Use of Marginal and Unconventional-Source Coal Ashes in Concrete (2024)

CHAPTER 2

Research Approach

Materials

Coal Ash Samples

The research team collected and tested 22 coal ashes (16 Class F ashes and six Class C ashes) to represent a broad distribution of ash sources and properties. The ashes were separated into two groups: standard and unconventional. The standard ashes are those that meet the AASHTO M 295-21 specification; the unconventional ashes are either harvested, out of scope, off-specification, or have been beneficiated. Beneficiated ashes were collected from all known commercial technologies currently available in North America. Eight ashes are from standard sources with in-specification bulk composition properties. The eight beneficiated ashes consist of ashes processed with a wide range of commercially available beneficiation technologies used by the industry (blending, drying, sieving/grinding, heat treatment, electroprocessing, and chemical treatment). The beneficiated ashes as a whole exhibit in-specification bulk composition properties but with some marginal and off-specification properties. The remaining seven unconventional ashes originated from a variety of alternative sources, including circulating fluidized bed (CFB) ash, cyclone collector ash, pond-impounded ash, and a bottom ash blend, and some are characterized with off-specification bulk composition properties. The coal ashes are listed in Table 5 and categorized by source.

Cement

The properties of the four cements used for testing are shown in Table 6. The values were provided by the cement manufacturer.

Characterization Methods

X-Ray Fluorescence

Chemical compositions of the ashes were determined using an x-ray fluorescence (XRF) spectrometer with borate fusion sample preparation (Barger 1985, ASTM C114-18). The samples were initially ground and mixed with a lithium borate flux in a crucible heated to a temperature exceeding 1,000^oC, and fused beads were subsequently analyzed by XRF for elements expressed as oxides (SiO₂, Al₂O₃, Fe₂O₃, CaO, MgO, K₂O, Na₂O, and SO₃). Alkali contents were also expressed as total alkali equivalent (Na₂O_eq).

LOI and Moisture Content

Moisture content and LOI were determined on 30 to 50 mg of powder using TGA by placing the material on a platinum pan and heating at 10°C per minute to 1,000°C in a nitrogen atmosphere

Table 5. Coal ash samples categorized as standard and unconventional.

per ASTM C311. Using 2–4 g of materials, a sample was weighed and dried to constant mass at 110°C to determine the moisture content. LOI was determined in two separate measurements, by heating this sample to either 750°C or 950°C. The different methods of measuring LOI were done to evaluate the effect of temperature and measurement type on the measured value to see if any of these resulted in values that better correlated to other adsorption measurements.

X-Ray Diffraction Quantitative Analysis

X-ray diffraction (XRD) scans were performed with a Bruker D8 XRD in flat plate reflection Johannsson mode. The diffractometer was operated at 40 kV and 40 mA and scans were run from 5° to 60° of 2θ with a step size of 0.02° using a CuKα x-ray source. Crystalline phase content was determined using Rietveld refinement performed by Profex software.

Particle Measurements

The particle size distributions (PSDs) of the investigated ashes were determined using Sympatic MYTOS equipment with dry-dispersion and laser diffraction measurement. A mean refractive index (RI) value of 1.55 (similar to the consistency of clay) was used for all samples. Each specimen was measured with five replicates and the average value was reported. The median particle size d₅₀,

Table 6. Cement properties.

the 45 µm retained percentage, and specific surface area (SSA) were extracted from these measurements. The percentage retained when wet-sieved on a 45 µm (no. 325) sieve was found using the procedure described in ASTM C311 and provided by the coal ash supplier. The percentage retained when wet-sieved on a 150 µm (no. 100) sieve was found using the procedure described in ASTM C136. The Blaine fineness tests followed ASTM C604 and were carried out on 2.5 g of coal ash powder tested in triplicate. Following ASTM C604, specific gravity was quantified from the volume of a known mass of powdered sample passing a 45-µm sieve determined using a helium pycnometer. Particle morphology was investigated using scanning electron microscopy (SEM) performed on a small subgram sample which was stuck on carbon tape and gold-coated. Imaging was done on settings of 5 kV voltage, 9 to 14 mm working distance, with 640 × 440 µm field of view.

Reactivity and Paste Testing Methods

R³ and Modified R³ Method

The R³ test for heat of hydration was carried out according to ASTM C1897-Method A, except for the mixing, which did not use high shear. Equipment and materials were preconditioned at 40°C ± 2°C in an oven. The dry ingredients were mixed SCM and calcium hydroxide (1:3 by mass) and SCM and calcium carbonate (2:1 by mass). The mixing solution was prepared by dissolving 4 g KOH and 20 g K₂SO₄ in 1 L distilled water. Powders were dry-mixed for 4 minutes by hand and the mixing solution was added at a liquid-to-solid ratio of 1.2 and hand-mixed for 4 more minutes. For the modified R³ test, the SCM to calcium hydroxide remained 1:3, but the solution was a 0.5 M KOH solution used at a liquid-to-solid ratio of 0.9. Post mixing, 6–7 g of mixture were inserted into a glass ampoule, sealed, and placed in the isothermal calorimeter preconditioned at 40°C ± 0.05°C for the R³ test and 50°C ± 0.05°C for the modified R³ test. Heat release was measured for 7 days and 10 days for the R³ test and modified R³ test, respectively, and later, TGA was carried out on 30–50 mg of material obtained from the paste bulk. TGA was tested as described in the Characterization Methods section, but run to 600°C. The bound water was calculated based on the TGA mass loss in the range of 150–350°C and the calcium hydroxide consumption was calculated using the mass loss in the 380–460°C range estimated using the tangential method and the initial calcium hydroxide content. Both were expressed as a function of paste mass before the start of the test. The bound water measurement was carried out differently than suggested in ASTM C1897 in order to obtain the entire mass loss curve, which allows the calculation of the calcium hydroxide consumption using just the TGA test.

Paste Measurements

Coal ashes and inert fillers replaced 30% of the cement by mass to enhance SCM reactivity effects. A w/cm of 0.40 was used for the mixture. For isothermal calorimetry, pastes were mixed by hand as detailed in the R³ and Modified R³ Method section. For hardened paste tests, pastes were mixed in a mixer following ASTM C305 and three 50-mm (2-in.) cubes were cast for each paste, covered and cured for 1 day in the molds, and then demolded and immersed in saturated lime water at 23°C ± 2°C until testing.

Paste Testing

Paste isothermal calorimetry was performed as detailed in the R³ and Modified R³ Method section, except that it was done at 23°C ± 0.05°C for 7 days. Paste bulk resistivity was measured at 7 and 91 days following ASTM C1876 on specimens cured in limewater with surface-dry specimens and corrected for geometry but not for degree of saturation. After bulk resistivity, compressive strength testing was performed at 7 and 91 days following ASTM C109 on the

pastes. TGA was performed on pieces from the core, crushed with a mortar and pestle to pass through a 75 µm sieve.

Modifications of SAI and Development of BRI

A subset of the tested materials—one Class C ash, two Class F ashes, and two inert fillers—was chosen specifically for this test using the same cement (Y. Wang, Burris, et al. 2021). Mortars were prepared by replacing 30% and 50% by mass of the cement with coal ash or inert materials. A w/cm of 0.40 and a sand-to-cementitious ratio of 2.75 were used for the mixtures. Mortars were cast into 50-mm (2-in.) cubes following ASTM C305 and cured in limewater at 23°C. For 50°C curing, mortars were demolded and stored in saturated limewater, sealed, and placed in an oven at 50°C ± 2°C. Strength was determined at 7 and 28 days, following ASTM C109. Bulk resistivity was determined at 7 and 28 days in surface-dry conditions at 1 kHz, following ASTM C1876. Pastes were hand-mixed as detailed in the R³ and Modified R³ Method section and subjected to isothermal calorimetry at 23°C ± 0.05°C or 50°C ± 0.05°C. On these pastes cured at either 23°C ± 1°C or 50°C ± 2°C temperature regime, calcium hydroxide was also determined at 7- and 28-day ages.

Mortar Testing Methods

Strength Activity Index and Modified Strength Activity Index

The SAI was performed following the procedures in ASTM C311 and ASTM C109 at 7, 28, and 91 days. A 20%-by-weight replacement rate of cement with coal ash was used. For SAI, the water requirement measurement was also determined for each of the ashes using the baseline flow of the control mixture to be within ± 5%. For the MSAI, the same procedures were followed and a 20%-by-mass replacement rate of cement with coal ash was also used. However, in the MSAI scheme, flow values were recorded for each of the ashes at a constant w/cm of 0.485 rather than constant flow. The baseline mixture proportions for all the tests used in this section are shown in Table 7.

Keil Hydraulic Index

The KHI test using inert filler materials (Keil 1952, Lea 1970, Pal et al. 2003, Sutter et al. 2013) was also performed. The KHI value, which assesses reactivity of the coal ash with respect to the datum of the control mixture and an additional inert material mortar, was also computed in triplicate at 7, 28, and 91 days. In this investigation, inert filler materials included powdered limestone, crystalline quartz, and basalt of similar particle sizes to the ashes. The KHI sample was proportioned similarly to the SAI control mixture, with an inert filler material at 20% cement-replacement level. A set of nine 50 mm × 50 mm (2 in. × 2 in.) cubes were made for each of the filler materials and their compressive strengths were determined in triplicate at 7-, 28-, and 91-day ages. The strength data were then used to compute KHI values for each filler material at the stated ages to

Table 7. Mortar mixture proportions (SAI, MSAI, TE, and KHI tests).

differentiate their contributions to strength in comparison to the coal ashes. KHI values (expressed in percentage) were computed using Equation 1.

where:

KHI = Keil hydraulic index

A = average strength of test mixture (SAI) at time t (kPa or psi)

B = average strength of the “control” mixture at time t (kPa or psi)

C = average strength of inert filler-cement mixture at time t (kPa or psi)

Total Efficiency Index

The TE test involves making an extra set of mortar cubes with a higher cementitious content and lower sand content to accompany the SAI and control mortar cubes (ASTM C311 Appendix Addition, Dunstan 2017, Dunstan 2019). The extra set of cubes consisted of an additional six 50 mm × 50 mm (2 × 2 in.) cubes to be tested at 7 and 28 days for compressive strengths. At 7 and 28 days, the compressive strength data of all three sets were then used to compute the water reduction and chemical reaction contributions to strength for each of the coal ashes or candidate SCMs used. In this investigation, only a few selected ashes and inert filler materials—including A, D, R, limestone, quartz, and basalt—were used to make the extra set of six cubes. Coal ashes A, D, and R were chosen to represent a standard Class F, standard Class C, and beneficiated ash, respectively. Components of CE and water-reduction efficiency (WRE), as well as the total strength efficiency (TSE), were determined in triplicate using Equations 2–4.

where:

TSE = total strength efficiency (%)

A = average strength of test mixture (SAI) at time t (kPa or psi)

B = average strength of the control mixture at time t (kPa or psi)

X = average strength of the strength efficiency sample with more cementitious content at time t (kPa or psi)

W_A = water content of the mortar mixture (SAI) expressed as % of the control

W_X = water content of strength efficiency mortar sample expressed as % of the control

CE = chemical efficiency (%)

WRE = water-reduction efficiency (%)

Bulk Resistivity

Bulk uniaxial resistivity (BR) measurements were made via the ProCEQ Resipod during the SAI, MSAI, and KHI tests at 3, 7, 28, 56, and 91 days to track reactivity and microstructural evolution of the ashes in the cement-mortar system.

Concrete Testing Methods

Concrete testing included measurements of various fresh and hardened properties ranging from slump, density, air content, slump loss to compressive strength, bulk resistivity, surface resistivity (SR), and chloride permeability properties.

Mixture Design

For concrete testing, a set of approximately 18 cylinders 10 cm diameter × 20 cm height (4 in. diameter × 8 in. height) were made for each ash using the general mixture design shown in Table 8. Concrete mixture designs adjusted for coal ash density are shown in Table 9. In all mixtures, air content was set to 2% of the total volume of 27 ft³. Coarse aggregate was ASTM C33 Size 57 (1 in.) limestone, and the fine aggregate was ASTM C33-compliant river sand. A moisture content test was performed for representative samples of both the coarse and fine aggregates in accordance with ASTM C566 (Total Evaporable Moisture Content of Aggregate by Drying) by placing the samples in a ventilated oven maintained at 100°C for 24 hours. Once the moisture contents for the coarse and fine aggregates were determined, the mixing water and the aggregates were readjusted per the National Precast Concrete Association (NPCA) method for aggregates in SSD conditions.

Table 8. Concrete baseline mixture prior to aggregate moisture content correction.

Table 9. Concrete mixture designs adjusted for coal ash density.

Fresh Property Measurements

Fresh properties including slump (ASTM C143/AASHTO T119), density or unit weight (ASTM C138/AASHTO T121), and air content (ASTM C231/AASHTO T152) were measured immediately after initial mixing. The targeted slumps at t = 0 seconds was set to a preferred range of 5–13 cm (2–5 in.) without consideration for the addition of a superplasticizer unless extremely low workability occurred. Air content correction was made by subtracting the aggregate correction factor G, which averaged 0.27% per ASTM C231 test on five replicates, from the measured air content. A total of 18 concrete cylinders was also molded immediately. Glenium 300 superplasticizer was added to the mixtures of coal ashes C, S, and U in increments of 20 mL until adequate workability level permitted ease of molding the concrete cylinders. After the samples were molded and placed in the curing chamber, an amount of wet concrete approximately equal to the volume of the cone frustrum was left in the mixer for slump loss measurement at 90-minute age. For the 90-minute slump measurement, the mixing procedure followed a scheme of 1 minute of mixing and 5 minutes of resting with the cover on until the 90-minute mark was reached. Then, a final mixing of 2 minutes was allowed before the 90-minute slump test was initiated.

Hardened Property Measurements

Hardened concrete properties were determined in triplicate at 1-, 7-, 28-, 91-, and 180-day ages. After demolding, concrete specimens were cured in a lime water bath at 23°C ± 2°C. Hardened properties including compressive strength using a Tinius Olsen universal testing machine (ASTM C39), BRI (AASHTO TP-119-15), SR (AASHTO TP 358), and various properties from the RCPT (ASTM C1202) were determined at the specified testing ages. Both BRI and SR measurements were taken in triplicate at 1-, 7-, 28-, 56-, 91-, and 180-day ages using the Wenner Probe Array (ProceQ Resipod). In the RCPT, electrical charge passage measurement was determined in duplicate at 28 and 91 days on 101.6 mm diameter × 50.8 mm height (4 in. diameter × 2 in. height) concrete disks trimmed from the concrete cylinder. Additionally, chloride penetration depth measurements expressed in millimeters were also taken in duplicate at 28 and 91 days for the concrete disks using the Nordtest method of NT Build 492, where the concrete disks were axially cleaved, sprayed with silver nitrate, and measured every 10 mm across the section until seven depth values had been obtained.

Sulfate Attack Measurements

SA testing using ASTM C1012 involved immersing coal ash mortar bars in sodium sulfate solution and measuring their expansions over time. Eight bars with dimensions of 25.4 mm × 25.4 mm × 285.75 mm (or 1 in. × 1 in. × 11.25 in.) were made for each of the investigated ashes at a 30% cement-replacement level. The mixture design for SA testing is shown in Table 10. Per the ASTM C1012 test method, once a compressive strength of at least 19.65 MPa (2,850 psi) was reached, the bars were immersed in sodium sulfate solution at a solution-to-total bar volume

Table 10. Mixture proportions for SA mortar bar test (per 8-bar set).

ratio of 4:1. Expansion measurements were then measured periodically at 1, 2, 3, 4, 8, 13, 15, 16, 24, 36, 48, 60, and 72 weeks. Fresh solution was prepared to replace the old solution after every expansion measurement. Expansion measurement due to SA was discontinued once warping of a single mortar bar was observed in the set. SA mitigation performance for the investigated ashes was then classified with respect to the 6-month (or 24-week) expansion limit and performance metrics given in ASTM C595 and ACI 318-19.

Adsorption Testing Methods

Five AEAs were used throughout this project and are shown in Table 11. Testing predominantly focused on use of the sodium lauryl sulfate (SLS) and vinsol resin (VR) types of admixtures, which are two commonly used AEA types in the United States. These should provide view of the variances in test results when using different types of admixtures with the various coal ash types.

Foam Index Test

The FIT was conducted according to ASTM C1827-20 except in instances where changes are explicitly mentioned, such as when using 50% SCM or 100% SCM instead of the standard 20% called for in ASTM C1827-20. The dilution percentage of AEA-water mixtures varied between 2.5% and 7.5% to better differentiate endpoints for ashes showing low or high adsorption capacity, respectively. The endpoint determination was normalized for all the ashes by accounting for the drop volume of the diluted AEA solution and the dilution percentage in accordance with ASTM C1827-20.

A variation of typical FIT testing was used to evaluate the effect of pore solution chemistry on removal of AEA from solution and impact on foaming. In this set of tests, the ASTM C1827-20 FIT procedure was followed, but cement and coal ash quantities were varied, with some samples omitting solids and some doubling the total quantity of solids in the mixture. Solution chemistry of these samples was varied using additions of Ca(OH)₂, Ca(NO₃)₂, KOH, K₂SO₄, NaCl, Na₂SO₄, Al(OH)₃, or K₂SiO₃, all American Chemical Society-grade. AEA was added dropwise to the solutions, shaken, and the foam examined, following the same procedure as in FIT testing. In some tests the quantity of AEA remaining in solution was measured. Filtrates of the solution were collected after mixing using 0.22 µm syringe filters and AEA remaining in solution was measured using a total organic carbon analyzer (TOC, Shimadzu) and a calibration curve developed by varying the concentrations of AEA dosage in deionized water (DI) solutions.

Iodine Number Test

The single-point iodine method was conducted to measure the adsorption of iodine onto coal ash using an iodine solution. The method quantifies iodine remaining in solution using a titration

Table 11. AEA types used in project testing.

method. 40 ± 0.01 g of coal ash was placed into a 100 g/L nitric acid solution and boiled on a magnetic stirring hot plate for 5 minutes. After cooling, the solution was filtered using Grade 54 90-mm-diameter filter paper with a 100 mL Buchner funnel. The solids were dried in an oven overnight at 105°C, recording mass after drying. The dried sample was redispersed in 100 mL of 0.025N iodine with 12 drops of starch solution. After stirring for 5 minutes, the sample rested for 15 minutes; 25 mL of the aliquot was retrieved from the sample and titrated using 0.025N sodium thiosulfate. The volume of titrant used was compared to a blank sample without coal ash. The iodine number (g of iodine/kg of coal ash) was calculated using Equation 5:

where:

V_F = volume of standard thiosulfate solution required for titration

V_B = volume of titrant required for titration of the blank solution

W_F = original mass of the sample, before boiling and acidification

Mortar Air Test

A testing method based on ASTM C185 was used to determine the air content of mortars relative to varying dosages of AEA. Cement-to-sand ratios were altered from the ASTM C185 method’s 1:4 cement to sand to 1:2.85 to reduce the water content required to achieve the specification’s mixture flow requirements of 87.5% ± 7.5% increased flow relative to the control OPC mortar. Six separate mortar mixtures were created using each coal ash to obtain a dosage curve. AEA was added to the mixing water prior to initial mixing to promote uniform distribution of the surfactant throughout the mixture and the mortar was mixed according to ASTM C305. AEA demand was calculated as the AEA dosage required (mL/g of coal ash) to obtain 18% air. To assess air system stability, after the dosage providing ∼18% air had been identified, the bowl was covered with a damp towel to prevent evaporation and allowed to rest for 1 hour, after which the mortar was remixed for 1 minute and the air content redetermined.

Foam Drainage

The foam drainage test was used to evaluate the stability of foam produced from mixtures of AEAs and cementitious materials in a diluted solution. Little is seen in the literature correlating results from the foam drainage test to air void content in concrete. The foam drainage method as described in X. Wang et al. (2019) blends a combination of water, cementitious material, and AEA, observing the foam for an hour to see the foam stability. This research used proportions of 96.0 g ± 0.5 g of cement, 24.0 g ± 0.5 g of coal ash, and 300 mL ± 0.5 mL of DI. The DI was added first, then the cementitious material, and the AEA last. A blender was used to blend the cementitious solution at 3,000 rpm for 30 seconds (it was difficult to find a blender with the required speed; the research team used a Waring Commercial Torq 2.0 TBB175). The solution was then promptly transferred to a 1,000 mL graduated cylinder. The volume of liquid at the bottom of the cylinder was filmed and recorded as a function of time for an hour. Foam height and volume of liquid were measured every 60 seconds for the first 10 minutes, then at 15-minute markers up to 1 hour.

Fluorescence-Based Method

A stock solution of diluted NP-10 surfactant (0.1% vol or 1,000 ppm) was diluted to a series of concentrations between 50 and 300 ppm and the maximum intensity (near 300 nm) was used to

Table 12. Coal ash sources used for concrete air content testing.

create a calibration curve between fluorescence intensity and solution concentration. Coal ash was added to the solution and mixed in centrifuge tubes using a tube roller for 1 hour. Filtrate was collected using 0.22 µm syringe filters. The concentration of surfactant in the filtrates was measured by fluorescence spectrophotometer with an excitation wavelength of 190 nm. The adsorption capacity was calculated by dividing the volume of NP-10 removed from the solution by the mass of coal ash used in each test.

Concrete Mixing, Fresh-State Testing, and Hardened Air Void Analysis

Linear traverse air void analysis testing was completed to verify the ability to adequately air-entrain concrete mixtures using beneficiated and unconventional coal ashes to acceptable levels of air content and spacing factor. Seven coal ashes were selected to evaluate air entrainment in concrete mixtures composed of (1) standard ashes and (2) beneficiated and unconventional ashes hypothesized to be most likely to be problematic with respect to air entrainment ability. The ashes are shown in Table 12. The SLS AEA was used in all mixtures.

Concrete mixtures were designed using the Ley Tarantula method (Cook et al. 2018) to obtain reasonable workability (slump) for all mixtures and used a cementitious materials content of 600 lb/yd³ and a water-to-cement ratio (w/c) of 0.45. Mixtures were designed to yield an air content of 6% ± 1%. The standard concrete mixture designs shown in Table 13 were used for all coal ash-concrete mixtures.

Slump was tested following ASTM C143; air content was tested following ASTM C231 using a Type B pressure meter. AEA dosages, slump, and volumetric air contents for each mixture are shown in Table 14.

Table 13. Concrete mixture proportions.

Table 14. AEA dosage and fresh concrete properties.

Two 10 cm diameter × 20 cm height (4 in. diameter × 8 in. height) cylinder samples were cast using each mixture following ASTM C39. Air void analysis testing was conducted using the procedure of Peterson et al. (2009). The samples were cured for 14 days and cut through the diameter of the cylinder along the longitudinal axis using a masonry saw. The split samples were then placed in an isopropyl alcohol bath for 24 hours to inhibit continued hydration. Samples were ground and polished using 60-, 180-, 260-, 800-, and 1,500-grit sandpaper. The surfaces were coated in black ink, applied to the surface in two perpendicular directions, and 2 mm wollastonite powder was embedded in the pores.

The samples were scanned using a flatbed scanner and analyzed to determine bulk hardened concrete air content and spacing factor using the Image J software according to the method in Peterson et al. (2001). The top and bottom halves of two replicate cylinders were analyzed and the results were averaged.

Statistical Analysis Methods

The primary objective of this section is to understand and quantify single-source variabilities and testing frequencies of pertinent parameters of standard and unconventional ashes. In this investigation, single-source variability assessment used a proposed and slightly modified combination of the statistical approaches of the Canadian Standard Association CSA A3004-A1 and the power analysis presented in Spencer et al. (2019). Using ASTM C311 as the datum of reference for sampling and testing guidelines either tonnage-wise or frequency-wise, statistical analysis of single-source coal ash variability under the proposed scheme was used to assess whether the ASTM C311 sampling guidelines are adequate for unconventional sources compared to standard sources. The current uniformity requirements in AASHTO M 295 were also assessed for these ashes. In addition, cross-source variability using the Levene test (F-statistic) was used to assess whether unconventional coal ash properties vary more or less than standard coal ash properties.

To provide a sample statistical analysis of single-source variation and cross-source uniformity comparison, historical testing data for various parameters of several coal ashes were attained from coal ash suppliers, including properties of coal ash measured directly by the suppliers during powerplant operation. Four ashes—standard ashes W and X and unconventional ashes Y and Z—were used for the statistical analysis. The investigated ashes are shown in Table 15 with commonly tested specification properties as described in ASTM C311.

Coal ash W was a standard powerplant ash produced commercially for use in concrete. W had historical testing data from 2009 to 2017 with measurements of various properties conducted on a 3,200-ton basis—which controlled over monthly measurements—using composite sampling. Coal ash X was another commercially produced coal ash with historical testing data tracked from 2011 to 2019. Measurements of properties of X were also conducted on composite samples per 3,200 tons, which also controlled over monthly measurements. X also had historical data

Table 15. Investigated ashes and tested historical data.

Note: W and X are standard ashes

Y and Z are unconventional ashes

Y_t^* is coal ash Y treated to reduce air-entrainment adsorption, which was measured posttreatment for Y

on FIT measurements. Only composite measurements, no daily or 400-ton measurements, were provided for moisture content, LOI, and fineness for these ashes.

Coal ash Y was a high-adsorption ash with historical testing data recorded from 2018 to 2019. Y’s properties—including moisture content, LOI, fineness, and FIT—were tested daily or per 400 tons, whichever came first, on regular grab samples. Coal ash Y_t is Y beneficiated for coal ash adsorption using an admixture. Only daily or per-400-tons FIT measurements were provided for the treated Y_t ash. These daily measurements were used for testing uniformity and for comparing variances under the Levene test. Also, to match the data collected from the other coal ashes that were tested per 3,200 tons on composite sampling, Y and Y_t data were converted to a 3,200-ton basis by averaging each set of eight measurements into one simulated composite sample.

Coal ash Z was a comingled and beneficiated bottom ash consisting of 60% coal ash and 40% bottom ash. Z had historical data from January 2022 to April 2022 and was tested on a 3,200-ton basis—which controlled over monthly measurements—using various grab samples from a storage silo to form a composite sampling scheme.

Data collected from all four of these ashes had a sample size of n = 40 when comparing composite samples. A sample size of n = 456 was analyzed for Y on data collected daily or per 400 tons. The procedure for testing single-source variability for quality assurance purposes followed four steps: (1) power law to evaluate the effect of sample size/testing frequency reduction, (2) CSA A3004-A1 to check against the prescribed minimum or maximum limit, (3) checking the uniformity requirements prescribed in AASHTO M 295, and (4) Levene’s variance analysis for variance comparison between standard ashes and unconventional ashes. The procedures are summarized separately in the following paragraphs. A sample detailed analysis procedure is provided in Appendix A.

The power law was used to see how the dataset for various parameters would change if the sample size were reduced for composite sampling. Generally, when a smaller sample size is chosen, signifying a reduction in the frequency of sampling and testing, the distribution of the data increases. This implies there is more variance in the dataset due to fewer data points being measured or sampled. If the change in distribution is not too significant and the data points of the reduced sample size fall within the 95% confidence interval (CI) of the narrower range of the original dataset, then sampling at a reduced frequency would be a satisfactory representation of the original unreduced dataset.

CSA A3004-A1 is an additional test to the power analysis, which allows both the original dataset and the reduced dataset of a given fly ash property to be compared directly to the specification limit of that property as recommended in AASHTO M 295. CSA uses a t-test and allows for computation of a t-statistic of a property based on the difference of the mean of the data for that property, whether reduced or unreduced, directly to the specification limit set forth in AASHTO M 295. If the computed t-statistic is greater than the t-statistic of the dataset at 95% or 99.5% CI

based on that sample size, then the frequency of testing for that property is likely adequate. The passing criterion in CSA represents the notion that all data points of a coal ash property do not exceed the maximum specification value of that property set forth by AASHTO M 295. Both the power law and CSA methods were only performed on composite samples collected every 3,200 tons and were not tested on regular—daily or per-400-tons—samples due to the limited dataset.

The uniformity measurements in AASHTO M 295 for density and fineness were measured by calculating the variation of an individual sample property compared to the average property value established by the 10 proceeding tests. In the specification, uniformity is tested on composite samples collected every 3,200 tons or monthly, whichever comes first. The variation must be within ±5% of the moving average of the proceeding 10 tests. In this study, uniformity is assessed on both composite (3,200-ton) samples and regular (daily or per-400-tons) samples.

The Levene F-statistic test allows for an analysis of variance (ANOVA) between two datasets. Levene test results were used to assess whether the variance in one dataset of coal ash is equal to the other and vice versa. If the Levene test detects equality in variance or no significant difference in an unconventional coal ash property compared to that property in a standard ash, then uniformity in that property between the two ashes is verified—meaning the two ashes exhibit a similar level of variation. In this study, the Levene test was assessed on both composite (3,200-ton) samples and regular (daily or per-400-tons) samples.