Recycled Materials and Byproducts in Highway Applications—Manufacturing and Construction Byproducts, Volume 8 (2013)

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.

1 Background Cement kiln dust (CKD) is generated during the production of the cement clinker and is a dust particulate mixture of partially calcined and unreacted raw feed, clinker dust, and ash, enriched with alkali sulfates, halides, and other volatiles (Adaska and Taubert 2008). According to EPA (2010), the definition of cement kiln dust is: “a fine-grained, solid, highly alkaline material removed from the cement kiln exhaust gases by scrubbers (filtration baghouses and/or electrostatic precip- itators). The composition of CKD varies by plants and over time at a single plant. Much of the material comprising CKD- reacted raw material, including a raw mix at various stages of burning, and particles of clinker.” Cement is produced using a rotary kiln to turn raw materials (limestone, clay, iron ore, and silica) into a sintered product referred to as a clinker. Gypsum is added at the end of the pro- cess to manage the rate of hydration. A rotary kiln is funda- mentally a long, slowly rotating cylinder tilted at a slight angle with the burner at the lower end. The raw materials enter the top end of the cylinder, are heated, then exit and cool. The sintered material at the end is referred to as “clinkers.” Kilns were first introduced in the 1890s and became popu- lar in the first part of the 1900s as improvements were made to provide continuous production and a more consistent final product in larger quantities (“Understanding Cement” 2010). There are three main types of kilns: • Long-wet kiln • Long-dry kiln • Precalciner kiln. The original kiln style, the long-wet kiln, feeds the raw material in as slurry and the length of the cylinder can be up to 656 ft long and 20 ft in diameter (Figure 1). The length is required because the material needs sufficient time to dry out the slurry water, which until recently was difficult to blend and add dry (“Understanding Cement” 2010). Once in the kiln, the materials are calcined then sintered to form the clinker. Some of these kilns are still in use. Newer dry kiln configurations add the dry, blended raw materials after passing through a pre-heating tower, using heat from recycling hot kiln gases (Figure 2). The heat exchange is accomplished by feeding the finely ground raw material, called raw meal, into the top of the preheater tower, then pass- ing through a series of cyclones in the tower through which the hot gases are circulated (“Understanding Cement” 2010). The high surface area and small particle size provide efficient heat transfer and approximately 30% to 40% of the decarbon- ation of the raw meal before it enters the kiln. Because the material enters preheated, the length and the diameter of the cylinder can be smaller but still produce the same quantity of clinker per hour. The precalciner kiln, the newest technology, is similar in concept to the dry kiln, but with the addition of a second burner, or precalciner (Figure 3). With the additional heat, about 85% to 95% of the material is decarbonated before entering the kiln (“Understanding Cement” 2010). The particulates for all types of the cement kilns are cap- tured from the exhaust gases using air pollution control devices such as cyclones, baghouses, and electrostatic precipitators (Adaska and Taubert 2008). The particles captured in this pro- cess are the CKD. Many cement plants recycle the CKD back into the kiln to optimize the process, save a small quantity of virgin raw materials, and avoid landfilling costs. The CKD is not reused by the plant when there are equipment limitations for handling the dust or it would make the cement product noncompliant with specifications. Physical and chemical ProPerties Different types of cement kilns generate CKD materials with different physical and chemical properties (Adaska and Taubert 2008). Long-wet and long-dry kiln CKD is typically partially calcined kiln feed fines enriched with alkali sulfates and chlorides. Alkali by-pass with precalciner kilns produce CKD that is more calcined with a coarser size and concen- trated with alkali volatiles (Table 1). These CKD byproducts also have the highest amount by weight of calcium oxide and the lowest loss on ignition (LOI) (Table 2). The type of burner fuel will also influence the CKD com- position. Gas- or oil-fired kilns contain higher proportions of soluble K2O compared with coal-fired kilns. The amounts of trace metals are not significant (Adaska and Taubert 2008; EPA 2010). Previous research by PCA (1992) reported that concentrations for eight Resource Conservation and Recovery Act (RCRA) metals (arsenic, barium, cadmium, chromium, lead, mercury, selenium, and silver) were well below the regu- lations at that time. chapter one cement kiln dust

2 Fresh cement kiln dust Properties Recent and extensive data on CKD chemistry was reported by Williams (2005) for a project that included an in-depth analysis of cement producers scattered throughout the United States and Canada. Approximately 18 companies and their subsidiaries, representing 100 cement plants, were identi- fied and contacted. Of these plants, 12 companies provided the chemical composition of CKD for a total of 32 plants (Table 3). The between-plant coefficient of variation (CV) was under 10% for only the SiO3 and Al2O3. The CV for the CaO, Fe2O3, and TiO2 was between 22% and 32% and ranged from 47% up to 143% for the other compounds. The loss on ignition CV was 35%. The coefficient of variations highlight the wide range of chemical and LOI properties between cement plants and the importance of evaluating the chemical properties of the CKD to be used on each project. By con- trast, the variability over time within one plant (Table 4) was When trace metals are present, they include antimony, barium, lead, manganese, strontium, thallium, and zinc (EPA 2008). Minor trace metals include beryllium, copper, hexava- lent chromium, mercury, nickel, silver, and thallium. Concen- trations of trace metals vary widely between sources of CKD. The trace metals are low enough to not be a concern. The EPA (2008) developed a materials characterization paper in sup- port of the advanced notice of proposed rulemaking identify- ing CKD as a nonhazardous material that is considered solid waste as long as stockpiles are managed properly. Traditionally, those agencies using CKD in highway applications focused on using byproducts from cement plants shortly after production. However, there are significant amounts of CKD that have been landfilled over the years. There is some movement to attempt to use these weathered stockpiles of CKD in highway applications to minimize the demands on raw materials. Fresh CKD and landfilled CKD can be expected to have different properties owing to envi- ronmental exposure. FIGURE 1 Long-wet kiln (after “Understanding Cement” 2010). Slurry Wet Kiln Burner Clinker Drying Calcining Sintering Wet Kiln Exhaust Gases Blended Dry Feed Kiln Burner Clinker Preheater Dry Kiln Exhaust Gases FIGURE 2 Dry kiln (after “Understanding Cement” 2010). Kiln Burner Clinker Preheating Towers Electrostatic Precipitators Cooling Precalciner Kiln Dry Feed FIGURE 3 Precalciner kiln (after “Understanding Cement” 2010). Particle Size, mm % by Weight Long-wet kiln Long-dry kiln Alkali by -pass from preheater/precalciner >0.1 5.0 0 2.0 <0.045 85.0 99.2 84.5 <0.003 77.3 98.8 66.0 <0.007 43.0 87.2 14.0 <0.001 12.0 12.0 3.0 <0.0006 7.5 5.6 2.0 Median Size, mm 9.4 3.0 2.2 After Todres et al. (1992); Adaska and Taubert (2008). TABLE 1 ExAMPLES OF KILN TyPE INFLUENCE ON PARTICLE SIzE

3 Constituent % by Weight Long-wet kiln Long-dry kiln Alkali by -pass from preheater/precalciner Typical type I ce me nt Si O 2 15.02 9.694 15.23 20.5 Al 2 O 3 3.8 3.39 3.07 5.4 Fe 2 O 3 1.88 1.1 25 2.6 CaO 41.01 44.91 61.28 63.9 MgO 1.47 1.29 2.13 2.1 SO 3 6.27 6.74 8.67 3 Na 2 SO 4 0.74 0.27 0.34 <1 K 2 O 2.57 2.4 2.51 <1 Loss on Ignition 25.78 30.24 4.48 0–3 Free Li me (CaO) 0.85 0.52 27.18 <2 Todres et al. (1992); Adaska and Taubert (2008). TABLE 2 ExAMPLES OF COMPOSITION OF CKD FROM DIFFERENT OPERATION SOURCES TABLE 3 CHEMICAL COMPOSITION By CEMENT KILN PLANT Cement Producer Plant Sample Location Percent by Weight (%) CaO SiO2 Al2O3 Fe2O3 MgO SO3 Na2O K2O TiO2 P2O5 Mn2O3 SrO LOI A 1 — 62.04 15.61 4.89 2.00 1.45 8.04 0.45 4.85 NR NR NR NR 25.52 2 — 48.21 15.88 4.61 1.87 0.71 3.60 0.58 11.92 0.36 0.08 0.05 0.04 9.77 3 52.19 15.75 3.90 2.06 3.16 11.44 0.34 5.65 NR NR 0.03 NR 1.78 4 1 54.11 14.53 5.30 2.10 2.08 12.70 3.28 NR 0.19 0.15 NR NR 14.71 2 56.29 14.41 5.49 2.26 2.13 12.99 4.03 NR 0.22 0.10 NR NR 13.59 3 48.21 13.46 4.96 2.18 2.43 12.77 3.78 NR 0.20 0.13 NR NR 13.53 4 57.17 13.24 5.15 2.13 2.37 13.06 4.46 NR 0.18 0.14 NR NR 12.70 5 54.56 13.29 4.95 2.13 1.98 14.27 6.90 NR 0.19 0.12 NR NR 15.88 combined 55.55 14.50 5.53 2.21 2.36 10.65 5.45 NR 0.31 0.16 NR NR 17.62 cond tower 56.90 12.00 4.38 2.00 2.33 10.60 3.21 NR 0.25 0.12 NR NR 15.30 recovery 51.20 12.10 4.69 1.82 1.82 9.72 2.84 NR 0.22 0.15 NR NR NR waste 53.70 11.30 4.01 1.78 2.12 11.00 2.64 NR 0.20 0.12 NR NR NR B 1 — 62.09 17.62 4.90 2.58 1.93 5.79 0.56 3.76 NR NR NR NR 4.94 2 — 49.00 13.00 4.04 2.02 0.67 8.10 0.26 3.51 NR NR NR NR 18.10 3 — 44.00 10.50 4.03 1.92 0.69 7.30 1.19 8.70 NR NR NR NR 19.84 4 — 41.93 12.66 3.17 0.78 1.75 6.15 0.40 9.71 NR NR NR NR 23.44 5 — 51.39 16.33 3.69 1.78 2.51 12.27 0.56 11.47 NR NR NR NR NR C 1 — 47.75 15.16 4.56 1.84 2.09 10.15 0.99 9.81 NR NR NR NR 1.77 D 1 — 39.41 16.57 3.51 2.22 2.42 2.35 0.09 0.71 0.18 0.00 0.04 0.02 32.16 E 1 — 48.05 14.56 5.59 3.31 0.78 6.06 0.32 1.61 0.28 0.13 NR NR 22.96 F 1 — 48.16 13.15 5.69 1.07 2.36 2.64 0.45 1.74 0.07 0.21 NR NR NR G 1 waste 44.64 11.48 3.72 1.47 3.15 5.14 0.44 2.03 0.19 0.04 NR NR 25.78 return 45.02 12.91 3.56 1.61 3.42 5.14 0.42 1.61 0.19 0.05 NR NR 23.68 H 1 — 0.00 15.55 5.39 0.40 0.00 5.83 0.01 1.35 NR NR NR NR 18.52 2 — 65.55 13.90 4.95 1.92 1.37 5.26 0.39 6.42 0.22 0.09 0.02 0.05 35.41 3 — 38.10 17.50 3.60 0.23 4.40 5.20 0.50 8.40 NR NR NR NR 22.00 I 1 — 43.20 14.04 3.16 2.08 1.42 2.36 0.09 0.95 0.23 0.23 NR NR 33.19 J 1 — 49.62 16.62 3.93 2.32 2.30 6.30 0.37 2.45 0.21 0.21 0.11 0.03 15.26 2 — 51.67 14.75 3.31 1.77 2.30 7.24 0.39 2.74 0.16 0.07 0.03 0.04 15.22 3 — 40.18 14.54 3.85 1.69 2.47 5.78 0.70 4.20 0.19 0.05 0.04 0.03 25.74 4 — 59.77 16.97 5.20 2.26 0.92 8.64 0.44 3.88 0.24 0.20 0.22 0.17 1.08 5 — 52.08 15.60 3.82 2.14 0.97 6.85 0.25 2.87 0.15 0.15 0.17 0.16 14.56 6 — 49.05 15.55 3.00 1.97 1.34 4.53 0.31 3.57 0.16 0.04 0.06 0.05 20.22 7 — 44.92 11.85 4.12 2.27 0.58 10.24 1.02 6.07 0.46 0.18 0.03 0.16 15.35 8 — 43.43 15.04 3.52 1.67 2.59 6.84 0.25 3.89 0.21 0.15 0.05 0.02 20.75 9 — 43.74 13.50 4.39 2.42 0.82 7.04 0.43 3.90 0.23 0.16 0.02 0.06 21.09 K 1 — 57.10 18.70 5.40 2.90 0.77 8.10 0.18 3.46 0.28 NR NR NR 19.50 2 baghouse 1 73.16 14.94 5.92 3.32 1.29 0.39 0.02 0.48 0.25 NR NR NR NR baghouse 2 72.98 15.14 5.91 3.30 1.29 0.40 0.00 0.48 0.25 NR NR NR NR kiln feed 67.41 20.67 5.62 3.68 1.29 0.20 0.04 0.57 0.30 NR NR NR NR 3 — 50.96 18.23 5.30 2.47 1.95 6.16 0.43 8.94 0.21 0.12 NR NR 16.01 L 1 min 50.00 11.20 5.00 1.50 0.70 3.70 0.20 3.00 NR NR NR NR 16.00 max 59.00 16.00 6.00 3.00 1.00 4.80 0.30 4.80 NR NR NR NR 22.00 2 — 60.00 17.00 5.00 2.80 1.09 7.20 0.18 3.30 0.38 0.22 NR 0.06 1.60 Statistics No. Plants 44 44 44 44 44 44 44 35 32 28 13 13 37 Average 50.99 14.71 4.56 2.07 1.76 7.16 1.14 4.37 0.23 0.13 0.07 0.07 17.47 Std Dev. 11.41 2.18 0.86 0.69 0.89 3.62 1.63 3.22 0.07 0.06 0.06 0.06 8.33 CV 22.4% 14.8% 18.9% 33.4% 50.6% 50.6% 143.2% 73.7 % 32.0% 46.9% 93.0% 81.3% 47.7% Min. 0.00 10.50 3.00 0.23 0.00 0.20 0.00 0.48 0.07 0.00 0.02 0.02 1.08 Max. 73.16 20.67 6.00 3.68 4.40 14.27 6.90 11.92 0.46 0.23 0.22 0.17 35.41 After Williams (2005). CV = coefficient of variation; NR = not reported.

4 less than 10% for seven compounds, and all but two were below 27%. There was also reasonably low variation between properties initially reported by the plant, those reported by the plant at the time of sampling, and those found when tested by an independent third party (Table 5). landfilled cement kiln dust Properties Sreekrishnavilasam et al. (2006) and Sreekrishnavilasam and Santagata (2006) evaluated the properties of landfilled CKD for a stockpile that had been generated over 12 years at one cement plant. Older CKD byproducts were located in the lower depths of the landfill. Three borings were used to evaluate the water content, LOI, free lime content, and pH at various depths in the landfill (0 to 78 ft), representing various ages of the stockpile. The first two borings (B1 and B2) were obtained between 0 and 50 ft located on the top terrain of the landfill. The third boring was taken from the lower terrain and repre- sented properties from about 15 ft to 75 ft in depth. Testing of the weathered CKD showed that the water contents in the stockpile ranged from 0% to about 65%, and the LOI ranged between 30% and 37%. The LOI values were at the high end of those reported in previous studies of fresh CKD and suggested that the reactivity of the landfilled CKD should be limited. The pH varied between 10.7 and 12.8 (Table 6). The chemical evaluation by x-ray defraction (xRD) showed a number of peaks for quartz and calcite for all of the samples. The landfilled CKD also showed a number of ettringite peaks along with similar calcite and quartz peaks. The ettringite was the result of the hydration reactions over time in the landfill. The free lime of the fresh CKD from this plant showed a low free lime content; therefore, the low free lime content in the landfill CKD was not unexpected. Particle size distribution analysis was conducted using a hydrometer test in water with sodium hexamethaphosphate as a dispersing agent. The landfilled CKD had a slightly higher mean particle size than the fresh CKD, which was attributed to the expansive reactions (e.g., ettringite) and change in mor- phology as seen in the scanning electron microscopy (SEM) photographs. Results indicated the properties of the landfilled Month %,thgieWybtnecreP CaO SiO2 Al2O3 Fe2O3 MgO SO3 Na2O K2O TiO2 P2O5 SrO Cl LOI 1/04 67.11 16.46 5.63 1.69 1.56 3.37 0.39 3.62 0.25 0.08 0.06 0.02 0.00 34.68 2/04 66.96 15.80 5.60 1.56 1.56 3.39 0.58 3.77 0.24 0.09 0.05 0.02 0.00 34.98 3/04 66.99 15.86 5.54 1.79 1.66 3.79 0.30 3.66 0.24 0.09 0.05 0.03 0.00 34.97 4/04 69.27 16.18 5.79 1.66 1.66 3.20 0.18 2.29 0.26 0.09 0.06 0.02 0.00 34.35 5/04 69.63 16.72 5.70 1.95 1.58 3.14 0.21 2.14 0.26 0.10 0.05 0.02 0.00 34.00 6/04 71.50 16.78 5.60 2.03 1.58 2.51 0.16 1.78 0.25 0.09 0.07 0.02 0.00 34.09 7/04 67.85 15.16 5.15 2.07 1.41 4.05 0.23 4.06 0.23 0.09 0.06 0.02 0.00 35.99 8/04 65.77 13.76 5.00 1.92 1.39 5.38 0.30 6.23 0.21 0.09 0.06 0.02 0.00 35.17 9/04 66.46 13.86 4.98 1.96 1.34 5.32 0.27 5.91 0.23 0.10 0.05 0.02 0.00 35.41 10/04 66.57 14.32 5.14 1.98 1.40 4.52 0.25 5.34 0.23 0.09 0.05 0.02 0.00 35.41 11/04 65.17 14.03 4.96 1.93 1.37 5.19 0.62 6.65 0.22 0.09 0.04 0.02 0.00 NR 12/04 64.00 13.39 4.71 1.79 1.37 6.01 0.42 7.81 0.21 0.09 0.04 0.02 0.00 NR Average 67.27 15.19 5.32 1.86 1.49 4.16 0.33 4.44 0.24 0.09 0.05 0.02 0.00 34.91 Std. Dev. 2.06 1.26 0.36 0.16 0.12 1.11 0.15 1.94 0.02 0.01 0.01 0.00 0.00 0.63 CV 3.1% 8.3% 6.8% 8.6% 8.1% 26.7% 45.9% 43.6% 7.3% 5.7% 16.6% 13.9% — 1.8% After Williams (2005). CV = coefficient of variation. Mn2O3 TABLE 4 MONTHLy CHANGE IN CKD CHEMICAL COMPOSITION IN PLANT H2

ID Sample Date Percent by Total Weight, % CaO Si O 2 Al 2 O 3 Fe 2 O 3 MgO SO 3 Na 2 O K 2 O Ti O 2 P 2 O 5 SrO Mn 2 O 3 LOI LH Quoted 9/04 41.93 12.66 3.17 0.78 1.75 6.15 0.4 9.71 NR NR NR NR 23.44 Received 2/05 38.56 15.46 1.77 1.25 1.3 4.23 0.27 7.62 NR NR NR NR 28.24 Tested 7/05 39.01 12.58 4.45 2.28 1.7 7.46 0.29 10.96 0.17 0.08 0.06 NR 25.9 HH Quoted 12/04 65.55 13.9 4.95 1.92 1.37 5.26 0.39 6.42 0.22 0.09 0.02 0.05 35.41 Received 2/05 70.25 14.75 5.14 1.89 1.47 2.64 0.17 2.51 0.23 0.08 0.02 0.04 35.26 Tested 7/05 51.94 10.62 3.94 2.68 1.02 2.26 0 1.87 0.19 0.03 2 NR 35.28 HL Quoted 9/04 62.09 17.62 4.9 2.58 1.93 5.79 0.56 3.76 NR NR NR NR 4.94 Received 3/05 51.94 13.27 4.04 1.97 1.55 3.79 0.57 6.13 0.28 0.09 0.11 NR 9.87 Tested 7/05 44.64 13.35 3.83 2.57 2.44 3.64 0.8 5.46 0.19 0.05 0 NR 8.07 LL Quoted 5/04 47.75 15.16 4.56 1.84 2.09 10.15 0.99 9.81 NR NR NR NR 1.77 Received 4/05 47.47 14.11 4.3 1.82 1.96 13.35 0.94 9.74 NR NR NR NR NR Tested 7/05 38.89 13.43 3.86 2.25 1.99 13.85 1.36 8.77 0.15 0.11 0.04 0 1.4 After Williams (2005). First letter represents high or low value of CaO. Second letter represents level high or low of LOI. Quoted: results reported by the plant during the survey. Received: results reported by the plant at time of sampling. Tested: results reported by an independent laboratory on the as-received CKD. NR = not reported. TABLE 5 COMPARISON OF CHEMICAL COMPOSITION OF CKD USED IN TESTING FOR VALUES QUOTED By THE PLANT, SAMPLING, AND WHEN USED

6 CKD were relatively consistent throughout the 12 years of the operation, although there were noticeable differences in the composition as a result of hydration over time. usage and Production As of 2006, fewer than 20 of the 118 cement plants in the United States managed 90% of the CKD disposed on-site (Adaska and Tauber 2008; PCA 2008). The CKD produced, beneficially reused, and landfilled has changed significantly over the last two decades (Table 7); the practice of CKD landfilling is being phased out as more plants are retrofitted to allow in-processing recycling. The peak amount sent to landfills was around 1995. By 2006, less than half of the amount of clinker-produced CKD was going to landfills. The quantity of CKD used in beneficial reuse applications was starting to increase each year. Improvements in the reduction of CKD produced during the clinker production Sam ple Fresh CKD Fresh CKD Boring Samples from CKD Stockpile Stockpiled Statistics B2-C B1-F B2-F B3-B B2-I B3-D Mean Std. Dev. Approxim ate Depth in Boring, ft — — 17 31 31 41 51 60 — — Water Content, % — — 2 37 21 24 5 2 — — LOI, % — — 33 34 32.7 — 34.3 33 — — pH — — 11.6 12.1 12.5 — 11.8 12.7 — — Composition, % by weight CaO 50.4 45.93 53.19 42.96 46.3 42.14 44.59 44.54 46.15 3.98 Si O 2 NA 9.30 8.70 7.62 7.82 7.10 7.99 12.37 8.80 1.91 Al 2 O 3 2.66 3.20 2.87 2.50 2.66 2.43 2.56 2.82 2.67 0.18 Fe 2 O 3 1.09 1.06 1.11 0.96 1.05 1.00 1.11 1.57 1.17 0.22 MgO 0.7 1.11 1.02 0.83 0.88 0.99 0.91 1.93 1.14 0.41 SO 3 3.50 2.30 4.92 4.62 3.76 4.17 4.12 2.59 3.91 0.81 Na 2 O 0.18 0.13 0.23 0.30 0.12 0.23 0.08 0.11 0.15 0.09 K 2 O 2.16 1.22 2.39 2.14 1.43 2.32 1.39 1.19 1.74 0.53 LOI 33.62 33.3 33 33.86 33.64 34.1 34.8 33.16 33.74 0.65 Total Alkali 1.6 0.93 1.80 1.71 1.06 1.76 0.99 0.89 1.34 0.41 Total Reactive Oxides (sodiu m equivalent) 15.14 12.36 18.59 7.49 11.99 6.49 9.24 12.00 11.66 4.36 After Sreekrishanavilasam et al. (2006). Oxide values expressed in percentage by mass; mean and standard deviation refer to tube samples alone . — = indicates no data. LOI = loss on ignition; NA = not available. TABLE 6 OxIDE COMPOSITION OF FRESH AND LANDFILLED CKD Year Plants Responding to Survey CKD Beneficially Reused On or Off Site CK D Sent to Landfills CK D Reclaim ed from Landfills Annual Clinker Production CKD Sent to a Landfill/Clinker Produced, 1,000 metr ic tons kg/ metr ic ton 1990 84 752 2,656 — 44,360 60 1995 94 651 3,147 — 61,729 51 1998 95 769 2,500 13 67,105 37 2000 92 575 2,223 79 68,263 33 2001 102 925 2,329 231 75,683 31 2002 101 665 1,990 103 77,637 26 2003 102 718 1,995 116 79,357 25 2004 102 918 1,993 69 83,945 24 2005 102 988 1,429 205 85,568 17 2006 101 1,160 1,403 361 86,687 16 After Adaska and Taubert (2008). TABLE 7 HISTORICAL CKD PRODUCTION AND USE

7 can be seen in Figure 4. The clinker production quantity increased significantly from 1990 to 2006; however, the amount of CKD landfilled significantly decreased over the same time. Although beneficial reuses accounted for some of the decrease, improvements in cement production technol- ogies also appeared to have a significant impact. In 2006, ten states represented 76% of the beneficially reused CKD in the United States; the states were Oklahoma, Texas, Pennsylvania, Ohio, Illinois, Indiana, California, Arkansas, Maryland, and Missouri. There were a range of beneficial reuse applications for CKD (Table 8). In some areas, the previously landfilled CKD was also being used, although at much lower levels. The U.S. cement industry has adopted a year 2020 vol- untary target for a 60% reduction from the 1990 baseline for the amount of CKD landfilled per ton of clinker produced (PCA 2008). literature review This section summarizes information obtained from the lit- erature and is organized as follows: • CKD regulatory history • Applications—bound • Applications—unbound • Environmental issues. The objective of this section is to provide the reader with a brief background on the evolution of the regulatory actions, currently researched applications for this byproduct, and a sampling of currently reported material properties and appli- cation product characteristics. After Adaska and Taubert (2008). Application Production Byproduct Recovered from Landfill Total Used in Application 1,000 metr ic tons Potential Use in Highway Applications Cement Additive/Blending 202 2 204 Soil/Clay Stabilization/Consolidation 588 23 611 Pave me nt Manufacturing 13 2 15 Concrete Products 0 0 0 Used in Structures or Highways 803 27 830 Returned to Kiln 0 126 126 Nonhighway Application Uses Wastewater Neutralization/Stabilization 13 3 16 Waste Stabilization/Soli dification 235 101 336 Mine Reclamation 168 0 168 Agricultural Soil Am endm ent 37 31 68 Sanitary Landfill Liner/Cover Material 17 0 17 Beneficial use Not Provided 4 0 4 Other Uses 2,080 315 2,395 TABLE 8 BENEFICIAL USES OF CKD FIGURE 4 Improvements in technology result in more clinker production and less CKD sent to landfills (after Adaska and Taubert 2008). 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 0 20,000 40,000 60,000 80,000 100,000 120,000 1985 1990 1995 2000 2005 2010 Q ua nti ty o f C KD L an dfi lle d, 1 ,0 00 to ns Q ua nti ty o f C lin ke r Pr od uc ed , 1 ,0 00 to n Year Annual Clinker Production CKD Sent to Landfills

8 cement kiln dust regulatory history Adaska and Taubert (2008) provided the history of legislation in the United States for the regulation of CKD byproducts. • In 1976, the RCRA required EPA to develop regulations governing the identification and management of hazard- ous wastes. • In 1980, the Solid Waste Disposal Act amendments amended RCRA (referred to as the Bevill Amendment) and exempted three special wastes including CKD from hazardous waste regulation until further study could be completed. This legislation required that EPA sub- mit a report to Congress evaluating the status of CKD management and potential risk to human health and the environment. • In 1993, the EPA Report on CKD was submitted to Con- gress. The conclusions from the report were that CKD posed little risk to human health and the environment. • In 1995, the EPA issued a final regulatory determination for CKD that recommended a more tailored set of stan- dards be developed for managing CKD to minimize any risk resulting from the mishandling of the byproduct. • In 1999, the EPA published Standards for the Manage- ment of Cement Kiln Dust; Proposed Rule, which clas- sified CKD as nonhazardous as long as specific manage- ment standards are met. If not, then it is to be classified as a “listed waste” and would need to comply with RCRA Subtitle C management standards. • In 1999, the American Portland Cement Association submitted formal comments opposing the use of federal authorities for CKD management. • In 2000, the EPA elected to retain the Bevill exclusion. • In 2001, the American Portland Cement Association filed a petition requesting that EPA withdraw the CKD proposed rule and reinstate the Bevill status for CKD. • In 2002, the EPA published a notice of data availability that explained they were considering a new approach to CKD management whereby it would finalize the pro- posed CKD management standards as RCRA Subtitle D (solid waste) rule and temporarily suspend the proposed RCRA Subtitle C (hazardous waste) portion rule for 3 to 5 years to assess how CKD management practices and state regulatory programs evolve. • In 2010, the EPA and industry were working together to resolve CKD issues. applications—Bound Clinkers Adaska and Taubert (2008) noted that some cement produc- ers have started to remove CKD from their on-site historical landfills to augment the raw materials used during cement production. Blended Cements Shah and Wang (2004) investigated the influence of differ- ent combinations of CKD and fly ash, grinding equipment and methods, chemical additions, and elevated curing tempera- tures on blended cements. Three activation methods were used for accelerating CKD-FA binder hydration: chemical (2% and 5% by weight of binder of NaOH), thermal (curing tempera- tures of 38% and 50%), and mechanical (simple blending, ball and mill grinding, grinding aid, and high-speed mixing). Test- ing included xRD and thermal gravimetric analysis (TGA) for chemical evaluations, and compressive strength for appli- cation properties. Results showed a combination of NaOH addition and elevated curing temperature was not beneficial to strength development. The blended cements and these activa- tion combinations resulted in a loss of strength. The elevated curing temperature was much more effective than NaOH on improving activation. An optimum combination of CKD-FA ratio, 2% NaOH, and a curing temperature of 100°F produced a blended cement strength that was comparable to that of ordinary portland cement. Ryou (2004) evaluated three different methods of mechan- ically improving reactivity (ball mill, vibratory mill, and attrition mill) of CKD (65%) and Class F fly ash (35%). The ball mill was a hollow cylinder that could be rotated on the mill rollers at different speeds. A tumbling media was used to impact the material as the cylinder rotates. The vibratory mill grinding was also a cylindrical container that held grind- ing media, but shook the materials in a horizontal or vertical direction at a high rate of speed. The impact energy varied with the amplitude of the movement. Typical settings were a rate of 1,200 rpm and amplitude of 0.5 in. The attrition mill imparted higher energy than either of the other two methods. A central shaft with arms continually stirred the particles and spherical media to provide a grinding action. Intense rolling and in-line impacts were produced by the differential velocity of the media moving around the agitator arms into the cavity behind the trailing edge. Results showed differences between types of grinding. As expected the no-grinding option showed the lowest levels of reactivity and compressive strength (Table 9). The ball grind- ing at either 4 or 8 h only slightly improved the reactivity. The attrition mill showed only a slight improvement in properties, similar to the 4 h in the ball mill. Four or eight hours of vibra- tory grinding produced the best improvement in reactivity. Materials from this method of grinding had the smallest mean diameter particle size (about 0.003 mm). Controlled Low Strength Materials Al-Harthy et al. (2004) investigated the use of CKD in the pro- duction of flowable fill [i.e., controlled low strength material (CLSM)] in Oman using water from regional oil production

9 activities. Water is considered a valuable commodity in Oman and using contaminated water in construction projects reduces the demand for fresh water. Flowable fill was produced using several sources of water with different properties (Table 10). The results showed that oily production water resulted in lower compressive strengths than when using ground water. How- ever, the 28-day compressive strengths were still above the minimum requirement of 2 to 24 psi for flowable fill. No dif- ferences were seen in slumps with the different water sources. Pierce et al. (2003) and Williams (2005) investigated the flow consistency and setting time of CLSM materials with CKD. Both the CaO and LOI values were used to select four CKDs to be used in the laboratory study (Table 11). The low CaO content CKD range was about 39% and the high con- tent CaO was between 44% and 52%. An LOI value of less than 8.1% was considered low and values greater than 25% were labeled as high. Particle surface area and particle size range parameters are shown in Table 12. The high LOI CKD samples generally had higher surface areas and smaller par- ticle sizes than the low LOI CKDs. Initial Vicat set times of the paste were approximately 480 min for the LH and HL (the high and low values of CaO), Grinding Method Hours of Grinding Cubes Max. Rate of Heat Evolution, kJ/kg-hr Time to Max. Rate, hour Percent Finer Than 0.0005 mm, % Mean Particle Size, mm Compressive strengths, psi 3 days 7 days 28 days Grinding (ball mill) 0 4 8 33 2.9 18.80 1.57 0.12 4 20 52 77 3.8 14.70 1.84 0.09 8 14 47 55 3.7 14.60 2.96 0.05 Vibratory Mill 4 33 68 119 7.7 14.50 5.10 0.00 8 46 78 130 6.6 16.60 5.16 0..3 Attrition Mill 4 9 28 68 3.0 13.20 2.72 0.09 8 9 30 68 3.2 15.40 3.25 0.08 After Ryou (2004). 65% CKD + 35% fly ash + water (w/cm = 0.5) Blends were mixed before grinding but without water. TABLE 9 INFLUENCE OF CKD PARTICLE SIzE ON COMPRESSIVE STRENGTH Source of Sample Time of Sample Collection Parameter Concentration pH TDS (mg/L) Chloride (mg/L) Hardness (mg/L) Alkalinity (mg/L) Sulfate (mg/L) Tap Water 2001 8.6 278 75 94 58 278 2002 8.3 398 86 182 114 65 Bahja Ground Water 2001 6.7 8,770 5,100 670 55 8 2002 8.2 233 93 17 16 5 Bahja Production Water 2001 7.4 66,300 44,500 13,000 59 281 2002 7.3 9,720 4,790 1,320 94 662 Rima Ground Water 2001 7.9 10,960 5,420 1,730 134 826 2002 7.4 9,850 4,820 2,250 169 759 Rima Production Water 2001 8.0 11,540 5,850 880 240 323 2002 8.5 586 223 13 72 5 Marmul Ground Water 2001 8.0 1,360 331 558 100 281 2002 8.0 1,540 383 588 147 548 Marmul Production Water 2001 7.3 4,900 2,040 166 606 233 2002 8.3 4,220 2,080 146 80 <1 Nimr Ground Water 2001 7.6 7,080 3,080 1,680 209 982 2002 7.8 7,050 3,160 1,670 219 782 Nimr Production Water 2001 7.3 423 4,000 490 399 330 2002 7.9 8,200 138 23 95 15 After Al-Harthy et al. (2004). TDS = total dissolved solids. TABLE 10 PROPERTIES OF WATER USED TO PRODUCE CKD CLSM MIxTURES

10 and between 780 and 1,380 for the HH and LL pastes (the high and low level of LOI). Ordinary portland cement usually achieved initial set in about 120 min by comparison. Final Vicat set times were between 1,200 and 1,680 min for the LH, HH, and HL pastes and greater than 1,680 for the LL paste. The final set time for ordinary portland cement was about 240 min. Hydration temperatures of the pastes were 74°F and 128°F for the LH and HL pastes, and 74°F and 108°F for the HH and LL pastes. Testing for the fresh CLSM properties included the deter- mination of the percent bleeding (ASTM D940), flowability (ASTM D6103), nonstandard inverted slump cone (ASTM C1611), and unit weight (ASTM D6023). The results are shown in Table 13. The bleeding evaluation, while not stan- dard for CLSM mixtures, was used in this study. In this test, the volume of excess water that accumulated on the surface of fresh mix placed in a graduated cylinder was measured. Flowability was assessed by placing the fresh mix in a non- absorbent cylinder (ASTM method) or an inverted slump cone on a clean, flat surface, then raising the container, and the largest diameter of the spread was determined. The bleeding decreased and the flowability decreased with increasing CKD. The unit weight increased with increasing CKD content. Fresh and hardened CLSM properties were determined for blends using Type I portland cement, Class F fly ash, sand, and city water. The level of the combined fly ash and CKD was held constant at 20%. Mixes varied the percent of fly ash, and hence the CKD, in 5% increments. Ratios of CKD to fly ash were: 3:1, 1:1, 1:3, and a 100% CKD. Water- to-cement ratio held constant at 0.85, but was adjusted as needed to maintain flowability. The set and hardening times of the CLSM mixes were determined using the pocket pene- trometer, penetration resistance (ASTM C403), and the Kelly ball (ASTM D6024). The combination of CKD and cement showed the fast- est set times using either the pocket penetrometer or the penetration resistance (Table 14). The rate of set was co- dependent on the levels of CaO and LOI percentages as well as the combination of CKD and fly ash. In other words, what combination works with one source of CKD may not work with another. This study showed that the best combination for the fastest rate of set was the highest CaO content with the lowest LOI CKD combined with fly ash. What data were available for the Kelly ball testing generally agreed with the pocket penetrometer and penetration resistance testing (Table 15). The compressive strengths with time are shown in Table 16. The compressive strengths were the highest for the high CaO and low LOI CKD, but decreased with increasing CKD con- tent. The concentration of CKD needed to achieve strength of 200 psi maximum (PCA 2010) decreased with increasing CaO content. CKD with high LOI provided adequate strengths, but lower concentrations were needed to gain strengths of around 150 to 200 psi. Williams (2005) used these data to develop a prediction equation for estimating the CLSM 28-day strength (Table 17): ) )( ( ) )( (s = + + +CaO LOI28day CKDc l f F b B Where: s28 day = 28 day compressive strength, MPa; CaO = calcium oxide content, %; LOI = loss on ignition; F = fineness, m2/kg; B = CKD content, percent of total mix, %; and c, l, f, and bCKD = coefficients selected from Table 17. Source Specific Surface Area (m2/kg) D95, µm D85, µm D50 , µm D10, µm Cement 300–500 45 35 13 2 LH 3,300 13 8 3 0.7 HH 3,900 30 5 2 0.6 HL 1,690 200 70 11 1 LL 230 200 170 30 5.2 After Williams (2005). Fineness determined using ASTM C204 (Blaine Fineness). First letter represents high or low value of CaO. Second letter represents high or low level of LOI. TABLE 12 CKD PROPERTIES USED IN LABORATORy STUDy ID Date % Total Weight CaO Si O 2 Al 2 O 3 Fe 2 O 3 MgO SO 3 Na 2 O K 2 O TiO 2 P 2 O 5 SrO Mn 2 O 3 L OI LH 7/05 39.01 12.58 4.45 2.28 1.7 7.46 0.29 10.96 0.17 0.08 0.06 NR 25.9 HH 7/05 51.94 10.62 3.94 2.68 1.02 2.26 0 1.87 0.19 0.03 2 NR 35.28 HL 7/05 44.64 13.35 3.83 2.57 2.44 3.64 0.8 5.46 0.19 0.05 0 NR 8.07 LL 7/05 38.89 13.43 3.86 2.25 1.99 13.85 1.36 8.77 0.15 0.11 0.04 0 1.4 After Williams (2005). First letter represents high or low value of CaO. Second letter represents high or low level of LOI. NR = not reported. TABLE 11 CHEMISTRy FOR CKD ByPRODUCTS USED IN CLSM MIxTURES

11 Mix CKD:Fly Ash w:cm Bleeding, % Flowability Unit Weight, pcf Inverted slump cone ASTM D6130 in. in. in. in. LH 5:15 0.80 2.4 13.0 13.0 31 30 128 13.5 13.5 10:10 0.83 0.4 9.5 9.5 21 20.5 130 10.5 10.5 15:5 0.85 0.2 7.8 7.8 18.5 18.5 134 8.3 8.3 20:0 1.00 0.6 8.5 8.5 20 22 126 8.5 8.3 9.0 9.5 HH 5:15 0.80 2.7 11.5 11.5 34 32 129 11.5 11.5 10:10 0.80 1.4 11.5 11.5 27.5 29 130 11.0 11.0 15:5 0.80 0.2 10.0 10.0 23.5 23.5 133 10.0 10.0 20:0 0.88 0 10.0 10.5 22 22.5 131 9.0 9.0 HL 5:15 0.74 3.4 11.5 10.5 23 23.5 132 11.5 11.0 10:10 0.74 2.3 8.5 8.5 20 19.5 134 8.5 8.5 15:5 0.80 0.7 8.5 8.5 ó ó 133 8.3 8.5 20:0 1.01 0.2 8.0 8.0 21 20.5 128 8.5 8.5 LL 5:15 0.70 3 9.0 10.0 22 23 130 9.5 9.5 10:10 0.78 2.1 10.0 10.0 22 23 132 9.5 9.5 15:5 0.89 0.6 8.3 8.3 19.5 20.5 130 8.5 8.3 20:0 1.07 1.1 8.5 8.5 21 21 125 8.0 8.3 After Williams (2005). First letter represents high or low value of CaO. Second letter represents high or low level of LOI. TABLE 13 FRESH CLSM PROPERTIES ID Mix Pocket Penetrometer Penetration Resistance 1 tsf, hours 4 tsf, hours 102 psi, hours 400 psi, hours LH CKD + FA <24 <23 <25 20–28 CKD <20.5 114 <23 never HH CKD + FA 17–24 42–44 <33 41–70 CKD never never never never HL CKD + FA 10–16 19–26 <21 16–23 CKD 14 31 <22 45 LL CKD + FA 28–35 57–72 <72 70–85 CKD 29 67 <67 100 After Williams (2005). First letter represents high or low value of CaO. Second letter represents high or low level of LOI. TABLE 14 RATE OF SET DATA FOR CKD CLSM

12 ID CKD:Fly Ash Indentation after Various Times, inches 24 hours 48 hours 72 hours 96 hours LH 5:15 3.5 3.4 — — 10:10 3.9 3.5 — 3.1 15:5 — — 4.5 — 20:0 — — 2.9 — HH 5:15 — 4 3.5 3.1 10:10 — — 3.5 — 15:5 — — — — 20:0 — 3.1 2.5 — HL 5:15 — 2.8 2.0 — 10:10 — — 2.8 2.5 15:5 — — 3.4 3 20:0 — — 4.0 — LL 5:15 — — 3.6 — 10:10 — — 4.0 — 15:5 — — 4.25 — 20:0 — — 4.25 — After Williams (2005). First letter represents high or low value of CaO. Second letter represents high or low level of LOI. — = indicates no data. TABLE 15 KELLy BALL INDENTATION DIAMETER AFTER 5 DROPS ID CKD:Fly Ash Esti mate d Co mp ressive Strength, psi 7 Days 14 Days 28 Days 56 Days 90 Days LH 5:15 124 145 173 165 180 10: 10 109 131 155 167 195 15: 5 70 90 110 140 148 20:0 — — — — — HH 5:15 67 100 120 140 160 10: 10 70 80 97 120 140 15: 5 58 68 70 78 89 20:0 — — 10 10 10 HL 5:15 200 375 720 1550 1750 10: 10 180 360 595 1150 1750 15: 5 90 200 310 565 720 20: 0 70 100 150 200 250 LL 5:15 75 225 530 725 — 10: 10 45 100 400 460 — 15: 5 30 80 210 445 — 20:0 25 60 75 140 — After Williams (2005); estimated from figures. First letter represents high or low value of CaO Second letter represents high or low level of LOI. — = indicates no data. TABLE 16 INFLUENCE OF CALCIUM OxIDE AND LOSS ON IGNITION OF CKD ByPRODUCTS ON COMPRESSIVE STRENGTH OF CONCRETE Coefficient Range Guidance for Selecting Coefficient c 6.1–18.0 High % CKD (>10%) = lower coefficients Low % CKD (<10%) = higher coefficients l 14.0–51.4 f 0.087–0.333 b CKD 6.3–21.9 TABLE 17 RANGE OF REGRESSION COEFFICIENTS FOR RANGE OF MATERIAL PROPERTIES IN THE WILLIAMS (2005) STUDy

13 The change in CLSM sample volume was also evaluated (Table 18). The volume changes increased with time and none of the samples showed any shrinkage. All of the mixes, except the low CaO/high LOI mixes, had volume changes of less than 2%. The change in sample mass shows that the low CaO/high LOI mixes lost mass, except at a ratio of 20:0, which gained the most mass. At 20:0 of the low CaO/high LOI, there was over 6.5% swell after 28 days and the highest percent increases in mass, which was attributed to water adsorption. All of the other mixes generally gained mass with time. Adaska and Taubert (2008) found in their literature review that CLSM slump and bleeding and strength decreased with increasing CKD content, although these mixes showed less bleeding than fly ash mixes. Longer set time for CKD and lower strengths compared with fly ash were also noted. With- out fly ash and with a high LOI, CKD mixes did not harden; the LOI has the greatest impact on hardened properties. Portland cement concrete Udoeyo and Hyee (2002) investigated five percentages of a single source of CKD. Results showed the slump decreased and setting times increased with increasing percentages of CKD. The compressive strength, splitting tensile, and modu- lus of rupture decreased with increasing percent of CKD. Research in Saudi Arabia by Daous (2004) used local CKD, cement, and fly ash from the combustion of heavy fuel oil to determine if this combination of materials could be used in the Middle East. The mortar cube testing showed adequate strength for 70% portland cement and 30% CKD. hot mix asphalt Taha et al. (2002) evaluated the influence of cement by-pass dust on asphalt binder and mix properties when used as mineral filler. When mixed with the asphalt, both filler materials (lime, CKD) used in the study decreased penetration and duc- tility with a corresponding increase in softening point with increasing mineral filler (Table 19). The CKD–asphalt binder appeared to have more ductility and a higher softening point than the lime–asphalt binder (Table 20). At 13% CKD the hot mix asphalt (HMA) mix properties met the specification requirements for both the wear and binder courses. At 5% CKD the voids in mineral aggregate (VMA) was under the Mix CKD:Fly Ash Diameter Change (3 in.), % Mass Change, % 7 days 14 days 28 days 56 days 90 days 14 days 28 days 56 days 90 days LH 5:15 — — — 1.22 1.22 0.9 0.03 0.84 0.36 10:10 — — — 1.13 1.04 0.28 –0.53 0.24 0.31 15:5 — — 1.39 1.3 1.74 0.33 –0.32 0.26 –0.04 20:0 — — 6.6 6.77 6.51 — 7.57 8.65 9.02 1:20:0 — — 0.95 0.78 0.78 0.29 0.39 0.42 0.43 HH 5:15 — 0.87 0.78 1.22 1.04 0.54 0.74 0.72 0.13 10:10 — 0.52 0.78 0.52 0.87 0.85 1.56 1.34 1.10 15:5 — 0.52 0.78 0.78 — –0.24 –0.24 0.56 0.02 20:0 — — 0.61 0.26 0.26 — — — — HL 5:15 0.78 1.04 — 1.04 1.13 0.91 1.15 1.23 1.06 10:10 0.78 1.04 1.13 0.95 1.04 0.47 1.06 1.02 0.50 15:5 1.04 1.30 1.04 1.30 1.04 0.39 0.27 0.21 0.39 20:0 0.95 1.04 1.04 1.04 1.04 0.61 0.28 0.29 0.18 LL 5:15 0.78 0.95 0.95 1.13 — 1.45 1.53 1.11 — 10:10 0.78 1.22 1.65 1.91 — 0.37 1.49 1.20 — 15:5 0.87 1.13 1.65 1.56 — 0.41 1.16 1.93 — 20:0 0.52 0.69 1.48 1.82 — 1.61 2.19 2.74 — After Williams (2005). First letter represents high or low value of CaO. Second letter represents high or low level of LOI. TABLE 18 VOLUME AND MASS CHANGES IN VARIOUS CLSM MIxES Binder Properties Asphalt Filler Content, % Lime CKD 2 5 7 10 15 2 5 7 10 15 Penetration, dmm 62 55 47 45 45 41 44 45 47 46 45 Ductility, cm 115 110 78 72 60 50 108 105 95 65 83 Softening Point, oC 41 47 50 53 56 57 41 45 43 48 50 Specific Gravity 1.03 After Taha et al. (2002). Values estimated from graphs in document. TABLE 19 INFLUENCE OF LIME AND CKD ON ASPHALT BINDER PROPERTIES

14 minimum value needed for either the wear or binder courses. The VMA was similar to the control mix, which also did not meet the specification; however, it is not clear that these are statistically significant. In this study, the use of the CKD helped improve the mix properties so that the mix met the specification requirements by increasing the VMA values. surface treatments Oikonomou and Eskioglou (2007) in Greece evaluated the influence of various sources of mineral filler in an emul- sion used for slurry seals. The mineral fillers included in the study were ordinary portland cement, fly ash, ladle furnace slag, CKD, and marble dust. The CKD had a free CaO con- tent of 3.8%. Testing evaluated mixing time, cone consistency, set time, cohesion, wet track abrasion for chip retention, wet stripping, and excess asphalt. The mixing time indicated how long the emulsion could be stirred before the emulsion began to break (i.e., emulsion separates from water). The cone consistency test was used to indicate the flow of the mix on a plate when poured from a standard mold. The cohesion test measured the cohesion at the interface between a rotating neoprene cylinder and the slurry seal test specimen at different times. Times of 30 min for set time properties and 60 min for resistance to damage from initial traffic were used. Wet track abrasion test loss evaluated the ability of the slurry seal to withstand traffic by measuring the loss of material with continued abrasion. Wet stripping checked the compatibility of the slurry seal sys- tem with the aggregate. Excess asphalt under loaded wheel conditions ensured that the mix did not show excess binder (flushing) under traffic. Results showed that the CKD provides the longest con- struction time window before breaking (highest mixing time) and acceptable, but lower, cohesion than seen for ordi- nary portland cement (Table 21). All of the mineral fillers provided acceptable values compared with the specification requirements. applications—unbound Stabilized Soils Stabilized soils are used for: • Drying up construction sites • Providing working platforms • Reducing soil plasticity index (PI) and clay/silt-sized particles • Improving compactability • Reducing shrinkage/swell of expansive soils • Improving strength and stability • Reducing moisture susceptibility • Utilizing local or recycled materials. Properties Control Mix (lime filler) 5% CKD 13% CKD Ministry of Transportation Spec. Wearing course Binder course Stability, lb 4,496 3,822 4,047 3,417 min 2,698 min Flow, 0.01 in. 8 9 10 8–16 8–16 Air Voids, % 4.5 4.5 4.5 3.5–5.5 4–6 Voids in Mineral Aggregate, % 14.2 13.9 16.7 15 min 14 min Voids Filled with Asphalt, % 70 68 72 63–75 55–70 After Taha et al. (2002). TABLE 20 HMA MARSHALL MIx DESIGN PROPERTIES Test Mineral Filler Type at 2% by Weight of Aggregate ASTM Specification International Slurry Seal Assoc. (ISSA) OPC HCFA LFS CKD MD Specification Test method Mixing Time, s 134 125 122 137 108 60–180 >120 TB106 Cone Consistency, cm 2.7 2.3 2.8 2.6 2.1 2–3 Set Time, s 175 143 151 196 228 <1 h Cohesion, 30 min, kg cm 17 14 15 15 13 >12 TB139 Cohesion, 60 min, kg cm 25 26 28 23 22 >20 TB130 WTAT, 1 h soak, g m-2 312 321 336 314 362 <807 <538 TB100 WTAT, 6 d soak, g m-2 540 506 512 532 544 <807 Wet Stripping, % >95 >95 >95 >95 >95 >90 TB114 Excess Asphalt by CWT, g m-2 392 429 382 416 371 <538 TB109 After Oikonomou and Eskioglou (2007). OPC = ordinary portland cement; HCFA = high calcium fly ash; LFS = ladle furnace slag; CKD = cement kiln dust; MD = marble dust; WTAT = wet track abrasion test. TABLE 21 INFLUENCE OF DIFFERENT MINERAL FILLERS ON EMULSIONS FOR SLURRy SEALS

15 The selection of treatment depends on the specific goals for a specific project, soil type, material availability, traf- fic loads, pavement design needs, environmental conditions (e.g., drainage, ground water table, and precipitation), and costs. Mix design is a key component of selecting appropri- ate stabilizer. The Oklahoma Department of Transportation (DOT) (2009) developed a soil modification mix design method that could be used for designing CKD stabilized soil in AASHTO M145 soil groups A-4, A-5, A-6, and A-7. The mix design method provided two methods for design. The first was an abbreviated laboratory test procedure selecting an optimum percentage from the Soil Modification table. The second method described a complete laboratory test procedure for designing stabilized soils. The end results produced general recommendations for each type of additive with each soil type (Table 22). In addition, soluble sulfates were measured using OHD L-49 (Method of Test for Determining Soluble Sulfate Con- tent in Soil) and, if the content was greater than 500 ppm, additional samples needed to be tested. Modification with calcium-based additives might not be appropriate for values greater than 1,000 ppm. Modification was not recommended if greater than 8,000 ppm. Soil dispersion was determined using the crumb test, ASTM D6572, which was originally called an aggregate coherence test, with seven different categories of soil–water reaction. There were only four categories (grades) of soil dispersion. If a grade of 3 or 4 was found, then the soil needed to be evalu- ated with the Pinhole Test procedure, ASTM D4647, which modeled the action of water flowing along a crack in an earth embankment. If the soil was dispersive, the agency represen- tative was notified. In this case, all exposed surfaces required special treatment to prevent erosion. The full mix design used a sample prepared with the addi- tive with minus No. 40 soil material at a moisture content equal to the plastic limit, covered and cured for 48 h, then dried and prepared for testing (AASHTO T87). The Atter- berg limits were determined and a plot of PI versus addi- tive percentage was prepared. The percent modification that reduced the PI by 2% per 1% of additive was selected as the optimum percent. Once the amount of additive was estimated, the target density and optimum moisture content were determined (AASHTO T99). Method D in this standard was used if the soil had more than 5% retained on the 19 mm sieve and Method A if 5% or less was retained on the 4.75 mm sieve; otherwise, Method C was used. The report included infor- mation about the AASHTO group classification of untreated soil, soluble sulfate content of soil, recommended percent of type of additive, density, and optimum moisture content for both the untreated and treated soil. Khoury and zaman (2007) evaluated the impact of 0 to 30 days of freeze/thaw cycles, after 28 days of curing, on stabilized aggregates for base materials using CKD, Class C fly ash, and fluidized bed ash. The percent added to each mix was held constant for each stabilizing material: 15% for CKD, 10% for each of the other two. Properties of the three addi- tives are shown in Table 23. Type of Additive Selected for Stabilization of Soil Percent Modification Needed for Different Soil Classifications, % Soil Group Classification—AASHTO M145 A-4 A-5 A-6 A-7 A-7-5 A-7-6 Portland Cement 3 3 3 Fly Ash 9 9 9 CKD, Precalciner Plant 4 4 4 CKD, Other Type 8 8 Hydrated Lime* NA NA 3 3** 3** *Reduce quantity by 20% when quicklime is used. **Use 4% when the liquid limit is greater than 50. NA = not available. TABLE 22 OKLAHOMA DOT SOIL MODIFICATION TABLE (2009) Property CKD Class C Fly Ash Fluidized Bed Ash Compounds, % by weight SiO2 + Al2O3 + Fe2O3 19 62 35 CaO 44 27 41 MgO 1.5 5.4 2.7 SO3 2.5 2 19 CaCO3 64 — 41 Other Properties Free Lime 2 to 3 — 18.2 Loss on Ignition, % 29 0.2 5.3 Percent Fineness — 11.5 55 Specific Gravity 2.74 2.69 2.87 — = no data reported. TABLE 23 PROPERTIES OF STABILIzING ADDITIVES USED IN THE KHOURy AND zAMAN (2007) STUDy

16 Testing focused on determining the resilient modulus values for the mixtures since this material property is used in mechanistic-empirical pavement design methods. Results were dependent on the particular combination of stabiliz- ing material and aggregate source (Table 24). The CKD pro- vided twice the initial modulus for limestone 2 as compared with limestone 1. The authors noted that both the fly ash and fluidized bed ash were more effective than the CKD with either of the limestone. The CKD provided higher modulus values for the sandstone aggregate mix compared with the other two additives. Miller and Azad (2000) noted in their literature review that in 1991 Kamon and Nontananandh suggested that a cementi- tious stabilizer should have a hydration modulus between that of alite and belite. Because CKD satisfies this requirement, it would act as a cementitious material. Typically, CKD had about one-third of the oxides as present in portland cement and approximately 6% to 10% of the total analytical calcium oxide was free lime (Table 25). Three Oklahoma natural soils were selected for the study (Table 26). The main variable was the level of plasticity (low, medium, high). Soils 2 and 3 reach a pH maximum of 12.3 at 15% CKD; soil 1 needs approximately 40% CKD to reach the same level. A higher pH is equated with greater chemical activity as a result of the cation exchange capacity of the clay fraction. The authors suggested that this was a function of the nature of the clay size fraction. They also noted that an upper limit of 15% was most likely the practical upper limit for cost- effective stabilization (based on authors experience). Results showed that pH and unconfined compressive strength decreased with increasing PI when using the same percent and source of CKD. At a PI of either 33 or 40 (CH or CL), the compressive strength was between 200% and 400% of the untreated soil. At a PI of 21, the unconfined com- pressive strength increased by about 600% to 1,300% for 7- to 28-day curing, respectively. Similar trends were seen with increasing pH; unconfined compressive strength increased with increasing pH. Changes in PI were more pronounced for the higher PI soils with the addition of 5% CKD. Parsons et al. (2004) evaluated the use of CKD from the three types of cement kilns for the Kansas DOT. Data from local consultants was provided to the University of Kansas by industry. These data were then sorted into one of two cat- egories of information; the first was for precalciner kilns and the second grouped both the long-wet and long-dry kilns together. The long-wet and long-dry data were combined because previous research showed that the material prop- erties were similar (Table 27). The precalciner CKD had much higher free lime content than the long-wet or long-dry kilns; therefore, it was expected to perform more like a lime stabilized soil. Materials and testing included the evaluation of lime, fly ash, cement, CKD (pre-calcined), and Permenzyme as stabilizers for eight different types of soils meeting one of the Unified Soils Classifications (CH, CL, ML, SM, or SP). Test methods used in the study were grain size analy- sis (ASTM D422), Atterberg limits (ASTM D4318), spe- cific gravity (ASTM D854), pH lime stabilization (ASTM D6276), moisture-density relationships (ASTM D698), swell (KDOT spec), freeze/thaw (ASTM D560), wet-dry condition- ing (ASTM 559), and unconfined compression (ASTM D1633 and D5102). The CKD content was determined based on pH and Atterberg limit results and the percent of quicklime accord- ing to ASTM D6276. The fly ash content was fixed at 16% and the percent of cement was determined by the amount needed to lower the PI below 10, or capped at a maximum of 9% for cost reasons. The proportion of the sand- and fine- sized particles in each soil type is shown in Table 28. Properties of soils modified with and without CKD are shown in Table 29. The CKD provided similar or slightly lower strengths than either the fly ash or lime. The pH ini- tially increased with increasing percent of CKD. Although the percent of CKD needed to attain the maximum pH varied with the soil type, the pH was consistently greater than 12 at 5% CKD for all mixtures. The CKD significantly increased the pH of the soil, slightly higher than lime (pH = 12.45). The CKD with the exception of one source of CH soil reduced the swelling potential. The CH soil used in this study contained sulfates and none of the additives worked well with this soil. For the other soils, the reductions in swelling when using the CKD were either similar to or slightly better than any of the other admixtures. Permeability of the CKD tended to decrease or remain steady with time (except for CH soil). Unconfined compressive strength after leaching showed an estimated retained strength of between 60% and 105% of the original strengths (except for CH soil). Most of the CKD samples did not survive the full 12 cycles of wet-dry testing; similar results were obtained for the other additives. Only four of the eight samples survived the 12 cycles of freeze/ thaw testing. Recommendations from the Parsons et al. (2004) study were that the CKD is an effective stabilizer for subgrade soils. Suggestions for using CKD in projects included: • Specify a free lime content of the CKD if it is used as a lime replacement. An alternative would be to pay the contractor based on the tonnage of free lime contained in the CKD. • Use pH testing with Atterberg limits for additional guidance for determining the optimum CKD content. • Evaluate the effectiveness of the CKD on reducing swell to confirm effectiveness, particularly with high plastic- ity soils. • Conduct strength testing, because it will be important if the subgrades are to be a substantial contributor to the strength of the pavement system.

Stresses Resilient Modulus, ksi for 15% CKD tsemiL 1 enotsemiL enotsdnaS 2 eno Aggregate: 97% CaO3 0.9 to 1.2% MgO3 0.25 to 6.2% SiO2 34% for LA abrasion Aggregate: 87% CaCO3 — 10% SiO2 26% for LA abrasion Aggregate: — — 94% SiO2 22% for LA abrasion Confining Stress, σ3 Deviator Stress, σd Seating Pressure, σs No. freeze/thaw cycles No. freeze/thaw cycles No. freeze thaw cycles 0 8 15 0 8 15 30 0 8 15 30 951 475 193 11,582 7,069 2,873 30,226 19,140 12,698 3,190 22,172 14,924 9,667 3,914 951 951 193 12,292 7,262 2,956 31,942 19,010 13,243 3,521 25,121 16,936 10,197 4,079 951 1,433 193 15,227 7,586 3,169 34,416 21,077 14,938 3,748 29,861 18,196 10,411 4,189 951 1,909 193 15,689 7,751 3,535 38,005 33,857 15,702 4,113 33,210 19,940 11,141 4,665 717 475 193 11,382 6,353 2,508 29,751 18,679 12,182 2,873 21,442 14,986 8,034 3,362 717 951 193 11,837 6,725 2,976 30,323 19,175 12,113 3,473 23,750 15,571 8,909 3,528 717 1,433 193 13,187 6,835 3,569 32,762 21,400 13,608 4,155 27,167 17,659 10,335 4,134 717 1,909 193 14,827 7,090 3,886 35,284 34,395 12,498 4,603 31,322 19,581 11,493 4,665 475 475 193 11,155 6,222 2,715 29,558 18,534 11,782 3,087 21,517 15,000 8,220 3,280 475 951 193 11,658 6,511 3,073 29,847 19,450 11,768 3,555 23,481 15,213 9,053 3,624 475 1,433 193 12,740 6,614 3,652 32,183 21,194 13,229 4,217 27,209 17,377 10,452 4,286 475 1,909 193 14,600 7,021 4,051 34,236 32,927 12,312 4,733 29,668 19,140 11,720 4,685 241 475 193 11,086 6,029 2,770 26,919 18,830 11,493 3,287 42,311 15,048 8,103 3,266 241 951 193 11,630 6,380 3,142 28,139 19,719 12,588 3,727 22,999 15,234 8,923 3,569 241 1,433 193 12,643 6,428 3,693 28,277 22,262 11,610 4,327 27,643 17,411 10,397 4,196 241 1,909 193 16,316 6,911 4,106 32,631 34,932 12,223 4,940 28,483 19,030 11,720 4,878 0 475 193 11,107 5,919 2,804 27,340 18,693 11,196 3,280 21,449 15,234 7,979 3,231 0 951 193 11,527 6,153 3,149 28,876 18,596 12,312 3,672 23,268 15,482 8,819 3,528 0 1,433 193 12,374 6,325 3,672 28,924 20,022 11,389 4,375 26,788 17,328 10,342 4,162 0 1,909 193 14,421 6,869 4,113 35,470 31,115 11,961 4,954 27,994 17,445 11,706 4,678 After Khoury and Zaman (2007). Cured for 28 days before freeze/thaw testing. TABLE 24 MODULUS VALUES FOR CKD STABILIzED LIMESTONE AGGREGATE

18 Property Precalciner Kiln Long-Kiln SiO2 0.40–1.11 26.71 % , Al2O3 0.5–5.3 9.4 % , Fe2O3 5.5–2.1 85.2 % , — 80.26 % ,OaC — 39.1 % ,OgM Na2 0.1–1.0 65.0 % ,O K2 0.01– 0.2 67.3 % ,O SO3 0.21–00.8 97.5 % , 0.1–3.0 — % ,selitaloV Available Lime Index, % CaO 33.7 — Water-Soluble Chlorides, % Cl — — — 70.0 % ,tnetnoC erutsioM — 49.4 % ,noitingI no ssoL Retained on No. 325 sieve, % 16.9 — 57–55 — % ,mm 570.0 gnissaP — 59.2 ytivarG cificepS 9.21–4.21 — Hp After Parsons et al. (2004). — = indicates no data. TABLE 25 COMPARISON OF PHySICAL AND CHEMICAL PROPERTIES OF TWO KANSAS CEMENT KILNS Properties Plasticity Index Levels High Medium Low USCS Classification CH CL ML % Finer Than 0.075 mm 98 94 52 % Finer Than 0.002 mm 51 42 21 32 84 55 % ,timiL diuqiL 6 33 04 % ,xednI yticitsalP 92.0 97.0 87.0 ytivitcA 76.2 27.2 28.2 ytivarG cificepS Optimum Moisture Content, % 23.3 16.0 14.0 Maximum Dry Unit Weight, lb/ft3 101.2 111.4 118.4 7.7 3.5 6.7 Hp Sulfate Content, SO4-2, mg/kg 137 171 ND Organics, % by weight 1.62 0.86 0.36 ND = not detectable. TABLE 26 SOIL PROPERTIES USED IN MILLER AND AzAD (2000) STUDy Property Precalciner Kiln Long-Kiln SiO2 0.41–0.11 26.71 % , AlsO3 0.5–5.3 9.4 % , Fe2O3 5.2–5.1 85.2 % , — 80.26 % ,OaC — 39.1 % ,OgM Na2 0.1–1.0 65.0 % ,O K2 0.01– 0.2 67.3 % ,O SO3 0.21–00.8 97.5 % , 0.1–3.0 — % ,selitaloV Available Lime Index, % CaO 33.7 — Water-Soluble Chlorides, % Cl — — Moisture Content, % 0.07 — — 49.4 % ,noitingI no ssoL Retained on No. 325 sieve, % 16.9 — Passing 0.075 mm, % — 55–75 — 59.2 ytivarG cificepS 9.21–4.21 — Hp After Parsons et al. (2004). TABLE 27 COMPARISON OF PHySICAL AND CHEMICAL PROPERTIES OF TWO KANSAS CEMENT KILNS

19 • Evaluate the sulfate in the CKD and soils. Soils or CKD materials with sulfates could potentially react with free lime and form expansive minerals, resulting in additional swelling where none previously existed. The percent- ages of sulfates in the CKD should be reported as part of the chemical analysis of the CKD. Adaska and Taubert (2008) noted that CKD could be used to stabilize highly expansive clay soils when com- bined with fly ash and limestone aggregates that produced a noncement concrete. The LOI was a significant factor in the effectiveness of CKD for stabilizing these soils. The CKD improved the unconfined compressive strength and reduced the PI when the LOI was low. A high LOI resulted in lower unconfined compressive strengths and higher PI. The higher LOI implied a higher amount of bound water and less CaO available to react. The CKD also improved freeze/thaw resistance and provided an alterna- tive to quicklime for subgrade stabilization. CKD has been reported in the literature to perform better than quicklime when the results were compared over time. Fresh CKD worked as well as hydrated lime. Larger quantities of CKD were needed compared with the hydrated lime for stabili- zation. CKD from stockpiles, rather than fresh from the Properties Unified Soils Classification System (USCS) Designation CH CH CH CL ML CL SM SP Native Soil Properties 69 07 43 21 8 5 21 5 dnaS % 4 03 66 88 29 59 88 59 seniF % — 02 53 03 63 56 35 07 timiL diuqiL — 3 61 4 61 63 13 54 xednI yticitsalP AASHTO Designation A-7-6 A-7-6 A-7-6 A-6 A-4 A-6 A-2-4 A-3 Unit Weight, lb/ft3 701 021 401 89 801 9.69 4.501 49 Optimum Moisture Content, % 25.7 20.3 25.3 18.5 13.7 19.9 9.9 2 Unconfined Compressive Strength at Optimum, lb/ft2 6,400 4,600 4,600 4,800 6,600 4,415 5,638 — Maximum Unconfined Compressive Strength, lb/ft2 8,600 7,500 6,400 7,500 6,600 6,200 5,638 — Moisture at Maximum Unconfined Compressive Strength, % 18.9 18.6 23.5 17 13.7 17.6 9.9 — 66.2 86.2 96.2 57.2 47.2 27.2 77.2 87.2 ytivarG cificepS RN 4.0 4.1 1 4.1 8.2 5.2 4.4 % ,llewS After Parsons et al. (2004). Sand refers to particles passing the 4.75 mm sieve (no. 4) and less than about 10% passing the 0.75 mm sieve (no. 200). Fines refer to particles passing the 0.075 mm sieve (no. 200). NR = not reported. TABLE 28 NATIVE SOIL PROPERTIES AND PERCENTAGES OF ADDITIVES Properties Soil Type CH CH CH CL ML CL SM SP Liquid Limit Before leaching 54 54 56 42 35 48 NP NP After leaching NP to 49 44 NP NP NP NP NP NP Plasticity Index Before leaching 13 17 17 10 6 12 NP NP After leaching NP to 11 9.5 NP NP NP NP NP NP Max. Dry Unit Weight lb/ft3 54 91 89.5 92 84.5 86 110 110 Opt. Moisture Content, % 23 21 20 23 17 24 16 5.5 Opt. Unconfined Compressive Strength, lb/ft2 17,500 18,000 12,250 23,000 12,250 15,750 14,000 780 Max. Unconfined Compressive Strength, lb/ft2 17,700 20,000 14,400 23,250 16,850 15,750 14,000 780 Moisture at Max. Unconfined Compressive Strength, % 23.5 23.5 23.0 24.5 23.0 24.0 16.0 5.5 enon 1.0 enon 2.0 0.1 1.7 4.1 % ,llewS Permeability with Time, cm/s 7 days 1.91E-05 3.80E-05 1.50E-06 2.30E-06 1.20E-05 1.10E-05 4.30E-08 2.60E-03 0-E02.3 50-E08.1 50-E93.1 41 8 2.70E-06 3.20E-06 5.75E-06 3.90E-08 2.80E-03 0-E08.4 50-E07.1 50-E53.1 12 8 4.50E-06 3.30E-06 5.15E-06 4.60E-08 2.90E-03 0-E03.2 60-E00.7 60-E01.7 82 7 2.85E-06 2.65E-06 1.20E-05 4.00E-09 3.50E-03 Freeze/Thaw Cycles to Failure 11 10 2 12 12 12 12 4 Wet-Dry Cycles to Failure NR 1 NR 7 2 3 12 NR After Parsons et al. (2004). NP = not provided. NR = not reported. TABLE 29 PROPERTIES OF CKD STABILIzED SOILS

20 cement plant, had a lower free lime content that resulted in poor reactivity. Peethamparan et al. (2008) evaluated CKD from four dif- ferent kiln types. One cement facility used a long-dry kiln with limestone, shale, sand, and iron ore as raw materials. A second source of CKD was from a long-dry kiln with a pre- heater that used limestone, clay, bottom ash, and iron scale as raw materials. The facility with the long-wet kiln used limestone, clay, sand, fly ash, and blast furnace slag, whereas the precalciner kiln used limestone, clay, bottom ash/fly ash, found sand/sludge, and iron waste in their cement produc- tion. The chemical properties of each CKD are shown in Table 30. Testing for differences in chemistry and morphol- ogy was conducted using xRD, thermal gravimetric analy- ses, differential TGA, and SEM. The properties of the mix measured the heat of hydration and unconfined compressive strength. Significant amounts of calcium hydroxide, syngenite, and ettringite were identified in hydrated free lime content CKDs. The amount of ettringite increased with increased curing periods. These reaction products were either low or nonexistent in CKD with lower free lime. The high free lime content CKD was also responsible for a higher strength gain compared with the lower lime content CKD. The authors attributed this to the increased formation of ettringite and the secondary C-S-H during hydration. The high free lime content was also responsible for higher temperatures during hydration. Both compressive strength and heat of hydration provided a good indication of the performance of the CKD used as a stabilizer in soils. Although not as effective at increasing the strength of the stabilized soil, the lower free lime content CKDs in the kaolinite improved the strengths from 100% to 300% after 7 days of curing. Peethamparan et al. (2009) evaluated the mechanisms for CKD stabilized Na-montmorillonite clays after more than 90 days of moist curing using xRD, SEM, and energy- dispersive x-ray spectroscopy (EDx). The results showed that extensive physicochemical changes occurred during curing. Calcium hydroxide was initially produced, but was quickly absorbed by the clay. SEM photographs showed that fractured surfaces of the bulk clay microstructure were significantly modified over time. The pH of the clay–CKD system was initially elevated to more than 13, but dropped over time to a stable value of 12.5 despite the absence of detectable calcium hydroxide. Some gypsum was produced by the anhydrite in the CKD and water reaction, which led to ettringite formation. The C-S-H reaction products were identified locally on the fracture surfaces of the CKD–clay system. The authors assumed that this was a function of the reaction of adsorbed calcium hydroxide with silica from the clay. Base and Subbase Texas Transportation Institute Report TTI-2003-1 is a syn- thesis that was prepared for TxDOT, which provides a good summary of pre-2001 information (Button 2003). The con- clusions at the end of this synthesis were: Chemical Composition by XRF Long Dry Precalciner Dry and Dry Preheater Long Wet Type I Cement Kaolinite Clay Percent by Weight (%) SiO2 12.18 16.42 11.91 15.39 20.48 45.73 Al2O3 4.24 3.62 2.17 4.66 4.21 37.36 TiO2 0.22 0.23 0.15 0.57 0.36 — P2O5 0.08 0.09 0.09 0.09 0.09 — Fe2O3 1.71 2.31 2.08 2.34 2.41 0.79 CaO 46.24 55.00 46.05 37.35 63.19 0.18 MgO 1.24 2.68 2.20 2.10 4.00 0.098 Na2O 0.51 0.17 0.33 0.81 0.19 0.059 K2O 4.89 2.89 1.43 7.00 0.28 0.33 Na2O equivalent 3.72 2.05 1.27 5.36 0.37 — Mn2O3 0.05 0.44 0.04 0.07 0.14 — SrO 0.04 0.03 0.07 0.02 0.04 — SO3 14.62 12.69 4.21 5.80 2.76 — Cl 0.59 0.74 0.35 3.26 — — LOI 14.22 3.92 29.63 27.65 1.76 — Free CaO 13.85 29.14 5.32 3.26 1.58 — Water-soluble Na2O 0.28 0.06 0.12 0.59 0.04 — Water-soluble K2O 2.95 1.68 0.93 6.33 0.16 — Raw Materials Limestone Shale sand Iron ore Limestone Clay Bottom ash/fly ash Foundry Sand/sludge Iron waste Limestone Clay Bottom ash Iron scale Limestone Clay sand Fly ash Blast furnace slag — — After Peethamparan et al. (2008). TABLE 30 PROPERTIES OF CKD FROM FOUR TyPES OF CEMENT KILN PROCESSES

21 • CKD can be used to stabilize subgrade soils and bases; the combination of CKD and fly ash significantly increase the compressive strength because of the poz- zolanic reactions. • High-quality bases for pavements can be obtained using CKD, but testing is needed to optimize the performance. • Full-depth reclamation HMA recycling successfully uses CKD to produce a base layer. • Specifications should set minimums for key components for testing or certification, or warranty performance. • Soluble sulfates and alkalis can lead to undesirable swelling. • Some cement plants may burn hazardous waste as kiln fuel, which may lead to hazardous materials in the kiln dust. • The CKD needs to be kept dry to preserve the reactivity of the material. • There is very little, or no, free lime or magnesia in stockpiled CKD. Aged stockpiled CKD should not be used as a component of stabilized base or subgrade soil unless conditioned by the addition of commercial lime to enhance short-term strength development. • Additional research is needed to assess the suitability of kiln dust as a pozzolan activator in stabilized base and soil applications. • Specifications are needed to define the physical and chemical properties that will provide acceptable per- formance. • Environmentally related properties are needed along with management guidelines. environmental considerations Approximately 5% of global CO2 emissions originate from the production of cement and is the third largest source of carbon emission in the United States. Life-cycle assessment is a method of evaluating the environmental impacts of tech- nologies from “cradle to grave” and may be performed on both products and processes. Huntzinger and Eatmon (2009) conducted a life-cycle assessment of portland cement manufacturing that compared the traditional manufacturing process with alternative tech- nologies and raw materials. LCA methodology consisted of four major steps (Figure 5): • Determination of the assessment scope and boundaries • Selection of inventory of outputs and inputs • Assessment of environmental impact data compiled in the inventory • Interpretation of the results and suggestions for im prove ment. The life-cycle assessment was conducted for four manu- facturing processes: • Production of ordinary portland cement • Blended cement (natural pozzolans) • Cement where 100% of waste CKD was recycled into the kiln process • Portland cement produced when CKD was used to seques- ter a portion of the process related to CO2 emissions. Quarrying Raw Materials Preheater Dry Mixing and Blending Raw Material Preparation (Grinding) Processing Raw Materials (Crushing) Finish Grinding Clinker Cooler Rotary Kiln Packaging Product Storage GypsumShipping 1 1 1 111 1 E 2 11 1 E 2 E 2 EE EE H H After Huntzinger and Eatmon 2009 E 1 2 H Particulate Emission Gaseous Emission Energy Heat FIGURE 5 Flow chart for life-cycle assessment for cement kiln dust (after Huntzinger and Eatmon 2009).

22 CKD recycling was used by a majority of the cement pro- ducers. Although a number of plants could reuse 100% of their CKD, the amount that was recycled ultimately depended on the chemistry of the CKD. For this study, the authors assumed 100% recycling of the CKD. The scope of this research was limited to the processes shown in the previous figure rather than the “cradle to grave” so that the focus of the analysis was on only the environmental impact of changing the four plant variables. The environmental impact of packaging and trans- portation default values in the Eco-indicator95 in the SimiPro software was used for this information. The life-cycle inventory used five major nonfuel raw materials consumed in the processes. The amount of material per ton of cement manufactured considered in the analysis is shown in parentheses: • Calcium oxides (1.41 tons), • Aluminum oxides (0.139 ton), • Silica (0.034 ton), • Ferrous oxides (0.015 ton), and • Calcium sulfate (0.05 ton). The SimaPro 6.0 software was used for the analysis. Energy usage varied widely depending on the plant. To control all variables except for the use of CKD, a blend of fuels was assumed as coal (70%), fuel oil (15%), and natu- ral gas (15%). The natural pozzolans were considered envi- ronmentally benign because they were typically the waste byproduct (e.g., fly ash and rice husks) of another process. The impact for the activities associated with these materials such as the collection procedures and transportation were considered to be the financial and environmental responsi- bility of those recyclers. A substitution of 25% by weight was assumed. CKD was considered to have the theoretical potential to sequester 0.4 ton of CO2 per ton of cement. Previous batch and column studies showed that CKD could readily seques- ter greater than 80% of the theoretical capacity at ambient temperatures and pressures. Results indicated that using the CKD for CO2 sequestering decreased the cement processing environmental impact score by 5%, which was the best of the choices. sPeciFications Lafarge (2008) cited the general reasons for using CKD as providing direct cementation of soils, containing reactive silica, providing ion exchange, and decreasing the moisture content of soils. A starting estimate of the dry weigh of CKD needed for the reduction of moisture was obtained by: W W W W W i f f CKD dry soil = −( ) +( )1 1604 0 1604. . Where: WCKD = dry weight of CKD; Wi = initial in situ moisture content; Wf = final moisture content; and Wdry soil = dry weight of in situ soil used. An example of a Michigan DOT project where CKD was used for the I-75 Ambassador Bridge was summarized. Mich- igan DOT had no specification for CKD in subgrade stabili- zation and asked Lafarge to develop such a specification. It was developed based on the anticipated uniformity of CKD properties within a given source, but with potentially widely variable properties between CKD sources. The specification was developed recognizing the need to be sufficiently broad to allow for more than one source of CKD to compete for the project. A review of existing specifications indicated few examples were available for CKD and those that were found appeared to be based on existing Type C fly ash specifica- tions. Since the CKD and Type C fly ash were very different in chemical and physical properties, the specification limited the amount of SO3 in the CKD to 10% without needing jus- tification. The original justification for the limit appeared to be a response to concerns with alkali silica reactivity (ASR) when using fly ash. Lafarge conducted an 18 month study and found levels of SO3 as high as 15% in the CKD produced blends, with vari- ous soil types showing acceptable expansions. Testing of the source material was considered to be critical; therefore, any problems with expansive behavior were identified for a given combination of CKD and soil. The final specification recom- mended the following: • CKD must conform to ASTM D5050-96. • Soil classification needed to be per AASHTO M145 and ASTM D2487. • Moisture and density testing per AASHTO T99 for both untreated and treated soils. • California bearing ratio (CBR) lab results must be above 10% using ASTM D1883. • Atterberg limits needed to be performed according to ASTM D4318. • Unconfined compressive strength must have a minimum of 125 psi at 7 days (ASTM D5120). agency survey results The most common use for CKD was in soil stabilization (Table 31). Other applications that used CKD were in portland cement concrete (PCC) and HMA. Eleven states indicated they had used CKD in highway applications (Table 32 and Figure 6). Three states have used a combination of cement and lime kiln dusts. The most common comment to the success or failure of a project was in identifying the correct soil for CKD stabi-

23 lization (Table 33). Texas noted that identifying sulfate and chloride contents was essential to using CKD in areas that will come in contact with reinforcement. document assessment survey Twenty-two documents were identified and reviewed for this byproduct. The summary of applications addressed in these documents showed that the most commonly researched uses were in PCC, geotechnical, and HMA products (Figure 7). The documents include research from national and interna- tional sources (Figure 8). The agencies reported focusing more on the use of CKD in geotechnical applications than either concrete or HMA. This suggests that these areas may benefit from increased use of this byproduct. summary oF cement kiln dust inFormation list of candidate Byproducts The list of the most commonly researched and used byprod- ucts include: • CKD, long-wet or long-dry kiln • CKD, precalciner kiln. The CKD byproducts are separated by the type of kiln used to collect the byproducts since the physical and chemical properties of the CKD are dependent on the type of kiln. test Procedures The test methods used to evaluate byproducts and highway applications are shown in Tables 34 and 35. material Preparation and Byproduct Quality control The following byproduct post-processing and quality control (QC) points need to be considered when using CKD in high- way applications: • Periodic composition testing to be done to track histor- ical changes in CKD byproduct over time. Changes in technology, burner fuel, and/or sources of raw materials can change the properties of the CKD. – CKD for PCC applications is most effective when there is a high concentration of CaO and a low LOI. • Post-processing of the CKD can improve reactivity by post-processing grinding of the CKD. TABLE 31 RESULTS FOR AGENCy SURVEy FOR CEMENT KILN DUST ByPRODUCTS USED IN HIGHWAy APPLICATIONS Question: Manufacturing or Misc. Construction Byproducts: Is your state using, or has ever used, these byproducts in highway applications? * Kiln dust, cement: airborne particles from the portland cement rotary kiln * Kiln dust, lime: airborne particles from the lime production process * Kiln dust, combination: blending of both cement and lime kiln dusts Type of Sand Byproduct Asphalt Cements or Emulsions Crack Sealants Drainage Materials Embank. Flowable Fill HMA Pavement Surface Treatment (nonstructural) PCC Soil Stability Cement Kiln Dust 0 0 0 0 0 2 0 3 7 Combination Kiln Dust 0 0 0 0 0 0 1 1 1 Embank. = embankment. TABLE 32 STATES USING CKD ByPRODUCTS IN HIGHWAy APPLICATIONS IN 2009 No. of Applications States Cement kiln dust Comb. dust — RO 2 1 CO, IL, IN, IA, KY, MO, NE, NM, NY, TX IA, MA, NY 2009 Kiln Dust, Cement 1 1 1 1 1 1 1 1 2 1 1 FIGURE 6 States currently using CKD in highway applications.

24 tnemmoC etatS CO Cement kiln dust did not perform well in the soil we were trying to stabilize. KY Kiln dust used for soil stabilization can be successful, but it depends on the type of soil involved (critical). PA CKD has proven to be an excellent soil stabilizer. TX The Amarillo District was happy with the performance of cement kiln dust used in soil stabilization and base applications. You need to be very conscious of the sulfate and chloride contents of the cement kiln dust especially in areas with reinforcement. TABLE 33 SUMMARy OF COMMENTS By STATES ON THEIR ExPERIENCE WITH USING CKD IN HIGHWAy APPLICATIONS Binders Emulsions Filler Full Depth Reclamation HMA Drainage Embankments Fill material Flowable fill Soil stabilization Aggregates Retaining walls Crack seals Chip seals Slurry Microsurface Interlayers Non-structural overlays Localized repairs Cement types Grouts Conventional concrete High performance concrete High strength concrete Cement types Mortar cements nonstructural concrete Pervious concrete Precast Pile grout Admixtures 0 2 4 6 8 10 12 14 16 Number of Documents Kiln Dust Concrete Pavement Preservation Geotechnical Asphalt 18 documents for CKD 3 documents for LKD 8 documents for combination kiln dusts FIGURE 7 Summary of information contained in the reviewed literature.

25 materials handling issues The following materials handling and stockpiling points need to be considered: • CKD byproduct could designate the type of kiln that generated the byproduct. • Age of stockpiles needs to be tracked and the age of the CKD be included in the byproduct information pro- vided to the user of the byproduct. • Fresh CKD is best if kept dry prior to use in a highway application. transformation of marginal materials No additional post-processing uses for transforming CKD into alternative products were noted in the literature or survey responses. design adaptations The following need to be considered when using CKD in highway applications: • CKD generally reduces the strength of PCC products. A combination of CKD and fly ash helps minimize the loss of strength. The best strengths are obtained when the CKD has a high CaO content and a low LOI. • CKD or CKD–fly ash decrease PCC workability and may require the use of superplasticizers in the PCC mix design. • The use of CKD improves soil properties such as plas- ticity and strength. Adding fly ash with the CKD proves further improvement. The increased strength of the soil is to be considered in designing applications. • The pH of water in contact with CKD stabilized soils will be increased. This could be considered during the project selection and design phases. construction issues CKD reactions will be slower in cold conditions; reactivity is improved with elevated curing temperatures. Failures, causes, and lessons learned None were described in the literature or the survey responses. Barriers The following barriers were identified: • The loss of reactivity of the CKD over time (weathered stockpiles) limits the use of landfilled CKD. • CKD chemical composition information from byproduct supplier. • CKD material specifications for individual highway applications (e.g., CaO and LOI properties for PCC producer). costs • Cost of landfilling needs to be high enough to encour- age CKD producers to initiate specific stockpiling and post-processing for highway applications. • About 15% CKD appears to be an upper limit for balanc- ing cost-effectiveness with desirable material properties. gaps The following gap was noted: • Further education for agencies for appropriate project selection (e.g., identification of soil properties good for stabilization with CKD) is needed. FIGURE 8 CKD research locations. AASHTO Method Title M145 Classification of soil and soil-aggregate mixtures for highway construction purposes T86 Investigations and sampling soils and rock for engineering purposes T87 Standard method of test for dry preparation of disturbed soil and soil-aggregate samples for test T88 Standard method of test for particle size analysis of soils T89 Standard method of test for determining the liquid limit of soils T90 Standard method of test for determining the plastic limit and plasticity index of soils T99 Standard method of test for moisture-density relations of soils using a 2.5 kg (5.5 lb) rammer and a 305 mm (12 in.) drop TABLE 34 AASHTO TEST METHODS USED WITH CKD ByPRODUCTS IN HIGHWAy APPLICATIONS

26 eltiT dohteM MTSA C204 Standard Test Methods for Fineness of Hydraulic Cement by Air-Permeability Apparatus C403 Standard Test Method for Time of Setting of Concrete Mixtures by Penetration Resistance D1633 Standard Test Methods for Compressive Strength of Molded Soil-Cement Cylinders D1833 Standard Test Method for CBR of Laboratory Compacted Soils D2487 Standard Practice for Classification of Soils for Engineering Purposes (Unified Soil Classification System) D422 Standard Test Method for Particle Size Analysis of Soils D4318 Standard Test Methods for Liquid Limit, Plastic Limit, and Plasticity Index of Soils D4647 Standard Test Method for Identification and Classification of Dispersive Clay Soils by the Pinhole Test D5050 Standard Guide for Commercial Use of Lime Kiln Dusts and Portland Cement Kiln Dusts D5102 Standard Test Method for Unconfined Compressive Strength of Compacted Soil-Lime Mixtures D5120 Standard Test Method for Inhibition of Respiration in Microbial Cultures in the Activated Sludge Process D560 Standard Test Methods for Freezing and Thawing Compacted Soil-Cement Mixtures D6023 Standard Test Method for Density (Unit Weight), Yield, Cement Content, and Air Content D6024 Standard Test Method for Ball Drop on Controlled Low Strength Material (CLSM) to Determine Suitability for Load Application D6103 Standard Test Method for Flow Consistency of Controlled Low Strength Material (CLSM) D6130 Standard Test Method for Determination of Silicon and Other Elements in Engine Coolant by Inductively Coupled Plasma-Atomic Emission Spectroscopy D6267 Standard Test Method for Using pH to Estimate the Soil-Lime Proportion Requirement for Soil Stabilization D6572 Standard Test Methods for Determining Dispersive Characteristics of Clayey Soils by the Crumb Test D698 Standard Test Methods for Laboratory Compaction Characteristics of Soils Using Standard Effort D854 Standard Test Method for Specific Gravity of Soil Solids by Water Pycnometer TABLE 35 ASTM TEST METHODS USED WITH CKD ByPRODUCTS IN HIGHWAy APPLICATIONS