CHAPTER 4

Conclusions and Proposed Draft Language for Specification

A large suite of standard and unconventional coal ashes were characterized and tested in pastes, mortars, and concretes for a variety of properties.

Conclusions

Coal Ash Characterization

Of the large number of coal ashes studied, a few showed unusual and potentially problematic properties, such as excessive SO₃, coarse particle size, or high LOI. However, many unconventional ashes did not show properties substantially different from standard ashes. This demonstrates, for many of these ashes, promise for their use in concrete. Specific conclusions include:

• All coal ashes showed significant amorphous contents: 57% for Class C coal ashes and 81% for Class F coal ashes. No significant difference in amorphous content or crystalline phase content was found between standard and unconventional coal ashes. Only a poor link was established between amorphous content and other reactivity measures, suggesting amorphous content should not be used as an indicator of reactivity.
• Laser diffraction median particle size, laser diffraction specific surface area, 45 mm sieve retention, and Blaine fineness were all well correlated with each other; any one of the tests could be used as a measure of fineness/particle size, although the type of information they provide is not the same.
• LOI values showed a complex dependency on both the measurement temperature and measurement atmosphere, driven by various oxidation and reduction mechanisms of carbon, sulfides, sulfur, and iron minerals. Due to the effects of atmosphere and high temperature, no LOI measurements are likely an indicator of adsorption, but the LOI using a furnace at 750°C is likely the best test.
• Coal ash particles showed many different morphologies such as spherical, angular, irregular, and spongy particles. Some evidence of potential contaminants and unusual particle morphologies was observed in several unconventional coal ashes.

Coal Ash Reactivity Testing

All tested coal ashes were reactive, and coal ash reactivity was significantly influenced by classification (C or F), but not by source or whether it was considered unconventional or standard.

• Both R³ and modified R³ reactivity tests were able to differentiate reactive coal ashes from inert materials at a heat release threshold of approximately 100 J/g SCM. Using other measures, the values are 30 g/100 g SCM for CH consumption, and 5 g/100 g SCM for bound water. Bound water and calcium hydroxide trends generally matched those of the heat release.

• Both R³ and modified R³ tests showed that all standard and unconventional ashes were reactive and showed no perceivable differences in heat release on average.
• Coarse ash, high-SO₃ ash, and the CFB ash showed somewhat different reactivity from other tested materials. Specifically, the low-fineness ashes generally showed lower reactivity.
• Both R³ and modified R³ tests showed reactivity of the ashes was heavily influenced by class, i.e., Class C or Class F, more so than by source, although heat release and kinetics were also shown to vary between the two tests.
• Common features include the notion of slow but sustained reaction kinetics of heat release for Class F ashes, and vice versa for Class C ashes, which manifested at early ages when Class C ashes generally generated higher heat release.
• R³ and modified R³ tests also established novel details about the reactivity behaviors of Class C and Class F ashes in simulated systems. The notion that curing temperature had a dominant effect on the reactivity of Class F ashes while additions of sulfate and carbonate dominated the effect of temperature in Class C ashes was established.
• The 3-day heat release correlated well with the ultimate heat release in both R³ and modified R³ tests, and the test duration could be reduced to 3 days for all materials except slow-reacting highly siliceous SCMs.

Paste Testing

None of the coal ashes negatively affected cement hydration; all contributed to strength development, especially at later ages. As all coal ashes were found to be reactive and many did not show unusual physicochemical properties, the team proposes that ash specifications be broadened to include many of these unconventional coal ashes.

• Differences between Class C and Class F coal ashes were significant, for both standard and unconventional coal ashes, and were much larger than any differences between standard and unconventional ashes.
• The CaO content of the ashes influenced the time to peak heat flow, 7-day heat release, 91-day bulk resistivity, 7-day compressive strength, 7-day bound water, and 91-day bound water; most properties increased linearly with CaO content.
• At 7 days, it was not often possible to differentiate coal ashes from inert materials, especially some of the slow-reacting Class F ashes.
• Coal ashes showed increased heat release, compressive strength, calcium hydroxide, and bound water compared to inert materials, especially at a later age of 91 days.
• While unconventional ashes generally showed promising properties, coarse particles, high SO₃, and high LOI resulted in poor performance, at least for some properties.
• Reactivity, determined using the R³ and the modified R³ tests, was able to predict early- and especially later-age cement paste properties, including strength, calcium hydroxide content, and bulk resistivity.

Mortar Testing

SAI, KHI, and TE testing on the coal ashes showed varied success in differentiating ash reactivity and influence on water demand regardless of ash source (standard or unconventional). BR measurements were more successful at differentiating inert and reactive materials than strength measurements, especially at high temperatures.

• Almost all investigated ashes except C, S, and U showed comparable water demands that were within ±5% of the control mixture.
• Except for one unconventional ash with off-specification fineness, the remaining ashes generated SAI values greater than the 75% limit by 56 days, and all ashes generated SAI values exceeding the 75% limit by 91 days.

• A proposed 80% SAI limit and use of MSAI with constant w/cm were found to be overly restrictive by failing reactive coal ashes.
• KHI detected low indices for off-specification fineness and low-reactivity ashes and very low indices for filler materials, circumventing concerns of false-positive classification of inert fillers as reactive SCMs, reinforcing the results of Sutter et al. (2013). It may be a good future alternative to using SAI in the AASHTO M 295 standard due to its ability to better differentiate reactivity.
• TE results were inconsistent from trial to trial and were less meaningful compared to other reactivity tests due to the difficulty in interpreting results.
• Measured BR values in SAI samples showed greater differentiation from the control compared to strength testing at later ages. This method could be used in a future specification, but more data using a larger set of ashes and robustness testing are needed.
• The BRI test showed especially clear differences between inert fillers and reactive ashes when the temperature of testing was increased. BR thus appears to be far more sensitive to reactivity than strength measurements, especially at high temperature.

Concrete Mechanical Property and Durability Testing

Fresh and hardened concrete property measurements showed that most unconventional coal ashes, with a few exceptions, are beneficial when used to replace a portion of cement.

• Both standard and unconventional ashes increased concrete slump on average (by 59% and 20%, respectively) compared to the control mixture. Certain ashes might have higher water demand, which can be mitigated by adjusting admixture dosages.
• All other fresh properties showed comparable values among the ashes by class and source and with respect to the control concrete mixture.
• By 28 and 180 days, both the standard and unconventional ashes provided concrete strengths comparable to the control concrete. Several off-specification ashes generated low concrete strengths, indicating the revised specification is rejecting ashes that reduce the performance of concrete.
• By 180 days on average, both standard and unconventional ashes generated comparable BR values, all of which were significantly greater than the control.
• RCPT results showed no significant differences between standard and unconventional ash sources at 28 and 91 days. All investigated ashes also performed better than the control concrete in terms of chloride-ingress resistance.
• As expected, significantly higher expansions and earlier deterioration due to SA were observed for Class C ashes with higher CaO, CaSO₄, and tricalcium aluminate (C₃A) contents.
• Most standard and unconventional ashes performed more efficiently in resisting SA expansion than the control mortar and filler materials, with the notable exception of certain “bad actor” ashes that would not normally be used for sulfate resistance (i.e., those with high calcium and sulfate contents).

Coal Ash Adsorption Testing

Very few differences in standard and unconventional source coal ash adsorption levels were identified. The FIT provided a good indication of overall ash adsorption, accounting for removal of AEA through all three important mechanisms: precipitation with calcium, surface chemisorption, and physisorption.

• Only ash J, the cyclone collector ash (LOI = 3.1%), showed higher-than-expected adsorption levels; the primary other high-adsorption ash was the 16.5% LOI ash U. Even with higher adsorption, all ashes achieved adequate concrete air entrainment levels.
• In general, high adsorption was neither frequent nor problematic in the set of samples tested and the team believes using unconventional ashes does not present more concern regarding high adsorption than using standard ashes.

• FIT provides a good indication of adsorption and is simple and inexpensive to use. FIT identified adsorption levels somewhat more accurately than LOI.
• An FIT limit of the OPC FIT + 2 standard deviations, or > 160% of the OPC FIT, is recommended as a limit for identifying coal ashes with potential significant increases in adsorption. This limit is not proposed to exclude ashes exceeding the value, but could be used to require additional testing of ashes with the mortar air test or concrete trial batching. Concrete producers can also use this limit as an indication of potential need to increase AEA dosing when using a particular batch of coal ash.
• Iodine number testing is not recommended for use in quantifying adsorption, especially with ashes with calcium contents > 12%. The iodine number test was also unable to identify high adsorption levels in the 16.5% LOI ash, suggesting carbon type may also play a role in the efficacy of this method.
• The fluorescence method developed during this project accurately compared differences in adsorption across the samples tested and provided a close fit with FIT results. However, the fluorescence molecule required by the method (NP-10) does not interact with calcium. As a result, it cannot be used to provide dosing recommendations for cementitious mixtures without additional quantification of AEA precipitation and chemisorption effects. A draft specification for the fluorescence adsorption method has been provided to AASHTO as Appendix 4.
• Of chemical considerations that may be related to use of unconventional ashes, available calcium contents have the most predominant effect. However, this effect is controlled when coal ashes are used or tested in cementitious mixtures, with the cement providing rapidly saturated and consistent Ca²+ conditions. As a result, coal ash chemistry, especially calcium content, does not affect the AEA requirements of cementitious mixtures, and physisorption effects (typically from carbon) provide the predominant mechanism of increased carbon sorption.
• If coal ashes with increased alkali and sulfate contents rapidly dissolve these substances into pore solution at levels high enough to alter overall solution chemistry, they may contribute to small increases in AEA dosage requirements.
• AEA composition did not significantly affect AEA dosage requirements in cement mortars and concretes except in the case of tall oil, which required much higher AEA dosages to entrain air to proper levels when used with coal ash. More significant differences were apparent in FIT testing results, so agencies using FIT may consider specifying a particular AEA for use in agency-required testing. Testing in this project suggests sodium lauryl sulfate and vinsol resin AEAs provide consistent and similar results.
• Cement type also significantly influenced FIT results. Portland limestone cement reduced FIT AEA dosage requirements relative to mixtures using OPC. Similar anecdotal reports from producers during the testing period also suggest that AEA dosing requirements are lower in PLC concrete as well.
• Due to differences in FIT resulting from varying AEA or cement, it may be better to determine FIT limits as a % increase over the FIT of a control cement + AEA mixture.

Coal Ash Uniformity Testing

Coal ash suppliers provided uniformity datasets for two standard and two unconventional ashes. The required composite sampling frequency—monthly or per 3,200 tons—specified in ASTM C311 for ash properties measured in AASHTO M 295 was assessed using the power analysis, t-test, uniformity assessment, and Levene test for all four ashes. Furthermore, regular sampling frequency—daily or per 400 tons—was assessed on fineness and foam index measurements for one unconventional ash. This was the only regular sampling data provided. Conclusions drawn here are only applicable to the tested ashes.

• The results of the power analysis and t-test on composite sampling (per 3,200 tons) conducted on pertinent properties (moisture content, density, LOI, fineness, AEA dosage, foam index)

• of two standard and two unconventional coal ashes indicated that a reduction of testing frequency from every 3,200 tons to every 6,400 tons did not negatively affect the ability of testing to catch outlier data nor fail to assess properties with respect to given limits. However, more data are needed to assess how regular sampling for moisture content, LOI, and fineness is affected by unconventional-source ashes.
• When comparing the uniformity of the moving mean for the composite samples and subsequently using the Levene test to compare variances, the unconventional ashes had comparable or lower variance in fineness and density compared to the standard ashes, but much higher variance in moisture content (except in one instance). When comparing foam index uniformity, the untreated unconventional coal ash Y had much higher variance compared to the same coal ash after treatment and another standard ash. The beneficiation significantly improved uniformity variance for foam index measurements.
• Additional uniformity testing on regular sampling data—daily or per 400 tons—for an unconventional ash showed it had much higher variability in fineness and foam index measurements compared to composite sampling. This suggests that load-to-load variability may not be captured by current uniformity testing, which is performed on composite sampling. More regular sample data are needed to further investigate load-to-load variability.

Proposed Draft Language for AASHTO M 295 Standard Specification

The research team proposes changes to the AASHTO M 295 specification in two steps: (1) harmonization with the current ASTM C618-23e1 and (2) updating the specification to accommodate other issues identified by this research in the harmonized specification.

Harmonization with ASTM C618

1. Change scope to allow harvested, bottom, and processed ashes.

   Reasoning: This has already been completed in ASTM C618-23e1. A large quantity of data and studies, including the work presented in this report, has shown that processed and harvested coal ash, bottom ash, and comingled ashes meeting all specification requirements show satisfactory performance when used in concrete. Ashes that didn’t perform well in this study were captured by recommended specification limits, regardless of source.

2. Add no. 100 sieve fineness limit to a maximum of 10% for harvested and bottom ash samples.

   Reasoning: In freshly produced ash, very coarse particles are uncommon; however, this may not be the case for harvested ashes, where particles may have agglomerated or comingled with coarse bottom ash, which might not be detected by the current 45-µm sieve test limit. Thus, the no. 100 sieve limit acts as a further quality control check to ensure adequate fineness for harvested materials.

3. Removal of the soundness limit as measured by autoclave.

   Reasoning: This change, already made in ASTM C618-23e1, was driven by work done by Hooton and others, who showed that the autoclave test does not predict long-term expansion. The test is unrealistic in measuring volume stability and adds an additional testing burden, while potentially preventing useful materials from being utilized in concrete.

4. Change in terminology from “coal fly ash” to “coal ash.”

   Reasoning: This change, already made in ASTM C618-23e1, is driven by change number (1) above. Since other ashes (bottom ashes and bottom ash/coal ash blends) will be covered, the name must be changed to coal ash and not be limited to fly ash.

5. Increase maximum allowable LOI from 5% to 6%.

   Reasoning: LOI was effective in identifying coal ashes requiring significant increases in concrete AEA dosing and should continue to be used. The team proposes reconciling the

5. AASHTO M 295 with ASTM C618-23e1, increasing allowable LOI from 5% to 6%. This study did not identify any performance issues with air entrainment in samples with LOI near this limit.

6. Allow use of Class F coal ashes with up to 12% LOI with accompanying successful field performance records or laboratory testing of air entrainment in mortar or concrete.

   Reasoning: This exception is listed in ASTM C618-23e1. Testing in this project indicated that variances in adsorption are less significant in mortar and concrete than in other testing methods. Additionally, in this study, concrete made with 17% LOI coal ash was able to be successfully air entrained. Therefore, increased levels of LOI may not translate to poor air entrainment in mortar or concrete mixtures and should not be used to exclude coal ashes. However, increased LOI likely indicates contamination with sulfates, organics, or clay, and so an upper limit of 12% should still be prescribed.

Updating the Specification

1. Change water requirement to “report only.”

   Reasoning: Coal ashes that exceed the current 105% limit can still be used in concrete, however workability may be reduced using a coal ash with high water demand. Workability issues can be resolved by the concrete producer in concrete mixture design using a water-reducing admixture.

2. Adding R³ reactivity bound water limit of 3.5 g/100 g dry paste limit when tested by ASTM C1897-Procedure B.

   Reasoning: A large volume of data has shown this number reliably differentiates reactive and inert materials. While the heat release and bound water data appear to be equivalent and interchangeable, Procedure A of ASTM C1897—heat release using isothermal calorimetry—may not be suitable for most labs due to the greater cost and difficulty in measurement. Results also appear to show variability between labs, possibly due to issues with baseline collection. The same recommended limit and note language as the broader “Standard Specification for Supplementary Cementitious Material for Use in Concrete” currently being balloted at ASTM have been used here. However, with further research, this limit might change.

3. Adding the FIT as “report only” under optional physical requirements in accordance with ASTM C1827.

   Reasoning: The FIT provides better differentiation between ashes with varying adsorption and AEA requirements than LOI measurements, which may incorrectly misidentify high-adsorption samples with varying forms of carbon. Use of sodium lauryl sulfate or vinsol resin AEAs and analysis of FIT results as a % change from a control mixture are proposed.

Lastly, the following changes are proposed for future iterations of the AASHTO M 295 specification to transition to a more performance-based approach, but are not quite ready to be implemented. Two of the changes relate to the SAI test, which is proposed to remain in the specification for now since it is a well-known test method and can still be useful in differentiating less reactive coal ash. However, this research has shown there are better methods that are not yet standardized for this purpose:

1. Use of the BRI following ASTM test method approval and/or introducing a report-only measurement of BR on SAI samples.

   Reasoning: This test is more sensitive than the SAI test. However, round-robin, robustness testing, and a specification is still required before incorporation into the specification. A draft specification of the BRI method has been provided to AASHTO as Appendix 3.

2. Use of KHI in place of SAI testing.

   Reasoning: This test is also more sensitive than the SAI test. However, round-robin, robustness testing, and a specification is still required before incorporation into the specification.

3. Using uniformity of FIT as a replacement for the current uniformity of 18% air content in mortar.

   Reasoning: Compared to measuring air content in mortar, the FIT a is more rapid and sensitive measurement. This enables it to be used more frequently by the supplier and provides a better indication of changes in adsorption. Further, FIT is less affected by mixture design variables, such as variances in flow, interactions between cementitious and sand particles, and mixing protocols.

Overall, more research is needed to include these changes into a future specification.

The draft language for AASHTO M 295 Standard Specification with ASTM C618-23 Harmonization has been provided to AASHTO as Appendix 1, and the draft language for AASHTO M 295 Standard Specification with ASTM C618-23 Harmonization and Additional Changes has been provided to AASHTO as Appendix 2.

Table 41 shows the changes in coal ash pass/fail compliance if the AASHTO M 295 Standard with ASTM C618-23 Harmonization and Additional Changes were to be implemented. Coal ashes M (if 56-day SAI data are permitted) and V would now pass after the changes are implemented. Furthermore, the explicit allowance of processed or beneficiated coal ashes would also officially pass coal ashes B, K, N, O, and R. All the ashes tested here were classified as reactive and generally performed well in mortar and concrete testing except for some notable exceptions regarding workability and SA performance. Therefore, Appendix B has been provided as a guide to DOTs if they are interested in using a coal ash in concrete that does not pass AASHTO M 295.

Table 41. Compliance of coal ashes under proposed AASHTO M 295-23.