without coupling fluid. Individual concrete specimens were laterally wrapped with aluminum foil-backed butyl rubber tape where the bottom was exposed to the freezing fluid and the top was open for vapor transmission. Figure 15 shows a sample placed in the testing pan setup. Painter’s tape was used to facilitate lifting the samples into and out of the F-T pans. Plastic wrap was used to minimize evaporation of the freezing fluid and corrosion of the interior of

Experimental Program 23 Figure 13. ABS stand to support both the freezing pans and the concrete specimens. Figure 14. ABS stand supporting the freezing pan. Figure 15. Concrete specimen support by ABS stand in container (with blue tape used to lift in and out of freezing fluid). the testing equipment. Figure 16 shows samples loaded into the testing arrangement before covering them with plastic. After the prerequisite cycles were completed (4, 8, 14, 28, 35, 42, and 56 cycles), each specimen was subjected to ultrasonic cleaning for 3 minutes consistent with RILEM TC 117 (Setzer et al. 1996). Figure 17 shows loose material releasing from a specimen during cleaning. Specimen mass and RDM (ASTM C215) were measured at each set testing cycle. Field Investigations Selection of Field Investigation Sites The goal of this part of the research was to investigate the characteristics of the entrained air-void system required for F-T durability of concrete based on pavement field observations. The research team examined concrete samples with a range of performance histories collected from different regions of the U.S. and Canada. The data were used to investigate the correlation

24 Entrained Air-Void Systems for Durable Highway Concrete between air-void system parameters and F-T performance to identify the key parameters and limits that can be used to predict performance. The research team made multiple requests to several state DOTs across the U.S. and Canada to identify pavements that may have been damaged due to insufficient air-void system char- acteristics. Many agencies expressed the opinion that they were not seeing air-void-related distresses in their systems and that there were no candidate sites under their jurisdiction. Some states—mostly Midwestern—identified locations that were considered of interest to this project; core specimens, historic performance data, and the available information regarding concrete mix design and material properties were collected from these locations. The sites investigated were recommended by the agencies in response to a request to locate examples exhibiting air system-related distress. As discussed below, many of these locations were later found to be deteriorating due to failure mechanisms that were not directly related to the air-void Figure 17. Cleaning concrete specimen in ultrasonic bath before testing. Figure 16. Concrete specimen support in coupling fluid in freeze–thaw tank.

Experimental Program 25 system. A general overview of the pavement conditions and available mix design information gathered from each of the investigated locations follows. Table 8 lists the coring locations. Pennsylvania The 18-year-old test section was reported by DOT staff to have been cast on a hot day. It was severely distressed, exhibiting a high concentration of small cracks on its surface, accompanied by a high concentration of wide-open longitudinal cracks and D-cracking patterns in most slabs. Another 18-year-old section, in contrast, exhibited acceptable performance with some scaling but no signs of cracking; the joint sealers were in good condition. The only notable problem in this section was the damage due to abrasion, which resulted in exposure of polished coarse aggregates at the pavement surface. New York According to the information provided by the New York State DOT (NYSDOT) personnel, this section, located on Alternate Route 7 (NY-7) in Albany, New York, was cast in the 1970s with 20-ft-long doweled panels. The distress in this pavement was mainly in the form of joint deterioration, transverse cracking (most likely due to heavy traffic loading), and longitudinal cracking. Because the observed distress patterns did not appear to be what is expected for F-T distress, except for slight scaling, cores extracted by the NYSDOT were examined for their air-void systems. Nevada Signs of distress due to F-T cycles, along with severe D-cracking, were identified on east- bound I-80 in Humboldt County, Nevada, by the material engineers at the Nevada DOT (NDOT). The pavement was about 11 in. thick and was cast in 1995. Missouri Coring was conducted on distressed sections on northbound US-63, about 0.5 mile north of the Boone County line. This pavement was about 12.5 in. thick and was cast in 2001. A variety of conditions was observed in the section, ranging from slight or no distress to moderate and severe distress; cores were extracted from each of these areas. Minnesota US-52 Pavement sections with damaged joints were identified on US-52 near Rochester, Minnesota. The pavement was 8.0 in. thick and was cast in 1992. The pavement was investigated even though the joint deterioration was likely to be a result of a combination of oxychloride formation and frost damage. Geographic Area State/Province East Coast New York and Pennsylvania Midwest Michigan, Minnesota (two sites), Missouri, Ohio, and South Dakota West Coast Nevada Canada Manitoba and Ontario Table 8. Geographic distribution of the investigated sites.

26 Entrained Air-Void Systems for Durable Highway Concrete Minnesota MnROAD Research Facility As part of the Strategic Highway Research Program (SHRP), test sections were built at the MnROAD facility in 1992 to monitor long-term resistance of concrete to freezing and thawing (Janssen 2006). The test sections were not exposed to any deicing salts or traffic loading. The slabs were in good condition, with slight scaling after more than 26 years of exposure to natural F-T cycles. The sections, located at the maintenance yard, consisted of 15 ft × 20 ft slabs. The slabs were quartered by making two saw cuts, resulting in four panels, each measuring 7.5 ft × 9.8 ft. South Dakota Two test sites were investigated in Sioux Falls, South Dakota. One site consisted of a bridge approach with signs of severe joint distress, which could be attributed to a combination of frost damage and chemical attack; no mid-panel cracking was identified. The other section was a pavement adjacent to the DOT facility in Sioux Falls. Signs of D-cracking and joint staining were observed along with slight scaling. The pavement thickness was about 10 in. in both sections. Michigan Pavement sections with damaged joints were identified on the M-6 freeway near Grand Rapids, Michigan. The pavement was 11.0 in. thick and was cast in 2004. The joint deteriora- tion could be a result of both oxychloride formation and frost distress. Manitoba, Canada Severe pavement distress was identified on Manitoba’s Provincial Trunk Highway 30 (PTH-30/MB-30). Ohio The Ohio DOT (ODOT) records indicated that, during the winter of 2013, premature distress was observed in the form of spalling at longitudinal joints on I-75 in Allen County. The pavement was cast in the summer of 2013, and the ODOT engineers decided to conduct examinations on hardened concrete specimens. The data provided by the ODOT material engineers were synthesized; the observations were then compared with laboratory data obtained from testing the core specimens obtained from other states. Ontario, Canada Premature distress in the form of severe joint deterioration was observed in one of the Ontario Ministry of Transportation projects. This project was an 18.5-mile-long section of ON-417 east of Ottawa. The eastbound and westbound lanes were cast in 2000 and 2004, respectively, each in three stages. The first signs of distress were observed about eight years after construction. Between 2014 and 2016, about 83 cores were extracted by the Ontario Ministry of Transpor- tation from different parts of the project to identify the cause for distress in various locations of the pavement. In addition to the distressed sections, cores were extracted from two other highways in the province (ON-410 and ON-401) that are exposed to a similar climate. ON-410 and ON-401were constructed with the same specification as ON-417 but did not exhibit any of the signs of joint spalling at the surface of the concrete pavement observed on ON-417. Cores from ON-410 and ON-401 showed some joint deterioration below the surface,

Experimental Program 27 but it was much less severe than that affecting the ON-417 cores. Test results provided by the Ontario Ministry of Transportation were used in the analyses. Field Investigation of Air-Void System Characteristics Results obtained from tests on core specimens extracted from the above sites, as well as hardened air-void system data obtained from laboratory tests, are discussed in this section. Also discussed are results of petrographic examination of cores obtained from selected pavements. Pennsylvania PennDOT staff provided the following mixture design information: • Cement: Type I cement, 588 lb/yd3 • SCM: N/A • Coarse aggregate: D-57, 1,825 lb/yd3 • Fine aggregate: 1,254 lb/yd3 • W/cm ratio: 0.47 • Fresh air (%): 4.5 to 7.5 Four cores per section were extracted and analyzed; data obtained from testing the air-void systems are listed in Table 9. Cores 4-C and 4-E were investigated for the air-void systems properties based on the ASTM C457 modified point count method. Data obtained from testing these samples were in agreement with the results obtained for each of the mixtures. Both cores contained secondary ettringite that partially filled some air voids; this appeared to be more noticeable in Core 3-E. Cores were also subjected to petrographic analysis. Photomicrographs of the cores and further details regarding the petrographic examinations are provided in Appendix B. Core 4-C, from the control pavement, revealed very limited distress in the concrete. The concrete was air entrained; air-void content was 2.3%. The core exhibited minor amounts of alkali-silica reaction (ASR) and minor damage. Core 4-E, extracted from panel with severe damage, revealed very limited damage in the concrete. The top surface of the core appeared to have been cut or ground and exposed bisected coarse and fine aggregate particle and apparent subsequent wear and weathering. Core # Pavement Distress Coring Location Air (%) S.F. (in.) S.S. (in.−1) 1-E High concentration of tiny cracks at surface, several wide-open longitudinal cracks, D-cracking pattern Mid-panel 2.46 0.011 634 3-E Several wide-open longitudinal cracks, D- cracking pattern 4.03 0.015 382 4-Ea 2.5 0.013 513 5-E Close to joint 2.39 0.013 531 1-C Very low concentration of tight and small cracks at surface Close to joint 2.75 0.011 624 2-C Mid-panel 1.96 0.023 345 4-Ca 2.3 0.011 615 5-C 3.99 0.012 470 Note: apetrography data; S.F. = spacing factor, S.S. = specific surface. Table 9. Hardened air-void system data obtained from cores in Pennsylvania.

28 Entrained Air-Void Systems for Durable Highway Concrete Aside from the top surface, the concrete appeared to be in fairly good condition. The air-void content was 2.5%; some of the microcracking in the concrete may be attributed to F-T. New York NYSDOT staff provided the following mixture design information: • Cement: Type I cement, 605 lb/yd3 • SCM: N/A • Coarse aggregate: quartzite, bulk specific gravity of 2.693, and 0.5% water absorption • Fine-to-total aggregate (% vol): 35.8 • Fine aggregate: bulk specific gravity of 2.572 and 1.9% water absorption • W/cm ratio: 0.44 • Design fresh air (%): 5.0 to 8.0 • Slump range: 1 in. to 3 in. (25 mm to 75 mm) The data obtained from testing the air-void systems of the core specimens are listed in Table 10. Nevada NDOT staff provided the following mixture design information: • Cement: Type IP cement, 605 lb/yd3 • SCM: N/A • Coarse aggregate: quartzite, bulk specific gravity of 2.693, and 0.5% water absorption • Fine-to-total aggregate (% vol.): 35.8 • Fine aggregate: bulk specific gravity of 2.572 and 1.9% water absorption • W/cm ratio: 0.44 • Design fresh air (%): 5.0 to 8.0 • Slump range: 1 in. to 3 in. (25 mm to 75 mm) The data obtained from testing the air-void systems of the core specimens are listed in Table 11. Three of the seven extracted cores were highly damaged and could not be used for hardened air-void analysis; the remaining specimens were used for laboratory tests. Two of the cores were also investigated using the ASTM C457 modified point count method; these cores exhibited large voids that inflated the air content. Petrographic examination of Core 3 revealed the development of ASR involving both the coarse and fine aggregate; this damage was considered minor and not representative of the more significant damage reported elsewhere in the pavement. The concrete appeared to be Core # Pavement Distress Coring Location Air (%) S.F. (in.) S.S. (in.−1) 5 Signs of distress due to heavy traffic loading, wide-open longitudinal and transverse cracking, no clear sign of F-T distress Mid-panel 4.47 0.007 796 6 3.82 0.009 618 7 6.74 0.008 511 8 Close to joint 5.69 0.009 538 9 Close to joint 7.13 0.009 467 Note: S.F. = spacing factor, S.S. = specific surface. Table 10. Hardened air-void system data obtained from testing cores in New York.

Experimental Program 29 air entrained, but an estimated half of the air-void volume was attributed to small entrapped and possibly coalesced air voids. Also, air content appeared to be lower near the surface of the concrete than at greater depth, largely due to reduction in the coarser entrapped air voids. Nonetheless, entrained air was present in modest amounts; evidence of F-T damage was not apparent in the sample. Missouri MoDOT staff provided the following mixture design information: • Cement: Type I cement, 485 lb/yd3 • SCM: Class C fly ash 86 lb/yd3 • Coarse aggregate: 11⁄2 in. limestone, 1,933 lb/yd3 • Fine aggregate: Missouri River sand, 1,179 lb/yd3 • W/cm ratio: 0.38 • AEA: 5 to 6 oz/yd3 • Fresh air (%): 4.4 to 6.6 The data obtained from testing the air-void systems of the core specimens are listed in Table 12. Cores 1 and 3 were also investigated using the ASTM C457 modified point count method; both cores contained secondary ettringite seen to be lining to partially filling air voids. Core 9 was taken in 2007 and tested in 2018. Cores 11 through 13 were extracted and tested in 2007. Core 1, representing the severely distressed pavement, revealed the development of ASR as well as possible evidence of alkali-carbonate reaction (ACR) and F-T distress to a portion of the coarse aggregate. Although both ACR and F-T damage were observed, it was unclear which distress mechanism was predominantly responsible for the distress reported in the pavement. Thus, both mechanisms were considered contributory to distress in the concrete. The concrete appeared to be air entrained throughout with only minor apparent reduction in air volume in the near-surface region. Core 3, extracted from a slightly distressed section, revealed the development of ASR and possibly ACR and F-T damage to a portion of the coarse aggregate. However, the occurrences of each of these distress mechanisms was not sufficient to assess their contributions to distress reported elsewhere in the pavement. The concrete was air entrained, but the volume of air was somewhat low and appeared to vary across the vertical profile of the core, with less air in the upper portion of the core. Evidence of F-T damage was not readily apparent in the sample. Core # Pavement Distress Coring Location Air (%) S.F. (in.) S.S. (in.−1) 3 Signs of severe D-cracking Mid-panel 4.52 0.013 408 3a 7.40 0.010 430 5 Signs of moderate D-cracking Close to joint 3.00 0.013 513 6 Mid-panel 4.63 0.014 375 6a 7.6 0.009 381 7 Slight distress Mid-panel 1.93 0.018 432 Note: aPetrography data; S.F. = spacing factor, S.S. = specific surface. Table 11. Hardened air-void system data obtained from testing cores in Nevada.

30 Entrained Air-Void Systems for Durable Highway Concrete Minnesota US-52 MnDOT staff provided the following mixture design information: • Cement: Type I cement, 493 lb/yd3 • SCM: Class C fly ash 87 lb/yd3 • Water: 244 lb/yd3 • Coarse aggregate over 3⁄4 in.: 1,123 lb/yd3, specific gravity of 2.66, and water absorption of 1.6% • Coarse aggregate under 3⁄4 in.: 746 lb/yd3, specific gravity of 2.65, and water absorption of 1.7% • Fine aggregate: 1,200 lb/yd3, specific gravity of 2.62, and water absorption of 0.8% • W/cm ratio: 0.42 • Design fresh air (%): 5.5 The data obtained from testing the air-void systems of the core specimens are listed in Table 13. Cores 1 and 4 were also investigated using the ASTM C457 modified point count method. Both cores contained secondary ettringite seen to be lining to partially filling air voids. Minnesota MnROAD Research Facility The available mixture design information is as follows (Janssen 2006): Slabs 2 and 3 were cast with 15% fly ash replacement and 2.7% and 2.0% air content in their fresh states, respectively. Slab 4 was cast without any fly ash and 2.5% air content in its fresh state. Slabs 6 and 7 were cast with 30% fly ash replacement and 3.8% and 1.4% air content in their fresh states, respectively. The data obtained from testing the air-void systems of the core specimens are summarized in Table 14. Core # Pavement Distress Coring Location Air (%) S.F. (in.) S.S. (in.−1) 1a Severe wide-open cracks Mid-panel 4.70 0.007 635 2 Moderate cracking 1.61 0.010 833 3a Slight distress 2.50 0.009 769 4 No distress, hand placed 3.06 0.019 333 5 Severe wide-open cracks 1.73 0.014 589 6 No distress, hand placed 4.38 0.014 390 7 Slight distress 1.90 0.017 457 8 Moderate cracking 1.11 0.016 633 9b Old core, adjacent to Core 3 2.45 0.011 657 11c Old core, adjacent to Core 2 2.49 0.014 670 12c Old core, adjacent to Core 4 3.53 0.010 546 13c Old core, adjacent to Core 8 2.58 0.007 891 Note: aPetrography data, b Extracted in 2007 and stored in DOT laboratory, cExtracted and tested in 2007; S.F. = spacing factor, S.S. = specific surface. Table 12. Hardened air-void system data obtained from testing cores in Missouri.

Experimental Program 31 Core # Pavement Distress Coring Location Air (%) S.F. (in.) S.S. (in.−1) 1a Severe joint damage, no distress at mid-panels Adjacent to jointb 4.70 0.006 727 2 Close to jointc 4.15 0.012 472 3 Mid-panel 4.53 0.011 492 4a No distress in joints or mid-panel Adjacent to joint 4.70 0.007 690 5 Close to joint 2.58 0.015 451 6 Mid-panel 4.49 0.015 357 8 Close to joint 4.96 0.014 353 9 Mid-panel 2.00 0.013 595 10 Severe joint damage, no distress at mid-panels Adjacent to joint 3.78 0.014 396 11 Mid-panel 6.59 0.012 372 Note: aPetrography data, bclosest possible to joint, cabout 1ft– 2 ft from joint; S.F. = spacing factor, S.S. = specific surface. Table 13. Hardened air-void system data obtained from testing cores in Minnesota, US-52. Core # Pavement Distress Coring Location Air (%) S.F. (in.) S.S. (in.−1) 2 Slight scaling, no significant distress Mid-panel 3.66 0.012 503 3 3.94 0.014 363 4 4.34 0.014 373 6 3.99 0.010 531 7 3.07 0.009 703 Note: S.F. = spacing factor, S.S. = specific surface. Table 14. Hardened air-void system data obtained from testing cores in Minnesota, MnROAD. South Dakota The available mixture design information provided by South Dakota DOT staff is summa- rized as follows: • Cement: Type I cement, 557 lb/yd3 • SCM: Class F fly ash 123 lb/yd3 • Water: 260 lb/yd3 • Coarse aggregate: 1,692 lb/yd3, quartzite type, specific gravity of 2.63, and water absorption of 0.4% • Fine aggregate: 1,175 lb/yd3, specific gravity of 2.63, water absorption of 1.3%, and fineness modulus of 2.81 • W/cm ratio: 0.38 • Design fresh air (%): 6.5 The data obtained from testing the air-void systems of the core specimens are listed in Table 15. Cores 5 and 6 were also investigated using the ASTM C457 modified point count method. Both cores contained secondary ettringite in the air voids.

32 Entrained Air-Void Systems for Durable Highway Concrete Michigan The available mixture design information provided by Michigan DOT staff is summarized as follows: • Cement: Type I cement, 526 lb/yd3 • SCM: N/A • Water: 236 lb/yd3 • Coarse aggregate: 1,692 lb/yd3, 3⁄4 in. maximum aggregate size, and specific gravity of 2.63 • Fine aggregate: 1,338 lb/yd3, specific gravity of 2.71, and fineness modulus of 2.65 • W/cm ratio: 0.45 • Design fresh air (%): 6.5 The data obtained from testing the air-void systems of the core specimens are summarized in Table 16. Manitoba, Canada Manitoba Infrastructure Department staff provided the following mixture design information: • Cement: Type I cement, 525 lb/yd3 • SCM: N/A • Water: 235 lb/yd3 • Coarse aggregate: 1,950 lb/yd3 • Fine aggregate: 1,390 lb/yd3 • W/cm ratio: 0.45 • Design fresh air (%): 5 to 7; minimum recorded: 3.2, maximum recorded: 12.1 Core # Pavement Distress Coring Location Air (%) S.F. (in.) S.S. (in. 2 Severe joint damage, no distress at mid-panels Adjacent to joint 2.86 0.014 456 3 Mid-panel 1.13 0.022 440 5a Signs of D-cracking and joint staining, good concrete at center of slabs Adjacent to joint 4.00 0.007 816 6a Mid-panel 5.30 0.009 556 6 2.83 0.014 479 8 Mid-panel 7.59 0.008 456 Note: aPetrography data; S.F. = spacing factor, S.S. = specific surface. −1) Table 15. Hardened air-void system data obtained from testing cores in South Dakota. Core # Pavement Condition Coring Location Air (%) S.F. (in.) S.S. (in.−1) 1 Severe transverse joint distress, no distress at mid-panels Close to transverse joint 4.48 0.013 418 2 3.97 0.012 467 7 Severe transverse joint distress, no distress at mid-panels Close to longitudinal joint 4.14 0.009 589 8 3.27 0.010 619 Note: S.F. = spacing factor, S.S. = specific surface. Table 16. Hardened air-void system data obtained from testing cores in Michigan.

Experimental Program 33 The data obtained from testing the air-void systems of the core specimens are listed in Table 17. Ohio The hardened air-void system of the concrete was investigated for all cores. Three types of concrete mixtures were used for casting the pavements. All mixtures were designed with a w/cm ratio of 0.44 and 6% air in the fresh state. ODOT staff provided the mixture design data listed in Table 18. The petrography data for the investigated cores indicated that the w/cm ratio ranged from 0.42 to 0.54 for areas with severe distress, 0.44 to 0.49 for areas of moderate distress, 0.39 to 0.50 for areas exhibiting light distress, and 0.43 to 0.49 for areas with no distress. Air-void system data are presented in Table 19. Ontario, Canada Ontario Ministry of Transportation staff provided the mixture design information listed in Table 20. Air-void system data are listed in Table 21. Summary of Findings from Field Analyses Based on the in situ condition assessments, the performance of investigated pavements was categorized into three classes—good, marginal, and poor—depending on the severity of the distress attributed to frost durability, while considering the age of the pavement. Table 22 presents a summary of the observed performance at the investigated sites. Core # Pavement Distress Air (%) S.F. (in.) S.S. (in.−1) 1 Moderate-severe joint distress, surface scaling 5.14 0.013 391 2 3.49 0.013 442 3 4.05 0.011 487 5 10.00 0.007 417 Note: S.F. = spacing factor, S.S. = specific surface. Table 17. Hardened air-void system data obtained from testing cores in Manitoba, Canada. Mix # Cement (lb/yd3) Fly Ash (lb/yd3) Coarse Aggregate (lb/yd3) Fine Aggregate 1 (lb/yd3) Fine Aggregate 2 (lb/yd3) Extracted Core # 1 475 110 1,880 1,275 – 1–6, 11, 12, 16–23 2 439 110 2,028 785 314 7–10 3 438 110 1,870 640 637 13–15 Source: Ohio DOT. Table 18. Concrete mix design information for sections investigated in Ohio.

34 Entrained Air-Void Systems for Durable Highway Concrete Core # Pavement Distress Coring Location Air (%) S.F. (in.) S.S. (in.−1) 1 Severe joint distress Longitudinal joint 2.75 0.006 1,061 2 Mid-panel 4.19 0.005 1,031 3 Transverse joint 5.13 0.005 946 4 Moderate joint distress Longitudinal joint 5.17 0.005 882 5 Mid-panel 3.55 0.004 1,281 6 Mid-panel 2.70 0.005 1,263 7 No distress Longitudinal joint 4.88 0.006 867 8 Close to transverse joint 2.46 0.005 1,280 9 Longitudinal joint 4.11 0.005 959 10 Mid-panel 2.08 0.004 1,446 11 Longitudinal joint 4.89 0.007 658 12 Mid-panel 4.76 0.005 1,011 13 Light joint distress Longitudinal joint 5.16 0.004 1,237 14 6.15 0.003 1,115 15 5.15 0.004 1,101 16 No distress Mid-panel 4.50 0.004 1,090 17 Mid-panel 3.49 0.005 1,221 18 Light joint distress Mid-panel 2.45 0.005 1,238 19 4.02 0.005 932 20 2.24 0.005 1,453 21 2.82 0.004 1,458 22 Mid-panel Mid-panel Mid-panel Mid-panel Mid-panel Mid-panel 2.33 0.005 1,387 23 Close to longitudinal joint 2.92 0.006 943 Source: Ohio DOT. Table 19. Hardened air-void data for pavement investigated in Ohio. Highway Stage Cementitious Content (lb/yd3) Slag Cement (%) Average 28-Day Compressive Strength (psi) Extracted Core # 417 1 Ranging from 505 to 600 for all mixtures Up to 25% permitted 5,295 14 2 4,770 5, 20 3 5,295 24, 35, 44 401 NA 6,310 C2T, C4T 410 NA 2C, 5C Source: Ontario Ministry of Transportation. Table 20. Concrete mix design information for sections investigated in Ontario, Canada.

Experimental Program 35 Core # Pavement Distress Coring Location Air (%) S.F. (in.) S.S. (in. −1) 5 Joint distress/good quality at mid-panel Mid-panel 5.1 0.009 432 14 5.5 0.007 685 20 5.1 0.005 838 24 6.0 0.005 584 35 2.3 0.006 914 44 3.1 0.006 1,016 C2T No surface distress Transverse joint 6.2 0.006 513 C4T Joint distress 4.7 0.006 592 2C No significant distress, very slight scaling Mid-panel 5.0 0.004 810 5C 5.7 0.004 1,054 Source: Ontario Ministry of Transportation. Table 21. Hardened air-void data for sections investigated in Ontario, Canada. Geographic Location Core # Type of Distress Scaling Rating (ASTM C672) Overall Performance Pe nn sy lv an ia 1-E F-T, D-Cracking, Scaling 3 Poor 3-E F-T, D-Cracking, Scaling 3 Poor 4-E F-T, D-Cracking, Scaling, ASR 3 Poor 5-E F-T, D-Cracking, Scaling 3 Poor 1-C Abrasion, Scaling 2 Good 2-C Abrasion, Scaling 2 Good 4-C Abrasion, Scaling, ASR 2 Good 5-C Abrasion, Scaling 2 Good N ew Y or k 5 Scaling 2–3 Good 6 Scaling 2–3 Good 7 Scaling 2–3 Good 8 Scaling 2–3 Good 9 Scaling 2–3 Good N ev ad a 3 F-T, D-Cracking, Scaling, ASR 2 Poor 5 F-T, D-Cracking, Scaling 2 Marginal 6 F-T, D-Cracking, Scaling, ASR 2 Marginal 7 F-T, Scaling 2 Good M is so ur i 1 F-T, D-Cracking, Scaling, ASR, ACR 1–2 Poor 2 F-T, Scaling 1–2 Marginal 3 F-T, Scaling, ASR, ACR 1 Good 4 F-T, Scaling 1–2 Good 5 F-T, Scaling 2–3 Poor 6 F-T, Scaling 1 Good 7 F-T, Scaling 1 Good 8 F-T, Scaling 1–2 Marginal Table 22. Summary of pavement distress survey at the investigated sites. (continued on next page)

36 Entrained Air-Void Systems for Durable Highway Concrete Geographic Location Core # Type of Distress Scaling Rating (ASTM C672) Overall Performance M ic hi ga n 1 F-T, Potential Oxychloride NA Poor Joints 2 Scaling 2 Good 7 Scaling 2 Good 8 Scaling 2 Good M an ito ba , C an ad a 1 F-T, Scaling, Potential Oxychloride 5 Poor 2 F-T, Scaling, Potential Oxychloride 5 Poor 3 F-T, Scaling, Potential Oxychloride 5 Poor 5 F-T, Scaling, Potential Oxychloride 5 Poor O hi o 1 Spalling NA Poor Joints 2 No Distress 0 Good 3 Spalling NA Poor Joints 4 No Distress NA Good 5 No Distress 0 Good 6 No Distress 0 Good 7 No Distress NA Good 8 No Distress 0 Good 9 Surface Crack NA Good 10 No Distress 0 Good 11 Spalling NA Good 12 No Distress 0 Good 13 Surface Crack NA Poor Joints 14 No Distress 0 Good 15 No Distress 0 Good M in ne so ta U S- 52 1 F-T, Potential Oxychloride NA Poor Joints 2 Scaling 2 Good 3 Scaling 1–2 Good 4 Scaling 1 Good 5 Scaling 1–2 Good 6 Scaling 1–2 Good 8 Scaling 1 Good 9 Scaling 1 Good 10 F-T, Potential Oxychloride NA Poor Joints 11 Scaling 2 Good 2 Scaling 1 Good 3 Scaling 1 Good 4 Scaling 1 Good 6 Scaling 1 Good 7 Scaling 1 Good So ut h D ak ot a 2 F-T, Potential Oxychloride NA Poor Joints 3 Scaling 2 Good 5 F-T, D-Cracking, Scaling 2 Good 6 Scaling 2 Good 8 Scaling 2 Good M in ne so ta M nR O A D Table 22. (Continued).

Experimental Program 37 The air content, spacing factor, and specific surface data were categorized into three levels— good, marginal, and poor—based on proposed limits: • Hardened air content – Good: greater than 5.0% – Marginal: 3.5 to: 5.0% – Poor: less than 3.5% • Spacing factor – Good: less than 0.008 in. – Marginal: 0.008 to 0.0108 in. – Poor: greater than 0.0108 in. • Specific surface – Good: greater than 600 in.–1 – Marginal: 390 to 600 in.–1 – Poor: less than 390 in.–1 Figures 18 through 20 present a summary of spacing factor, specific surface, and air content data obtained for the investigated cores. These observations suggest that when exposed to F-T cycles, concretes with similar air-void systems can exhibit a wide range of performance from good durability to severe distress. Other factors often contribute to distress in cold climates, including that caused by alkali-susceptible aggregates or chemical attack by deicers, with multiple mechanisms often having a synergistic effect on each other. In other words, securing a specific range of air content, specific surface, or spacing factor values may not necessarily guarantee frost durability. Further analysis was, therefore, limited to the performance of cores that exhibited acceptable durability against F-T. A summary of the data obtained from such cores is presented in Figure 21 and Figure 22. Geographic Location Core # Type of Distress Scaling Rating (ASTM C672) Overall Performance 16 No Distress 0 Good 17 No Distress 0 Good 18 No Distress 0 Good 19 No Distress 0 Good 20 No Distress 0 Good 21 No Distress 0 Good 22 No Distress 0 Good 23 No Distress 0 Good O nt ar io , C an ad a 5 Scaling 1 Good 14 Scaling 1 Good 20 Scaling 1 Good 24 Scaling 1 Good 35 Scaling 1 Good 44 Scaling 1 Good C2T No Distress NA Good C4T F-T, Scaling, Potential Oxychloride NA Poor Joints 2C Scaling 1 Good 5C Scaling 1 Good Note: F-T = freeze–thaw cracking, ASR = alkali-silica reaction, ACR = alkali-carbonate reaction. Table 22. (Continued).

38 Entrained Air-Void Systems for Durable Highway Concrete Note: MB-CA = Manitoba, Canada; MI = Michigan; MNROAD = MnROAD; MN- US52 = Minnesota US-52; MO = Missouri; NV = Nevada; NY = New York; OH = Ohio; ON-CA = Ontario, Canada; PA = Pennsylvania; SD = South Dakota; SF = spacing factor. Figure 18. Summary of spacing factor data. Note: MB-CA = Manitoba, Canada; MI = Michigan; MNROAD = MnRoad; MN- US52 = Minnesota US-52; MO = Missouri; NV = Nevada; NY = New York; OH = Ohio; ON-CA = Ontario, Canada; PA = Pennsylvania; SD = South Dakota; SS = specific surface data. Figure 19. Summary of specific surface data.

Experimental Program 39 Note: MI = Michigan; MNROAD = MnROAD; MN-US52 = Minnesota US-52; MO = Missouri; NV = Nevada; NY = New York; OH = Ohio; ON-CA = Ontario, Canada; PA = Pennsylvania; SD = South Dakota. Figure 21. Hardened air content and spacing factor data of cores extracted from concrete segments with acceptable durability. Note: MB-CA = Manitoba, Canada; MI = Michigan; MNROAD = MnROAD; MN-US52 = Minnesota US-52, MO = Missouri; NV = Nevada; NY = New York; OH = Ohio; ON-CA = Ontario, Canada; PA = Pennsylvania; SD = South Dakota. Figure 20. Summary of hardened air-void data.

40 Entrained Air-Void Systems for Durable Highway Concrete In Figure 21, most of the data points for samples exhibiting acceptable durability fall in the right-bottom corner, corresponding to a spacing factor lower than 0.014 in. and hardened air content higher than 2.2%. In Figure 22, most of the data points for samples exhibiting acceptable durability fall in the top-right corner, with a specific surface greater than 400 in.–1 and hardened air content greater than 2.2%. These observations indicate that cut-off limits of 0.008 in. and 600 in.–1 for the spacing factor and specific surface, respectively, may not guarantee acceptable performance. The field data also contained several cases of distressed concrete, where the hardened air-void system measurements may be considered satisfactory. Thus securing a “good” air-void system cannot be a guarantee of long-term durability A review of the hardened air-void system characteristics of the cores obtained from different geographic locations suggests that the following limits are in good agreement with long-term F-T durability of the investigated concrete pavements: • Hardened air content greater than 2.2%, • Spacing factor less than 0.014 in., and • Specific surface greater than 400 in.–1. Laboratory Investigation The goal of this step of the research was to investigate the test methods for measuring air-void characteristics and to establish correlations with field observations discussed earlier. The study included investigation of the following: • Air-void systems of concrete in the fresh state, • Air-void systems of hardened concrete, and • Clustering of air voids in hardened concrete specimens. Note: MI = Michigan; MNROAD = MnROAD; MN-US52 = Minnesota US-52; MO = Missouri; NV = Nevada; NY = New York; OH = Ohio; ON-CA = Ontario, Canada; PA = Pennsylvania; SD = South Dakota. Figure 22. Hardened air content and specific surface data of cores extracted from concrete segments with acceptable durability.

Experimental Program 41 A test matrix of 144 concrete mixtures was used for evaluation in this study (details are provided in Appendix A). Concrete mixtures were developed in the laboratory and their air-void systems were tested in the fresh and hardened states; the results were compared with the core data described in the previous section. Measurement in the Fresh State The SAM was used to evaluate the air-void systems of concrete in the fresh state. In addition to the total volume of air in the fresh state, the SAM provides a number, ranging from zero to almost 1.0, which is an index of the quality of the air-void system in fresh concrete. A lower SAM number suggest an air-void system of a higher quality and a higher percentage of uniformly distributed small air bubbles. Repeatability of the SAM results was investigated in this project. Repeatability of Air Content Using SAM For each mixture, two measurements were made by two operators using two different SAMs to investigate the repeatability of measurements; average values were compared with the measurements of the hardened air-void systems. As previously noted, all concrete mixtures were designed to exhibit two levels of air content in the fresh state: marginal air content (3% to 4.5%) and acceptable air content (5% to 7.5%). Repeatability of the test data obtained from the SAM was initially shown to be one of the main challenges in the laboratory. Trouble- shooting the test equipment and training the operators in how to comply with the test procedures improved the repeatability between devices and operators. Figure 23 presents the correlation between values of air content measured in the fresh state using two different meters. The line of equality falls in the middle of the 95% confidence interval. The average standard deviation of fresh air content for the 144 mixtures was 0.25%. Figure 23. Repeatability of air content measurement using two different SAMs.

42 Entrained Air-Void Systems for Durable Highway Concrete Figure 24 presents the histogram of the variation in fresh air content measured using two different SAMs for the data obtained for the 144 concrete mixtures. For 75 (52%) of the mixtures, variation in air content measured by two different meters was within 0.3%. For 120 (83%) of the mixtures, variation in air content measured by two different meters was within 0.6%. Only two of the measurements exhibited differences in air content of more than 1.0%. Based on these data, it can be concluded that the fresh air content measure- ments were repeatable regardless of the binder composition, aggregate type, admixture type or combination, or total fresh air content. Repeatability of SAM Number Measurements Figure 25 presents the variations in SAM numbers measured using two different meters for 144 concrete mixtures. A linear correlation was observed for two SAM number measurements. A cut-off limit of 0.20 for the SAM number, as suggested by Ley et al. (2017), was used for the comparisons in Figure 25. The shaded areas in this graph represent agreement between the two SAM number measurements and show that both meters reported “acceptable” or “unacceptable” test data for the same sample. For 129 of the 144 mixtures, the SAM numbers obtained by two different operators fell into the zones of agreement with an overall 90% agreement between two SAM number measurements. The average standard deviation of SAM number measurements for 144 mixtures was 0.05, which is in agreement with previous findings reported by Ley et al. (2017). The data in Figure 25 also suggest higher scatter in data for SAM number values that are higher than 0.40 because of the random distribution of bubbles in a poor air-void system. Further breakdown of SAM data, shown in Figure 26, reveals that the difference in the two SAM numbers for 39 of the mixtures was no more than 0.03. Based on these data, it can be concluded that the SAM number measurements were sufficiently repeatable regardless of the binder composition, aggregate type, admixture type or combination, or total fresh air content. Figure 24. Variations in air content measurements using two different SAMs.

Experimental Program 43 SAM#—Measurement 1 SA M #— M ea su re m en t 2 Figure 25. Repeatability of SAM number measurements using two different meters. Figure 26. Variations in SAM number measurement using two different meters.

44 Entrained Air-Void Systems for Durable Highway Concrete SAM Number Versus Fresh Air Content: Effect of Concrete Mixture Properties Data obtained from this study reveal that correlations between air content and SAM number can vary depending on the mixture chemistry. Figure 27 through Figure 29 present the cor- relation between the average SAM number and average air content measurements that were obtained for the mixtures prepared with LA cement. Note: AEA = air-entraining agent, FA 4 = higher-LOI fly ash, LA = low-alkali. Figure 27. Correlation between SAM number and fresh air content for concrete made with low-alkali cement and higher-LOI fly ash. Note: AEA = air-entraining agent, FA 5 = lower-LOI fly ash, LA = low-alkali. Figure 28. Correlation between SAM number and fresh air content for concrete made with low-alkali cement and lower-LOI fly ash.

Experimental Program 45 For all mixtures, a linear correlation was observed between the SAM number and air content for concrete in the fresh state, and the correlations were similar for the mixtures prepared with stable or unstable AEA for each SCM type. When using LA cement regardless of SCM type and AEA stability, an air content of 6.5% to 7% in the fresh concrete can result in a SAM number of 0.20 or less, which is considered an acceptable SAM number. Data for the mixtures prepared with TIL cement also revealed similar trends regardless of the AEA stability and the type of SCM incorporated. Data obtained for the mixtures prepared with HA cement did not show strong correlations, indicating some effect of alkali content on the variability of the test method. Data obtained from comparing SAM numbers for different concrete mixtures are presented in Appendix C. Measurement of Hardened Concrete Two different test methods were used to investigate the air-void system in the hardened state, based on recommendations and methodology described by ASTM C457. The first method involved the use of an automated fixed-focus optical microscope. The second method involved use of a flatbed scanner to obtain high-quality images from treated concrete surfaces, followed by image analysis. Hardened air content, spacing factor, and specific surface data obtained using both methods are discussed in the following sections. Hardened Data from Fixed-Focus Optical Microscope The hardened concrete specimens were tested using the automated system to measure spacing factors and specific surface areas. Data from duplicate tests for each of the 144 mixtures were averaged and reported for each concrete mixture. The data for the 144 laboratory mix- tures revealed average standard deviations of 0.0026 in., 64 in.-1, and 0.6% for spacing factor, specific surface, and hardened air content measurements, respectively (Appendix C provides further details). Figure 30 presents the relationship between the hardened air content and spacing factors. Note: AEA = air-entraining agent, LA = low-alkali, Nat. = natural pozzolan. Figure 29. Correlation between SAM number and fresh air content for concrete made with low-alkali cement and natural pozzolan.

46 Entrained Air-Void Systems for Durable Highway Concrete Air Content. The data from the optical microscopy tests were compared with those from the SAM tests conducted on the fresh concrete. Average values of air content obtained from the SAM versus those determined using the automated system are presented in Figure 31. In general, the air content values obtained in the fresh state were higher than those measured in the hardened state; a breakdown of this variation is presented in Figure 32. In 59% of the mixtures tested, the difference between the fresh and hardened air content was ≤ 1.0%. Fine Aggregate Type: No significant difference was observed for the two groups of mixtures made with or without manufactured sand. Paste Components: Cement Properties, SCM Type, and Chemical Admixture Stability: In general, results obtained for LA and TIL cements indicated correlations between the fresh and hardened air content measurements. For HA cement, a higher scatter in results was observed. The type of SCM also had a major effect on the relationships. SAM Number Versus Spacing Factor. In general, it was observed that an increase in the SAM number corresponds to an increase in the spacing factor. Figure 33 presents the logarithmic correlation between the spacing factor and the SAM number. Figure 34 presents the distribution of SAM numbers compared with the spacing factor results from testing the hardened concrete samples using the fixed-focus optical microscope. Data obtained from testing 144 concrete mixtures were analyzed by comparing a range of SAM cut-off limits with a spacing factor limit of 0.008 in., to divide the graphs into four different zones: • Good air-void system at bottom left, • Poor air-void system at top right, • False positive SAM number (i.e., good SAM number but poor spacing factor) at top left, and • False negative SAM number (i.e., poor SAM number but good spacing factor) at bottom right. The number of data points that fell in each of the zones were counted and are shown in Table 23. Figure 30. Correlation between average hardened air content and spacing factors obtained using the fixed-focus optical microscope.

Experimental Program 47 Figure 31. Correlation between average fresh air content obtained using SAM test and average hardened air content obtained with fixed-focus optical microscope. Figure 32. Extent of variations in fresh air content and air content measurement from fixed-focus optical microscope.

48 Entrained Air-Void Systems for Durable Highway Concrete Figure 33. Correlation between SAM number and spacing factor. Figure 34. Distribution of SAM numbers with cut-off limit of 0.20 and spacing factors with cut-off limit of 0.008 in. SAM Limit Agreement in Relation to Good Air Agreement in Relation to Poor Air False Negative False Positive 0.20 7 (5%) 116 (81%) 9 (6%) 12 (8%) 0.25 11 (8%) 105 (73%) 5 (3%) 23 (16%) 0.30 14 (10%) 100 (69%) 2 (1%) 28 (19%) 0.35 14 (10%) 92 (64%) 2 (1%) 36 (25%) 0.40 16 (11%) 79 (55%) 0 (0%) 49 (34%) Table 23. Comparison of agreement between SAM number cut-off limits and spacing factor limit of 0.008.

Experimental Program 49 Shifting the limit for a good SAM number from 0.2 to a higher value (e.g., 0.40) lowers the probability of obtaining false negatives from the SAM test, while increasing the probability of observing false positives. Finding an optimal cut-off limit seems necessary to balance the two types of error. The trends of variation in probability of false negative and false positive errors as a function of the SAM number are presented in Figure 35. The point of intersection of the two lines could be considered an optimum SAM number. The effect of concrete mixture properties on the correlations was also assessed; no significant difference was observed for the two groups of mixtures made with natural or manufactured sand. In general, results obtained for LA and TIL cements indicated linear correlations between the SAM number and spacing factor measurements. The amount of scatter in spacing factor data was lower when TIL and LA cements were used. Data obtained for HA cement indicated no such correlation and scatter in spacing factor data. The type of SCM had a marked effect on the relationships. The best correlations were observed for the mixtures prepared with lower-LOI fly ash (FA 5), followed by higher-LOI fly ash (FA 4) and natural pozzolan. Hardened Data from Flatbed Scanner The hardened concrete specimens used for measuring the properties of the hardened air- void system with the fixed-focus optical microscope were also tested using the flatbed scanner method to investigate both the total air content and spacing factor. Optimization of the scanning thresholds for air content and void frequency was necessary to ensure accuracy of the scanner data; the results are discussed in Appendix D. Hardened Air Content and Spacing Factor Measurements Using Flatbed Scanner All mixtures were analyzed using the flatbed scanner; one sample was tested for each mixture. Afterward, the same samples were also analyzed using the fixed-focus optical microscope. Figure 35. Probability of false negative and false positive observations as a function of SAM cut-off limit.

50 Entrained Air-Void Systems for Durable Highway Concrete Air Content. In general, good agreement was observed between the hardened air content measurements obtained from the scanner and the automatic microscope test methods, as shown in Figure 36. Spacing Factor. Similar results and a reasonable linear correlation between the spacing factor data obtained using the two hardened concrete methods were observed, as shown in Figure 37. The highlighted areas suggest agreement between the spacing factor measurements; the area on the left suggests a good spacing factor with values lower than 0.008 in., and the areas on the right correspond to spacing factors higher than 0.008 in. Clustering of Air Voids Cement type, coarse aggregate type, quality of fly ash, stability of AEA, temperature of the mixing water, and retempering were monitored to investigate their impact on clustering. All mixtures produced for clustering analysis were examined for clustering rating before and after retempering. The mixtures were also investigated for air content in the fresh state, as well as compressive strength at 7 and 28 days before and after retempering; results are detailed in Appendix E. Figure 38 shows the clustering ratings observed for the mixtures prepared with LA cement and coarse limestone aggregate. The results indicated little influence of w/cm on the clustering rate. However, the following factors increased the clustering rate: • Use of HA cement and, to a lesser extent, TIL cement; • Retempering the mixtures regardless of the fly ash quality, AEA stability, or water temperature; Figure 36. Comparison between the hardened air content measured by flatbed scanner and fixed-focus optical microscope.

Experimental Program 51 Figure 37. Comparison between the spacing factor values measured by flatbed scanner and fixed-focus optical microscope. Note: LA = low-alkali, FA 4 = higher-LOI fly ash, FA 5 = lower-LOI fly ash, S = stable AEA, U = unstable AEA, 70 and 90 = temperature of mixing water in °F. Figure 38. Clustering rates for mixtures prepared with low-alkali cement and coarse limestone aggregate.

52 Entrained Air-Void Systems for Durable Highway Concrete • Increase in temperature of the mixing water from 70°F to 90°F regardless of the fly ash quality, admixture stability, or retempering; • Use of an unstable AEA; and • Use of limestone aggregate. The changes in compressive strength values were compared with changes in clustering ratings obtained for concrete mixtures prepared with the same cementitious materials, aggregates, and admixtures, but with different temperatures of the mixing water. Data were grouped into before and after retempering categories. Figures 39 and 40 present the effect of clustering on the change in compressive strength of the mixtures prepared with coarse limestone aggregate before and after retempering, respectively. The figures show that an increase in the clustering rate corresponds to a reduction in compres- sive strength values. With average before-retempering compressive strengths limited to 5,000 and 6,000 psi at 7 and 28 days, respectively, the data presented in these two figures indicate an up to 35% change in compressive strength due to clustering. Similar trends were observed for mixtures with coarse gravel aggregate. Correlations Between Laboratory and Field Data As observed earlier from field data, most of the cores taken from sites with good durability against F-T cycles exhibited the following hardened air-void system characteristics: • Spacing factor less than 0.014 in., • Specific surface greater than 400 in.–1, and • Hardened air content greater than 2.2%. The average standard deviations for laboratory samples were 0.0026 in., 64 in.–1, and 0.6% for spacing factor, specific surface, and hardened air content measurements, respectively. Considering these values, the following limits for air-void system properties are obtained: • Maximum spacing factor = 0.014 – 0.0026 = 0.0114 in., • Minimum specific surface = 400 + 64 = 464 in.–1, and • Minimum hardened air content = 2.2 + 0.6 = 2.8%. Figure 39. Change in compressive strength versus change in clustering rate, before retempering, for concrete made with coarse limestone aggregate.

Experimental Program 53 The relationship established in the laboratory between spacing factor and specific surface, previously discussed, is in good agreement with these observations. For a spacing factor of 0.0114 in., the obtained trend line yields a specific surface of 458 in.–1. Moreover, based on the relationship between spacing factor and hardened air content pre- sented in this chapter, a spacing factor of 0.0114 in. corresponds to an air content of 4.2%. This value is higher than the minimum of 2.8% obtained for hardened concrete exhibiting durability against F-T cycles, and is recommended as the desired minimum. Considering the established correlation between fresh and hardened air content, one would expect concrete with a fresh air content of 4.8% to be adequate for securing F-T durability. A similar approach was used to select a desired SAM number that would indicate a prop- erly hardened air-void system. Based on data collected in the laboratory, a spacing factor of 0.0114 in. corresponds to a SAM number of 0.42, and a SAM number of 0.46 correlates well with a fresh air content of 4.8%. The average standard deviation in SAM number measure- ments on 144 concrete mixtures was approximately 0.05. Considering that two standard deviations (i.e., 2 × 0.05 = 0.10) ensures a 95% confidence level, the desired SAM number will be 0.42 – 0.10 = 0.32, which is rounded down to 0.30. Summary The data obtained from testing the concrete mixtures in the fresh and hardened states suggest the following: • Good repeatability can be obtained for the SAM test as long as operators follow the test procedures and meters are properly calibrated. • Data from the SAM test can be used for analyzing the quality of the air-void system in the fresh state. SAM data were repeatable and agreement was observed between the SAM numbers and spacing factor data. • A SAM number of 0.20 corresponds well with a spacing factor of 0.008 in. or less. • The correlation between SAM number and spacing factor is sensitive to the mixture chemistry. Figure 40. Change in compressive strength versus change in clustering rate, after retempering, for concrete made with coarse limestone aggregate.

54 Entrained Air-Void Systems for Durable Highway Concrete • Strong correlations exist between the air content and spacing factor data obtained from the flatbed scanner and those from the fixed-focus optical microscope. • The type of cement is a primary factor in governing the extent of clustering for mixtures not exposed to retempering. • Retempering, increase in the temperature of the mixing water, AEA stability, quality of fly ash, and type of coarse aggregate influence the clustering rate. • Using cements with LA content and avoiding retempering reduce the risk of clustering. Using a stable AEA and lower-LOI fly ash also helps reduce the risk of clustering. • Variations in compressive strength up to a 35% change occur due to clustering. • Combining the field observations with the data obtained from fixed-focus optical microscopy indicates that concrete mixtures having a spacing factor less than 0.0114 in., specific surface greater than 464 in.–1, and hardened air content greater than 4.2% will exhibit durability against F-T cycles. • Combining the field observations with the flatbed scanner data indicates that concrete mixtures having a spacing factor less than 0.0122 in., specific surface greater than 445 in.–1, and hardened air content greater than 4.3% will exhibit durability against F-T cycles. • Given the correlations between the spacing factor, SAM number, and fresh air content, the following fresh concrete properties can be recommended for securing a proper air-void system: a SAM number less than 0.30 and fresh air content greater than 5.0%. Test Methods for Evaluating Freeze–Thaw Durability The most commonly used F-T test method in the U.S. is AASHTO T 161 Procedure “A.” The equipment for variant A of the test is relatively inexpensive (∼$22,000) and compact compared with Procedure “B” version equipment. The relative low cost stems from the basic construction of a refrigerant-based cooling system and resistive heating elements. These units can be easily repaired by trained heating and air conditioning technicians and have the potential to be economically modified using readily available controls. Consequently, all proposed factors and modifications to the test method were based on the use of this piece of equipment. A plethora of historical testing and procedure development allowed the recommendation of specific modifications, including the cooling rate, the degree of saturation, and the sample conditioning method. Procedural and functional modifications to the testing method included the following: • Procedural modifications including longer curing times to better encompass requirements for higher SCM mixtures and the introduction of a drying period to better simulate actual field exposure conditions, • Functional modifications including introducing water through capillary suction and changing the freezing solution to incorporate deicing salts, and • A combination of modification approaches including application of a concentrated pre-storm brine followed by drying. The F-T laboratory investigation was performed in three phases. The first phase involved modification of the AASHTO T 161 “A” F-T cabinet for the proposed testing arrangement. The F-T cabinet effectively has two relays that switch between heating and cooling using a temperature sensor (or a mechanical temperature switch on the older equipment). A micro- computer was programmed to control low-voltage heating and cooling relays. A description of the conversion process is provided in Appendix F. The second phase involved evaluating three separate variants of the proposed test not encompassed by current test methods, including the impact of curing time (CDF-A), deicer type (CDF-B), and deicer pretreatment (CDF-C). The third phase involved evaluating a range

Experimental Program 55 of concrete samples with varying levels and qualities of air systems as determined in this project for selected test parameters. The test plan was designed to explore F-T testing conditions and determine air-void parameter aspects that define F-T performance. Freeze–Thaw Test Mixtures and Methods In order to evaluate F-T test methods, six different concrete mixtures were selected for testing with one mixture (Mixture 29) retempered three times to induce air-void clustering around the coarse aggregate. Table 24 shows the specific parameters for each mixture. Mixture designa- tions include the number and three letters that describe the air system. For example, Mixture 2 (A, S, N) has an acceptable air volume, stable air system, and no SCM influence. Concrete mixture proportions and attributes are detailed in Appendix A. The objective was to encompass the range of air-void systems that may be observed during field construction. Mixture 2 was the control evaluated through all F-T variants (refer to Table 7). All other mixtures were evaluated using AASHTO T 161 “A” along with variants CDF-A:FT1, CDF-B:FT4, and CDF-C:FT5, which included water or NaCl as the deicer solution. Mixture 29 represented a mixture with low air, an unstable air system, and significant SCM influence. The other five mixtures had varying air system performance design to quantify which aspect of a “marginal” air system controlled F-T durability. Table 25 presents the air system values used for classification purposes of good, marginal, and poor parameters. The complete air-void analysis for all mixtures was presented previously in this report (see Appendix C for further details). A summary of the relevant air-void parameters for F-T performance are listed in Table 26. Mixture Designation Air Volume Air System SCM Influence Other 2 (A, S, N) Acceptable Stable None – 14 (A, NS, N) Acceptable Not Stable None – 1 (M, S, N) Marginal Stable None – 13 (M, NS, N) Marginal Not Stable None – 17 (M, S, S) Marginal Stable Significant – 29 (M, NS, S) Marginal Not Stable Significant – 29 Retempered Acceptable – – Clustering Note: A = acceptable air content in fresh state, S = stable AEA, N = no significant effect due to SCM use, M = marginal air content in fresh state, NS = unstable AEA. Table 24. Concrete mixture parameters considered for freeze–thaw testing. Characteristic Parameter Rating Good Marginal Poor Fresh Air Volume (%) > 5% 4%–5% < 4% SAM Number < 0.30 0.30–0.40 > 0.40 ASTM C457 Total Air Volume (%) > 4.5% 3.5%–4.5% < 3.5% ASTM C457 Spacing Factor < 0.0114 0.0114–0.0140 > 0.0140 Note: Values of fresh air content are based on ACI 201 (2016) recommendations for mixtures made with crushed aggregates. For mixtures prepared with rounded aggregates, the air content can be lowered, typically by 1%. Table 25. Air system values used for rating expected freeze–thaw performance.

56 Entrained Air-Void Systems for Durable Highway Concrete In only two cases (Mixtures 14 and 29 retempered), the SAM number predicted good F-T performance. Results A summary of findings obtained from testing hardened concrete samples is presented below; detailed results are provided in Appendix G. AASHTO T 161 “A” Figure 41 shows the control test results across all concrete mixtures. The control testing conditions using AASHTO T 161 “A” indicated acceptable (> 60% RDM) performance to 300 cycles for all mixtures with expected good-to-marginal performance (2, 14, 17, and 29 retempered [29RT]). Mixture 17 with expected marginal-to-poor performance Mixture Designation Fresh Hardened Anticipated Performance Air (%) SAM Air (%) S.F. (in.) 2 7.1 0.44 5.8 0.009 Good–Marginal 14 5.8 0.39 5.0 0.007 Good 1 3.2 0.69 2.3 0.022 Poor 13 3.1 0.70 2.6 0.019 Poor 17 4.0 0.74 3.9 0.017 Marginal–Poor 29 3.8 0.80 2.5 0.022 Poor 29 Retempered 6.7 0.31 5.1 0.010 Good Note: S.F. = spacing factor. Table 26. Air testing results and anticipated freeze–thaw performance. Note: A = acceptable air content in fresh state, S = stable AEA, N = no significant effect due to SCM use, M = marginal air content in fresh state, NS = unstable AEA. 60% 65% 70% 75% 80% 85% 90% 95% 100% 0 50 100 150 200 250 300 R el at iv e D yn am ic M od ul us ( % ) Cycles 2 (A, S, N) 14 (A, NS, N) 1 (M, S, N) 13 (M, NS, N) 17 (M, S, S) 29 (M, NS, S) 29RT (A, NS, S) Figure 41. Relative dynamic modulus versus freeze–thaw cycles for AASHTO T 161 “A”.

Experimental Program 57 achieved a durability factor > 90% until 240 cycles. In general, the mixtures performed as predicted or better, except for Mixture 29R, which performed worse than predicted probably due to the form of the air-void system induced by the retempering. Mass loss and RDM did not correlate well. CDF-A: Influence of Curing Time The CDF-A:FT1 testing variant included concrete cured under plastic and burlap for the initial 24 hours. Following initial curing, samples were demolded and cured in limewater until day 7, after which, samples were dried for 21 days, and then conditioned in limewater for an additional 7 days before starting F-T cycling in deionized water for 140 cycles or 60% RDM. Figure 42 shows the results. Additional variants of CDF-A included sample curing in limewater to 28 and 56 days for the control Mixture 2. Average RDM is shown in Figure 43 for Mixture 2, which shows that duration of curing did not affect sample performance. CDF-B: Influence of Deicer Solution The testing variant CDF-B involved the same curing and conditioning procedure as CDF-A, with freezing performed in 3% NaCl. F-T performance was similar to that of testing performed in deionized water, with all samples completing 56 cycles with RDM values above 90% except for Mixture 1 (see Figure 44). CDF-C: Brine Pretreatment F-T testing variation CDF-C involved conditioning the dried specimens in high concentration brine solutions to simulate brine pretreatment on pavement before a freezing event. CDF-C conditioned the dried samples in 35% NaCl for 7 days, then re-dried them for 21 days before finally resaturating them in deionized water before starting F-T cycling in deionized water. Figure 45 shows that F-T performance was similar to that observed with both water (CDF-A) and deicer solution (CDF-B). Figure 42. Relative dynamic modulus versus freeze–thaw cycles for CDF-A:FT1. Note: A = acceptable air content in fresh state, S = stable AEA, N = no significant effect due to SCM use, M = marginal air content in fresh state, NS = unstable AEA. 60% 65% 70% 75% 80% 85% 90% 95% 100% 105% 0 10 20 30 40 50 R el at iv e D yn am ic M od ul us (% ) Cycles 2 (A, S, N) 14 (A, NS, N) 1 (M, S, N) 13 (M, NS, N) 17 (M, S, S) 29 (M, NS, S) 29RT (A, NS, S)

58 Entrained Air-Void Systems for Durable Highway Concrete 60% 65% 70% 75% 80% 85% 90% 95% 100% 105% 0 10 20 30 40 50 R el at iv e D yn am ic M od ul us (% ) Cycles Mix 2—FT 1 (7d) Mix 2—FT2 (28d) Mix 2—FT3 (56d) Figure 43. CDF-A testing at various curing ages (7, 28, and 56 days). 60% 65% 70% 75% 80% 85% 90% 95% 100% 105% 0 10 20 30 40 50 R el at iv e D yn am ic M od ul us (% ) Cycles 2 (A, S, N) 14 (A, NS, N) 1 (M, S, N) 13 (M, NS, N) 17 (M, S, S) 29 (M, NS, S) 29RT (A, NS, S) Figure 44. Relative dynamic modulus versus freeze–thaw cycles for CDF-B:FT4. All mixtures, except for Mixture 1, had good performance through 56 cycles with RDM values above 90%. More scaling was observed for most of the samples compared with freezing in deionized water. The 35% NaCl solution had a decrease in RDM values and mass loss resulting from scaling. Performance Across Testing Solutions Figure 46 through Figure 48 show the performance of the control mixture and Mixture 2 for NaCl, CaCl2, and MgCl2 treatments, respectively. F-T performance in all conditions was good, with RDM values above 90%. ASTM C672: Deicer Scaling The visual deicer scaling results are shown in Figure 49. Performance ranged from a rating of 2.0 to 4.3. The control mixture, Mixture 2, and Mixture 13 had the best performance.

Experimental Program 59 60% 65% 70% 75% 80% 85% 90% 95% 100% 105% 0 10 20 30 40 50 R el at iv e D yn am ic M od ul us (% ) Cycles 2 (A, S, N) 14 (A, NS, N) 1 (M, S, N) 13 (M, NS, N) 17 (M, S, S) 29 (M, NS, S) 29RT (A, NS, S) Figure 45. Relative dynamic modulus versus freeze–thaw cycles for CDF-C:FT7. 60% 65% 70% 75% 80% 85% 90% 95% 100% 105% 0 10 20 30 40 50 R el at iv e D yn am ic M od ul us (% ) Cycles Mix 2—CDF A:FT 1 (water) Mix 2—CDF B:FT4 (3% NaCl freezing solution) Mix 2—CDF C:FT7 (35% NaCl brine pretreatment) Figure 46. Relative dynamic modulus versus freeze–thaw cycles for Mixture 2 across NaCl treatments. Mixture 14 possessed acceptable air-void volume but contained non-stable AEA and had the worst performance. Summary and Conclusions from Freeze–Thaw Testing This work was performed to identify air-void characteristics that are most important for F-T durability and to evaluate new and better techniques for both air-void system and dura- bility determination. The standard F-T chest used to evaluate AASHTO T 161 “A” was modified functionally and procedurally to incorporate components of ASTM and RILEM F-T and deicer scaling testing. Five different F-T techniques were evaluated across seven different concrete mixtures. The concrete mixtures represented good, poor, and various marginal air-void systems.

60 Entrained Air-Void Systems for Durable Highway Concrete The F-T techniques altered freezing fluid, method of fluid exposure, timing of F-T cycles, and conditioning techniques. Table 27 shows a summary of the F-T results. The proposed 56-cycle evaluations were not sufficient to elucidate any performance differ- ences for the control mixture, which possessed a borderline good air-void system. All F-T test variants produced durability factors above 97% on the control mixture. Of the five F-T test methods/variants evaluated across all of the concrete mixtures, AASHTO T 161 “A” performance agreed with the anticipated performance predicted from the air system 60% 65% 70% 75% 80% 85% 90% 95% 100% 105% 0 10 20 30 40 50 R el at iv e D yn am ic M od ul us (% ) Cycles Mix 2—CDF A:FT 1 (water) Mix 2—CDF B:FT5 (3% CaCl2 freezing solution) Mix 2—CDF C:FT8 (74% CaCl2 brine pretreatment) Figure 47. Relative dynamic modulus versus freeze–thaw cycles for Mixture 2 across CaCl2 treatments. 60% 65% 70% 75% 80% 85% 90% 95% 100% 105% 0 10 20 30 40 50 R el at iv e D yn am ic M od ul us (% ) Cycles Mix 2—CDF A:FT 1 (water) Mix 2—CDF B:FT6 (3% MgCl2 freezing solution) Mix 2—CDF C:FT9 (54% MgCl2 brine pretreatment) Figure 48. Relative dynamic modulus versus freeze–thaw cycles for Mixture 2 across MgCl2 treatments.

Experimental Program 61 Note: A = acceptable air content in fresh state, S = stable AEA, N = no significant effect due to SCM use, M = marginal air content in fresh state, NS = unstable AEA. 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 0 5 10 15 20 25 30 35 40 45 50 Sc al in g Se ve ri ty R at in g Cycles Mix 2 (A, S, N) Mix 14 (A, NS, N) Mix 1 (M, S, N) Mix 13 (M, NS, N) Mix 17 (M, S, S) Mix 29 (M, NS, S) Mix 29RT (A, NS, S) Figure 49. Scaling versus deicer cycles for ASTM C672. parameters for six of the seven concrete mixtures. The modified CDF variants agreed for four of the seven concrete mixtures, and ASTM C672 deicer scaling agreed for four. Although often considered an aggregate quality test, AASHTO T 161 “A” did provide F-T evaluations that correlated well with concrete air system parameters. After the remaining testing variants had been completed, all samples representing the 7-day cure tested in deionized water (CDF-A:FT1) were tested until 140 cycles. Table 28 shows the durability factor calculated at either 300 or 140 cycles for the respective test. Mixture AnticipatedPerformance AASHTO T 161 “A” CDF-A: FT1 CDF-A: FT2 CDF-A: FT3 CDF-B: FT4 CDF-B: FT5 CDF-B: FT6 CDF-C: FT7 CDF-C: FT8 CDF-C: FT9 ASTM C672 2 Good–Marginal 99 96 97 98 97 98 99 93 99 98 2.3 14 Good 100 100 — — 100 — — 100 — — 4.3 1 Poor 53 40 — — 29 — — 53 — — 3.3 13 Poor 100 100 — — 99 — — 100 — — 3.7 17 Marginal–Poor 71 98 — — 99 — — 99 — — 2.7 29 Poor 38 97 — — 98 — — 98 — — 2.0 29RT Good 77 100 — — 95 — — 100 — — 3.7 Table 27. Summary of air-void system and freeze–thaw results and durability factor ratings. Mixture Designation Anticipated Performance Durability Factors AASHTO T 161 “A” (300 cycles) CDF-A:FT1 (140 cycles) 2 Good–Marginal 99 93 14 Good 100 100 1 Poor 53 16 13 Poor 100 98 17 Marginal–Poor 71 68 29 Poor 38 43 29 Retempered Good 77 63 Table 28. AASHTO T 161 “A” and CDF-A:FT1 results.

62 Entrained Air-Void Systems for Durable Highway Concrete The extended CDF-A:FT1 test and AASHTO T 161 “A” yielded comparable performance predictions, matching anticipated performance from air system parameters. Observations showed that retempering improved the air system for Mixture 29, and measured air system parameters did not adequately predict performance, as good performance was anticipated and marginal performance was observed. Although Mixture 1 and Mixture 13 had very similar air system properties, Mixture 13 had good performance and Mixture 1 had poor performance in all tests. Table 29 shows a comparison of visual ratings for the deicer scaling and F-T test methods. Scaling performance was highly variable and did not correlate with F-T test results obtained from the RDM at the number of cycles shown. The primary purpose of this study was to define and evaluate components of the concrete air-void system that greatly influenced durability. For six of the seven concrete mixtures evaluated for F-T, the fresh air-void volume, hardened air-void volume, and spacing factor parameters provided a good indication of laboratory performance. However, from an air system perspective, while Mixture 1 and Mixture 13 were nearly identical—both had low air and high spacing factors and should have had poor F-T durability—Mixture 1 exhibited poor F-T durability in all cabinet-based tests, but Mixture 13 had excellent performance in all tests (see Table 29). Mixture Designation Visual Rating ASTM C672 CDF-A:FT1 CDF-B:FT4 CDF-C:FT7 2 2.3 2 2 2 14 4.3 5 1 1 1 3.3 4 4–5 5 13 3.7 1–2 2 2 17 2.7 1 2 4 29 2.0 4 1 2 29 Retempered 3.7 3 5 3 Note: Shading represents mixtures with nearly identical air-void volumes and spacing factors but differing test results. Table 29. Visual rating for different test methods.