Contents lists available atScienceDirect
Thermochimica Acta
journal homepage:www.elsevier.com/locate/tca
Thermogravimetric analysis of (co-)combustion of oily sludge and litchi
peels: combustion characterization, interactions and kinetics
Chao Liu
a, Jingyong Liu
a,⁎, Guang Sun
a, Wuming Xie
a, Jiahong Kuo
a, Shoupeng Li
a, Jialin Liang
a,
Kenlin Chang
b, Shuiyu Sun
a, Musa Buyukada
c, Fatih Evrendilek
c,daSchool of Environmental Science and Engineering, Institute of Environmental Health and Pollution Control, Guangdong University of Technology, Guangzhou, 510006, China
bInstitute of Environmental Engineering, National Sun Yat-Sen University, Kaohsiung, 80424, Taiwan cDepartment of Environmental Engineering, Abant Izzet Baysal University, Bolu, 14052, Turkey dDepartment of Environmental Engineering, Ardahan University, Ardahan, 75002, Turkey
A R T I C L E I N F O Keywords: Oily sludge Litchi peels Co-combustion Thermogravimetric analysis Master-plots method A B S T R A C T
The thermal characteristics and kinetics of oily sludge, litchi peels and their blends were comprehensively evaluated using thermogravimetric experiments under air atmosphere. Results showed that devolatilization, ignition and burnout indices of litchi peels were higher than those of oily sludge, with better combustion characteristics. However, co-combustion performance decreased with increased litchi peels fraction. Such in-teractions as inhibition effect on devolatilization and promotion effect on char oxidation of oily sludge occurred. Average activation energy was estimated at 118.1 kJ/mol for oily sludge and 215.2 kJ/mol for litchi peels using the Kissinger-Akahira-Sunose integral method and reached its minimum value with the addition of 15%–20% litchi peels. Trend of activation energy was consistent with the interaction analysis during co-combustion pro-cess. Kinetic triplets were also estimated using the integral master-plots method, while the most suitable reaction model for three stages of oily sludge and their blends was proposed.
1. Introduction
Petroleum industries generate large quantities of oily sludge (OS) that accumulates in crude oil and refinery product tanks, and desalters during oil production and processing [1]. Rapid development of the petroleum industry has increased the production of OS, a kind of black brown sticky semisolid material mainly composed of water, oil emul-sion, and suspended solid particles [2]. More importantly, OS contains a large quantity of toxic hazardous substances (e.g., benzene, anthracene, pyrene, phenols, and other toxic smelly organics), pathogens, heavy metals (e.g., Cu, Pb, Cr, Zn, and Hg),flocculants, detergents, pesticides, and other water treatment agents [3,4]. Due to its adverse effects on human and environmental health, it has been classified as a priority environmental pollutant by the US Environmental Protection Agency and regarded as dangerous solid waste (HW08) in China [5]. Complex OS ingredients and issues with its dehydration result in a high disposal and treatment cost, which significantly restricts the sustainable devel-opment of petrochemical industries [6].
The commonly used technologies for recycling OS include extrac-tion [7], centrifugal [8], pyrolysis [9], gasification [4], incineration
[10,11], co-combustion technology [12], thermochemical cleaning [13], modulation mechanical separation [14], and ultrasonic treatment [15]. However, each has its own limitations due to differences in sampling location, composition and performance of OS [1]. Co-com-bustion with biomass has becoming a major treatment approach owing to its volume reduction, and energy conservation [11,16]. The complex compositions of fuels such as OS, biomass, coals, and other solid wastes cause the co-firing characteristics of their blended fuels to fluctuate significantly, thus rendering it difficult to quantify their co-firing properties [17].
For co-combustion characteristics of OS, sewage sludge, biomass, coal, and their blends, thermogravimetric analysis (TGA) was widely adopted to gain a better understanding of reaction kinetics and me-chanisms involved in co-firing process [18,19]. Font et al. [20] pointed out significant differences in co-combustion characteristic of sludge depending on its different physical and chemical properties. Chen et al. [21] obtained the pyrolysis characteristics, byproducts and reaction kinetics of OS by TGA. High moisture and ash contents, low calorific value and high viscosity of OS were reported to lead to unsteady combustion and incomplete burnout when burnt alone, thus pointing to
https://doi.org/10.1016/j.tca.2018.06.009
Received 17 March 2018; Received in revised form 28 May 2018; Accepted 11 June 2018
⁎Corresponding author.
E-mail addresses:www053991@126.com,Liujy@gdut.edu.cn(J. Liu).
Available online 15 June 2018
0040-6031/ © 2018 Elsevier B.V. All rights reserved.
the need for auxiliary biofuels [11,22]. For example, Liao et al. [23] found that herbaceous biomass (straw, and sugarcane) had a faster devolatilization rate and a lower ignition temperature with better ig-nition characteristics than did woody biomass (eucalyptus bark, leaves and branches). Huang et al. [24] and Chen et al. [25] observed that co-firing with water hyacinth and coffee residue enhanced the compre-hensive combustion characteristics of sludge. Mu et al. [26] concluded that strong interactions during co-pyrolysis of OS and lignite increased the yield of pyrolysis gas. Huang et al. [27] found that co-firing with water hyacinth reduced emissions of air pollutants of sewage sludge. Although there exist studies about co-pyrolysis/co-combustion of OS and coal [10,11,21], a few studies have been conducted about the co-pyrolysis/co-combustion of OS and biomass thus far.
Recently, co-combustion with biomass has been started to be uti-lized for large-scale power generation [28,29]. Litchi (Litchi chinensis Sonn) is the main fruit tree planted in (sub)tropical nations including China, India, and Thailand among which China ranksfirst globally in terms of its production and planted area [30]. Currently, the annual production of litchi in China was estimated at about 150 × 104tons
[31], more than 70% of its total global yield [32]. Naturally, litchi peels (LP) is regarded as agricultural waste produced in large volumes and accounts for more than 15% of the quantity of fresh litchi [33]. Though studied mainly focused on extraction of chemical raw materials [34], development of functional medicine [35], utilization of carbon mate-rials [36], preparation of adsorbents [37] and dry preparation of fuel [38], LP has not been explored in terms of its potential as a biofuel. In particular, there still remains a big gap in the exploration and quanti-fication of (co-)combustion behaviors and kinetics of OS. Therefore, studies are needed to provide guidance and baseline for optimally de-signing and operating the co-combustion process of OS and biofuels such as LP.
Therefore, the present study aims to (1) quantify (co-)combustion characteristics of OS and LP using thermogravimetric experiments
underfive blend ratios and four heating rates; (2) evaluate ignition, burnout and comprehensive combustion characteristics, and their in-teractions; and (3) estimate (co-)combustion kinetic triplets of OS and its blends by using the iso-conversional and integral master-plots methods.
2. Materials and methods
2.1. Experimental materials
Oily sludge used in the experiments was collected from a waste-water treatment plant of Liaohe Oil Field Company, Panjin city, Liaoning Province, China where it was pretreated with two stages of biological treatment. The raw OS material is black, lumpy and sticky. Litchi peels were obtained from a vegetable market of Guangzhou Higher Education Mega Center, Guangzhou, Guangdong Province, China. Pure LP and OS samples were all passed through a 200-mesh (0.075 mm) screen before being fully blended in agate mortar and stored in a desiccator for subsequent experiments.
The blending of OS and LP samples was carried out using 5, 15, 25, 35 and 45% (by weight) of LP and was coded as follows: 95OS5LP, 85OS15LP, 75OS25LP, 65OS35LP and 55OS45LP, respectively. The ultimate, proximate, calorific value and ash composition analyses of OS and LP are presented inTable 1. Ash composition was analyzed using a ZSX Primus II sequential XRF spectrometer (Rigaku, Japan). As can be seen fromTable 1, ash content and calorific values of OS were higher than those of LP, whereasfixed carbon and volatile contents of LP were higher than those of OS. Ash composition also differed significantly in that OS contained high Fe2O3, Al2O3 and SiO2 concentrations, there
might exist some minerals such as Cu, Zn, V, and Cr in oily sludge. However, LP contained more alkali and alkaline earth metals (K2O,
CaO, and MgO). Thus, tendencies of shagging, fouling and corrosion also need to be considered during co-combustion.
Nomenclature
Symbols/abbreviations
Ti ignition temperature (°C)
Tp ignition temperature (°C)
Tb burnout temperature (°C)
Tv initial devolatilization temperature (°C)
-Rp maximum weight loss rate (%∙min−1)
-Rv average weight loss rate (%∙min−1)
ΔT1/2 temperature interval at the half value of -Rp(°C)
Ci ignition index (%∙min∙°C3)
Cb burnout index (%∙min∙°C4)
Dv volatile matter release index (%∙min∙°C3)
S comprehensive combustibility index (%2∙min2∙°C3)
T absolute temperature (K) β heating rate (°C∙min) k(T) reaction rate constant α conversion rate (%) f(α) differential reaction function G(α) integral reaction function P(u) temperature integral
A apparent pre-exponential factor (s−1) R universal gas constant (J∙mol−1∙K−1)
E apparent activation energy (kJ∙mol) Ea average apparent activation energy (kJ∙mol)
Table 1
The ultimate, proximate, calorific value and ash composition analyses of OS and LP on an air-dried basis.
Samples Ultimate analysis (%) Proximate analysis (%) Qnete
(MJ/kg)
C H O N S Ma Vb Ac F
Cd
OS 63.49 7.91 4.99 0.86 0.665 1.37 68.88 20.72 9.03 28.81
LP 47.39 6.32 36.81 0.81 0.019 5.75 75.53 3.30 15.42 17.49
Ash composition analysis (%)
SiO2 Al2O3 Na2O MgO K2O CaO MnO Fe2O3 TiO2 P2O5
OS 32.404 6.499 0.658 0.209 1.124 2.638 0.093 51.442 1.671 0.480 LP 1.683 0.298 1.081 3.219 43.421 37.591 1.027 1.114 0.022 6.004 a M, moisture. b V, volatile matters. c A, ash. dF C,fixed carbon. e Q
2.2. Experimental design and procedure
Thermogravimetric analysis was carried out using a NETZSCH STA 409 PC simultaneous analyzer in the air atmosphere (21% O2/79% N2).
To eliminate the effects caused by the mass and heat transfer limita-tions, a small sample (9± 0.5 mg) was loaded into an Al2O3ceramic
crucible for each run under the non-isothermal conditions. The sample was heated up from an ambient temperature to 1000℃ at the four heating rates of 10, 20, 30 and 40℃/min with a gas inflow of 50 mL/ min. Before the start of the experiment, blank tests were performed to provide the baseline to eliminate instrumental systematic errors, and repetitions were also performed to ensure an experimental error below ± 3%, showing good stability of the TGA instrument as well as reliable and reproducible results. The NETZSCH–T4–Kinetic 2 software was utilized to derive thermogravimetry (TG) and derivative thermo-gravimetry (DTG) curves.
2.3. Characterization of (co-)combustion performances
To quantify effects of the heating rates and the blend ratios on (co-) combustion performance, the combustion characteristic parameters of ignition temperature (Ti), peak temperature (Tp), burnout temperature
(Tb), maximum weight loss rate (-Rp), average weight loss rate (-Rv),
and temperature interval at the half value of -Rp(ΔT1/2) were used each
of which was detailed in related literature [39–41].
The additional combustion indices of ignition (Ci), burnout (Cb),
volatile matter release (Dv), and comprehensive combustibility (S) were
adopted to capture combustion performance under different conditions [42,43]. These indices can be described as a function of characteristic temperatures and weight loss rates. Dv represents the release
perfor-mance of volatile matters in fuel, while S represents the comprehensive characteristics including ignition and burnout [43].
= − × × D R T T ΔT V P P V 1/2 (1) = − × C R t t i P i p (2) = − × × C R Δt t t b P P b 1/2 (3) = − × − × S R R T T ( P) ( V) i2 b (4)
where ti, tp, tb andΔt1/2 represent ignition time, peak time, burnout
time, and time interval at the half value of -Rp, respectively. Tv
re-presents the initial devolatilization temperature.
The higher an S value is, the better the ignition and burnout per-formances are. The higher a Dv value is, the more centralized the
combustion region of char residues is and the better the burnout per-formance is [43].
2.4. Kinetic analysis theory
In the non-isothermal experiments carried out under a thermo bal-ance, the sample mass was measured as a function of temperature. The rate of degradation or conversion (dα/dt) is a linear product of a tem-perature-dependent rate constant [k(T)] and a temperature-in-dependent conversion [f(α)] thus:
= = ⎛ ⎝− ⎞ ⎠ dα dt f α k T A E RT f α ( ) ( ) exp ( ) (5) whereα is conversion rate of the solid reactant; k(T) is a reaction rate constant; A is apparent pre-exponential factor (min−1); E is apparent activation energy (J/mol); R is the universal gas constant (8.314 J/mol/ K); T is absolute temperature (K); t is reaction time (min); and β is heating rate (℃/min).
The conversion rate (α) is defined as follows:
= − − α m m m m t f 0 0 (6)
where m0, mtand mfrepresent sample weights at the initial moment,
moment t and the end of the reaction.
The integration form of Eq. (5) can be expressed as follows:
∫
∫
∫
= = ⎛ ⎝− ⎞⎠ = − = ∞ G α d α f α A β E RT d T AE βR x x dx AE βRp u ( ) ( ) ( ) exp ( ) exp( ) ( ) α T x 0 0 2 (7) The Kissinger–Akahira–Sunose (KAS) method [44] can be expressed as follows: ⎟ ⎜ ⎟ ⎛ ⎝ ⎞ ⎠ = ⎛ ⎝ ⎞ ⎠ − β T AR EG α E RT ln ln ( ) 2 (8) For a givenα, the E value can be calculated from the slope of the ln [β/T2] versus 1/T plot, where u = E/RT and p(u) is the temperature integral without analytical solution which can be approximately solved by using empirical equations. When Tang-Liu-Zhang-Wang-Wang ap-proximation was selected for the master-plots method, the deviation of the numerical solution for P(u) at u﹥14 was below 0.1% [45].
= − × + P u u u u ( ) exp( ) (1.00198882 1.87391198) (9)
For a single-step reaction, the kinetic triplets are invariable. Through estimating E value by iso-conversional methods in advance, a proper reaction model can be found to simulate TG data. Adopting a reference atα = 0.5, Eq. (7) becomes as follows:
=
G AE
βR P u (0.5) ( ) ( 0.5)
(10) where u0.5= E/RT0.5. G(0.5) is the integral reaction model atα = 0.5;
T0.5is the temperature atα = 0.5.
The integral master-plots equation can be obtained dividing Eq. (7) by Eq. (10). = G α G P u P ( ) (0.5) ( ) (0.5) (11)
Employing various common G(α) functions (Table S-1), the plots of G(α)/G(0.5) versus α on the left hand of Eq. (11) represent the theo-retical master plots. Moreover, the experimental master plots are gen-erated plotting P(u)/P(u0.5) versusα on the right hand of Eq. (11) from
TG data obtained at distinct heating rates. For a givenα, the equality would be achieved using a proper kinetic model. Otherwise, an in-appropriate kinetic model of G(α) leads to a significant difference be-tween theoretical and experimental master plots. Thus, the reaction mechanism of each phase of the co-combustion process can be de-termined using the method of integral master-plots.
3. Results and discussion
3.1. Combustion characteristics of pure OS and LP
Fig. 1shows the TG-DTG curves of pure OS and LP combustion at a heating rate of 20℃/min. There were four main groups of weight loss during the OS combustion (Fig. 1a). Stage 1 (< 350℃) included a very small amount of water evaporation and the release of light volatiles, with 21.6% total weight loss. Stage 2 (350 to 485℃) corresponded mainly to devolatilization and decomposition of heavy volatiles during which the weight loss rate reached its maximum (12.51%/min) at 445.3℃, with 37.3% total weight loss. OS was reported to contain various of light and heavy oil components and other organic ingredients which can be divided into light and heavy volatiles according to their
boiling points [10,46,47]. In the present study, the range of 150 to 350℃ was attributed to the release of light oil components with rela-tively low boiling points, while the range of 350 to 485℃ corresponded to the decomposition and combustion of heavy volatiles with a higher boiling point. In these processes, some light components may undergo an incomplete release and combustion, while heavy components and bio-refractory macromolecule organic compounds began to undergo complex reactions such as cracking and condensation at high tem-peratures, thus resulting in a charring process. Stage 3 (485 to 550℃) corresponded to the combustion of refractory organic residues andfixed carbon, with 21.0% weight loss, much more than content of fixed carbon according to the proximate analysis fromTable 1. The reason for this may be that some heavy oil components did not decompose com-pletely or were transformed into compounds with a more stable struc-ture through the secondary cracking and condensation reactions at higher temperatures. When the combustion of volatiles ended, oxygen
was diffused to the exposed fixed carbon surface. At the same time, the fixed carbon was rapidly burned out. Stage 4 (> 550 ℃) was the sta-bilization stage of residues, with only 1.6% weight loss due to the de-composition of carbonates and other inorganic minerals [10]. The de-volatilization and combustion process of OS that occurred in the range of 150 to 550℃was consistent with the previous findings (200 to 650℃) about the separation of different oil components of OS [46,47].
Similarly, the weight loss of LP included four stages (Fig. 1b). Stage 1 (< 157℃) involved a small amount of water evaporation. Stage 2 (157 to 438℃) was due to devolatilization of volatiles with 67.5% weight loss. Although the combustion curves of biomass vary with biomass species and relative contents of hemicellulose, cellulose, and lignin, the devolatilization stages of hemicellulose, cellulose, and lignin usually occur in the range of 160 to 240℃, 240 to 360℃ and 160 to 627℃, respectively [23]. Moreover, lignin has the widest temperature range for decomposition during which lignin decomposes slowly after (hemi)cellulose due to its high thermal stability. Thus, the range of 157 to 375℃ may have overlapping decomposing stages of (hemi)cellulose and lignin and had the maximum weight loss rate (10.3%/min) at 310.1℃ which can be attributed mainly to cellulose decomposition. An obvious shoulder at about 240℃ corresponded to the decomposition of hemicellulose, while a considerable amount of lignin was decomposed in the range of 375 to 438℃ with 14.4% weight loss. Stage 3 (438 to 511℃) corresponded to the burnout of residues and combustion of fixed carbon. As the release of volatiles was completed, oxygen could diffuse to the highly reactive surface of biomass char, thus rapidly increasing the oxidation rate. The weight loss was also higher for thefixed carbon content in stage 3, which may be caused by the coking process of cel-lulose and lignin. Stage 4 (> 511℃) involved the burnout of the re-sidues with only 1.9% weight loss as well as the decomposition of a small amount of minerals and carbonate. Consistent with 17.7% of the OS residues found in the present study, OS was reported to contain more non-combustible materials such as kaolin, quartz, mica and other inorganic minerals [48]. Therefore, the addition of LP appeared to ameliorate the combustion characteristics of OS.
Given the TG/DTG curves of OS and LP, their weight loss perfor-mances differed dramatically although their combustions both occurred in the range of 160 to 580℃. The weight loss of LP occurred intensively during devolatilization of volatiles in the earlier stage at a relatively low temperature zone, whereas that of OS mainly occurred during combustion of heavy volatiles and char oxidation in the later stage. The combustion characteristic parameters of OS and LP at the four heating rates are summarized in Table 2. Mass residue, burnout time, and temperature for maximum weight loss rate and char oxidation were lower for LP than OS. This indicated that LP performed better com-bustion characteristics than OS which related to different structure and composition of the two materials. OS contained a large amount of ash which weakened heat transfer efficiency in the later stage of combus-tion and hindered oxygen diffusion to the char surface, thus causing the delay combustion of thefixed carbon at a higher temperatures [49]. While the combustible components of LP burnt more concentrated and Fig. 1. TG-DTG curves of OS (a) and LP (b) in air atmosphere at 20℃/min.
Table 2
Combustion characteristic parameters of OS and LP at four heating rates.
Sample β Ti Tp Tb ti tp tb -Rp -Rv ΔT1/2 Mf OS 10 212.5 431.8 519.5 18.6 40.1 49.1 6.3 1.6 94 18.4 20 219.6 445.3 550.3 10.1 20.7 26.1 12.5 3.0 114 17.7 30 227.3 459.9 580.4 7.4 14.3 18.4 16.9 4.2 139 18.8 40 231.4 467.4 615.3 5.9 11.0 14.7 20.8 5.3 156 18.3 LP 10 237.5 307.4 482.1 21.0 27.7 45.3 5.0 2.0 106 3.2 20 244.3 310.1 511.0 11.2 14.0 24.1 10.3 3.9 100 2.9 30 247.9 313.4 521.8 7.9 9.8 16.5 16.1 5.5 104 3.7 40 252.7 319.1 539.0 6.3 7.7 12.8 23.8 7.1 93 3.1
β: heating rate,℃/min; Ti: ignition temperature,℃; Tp: temperature corresponding to the maximum weight loss rate; Tb: burnout temperature,℃; ti: ignition time,
violently, which can provide sufficient heat to promote the fixed carbon combustion [50].
Based on the tangent method of the TG-DTG curve [39], the ignition temperatures (Ti) of OS and LP were estimated at 219.6℃and 244.3℃,
respectively, while the burnout temperatures (Tb) of OS and LP were
550.4℃ and 511.1℃ with 98% total mass loss. The volatile content of LP was higher and released more violently, whereas the light volatiles of OS were precipitated earlier. This can be attributed to the differences between OS and LP in composition, structure, properties and reactions. As LP consisted mainly of hemicellulose, cellulose, lignin, and a very small amount of extracted organics, the amorphous structure of hemi-cellulose resulted in its relatively poor thermal stability. However, cellulose was difficult to decompose as a macromolecule compound consisting of D-glucose bonded by glycoside linkage of crystal structure with high degree of polymerization. Due to polysaccharides as the main ingredient of lignin consisting of phenyl propane monomers linked by carbon–carbon and ether bonds to form high molecules, lignin had higher thermal stability and was more difficult to break down. How-ever, the petroleum hydrocarbons with low boiling point and high volatility of OS were easier to decompose with heating [51]. Mean-while, the organic matters such as protein, fat and oil of OS are easy to decompose through the heating during the wastewater treatment pro-cess [52]. Given the higher burnout temperature of OS, it was more difficult to burnout because it contained more non-flammable mate-rials. Large amount of ash may cover the unburn particle surface, thus weakening the efficiencies of heat transfer and oxygen diffusion, re-sulting in a delayed combustion offixed carbon [53]. The ash content of OS is rich in heavy metals and metallic salts which can participate in complex reactions at high temperatures [54], while the higher alkali metal Na and K contents of biomass ash volatilize easily at high
temperatures [55].
Overall, the differences in the volatile content, component, struc-ture and reactivity of OS and LP led to different combustion perfor-mances. Combustion of LP as lignocelluloses biomass occurred mainly in the early stage of devolatilization and burning (low temperature zone), while OS was devolatilized easily and combusted mostly in the late stage of burning of heavy volatiles andfixed carbon (high tem-perature zone).
3.2. Effects of heating rate on combustion
Fig. 2shows the TG/DTG curves of OS and LP at the four heating rates (10, 20, 30 and 40℃/min) in the air atmosphere, with their characteristic parameters summarized in Table 2. With the raised heating rate, the TG/DTG curves of OS did not change and were dis-tinguished clearly, but the temperature range became wider. However, the DTG curve of LP changed significantly where the peak shapes were overlapped and only two large peaks appeared under the higher heating rate. With the increased heating rate, all the DTG curves shifted toward a higher temperature zone, the ignition and burnout temperatures in-creased by 18.9℃ and 15.2℃, and 19.5℃ and 95.8℃ for OS and LP, respectively. This indicated that the entire decomposition process was delayed with the increased heating rate [56]. This can be because the heat transfer of the sample particles was less effective at the high than low heating rate. Also, the reaction temperature, the release of vola-tiles, and the diffusion concentration of oxygen appeared to play a role in the combustion process.
The weight loss peaked in the volatilization and combustion stages at above 10℃/min and in the oxidation of char at 10℃/min. This may be because the slow decomposition of cellulose and lignin at the low
heating rate facilitated the generation of more highly reactive char by the charring process, which increased the weight loss rate [57].
The amount of the combustion residuals of OS and LP did not change and was lowest at 20℃/min, which indicated its insignificant effect on the residual mass [58]. With the increased heating rate, the ignition (ti) and burnout (tp) times of OS and LP were shortened,
burning significantly ahead of their schedules. However, the maximum (-Rp) and average (-Rv) weight loss rates increased dramatically with the
increased heating rate which indicated more violent combustion. With the increased heating rate, -Rpand -Rvincreased by 3.2 times for both
OS and LP and 4.73 and 3.45 times for OS and LP, respectively, which indicated their enhanced combustion intensity. This could be due to the limitations on the heat and mass transfers. With the increased heating rate, it would take a shorter time to reach a certain ambient tempera-ture. Temperature of the particle surface was slightly higher than that of the interior which indicated their increased temperature difference, thus enhancing the heat transfer from the surface to the core [59,60].
3.3. Co-combustion characteristics of blends
Fig. 3shows the co-combustion TG/DTG curves of OS and LP under thefive LP blend ratios plus the pure samples at 20℃/min in the air atmosphere. The TG curves of the blends located in between those of the pure OS and LP. With the increased blend ratio, the dominance of the TG curve shifted from OS to LP which implied that the addition of LP affected the OS combustion characteristics. The DTG curves showed three main peaks of weight loss for the co-combustion when the water evaporation process was excluded. However, the shapes of the DTG curves were almost the same as those of OS, and the characteristic temperature of the blends in each stage located in between the pure OS and LP. With the increased blend ratio, the devolatilization rate of light volatiles increased, whereas the combustion rate of heavy volatiles and fixed carbon decreased which was consistent with the findings in re-lated literature [24,25].
The (co-)combustion characteristic parameters of OS and LP were summarized inTable 3. The ignition temperatures (Ti) of all the blends
fluctuated around that of OS (219.6℃), and the maximum change was only 6.7℃which indicated a slight effect of the added LP on the ignition performance of OS. This may be related to the fact that both OS and LP had a high volatile content and a similar devolatilization temperature. The increased blend ratio did not change the burnout time significantly, diminished the combustion residue mass (Mf) from 17.7% to 12.4% and
grew the burnout temperature (Tb) from 550.4 to 560.4℃. Though the
co-combustion residues decreased, the increased burnout temperature of OS was not propitious for improving the OS burnout performance.
With the increased blend ratio, the maximum weight loss of light volatiles occurred for OS during the devolatilization processes (stage 1). -Rp1increased from 4.0 to 5.9%/min, and the characteristic
tempera-ture (T1) reduced slightly, the weight loss of Mf1increased from 21.6 to
37.1% due to the high volatile content of LP. Thus, the co-combustion during stage 1 appeared to be promoted.
For the decomposition and combustion of the heavy volatiles of OS (stage 2), -Rp2increased slightly with the LP blend ratio of 5% but
re-duced sharply from 12.9 to 9.0%/min when the ratio was between 15% and 45%. The weight loss of Mf2 decreased and the characteristic
temperature (T2) did not change significantly. For the combustion of
organic residues andfixed carbon (stage 3) of OS, the maximum weight loss rate (-Rp3) decreased, while the weight loss of Mf3increased, and
the characteristic temperature (T2) reduced from 521.8 to 478.9℃. The
above co-combustion characteristic parameters indicated that the ad-dition of LP had differently affected OS combustion at the different stages which may be related to their complex components and inter-actions during co-combustion as discussed in Section3.4.
3.4. Interactions between OS and LP during co-combustion
To further explore the interactions between OS and LP, the theo-retical TG/DTG curves were obtained by calculating the average weights of the individual fuels as can be seen in Eq. (12) [51]:
= +
Wol η Wos os η Wlp lp (12)
= −
ΔW TGexp erimental TGcalculated (13)
where Wolis the calculated value (%) of TG-DTG curve at the
corre-sponding temperature;ηosandηlprepresent the weight percentages of
OS and LP in the blend (%), respectively; and Wosand Wlpare the
ex-perimental mass losses or mass lose rates for each sample (%), respec-tively. The theoretical TG curves obtained were compared with the experimental curves in Fig. 4a. ΔW was introduced in Eq. (13) to characterize the strength of the interactions [61] and the results are given inFig. 4b.
The theoretical and experimental TG curves had the same trend with the different temperatures and blend ratios. At below 270℃, the theoretical and experimental TG curves overlapped, with ΔW being close to zero which indicated occurrence of no interaction. At between 270 and 500℃, the experimental curves lagged behind the theoretical curves regularly andΔW had the same trend with the blend ratios. At the same temperature, ΔW increased with the increased blend ratio. With the same blend ratio,ΔW increased with the temperature. The three peaks were observed on theΔW curves at 355℃, 430℃ and 500℃ which indicated strong interactions between LP and OS during
co-Fig. 3. TG (a) and DTG (b) curves for co-combustion of OS and LP withfive blend ratios in air atmosphere.
combustion. TheΔW curves shifted towards the high temperature with the increased blend ratio. With the 5% LP blend ratio, ΔW was ex-tremely small (< 0.6%/min) which suggested that the LP addition had no significant effect on OS combustion. With the 45% LP blend ratio, ΔW reached 6.8% which indicated a significant inhibition effect on OS combustion from 270 to 500℃. At above 500℃ until the complete combustion, theΔW values declined sharply even to below zero which pointed to the gradually weakened effect of inhibition and even to the facilitation of OS combustion.
The volatiles of OS mainly include light paraffin with a low boiling point, colloid substance with a high boiling point, and heavy compo-nents such as asphalting. During the volatilization process, the heavy components may produce a large amount of sticky oils or tars with the increased temperature. Previous results showed that bio-oils derived from pyrolysis of microalgae coated textile dyeing sludge powder so as to render it difficult to release its volatiles [62]. Therefore, the possible reason for the inhibition is that the OS powder could be coated by these sticky liquid products, thus rendering the volatiles difficult to release. Similar results were found in related literature [63]. During OS com-bustion, the particles were condensed gradually, and a large amount ash was generated at the same time which in turn influenced the oxygen diffusion and heat transfer. After the addition of LP, the calorific value declined significantly, and the heat required for co-combustion became insufficient which inhibited or weakened the decomposition rate of OS in this temperature range [64].
Fig. 4c shows the comparison of experimental versus calculated TG/ DTG curves with the 45% LP blend ratio. At below 270℃, the experi-mental TG curve coincided with the calculated one. However, at be-tween 270 and 500℃, the experimental TG curve lagged behind the calculated one which indicated that the devolatilization and combus-tion processes needed a higher temperature. In the range from 270 to 500℃, two peaks at 310℃ and 452℃ were detected in the DTG curves which corresponded to stages 1 and 2. Consistent with the above in-teraction conclusions, the experimental loss rate was smaller than the calculated one, and the characteristic temperature remained basically unchanged. At above 500℃, the experimental mass loss occurred more rapidly than did the calculated one, and the experimental mass loss rate was much higher than the calculated one at 564℃, which enhanced the combustion of organic residues andfixed carbon of OS (stage 3).
However, the mechanism of the synergistic effects between OS and LP during co-combustion was not very clear. For example, after the removal of the mineral matter of paper mill sludge, its burning rate and the peak temperature were reported to significantly increase, and the presence of the mineral matter inhibited its combustion rate which acted as an inert material [65]. Therefore, the addition of LP with a lower ash content to OS reduced their mineral matter. The char formed during the LP decomposition appeared to play a catalytic role in the complete degradation of OS organic residues and fixed carbon. This might explain that the interaction facilitated the combustion of OS above 500℃ in stage 3, and the experimental DTG curve shifted to low
temperatures compared to the calculated one.
3.5. (Co-)combustion parameters of OS and LP at four heating rates
(Co-)combustion parameters of OS and LP at four heating rates are shown inTable 4. The Dvvalue of LP was higher than that of OS at the
heating rates, thus indicating better volatile release characteristics for LP than OS. This was mainly because LP had a higher volatile content, and its volatilization occurred in a lower temperature zone. The change in Dvof the blends was related to both heating rate and blend ratio.
With the increased heating rate, Dvof both OS and LP (especially LP)
increased. For example, Dvvalues of OS and LP were 7.02 × 10−7and
7.58 × 10−7%/min/℃3whenβ = 10℃/min and increased by 2.3 and 5.8 times whenβ = 40℃/min, respectively. This suggested that higher heating rate was helpful for the release of LP volatiles. At 20℃/min, Dv
with the 5% LP blend ratio increased from 1.2485 × 10−7 to 1.3624 × 10−7%/min/℃3and decreased gradually to 0.9577 × 10−7
%/min/℃3with the increased blend ratio. This indicated that co-firing
of OS with LP was not conducive to the release and combustion of volatiles. The results were consistent with the above interaction ana-lysis.
At the same heating rate, the maximum volatile release time of LP was obviously shorter than that of OS, and the ignition index (Ci) of LP
was higher than that of OS. With the addition of LP, the tiand tpvalues
did not change significantly, whereas the maximum volatile release rate (-Rp) of OS reduced. The ignition index of the blends was lower than
that of the individual samples. Similarly, the burnout index (Cb) of LP
was 1.1, 1.9, 2.8 and 4.4 times higher than that of OS when the heating rate increased from 10 to 40℃/min, respectively. For example, Cbof the
blend increased from 11.5 to 12.9%/min/℃4at 30℃/min with below
the 15% LP blend ratio and decreased gradually to 9.0%/min/℃4with
above the 15% LP blend ratio. This indicated that even a small pro-portion of LP was helpful to improve the burnout performance of OS.
The S value of OS was higher than that of LP at below 20℃/min but became the opposite at the higher heating rate. The addition of 5% LP increased the S value slightly which decreased sharply with increased LP fraction. The S values of OS and LP increased by 7.6 and 12.8 times with the increased heating rate, respectively. Overall, the comprehen-sive combustion characteristics of the fuels became better with the creased heating rate. The addition of the smallest amount of LP in-creased the comprehensive combustion performance of OS. However, the overall combustion characteristics of the blend became worse, suggesting that the co-firing of LP was not beneficial to enhance the comprehensive combustion performance of OS.
3.6. Kinetic analysis
3.6.1. Activation energy estimated by KAS method
Fig. 5shows changes in E withα for OS, LP and their blends during (co-)combustion. The change in E indicated that the (co-)combustion Table 3
(Co-)combustion characteristic parameters at a heating rate of 20℃/min.
Samples Stage1 Stage2 Stage 3
Ti Tb T1 -Rp1 Mf1 T2 -Rp2 Mf2 T3 -Rp3 Mf3 -Rp -Rv Mf 100OS 219.6 550.4 299.0 4.0 21.6 445.3 12.5 37.7 522.8 8.1 21.2 12.51 3.03 17.75 95OS5LP 220.0 550.5 298.2 4.2 23.1 446.7 12.9 35.4 518.4 8.0 23.1 12.97 3.11 16.72 85OS15LP 219.7 557.6 297.7 4.6 27.5 442.5 11.0 32.0 513.2 7.7 23.1 11.08 3.13 15.88 75OS25LP 224.5 560.4 289.4 5.1 31.1 445.2 9.9 28.1 511.7 7.5 24.4 9.91 3.17 14.19 65OS35LP 224.0 558.4 298.3 5.4 34.1 444.9 9.7 26.6 506.5 7.5 24.0 9.71 3.20 12.95 55OS45LP 226.3 554.4 290.6 5.9 37.1 444.1 9.0 24.1 / 7.4 24.1 9.01 3.24 12.58 100LP 244.3 511.1 306.1 10.3 / 424.6 5.3 / 454.1 8.4 / 10.36 3.90 2.98
Ti: ignition temperature,℃; Tb: burnout temperature,℃; T1, T2and T3: are temperatures of thefirst, second and third peaks, respectively. -Rp: maximum weight loss
rate, %/min; -Rp1, -Rp2and-Rp3are values of thefirst, second and third peaks, respectively; -Rv: average weight loss rate, %/min; Mf: combustion residue mass; and
process underwent different stages involving many complex reactions. E needed for OS was obviously smaller than LP during the whole de-composition process. It implied that OS was burned more easily than LP which might attributed to higher calorific value of OS, while LP pos-sessed a complicated composition and stable structure.
For example, E estimates at the ignition temperatures (α≈0.1) were 129.3 kJ/mol for OS and 211.4 kJ/mol for LP which indicated a better ignition performance of OS. This was consistent with the above result that Tiof OS was lower than that of LP. The change in E of OS withα
showed that E increased initially, with a peak in stage 1 corresponding to devolatilization of the light volatiles (α ≈ 0.1–0.3). E remained stable in stage 2 due to devolatilization and combustion of the heavy
volatiles (α ≈ 0.3–0.6) of OS. Eventually, E reduced gradually during the oxidation of char (α≈0.6–0.9). This trend may be because the de-composition and combustion of the heavy volatiles of OS were domi-nant where the heat was sufficient to maintain a stable and continuous combustion in the medium term. However, more activation energy appeared to be needed in the transition from stage 1 to 2. The relatively low level of E needed for stage 3 was related mainly to the high calorific value and lowfixed carbon content of OS.
As for LP, E underwent two gradually increasing peaks at 300℃ and 340℃ and then decreased sharply during the devolatilization and combustion of its volatiles (α ≈ 0.1–0.75), In the combustion of bio-mass char (α≈0.75–0.9), E increased initially and then decreased which might be attributed to the different reactions or mechanisms involved in the pyrolysis of the three components [66]. Ball et al. [67] pointed out that the charring process of lignin pyrolysis was a low ac-tivation energy exothermal process, whereas the volatilization decom-position of cellulose was a high activation energy endothermal process. Thus, the increase in E in thefirst two peaks probably corresponded to the successive decomposition and combustion of (hemi)cellulose and cellulose, while its decrease may be caused by the heat supplement releasing from the decomposition and combustion of residual lignin. Our results are in close agreement with the previousfindings that E needed for the decomposition of cellulose,fixed carbon, hemicellulose and lignin decreased in the same order [68].
With the increased blend ratio, E needed for the devolatilization of light components of OS (stage 1) increased sharply from 135.5 to 165.8 kJ/mol. The corresponding peak also gradually shifted to a higher temperature zone, indicating that devolatilization stage became more difficult. However, in the devolatilization and combustion of heavy components as well asfixed carbon oxidation of OS (stages 2 and 3), E was lower for OS and decreased gradually with the increased blend ratio from 108.6 to 86.8 kJ/mol. E needed for the devolatilization of the light volatiles of OS increased. In other words, the devolatiliza-tion of the light volatiles became more difficult. It is consistent with our previous interaction analysis that the release of the volatiles of the blends was inhibited. With the increased blend ratio, E needed for the decomposition and combustion of the heavy components and fixed carbon of OS reduced which was also consistent with the conclusion that the char oxidation of OS was enhanced.
Fig. 6shows changes in average apparent activation energy (Ea)
with the blend ratio during co-combustion. Ea estimates by the KAS
method were 215.3 kJ/mol for LP and 118.2 kJ/mol for OS. With the increased blend ratio, Earemained basically consistent with that of OS
even when 5–45% of LP was added. The minimum Eavalue (112.5 kJ/
mol) was obtained when the blend ratio was between 15% and 20%.
3.6.2. Kinetic model determined by integral master-plots method
The integral master-plots approach was adopted to estimate the kinetic parameters of apparent activation energy (E), apparent pre-ex-ponential factor (A), and mechanism function (f(a)), for the three thermal decomposition steps of OS and its blends. The activation energy can be determined accurately using the iso-conversional methods without previously assuming a reaction model as was highly re-commended by the Kinetics Committee of the International Confederation for Thermal Analysis and Calorimetry (ICTAC Kinetics Committee) [69]. Thus, the E values for the three decomposition stages of OS and its blends were estimated and are presented together with coefficients of determination (R2) in Tables S-2–S-4, respectively. All
the R2values were in the range of 0.9542 to 0.9999 indicating that the
results were reliable. For OS as an example, the plots of P(u)/P(u0.5)
versusα from the TG data obtained at the four heating rates are shown in Fig. S-1. The P(u)/P(u0.5) plots of each stage for OS degradation at
the four heating rates were almost identical, thus indicating that each stage of OS was best described by a single kinetic model [70,71].
The experimental and theoretical master-plots of P(u)/P(u0.5) were
compared for each stage for OS degradation at 20℃/min (Fig. 7a–c). Fig. 4. (a) Comparison of calculated versus experimental TG curves; (b)
Variations in profiles of ΔW at five blend ratios; (c) Comparison of experimental versus calculated TG/DTG curves of 55OS45LP sample.
The P(u)/P(u0.5) plots of the three stages were close to the theoretical
master-plots; namely, D3, R2 and A3 models, respectively. Moreover, the plots of ln[β/T2] versus 1/T for the given conversion degrees should
be straight lines if a suitable reaction function was chosen (Fig. S-2). Based on E estimates by master-plots and KAS methods, and R2
values, the most potential models were determined. Kinetic triplets (E, A, and f(a)) for the three stages of OS and its blends at 20℃/min are shown inTable 5. D3, R1 or R2 and A3 models appeared to stage 1, 2 and 3, respectively. The activation energy highly depends on the acti-vated molecule level, diffusion limitation and organic impurities during the thermal decomposition of the solid samples. E1 increased form
96.8–143.0 kJ/mol, while A1showed an opposite trend which indicated
that devolatilization became more difficult. This was consistent with the interaction analysis results. The trend of E2was also consistent with
Fig. 5, and stage 3 had the lowest E and A due to the low levels of activated molecule andfixed carbon.
To prove the validity of kinetic results, the experimental and esti-mated conversionα of OS at 20℃/min were compared inFig. 7d–f. The estimated kinetic parameters could accurately predicte and reproduce the thermal decomposition process of OS and its blends, indicating that the results of kinetic analysis were correct.
Table 4
Characteristics parameters of co-combustion profiles.
Samples β ti tp tb Tv ΔT1/2 Δt1/2 Ci Cb×103 Dv×106 S×106 OS 10 18.5 40.0 49.1 223 94.2 9.5 0.008 0.33 0.702 0.444 20 10.1 20.6 26.1 196 114.8 6.7 0.059 3.42 1.248 1.427 30 7.3 14.3 18.4 183 139.4 5.5 0.161 11.50 1.446 2.404 40 5.8 11.0 14.7 175 156.6 4.7 0.322 26.86 1.629 3.383 95OS5LP 10 18.6 40.3 48.7 219 92.7 9.4 0.008 0.33 0.713 0.446 20 10.2 20.7 26.1 196 108.8 6.4 0.061 3.70 1.362 1.489 30 7.3 14.2 18.2 185 132.4 5.3 0.167 12.53 1.551 2.552 40 5.8 10.9 14.6 173 / / 0.341 / / 3.678 85OS15LP 10 18.6 40.8 48.8 213 106.2 10.7 0.006 0.24 0.529 0.380 20 10.1 20.5 26.5 193 108.9 5.4 0.053 3.73 1.191 1.289 30 7.4 14.1 18.3 183 138.4 4.5 0.144 12.94 1.318 2.198 40 5.9 10.8 14.6 174 155.5 4.7 0.310 26.33 1.594 3.328 75OS25LP 10 18.6 41.0 49.1 210 112.5 11.3 0.006 0.20 0.453 0.345 20 10.3 20.6 26.6 192 111.7 5.6 0.046 3.21 1.038 1.112 30 7.4 14.1 18.4 182 142.7 4.6 0.128 11.09 1.145 1.946 40 5.9 10.8 14.5 173 162.3 4.9 0.277 23.03 1.389 3.003 65OS35LP 10 18.5 41.0 49.1 207 114.3 11.5 0.005 0.18 0.423 0.326 20 10.3 20.6 26.5 190 108.8 5.4 0.045 3.24 1.055 1.110 30 7.5 14.0 18.4 181 144.8 4.7 0.113 9.76 1.018 1.710 40 5.9 10.7 14.6 173 166.3 4.0 0.254 25.86 1.244 2.749 55OS45LP 10 18.8 40.7 49.5 208 116.8 11.8 0.005 0.17 0.389 0.298 20 10.4 20.6 26.3 190 111.5 5.5 0.041 2.97 0.957 1.028 30 7.5 14.0 18.4 181 147.5 4.8 0.106 9.09 0.944 1.633 40 5.9 10.7 14.5 173 169.9 4.0 0.229 23.10 1.108 2.498 LP 10 20.9 27.7 45.3 204 106.0 10.4 0.008 0.38 0.758 0.386 20 11.2 14.0 24.1 196 100.8 4.7 0.065 6.51 1.713 1.325 30 7.9 9.8 16.5 187 104.1 3.0 0.207 32.41 2.647 2.814 40 6.3 7.7 12.8 180 93.6 1.9 0.489 120.70 4.440 4.984
β: heating rate; ti: ignition time; tp: corresponding time of maximum weight loss rate; tb: burnout time;Δt1/2: time interval at the half value of -Rp;ΔT1/2: temperature
interval at the half value of -Rp; -Rv: initial release temperature of volatile matter (℃) (namely the corresponding temperature at mass loss rate of 0.1 mg/min); Ci:
ignition index,%/min/℃3; C
i: burnout index, %/min/℃4; Dv: volatile matter release index, %/min/℃3; S: comprehensive combustibility index, %2/min2/℃3.
Fig. 7. G(a)/G(0.5) versus α for various reaction models and P(u)/P(0.5) versus α for three stages (a) 1, (b) 2 and (c) 3 of OS at 20 ℃/min; Calculated and experimental conversion data for three stages (d) 1, (e) 2 and (f) 3 of OS at 20℃/min.
Table 5
Kinetic triplets (E, A, and f(a)) for various stages of OS and its blends at 20℃/min by master-plots method.
Samples Stage1 Stage2 Stage3
f1(a) E1(kJ/mol) R2 A1(s−1) f2(a) E2(kJ/mol) R2 A2(s−1) f3(a) E3(kJ/mol) R2 A3(s−1)
OS [1-(1-a)1/3]2 96.8 0.9999 1.816 × 108 1-(1-a)1/2 123.2 0.9964 8.252 × 109 [-ln(1-a)]1/3 73.5 0.9878 3.092 × 104
95OS 5LP [1-(1-a)1/3]2 107.9 0.9996 2.613 × 108 1-(1-a)1/2 131.2 0.9922 1.181 × 109 [-ln(1-a)]1/3 78.8 0.9873 9.464 × 104 95OS 15LP [1-(1-a)1/3]2 109.1 0.9978 9.244 × 107 1-(1-a)1/2 107.2 0.9986 4.381 × 108 [-ln(1-a)]1/3 71.9 0.9832 4.980 × 104
95OS 25LP [1-(1-a)1/3]2 122.8 0.9947 4.781 × 107 a 111.8 0.9941 2.769 × 107 [-ln(1-a)]1/3 73.8 0.9808 1.029 × 104
95OS 35LP [1-(1-a)1/3]2 117.4 0.9946 4.005 × 107 a 112.4 0.9926 3.324 × 107 [-ln(1-a)]1/3 76.1 0.9750 5.786 × 103
4. Discussion
Since biomass usually contains significantly higher amounts of AAEMs (mainly K, Na, Ca, and Mg) than other fuels such as coal, and petroleum, it is prone to volatilize and lead to serious issues with the thermal utilization process such as slagging, agglomeration, deposition, and heated side corrosion [72,73]. Since LP contained over 80% AAEMs, while OS contained more thermal stable metal oxides (SiO2,
Fe2O3, and Al2O3) (Table 1), this big difference in ash composition
appeared to significantly influence co-combustion.
AAEMs can be present in biomass mainly as ionic salts (e.g. KCl), organic compounds (e.g. R−COO–K+) and minerals (e.g. K
2O3Si).
Kawamoto et al. [74] pointed out that K+, Na+, Ca2+, and Mg2+
promoted the generation of solid chars, small molecular compounds, and H2O in cellulose pyrolysis. The complex release characteristic of
AAEMs is affected by temperature, heating rate, and contents of Cl, S and other mineral elements during which Cl and S can increase the mobility of Na and K. However, there are studies showing that Cl and S can be released, while the K content of biomass remains intact at low temperatures between 200 and 700℃ [75,76]. Fuentes et al. [77] showed that K catalyzed both volatile release and char oxidation stages by shifting their DTG characteristic peaks to lower temperatures. Also, Saddawi et al. [78] found that characteristic temperature for devolati-lization decreased above 50℃for 2% K(Na)-impregnated willow com-pared to demineralized willow. During the partial release of AAEMs from biomass with decomposition of its lignin and (hemi)cellulose during pyrolysis, the release ratio of Na was found to be higher than that of K due to its tightly bonded with carbon [79]. Moreover, com-paring to alkali metal, alkaline earth metal was hard to vaporize during pyrolysis process, a low release of K below 1000 K from devolatilization of various biomass samples was reported relative to the subsequent releases during char combustion [80,81]. The average activation energy during the pyrolysis stage was found to be 223 kJ/mol for Na and 185 kJ/mol for K, with the releases of up to 55% of Na and less than 10% of K [82].
Overall, the release amount of K and Ca during devolatilization (stages 1 and 2) was very limited, while most of AAEMs remained in solid phrase which appeared to participate in char combustion (stage 3) and particle shrinking process at higher temperatures. Negative ΔW values between 200 and 500℃ (Fig. 4b) indicated that the weight loss due to devolatilization was suppressed with the added LP. Moreover, the characteristic temperatures (Ti, T1and T2) were very close to those
of OS (Table 3). More importantly,Fig. 4c showed that the measured T1
and T2were very close to the calculated ones, and the mass loss rate
measured due to devolatilization was lower than the calculated one, indicating no obvious catalytic effect on devolatilization.
AAEMs had a catalytic impact on the devolatilization and oxidation processes, whereas no obvious catalytic effect on devolatilization of the volatiles of OS occurred when compared to the oxidation of char in our study. One of the reasons behind this case may be that catalytic effect was significantly weakened by the low initial AAEMs content of the blends (below 1.18%), given the ash content of LP (3.30%) [83]. Another reason may be that the catalysis effect of AAEMs may be diminished by the high thermal stable metal oxide content of OS (20.72%) with higher melting and volatilization temperatures than alkali metals. For example, the presence of Fe2O3 was reported to inhibit the decomposition of
sludge organic matter [65,84]. Alternatively, sludge powder coating by sticky liquid products such as bio-oils and pyrolysis oils may render volatiles more difficult to release. However, the experimental mass loss rate was higher than the calculated one above 500℃ (stage 3) which may be attributed to the catalytic effect of AAEMs.
5. Conclusions
(Co-)combustion characteristics of oily sludge, litchi peels and their blends were comprehensively evaluated using thermogravimetric
experiments, and combustion and kinetic parameters. Our results showed that LP has a better combustion performance than OS according to the indices of ignition, burnout, and volatile matter release. Strong interactions emerged in the three co-combustion stages were the in-hibition effect on devolatilization and the enhancement effect on char oxidation. Average apparent activation energy by the KAS method was estimated at 215.3 kJ/mol for LP and 118.2 kJ/mol for OS and reached its minimum (112.5 kJ/mol) with 15% to 20% LP added. Also, average activation energy increased from 135.5 to 165.8 kJ/mol but decreased from 108.6 to 86.8 kJ/mol in the devolatilization and char combustion stages, respectively. Kinetic triplets for the stages 1 to 3 were also es-timated using the integral master-plots method, with D3, R2 and A3 as the best reaction models for oily sludge and its blends, respectively.
Acknowledgments
This study was financially supported by the Scientific and Technological Planning Project of Guangzhou, China (No. 201704030109; 2016201604030058), the Science and Technology Planning Project of Guangdong Province, China (No.2017A040403059;2017A040403045;2017A050501036), and the Guangdong Special Support Program for Training High Level Talents (No. 2014TQ01Z248).
Appendix A. Supplementary data
Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.tca.2018.06.009.
References
[1] G. Hu, J. Li, G. Zeng, Recent development in the treatment of oily sludge from petroleum industry: a review, J. Hazard. Mater. 261 (2013) 470–490. [2] G. Jing, T. Chen, M. Luan, Studying oily sludge treatment by thermos chemistry,
Arab. J. Chem. 9 (2016) 457–460.
[3] B. Lin, J. Wang, Q. Huang, Y. Chi, Effects of potassium hydroxide on the catalytic pyrolysis of oily sludge for high-quality oil product, Fuel 200 (2017) 124–133. [4] Y. Zhang, Q. Zhao, J. Jiang, K. Wang, L. Wei, J. Ding, et al., Acceleration of organic
removal and electricity generation from dewatered oily sludge in a bioelec-trochemical system by rhamnolipid addition, Bioresour. Technol. 243 (2017) 820–827.
[5] M. Xu, J. Zhang, H. Liu, H. Zhao, W. Li, The resource utilization of oily sludge by co-gasification with coal, Fuel 126 (2014) 55–561.
[6] N. Buyukkamaci, E. Kucukselek, Improvement of dewatering capacity of a petro-chemical sludge, J. Hazard. Mater. 144 (2007) 323–327.
[7] E. Zubaidy, D. Abouelnasr, Fuel recovery from waste oily sludge using solvent ex-traction, Process Saf. Environ. 88 (2010) 318–326.
[8] A. Cambiella, J. Benito, C. Pazos, J. Coca, Centrifugal separation efficiency in the treatment of waste emulsified oils, Chem. Eng. Res. Des. 84 (2006) 69–76. [9] L. Qin, J. Han, X. He, Y. Zhan, F. Yu, Recovery of energy and iron from oily sludge
pyrolysis in afluidized bed reactor, J. Environ. Manage. 154 (2015) 177–182. [10] L. Zhou, X. Jiang, J. Liu, Characteristics of oily sludge combustion in circulating
fluidized beds, J. Hazard. Mater. 170 (2009) 175–179.
[11] J. Liu, X. Jiang, L. Zhou, H. Wang, X. Han, Co-firing of oil sludge with coal–water slurry in an industrial internal circulatingfluidized bed boiler, J. Hazard. Mater. 167 (2009) 817–823.
[12] E. Nkhalambayausi, C. Mampholo, O. Fayemiwo, F. Bezza, Biosurfactant assisted recovery of the C5–C11 hydrocarbon fraction from oily sludge using biosurfactant producing consortium culture of bacteria, J. Environ. Manage. 196 (2017) 261–269. [13] M. Nazem, O. Tavakoli, Bio-oil production from refinery oily sludge using
hydro-thermal liquefaction technology, J. Supercrit. Fluids 127 (2017) 33–40. [14] T. Hwa, S. Jeyaseelan, Comparison of lime and alum as oily sludge conditioners,
Water Sci. Technol. 36 (1997) 117–124.
[15] J. Zhang, J. Li, R. Thring, L. Liu, Application of ultrasound and Fenton’s reaction process for the treatment of oily sludge, Procedia Environ. Sci. 18 (2013) 686–693. [16] B. Cieślik, J. Namieśnik, P. Konieczka, Review of sewage sludge management:
standards, regulations and analytical methods, J. Clean. Prod. 90 (2015) 1–15. [17] L. Yu, J. Zheng, X. Yuan, H. Song, J. Peng, M. Ren, Characterization and mass balance of PCDD/Fs during the co-combustion of sewage sludge in a grate-type municipal solid waste incineration, Procedia Environ. Sci. 31 (2016) 303–308. [18] S. Deng, X. Wang, H. Tan, H. Mikulčić, F. Yang, Z. Li, et al., Thermogravimetric
study on the Co-combustion characteristics of oily sludge with plant biomass, Thermochim. Acta 633 (2016) 69–76.
[19] G. Hu, J. Li, X. Zhang, Y. Li, Investigation of waste biomass co-pyrolysis with pet-roleum sludge using a response surface methodology, J. Environ. Manage. 192 (2017) 234–242.
[20] R. Font, A. Fullana, J. Conesa, F. Llavador, Analysis of the pyrolysis and combustion of different sewage sludges by TG, J. Anal. Appl. Pyrol. 58–59 (2001) 927–941. [21] J. Chen, L. Mu, B. Jiang, H. Yin, X. Song, A. Li, TG/DSC-FTIR and Py-GC
investigation on pyrolysis characteristics of petrochemical wastewater sludge, Bioresour. Technol. 192 (2015) 1–10.
[22] A. Demirbas, Combustion characteristics of different biomass fuels, Prog. Energy Combust. 30 (2004) 219–230.
[23] Y. Liao, C. Zeng, X. Ma, J. Song, Thermogravimetric analysis of pyrolysis and combustion characteristics of typical biomass in south China, J. S. China Univ. Technol. (Nat. Sci. Ed.) 41 (2013) [in Chinese]..
[24] L. Huang, J. Liu, Y. He, S. Sun, J. Chen, J. Sun, et al., Thermodynamics and kinetics parameters of co-combustion between sewage sludge and water hyacinth in CO2/O2 atmosphere as biomass to solid biofuel, Bioresour. Technol. 218 (2016) 631–642. [25] J. Chen, J. Liu, Y. He, L. Huang, S. Sun, J. Sun, et al., Investigation of co-combustion
characteristics of sewage sludge and coffee grounds mixtures using thermogravi-metric analysis coupled to artificial neural networks modeling, Bioresour. Technol. 225 (2017) 234–245.
[26] L. Mu, J. Chen, P. Yao, D. Zhou, L. Zhao, H. Yin, Evaluation of co-pyrolysis pet-rochemical wastewater sludge with lignite in a thermogravimetric analyzer and a packed-bed reactor: pyrolysis characteristics, kinetics, and products analysis, Bioresour. Technol. 221 (2016) 147–156.
[27] L. Huang, C. Xie, J. Liu, X. Zhang, K. Chang, J. Kuo, et al., Influence of catalysts on co-combustion of sewage sludge and water hyacinth blends as determined by TG-MS analysis, Bioresour. Technol. 247 (2018) 217–225.
[28] M. Roni, S. Chowdhury, S. Mamun, M. Marufuzzaman, W. Lein, S. Johnson, Biomass co-firing technology with policies, challenges, and opportunities: a global review, Renew. Sustain. Energy Rev. 78 (2017) 1089–1101.
[29] Z. Bundhoo, R. Mohee, Ultrasound-assisted biological conversion of biomass and waste materials to biofuels: a review, Ultrason. Sonochem. 40 (2018) 298–313. [30] N. Pandey, S. Joshi, C. Singh, S. Kumar, S. Rajput, R. Khandal, Enhancing shelf life
of litchi (Litchi chinensis) fruit through integrated approach of surface coating and gamma irradiation, Radiat. Phys. Chem. 85 (2013) 197–203.
[31] B. Yang, M. Zhao, Y. Liu, Z. Bao, Characterization of litchi pericarp polysaccharide, Nat. Prod. Res. Dev. 17 (2005) 685–687.
[32] D. Liang, F. Lin, G. Yang, X. Yue, Z. Zhang, et al., Radiation physics and chemistry quality preservation of litchi fruit during frozen storage, LWT—Food Sci. Technol. 60 (2015) 948–956.
[33] P. Spencer, V. Palmer, The enigma of litchi toxicity: an emerging health concern in southern Asia, Lancet Glob. Health 5 (2017) 383–384.
[34] H. Kessy, K. Wang, L. Zhao, M. Zhou, Z. Hu, Enrichment and biotransformation of phenolic compounds from litchi pericarps with angiotensin I-converting enzyme (ACE) inhibition activity, LWT—Food Sci. Technol. 87 (2018) 301–309. [35] G. Jiang, S. Lin, L. Wen, Y. Jiang, M. Zhao, F. Chen, et al., Identification of a novel
phenolic compound in litchi (Litchi chinensis Sonn.) pericarp and bioactivity eva-luation, Food Chem. 136 (2013) 563–568.
[36] Z. Sun, S. Wang, L. Yan, M. Xiao, D. Han, Y. Meng, Mesoporous carbon materials prepared from litchi shell as sulfur encapsulator for lithium-sulfur battery appli-cation, J. Power Sources 324 (2016) 547–555.
[37] Y. Yi, J. Lv, Y. Liu, G. Wu, Synthesis and application of modified Litchi peel for removal of hexavalent chromium from aqueous solutions, J. Mol. Liq. 225 (2017) 28–33.
[38] M. Nagle, K. Habasimbik, B. Mahayothee, M. Haewsungcharern, M. Janjai, J. Müller, Fruit processing residues as an alternative fuel for drying in northern Thailand, Fuels 90 (2011) 818–823.
[39] F. Meng, J. Yu, A. Tahmasebi, Y. Han, Pyrolysis and combustion behavior of coal gangue in O2/CO2 and O2/N2 mixtures using thermogravimetric analysis and a drop tube furnace, Energy Fuels 27 (2013) 2923–2932.
[40] J. Chen, L. Mu, J. Cai, P. Yao, X. Song, H. Yin, et al., Pyrolysis and oxy–fuel com-bustion characteristics and kinetics of petrochemical wastewater sludge using thermogravimetric analysis, Bioresour. Technol. 198 (2015) 115–123. [41] J. Chen, Y. Wang, X. Lang, X. Ren, S. Fan, Comparative evaluation of thermal
oxidative decomposition for oil-plant residues via thermogravimetric analysis: thermal conversion characteristics, kinetics, and thermodynamics, Bioresour. Technol. 243 (2017) 37–46.
[42] X. Li, Y. Lv, B. Ma, S. Jian, H. Tan, Thermogravimetric investigation on co-com-bustion characteristics of tobacco residue and high-ash anthracite coal, Bioresour. Technol. 102 (2011) 9783–9787.
[43] X. Liu, M. Chen, D. Yu, Oxygen enriched co-combustion characteristics of herbac-eous biomass and bituminous coal, Thermochim. Acta 569 (2013) 17–24. [44] A. Jain, A. Mehra, V. Ranade, Processing of TGA data: analysis of isoconversional
and modelfitting methods, Fuel 165 (2016) 490–498.
[45] T. Wanjun, L. Yuwen, Z. Hen, W. Zhiyong, W. Cunxin, New temperature integral approximate formula for non-isothermal kinetic analysis, J. Therm. Anal. Calorim. 74 (2003) 309–315.
[46] H. Schmidt, W. Kaminsky, Pyrolysis of oil sludge in afluidised bed reactor, Chemosphere 45 (2001) 285–290.
[47] Z. Wang, Q. Guo, X. Liu, C. Cao, Low temperature pyrolysis characteristics of oil sludge under various heating conditions, Energy Fuels 21 (2007) 957–962. [48] D. Lai, Z. Chen, Y. Shi, L. Lin, J. Zhan, S. Gao, Pyrolysis of oil shale by solid heat
carrier in an innovative moving bed with internals, Fuel 159 (2015) 943–951. [49] C. Chen, X. Ma, K. Liu, Thermogravimetric analysis of microalgae combustion under
different oxygen supply concentrations, Appl. Energy 88 (2011) 3189–3196. [50] Z. Xie, X. Ma, The thermal behaviour of the co-combustion between paper sludge
and rice straw, Bioresour. Technol. 146 (2013) 611–618.
[51] C. Chao, H. Chiang, C. Chen, Pyrolytic kinetics of sludge from a petrochemical factory wastewater treatment plant––a transition state theory approach, Chemosphere 49 (2002) 431–437.
[52] X. Yang, Z. Jiang, Kinetic studies of overlapping pyrolysis reactions in industrial
waste activated sludge, Bioresour. Technol. 100 (2009) 3663–3668.
[53] C. Chen, F. Chen, Z. Cheng, Q. Chan, S. Kook, G. Yeoh, Emissions characteristics of NOx and SO2 in the combustion of microalgae biomass using a tube furnace, J. Energy Inst. 90 (2017) 806–812.
[54] N. Sonoyama, K. Nobuta, T. Kimura, S. Hosokai, J. Hayashi, T. Tago, et al., Production of chemicals by cracking pyrolytic tar from Loy Yang coal over iron oxide catalysts in a steam atmosphere, Fuel Process. Technol. 92 (2011) 771–775. [55] X. Wei, U. Schnell, K. Hein, Behaviour of gaseous chlorine and alkali metals during
biomass thermal utilization, Fuel 84 (2005) 841–848.
[56] Y. Yu, S. Kim, J. Lee, K. Lee, Kinetic studies of dehydration, pyrolysisand com-bustion of paper sludge, Energy 27 (2002) 457–469.
[57] J. Liu, L. Huang, X. Ning, J. Kuo, J. Sun, K. Chang, et al., Analysis of co-combustion characteristics of sewage sludge and water hyacinth, Acta Sci. Circum. 36 (2016) 2955–2967.
[58] W. Buah, A. Cunliffe, P. Williams, Characterization of products from the pyrolysis of municipal solid waste, Process. Saf. Environ. 85 (2007) 450–457.
[59] C. Chen, X. Ma, Y. He, Co-pyrolysis characteristics of microalgae Chlorellavulgaris and coal through TGA, Bioresour. Technol. 117 (2012) 264–273.
[60] Y. Park, J. Kim, S. Kim, Y. Park, Pyrolysis characteristics and kinetics of oak trees using thermogravimetric analyzer and micro-tubing reactor, Bioresour. Technol. 100 (2009) 400–405.
[61] H. Haykiri-Acma, S. Yaman, Interaction between biomass and different rank coals during co-pyrolysis, Renew. Energy 35 (2010) 288–292.
[62] X. Peng, X. Ma, Y. Lin, Z. Guo, S. Hu, X. Ning, et al., Co-pyrolysis between micro-algae and textile dyeing sludge by TG-FTIR: kinetics and products, Energy Convers. Manage. 100 (2015) 391–402.
[63] S. Grierson, V. Strezov, G. Ellem, R. McGregor, J. Herbertson, Thermal character-isation of microalgae under slow pyrolysis conditions, J. Anal. Appl. Pyrol. 85 (2009) 118–123.
[64] Y. Liao, X. Ma, Thermogravimetric analysis of the co-combustion of coal and paper mill sludge, Appl. Energy 87 (2010) 3526–3532.
[65] D. Vamvuka, N. Salpigidou, E. Kastanaki, S. Sfakiotakis, Possibility of using paper sludge in co-firing applications, Fuel 88 (2009) 637–643.
[66] H. Yang, R. Yan, H. Chen, D. Lee, C. Zheng, Characteristics of hemicellulose, cel-lulose and lignin pyrolysis, Fuel 86 (2007) 1781–1788.
[67] R. Ball, A. McIntosh, J. Brindley, Feedback processes in cellulose thermal decom-position: implications forfire-retarding strategies and treatments, Combust. Theor. Model. 8 (2004) 281–291.
[68] Q. Bach, K. Tran, Ø. Skreiberg, Combustion kinetics of wet-torrefied forest residues using the distributed activation energy model (DAEM), Appl. Energy 185 (2017) 1059–1066.
[69] S. Vyazovkin, A.K. Burnham, J.M. Criado, L.A. Pérez-Maqueda, C. Popescu, N. Sbirrazzuoli, ICTAC kinetics committee recommendations for performing kinetic computations on thermal analysis data, Thermochim. Acta 520 (2011) 1–19. [70] S. Zou, Y. Wu, M. Yang, C. Li, J. Tong, Pyrolysis characteristics and kinetics of the
marine microalgae Dunaliella tertiolecta using thermogravimetric analyzer, Bioresour. Technol. 101 (2010) 359–365.
[71] C. Chen, X. Ma, Y. He, Co-pyrolysis characteristics of microalgae Chlorella vulgaris and coal through TGA, Bioresour. Technol. 117 (2012) 264–273.
[72] Y. Niu, H. Tan, S. Hui, Ash-related issues during biomass combustion: alkali-in-duced slagging, silicate melt-inalkali-in-duced slagging (ash fusion), agglomeration, corro-sion, ash utilization, and related countermeasures, Prog. Energy Combust. Sci. 52 (2016) 1–61.
[73] C. Gilbe, M. Ohman, E. Lindstrom, D. Bostrom, R. Backman, R. Samuelsson, J. Burvalll, Slagging characteristics during residential combustion of biomass pel-lets, Energy Fuels 22 (2008) 536–543.
[74] H. Wang, R. Srinivasan, F. Yu, P. Steele, Q. Li, B. Mitchell, Effect of acid, alkali, and steam explosion pretreatments on characteristics of bio-oil produced from pine-wood, Energy Fuels 25 (2011) 758–764.
[75] S.B. Saleh, J.P. Flensborg, T.K. Shoulaifar, et al., Release of chlorine and sulfur during biomass torrefaction and pyrolysis, Energy Fuels 28 (2014) 738–746. [76] T. Khazraie Shoulaifar, N. DeMartini, M. Zevenhoven, F. Verhoeff, J. Kiel, M. Hupa,
Ash-forming matter in torrefied Birch wood: changes in chemical association, Energy Fuels 27 (2013) 684–690.
[77] M.E. Fuentes, F.J. Nowakowski, M.L. Kubacki, J.M. Cove, T.G. Bridgeman, J.M. Jones, Survey of influence of biomass mineral matter in thermochemical conversion of short rotation willow coppice, J. Energy Inst. 81 (2008) 234-24. [78] A. Saddawi, J.M. Jones, A. Williams, Influence of alkali metals on the kinetics of the
thermal decomposition of biomass, Fuel Process. Technol. 104 (2012) 189–197. [79] L. Jiang, S. Hu, J. Xiang, S. Su, L. Sun, K. Xu, Y. Yao, Release characteristics of alkali
and alkaline earth metallic species during biomass pyrolysis and steam gasification process, Bioresour. Technol. 116 (2012) 278–284.
[80] S.S. Kim, Y.S. Kang, H.D. Lee, J.K. Kim, S.C. Hong, Release of potassium and sodium species during combustion of various rank coals, biomass, sludge and peats, J. Ind. Eng. Chem. 18 (2012) 199–203.
[81] P.E. Mason, L. Darvell, J.M. Jones, A. Williams, Observations on the release of gas-phase potassium during thecombustion of single particles of biomass, Fuel 182 (2016) 110–117.
[82] H. Fatehi, Z.S. Li, X.S. Bai, M. Aldén, Modeling of alkali metal release during bio-mass pyrolysis, Proc. Combust. Inst. 36 (2017) 243–251.
[83] A. Carvalho, M. Rabaçal, M. Costa, M.U. Alzueta, M. Abián, Effects of potassium and calcium on the early stages of combustion of single biomass particles, Fuel 209 (2017) 787–794.
[84] J.G. Shao, R. Yan, H.P. Chen, H.P. Yang, D.H. Lee, Catalytic effect of metal oxides on pyrolysis of sewage sludge, Fuel Process. Technol. 91 (2010) 111–118.
