Incorporating Physical and Chemical Characteristics of Fly Ash

Incorporating Physical and Chemical Characteristics of Fly Ash in Statistical Modeling of Binder Properties FINAL EXAM - MSCE Prasanth Tanikella Major Professor: Jan Olek Department of Civil Engineering July 23 rd , 2009 Prasanth Tanikella - Purdue University 1

Objectives and Hypothesis • The goal of this research was to: – Characterize two sets of fly ashes (Class C and Class F) – Statistically verify the importance of their physical and chemical properties on the performance of binary and ternary paste systems • Scope of the Project (3 Phases) – Phase 1 – Characterization of Fly Ashes – Phase 2 – Effect of Fly Ashes on the Properties of Binary Paste Systems (cement + fly ash) – Phase 3 – Effect of Fly Ashes on the Properties on Ternary Paste Systems (cement + two different fly ashes) Prasanth Tanikella - Purdue University 2

Phase 1 – Characterization of Fly Ashes Collected 20 different fly ashes (13 Class C and 7 Class F) • 15 of them ( 9 Class C ashes and 6 Class F ashes) are currently on the INDOT’s list of approved pozzolanic materials • A database summarizing the physical and chemical characteristics of the collected fly ashes and the impact of these properties on the behavior of binders would benefit the engineers, contractors and concrete producers Test Methods • Total Chemical Analysis and loss-on ignition Soluble Sulfates and Alkalis Particle Size Distribution Magnetic Particles Crystalling component and glass fraction Morphology Strength Activity Index ASTM C 311 Ion Chromatography Laser Particle Size Analyzer and Sedimentation Analysis Teflon coated bar magnet X-ray Diffraction SEM ASTM C 311 Prasanth Tanikella - Purdue University 3

Phase 1 Results Range of chemical compositions Ca. O (%) Si. O 2 (%) Al 2 O 3 (%) Fe 2 O 3 (%) Sulfate Alkali (%) Content as Na 2 O (%) F 1 - 9 39 - 56 18 -29 5 - 25 0. 4 - 2 1. 4 – 2. 6 1. 4 – 2. 4 C 17 -28 32 -44 17 - 22 6 - 10 0. 05 – 1. 3 1. 6 - 3. 9 0. 25 - 0. 9 CLASS LOI(%) Ca. O (%) Si. O 2 (%) Al 2 O 3 (%) Fe 2 O 3 (%) Sulfate (%) Alkali Content as Na 2 O (%) F 1 - 9 39 - 56 18 -29 5 - 25 0. 4 - 2 1. 4 – 2. 6 1. 4 – 2. 4 C 17 -28 32 -44 17 - 22 6 - 10 0. 05 – 1. 3 1. 6 - 3. 9 0. 25 - 0. 9 CLASS * INDOT list of approved fly ashes Prasanth Tanikella - Purdue University LOI(%) 4

Phase 1 Results XRD – Typical Class F Fly Ash XRD pattern for Elmer Smith fly ash § § • • • Typical X-ray patterns for Class F fly ashes Includes 1. Quartz – Si. O 2 2. Mullite – Al 6 Si 2 O 13 3. Anhydrite – Ca. SO 4 4. Hematite – Fe 2 O 3 5. Magnetite – Fe 3 O 4 6. Lime – Ca. O Measured magnetic content is generally very high (with two exceptions) A hump, representing a silica-type glass with a maximum at 2θ=~25° is visible Glass “hump” is generally higher than that observed for Class C ashes XRD pattern for Miami 7 fly ash Prasanth Tanikella - Purdue University 5

Phase 1 Results XRD - Typical Class C Fly Ash § § • • • X-ray pattern for a typical Class C fly ash Includes 1. Quartz – Si. O 2 2. Anhydrite – Ca. SO 4 3. Merwinite – Ca 3 Mg(Si. O 4)2 4. Periclase – Mg. O 5. Lime – Ca. O Glass peak is similar for all the ashes of this type Magnetite might be present in the fly ash, either in crystalline form or in the glass A hump, representing a calciumaluminate type of glass with a maximum at 2θ=~30° is visible XRD pattern for Hennepin fly ash Prasanth Tanikella - Purdue University 6

Phase 1 Results XRD – Glass Content Estimation § • • Glass content was empirically estimated by calculating the area under the glass hump Three softwares were used for the purpose xy. Extract – To extract points from the XRD pattern Lab. Fit – To fit the curve very precisely through the extracted points Sicyon Calculator – To integrate the fitted curve Prasanth Tanikella - Purdue University 7

Phase 1 Results Particle Size Distributions Comparison of PSDs for Class F and C ashes 100. 0 90. 0 80. 0 Undersize Percentage (%) • Class F and Class C ashes form two different bands of PSDs • The band of Class C ashes is shifted towards the left of the band of Class F ashes Class C 70. 0 60. 0 50. 0 40. 0 30. 0 20. 0 Class F 10. 0 0. 1 1. 0 10. 0 Diameter (microns) Prasanth Tanikella - Purdue University 100. 0 8

Phase 1 Results Discrepancies in PSD 100 Discrepancies observed in PSD § The pipette analysis seems to work well for particles larger than 5 micron § The results below 5 microns seem to diverge from either of the curves § Even though the sedimentation technique does not work well for particles smaller than 5 microns, based on the data it is reasonable to assume that the PSD based on Lab 1 (Purdue) data is accurate § Fly Ash Petersberg 90 80 70 60 50 40 30 20 10 0 0. 1 1 10 1000 100 90 Fly Ash Trimble 80 70 60 50 40 Lab 1 30 Lab 2 20 Pipette 10 0 0. 1 Prasanth Tanikella - Purdue University 1. 0 100. 0 9 1000. 0

Results Morphology of class F (Type I) ashes § § There is a large variation in the sizes and shapes of the particles Particles with rugged surface are generally magnetic, contrary to the class C fly ashes Many hollow particles present Relatively smaller number of unburnt carbon particles, but bigger particles have been observed, which is consistent with the higher LOIs values observed in Class F ashes Zimmer Petersburg Prasanth Tanikella - Purdue University Elmer Smith Mill Creek 10

Results Morphology of class C ashes § § § Wide range of sizes of spherical particles Many hollow particles with shell generally composed of silica and alumina Frequent irregularly-shaped particles (often with rugged surfaces) predominantly composed of sulfates or magnesium, or rarely sodium Labadie Will County Prasanth Tanikella - Purdue University Kenosha Rush Island 11

Phase 1 Summary – Phase 1 Characterization of fly ashes • • Significant variations in the chemical and physical characteristics of fly ashes observed The strength activity index of Class C ashes was higher than Class F ashes The glass content for all the Class C ashes was higher than the glass content for all but two Class F ashes, thus indicating that although Class C fly ashes have less glass than these two Class F ashes, the glass in Class C ashes is more reactive The morphology of the ashes was similar irrespective of the class, with a few exceptions The particle size distributions of class C and class F ashes were significantly different All mean particle sizes in class F were larger than mean particle sizes in class C ashes, resulting in a lower surface area of class F ashes The LOI values of all class F ashes were higher than that of the C ashes Prasanth Tanikella - Purdue University 12

Phase 2 - Evaluation of the hydration characteristics of cement-fly ash binder systems § Binder systems consisted of portland cement with 20% (by weight) replaced by fly ash § Pastes with constant water/binder ratio (0. 41) were tested for various properties including, Ø Initial Time of Set – Vicat needle (ASTM C 191) Ø Heat of Hydration – Isothermal Calorimetry (at a constant temperature of 21 o. C) Ø Amount of Calcium Hydroxide at ages 1, 3, 7 and 28 days - TGA Ø Non-evaporable water content at 1, 3 7 and 28 days – TGA Ø Rate of strength gain at 1, 3, 7 and 28 days – Strength activity index (ASTM C 311) Prasanth Tanikella - Purdue University 13

Phase 2 Initial Setting Time - Results § § Range of set time for Class C ashes – (1 hour to 4. 5 hours) Range of set time for Class F ashes – ( 2. 5 hours to 3. 5 hours) Prasanth Tanikella - Purdue University 14

Phase 2 A Typical Calorimeter Curve Time of Peak Heat Total Heat • Ø Ø Ø Data acquired from the calorimeter curve Peak heat of hydration (W/kg) Time of peak heat of hydration (minutes) Total heat of hydration (J/kg) – ( Area under the curve from 60 minutes to 3 days) Prasanth Tanikella - Purdue University 15

Phase 2 Peak Heat of Hydration - Results § § § Most ashes tend to reduce the peak heat of hydration compared to cement Class F ashes in general have a higher peak heat of hydration than Class C ashes Kenosha, the fly ash with the lowest peak heat of hydration had a flash set Prasanth Tanikella - Purdue University 16

Phase 2 Time of Peak Heat of Hydration - Results § § § Most ashes tend to delay the occurrence peak heat of hydration compared to cement Class C ashes in general have a higher time of peak heat than Class C ashes Kenosha, the fly ash with the lowest peak heat of hydration had longest time of peak heat Prasanth Tanikella - Purdue University 17

Phase 2 Total Heat of Hydration - Results § § § Most ashes tend to reduce the total heat of hydration compared to cement Most Class C ashes have a similar total heat of hydration Quite a few of the Class F ashes have a similar total heat of hydration as that of most Class C ashes Prasanth Tanikella - Purdue University 18

Phase 2 Thermo-gravimetric Analysis (TGA) § Calcium hydroxide content and nonevaporable water content were estimated using TGA at various ages (1, 3, 7 and 28 days) § Calcium Hydroxide content between 480 o. C and 550 o. C (carbonation taken in to account) § Non-evaporable water content calculated according to Barneyback, 1983. Prasanth Tanikella - Purdue University 19

Phase 2 Calcium Hydroxide Content at 1 day - Results § § Most ashes tend to reduce the amount of calcium hydroxide at 1 day compared to plain cement paste (with some exception) Class F ashes have a slightly higher CH content than Class C ashes at early ages Prasanth Tanikella - Purdue University 20

Phase 2 Calcium Hydroxide Content at 28 days - Results § § Most of the ashes show a higher amount of calcium hydroxide at 28 day compared to plain cement paste Difference in the rates of reactions in the fly ashes Prasanth Tanikella - Purdue University 21

Phase 2 Non-evaporable Water Content at 1 day - Results § § Most ashes tend to lower the amount of non-evaporable water content at 1 day compared to plain cement paste Plain cement has a higher degree of hydration than most of the fly ash pastes Prasanth Tanikella - Purdue University 22

Phase 2 Non-evaporable Water Content at 28 days - Results § § Most of the Class C ashes show a higher amount of non-evaporable water at 28 day compared to Class F ashes Difference in the rates of reactions in the fly ashes Prasanth Tanikella - Purdue University 23

Phase 2 Strength Activity Index at 28 days - Results § All of the Class C ashes show a higher strength at 28 days compared to plain cement paste while Class F ashes show a lower strength comparatively Prasanth Tanikella - Purdue University 24

Phase 2 Statistical Modeling of Binary Binders STEP 1 - Perform linear regression analysis for each of the 16 dependent variables (hydration related properties of ashes) using all the data points (13 Class C and 7 Class F binary pastes) STEP 2 - Prepare a table with a list of models containing the sets of independent variables that must affect the dependent variables, in a decreasing order of "Adj-R 2" (only models with the best 10 adj. R 2 values were included) STEP 3 - Perform linear regression analysis for the same set of 16 dependent variables as in Step 1, but using only those independent variables that were selected based on Step 2 for both Class C and Class F ashes separately STEP 4 - If both the model for Class C and Class F ashes are statistically significant, the set of variables selected in Step 2 is used in the formulation of the experiments for the ternary paste systems Prasanth Tanikella - Purdue University 25

Phase 2 Statistical Modeling of Binary Binders Independent Variables Abbreviations Mean Particle Size meansize Specific surface area measured using Blaine's Physical Properties apparatus blaines Specific surface area measured using laser Spsurface particle size analyzer Chemical Properties Physico-chemical Properties Calcium oxide content cao Sum of silicon, aluminum and iron oxide contents SAF Magnesium oxide content mgo Aluminum oxide content Alumina Sulfate content sulfate Loss-on ignition carbon Glass content measured using X-ray diffraction Prasanth Tanikella - Purdue University glass 26

Dependent Variables Prasanth Tanikella - Purdue University 27

Phase 2 Ten Models with the highest Adj-R 2 – Set Time Model Number of Variables in Number the model 1 Adjusted R 2 Variables in the model 3 0. 2447 0. 3706 sulfate, alumina, glass 2 5 0. 2298 0. 4437 sulfate, SAF , mgo, alumina, glass 3 2 0. 223 0. 3093 sulfate, alumina 7 0. 2189 0. 5226 1 0. 217 0. 2605 7 0. 2099 0. 5172 6 0. 2095 0. 473 8 4 0. 2089 0. 3847 sulfate, SAF, mgo, alumina 9 2 0. 2032 0. 2917 sulfate, carbon 5 0. 2008 0. 4228 4 5 6 7 10 Prasanth Tanikella - Purdue University spsurface, meansize, sulfate, carbon, SAF, alumina, glass sulfate spsurface, meansize, sulfate, carbon, cao, alumina, glass spsurface, sulfate, SAF, mgo, alumina 28

Phase 2 ANOVA Table (Class C Ashes) – (SAI) at 28 days Source DF Model 3 Error Total Sum of Mean Square F Value p-Value 470. 0203 156. 67343 4. 444977 0. 0407 8 281. 9784 35. 2472975 11 751. 9987 R 2 0. 625 Adj - R 2 0. 4844 Standard Error t-Value p-Value Squares Parameter Variable DF Intercept 1 142. 0434 1. 66837 85. 13902 0. 0036 meansize 1 -1. 573 0. 10387 -15. 1439 0. 0246 sulfate 1 -15. 7847 0. 01841 -857. 398 0. 0135 SAF 1 0. 0496 0. 11843 0. 418813 0. 9266 Estimate Prasanth Tanikella - Purdue University 29

Phase 2 Observed Vs Predicted (Class C Ashes) – Set Time 5. 5 Predicted Setting Time (Hours) 5 4 3. 5 3 2. 5 2 1. 5 1 1 1. 5 2 2. 5 3 3. 5 4 4. 5 5 Observed Setting Time (Hours) Prasanth Tanikella - Purdue University 30

Phase 2 ANOVA Table (Class F Ashes) – (SAI) at 28 days Source DF Sum of Squares Mean Square F Value p-Value Model 3 107. 65676 35. 88559 40. 13 0. 0244 Error 2 1. 78839 0. 894195 Total 5 109. 44515 R 2 0. 9837 Adj - R 2 0. 9591 Parameter Standard Estimate Error t-Value p-Value Variable DF Intercept 1 126. 13758 17. 78975 7. 090464 0. 0193 meansize 1 -0. 67193 0. 23397 -2. 87186 0. 1029 sulfate 1 -9. 27674 1. 16409 -7. 96909 0. 0154 SAF 1 -0. 00329 0. 1415 -0. 02325 0. 9835 Prasanth Tanikella - Purdue University 31

Phase 2 Observed Vs Predicted (Class F Ashes) – Set Time 4 3. 8 Predicted Set Time (Hours) 3. 6 3. 4 3. 2 3 2. 8 2. 6 2. 4 2. 2 2. 4 2. 6 2. 8 3 3. 2 Observed Set Time (Hours) Prasanth Tanikella - Purdue University 3. 4 3. 6 3. 8 4 32

Phase 2 Summary- Phase 2 Binary Binder Systems Property Set Time Peak Heat Timepeak Total Heat Ca(OH)2 • • • Most Influencing Variables Sulfate, alumina, glass Spsurface, SAF, glass Spsurface, Meansize, Mg. O Meansize, carbon, SAF Blaines, Spsurface, sulfate, cao, glass, carbon, alumina Significant Variables None Spsurface, Ca. O Spsurface Meansize Blaines Wn Blaines, carbon, alumina, sulfate SAF, mgo Blaines SAI 7 Day SAF, Ca. O, Glass SAF, Ca. O SAI 28 Day Meansize, sulfate, SAF Meansize, Sulfate Physical characteristics of fly ash had a higher effect than chemical characteristics of fly ash Surface area was found to be the most influencing variable affecting most of the properties of the binder system at both early and later ages Variables including SAI (at later ages) and time of peak heat of hydration can be predicted accurately using the respective statistical models Prasanth Tanikella - Purdue University 33

Phase 3 – Ternary Binder Systems § Ternary Binder System – Cement + 2 different fly ashes § Total 20 % of the cement replaced with the mixture of fly ashes at specific percentages § Water/binder ratio was 0. 41 (unless specified in the standard) OBJECTIVES 1. To ascertain the applicability of the weighted sum of the models chosen for the binary paste systems to predict the properties of ternary binder systems. 2. The analysis of which of the chosen independent variables (from binary binder models) have the highest effect on the properties of ternary systems Prasanth Tanikella - Purdue University 34

Phase 3 Ternary Binder Systems Experimental Design § Full factorial design consists of 180 experiments when the ratio of the two fly ashes is fixed § Fractional factorial design – Orthogonal Array Technique (Taguchi Method) Requirements of a Fractional Factorial Design § Reduction in the number of experiments § The data should be a representative data set of the full factorial design § The quality of the inferences obtained should be similar to the inferences from the full factorial design Prasanth Tanikella - Purdue University 35

Phase 3 Orthogonal Array Technique – Taguchi Method § § A special test matrix is prepared Columns – Factors (Dependent Variables) Rows – Each row is an experiment (Mix design) Values in the table – Factors levels, levels at which the three factors are varied Factors Experiment A B C D 1 1 1 1 1 2 2 2 3 1 3 3 3 4 2 1 2 3 5 2 2 3 1 6 2 3 1 2 7 3 1 3 2 8 3 2 1 3 9 3 3 2 1 Prasanth Tanikella - Purdue University 36

Phase 3 Test Matrix for Set Time § Columns – Factors (Sulfate, Alumina, Glass) § Rows – Each row is an experiment (Mix design) § Factor Levels – 33. 33 , 50 and 66. 67 percentile of the available data set Experiment Glass Sulfate Alumina Levels 1 1 (0. 4347) 1 (18. 75) 1 (1. 294) Factors 2 1 2 2 3 1 3 3 4 2 1 2 5 2 2 3 6 2 3 1 7 3 1 3 8 3 2 1 9 3 3 2 1 2 3 0. 4347 0. 5281 0. 7593 Alumina (%) 18. 75 19. 28 20. 07 Glass 1. 294 1. 476 1. 513 Sulfate (%) Prasanth Tanikella - Purdue University 37

Phase 3 Scaled Standard Deviation - SSD • It is practically not possible to choose two ashes (in any proportions) having a target combination of three different factors Standardizing the Error in the Fly Ash Combinations • Scaled Standard Deviation (SSD) to evaluate the error of the combination • SSD values up to 0. 3 were found to give a good approximation of the target values Prasanth Tanikella - Purdue University Initial Time of Set (Hours) SSD = 4. 5 4 3. 5 3 2. 5 2 0 0. 2 SSD 0. 4 0. 6 38

Phase 3 Analysis of the Data - Additivity Model 1 – The two models obtained for Class C and Class F ashes from the binary binder, with the chosen independent variables (factors) were used to predict the properties (dependent variables) for both the Classes of ashes separately. The two predicted values were then added in the proportions of the added fly ashes to obtain the final value of prediction for the ternary binder system. This value was compared with the experimentally observed values. Model 2 – The best models obtained for Class C, Class F ashes individually were used to predict the properties of the ashes in the mixture separately, and the predicted values of the properties were added in the proportion of the ashes to obtain the final value of the predicted properties of the ternary binder systems. Model 3 – The model obtained for the entire set of Class C and Class F ashes together using all the 20 data points, containing the best three chosen independent variables was used to predict the properties of Class C and Class F ashes separately. Prasanth Tanikella - Purdue University 39

Phase 3 Analysis(Objective 2) – Influencing Variables Analysis of Variance (ANOVA) – Factor level ANOVA • Total Sum of Squares , ST = • Variation caused by a single factor A, SA = - where, NA 1 = total number of experiments in which level 1 of factor A is present A 1 = the sum of the results of level 1 of factor A (Xi) • Mean squares (Variance): VA = • Pure sum of squares: SA’ = SA – (Ve x f. A) • Percent Influence: PA = where, f. A is the degrees of freedom for factor A Ve is the variance for the error term, which is calculated as Se = error sum of squares fe = error degrees of freedom T = sum of the results (Xi) and N is total number of results Prasanth Tanikella - Purdue University 40

Phase 3 Test for Additivity – SAI 28 days (%) Model 1 Model 2 Model 3 Exp No. Observed Predicted 1 125. 3 110. 3 104. 9 80. 5 2 123. 1 109. 3 104. 2 78. 8 3 119. 4 106. 0 103. 6 76. 5 4 118. 1 103. 7 102. 6 75. 1 5 118. 5 108. 1 101. 7 70. 4 6 115. 6 98. 8 96. 1 61. 9 7 113. 4 104. 7 104. 0 65. 2 8 117. 3 94. 9 90. 4 57. 1 9 112. 2 96. 3 90. 2 54. 3 • Observed Strength higher than most predicted for all the combinations of the fly ash Prasanth Tanikella - Purdue University 41

Phase 3 Percent Influence Set Time Peak Heat Time Peak Wn 28 Day SAI 28 Day Sulfate Alumina Glass Error 26. 81 7. 07 14. 1 52. 01 Spsurface SAF Glass Error 39. 46 37. 76 2. 84 19. 94 Spsurface Meansize Mgo Error 63. 44 4. 2 0. 77 31. 59 Blaines Carbon Alumina Error 49. 88 10. 91 14. 35 24. 85 Meansize Sulfate SAF Error 66. 62 15. 1 3. 77 15. 52 • Observed Strength higher than most predicted for all the combinations of the fly ash Prasanth Tanikella - Purdue University 42

Phase 3 Summary - Phase 3 Ternary Binder Systems • None of the properties were found linearly additive Reasons could be: 1. Variables chosen in the binary binder systems can not explain a significant variation in the dependent variables 2. A few of the binary binder models were not significant and the error carried into the analysis of the ternary binder systems 3. The chosen variables might not be “linearly” related to the properties of the binary binder systems • Weighted linear combinations of strength activity index at 28 days suggest a synergistic effect in the addition of two ashes to the binder system • Physical properties of the fly ashes were more influencing than the chemical composition in most of the properties • Surface area of fly ashes has the highest effect on the properties • The most influencing variables on ternary binder systems were similar to the ones for binary binder systems Prasanth Tanikella - Purdue University 43

Conclusions • • • Class C and Class F ashes were significantly different in both their physical characteristics and chemical composition There was significant difference in the effect of the two classes on binder properties Both physical and chemical characteristics of fly ash had an effect on the binder systems The sets of variables affecting each of the properties were unique The signs of the coefficients in the models indeed pointed out the type of effect on the property The statistical analysis of the properties of binary binders allowed us to draw inferences about the characteristics of fly ash which held the highest importance Prasanth Tanikella - Purdue University 44

Conclusions • • Some of the properties could not be accurately predicted by the statistical models with good significant as there were errors introduced by the limited number of variables chosen for modeling Statistical analysis on the properties of ternary systems indicated that these properties are not a weighted linear combination of binary binder models The statistical analysis of the properties of the ternary systems allowed us to draw inferences about the most significant variables and also about their relative percent influence Specific surface area of the fly ash had the highest impact on all the properties of binder systems Prasanth Tanikella - Purdue University 45

THANK YOU Prasanth Tanikella - Purdue University 46

Phase 1 Results Chemical composition of fly ashes PROPERTY Ca. O(%) Si. O 2(%) Al 2 O 3(%) Fe 2 O 3(%) Sulfate (%) Alkali Content as Na 2 O (%) LOI(%) FLY ASH Class Petersburg F* 1. 86 43. 82 21. 74 25. 29 0. 87 2. 29 1. 39 Elmer smith F* 9. 31 41. 6 17. 74 22. 02 0. 60 2. 32 2. 37 Trimble F 2. 5 46. 91 21. 08 19. 9 1. 09 2. 45 1. 89 Miami 8 F* 3. 98 55. 52 26. 02 4. 62 0. 76 2. 55 2. 43 Mill creek F* 5. 42 47. 48 19. 99 18. 52 0. 69 2. 55 1. 38 Zimmer F* 4. 94 38. 66 18. 96 24. 9 2. 08 1. 44 1. 48 Miami 7 F* 1. 25 55. 89 29. 45 4. 96 0. 42 2. 20 2. 31 Rockport C 16. 98 43. 65 21. 76 6. 58 0. 45 2. 08 0. 9 Joppa C* 26. 23 35. 75 18. 01 6. 36 0. 07 2. 31 0. 35 Kenosha C 23. 35 37. 78 20. 11 5. 87 0. 53 2. 18 0. 38 Miller C* 24. 62 36. 38 18. 74 6. 03 0. 52 2. 08 0. 44 Hennepin C* 21. 8 40. 36 19. 38 5. 91 0. 35 1. 99 0. 61 Joliet C* 26. 98 32. 12 17. 88 6. 41 1. 28 3. 92 0. 49 Vermilion C* 23. 92 39. 13 18. 77 6. 19 0. 22 1. 91 0. 43 Will county C* 26. 97 32. 3 18. 55 6. 47 0. 43 3. 06 0. 35 Rush island C 27. 66 34. 23 16. 91 6. 86 0. 05 2. 26 0. 17 Baldwin C* 25. 23 35. 06 19. 39 6. 25 0. 28 2. 24 0. 49 Labadie C 24. 26 37. 03 19. 28 6. 46 1. 14 1. 94 0. 25 Schafer C* 20. 29 41. 9 19. 32 6. 76 0. 48 1. 83 0. 44 Edwards C* 24. 28 33. 15 19. 21 10. 11 0. 75 1. 63 0. 43 * INDOT list of approved fly ashes Prasanth Tanikella - Purdue University 47

Phase 1 Results Physical characteristics of fly ashes PROPERTY Specific Surface - LPSD Strangth activity (cm 2/g) index (%) FLY ASH Class Blaine’s Specific surface (cm 2/g) Mean Size (microns) Magnetic particles (%) Specific Gravity Petersburg F* 2391 28. 37 9849 104. 3 37. 72 2. 63 (2. 55) Elmer smith F* 3092 33. 24 6344 109. 9 32. 99 2. 64 (2. 52) Trimble F 3253 27. 35 8857 109. 1 26. 39 2. 69 Miami 8 F* 3600 31. 58 13012 112. 3 4. 18 2. 22 (2. 21) Mill creek F* 3739 26. 35 10295 125. 7 24. 9 2. 60 (2. 46) Zimmer F* 3782 26. 1 11308 96. 2 35. 32 2. 68 (2. 64) Miami 7 F* 4088 30. 41 12592 118. 2 3. 68 2. 26 (2. 22) Rockport C 4354 32. 2 11963 134. 1 3. 5 2. 56 Joppa C* 4371 18. 37 17597 135. 9 0. 31 2. 72 (2. 70) Kenosha C 4452 17. 35 16577 121. 2 0 2. 80 Miller C* 4851 24. 93 17089 123 0 2. 63 (2. 66) Hennepin C* 5125 16. 88 16457 136. 5 0. 07 2. 70 (2. 35) Joliet C* 5356 14. 48 19776 116. 7 0 2. 84 (2. 46) Vermilion C* 5536 13. 85 17928 136. 7 0. 12 2. 69 (2. 64) Will county C* 5907 14. 85 19646 140. 2 0 2. 84 (2. 49) Rush island C 5924 20. 77 17477 127. 7 0 2. 81 Baldwin C* 6102 21. 99 15492 127. 2 0 2. 72 (2. 66) Labadie C 6269 16503 118. 6 2. 89 2. 75 Schafer C* 6428 18. 87 14679 118. 8 2. 7 2. 58 (2. 59) Edwards C* 7306 15. 08 22075 133 3. 34 2. 63 (2. 65) * INDOT list of approved fly ashes. 48 SG listed in () indicate values from INDOT’s list of approved fly ashes.

Phase 1 Analysis Andreasen Pipette Analysis An attempt was made to resolve the differences in the observed PSD using the Andreasen Pipette Particles suspended in dispersing solution Particles settle at different rates. The rates depend on the radius and density of the particles Stokes law used to calculate the particle size Prasanth Tanikella - Purdue University 49

Phase 1 Results Morphology General Inferences § Fly ash particles were generally spherical in shape § Fly ash particles were found inside some of the hollow spherical particles § A few pieces of carbon (usually of a very large size) can be seen with a “Swiss Cheese” structure § Particles with rugged surface were generally magnetic in Class F ashes, contrary to the class C fly ashes which contained sulfate, magnesium and sodium Prasanth Tanikella - Purdue University 50

Phase 2 - Evaluation of the hydration characteristics of cement-fly ash binder systems • When used as a substitute for part of the cement, fly ash offers a lot of benefits, both in terms of early and later hydration characteristics and in terms of the economy • As seen from Phase 1, no two fly ashes are entirely similar with respect to their chemical and physical properties • As a consequence, their incorporation into cementitious binder systems can result in highly variable hydration related characteristics • It is important to understand estimate the properties of cement-fly ash binder systems for its field application • In addition, we can also estimate the amount and type of fly ash(s) to be added to the binder system to achieve some required properties using models that account for variable characteristics of the fly ashes Prasanth Tanikella - Purdue University 51

Phase 2 Initial Setting Time § Initial setting time experiments were performed according to ASTM C 191 § Set time value of a binary binder (fly ash + cement) – Average of duplicates § Manual vicat needle was used for the measurements § Water/binder ratio selected based on the consistency of the binder (ASTM C 187) Mixing process § Dry mixing of the powder by hand § Rest of the procedure, as mentioned in the standard Prasanth Tanikella - Purdue University 52

Phase 2 Calcium Hydroxide Content at 7 days - Results § At 7 days, the amount of calcium hydroxide in fly ash pastes is similar or higher than that in plain paste Prasanth Tanikella - Purdue University 53

Phase 2 Strength Activity Index at 3 days - Results § Most of the ashes show a lower strength at 3 days compared to plain cement paste Prasanth Tanikella - Purdue University 54

Phase 2 Strength Activity Index at 7 days - Results § § § Few of the Class C ashes show a higher strength at 7 days compared to plain cement paste (could be due to the inception of hydration in fly ashes) No pozzolanic reaction at this age Difference in the rates of reactions in the fly ashes can be clearly observed Prasanth Tanikella - Purdue University 55

Phase 2 Strength Activity Index at 28 days - Results SAI at 28 days using Cement 2 130 125 R 2 = 0. 8364 120 115 110 105 100 95 90 85 80 90. 0 § § 100. 0 110. 0 120. 0 130. 0 SAI at 28 days using Cement 1 140. 0 150. 0 Correlation between strength activity index at 28 days, measured with two different cements R 2 = 0. 84 Prasanth Tanikella - Purdue University 56

Phase 2 ANOVA Table (Class C Ashes) – Set Time Source DF Sum of Squares Mean Square F Value p-Value Model 3 3. 269 1. 089 1. 65 0. 2543 Error 8 5. 292 0. 6615 Total 11 8. 561 R 2 0. 3818 Adj - R 2 0. 15 Standard Error t-Value p-Value Parameter Variable DF Intercept 1 4. 456 4. 112 1. 08 0. 3101 sulfate 1 1. 178 0. 644 0. 183 0. 1048 alumina 1 -0. 085 0. 235 -0. 36 0. 7267 glass 1 -0. 583 0. 619 -0. 94 0. 3738 Estimate Prasanth Tanikella - Purdue University 57

Phase 2 ANOVA Table (Class F Ashes) – Set Time Source DF Sum of Squares Mean Square F Value p-Value Model 3 0. 44358 0. 14786 1. 63 0. 3487 Error 3 0. 27189 0. 09063 Total 6 0. 71547 R 2 0. 62 Adj - R 2 0. 24 Standard Error t-Value p-Value Parameter Variable DF Intercept 1 1. 26093 0. 99826 1. 26 0. 2958 sulfate 1 0. 46946 0. 25233 1. 86 0. 1598 alumina 1 0. 07325 0. 0769 0. 95 0. 4111 glass 1 -0. 0845 0. 53944 -0. 16 0. 8855 Estimate Prasanth Tanikella - Purdue University 58

Phase 2 ANOVA Table (Class C Ashes) – Set Time Source DF Sum of Squares Mean Square F Value p-Value Model 3 3. 269 1. 089 1. 65 0. 2543 Error 8 5. 292 0. 6615 Total 11 8. 561 R 2 0. 3818 Adj - R 2 0. 15 Standard Error t-Value p-Value Parameter Variable DF Intercept 1 4. 456 4. 112 1. 08 0. 3101 sulfate 1 1. 178 0. 644 0. 183 0. 1048 alumina 1 -0. 085 0. 235 -0. 36 0. 7267 glass 1 -0. 583 0. 619 -0. 94 0. 3738 Estimate Prasanth Tanikella - Purdue University 59

Phase 2 ANOVA Table (Class F Ashes) – Set Time Source DF Sum of Squares Mean Square F Value p-Value Model 3 0. 44358 0. 14786 1. 63 0. 3487 Error 3 0. 27189 0. 09063 Total 6 0. 71547 R 2 0. 62 Adj - R 2 0. 24 Standard Error t-Value p-Value Parameter Variable DF Intercept 1 1. 26093 0. 99826 1. 26 0. 2958 sulfate 1 0. 46946 0. 25233 1. 86 0. 1598 alumina 1 0. 07325 0. 0769 0. 95 0. 4111 glass 1 -0. 0845 0. 53944 -0. 16 0. 8855 Estimate Prasanth Tanikella - Purdue University 60

Phase 2 ANOVA Table (Class C Ashes) – (SAI) at 7 days Source DF Model 3 Error Total Sum of Mean Square F Value p-Value 873. 541 291. 180337 4. 017362 0. 0514 8 579. 8439 72. 480485 11 1453. 385 R 2 0. 601 Adj - R 2 0. 4514 Standard Error t-Value p-Value Squares Parameter Variable DF Intercept 1 -521. 432 308. 59493 -1. 6897 0. 1296 SAF 1 5. 86739 3. 05711 1. 91926 0. 0912 cao 1 9. 32163 4. 97522 1. 873612 0. 0979 glass 1 17. 94111 7. 88917 2. 274144 0. 0525 Estimate Prasanth Tanikella - Purdue University 61

Phase 2 ANOVA Table (Class F Ashes) – (SAI) at 7 days Source DF Sum of Squares Mean Square Model 3 229. 16838 76. 38946 Error 2 27. 53575 13. 76788 Total 5 256. 70413 R 2 0. 8927 Adj - R 2 0. 7318 Parameter Standard Estimate Error F Value p-Value 5. 55 0. 1565 t-Value p-Value Variable DF Intercept 1 -234. 5254 89. 77232 -2. 61245 0. 1206 SAF 1 3. 44957 0. 98377 3. 50648 0. 0726 cao 1 5. 08465 1. 32791 3. 829062 0. 0619 glass 1 -1. 29999 3. 55845 -0. 36532 0. 7499 Prasanth Tanikella - Purdue University 62

Phase 2 ANOVA Table – Heat of Hydration Model Peak. Heat Class C Class F Time. Peak Class C Class F Total. Heat Class C Class F p-value 0. 0484 0. 4564 0. 1722 0. 0698 0. 0562 0. 5101 R 2 0. 5662 0. 5346 0. 4103 0. 8778 0. 6461 0. 6999 Adj-R 2 0. 4216 0. 0685 0. 2138 0. 7556 0. 4692 0. 0998 Variables spsurface SAF Glass -0. 000291 -0. 168 0. 6817 0. 0084 0. 0172 0. 1447 spsurface SAF Glass -0. 0000264 -0. 09952 0. 6113 0. 1868 0. 2057 0. 2479 spsurface meansize SAF -0. 0301 -10. 1287 63. 35305 0. 0533 0. 1302 0. 1064 spsurface Meansize SAF -0. 0145 -12. 4914 13. 5769 0. 0597 0. 0656 0. 2062 meansize Carbon SAF 1. 42925 -37. 928 -3. 8206 0. 0151 0. 1296 0. 2259 meansize Carbon SAF -2. 2513 32. 8216 -4. 8452 0. 8169 0. 5248 0. 5438 Prasanth Tanikella - Purdue University Statistic cao -4. 764 0. 3643 cao -3. 2055 0. 7719 Coefficient p-value Coefficient p-value 63

Phase 2 ANOVA Table – Calcium Hydroxide Content Model CH 1 Day Class C Class F CH 7 Days Class C p-value R 2 Adj-R 2 Variables 0. 0565 0. 5909 0. 4375 blaines carbon alumina 0. 0001918 0. 07257 0. 02428 0. 0109 0. 8861 0. 7195 0. 0458 0. 9692 0. 9231 blaines carbon alumina -0. 000006 -0. 41087 -0. 0276 0. 8998 0. 017 0. 0484 0. 0171 0. 7013 0. 5893 blaines cao glass 0. 000324 0. 05958 1. 03531 0. 0268 0. 3755 0. 0111 CH 28 Days Class C 0. 0135 Coefficient p-value 0. 0005233 0. 0000418 0. 38429 0. 0021 0. 4252 0. 1818 Class F 0. 1602 0. 8901 0. 7253 blaines spsurface sulfate Coefficient p-value blaines Coefficient p-value 0. 6136 spsurface sulfate 0. 719 Statistic -0. 0002886 0. 0001469 0. 29562 0. 227 0. 093 0. 1728 Prasanth Tanikella - Purdue University 64

Phase 2 ANOVA Table – Non-evaporable Water (Wn) Model p-value R 2 Adj-R 2 Variables Wn 1 Day Class C 0. 0684 0. 5694 0. 4079 blaines carbon alumina Statistic 0. 000132 -0. 64483 -0. 01391 Coefficient 0. 0372 0. 1874 0. 8209 p-value Wn 3 Days Class C 0. 0333 0. 6443 0. 511 Class F 0. 1576 0. 892 0. 7299 Wn 28 Days Class C 0. 1618 0. 719 0. 6136 sulfate 0. 32747 0. 0135 sulfate SAF mgo Statistic -0. 01468 0. 04352 Coefficient 0. 4483 0. 7228 p-value SAF mgo Statistic 1. 04884 0. 0693 blaines 0. 00022 0. 0531 0. 02403 -0. 32493 Coefficient 0. 3505 0. 1065 p-value carbon alumina Statistic 0. 63946 0. 00992 Coefficient 0. 4538 0. 929 p-value Prasanth Tanikella - Purdue University 65

Phase 3 Test for Additivity – Set Time (minutes) Exp. No Model 1 Model 2 Model 3 Observed Predicted 1 120 157. 4 157. 5 128 2 155 150. 2 162. 4 210. 8 3 160 156. 8 168. 9 210. 2 4 195 151. 5 153. 8 147. 2 5 125 152. 9 165. 2 26 6 230 173. 6 168. 3 177. 9 7 170 144. 7 152. 4 119. 2 8 225 151. 6 129. 3 9 190 153. 2 161 122. 7 • Observed set time was not the same as predicted set time for most of the cases Prasanth Tanikella - Purdue University 66