Preliminary Examination of the System Fly Ash-Bottom Ash-Flue Gas Desulphurization Gypsum-Portland Cement-Water for Road Construction

is paper describes an investigation into the use of three power plant wastes: �y ash, �ue gas desulphurization gypsum, and bottom ash for subbase layers in road construction. Two kinds of mixtures of these wastes with Portland cement and water were made: �rst with �y ash consisting of coarser particles (<1.651mm) and second with �y ash consisting of smaller particles (<0.42mm). e mass ratio of �y ash-Portland cement-�ue gas desulphurization gypsum-bottom ash was the same (3 : 1 : 1 : 5) in both mixtures. For both mixtures, the compressive strength, the mineralogical composition, and the leaching characteristics were determined at di�erent times, 7 and 28 days, a�er preparation. e obtained results showed that both mixtures could �nd a potential use for subbase layers in road construction.


Introduction
Fly ash (FA), bottom ash (BA), and �ue gas desulphurization wastes (FGD gypsum) are three byproducts, that is, wastes, from power plants.e disposal of these wastes is an economic problem and an ecological problem (pollution of air, waters, and soil), and consequently there is intensive search to increase the use of these wastes as construction materials, especially if they are used in bulk (in the large volume building material industry, in civil engineering).
e application of wastes from power plants (FA, BA, and FGD gypsum) as construction materials is connected with their characteristics.
FA with a low CaO content (class F according to ASTM 618) may exhibit pozzolanic properties (itself possessing little or no cementing value, but, in the presence of moisture, it reacts chemically with calcium hydroxide at ambient temperature to form compounds that possess cementing properties) and could be applied as a substitute for Portland cement (PC) in concrete [1][2][3][4][5].FA with a high CaO content (class C, according to ASTM 618) has alone hardening properties in the presence of moisture, without the addition of calcium hydroxide [1].
BA is a more or less inert material; that is; bottom ash usually does not exhibit pozzolanic properties, but it could be a potential aggregate in concrete as a replacement for natural sand (its particles are coarse, sized between sand and gravel, fused, and with a glassy texture) [2,6,7].
FGD gypsum could be applied as chemical activator of FA in FA/cement system, based on the ability of gypsum to react with aluminium oxide in the glass phase of �y ash to produce ettringite that contributes to the strength in the early ages [8].
Fly ash from Serbian power plant "Nicola Tesla" (class F) is a powdered material consisting of small particles (<1.651 mm) with good pozzolanic properties, which means that it is a reactive material with potential suitability for use as a replacement for Portland cement.
Flue gas desulphurisation gypsum from the Bohemian Power Plant (Hvaletice) is a powdered material (its particles are smaller than 0.104 mm), mainly composed of CaSO 4 ⋅2H 2 O, with potential use as an FA activator.
Bottom ash from the Serbian power plant "Nicola Tesla" has a lower quality (high loss on ignition, low density, low bulk density, and low compactability) than natural aggregates (sand and gravel) and as such, has the potential for use as road subbases, when the strength requirements are not very high.e aim of this work was to investigate the possibilities of the potential utilisation of FA and BA from the Serbian power plant "Nicola Tesla" and FGD gypsum from the Bohemian Power Plant-Hvaletice (no Serbian plant has an FGD system installed yet) for subbase layers in road construction.
Due to the promised properties (good reactivity) of FA and FGD gypsum (exposed above) and for economic and ecological reasons (recycling of waste materials and avoidance of the necessity of their disposal), two mixtures with a low content of Portland cement and a high content of power plant wastes (FA, FGD gypsum, and BA) were prepared.ese mixtures had the same mass composition (FA : PC : FGD gypsum : BA = 3 : 1 : 1 : 5), but they were different in terms of the size of the FA particles.For both mixtures, the compressive strength, mineralogical composition, and leaching characteristics were determined 7 and 28 days aer preparation.ese properties are of primary importance for a preliminary examination and assessment for any possible application of the materials in construction.
e compressive strength is an essential parameter for mechanical characterization and indicator of the overall quality of cement materials.e mineralogical composition and leaching characteristics are very important properties for an assessment of the applicability of materials for construction.e mineralogical composition in�uences the compressive strength and durability of materials.e leaching characteristics predict the environmental impact of the materials.

Experimental
First, the chemical properties (chemical composition and mineralogical composition) of the materials (FA, BA, and FGD gypsum) were investigated.en, the compressive strength, mineralogical composition, and leaching behaviour of mixtures made from these materials and Portland cement (PC) and water (W) were determined.
e chemical compositions of materials (presented as oxide equivalents) were determined by classic chemical analysis.
e mineralogical compositions of FA and BA were determined by the powder XRD technique using a Philips PW 1729 X-ray generator and a Philips PW 1710 diffractometer.e radiation source was an X-ray LLF tube with copper radiation and a graphite monochromator.e radiation wavelength  CuK1 was 0.15405 nm.e anode tube load was 40 KV and 30 mA. e samples were pressed into standard aluminium frames and measured in the 2 range from 5 ∘ to 50 ∘ .e recorded data were interpreted using JCPDS cards.
FGD gypsum was examined by qualitative IR analysis.e IR absorption spectrum of FGD gypsum was recorded on a Perkin Elmer 782 spectrophotometer in the range from 4000 to 400 cm -1 , using the KBr pressed disc technique.e mixtures (pastes) �y ash-Portland cement-�ue gas desulphurization gypsum-bottom ash-water (FA-PC-FGD gypsum-BA-W) were made using (1) raw �y ash (consisting of particles smaller than 1.651 mm) and (2) sieved �y ash (consisting of particles smaller than 0.42 mm).e mass ratio of solid materials in these mixtures was the same (FA : PC : FGD gypsum : BA = 3 : 1 : 1 : 5).
To simulate a stabilization (necessary for the use of the mixtures in road construction) cylindrical specimens (diameter 10.2 cm and height 11.7 cm), so-called Proctor specimens, of both mixtures with the optimum moisture content were made by compaction.e values of the optimum moisture content for both mixtures (FA : PC : FGD gypsum : BA = 3 : 1 : 1 : 5) with FA of different particle size (smaller than 1.651 mm and smaller than 0.42 mm) were determined by the Standard Proctor test (ASTM D698).ese values were practically the same and about 38%.
Six Proctor specimens of these mixtures were stored for 7 days, and another six of one were stored for 28 days in a humid room (20 ± 2 ∘ C, 85% humidity).Aer storage, the following properties of the specimens were determined: compressive strength, mineralogical composition, leaching behaviour, water content, and absorption of water.e compressive strength was determined according to the Serbian Standard SRPS U.B1.030 (which is in accordance with European Standard EN 14227-4).
e values of compressive strength given in this work present an average value of the compressive strength of three specimens of both hardening mixtures aer 7 and 28 days.
Determination of mineralogical composition and leaching behaviour were performed with pulverized crushed (during the determination of their compressive strength) samples.e mineralogical composition of the samples was examined in the same manner as for FA and BA (described above).
e leaching behaviour of specimens was determined according to the method described in European Standard EN 12457.is method consists of placing 100 g of pulverised material in 1000 cm 3 of distilled water, agitating for 24 h (5 rpm) and then �ltering the material-water suspension.
In the present study, tap water was used as the liquid medium (pH = 6) instead of distilled water.e liquid phase (extract) was analyzed by atomic absorption spectroscopy (Perkin Elmer "Analyst 300" spectrophotometer).e measurements were performed in an air/acetylene �ame and an N 2 �/acetylene �ame.External standard solutions were prepared from 1000 mg/dm 3 stock metal solutions.e results of leaching test were compared with waste acceptance criteria (WAC) for granular wastes (limit values for the compliance test using EN 12457 at L/S = 10).
Water content and absorption of water of monolithic (noncrushed) cylindrical (Proctor) specimens were determined according to SRPS U.B1.012 and SRPS B.B8.010., respectively.ree specimens of both mixtures aer 7 and 28 days were dried to constant mass at 85 ∘ C. Based on the masses of the wet and dry specimens, the water content was calculated in % (as mean value of three measurements).
Subsequently, the same dry specimens were immersed in tap water for 24 h.e water absorption (average value in %) was calculated using the masses of the specimens aer and before immersion in water.

Results and Discussion
e content of main component (presented as oxide equivalents) of the materials is given in Table 1.
Based on the chemical analysis results, it is evident that the �y ash from the Nicola Tesla Power Plant could be classi�ed as a low calcium �y ash (class F, according to ASTM 618).It satis�es ASTM C618 Standard speci�cation for use as a pozzolan for replacement of Portland cement in concrete because its FAS content (Fe 2 O 3 + Al 2 O 3 + SiO 2 ) was greater (80.66%) than that required by ASTM 618 (70%).In addition, the contents of SO 3 (2.74%) and Na 2 O (0.41%), as well as the loss on ignition (1.85%) were lower than the allowed maximum values proposed by ASTM 618 (5%, 1.5%, and 6%, resp.) for the utilization of FA as a pozzolan.
e content of oxides: SiO 2 , Al 2 O 3 , and Fe 2 O 3 (49.39%,19.70%, and 5.83%, resp.) was lower but the loss on ignition was higher (8.91%) in the bottom ash than in the �y ash.
Based on the results of the chemical analysis of the FGD gypsum from the Power Plant "Hvaletice, " it is evident that it was a sufficiently pure substance (32.80%CaO, 45.72% SO 3 , 17.18% crystal water, and 4.3% impurities).
Based on the X-ray diffraction study, it could be stated that the major crystalline phase in the �y ash was quartz-SiO 2 .e other crystalline phases present in small amounts in the �y ash were mullite, Al e major crystalline phase in the bottom ash was quartz, then calcite, CaCO 3 , and small amounts of anhydrite, feldspar, diopside, and hematite.In addition to the mentioned crystalline phases, a signi�cant amount of amorphous aluminosilicates (glass) was evidenced by the broad "hump" (between 2  2 ∘ and 2  3 ∘ ) in the XRD diffractogram of the �y ash and bottom ash.e SiO 2 , present as quartz or in the crystalline aluminosilicates phase, is inert.Similarly, the part of the alumina in the crystalline phase is inactive.Both the silica and alumina in the amorphous aluminosilicates (glass) are reactive [9].
e results presented in Table 2 are the average value of three determinations of the compressive strength of the specimens S 1 (18.6%FA, 6.2% PC, 6.2% FGD gypsum, 31% BA, and 38% W, the FA consisted of particles smaller than 1.651 mm) and S 2 (18.6%FA, 6.2% PC, 6.2% FGD gypsum, 31% BA, and 38% W, the FA consisted of particles smaller than 0.42 mm) aer 7 and 28 days.e estimated standard deviation for the compressive strength is also given in Table 2.
e low values of standard deviations for the compressive strength of both samples (aer 7 and 28 days) show that there was no spread of the obtained compressive strength results.
e results in Table 2 show that the values of the compressive strength of the specimens were relatively low.e specimen S 2 (made with FA consisting of smaller particles) had greater compressive strength values than the specimen S 1 , both aer 7 and 28 days of hardening (by 6.5 and 3.9%, resp.).e reason for this is, probably, the higher reactivity of the smaller particles and consequently higher content of the products of hydration reactions (it is known that the compressive strength is in a functional relation with the content of the products of hydration reactions (12); the higher is their content, the greater is the compressive strength).Moreover, it is evident that the compressive strengths of both samples (S 1 and S 2 ) aer 28 days were considerable higher than those aer 7 days (by about two times).
For the assessment of the validity of the obtained results for compressive strength of the specimens S 1 and S 2 , they were compared with the criteria for compressive strength of subbase layers proposed by the Public Enterprise-Roads of Serbia (Table 3).
Comparing the obtained results (Table 2) with the criteria in Table 3, it could be concluded that both mixtures prepared in this study were suitable as a subbase only for heavy and medium traffic load.
e XRD diffractograms of the specimens aer 7 and 28 days of hardening are presented in Figure 1.
In the diffractograms of all specimens (Figure 1), the presence of a broad hump between 2 = 20-38 ∘ could be attributed to the presence of an amorphous C-S-H phase [11].
Quartz, mullite, and feldspar, detected in all specimens, are inherent to �y ash and bottom ash.ey are inert and did not participate in the reactions occurring in the system FA-PC-FGD gypsum-BA-W.Ettringite and C-S-H are the products of the hydration reactions that occurred.Ettringite is formed by reaction between calcium aluminate, from Portland cement and FA, and gypsum from FGD gypsum and Portland cement (a small amount of gypsum is present in Portland cement).
e reaction of ettringite formation can be written as [12] 3CaO ⋅ e liberated Ca(OH) 2 from reaction (2) could react with active silicates from �y ash (present in the amorphous phase of FA) forming thereby also calcium silicate hydrate (C-S-H).
e gypsum phase was present in all the specimens despite it is being a reactant in reaction (1); its quantity was even greater in the specimens aer 28 days than in the specimens aer 7 days (compare the peak intensities at   0.7630 nm and 2  .∘ in Figure 1; the peak at   0.4280, and 2  20.74 ∘ is not convenient for comparison because it overlaps with that of quartz at   0.427 nm and 2  20.8 ∘ ).In this way, it could be concluded that gypsum was formed in the specimens as a new (secondary) phase during carbonation of the specimens, via reaction with atmospheric CO 2 [13], according to the following equation [14]: (Some amount of the gypsum in the specimens may have originated from the FGD gypsum that did not react in the hydration reactions (1)).
Carbonation of the other product of hydration reaction, that is, C-S-H, very probably occurred, following the equation [14,15]: Carbonation of the specimens could be con�rmed by comparing the main peaks intensities of ettringite at 0.9650 nm (2  .6 ∘ ), and, gypsum at 0.7630 nm (2  .∘ ).e other main peak of gypsum at 0.4280 nm (2  20.74 ∘ ) is very near to the peak of quartz at 0.4257 nm (2  20.8 ∘ ), and, for this reason, it could not be taken into consideration (as was stated above).e peak intensities (in cps) of ettringite (2  .6 ∘ ) and gypsum (2  .∘ ), taken from the X-ray diffractograms of the specimens (Figure 1), are presented in Table 4.According to literature data (18)(19)(20), different forms of CaCO 3 (calcite, vaterite, aragonite, and amorphous) appear during C-S-H gel carbonation.Due to this, carbonation could not be estimated based on the intensities of the calcite peak, and hence, for this reason, the intensity of the main peak of calcite (2 = 29.6 ∘ ) is not presented in Table 4.
As can be seen from Table 4, the intensities of the peak of gypsum increased with time for both mixtures, which indicated carbonation of these mixtures (3).e decrease of the peaks of ettringite with time in the �rst mixture also  indicated to carbonation phenomena.e increase in the size of both the peak of ettringite and of gypsum with time for second mixture could be explained by the higher reactivity of the smaller FA particles.e higher reactivity of the smaller FA particles resulted in the formation of a larger content of both ettringite and gypsum.
e results of the leaching test (European Standard EN 12457), that is, the concentrations of the trace elements in the extracts of powdered specimens (determined by atomic absorption spectroscopy) are presented in Table 5.In addition, the Waste Acceptance Criteria (WAC) for granular wastes and the limit values for compliance with EN12457 test at L/S = 10 are also included in this table.
e results showed that both mixtures could be considered as hazardous material regarding the leached amounts of Cd, Pb, Ni, and Sb.
Comparing the metal concentrations in the extracts of the pulverized hardened specimens (S 1 and S 2 ) aer 7 days with those aer 28 days, a signi�cant increase in the concentration of Sb, moderate increases in Ni and Cr, and minor increase in the Cd concentrations in the extracts of both specimens aer 28 days were found (Table 5).
ese results could be correlated with the results of the X-ray analysis.e formation of ettringite by the hydration reaction (1), evidenced by X-ray analysis in the specimens aer 7 days, can immobilise metal ions.e immobilisation is achieved by the substitution of Ca 2+ , Al 3+ , and SO 4 2− ions in the ettringite crystal structures (Ca 6 Al 2 (SO 4 ) 3 (OH) 12 ⋅26H 2 O) by metal ions and metal oxyanions [16,17].e carbonation phenomena and decomposition of ettringite (3) in specimens aer 28 days (con�rmed by the augmentation of the content of gypsum, see favour the breakdown of the ettringite structure, which affects the leaching processes and the transfer of metal ions and oxyanions to the leachates.e average values (of three determination) of water content and absorption of water of monolithic (noncrushed) cylindrical (Proctor) specimens, aer 7 and 28 days, are presented in Table 6.
e result presented in Table 6 shows that the water content was higher, but the absorption of water lower aer 28 days than aer 7 days in both specimens.ese results also con�rmed the carbonation of the specimens.During carbonation, the ettringite decomposed with the release of a lot of water and, consequently, the specimens aer 28 days had a greater content of water than those aer 7 days.In addition, it is known [18,19], that, during carbonation, the structural properties of the specimens improve, for example, their pore volume decreases which are associated with the deposit of the formed CaCO 3 .For this reason, the absorption of water by S 1 and S 2 (Table 6) was lower for the specimens aer 28 days than for the specimens aer 7 days.

Conclusions
(1) Regarding the compressive strength values, the mixture FA-PC-FGD gypsum-BA with the mass composition 3 : 1 : 1 : 5 satis�ed the criteria for subbase layers for heavy and medium traffic load which were proposed by the Public Enterprise-Roads of Serbia.
(2) e examinations realised in this study indicated that carbonation reactions occurred in the mixtures FA, PC, FGD gypsum, BA, and W (3 : 1 : 1 : 5), in addition to hydration reactions.e results that con�rmed this phenomenon were (a) augmentation of the peaks at 0.7630 nm in the X-ray diffractograms of the specimens that could only be explained by the formation of gypsum as a new, second phase (different from starting FGD gypsum), (b) considerable improvement of the compressive strength of the specimens in the period from 7 to 28 days (the compressive strength increased about two times), which could be related to CaCO 3 formation during carbonation (CaCO 3 has a very low solubility and would contribute to a clogging of the pores and in this way improve the compressive strength), (c) increase in the water content in the monolithic cylindrical specimens during the period from 7 to 28 days, which is very probably related to the release of high amounts of water during the reaction of ettringite carbonation (3), (d) decrease in water absorption by the monolithic cylindrical specimens during the period from 7 to 28 days, which is very probably a consequence of the decrease in the pore volume caused
FA consisted of smaller particles (<0.42 mm), aer 7 days

Table 4 )
T 5: Results of the batch leach test and regulatory limits for the determined elements.