Predicting the steam drying behavior of brown coals and lignites

Fuel 159 (2015) 345–353

Contents lists available at ScienceDirect

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

Predicting the steam drying behavior of brown coals and lignites Stephen Niksa ⇑, Balaji Krishnakumar Niksa Energy Associates LLC, 1745 Terrace Drive, Belmont, CA 94002, USA

h i g h l i g h t s  Our drying model explicitly accounts for monolayer, multilayer, and bulk moisture in low-rank coals.  Moisture forms are estimated from a database of equilibrium moisture contents.  The drying analysis is validated with laboratory tests on steam drying in autoclaves and a fluidized bed.  One key aspect is whether or not a target moisture level requires removal of any monolayer moisture at all.  The three forms of moisture in any particular coal sample are determining factors for drying times.

a r t i c l e

i n f o

Article history: Received 16 April 2015 Received in revised form 7 June 2015 Accepted 9 June 2015 Available online 16 June 2015 Keywords: Drying Moisture Low-rank coals Steam drying technology

a b s t r a c t Our drying model explicitly accounts for the three forms of moisture in low-rank coals – monolayer, multilayer, and bulk – as a basis to predict the distinctive drying behavior and energy requirements of individual low-rank coal samples. The mass loadings of the moisture forms are estimated from a database of equilibrium moisture contents that were correlated to parameters from a coal’s proximate and ultimate analyses. The drying analysis is validated with laboratory tests on steam drying in autoclaves and a fluidized bed that cover broad ranges of steam pressures and temperatures for an assortment of Australian brown coals. Then the analysis determines which forms of moisture and operating conditions most affect drying times. For some selections of samples, drying time will be proportional to total moisture, and for others, there will be exceptions. One key aspect is whether or not the target moisture level requires removal of any monolayer moisture at all. Coals with abundant bulk and multilayer moisture can achieve targets on residual moisture by vaporizing only a portion of the multilayer moisture. Consequently, their drying times will be relatively modest. But coals with little bulk moisture and relatively abundant monolayer moisture will require relatively long drying times simply because monolayer moisture is always released much more slowly than multilayer moisture. Hence, the levels of the three forms of moisture in any particular coal sample are determining factors for drying times. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction The very high moisture levels in brown coals and lignites are problematic because they raise transport costs, lower calorific values, and complicate handling and grinding operations. An assortment of technologies has been developed to remove most of the moisture at both mines and power plants, as surveyed recently [1]. The study in this paper supports steam fluidized bed drying (SFBD) technology [2–6], in which millimeter-sized fuel particles are fluidized in slightly superheated steam at near-atmospheric pressures from 120 to 250 °C to bring down the moisture levels ⇑ Corresponding author. Tel.: +1 650 654 3182; fax: +1 650 654 3179. E-mail address: [email protected] (S. Niksa). http://dx.doi.org/10.1016/j.fuel.2015.06.031 0016-2361/Ó 2015 Elsevier Ltd. All rights reserved.

to about 10 wt.%. Two main advantages are that the drying atmosphere is inert, which eliminates spontaneous combustion hazards, and that most of the heat of vaporization can be recovered by circulating steam from the product stream through a compressor and heat exchanger in the fluidized bed. The technological imperatives have already spawned numerous computational schemes to predict the operating conditions needed to achieve a target dryness in the product [7–14], and to identify the best factors for process management and optimization [15–20]. Collectively, this work provides essential information on transport coefficients and the relevant heat transfer mechanisms, but it is limited by the universal premises that (i) the moisture in brown coals and lignites is a homogeneous phase that occupies the internal porosity; and (ii) its evaporation can be described with

346

S. Niksa, B. Krishnakumar / Fuel 159 (2015) 345–353

Nomenclature a Ai C Cp,i Di Ei h

DHi DHVAP ki mP M0 ni p r rb RDRY Ri t tbMAX

initial particle radius, cm pseudo-frequency factor for release of multi- or monolayer moisture, g/cm3 s constant in the BET adsorption isotherm, Eq. (8) specific heat of a region of coal plus moisture form i = b, M, m, cal/g K mass diffusivity of multi- and monolayer moisture along pore walls, cm2/s activation energy for release of multi- or monolayer moisture, kJ/mole convective heat transfer coefficient at the external particle surface, cal/cm2 s K specific enthalpy of vaporization of multi- or monolayer moisture, cal/g standard specific enthalpy of vaporization of bulk moisture, cal/g overall mass transfer coefficient for multi- and monolayer moisture from the external surface, g/cm2 s particle mass, g initial mass fraction of total moisture power-law temperature coefficient in Ri, Eq. (3) pressure, MPa radial position, cm position of the bulk vaporization front, cm drying rate, g moisture/s rate of multi- or monolayer moisture release, g/cm3 s time, s time when the bulk vaporization front arrives at the center of the particle and vanishes, s

the mathematical analyses for drying that are routinely applied to most other granular, porous materials. As explained in more detail elsewhere [21,22], there are actually three distinct types of moisture in very low-rank coals: (i) Monolayer moisture fully interacts with functional groups in the coal structure via hydrogen bonds and ionic bonding complexes; (ii) Multilayer moisture forms a hydrogen bonded layer over monolayer moisture via long-range interactions; and (iii) Bulk moisture is simply water that volumetrically fills the bulk of the pore voidage, without chemical interactions with coal components. Monolayer and multilayer moisture are primarily stabilized by metal carboxylates, carboxylic acids, and phenolic hydroxyls in the coal structure. Moisture on metal carboxylates is ionically bound and therefore immobile, whereas moisture on carboxylic acids and phenolic hydroxyls is hydrogen bonded and fully mobile, as are the acidic protons and the hydroxyl protons in the coal structure. A thermochemical equilibrium determines the concentrations of metal carboxylates, carboxylic acids, and the concentrations of metallic cations dissolved in the monolayer and multilayer moisture. This equilibrium may be affected by the pH of the bulk moisture, such that bulk moisture acts as a reservoir that responds to changes in the proportions of metal carboxylates and carboxylic acids. During drying, the three forms of moisture are released sequentially [21]. Bulk moisture is released via evaporation and requires only the normal enthalpy of vaporization. Multilayer moisture is released by supplementing the enthalpy of vaporization with the free energy of formation of the hydrogen bonded multilayer. Monolayer moisture is released by supplementing the enthalpy of vaporization with the energy requirement to break the hydrogen bonds to phenolic hydroxyls and carboxylic acids, and to convert the metal carboxylates into an organic crosslink plus a free metallic

tMMAX tPMAX tM10% T xS yi yi,mono EQ

yi

time to eliminate multilayer moisture, s duration of the preheating stage, s time to dry to 10 wt.% moisture, s temperature, K steam concentration, as a mole fraction mass fraction of moisture of form i mass fraction of a monolayer of multi- or monolayer moisture in the BET isotherm, Eq. (8) mass fraction of multi- or monolayer moisture on the external surface in adsorption equilibrium with the drying environment

Greek Symbols ai thermal diffusivity of the coal particle plus moisture form i, cm2/s gDRY drying efficiency ki thermal conductivity of the coal particle plus moisture form i, cal/cm s K qi density of the coal particle plus moisture form i, g/cm3 Subscripts b region with bulk, multilayer, and monolayer moisture f free stream condition M region with multilayer and monolayer moisture m region with monolayer moisture only 0 initial condition

acid. During drying at temperatures above about 150 °C, metal carboxylates, carboxylic acids, and phenolic hydroxyls spontaneously decompose into CO2, H2O, and organic crosslinks within coal macromolecules. At temperatures above about 250 °C, these decompositions become extensive enough to affect the thermochemical equilibrium that determines the concentrations of cations, carboxylic acids and carboxylates dissolved in multi- and monolayer moisture. Unfortunately, we were unable to locate a single quantitative analysis of brown coal or lignite drying that explicitly recognized the different forms of moisture, and their distinctive drying behavior. Bulk moisture vaporization is the only evaporation mechanism in all the available models. Here we show that these different forms of moisture are an essential aspect of the drying behavior of these fuels, without which there is little hope of describing the distinctive behavior of individual fuel samples. This paper introduces the elements of our drying analysis; validates the model predictions in comparisons with several reported drying histories from lab-scale steam drying experiments; and uses parametric studies to identify the factors that most strongly affect drying times.

2. Mathematical analysis 2.1. Phenomenology The reaction system is a single, spherical particle of very low rank coal with specified proximate and ultimate analyses and particle diameter. Initially, the particle holds specified levels of monolayer, multilayer, and bulk moisture, expressed as fractions of the total fuel mass. At some specified time, the particle is injected into

347

S. Niksa, B. Krishnakumar / Fuel 159 (2015) 345–353

qf

qf

Monolayer, Multilayer, and Bulk Moisture

a

a

rb Monolayer, Multilayer, and Bulk Moisture

PREHEATING

Monolayer & Multilayer Moisture

BULK VAPORIZATION

qf

rm a REACTION CONTROL

Monolayer & Multilayer Moisture

Monolayer Moisture

Fig. 1. Three regimes of brown coal drying.

a drying environment with specified uniform temperature, pressure, and steam concentration. The description of the drying environment must determine the external heat flux to the particle. For example, in a fluidized bed dryer, sufficient information would need to be provided to evaluate an overall heat transfer coefficient. The external heat flux would then be evaluated as the product of the heat transfer coefficient and the instantaneous temperature difference between the particle and fluidizing gas. As seen in Fig. 1, the drying process proceeds through three regimes: (1) preheating; (2) bulk vaporization; and (3) drying under Chemical Reaction Control. Preheating simply increases the particle temperature while the particle absorbs the external heat flux without releasing appreciable moisture. Temperature profiles are established from the external surface to the center while the external heat flux is conducted through the coal and moisture. The preheating stage lasts as long as necessary to bring the external surface to the boiling point for the specified drying conditions, where the first bulk moisture begins to vaporize. As soon as the external surface reaches the boiling point, a bulk moisture vaporization front starts to move toward the center. The conduction flux into the front from the exterior shell provides sufficient energy to vaporize only the bulk moisture, and to continue to conduct energy toward the center. The net conduction flux across the vaporization front equals the latent heat of vaporization, so the front remains at the boiling point while it moves toward the center.

The exterior shell beyond the vaporization front continues to conduct the external heat flux, but at a slower rate due to the replacement of bulk water by steam within the internal pore system, which lowers the thermal conductivity. The steam escaping through the shell also prevents multilayer moisture from evaporating, even while the temperature through the outer layer continues to increase via heat conduction. Only after the bulk vaporization front vanishes at the particle center does multilayer moisture start to evaporate. At this point, the mass fraction of multilayer moisture is fixed at the initial value throughout the particle. It then diminishes at the fastest rate at the external surface where the local temperature is hotter, while the drying endotherm suppresses radial temperature gradients. Eventually, the concentration of multilayer moisture vanishes from the external surface, which leaves only monolayer moisture. This region expands into an exterior shell due to the continuous conduction of the external heat flux. So there are now two domains, a core that contains a concentration profile of multilayer moisture on a uniform mass fraction of monolayer moisture and a shell that contains a profile of monolayer moisture. Since both regions have concentration profiles determined by the decomposition rates of bound moisture, this period is labeled as the Reaction Control regime in Fig. 1. Within the boundary between these regions, the mass fraction of multilayer moisture increases for progressively smaller radial positions, and the mass fraction of monolayer moisture remains at its initial value. On this boundary,

348

S. Niksa, B. Krishnakumar / Fuel 159 (2015) 345–353

Table 1 Domains in time and radial extent, and moisture release terms for the four stages of moisture release. Stage

Domains

STi

SM i

Time

Radius

Preheating Bulk Moisture

0
0
0 0 0

0 0 0

Multilayer Monolayer

tbMAX < t < tMMAX tMMAX < t < tmMAX

0 < r < rM 6 a rM < r < rm 6 a

RM (TM, yM)DHM/qMCp,M Rm(Tm, ym) DHm/qmCp,m

RM(TM, yM)/qM Rm(Tm, ym)/qm

there is no multilayer moisture but the mass fraction of monolayer moisture remains at the initial level. Beyond this boundary, the amount of monolayer moisture diminishes for progressively greater radial positions. In this configuration, monolayer moisture persists at the initial level until all multilayer and bulk moisture have been released from a given position. Similarly, multi- and monolayer moisture persist at their initial levels until all bulk moisture has been released from the entire particle, due to an equilibration of the mono- and multilayer moisture with the escaping steam from the bulk vaporization front. In this way, this analysis depicts the sequential elimination of the three forms of moisture. 2.2. Conservation equations The sequence of events in the previous section is implemented with pairs of conservation equations for energy and moisture for each of the four stages: preheating, bulk moisture vaporization, and release of multi- and monolayer moisture. All energy balances are transient equations in the radial direction of the following form:

@T i 1 @ @T ¼ 2 ai r2 i  STi r @r @t @r

where i ¼ b; M; m

ð1Þ

where Ti is absolute temperature; t is time; r is radial position; ai is thermal diffusivity; and STi is the enthalpy absorption rate for moisture release. The absolute temperature is a function of time and radial position, Ti(r,t). Note that only three subscripts are used to label the four stages in this analysis. This is because ‘‘b’’ denotes regions that contain bulk, multilayer, and monolayer moisture, which arise during both preheating and bulk moisture release. Similarly, ‘‘M’’ denotes regions with multi- and monolayer moisture, whereas ‘‘m’’ denotes regions with only monolayer moisture. The domains in time and radial position and the enthalpy absorption rates for the four stages are collected in Table 1. Preheating occurs throughout the entire particle, and lasts until the outer surface reaches the boiling point, which is denoted as tPMAX. Since preheating releases no moisture, there is no enthalpy requirement. Bulk moisture release partitions the particle into a shell and core, and lasts until the vaporization front reaches the center at tbMAX. Since bulk moisture is only released at the vaporization front, there is no enthalpy requirement within the shell of remaining bulk moisture. But this enthalpy requirement does factor into the following flux condition that determines the instantaneous location of the vaporization front:

dr @T ðr ; tÞ @T ðr ; tÞ qb;0 yb;0 DHVAP b ¼ kM M b þ kb b b @r @r dt

ð2Þ

where rb is the position of the vaporization front; subscript ‘‘0’’ denotes the initial values for density, qb, and bulk moisture wt. fraction, yb; DHVAP denotes the normal heat of vaporization at the boiling point for the ambient conditions; and ki denotes thermal conductivity. During the reaction control stage, the radial domain is divided into a core of radius rM that contains both multi- and monolayer moisture, and a shell with only monolayer moisture. Similarly,

once multilayer moisture has completely evaporated, a core of radius rm contains only monolayer moisture, and the surrounding shell is in adsorption equilibrium with the ambient conditions. The enthalpy absorption rates for multi- and monolayer moisture release contain distinctive release rates and endotherms. The release rates are given by

  Ei n y Ri ¼ Ai T i i exp  RT i i

where i ¼ M; m

ð3Þ

where Ai is a pseudo-frequency factor; ni is a power-law exponent; and Ei is a desorption energy. For both forms of moisture, the assigned values are 1.5 for ni and 4.2 kJ/mole for Ei, which are within the range of values associated with a physical adsorption/desorption processes with weak chemical interactions. The two pseudo-frequency factors are the only adjustable parameters in the analysis. Both values were set to interpret the datasets in this paper, and are the same for all coal samples. The values are 1  107 g/cm3/s for multilayer moisture and 5  108 for monolayer moisture. The vaporization endotherms for multi- and monolayer moisture were enhanced by 5% and 25%, respectively, over the normal enthalpy of vaporization, based on reported values [23]. In sensitivity studies, variations in the enhancements to DHi of ±5% did not significantly affect the drying times suggesting that the release rates for multi- and monolayer moisture are the predominant factors. The boundary conditions on the energy balance are symmetry in the radial profiles at the particle center, and a convective heat flux into the external surface, according to

@T i ð0; tÞ @T i ða; tÞ ¼ 0 and ki ¼ hðT f  T b ða; tÞÞ @r @r

ð4Þ

where Tf is the temperature in the free stream and h is the convective heat transfer coefficient. Throughout the release of bulk moisture, the mass loss can be evaluated from the movement of the bulk vaporization front, according to

    dmp dr b with mp tMAX ¼ qb;0 yb;0 4pr 2b ¼ mp;0 P dt dt

ð5Þ

where mP is the instantaneous particle mass. For multi- and monolayer moisture, the moisture analysis is analogous to the thermal analysis and of the form

@yi 1 @ @y Di r 2 i  SM ¼ 2 i r @r @t @r

where i ¼ M; m

ð6Þ

where Di are mass diffusivities; and SM i are the moisture release rates specified in Table 1 and Eq. (3). The diffusivities were set to very low values, because monolayer moisture is actually immobile, and surface diffusion of multilayer moisture is relatively slow. The boundary conditions on the moisture fraction equations are symmetry in the radial profiles at the particle center, and a convective mass flux off the external surface, according to

  @yi ð0; tÞ @y ða; tÞ ¼ 0 and qi Di i ¼ ki EQ yi ðxS;f ; T f ; pf Þ  yi ða; tÞ @r @r

ð7Þ

349

S. Niksa, B. Krishnakumar / Fuel 159 (2015) 345–353

where ki is a convective mass transfer coefficient and EQyi is the surface coverage of the moisture layer in equilibrium with the mole fraction of steam in the free stream, xS,f, at the free-stream temperature and pressure, pf. The surface coverages are evaluated with BET adsorption isotherms, according to EQ y

pf 1 C  1 pf þ ¼ Cy Cy ðp  p Þ i sat f i;mono i;mono psat

ð8Þ

where all parameters are based on the values reported by Allardice and Evans [23] with extrapolations to the hotter temperatures in our applications. The analysis is completed with definitions for the thermophysical properties as functions of the three forms of moisture. Particle densities were resolved into contributions for the organic mass assigned with the correlations of Strugala et al. [24] and the three moisture forms. Specific heats were evaluated from Merrick’s correlations [25], along with contributions for moisture weighted by the mass fractions of the three forms. For thermal conductivity, the weighting factors were based on the component volume fractions plus the correlation from Atkinson and Merrick [26] for the coal contribution. Convective heat and mass transfer coefficients were evaluated from the Ranz–Marshall correlation for Nusselt number [27], and the analog for Sherwood number. Once the profiles of temperature and the mass fractions of multi- and monolayer moisture have been evaluated, the bulk moisture loss is evaluated from the particle mass loss, as follows:

Z a dmp ¼  4pr 2 RM dr for tMAX > t > tMAX M b dt 0

  with mp t MAX b

¼ mp;0 ð1  yb;0 Þ

ð9aÞ

Z rM Z a dmp ¼ 4pr2 RM dr  4pr 2 Rm dr for t > t MAX M dt rM 0

ð9bÞ

As long as the organic coal mass does not decompose, the mass loss rate in Eq. (9b) equals the drying rate due to the joint elimination of multi- and monolayer moisture. The equations were solved in MatLab, and each simulation of a complete drying history took 5–10 min on a quad-core processor, although some cases took as long as 25 min. 2.3. Drying efficiency and the drying rate Once the bulk vaporization front disappears, the moisture concentrations disperse into radial profiles, even while some forms of moisture have not begun to be released. The expressions in this section first evaluate the drying efficiency as a function of time, and then define the drying rate in terms of the drying efficiency. The drying efficiency, gDRY, is defined from the mean mass fractions of the three forms of moisture, according to

gDRY ðtÞ ¼ 1 

hyb i þ hyM i þ hym i M0

ð10aÞ

where M0 = yb,0 + yM,0 + ym,0 and the quantities in brackets are the mean mass fractions of each form of moisture at time t. These mean values are obtained by integrating the solutions in the previous sections over the particle volume, so that

hyi i ¼ yi;0 hyb i ¼

if t < t MAX P

 3 r b ðtÞ yb;0 ; a

ð10bÞ

hyM i ¼ yM;0

and hym i ¼ ym;0

if tMAX < t < tMAX P b ð10cÞ

Ra hyb i ¼ 0; hyM i ¼

0

r 2 yM ðr;tÞ dr a3 3

and hym i ¼ ym;0 if t MAX  t  t MAX b M ð10dÞ

Ra hyb i ¼ 0; hym i ¼

r M3 3

0

hyM i ¼ ym;0 þ

r 2 yM ðr; tÞ dr a3 3

Ra rM

r 2 ym ðr; tÞ dr

a3

and if t > t MAX M

ð10eÞ

3

Once the drying efficiency has been evaluated throughout a specified drying period, the drying rate in g moisture/s can be determined by differentiation, according to

RDRY ðtÞ ¼ mp;0 M 0

dgDRY dt

ð10fÞ

2.4. Initial mass fractions for three forms of moisture According to this analysis, the distinctive drying behavior of individual brown coal samples is determined by the initial mass fractions of the three types of moisture within the coal porosity. The sum of these three forms is given by the difference between the moisture reported in a proximate analysis and bed moisture, which fills interstitial voids in any aggregate coal sample. Hence, total moisture in the analysis is proximate moisture less bed moisture. The accompanying moisture estimation scheme is based on the equilibrium moisture content (EMC) as a function of relative humidity (RH) at 30 °C, which has already been reported for numerous brown coal samples [23,28–30]. Monolayer moisture was assigned as the EMC at RH of 22%, after Allardice and Evans [28] and Hall et al. [30]. Bulk moisture was assigned from the EMC at RH of 93%. Bed moisture was based on the EMC at RH of 98%. Once the bulk and monolayer moisture fractions are calculated, the multilayer moisture is assigned by difference with the proximate moisture level minus bed moisture. The levels of the three estimated types of moisture were then correlated with variables from the proximate and ultimate analysis, which are the only sample-specific input requirement. The regression for bulk moisture is

Mb ¼ 1:334MTOT  35:080

O  24:695H þ 5:631O H

 1:667C þ 234:88

ð11aÞ

where C, H, and O are the elemental mass percentages on a daf basis. The regression has a correlation coefficient (r2) of 0.91. The regression for monolayer moisture is

 O Mm ¼ 0:5586M TOT þ 7:788  0:9727M b þ 1:554C þ 0:717O H  M TOT þ9:491H  179:36 ð11bÞ 100 for which the correlation coefficient is 0.78. Both regressions were most sensitive to the total moisture content and the O/H ratio. When applied to an assortment of Australian brown coals, this estimation procedure gave broad variations in the levels of the three types of moisture among individual samples. Bulk moisture constituted 57–80% of the total moisture; multilayer moisture constituted 19–34%; and monolayer moisture constituted 1–11%. These variations in the level of each form of moisture are definitely large enough to significantly affect the drying times under some, but not all, drying conditions. 3. Results 3.1. Validation with reported drying histories This section presents validations of the model results with data from three laboratory studies that characterized steam drying to

S. Niksa, B. Krishnakumar / Fuel 159 (2015) 345–353

1.0

Particle Mass Fraction

0.9 0.8 0.7 0.6 0.5 0.4 0.3 180

190

200

210

220

230

Temperature, C Fig. 2. Residual mass fractions for different degrees of superheat for the tests of Bongers et al. [6].

1.0

Particle Mass Fraction

moderate temperatures for a wide range of pressures. In the study of Bongers et al. [31], pressurized steam drying of low ash Loy Yang (LYLA) coal was examined in a batch autoclave from 180 to 230 °C at 1 MPa. The test coal had 62% moisture with 41% bulk moisture; 16% multilayer; and 5% monolayer moisture. At 1 MPa, the dried coals reached their EMCs after 20 h of drying at all temperatures. For LYLA, increasing the residence time to 112 h did not change the EMC. Conversely, increasing the drying temperature above 185 °C at 1 MPa to impose five degrees of superheat was sufficient to attain the EMC. Since the particle size of the coal was not provided, this value was estimated by a calibration simulation at 182 °C and 1 MPa and uniformly applied to the remainder of the test conditions. The evaluation of the residual particle mass fractions is presented in Fig. 2. After a residence time of 20 h, the measured moisture content of the coal decreased from 62% to 5% at 222 °C whereas the calculated values dropped from 62% to 1% moisture. While the general profile of the predicted residual coal mass as a function of temperature is consistent with the measured values, the relatively steep drop in the moisture level at 183 °C is depicted as a more gradual removal of moisture. The steam drying experiments of Favas et al. [32] were conducted between 130 and 350 °C for 30 min after the autoclave reached the test temperature. Cases to 250 °C are considered here, because the analysis does not account for the extensive loss of volatiles at hotter temperatures. The size of the coal particles was estimated by calibration runs at 150 °C at the corresponding steam saturation pressure for different particle sizes. For 5 mm particles, the calculated preheating time was approximately 15 min, similar to the value reported by Favas et al. This size was used in all other simulations. The coal contained 60% moisture with 37% bulk, 18% multilayer and 5% monolayer moisture. Unfortunately, the test pressure was not reported. To calibrate it, the steam pressure in simulations was adjusted continuously until the calculated moisture content of the coal after 30 min matched the measured values at 130 and 230 °C. At 130 °C, the estimated degree of superheat was 2.15 °C whereas at 230 °C, it was 17 °C. For the rest of the temperatures, a linear interpolation was used to first calculate the degrees of superheat and then to estimate the corresponding pressure. The predicted coal mass fractions are evaluated in Fig. 3. In the comparison at 250 °C, the residual coal mass was corrected to represent only moisture loss since the coal recovery was only 95.5%, with the remainder being coal decomposition products. Throughout the temperature range, the measured and predicted

0.9

0.8

0.7

0.6

0.5 125

150

175

200

225

250

Temperature, C Fig. 3. Residual mass fractions at different temperatures for the tests of Favas et al. [7].

1.0

Particle Mass Fraction

350

0.8

0.6

0.4

yb 0.2

yM

ym

0.0 0

1000

2000

3000

4000

5000

6000

Time, s Fig. 4. Predicted drying curve for the test at 140 °C from Beeby et al. [8].

values are in excellent agreement indicating an adequate representation of the degrees of superheat. For these tests, the residence time was relatively short so only bulk moisture was removed, and the rates of multi- and monolayer moisture loss did not come into play except for a marginal contribution from multilayer moisture at 250 °C. Beeby et al. [33] monitored steam drying at atmospheric pressure from 101 to 178 °C over extended holding times sufficient to achieve the EMCs. The coal properties were not reported so we used the properties of Yallourn coal (YL) in other tests reported by the authors [34]; the total moisture was 64%, and the bulk, multilayer and monolayer moisture fractions were 42%, 14%, and 8%, respectively. While particle sizes were not reported, the associated cost analysis used 1 mm, which we used in our calculations. The EMC reported at 140 °C was used to estimate the residence times in all tests. The reported residual moisture fraction was 0.44, which corresponds to the predicted moisture after 4900 s. This time was applied as the residence time for all other temperatures in this dataset. The predicted drying curve for 140 °C appears in Fig. 4, along with its resolution into the contributions for bulk moisture vaporization and the release of multi- and monolayer moisture. The preheating period is too short to see on this time scale. Bulk

351

S. Niksa, B. Krishnakumar / Fuel 159 (2015) 345–353

moisture is removed much faster than both other forms, as expected. At the end of the bulk vaporization period of 120 s, the residual moisture was 38 wt.%. Subsequent removals of multiand monolayer moisture are relatively very slow, which introduces two dramatic reductions in the slope of the drying curve. About 3000 s was required to completely remove multilayer moisture, at which point, the moisture content was about 18%. Further removal of monolayer moisture is even slower, such that an additional 1900 s only decreased the moisture to about 16. About 6800 s is required to reduce the coal moisture to 10 wt.%, and over 15,000 s are needed for complete dryness. The predictions for the other test temperatures are in excellent agreement with the measured values, as seen in Fig. 5. At a low superheat condition at 101 °C, only bulk moisture was removed, whereas multi- and monolayer moistures were removed at the higher temperatures. The most distinctive feature for this dataset is that, at 101 °C after 1600 s, there was no further advance of the bulk vaporization front toward the center, so bulk moisture vaporization ceased. Evidently, the degree of superheat and the temperature gradients within the particle were too weak to drive the front to remove additional moisture. 3.2. Sensitivities to fuel properties and operating conditions Since the analysis accurately depicts the most important aspects of drying, it can be used to identify the fuel properties and operating conditions that govern drying times and efficiencies. We first consider how variations in the three forms of moisture affect drying times. Four samples that nearly cover the domain of moisture variations for Australian brown coals were compiled from the reported properties of Yallourn (YL), low ash Loy Yang (LYLA), Morewell No. 1 (MOR1), and Bowmans (BOW) coals. As seen in Table 2, the total moisture in these coals varies between 52 and 64 wt.%, based on our correlations with coal composition. Monolayer moisture varies between 2.3 and 7.2 wt.%; multilayer moisture varies between 8.9 and 20.3 wt.%; and bulk moisture varies between 36 and 43.2 wt.%. LYLA has the most multilayer moisture and the least monolayer moisture; YL has the most bulk and monolayer moisture; MOR1 has the least bulk moisture; and BOW has the least multilayer moisture. In these simulations, 2 mm coal particles were steam dried at a slightly subatmospheric pressure and 145 °C. The feed steam contained 5% nitrogen. The simulations were then run to complete

1.0

Particle Mass Fraction

0.9 0.8 0.7 0.6

Table 2 Predicted times for preheating, and to remove bulk and multilayer moisture, and to reach 10% residual moisture, and the drying efficiency for diverse brown coals. Sample

M0/yb/yM/ym

tPMAX, s

tbMAX, s

tMMAX, s

tM10%, s

gDRY, %

YL LYLA MOR1 BOW

64/43/14/7 62/39/20/2 59/36/17/6 52/39/9/5

42 41 38 36

415 383 323 381

6200 9577 6856 5715

10,500 8147 9555 5620

95 93 94 91

dryness. The times corresponding to the different stages of drying are reported in Table 2 for the four coals. The table reports the times to finish preheating (tPMAX); to release bulk (tbMAX) and multilayer (tMMAX) moisture, and to reduce total moisture to 10 wt.% (tM10%). These times are cumulative and do not express drying times for individual stages. The corresponding drying efficiency, gDRY, appears in the final column. Drying times are not solely a function of the total moisture in the coal. Time tM10% was longest for YL and shortest for BOW, and these coals had the highest and lowest total moisture, respectively. But the times for LYLA and MOR1 did not correlate with their total moisture contents. For all coals, preheating took negligible portions of the total drying period. Similarly, the variation in bulk moisture from 36% to 43.2% changed the bulk vaporization period by less than 100 s, which is insignificant compared to tM10%, and the time to eliminate bulk moisture was never more than 7% of the drying time. Consequently, variations in the bulk moisture levels for Australian brown coals do not significantly impact drying times under typical SFBD conditions. Since bulk moisture vaporization times are similar, the time for multilayer moisture removal is determined by the level of multilayer moisture, as expected. But the drying time to 10% moisture does not correlate with the monolayer moisture levels. The comparisons between tMMAX and tM10% reveal that coals with more multilayer moisture reach a target dryness faster than those with more monolayer moisture. The 10 wt.% moisture target was reached during multilayer moisture vaporization for LYLA and BOW, so that no monolayer moisture was released during these simulations; in contrast, some monolayer moisture had to be released from YL and MOR1 to reach the target. Consequently, LYLA and BOW dried faster than YL and MOR1 simply because multilayer moisture is released faster than monolayer moisture. The times to achieve a target moisture level are proportional to the total moisture in the coal, provided that the coals are first sorted into two groups, one that meets the target without releasing any of its monolayer moisture, and the other that requires the release of appreciable monolayer moisture. Within each group, drying times are proportional to total moisture. But the correlation to total moisture breaks down when samples from both groups are taken together. Table 3 presents results on the impact of variations in the processing conditions on brown coal drying. In all cases, the coal contained 60% moisture allocated as 30% bulk, 10% multilayer, and 20% monolayer, and was injected into a slightly subatmospheric bed with 5% N2 in steam. The baseline case operated with

0.5 Table 3 Predicted times for preheating, and to remove bulk and multilayer moisture, and to reach 10% residual moisture for various sizes and temperatures.

0.4 0.3 100

120

140

160

180

Temperature, C Fig. 5. Residual mass fractions at different temperatures for the tests of Beeby et al. [8].

Case

tPMAX, s

tbMAX, s

tMMAX, s

tM10%, s

Baseline 0.5 mm 5.0 mm 170 °C 120 °C

39 2 241 26 64

294 19 1746 180 640

2289 400 3958 1878 3008

3692 2380 5566 3070 4670

352

S. Niksa, B. Krishnakumar / Fuel 159 (2015) 345–353

2 mm particles at 145 °C. Temperatures were varied from 120 to 170 °C, and sizes were varied from 0.5 to 5 mm. Simulations were run at all conditions until the total remaining moisture was 10%, and this target required release of about three-fourths of the monolayer moisture for the subject coal properties. The drying efficiency was 93% in all cases. Drying times are most sensitive to variations in the particle size. Drying times more-than-tripled as size was increased from 0.5 to 5 mm. Drying times also diminished as the temperature was reduced from 170 to 120 °C, but only by one-third. The results in Table 3 also show that the relative contributions from the stages of drying to the total drying time are strongly affected by particle size. For the smallest size, preheating and bulk moisture vaporization account for less than 1% of the drying time, whereas for the largest size, these two stages take just under one-third of the drying time. Consequently, the total drying time for the largest particle size comprises comparable contributions for the three types of moisture, but for small particles, the time to release monolayer moisture predominates.

4. Discussion and conclusions Our current validation database represents temperatures from 100 to 250 °C, pressures from 0.1 to more than 1 MPa, and drying times to several thousand seconds. Throughout this domain of drying conditions, the simulation results accurately describe both the residual moisture fraction and the moisture content as functions of the steam drying conditions. This performance is especially noteworthy because none of the drying rate parameters were adjusted to improve the agreement with data in any of the test cases. All the simulation results for all the data comparisons are based on the same drying parameters. Better quantitative performance for the analysis can only be developed with additional validation datasets, and the identification of the primary factors for drying very low rank coals in this work should yield better validation data. At a minimum, validation data must include the proximate and ultimate analyses and the mean particle size or particle size distribution of the coal; and the temperature, pressure, steam partial pressure, and residence time in every test. For drying temperatures near the normal boiling point of water, bulk moisture vaporization stops even before the bulk vaporization front reaches the particle center, so that only part of the bulk moisture is removed. The drying curve and the moisture concentration profiles remain steady because the temperature gradients within the particle are too weak to remove additional bulk moisture. But for the hotter temperatures and greater degrees of superheat in many commercial drying processes, both the preheating stage and the bulk vaporization stages make negligible contributions to the total drying times unless the particle size is larger than several millimeters. Since most of the moisture in all brown coals is bulk moisture, its rapid elimination obscures the relationship between total moisture and the time needed to achieve a target moisture level. But the drying curves of residual moisture vs. time from this analysis clearly depict the sudden deceleration in the drying rate when bulk moisture is eliminated and multilayer moisture begins to vaporize, and a second, more moderate deceleration, when multilayer moisture is eliminated and monolayer moisture begins to vaporize. Indeed, it is these two transitions, rather than the total moisture level, that determine drying times to a target moisture level. Sensitivity studies for coals with different levels of the three moisture types clearly demonstrate that the drying time is not solely a function of the total moisture content of the coal. For a selection of different coals, the times to achieve a target moisture level are proportional to the total moisture in the coal, but only

if the coals are first sorted into two groups, one that meets the target without releasing any of its monolayer moisture, and the other that requires the release of appreciable monolayer moisture. A key aspect is whether or not the target moisture level requires removal of any monolayer moisture at all. Coals with abundant bulk and multilayer moisture can achieve targets of, say, 10% residual moisture by vaporizing only a portion of the multilayer moisture. Consequently, their drying times will be relatively modest. But coals with little bulk moisture and relatively abundant monolayer moisture will require relatively long drying times simply because monolayer moisture is always released more slowly than multilayer moisture. Hence, the distribution of the three forms of moisture in any particular coal sample is a determining factor for both drying time and enthalpy requirements. These conclusions were demonstrated with simulations for diverse Australian brown coal samples, based on our correlation of EMCs for broad ranges of RH to estimate the initial levels of the three forms of moisture. Whereas the mathematical analysis can be applied to very low rank coals from other regions of the world without modification, new correlations will need to be developed to estimate the moisture levels for other regions, because numerous climatic and geographical factors acting on geological time scales determine the cation distributions which, in turn, affect the forms of moisture in coal beds.

Acknowledgement This work was sponsored by the Japan Coal Energy Center.

References [1] Rao Z, Zhao Y, Huang C. Recent developments in drying and dewatering for low rank coals. Prog Energy Combust Sci 2015;46:1–11. [2] Fushimi C, Kansha Y, Aziz M, Mochidzuki K, Kaneko S, Tsutsumi A, et al. Novel drying process based on self-heat recuperation technology. Drying Technol 2010;29:105–10. [3] Aziz M, Kansha Y, Tsutsumi A. Self-heat recuperative fluidized bed drying of brown coal’’. Chem Eng Process 2011;50:944–51. [4] Katalambula H, Gupta R. Low grade coals – a review of some prospective upgrading technologies. Energy Fuels 2009;23:3392–405. [5] Osman H, Jangam SV, Lease JD, Mujumdar AS. Drying of low rank coal – a review of recent patents and innovations. Drying Technol 2011;29(15):1763–83. [6] Stanmore BR, Boyd AR. Removal of water from brown coal by treatment with steam. Fuel Process Technol 1977–78;1:305–13. [7] Agarwal PK, Genetti WE, Lee YY, Prasad SM. Model for drying during fluidizedbed combustion of wet low-rank coals. Fuel 1984;63:1020–7. [8] Agarwal PK, Pedler I. Drying, devolatilization, and volatiles combustion for single coal particles: a pseudo steady-state approach. Fuel 1986;65:640–3. [9] Chen Z, Wu W, Agarwal PK. Steam drying of coal. Part 1. Modeling the behavior of a single particle. Fuel 2000;79:961–73. [10] Li X, Song H, Wang Q, Meesri C, Wall T, Yu J. Experimental study on drying and moisture re-adsorption kinetics of an Indonesian low rank coal. J Environ Sci Suppl 2009:S127–30. [11] Looi AY, Golonka K, Rhodes M. Drying kinetics of single porous particles in superheated steam under pressure. Chem Eng J 2002;87:329–38. [12] McIntosh MJ. Mathematical model of drying in a brown coal mill system. Part 1. Formulation of model. Fuel 1976;55:47–52. [13] McIntosh MJ. Mathematical model of drying in a brown coal mill system. Part 2. Testing of model. Fuel 1976;55:53–8. [14] Woo MW, Stokie D, Choo WL, Bhattacharya S. Master curve behavior in superheated steam drying of small porous particles. Appl Therm Eng 2013;52:460–7. [15] Potter OE, Keogh AJ. Drying high moisture coals before liquefaction or gasification. Fuel Process Technol 1981;4. 271–27. [16] Ross D, Doguparthy D, Huynh D, McIntosh M. Pressurized flash drying of Yallourn lignite. Fuel 2005;84:47–52. [17] Stakic M, Cvetinovic D, Skobalj P, Spasojevic V. An initial study on feasible treatment of Serbian lignite through utilization of low-rank coal upgrading technologies. Chem Eng Res Des 2014;92:2383–95. [18] Agarwal PK, Genetti WE, Lee YY. Coupled drying and devolatilization of wet coals in fluidized beds. Chem Eng Sci 1986;41(9):2373–83. [19] Chen Z, Agarwal PK, Agnew JB. Steam drying of coal. Part 2. Modeling the operation of a fluidized bed drying unit’’. Fuel 2001;80:209–23.

S. Niksa, B. Krishnakumar / Fuel 159 (2015) 345–353 [20] Lukowicz H, Kochaniewicz A. Analysis of waste heat obtained from coal-fired units in organic Rankine cycles and for brown coal drying. Energy 2012;45(1):203–12. [21] Allardice DJ, Chaffee AL, Jackson WR, Marshall M. Water in brown coal and its removal. In: Li C-Z, editor. Advances in the science of Victorian brown coal. Amsterdam: Elsevier; 2004. p. 85–125. [22] Yu J, Tahmasebi A, Han Y, Yin F, Li X. A review on water in low rank coal: the existence, interaction with coal structure and effects on coal utilization. Fuel Process Technol 2013;106:9–20. [23] Allardice DJ, Evans DG. The brown-coal/water system: Part 1. The effect of temperature on the evolution of water from the brown coal. Fuel 1971;50:201–10. [24] Strugala A. Empirical formulae for calculation of real density and total pore volume of hard coals. Fuel 1994;73(11):1781–5. [25] Merrick D. Mathematical models of the thermal decomposition of coal. 2. Specific heats and heats of reaction. Fuel 1983;62:540–70. [26] Atkinson B, Merrick D. Mathematical model of the chemical decomposition of coal. Fuel 1983;62:540–6. [27] Ranz WE, Marshall WR. Evaporation from drops. Chem Eng Prog 1952; 48:141–6.

353

[28] Allardice DJ, Evans DG. The brown-coal/water system: Part 2. Water sorption isotherms on bed-moist Yallourn brown coal. Fuel 1971;50:236–53. [29] Allardice DJ, Clemow LM, Favas G, Jackson WR, Marshall M, Sakurovs R. The characterization of different forms of water in low rank coals and some hydrothermally dried products. Fuel 2003;82. 661–07. [30] Hall SF, Readett DJ, Quast KB, Ketteridge IB. CHEMECA 1987, Melbourne. [31] Bongers GD, Woskoboenko F, Jackson WR, Patti AF, Redlich PJ. The properties of the coal and steam condensate from pressurized steam drying. In: Pajares JA, Tascon JMD, editors. Coal science, vol. 24. Amsterdam: Elsevier Science; 1995. p. 901. [32] Favas G, Chaffee AL, Jackson WR. Exploring the horizons of brown coal drying technologies for power generation. Evaluation of products from various processes. In: Proc 11th int conf on coal sci, San Francisco(CA): NETL, US DoE; 2001. [33] Beeby C, Dodds C-R, Ho H, Potter OE, YamaokaY. Steam drying of coal. In: Proc 1985 Int conf on coal sci, Sydney(Australia): Pergamon Press; 1985. p. 513. [34] Potter OE. http://www.ret.gov.au/energy/Documentsw/ewp/draft-ewp 2011/ submissions/75.OwenPotterSubmissions2.pdf.