Fuel 194 (2017) 157–165

Contents lists available at ScienceDirect

Fuel

journal homepage: www.elsevier .com/locate / fuel

Full Length Article

Determination of the initial ash sintering temperature by coldcompression strength tests with regard to mineral transitions

http://dx.doi.org/10.1016/j.fuel.2016.12.0660016-2361/� 2017 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.E-mail address: ronny.schimpke@iec.tu-freiberg.de (R. Schimpke).

Ronny Schimpke ⇑, Mathias Klinger, Steffen Krzack, Bernd MeyerDepartment of Energy Process Engineering and Chemical Engineering, TU Bergakademie Freiberg, Fuchsmühlenweg 9, Reiche Zeche, 09596 Freiberg, Germany

h i g h l i g h t s

� Initial ash sintering temperatures were detected by cold compression strength tests.� Three different ashes were investigated at oxidizing, inert, and reducing conditions.� Sintering relevant phases were found by SEM/EDX, high temperature XRD, and TG-DSC.� Silicates, iron-aluminum-oxides, carbonation, and iron sulfide melts induce sintering.� Sintering can be connected with melt formation or solid phase sintering.

a r t i c l e i n f o

Article history:Received 1 September 2016Received in revised form 18 December 2016Accepted 20 December 2016

Keywords:CoalCombustionGasificationAsh depositionAgglomerationFoulingAsh sinteringCold compression strength

a b s t r a c t

Ash deposition, fouling, and slagging are commonly undesirable occurrences in coal utility boilers andgasifiers. Ash deposit formation is usually initiated by sintering of the particles. The sintering character-istics of the ash have to be evaluated to understand the process of deposit formation. Cold compressionstrength (CCS) tests on heat treated ash pellets produced under defined conditions can be applied for thedetermination of the initial sintering temperature (IST). Scanning electron microscopy (SEM) incombination with energy dispersive X-ray spectroscopy (EDX), high temperature X-ray diffraction anal-ysis (HT-XRD), and thermogravimetric differential scanning calorimetry (TG-DSC) are analytical methodsto characterize the mineral transitions responsible for sintering.CCS tests were done on ashes of three different coals to determine their IST. For this purpose ash pellets

were pretreated at different temperatures. The influence of oxidizing, inert, and reducing conditions atambient pressure was investigated. The IST were compared with the characteristic temperatures fromash fusibility tests (AFT). HT-XRD, TG-DSC, and SEM/EDX were applied for a more detailed characteriza-tion of sintering.The IST determined by CCS tests are considerably lower than the respective initial deformation temper-

ature (IDT) obtained by AFT. Thus, the AFT is not suitable to detect IST. Major mineral transitions werefound which are responsible for initial sintering in the case of the investigated ashes. The formation ofsilicates or the crystallization of iron-aluminum-oxides were found to induce sintering at oxidizingand inert conditions. Sintering at reducing conditions was initiated by carbonation of calcium com-pounds, silicate formation, or melting of iron sulfide. These phases can be understood as the relevantphases for initial strengthening of fouling deposits in coal utility boilers and gasifiers in the case of theinvestigated ashes.

� 2017 Elsevier Ltd. All rights reserved.

1. Introduction

The thermochemical conversion of coals, e.g. for syngas produc-tion or energetic utilization, involves various drawbacks that comealong with the coal’s ash behavior like slagging and fouling. Hence,

the knowledge of the temperature of agglomeration onset gainshigh importance. The initial sintering temperature (IST) is com-monly used to specify the initiation of ash particle agglomeration[1]. Sintering describes a heat induced mass transport in a looseor packed bed of particles [2]. The particle bed is strengthenedand shrinks due to inter-particle material bonds and filling of thevoid fraction. The driving force is the reduction of the free surfaceenergy, which is kinetically controlled by the temperature [3]. The

Table 1X-ray fluorescence analysis of the 500 �C ashes.

LBK SAK HOK

CO2 5.39 6.38 20.10Na2O 0.06 0.68 4.10MgO 9.12 3.08 14.96Al2O3 3.79 21.50 3.78SiO2 13.78 44.24 0.82SO3 21.57 1.75 11.90Cl 0.23 – 0.42

K2O 0.29 0.69 0.60CaO 23.09 7.35 31.39TiO2 0.19 2.17 0.21

Fe2O3 21.95 10.20 13.46BaO 0.11 0.29 0.30P2O5 0.03 1.18 0.03

Traces 0.41 0.49 0.39

RB=A 3.07 0.32 13.43

Nomenclature

AbbreviationsAFT ash fusion test (–)BSE back scattered electron (–)CCS cold compression strength (–)EDX energy dispersive X-ray spectroscopy (–)FT fluid temperature (�C)HT hemispherical temperature (�C)HT-XRD high temperature X-ray diffraction (–)IDT initial deformation temperature (�C)IST initial sintering temperature (�C)SEM scanning electron microscopy (–)

ST softening temperature (�C)TG-DSC thermogravimetric differential scanning calorimetry (–)

SymbolsA0 base area (m3)Fmax maximum force (N)DRH

�m reaction enthalpy (J/mol)

RB=A base to acid ratio (–)x50;3 median particle diameter (m)rD compression strength (Pa)

158 R. Schimpke et al. / Fuel 194 (2017) 157–165

temperature at which the surface energy starts to decrease due tomass transport can be called the initial sintering temperature.Above this temperature the ash particles tend to agglomerate.The determination of the IST and its relation to mineral transitionsare the main focus of the present work. A well established experi-mental investigation to determine ash melting characteristics isthe ash fusibility test (AFT) according to international standards(e.g., ASTM D 1857 [4], DIN 51730 [5], ISO 540 [6]). These stan-dards mention the initial deformation temperature (IDTAFT) as thelowest characteristic temperature. However, agglomeration mayalready occur several hundred degrees below the IDTAFT, as Stall-mann and Neavel observed [7,8]. Huffman et al. [9] observed firstmolten phases already 200–400 K below IDTAFT, which lead toash sintering and agglomeration. Hence, a reliable IST cannot bedetected by AFT.

Several approaches can be found for the determination of theIST. Neuroth et al. [10] or Bartels et al. [1] present overviews aboutpossible methods: Dilatometry [1,11,3], combined differentialthermal analysis and thermogravimetric analysis [12,13], thermalconductivity analysis [14,15], electric conductance [16], thermo-chemical equilibrium calculations [17] or yield stress [1,18]. Theresults of the IST vary between the different methods. This makesthe IST a value which is specific to the applied method.

In this work the main focus lies on the cold compressionstrength (CCS) test of previously heat treated ash pellets. Pelletizedash samples are treated at different temperature levels for a speci-fic dwell time. After cooling down of the pellets to ambient tem-perature they are tested for their compression strength. Drawingthe compression strength over temperature an increase in thecompression strength indicates the ISTCCS. Similar experimentswere carried out, e.g., by Hupa et al. [19], Ots and Zelkowski [20],Nel et al. [12], Skrifvars [21,22], and Al-Otoom et al. [14]. TheCCS measurements are widely accepted as a reliable indicator forthe IST [1].

For this reason this method is used in the present work for thedetermination of the ISTCCS of three different coal ashes. The ISTCCSwill be compared with IDTAFT and a self-defined ISTAFT.

Additionally to the temperature, the surrounding atmospherehas a strong influence on the sintering occurrences of ash systems.In iron containing ashes and reducing conditions, e.g., elementaliron and wustite (FeO) are formed instead of hematite (Fe2O3)[23]. Whereas, wustite is a strong fluxing agent [24,9]. Wustiteand iron sulfides, such as pyrite (FeS2) and pyrrhotite (Fe1�xS), forman eutectic at a temperature of 910 �C [25,26]. Together with cal-cium sulfide wustite forms an eutectic at 890 �C [26]. Furthermore,iron silicates can be built with relatively low melting temperaturesin the presence of silicon oxide. To account for the influence of theatmosphere the following experiments were conducted in oxidiz-ing, inert, and reducing atmosphere. Air was used for oxidizing,

nitrogen (99.999% purity) for inert, and a mixture of 40 vol.% car-bon dioxide in carbon monoxide for reducing conditions. The sin-tering tendencies at different atmospheres were evaluated withthe support of high temperature X-ray diffraction analysis (HT-XRD), thermogravimetric differential scanning calorimetry (TG-DSC) and scanning electron microscopy (SEM). The results will becompared with the standardized AFT.

2. Material characterization

Three ashes from different coals were analyzed for their sinter-ing tendency. The coals are a Lusatian lignite (LBK), a hearth fur-nace coke made of Rhenish lignite (HOK) and a South Africanhardcoal (SAK). The coals were ashed in a muffle furnace at500 �C to avoid devolatilization of chlorides, alkaline, and alkalineearth metals [27,28,1] and to suppress the decomposition of car-bonates and sulfates [29]. Thus, components which essentiallyaffect the sintering behavior remain inside the ash. The resultingashes consist of fine particles with diameters of x < 300 lm andmedian diameters of x50,3 = 19 lm (LBK), x50,3 = 13 lm (SAK) andx50,3 = 10 lm (HOK) (optical measurement with a CamSizer XT,Retsch Technology).

2.1. Elemental composition

Table 1 shows the elemental distribution of the three ashes andthe resulting base-to-acid ratio

RB=A ¼ Fe2O3 þ CaOþMgOþ Na2Oþ K2OSiO2 þ Al2O3 þ TiO2

: ð1Þ

R. Schimpke et al. / Fuel 194 (2017) 157–165 159

The base-to-acid ratio refers to a completely oxidized ash at 500 �C.The values of carbon dioxide result from carbonates, which remainstable at an ashing temperature of 500 �C in contrast to regular ash-ing at 815 �C. Also the fractions of chlorine and sodium oxide wouldtypically be lower at 815 �C. LBK and HOK show high amounts ofbasic oxides like calcium oxide, iron(III) oxide, and magnesiumoxide together with high amounts of sulfur trioxide. Calcium andiron can work as intense network modifiers, especially in reducingatmosphere. Whereas, the high sulfur contents involves the possi-bility of the formation of low melting sulfidic or sulfatic eutectics[30]. The HOK is strongly basic with a base to acid ratio ofRB=A = 13.43 due to very low amounts of the acidic oxides silicondioxide and aluminum oxide but high contents of magnesium oxideand calcium oxide. The LBK is less basic with RB=A = 3.07 because of asilicon dioxide content of 13.78 wt.%. The SAK is strongly acidic(RB=A = 0.32) with high amounts of silicon dioxide and aluminumoxide. However, the base to acid ratio only considers alumosilicatenetworks. The possibility of sulfidic or sulfatic melt occurrence isneglected.

2.2. Mineral composition

The mineral compositions of the 500 �C ashes were obtained byX-ray diffraction (XRD) applying a Bruker D8-II Discover with acobalt tube. The <10 lm ground ashes were mixed with 10 wt.%of zinc oxide as a standard. The diffractograms were measured inthe range of 15–80� 2h. In Table 2 it can be seen that in all ashescalcium is mainly bound in anhydrite and carbonates (calcite, dolo-mite/ankerite). The sulfur, verifiable by X-ray diffraction, is boundsulfatic. The LBK is dominated by sulfates, iron oxides, and quartz.The SAK has the main components quartz, iron oxdides, and anamorphous fraction of 37.1 wt.%, which is completely composedof amorphous metakaolin, as SEM/EDX reveals. Similar observa-tions were made by Matjie et al. [31] who investigated a SouthAfrican hardcoal of comparable composition. The HOK is stronglydominated by carbonates with some periclase and iron oxides.

Table 2Mineral composition of the 500 �C ashes in wt.%.

Designation Formula

Periclase MgOQuartz SiO2

Anhydrite CaSO4

Hematite Fe2O3

Maghemite/Magnetite Fe2O3/Fe3O4

Calcite CaCO3

Brownmillerite/Srebrodolskite Ca2(Al,Fe)2O5/Ca2Fe2O5

Dolomite/Ankerite CaMg(CO3)2/CaFe(CO3)2Sodium sulfate Na2SO4

Grossite CaAl4O7

Amorphous phase

Table 3Ash fusion test of the 500 �C ashes.

�C Oxidizing

LBK SAK HOK LBK

ISTAFT 760 856 840 840IDTAFT 1352 1254 1340 1270STAFT 1385a,b 1273 1350a,b 1355HTAFT 1388 1281 1465c 1371FTAFT 1407 1318 >1600 1375

a No distinct spherical shape.b Height considerably less than the base.c No distinct hemispherical shape.

2.3. Ash fusion test

Ash fusion tests were done according to DIN 51730 [5]. The onlydifference is the usage of 500 �C ashes. The results of the ash fusiontests are listed in Table 3. Additionally to the DIN 51730, an ISTAFTis defined by the onset of shrinking of the sample. It has to be men-tioned that shrinking may take place due to decomposition reac-tions, like carbonate decomposition, which actually cause nosintering effects. However, the ISTAFT gives a first indicative valueand is used for comparison. In reducing conditions the initial defor-mation temperature (IDTAFT), the softening temperature (STAFT), thehemispherical temperature (HTAFT), and the fluid temperature(FTAFT) are lower than in oxidizing and inert conditions for all ofthe examined ashes. However, the ISTAFT is only distinctly lowerin the case of the both basic and sulfatic ashes LBK and HOK. TheISTAFT of the acidic SAK is hardly effected by the atmosphere.

3. Experimental

3.1. High temperature X-ray diffraction

For the determination of phase reactions during heat treatmentof the ashes X-ray diffraction analyses were applied at elevatedtemperatures. For this purpose also a Bruker D8-II Discover witha cobalt tube was applied. Additionally, a TCP-radiation heatingchamber from mri Physikalische Geräte GmbH and a Vantec-2000 2D detector were used. For each sample the compositionsin oxidizing, inert, and reducing conditions were measured. Thechamber was flushed with the particular gas mixture during themeasurements. No standard could be added to avoid reactionswith ash components. For each test run a thin isopropanol-ash-suspension was placed on a platinum strip heater. Diffrac-tograms were taken between 500 and 1200 �C with steps of100 K. During the cool down steps of 200 K were used. An angularrange of 25–80� 2hwas applied with a recording time of 20 min foreach temperature.

LBK SAK HOK

4.2 – 17.911.7 23.5 –36.3 4.8 3.90.5 7.6 –

20.9 7.7 9.49.8 5.6 51.4– – 2.7– 9.5 –– – 3.4– 4.2 –

16.5 37.1 11.3

Inert Reducing

SAK HOK LBK SAK HOK

883 840 526 825 6911250 1330 1291 1153 12641275 1340a,b 1330a,b 1218 1350b

1286 1424c 1351c 1235 1383c

1355 >1600 1400 1279 1484

160 R. Schimpke et al. / Fuel 194 (2017) 157–165

3.2. Thermogravimetric differential scanning calorimetry

The formation of first melts is endothermic and not connectedwith a mass change. Other ash phase reactions are either endother-mic or exothermic and may come along with a mass change. Eachoccurrence can be observed by thermogravimetric differentialscanning calorimetry (TG-DSC), which measures the sample massand heat flux during heating. For this purpose a Netzsch STA 409was applied. The samples with a net weight between 10 and12 mg were heated to 1200 �C with a rate of 10 K/min. The furnacewas flushed with the particular gas mixture.

3.3. Cold compression strength tests

The dried ash was compressed with a force of 6 MN/m2 to formcylindrical pellets with 15 mm diameter and 15 mm height. Thepellets were subsequently measured in height, diameter, and mass.Respective target temperatures were reached with 10 K/min andhold for 24 h, while flushing with the particular gas. The targettemperatures were set by 100 K increments beginning at 500 �C.After cooling down the pellets were measured again in height,diameter, and mass. Possible changes in shape due to sinteringeffects made special measurement techniques necessary. First,the base areas had to be abraded in some cases to obtain two evensurfaces in parallel to each other (based on DIN EN 993-5 [32]). Thediameter was measured on both base areas at two positionsorthogonal to each other. The four achieved values were averagedand applied to the calculation of the base area A0 [32].

The test equipment applied for the CCS tests was a ‘‘inspectmini 3 kN” from Trilogica GmbH. Its maximum test force is 3 kNwith a distance resolution of 1 lm. Two parallel plates cover thecylindrical pellet with the upper moving down with a constantrate. For an online measurement of the force acting on the pelletthe upper plate is equipped with a resistive wire strain. Force-distance-diagrams can be plotted, where the maximum force Fmax

represents the fracture of the pellet. Fmax is utilized for the calcula-tion of the compression strength rD

rD ¼ Fmax

A0; ð2Þ

with the pellet base area A0. The abort criterion of the strength testwas set to 20% in force loss and 33.3% loss in pellet height. The cho-sen values for the dwell time at a set temperature, the pellet height,and the applied strain rate are based on preliminary tests. Thosewere done on ash of a comparable hearth furnace coke, which can

0

2

4

6

8

10

12

14

16

18

20

500 600 700 800 900 1000 1100 1200

com

pres

sion

stre

ngth

inM

Pa

temperature in °C

linear fluctua�ng

linear(*): ISTCCS = 1034 °Cfluctua�ng(+): ISTCCS = 850 - 900 °C

(*)

(+)

Fig. 1. Examples of linear and fluctuating CCS results in the case of the LBK at inert(linear) and reducing (fluctuating) conditions and the procedure of ISTCCSdetermination.

be found in an earlier work [33]. The compression strength meanvalue and the standard deviation of six pellets were obtained forevery temperature. The results were drawn in a diagram with thecold compression strength related against temperature. The com-pression strength either increased almost linear or showed fluctuat-ing values at temperatures above the ISTCCS. Two examples can befound in Fig. 1, where the curves of the LBK at inert (linear) andreducing (fluctuating) conditions are drawn. In the case of a linearincrease the ISTCCS was calculated by the intersection of two linearregression lines. A temperature range was determined for fluctuat-ing compression strength values between the temperature of thefirst increased value and the lower adjacent temperature.

4. Results and discussion

The results of the CCS and the TG-DSC are presented in the fol-lowing. The diffractograms of the HT-XRD runs can be found in thesupplementary material (Section A) and are considered in the dis-cussions. Additionally, several heat treated ash pellets were pre-pared for scanning electron microscopy in back scatteredelectron mode (BSE). Pellets prepared in the regions of the ISTCCSwere chosen for this purpose. Those were fixed with epoxy resinand cut to prepare polished thin sections. Mentioned mineralphases were indicated by energy dispersive X-ray spectroscopy(EDX), according to the phases found by HT-XRD. Oxidizing andinert conditions are conjoined in the discussions, since they inducesimilar ash reactions.

4.1. Lusatian lignite (LBK)

4.1.1. Oxidizing and inert conditionsIn oxidizing and inert conditions the CCS of the pellets, as

shown in Fig. 2(a), increase simultaneously with a formation of sil-icates, as obtained by HT-XRD. The ISTCCS could be determined bylinear regression in both cases. It is 1008 �C for oxidizing condi-tions and 1034 �C for inert conditions. In comparison to the ashfusion tests (AFT) these values are 248 and 194 K higher than theISTAFT as well as 344 and 236 K lower than the IDTAFT for oxidizingand inert conditions, respectively. At these atmospheres the deter-mination of a clear increase in endothermic behavior to indicatethe formation of melt would be incorrect, due to overlapping ashreactions and melt formation.

In both oxidizing and inert conditions the HT-XRD of the LBKshows the decomposition of a small amount of carbonates between600 and 700 �C. This results in the formation of calcium oxide andcarbon dioxide. This is reflected by endothermic peaks and a massloss in the TG-DSC (Fig. 2(b)). Calcium and magnesium silicates likeakermanite/gehlenite (Ca2MgSi2O7/Ca2Al2SiO7), enstatite (Mg2Si2-O6), larnite (Ca2SiO4), and diopside/hedenbergite (CaMgSi2O6/CaFeSi2O6) can be found in the HT-XRD at elevated temperaturesstarting at 1000 �C. This goes along with a dissipation of the previ-ously formed calcium oxide. Additionally, the anhydrite decompo-sition occurs above 1150 �C and 1050 �C at oxidizing and inertconditions, respectively, resulting in a further formation of calciumsilicates. Anhydrite is decomposed by the following equation [34]:

CaSO4 � CaOþ SO2 þ 12O2 DRH

�m ¼ þ502:2 kJ=mol ð3Þ

The gas release is indicated by a mass loss in the TG curve. The TGcurve also reflects the influence of the oxygen content of the sur-rounding atmosphere. The anhydrite decomposition is slightlyinhibited in air. This results in a small residual amount of anhydriteat 1200 �C in oxidizing conditions compared to a complete decom-position in inert conditions.

These observations are reflected by the SEM pictures (Fig. 2(c)).In both oxidizing and inert conditions calcium silicates are formed

0

2

4

6

8

10

12

14

16

18

20

200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400

com

pres

sion

stre

ngth

inM

Pa

temperature in °C

oxidizing inert reducing

oxidizing: ISTCCS = 1008 °Cinert: ISTCCS = 1034 °Creducing: ISTCCS = 850 - 900 °C

-4.5

-3.5

-2.5

-1.5

-0.5

0.5

1.5

2.5

3.5

4.5

5.5

50

55

60

65

70

75

80

85

90

95

100

200 300 400 500 600 700 800 900 1000 1100 1200

DSC

inm

W/m

g

rel.

mas

sin

wt.-

%

temperature in °C

oxidizing inert reducing

anhydrite-decomp.

red. anhydrite-decomposi�on

calcitedecomposi�on

calciteforma�on

TG

DSC FeS mel�ng

exo.

S-elimina�on:Fe1-xS FeS

(1)

(2)(3)

(1)

(2)

Fig. 2. Experimental results for the LBK.

R. Schimpke et al. / Fuel 194 (2017) 157–165 161

at the boundaries of quartz particles. These reaction rims arealready observed at 950 �C, which is even lower than the obtainedISTCCS. No evidence for the existence of molten phases can befound, but solid bridges between particles have been observed.Thus, the occurrence of solid phase sintering can be assumed attemperatures close to the ISTCCS.

4.1.2. Reducing conditionsThe CCS tests at reducing conditions show fluctuations in com-

pression strengths (Fig. 2(a)). Thus, linear regression is not suitablein that case. A first increased compression strength at 900 �C leadsto an ISTCCS of 850–900 �C. This is at least 324 K higher than ISTAFTand 391 K lower than IDTAFT. The responsible component for sin-tering is iron(II)-sulfide (FeS). In pellets, treated at 900 and1000 �C, solidified melts of iron sulfide are found by SEM/EDX, asshown in Fig. 2(d). This agrees with the beforehand mentionedeutectic at 910 �C in a FeO-FeS-system [25,26]. A matrix and sinter-ing bridges of liquid iron sulfide are formed and lead to an increasein compression strength.

HT-XRD reveals that anhydrite is decomposed at temperaturesbelow 700 �C in reducing conditions. According to Kuusik et al.[35] two reaction mechanisms may take place:

CaSO4 þ CO � CaOþ SO2 þ CO2 DRH�m ¼ þ219:2 kJ=mol ð4Þ

CaSO4 þ 4CO � CaSþ 4CO2 DRH�m ¼ �171:1 kJ=mol ð5Þ

The decomposition more likely proceeds according to Eq. (5) withincreasing carbon monoxide content [35]. Because a high carbonmonoxide content was present during the experiments, thisreaction is assumed to take place. This is confirmed by the massloss and the exothermic peak in the TG-DSC curves between 500and 600 �C, shown in Fig. 2(b). The small mass increase between600 and 650 �C is due to the conversion of calcium sulfide intocalcite and iron sulfide. SEM pictures of a 650 �C sample show sev-eral iron sulfide particles of the composition Fe1�xS, with x beingin the range of 0.2–0.4 (see Section A). The HT-XRD showsdiffraction patterns representing hexagonal pyrrhotite at 600,700, and 800 �C. Sulfur is released as carbonyl sulfide (COS) atabout 720 �C by forming stoichiometric pyrrhotite with x=0, asSEM/EDX of a 800 �C sample reveals. Oldhamite (CaS) is formedabove 800 �C due to the thermal decomposition of calcite. Theincreased carbon dioxide content (40 vol.%) in the reducingatmosphere forces the calcite decomposition to higher tempera-tures compared to oxidizing and inert conditions. The TG-DSCshows that the calcite decomposition occurs at about 820 �C (seeFig. 2(b)). The largest fractions of oldhamite are present at 900and 1000 �C. It is involved into the production of silicates like lar-nite, akermanite/gehlenite, diopside/hedenbergite, and merwinite(Ca3Mg(SiO4)2) at higher temperatures. Some quartz particlesshow boundaries of sodium potassium silicates in samples treatedat 800 and 900 �C in reducing conditions. However, it seems thatthose silicates do not induce sintering.

162 R. Schimpke et al. / Fuel 194 (2017) 157–165

4.2. South African hardcoal (SAK)

Well-defined straight regression lines could be drawn for thethree different atmospheres of the SAK. Exothermic peaks of com-parable reaction heat and without mass changes are measured inall atmospheres at 990 �C by TG-DSC. Since akermanite/gehleniteis the only silicate that was detected at 1000 �C in all atmospheresby HT-XRD, the formation of this mineral is made responsible forthe exothermic effects. However, the CCS tests of the SAK (Fig. 3(a)) show that sintering obviously occurs way below this intensesilicate formation in all atmospheres.

4.2.1. Oxidizing and inert conditionsThe behavior of the SAK at inert conditions is also comparable

to oxidizing conditions. The ISTCCS are 787 �C at oxidizingconditions and 742 �C at inert conditions. Hence, these values are424–508 K lower than the IDTAFT and 200–250 K lower than theakermanite/gehlenite peak in the DSC curve. Even the ISTAFT is69–141 K higher than the ISTCCS. The SEM picture of the sinteredpellet in Fig. 3(c), heat treated at 900 �C in oxidizing conditions,demonstrates the solid phase reactions that already took place.Boundaries of calcium silicates are formed around quartz andmetakaolin particles. Thus, mobile calcium phases are present withthe capability to form sintering necks resulting in an increase ofcompression strength.

In both atmospheres a carbonate decomposition occursbetween 600 and 710 �C, shown by an endothermic mass loss in

0

2

4

6

8

10

12

14

16

18

20

200 300 400 500 600 700 800 900 1000 1100 1200 1300

com

pres

sion

stre

ngth

inM

Pa

temperature in °C

oxidizing inert reduzing

oxidizing: ISTCCS = 787 °Cinert: ISTCCS = 742 °Creducing: ISTCCS = 729 °C

1

rel.

mas

sin

wt.-

%

(1)

(2)

Fig. 3. Experimental re

the TG-DSC curves of Fig. 3(b). The HT-XRD reveals that calciumoxide remains in the ash and begins to react with quartz and meta-kaolin to form akermanite/gehlenite, diopside/hedenbergite, anor-thite (CaAl2Si2O8), and sillimanite (Al2SiO5) above 900 �C. A silicatemelt forms below 1200 �C, which is indicated by a crystallization ofanorthite during cool down. Magnetite/maghemite is oxidized tohematite at temperatures lower than 800 �C. The hematite is stableup to 1100 �C and seems not to be involved into any reactions rel-evant for initial sintering effects. A decrease in hematite is mea-sured at 1200 �C by means of HT-XRD. Anhydrite is decomposedbetween 1000 and 1100 �C under inert and oxidizing conditions,respectively. Sulfur oxides are released to the gas phase, while cal-cium is bound in calcium silicates. The TG-DSC curves in Fig. 3(b)reveal similar trends of the anhydrite decomposition in oxidizingand inert atmosphere as the LBK showed. The decomposition isslightly suppressed in air. Therefore, it is shifted to a highertemperature.

4.2.2. Reducing conditionsThe ISTCCS at reducing conditions is 729 �C, which is only

slightly lower compared to oxidizing and inert conditions. It is96 K lower than ISTAFT and 424 K lower than IDTAFT. No noticeablereaction heat can be observed in the TG-DSC up to 850 �C as shownin Fig. 3(b), but a mass loss of about 6 wt.% beginning at around600 �C. From 600 to 850 �C a reduction of the iron oxides hematiteand magnetite/maghemite to wustite occurs according to the HT-XRD. These reductions are of low reaction enthalpies and show

-3.5

-3.0

-2.5

-2.0

-1.5

-1.0

-0.5

0.0

0.5

80

85

90

95

00

200 300 400 500 600 700 800 900 1000 1100 1200

DSC

inm

W/m

g

temperature in °C

oxidizing inert reducing

anhydrite-decomp.

calcitedecomposi�on

TG

DSC

exo.

silicate forma�on

(1)

(2)

(3)

sults for the SAK.

R. Schimpke et al. / Fuel 194 (2017) 157–165 163

no intensive DSC signal. The calcite decomposition is forced tohigher temperatures in the reducing atmosphere, as alreadyobserved for the LBK. Dolomite/ankerite is decomposed below800 �C, which goes along with the formation of calcite, as theHT-XRD shows. In the TG-DSC experiments the calcite decomposi-tion starts at about 850 �C and is finished at 890 �C. The residualcalcium oxide was measured in the HT-XRD as lime at 1000 and1100 �C. It is incorporated into the silicate formation above900 �C where akermanite/gehlenite and huge amounts of anorthiteand sillimanite are formed. The anhydrite decomposition proceedsgradually. Anhydrite still can be detected at higher temperaturesup to 1000 �C in the HT-XRD. Thus, it is present at temperaturesof about 400 K higher compared to the LBK. Kuusik et al. [35]and Wheelock and Boylan [36] mentioned the dependence ofanhydrite decomposition on the accompanying ash componentsand the sulfur oxides in the surrounding atmosphere. The formeris the main reason for the different occurrences between the LBKand the SAK. That means that anhydrite and calcite are hardlydecomposed below 850 �C. With regard to sintering occurrences,this results in a lack of calcium available for the formation ofcalcium silicates. Instead, higher amounts of forsterite/fayalite(Mg2SiO4/Fe2SiO4) were measured in the HT-XRD above 900 �C.The wustite content gradually decreases between 900 and 1200 �C.

Fig. 3(d) shows a SEM picture of a pellet heat treated in reducingconditions at 800 �C. Even at this low temperature fayalite-like sil-icates could be measured at quartz and metakaolin boundaries byEDX. It can be seen that sufficient iron diffusion into surroundingquartz and metakaolin particles occurred around wustite grains.This leads to a formation of iron silicates with the ability to formsintering bridges and to increase the compression strength of thesample, although a lack of calcium is present below 850 �C.

4.3. Hearth furnace coke (HOK)

The CCS test results of the HOK are drawn in Fig. 4(a). The esti-mation of the ISTCCS is not possible by regression lines because ofthe fracture forces exceeding the equipment‘s maximum measur-able force. Hence, the ISTCCS is determined by the first measuredincrease of the compression strength at all conditions.

4.3.1. Oxidizing and inert conditionsISTCCS of 700–750 �C and 800–850 �C are obtained for oxidizing

and inert conditions, respectively. In both cases the compressionstrength increases only slightly up to 900 �C. The ISTCCS were indi-cated by an increase without overlapping error bars. The obtainedvalues are up to 140 K lower compared to the ISTAFT. Similar to theLBK and the SAK, a big difference exists in comparison to the IDTAFT,which is 450–590 K higher. First stabilization of the pelletsoccurred due to crystallization spots of brownmillerite/srebrodol-skite at about 800 �C. This crystallization continues with increasingtemperature leading to a network of crystallized brownmillerite/srebrodolskite, as can be seen in Fig. 4(c), which shows a sampletreated at 1000 �C in oxidizing conditions. A less intense formationof brownmillerite/srebrodolskite at inert conditions is reflected bylower compression strength values, as shown in Fig. 4(a).

The TG-DSC curves of the HOK are given in Fig. 4(b). Only thedecomposition of calcite caused intensive signals in oxidizingand inert conditions. It is indicated by an endothermic effect andmass loss of about 22 wt.% between 600 and 700 �C. The HT-XRDshows that anhydrite is decomposed below 1000 �C at both atmo-spheres. Because of the low amount it is hardly measured by theTG-DSC. The formation of brownmillerite/srebrodolskite andcalcium oxide starts at 600 �C. The fraction of brownmillerite/sre-brodolskite increases with temperature up to 1000 �C. This goesalong with a decrease in the contents of calcium oxide, hematite,and magnetite/maghemite between 600 and 800 �C. While iron

oxides could not be measured anymore at 800 �C, calcium oxideconsumption continues up to 1200 �C, where a part of it is still pre-sent. At inert conditions less calcium oxide is consumed up to1200 �C resulting in less brownmillerite/srebredolskite formation.Above 1100 �C brownmillerite/srebrodolskite decrease consider-ably. The formation of melt is a possible explanation, since no fur-ther phases can be measured and an endothermic behavior can beseen in the DSC curve above 900 �C. Additionally, low amounts ofpseudowollastonite (CaSiO3) and akermanite/gehlenite are formedbetween 1000 and 1200 �C.

In the work of Neuroth [18] comparable coal ashes were exam-ined on their sintering temperatures in oxidizing conditions. Thesintering temperatures were in the range of 700–800 �C, which isin very good agreement with the obtained values of the presentstudy. Neuroth [18] found that mainly sodium and calcium sulfaticmelts are responsible for initial sintering of the ashes. Calcium sul-fate and sodium sulfate were both also detected in the HOK used inthis study. It neither can be proven nor excluded whether thosemelts play a role in the initial sintering of the HOK or not.

4.3.2. Reducing conditionsReducing conditions lead to an intensive increase in strength of

the pellets already starting at 700 �C. At 800 and 900 �C the maxi-mum measurable force of 3 KN is exceeded, which makes linearregression impossible. Thus, an ISTCCS of 650–700 �C is determinedfor the HOK in reducing conditions. The calcite content measuredby HT-XRD starts to decrease at 700 �C and gradually declines untilno calcite was measured at 900 �C. At 800 �C crystalline calcite isreduced to about a half of its initial content. Whereas, no calciumoxide and only a few brownmillerite/srebrodolskite is formed. Cal-cium oxide appears at 900 �C and remains stable until 1200 �C.Only little of it seems to be involved into the formation of brownmillerite/srebrodolskite. The SEM picture (Fig. 4(d)) shows a pelletheat treated at 800 �C in reducing conditions. A huge intercon-nected network of calcium-magnesium-carbonate can be seen,where calcium is dominant. Particles like oldhamite, periclase(MgO), wustite, and brownmillerite/srebrodolskite are incorpo-rated into this network. Electron back scatter diffraction indicatedthat most of the network is of an amorphous structure with crys-talline calcite portions enclosed. The carbonate formation is achemical reaction sintering between gas components and solidash material. Anthony et al. [37] and Skrifvars et al. [22] describethis effect as molecular cramming or chemical hardening. Limereacts with the carbon dioxide in the surrounding gas to form car-bonates. That way pores, interstitial spaces, and junctions betweenparticles are filled with carbonates. The resulting network is thereason for the intensive increase in compression strength between650 and 700 �C. Skrifvars et al. [22] also measured an increase incompression strength of a mixture of lime and petrol coke ash trea-ted in a carbon dioxide rich gas atmosphere. A compressionstrength increase was detected between 600 and 700 �C.

Anhydrite is already decomposed at 600 �C as the HT-XRD and alittle exothermic mass loss in the TG-DSC show. But no sulfur con-taining phase is determined at 600 �C. Sulfur seems to be bond inthe amorphous phase, because the crystalline sulfidesCa0:17 Mg0:83S and oldhamite (CaS) do not appear below 700 �Cand 800 �C, respectively.

The iron oxides are reduced to wustite at temperatures above700 �C. Part of it is incorporated into iron carbide (Fe2C), whichwas measured at 600, 700, and 800 �C. A small amount of complexoxides like brownmillerite/srebrodolskite and CaFe5O7 is formed inreducing conditions. Strong endothermic effects can be seen over awide temperature range beginning at about 700 �C. The endother-mic mass loss between 720 and 760 �C is due to the sulfur releaseof pyrrhotite, as it is detected for the LBK. Between 780 and 840 �Cthe endothermic iron carbide decomposition and between 820 and

0

5

10

15

20

25

30

35

200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400

com

pres

sion

stre

ngth

inM

Pa

temperaturein °C

oxidizing inert reducing

(1) max. measurable force

oxidizing: ISTCCS = 700 - 750 °Cinert: ISTCCS = 800 - 850 °Creducing: ISTCCS = 650 - 700 °C

-4.5

-3.5

-2.5

-1.5

-0.5

0.5

1.5

2.5

3.5

4.5

5.5

50

55

60

65

70

75

80

85

90

95

100

200 300 400 500 600 700 800 900 1000 1100 1200

DSCi

nmW

/mg

rel.m

assi

nwt.-

%

temperaturein°C

oxidizing inert reducing

red. anhydrite-decomposi�on

calcitedecomposi�on

TG

DSC

exo.

S-elimina�on:Fe1-xS FeS

reduc�on ofiron oxides

(1)(1)

Fig. 4. Experimental results for the HOK.

500

600

700

800

900

1000

1100

1200

1300

1400

1500

1600

Oxid. Inert Red. Oxid. Inert Red. Oxid. Inert Red.

tem

pera

ture

in °C

LBK SAK HOK

ISTCCS ISTAFT IDTAFT

Fig. 5. Comparison of the ISTCCS with sintering relevant temperatures determinedby AFT.

164 R. Schimpke et al. / Fuel 194 (2017) 157–165

850 �C the endothermic calcite decomposition occur. An endother-mic effect without mass loss reflects the melt formation of pyrrho-tite starting at 910 �C. This melt further strengthens the ash pelletsabove 900 �C.

5. Conclusion

The measurement of CCS was confirmed to be a valuablemethod for the determination of initial sintering temperatures. It

could be shown that the combination with TG-DSC, HT-XRD, andSEM/EDX is an efficient approach to investigate initial sinteringcharacteristics. However, it still has to be differentiated betweenthe initialization of sintering by mass transport and a strengthen-ing of the sample. Sintering initiating mass transport may alreadyoccur a few degree below the ISTCCS. Not all influences on ash sin-tering could be considered in the present work. For example, theparticle size which affects solid-solid contacting and reaction, orhigher dwell times that might further decrease the ISTCCS.

Fig. 5 shows a comparison of the obtained ISTCCS with the sinter-ing relevant temperatures of the AFT. The highest ISTCCS wereobtained for the LBK, whereas the HOK and the SAK start sinteringat similar temperature ranges. Because the HOK and the SAK areeither highly basic or acidic, a dependence of the IST on the basicitycannot be confirmed. First mobile phases were detected by SEM attemperatures below the ISTCCS in the case of the LBK. The ISTAFT ofthe LBK in the three considered atmospheres, which are estimatedby a loss of sample height, are considerably low for an actual sin-tering. In the case of the SAK the ISTCCS are much lower than anysintering relevant appearances detected in the AFT, HT-XRD orTG-DSC. Smaller differences between the ISTAFT and the ISTCCS wereobtained for the HOK. Hence, a sufficient comparability betweenCCS and AFT exists only in the case of the HOK in all atmospheres.All considered samples and atmospheres have in common that theIDTAFT is at least 200 K higher than the ISTCCS. Eventually, it meansthat the AFT delivers insufficient results for the identification ofany sintering occurrences. AFT results have to be confirmed byother measurement techniques to determine IST.

R. Schimpke et al. / Fuel 194 (2017) 157–165 165

It could be shown that solid phase sintering was present at oxi-dizing and inert conditions in terms of silicate formation (LBK,SAK) and crystallization of brownmillerite/srebrodolskite (HOK).Furthermore, calcite and iron sulfide were detected as the reasonfor sintering effects in reducing conditions for HOK and LBK,respectively. Two important reactions include the decompositionof anhydrite and calcite, which provide lime for further reactionswith other ash components. It can be assumed that a solid phasesintering usually takes place at temperatures some degrees lowerthan first melts appear.

The relevance of solid phase sintering decreases for practicalapplications with increasing relative velocities and decreasing con-tact time of the particles. However, these temperatures will resultin the strengthening of loose fouling deposits in coal utility boilersand gasifiers, leading to a continuous increase in thickness. Thiswill already occur at temperatures where no melts are present –only induced by solid phase sintering.

Acknowledgments

The authors gratefully acknowledge the financial support of theGerman Federal Ministry of Economics and Energy and the PoernerGroup in the framework of the COORVED project (R&D number0327865). Special thanks are due to Marcus Schreiner for varioushints and productive discussions.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.fuel.2016.12.066.

References

[1] Bartels M, Lin W, Nijenhuis J, Kapteijn F, van Ommen JR. Agglomeration influidized beds at high temperatures: mechanisms, detection and prevention.Prog Energy Combust Sci 2008;34(5):633–66.

[2] Salmang H, Scholze H, Telle R, et al. Keramik. seventh ed. Berlin: Springer;2007. ISBN 978-3-540-63273-3.

[3] Siegell JH. High-temperature defluidization. Powd Technol 1984;38(1):13–22.[4] ASTM International D05 Committee, ASTM D1857: Test Method for Fusibility

of Coal and Coke Ash, 2010, doi:10.1520/D1857_D1857M-04R10.[5] Deutsches Institut für Normung e.V., DIN 51730: Prüfung fester Brennstoffe -

Bestimmung des Asche-Schmelzverhaltens, September 2007.[6] International Organization for Standardization, ISO 540:2008(E): Hard coal

and coke - Determination of ash fusibility, 2008.[7] Stallmann JJ, Neavel RC. Technique to measure the temperature of

agglomeration of coal ash. Fuel 1980;59(8):584–6.[8] Vélez JF, Chejne F, Valdés CF, Emery EJ, Londoño CA. Co-gasification of

Colombian coal and biomass in fluidized bed: an experimental study. Fuel2009;88(3):424–30.

[9] Huffman GP, Huggins FE, Dunmyre GR. Investigation of the high-temperaturebehaviour of coal ash in reducing and oxidizing atmospheres. Fuel 1981;60(7):585–97. http://dx.doi.org/10.1016/0016-2361(81)90158-7.

[10] Neuroth M, Schreck T, Simmat R, Nover G, Müller M, Muhammadieh M, et al.Studies of coal ash sintering in lignite combustion. VGB PowerTech 2011;91(11):83–93.

[11] He Y. Characterisation of spouting behaviour of coal ash with thermo-mechanical analysis. Fuel Process Technol 1999;60(1):69–79. http://dx.doi.org/10.1016/S0378-3820(99)00024-7.

[12] Nel MV, Strydom CA, Schobert HH, Beukes JP, Bunt JR. Comparison of sinteringand compressive strength tendencies of a model coal mineral mixture heat-treated in inert and oxidizing atmospheres. Fuel Process Technol 2011;92(5):1042–51.

[13] Van der Drift A, Olsen A, Prij J, Beesteheerde J, der Klein C. Conversion ofbiomass, prediction and solution methods for ash agglomeration and relatedproblems. Project Report of ECN.

[14] Al-Otoom AY, Bryant GW, Elliott LK, Skrifvars BJ, Hupa M, Wall TF.Experimental options for determining the temperature for the onset ofsintering of coal ash. Energy Fuels 2000;14(1):227–33.

[15] Rezaei HR, Gupta RP, Bryant GW, Hart JT, Liu GS, Bailey CW, Wall TF, MiyamaeS, Makino K, Endo Y. Thermal conductivity of coal ash and slags and modelsused. Fuel 2000;79(13):1697–710.

[16] Raask E. Sintering characteristics of coal ashes by simultaneous dilatometry-electrical conductance measurements. J Thermal Anal Calor 1979;16(1):91–102.

[17] Zevenhoven-Onderwater M, Backman R, Skrifvars BJ, Hupa M. The ashchemistry in fluidised bed gasification of biomass fuels. Part I: predicting thechemistry of melting ashes and ash-bed material interaction. Fuel 2001;80(10):1489–502.

[18] Neuroth M. Thermochemical simulation and experimental investigation ondeposit formation in coal utility boilers in the case of rhenish lignite: Finalreport of the research project 0326887; 2007.

[19] Hupa M, Skrifvars BJ, Moilanen A. Measuring the sintering tendency of ash by alaboratory method. J Inst Energy 1989;62(452):131–7.

[20] Ots A, Zelkowski J. Fusion and sintering characterization of coal ashes. In:Proceedings of the 11th international conference on coal science. San Francisco(USA). p. 1–4.

[21] Skrifvars BJ, Öhman M, Nordin A, Hupa M. Predicting bed agglomerationtendencies for biomass fuels fired in FBC boilers: a comparison of threedifferent prediction methods. Energy Fuels 1999;13(2):359–63.

[22] Skrifvars B-J, Hupa M, Anthony EJ. Mechanisms of bed material agglomerationin a petroleum coke-fired circulating fluidized bed boiler. J Energy ResourTechnol 1998;120(3):215–8. http://dx.doi.org/10.1115/1.2795038. ISSN 0195-0738.

[23] Zelkowski J. Kohlecharakterisierung und Kohleverbrennung: Kohle alsBrennstoff, Physik und Theorie der Kohleverbrennung, Technik. Essen,Germany: VGB PowerTech Service GmbH; 2004.

[24] Jing N, Wang Q, Luo Z, Cen K. Effect of different reaction atmospheres on thesintering temperature of Jincheng coal ash under pressurized conditions. Fuel2011;90(8):2645–51. http://dx.doi.org/10.1016/j.fuel.2011.04.013. , <http://www.sciencedirect.com/science/article/pii/S0016236111002201>.

[25] McLennan AR, Bryant GW, Bailey CW, Stanmore BR, et al. An experimentalcomparison of the ash formed from coals containing pyrite and sideritemineral in oxidizing and reducing conditions. Energy Fuels 2000;14(2):308–15. http://dx.doi.org/10.1021/ef990092h. ISSN 0887-0624.

[26] Allibert M. Slag atlas, 2nd ed. Düsseldorf: Verl. Stahleisen; 1995 [ISBN 3-514-00457-9].

[27] Glazer MP, Khan NA, de Jong W, Spliethoff H, Schürmann H, Monkhouse P.Alkali metals in circulating fluidized bed combustion of biomass and coal:measurements and chemical equilibrium analysis. Energy Fuels 2005;19(5):1889–97.

[28] Bläsing M, Müller M. Mass spectrometric investigations on the release ofinorganic species during gasification and combustion of German hard coals.Combust Flame 2010;157(7):1374–81.

[29] Bellin A, Filla L, Meyer B, Engelhardt J, Adlhoch W. Einfluss derMineralstoffzusammensetzung von Braunkohle auf das Schmelz- undVerschlackungsverhalten bei deren Veredelung. EUR(Luxembourg) ISSN1018-5593.

[30] Li S, Whitty KJ. Investigation of coal char–slag transition during oxidation:effect of temperature and residual carbon. Energy Fuels 2009;23(4):1998–2005. http://dx.doi.org/10.1021/ef801082j.

[31] Matjie RH, Li Z, Ward CR, French D. Chemical composition of glass andcrystalline phases in coarse coal gasification ash. Fuel 2008;87(6):857–69.http://dx.doi.org/10.1016/j.fuel.2007.05.050. <http://www.sciencedirect.com/science/article/pii/S001623610700275X>.

[32] Deutsches Institut für Normung e.V., DIN EN 993-5: Prüfverfahren für dichtegeformte feuerfeste Erzeugnisse - Teil 5: Bestimmung der Kaltdruckfestigkeit,Dezember 1998.

[33] Schimpke R, Krzack S, Bräutigam N, Meyer B. Determination of coal ashsintering characteristics by compression strength test at differentatmospheres. In: Proceedings of the 31th Pittsburgh coal conference. p. 1–6.

[34] Swift WM, Panek AF, Smith GW, Vogel GJ, Jonke AA. Decomposition of calciumsulfate: a review of the literature; 1976, doi:http://dx.doi.org/10.2172/7224692 <http://www.osti.gov/scitech/servlets/purl/7224692>.

[35] Kuusik R, Salkkonen P, Niinistö L. Thermal decomposition of calcium sulphatein carbon monoxide. J Therm Anal 1985;30(1):187–93. http://dx.doi.org/10.1007/BF02128129. ISSN 0022-5215.

[36] Wheelock TD, Boylan DR. Reductive decomposition of gypsum by carbonmonoxide. Ind Eng Chem 1960;52(3):215–8. http://dx.doi.org/10.1021/ie50603a023. ISSN 0019-7866.

[37] Anthony EJ, Iribarne AP, Iribarne JV. A new mechanism for FBC agglomerationand fouling in 100 percent firing of petroleum coke. J Energy Resour Technol1997;119(1):55–61. http://dx.doi.org/10.1115/1.2794223. ISSN 0195-0738.