Chemical Engineering Journal 374 (2019) 338–346

Contents lists available at ScienceDirect

Chemical Engineering Journal

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

Kinetic study of the effect of in-situ mineral solids on pyrolysis process of oil T
sludge
⁎ ⁎
Yi Qu, Aimin Li , Dong Wang, Lei Zhang, Guozhao Ji
Key Laboratory of Industrial Ecology and Environmental Engineering of Ministry of Education, School of Environmental Science and Technology, Dalian University of
Technology, Linggong Road 2, Dalian 116023, China

H I GH L IG H T S

• The effect of solids contained in the oil sludge on pyrolysis process was studied.
• The presence of mineral solids could reduce the activation energy of pyrolysis.
• The solids provided more surface for the devolatilization of light organics.
• The solids offered phase interface for thermal decomposition.
• The solids participated in the char formation as the nucleating agent.

A R T I C LE I N FO A B S T R A C T

Keywords: Oil sludge (OS) is composed of water, oil and inert solids (silica, clay minerals, etc.). Although solids were
Oil sludge ubiquitous in oil sludge, its potential effects on altering pyrolysis characteristics had rarely been studied.
Solids Therefore, the role that solids played during the pyrolysis process of oil sludge was investigated by thermo-
Solid-free gravimetric analysis (TGA) in this study. The kinetic triplets of both OS (contained solids) and extracted oil (EO,
Pyrolysis
solid-free) from OS during the pyrolysis processes were determined by a combined method, including the in-
Mechanism
Kinetic
tegral master-plots method and iso-conversional method. The activation energy values for OS pyrolysis varied
between 28 and 107 kJ mol−1, and pre-exponential factor ranged from 10 to 104 s−1. Whereas, the activation
energy values for EO pyrolysis ranged from 67 to 193 kJ mol−1, and its pre-exponential factor varied from 106 to
1010 s−1. These results suggested that inert solids had positive influences on the pyrolysis of OS because the
activation energy for OS pyrolysis was obviously lower than that for EO pyrolysis. The comparison of kinetic
models between OS and EO illustrated that solids not only provided sufficient surface for the devolatilization of
hydrocarbon but also acted as nucleating agent in char formation.

1. Introduction Traditional disposal such as landfilling and incineration wasted this

Oil sludge (OS), which is generated from the petroleum industry, is
a complex mixture mainly containing water, organic and inorganic
residues. Due to the content of pathogenic bacteria, parasite and heavy
metals that are threats to environment and human health, OS is re-
garded as a hazardous waste. In the past several years, more than one
billion tons of OS were generated annually by petroleum industries all
over the world [1]. Unfortunately, most of the OS was disposed in
landfills or incinerated which still posed risks to the environment.
Therefore, there is an urgent need for the effective and affordable
technique for OS disposal. OS contains a high percentage of hydro-
carbons, which are valuable resources for fuel and energy recovery.
valuable content of oil sludge. Hence, oil recovery is also one of the best
environmental choices of OS treatment, which has been intensively
researched. Various methods have been exploited for the oil recovery
from OS such as freezing [2], solvent extraction [3], microwave irra-
diation [4], centrifugation [5], ultrasonic irradiation [6], pyrolysis
[7,8], etc.
Among all the recovery methods, pyrolysis technology is regarded
as an ideal recovery method for OS treatment, which could recover oil
product with little secondary pollutants. Meanwhile, pyrolysis has
crucial influences on other thermal utilization (combustion and gasifi-
cation) of OS since pyrolysis is the first stage of combustion or gasifi-
cation process. Many researchers have carried out investigations

⁎
Corresponding authors.
E-mail addresses: leeam@dlut.edu.cn (A. Li), guozhao.ji@uqconnect.edu.au (G. Ji).

https://doi.org/10.1016/j.cej.2019.05.183

Available online 27 May 2019

1385-8947/ © 2019 Elsevier B.V. All rights reserved.
Y. Qu, et al.
focusing on pyrolysis behaviors and kinetics of OS. According to
Shishkin terminology [7], there were seven pseudo-components de-
composed during OS pyrolysis such as char, condensed aromatics-as-
phaltenes, paraffin base oils, heavy kerosene-gas oil, light kerosene-gas
oil, benzine-ligroin, and water. Chen’s research [8] showed that OS was
thermally degraded via three phases: devolatilization of light organic,
thermal decomposition of medium molecular weight substances, con-
tinuously cracking of the heavy carbonaceous substances with solid
char formation.
Most of the relevant studies regarded the pyrolysis of OS just as the
thermal degradation of oil components in OS. In fact, OS consisted of
not only water and oil but also solids (such as silica and clay minerals).
Although solids were ubiquitous in OS, the interactions between solids
and oil components during pyrolysis of OS sludge were neglected.
Nevertheless, some researchers had found that solids have considerable
influence on the thermal decomposition of oil components. Drici and
Vossoughi [9] studied how solid matrix affected the combustion ki-
netics of crude oil. They observed that solids might decrease the
average reaction activation energy by increasing the surface area of
solids. Besides, solids have effects on the coke formation in vacuum
residues and bitumen. Sanaie [10] studied the effects of solids on coke
formation of Cold Lake bitumen. The results showed that native clays in
Athabasca bitumen result in a lower coke yield, from 4% to 3%. Chen
[11] found adding solids can increase the mass loss which is a preferred
effect on the pyrolysis process of OS. On basis of the above studies, it
was obvious that solids have beneficial effects on the pyrolysis process
of OS. However, most researchers treated these solids as thermal in-
ertia, and few researchers had investigated the effects of solids on
pyrolysis of OS. A deeper understanding of the effect from solid content
would provide an instructive guidance on promoting the efficiency of
oil sludge pyrolysis.
Thermogravimetric analysis (TGA) is an effective technology in in-
vestigating the thermochemical conversions of solid waste and fossil
fuels. It was widely applied to calculate the kinetic triplets of thermal
degradation processes [12], such as oil refinery sludge [13], sewage
sludge [14], lignite [15], char [16], etc. Many researchers [17–19] have
investigated the pyrolysis models and mechanisms by handling TGA
data with the master-plots method. The master-plots method could be
used to estimate the appropriate kinetic model that could reveal the
solid-states reaction mechanisms.
Therefore, the present study aimed to clarify the effects of solids
during the OS pyrolysis process via TGA. For investigation, OS was
subjected to Soxhlet extraction and then separated into extracted oil
(EO) and extracted residues (ER). EO contained almost all active con-
stituents during pyrolysis of OS without solids, while ER was mainly
composed of inert solids. Then, the pyrolysis properties of both OS and
EO at three different heating rates were studied via TGA, and the ki-
netic triplets of both OS and EO were estimated by a combined method.
In this combined method, the activation energy was obtained by two
iso-conversional methods (FWO and KAS). While kinetic model and pre-
exponential factor were obtained by the integral master-plots method.
Finally, by comparing the kinetic models of OS and EO, the mechanisms
of solid on OS pyrolysis was revealed, which will lead to a more effi-
cient oil recovery from pyrolysis.
2. Methods
2.1. Materials
OS used in this research was collected from Panjin oil field,
Liaoning, China. The OS was viscous and finely gritty black mixture.
Table 1 displayed the properties of OS sample, including ultimate
analysis, proximate analysis, higher heating value (HHV) and oil con-
tent. The ultimate analysis was performed using an elemental analyzer
(Vario EL, ELEMENTAR, Germany). The oxygen content was calculated
by the subtraction method in the percentage of weight. The heating
339
Chemical Engineering Journal 374 (2019) 338–346
Table 1
Properties and composition of oil sludge and extracting re-
sidue.
Sample Oil sludge
Proximate analysis (wt% wet basis)
Moisture 21.5
Volatile 68.3
Fixed carbon 3.72
Ash 7.31
Oil content (wt%)a 63.45
Ultimate analysis (wt%)
C 53.84
H 10.09
N 0.61
S 0.31
Ob 35.15
HHV (MJ/kg) 21.35
a
By a Soxhlet extractor.
b
By difference.
values of the samples were measured by a bomb calorimeter.
The extracted oil (EO) was obtained by means of Soxhlet extraction
(STX-06, HongJi, China). OS was extracted with dichloromethane for
48 h at 333 K, and then extracting solution was filtrated with a mem-
brane filter with pore diameter of 0.25 µm to ensure the EO was solid-
free. The solvent of the filtered oil was eliminated by rotary evaporation
(RE2000E, GongYi, China) until the weight did not change. The filtered
solids (extracting residue, ER) were vacuum-dried for 2 h at 378 K, and
EO and ER were stored at 277 K for further use. The X Ray Fluorescence
(XRF) results of ER were present in Table S.1, the main content in ER
was SiO2.
2.2. Experimental methods
Fig. 1 presented the schematic diagram of this work. Thermo-
gravimetry analysis (TGA) and differential thermogravimetry (DTG)
were carried out on a thermal analyzer (TG/DTA 6300, Seiko, Japan)
for OS and EO. The sample of OS used in this research was fresh OS
rather than dry OS which was closer to practical application. OS and EO
were heated to 1073 K at 20, 60, and 100 K min−1, respectively. The
sample weights in TGA were kept at 10 ± 0.2 mg. High purity nitrogen
(99.999%) was used to keep the atmosphere inert as the carrier gas. A
high flow rate of 100 mL min−1 was adopted to blow away the non-
condensable gas products and minimize secondary gas-fuels reactions.
Parallel tests of TGA under same operating conditions but with no
samples was carried out to eliminate the buoyancy effect. All experi-
ments were reduplicated three times to ensure the validity of results.
3. Results and discussion
3.1. Thermal characteristic analysis
Fig. 2 exhibited the plots of thermogravimetric analysis (TGA),
differential thermogravimetry (DTG) of OS and EO under different
heating rates. For more detailed analysis of the pyrolysis process, the
DTG curves were separated into three individual peaks by the tool of
peak fitting in Origin. Fig. 2(b) showed a representative peak fitting
DTG curves of OS at the heating rate of 100 K min−1, as an example.
The curves of peak fitting demonstrated that the pyrolysis process of OS
was comprised of three different phases and each phase represented
different thermal reaction. The initial phase (PH 1) was the devolati-
lization of water and light hydrocarbons. Then, medium molecular
weight components such as aromatics were decomposed in PH 2. Fi-
nally, the last phase (PH 3) was ascribed to the continuously thermal
cracking of heavy carbonaceous substances and chars were generated
Y. Qu, et al. Chemical Engineering Journal 374 (2019) 338–346

Fig. 1. Schematic diagram of the research.

simultaneously by the secondary reactions. It must be pointed out that
the overlapping phenomenon was evidently presented between dif-
ferent phases. Thermal hysteresis caused by the rising of the heating rate
was also found, which was consistent with other studies in solid pyr-
olysis [20]. This phenomenon manifested as the thermal profiles shift to
the higher temperature zone gradually with the increment of heating
rate, which caused by the limitations of heat and mass transfer.
As exhibited in Fig. 2(d), the pyrolysis processes of EO also included
three phases owing to the fact that main reactants of EO were almost
coincident with OS. Simultaneously, the effects of heating rates on
pyrolysis of EO exhibited the same trend as that of OS. However, the
pyrolysis characteristic of EO was quite different. To demonstrate the
differences, pyrolysis feature parameters of both OS and EO were
compared in Table 2. By contrast, the maximum weight loss rate of OS
in PH 1 was much higher than that of EO, which indicated that OS
reacted more intensely than EO in PH 1. Nevertheless, the decom-
position of heavy oil (PH 3) in OS was more stable than in EO on ac-
count of the observation that the maximum weight loss rate of OS was
less than half of EO.
To further investigate the effect of solids on pyrolysis, a comparison

Fig. 2. The TG and DTG curves with the heating rate of 20, 60, and 100 K min−1 (a) TG of OS; (b) DTG of OS; (c) TG of EO; (d) DTG of EO.

340
Y. Qu, et al. Chemical Engineering Journal 374 (2019) 338–346

Table 2
Pyrolysis characteristic parameters for OS and EO at 20, 60, and 100 K min−1.
βa Ti b Tp1c −Rp1d Tp2c −Rp2d Tp3c −Rp3d Tfe M rf
K min−1 K K wt.% min−1 K wt.% min−1 K wt.% min−1 K wt%

OS 20 298.6 371.5 5.73 590.5 3.64 720.1 4.27 805.4 15.19

60 305.9 435.9 18.83 655.4 8.56 775.2 15.67 872.4 14.29
100 322.6 453.0 26.55 689.5 16.73 787 31.97 879.4 12.97

EO 20 333.3 532.6 3.21 651.2 4.43 741.8 10.51 792.0 7.54

60 378.1 567.0 11.61 687.3 13.64 769.1 30.52 823.8 6.01
100 354.8 524.5 8.16 682.9 30.02 780.7 43.33 834.9 4.76

a
β, the heating rate.
b
Ti, the initial decomposition temperature.
c
Tp, the peak temperature of each phase; Tp1, Tp2, and Tp3 are the values of Tp of PH 1, PH 2, and PH 3, respectively.
d
Rp, the maximum weight loss rate of each phase; −Rp1, −Rp2, and −Rp3 are values of −Rp of PH 1, PH 2, and PH 3, respectively.
e
Tf, the terminal decomposition temperature.
f
Mr, the residual weight after pyrolysis experiment.

Fig. 3. The comparison between the TGA curve of original oil sludge and the summation of TGA curves from oil, water, and residue. (a) TGA curves of extracted oil,
extracted residue and water; (b) TGA curve of original oil sludge and the summation of TGA curves; heating rate was 100 K min−1.

between the TGA curve of original oil sludge and the summation of TGA [22]:
curves from oil, water, and residue was performed. The contribution of mp − m
oil, water and residue was multiplied by the mass percentage of each αp = (p = 1, 2, 3)
mp − mp + 1 (3)
component based the proximate analysis data. As the experimental TGA
curve of the original oil sludge was obvious located below the calcu- where mp is the initial weight of a certain phase, m is the weight of
lated curve (Fig. 3 (b) and Fig. S.1), the difference of the two curves sample at time t in a certain phase, and mp+1 is the final weight of each
indicated that the in-situ mineral solids accelerated pyrolysis process of phase.
oil. Then, substituting k(T) using the Arrhenius equation, Eq. (2) can be
transformed into:
3.2. Activation energy calculation dα A E
= exp ⎛− ⎞ dT
f (α ) β ⎝ RT ⎠ (4)
The pyrolysis of OS as solid-state reaction could be explained as
follows, which is similar to biomass pyrolysis [21]: where β represents the heating rate which is a constant (β = dT/dt) in
this study.
Solid fuel → Volatile matter (Tars + Gases) + Chars. (1)
And the integrated form of Eq. (4) can be shown as follow:
Although the pyrolysis of OS is a series of complex reactions, the α
dα A T E
mathematical models can be used to describe the thermal analysis data. G (α ) = ∫ = ∫ exp ⎛− ⎞ dT
0 f (α ) β T0 ⎝ RT ⎠ (5)
The overall pyrolysis of OS was the superposition of different phases,
and each phase followed an independent model [8]. The reaction where G(α) represents the integral form of the reaction function f(α)
conversion of the sample in pyrolysis can be described as: [23].
The iso-conversional methods have been highly appreciated for the
dα
= k (T ) f (α ) accurate estimation of the activation energy (E) because the assumption
dt (2)
of the reaction mechanism is not required. In the present work,
where dα/dt is the degradation rate; α is the degradation degree; t is the Kissinger-Akahira-Sunose (KAS) method and Flynn-Wall-Ozawa (FWO)
reaction time; T is the absolute temperature; f(α) is the differential form method were chosen to calculate the activation energy of pyrolysis for
of reaction model; k(T) is the pyrolysis rate constant dependent on their higher accuracy [24–26].
temperature and can be described with the Arrhenius equation: k The expression of FWO method is:
(T) = A exp(−E/RT), (E, A, and R are the activation energy, the pre-
AE E
exponential factor, and the universal gas constant, respectively). ln β = ln − 5.331 − 1.0516
RG (α ) RT (6)
The degradation degree of each phase can be expressed as follows

341
Y. Qu, et al. Chemical Engineering Journal 374 (2019) 338–346

Table 3 from PH 1 to PH 3, because the thermal decomposition of samples was

Activation energy (E) obtained from FWO and KAS method for the pyrolysis of harder at higher temperatures [27]. In PH 2 and PH 3, the differences of
OS. activation energy between OS and EO became smaller. The comparison
α PH 1 PH 2 PH 3 of the E value between OS and EO indicated that the presence of solids
would promote thermal decomposition of OS by reducing the reaction
FWO KAS FWO KAS FWO KAS energy barrier. The mechanism of solid effect on OS would be further
E kJ E kJ E kJ E kJ E kJ mol−1 E kJ mol−1
inferred by the kinetic model estimated from the master-plots method.
mol−1 mol−1 mol−1 mol−1
The activation energy values in PH1 and PH2 of oil sludge were
0.2 25.47 20.82 37.97 30.59 98.94 92.15 relatively lower than the reported values in open literatures [28,29].
0.25 25.49 20.73 40.54 33.12 99.60 92.75 This was mainly ascribed to the difference between oil sludge samples
0.3 25.41 20.55 43.07 35.63 100.04 93.13
in this study and other studies. This study used raw oil sludge which
0.35 25.95 21.01 45.35 37.87 102.72 95.89
0.4 26.91 21.91 47.80 40.30 103.02 96.19 was closer to practical application, but other studies generally used
0.45 28.01 22.95 50.75 43.28 105.07 98.21 dried oil sludge. The presence of water and high volatile (Table 1) in the
0.5 28.94 23.82 52.95 45.45 107.10 104.78 raw oil sludge was easier to be decomposed and led to relatively lower
0.55 29.19 23.97 55.77 48.30 109.20 102.42 activation energy.
0.6 29.37 24.03 57.88 50.38 110.53 103.75
0.65 29.70 24.26 60.71 53.24 111.00 104.18
0.7 29.87 24.29 63.74 56.31 113.87 107.13 3.3. Kinetic model determination
0.75 31.12 25.46 67.34 59.98 115.47 108.74
0.8 32.04 26.26 69.95 62.59 118.37 111.71 To describe the kinetic model of various phases for OS and EO, the
average 28.27 23.08 53.37 45.93 107.30 100.84
integral master-plots methods were employed. As the decomposition
reactions of samples are very slow at initial temperature T0, and the
Meanwhile, the KAS method can be described as: conversion α is almost zero until T0, the integral of Eq. (5) can be de-
scribed by the master-plots method [17]:
β AR E α
d(α ) A T E A T E AE
ln( ) = ln[ ]− G (α ) = ∫ = ∫ exp ⎛− ⎞ dT ≈ ∫ exp ⎛− ⎞ dT = P (u)
T2 EG (α ) RT (7) f (α ) β T0 ⎝ RT ⎠ β 0 ⎝ RT ⎠ βR
0

For a certain α, the values of E are related to the slopes of plots by (8)
drawing ln β versus 1/T for FWO and ln[β/T2] versus 1/T for KAS, where P(u) is the temperature integral and u = E/RT.
respectively. In this study, the range of α was 0.20–0.80 with a step of Empirical equations such as Tang–Liu–Zhang–Wang–Wang approx-
0.05. imation can be used to approximately solve P(u), due to the divergence
Based on Eq. (6) and Eq. (7), the E values for each phase of OS and of the numerical solution is lower than 0.1% [30].
EO pyrolysis were estimated by the TGA experimental data with various
exp (−u)
heating rates of 20, 60, and 100 K min−1. Figs. S.2 and S.3 exhibited the P (u) =
u × (1.00198882u + 1.87391198) (9)
plots of FWO method and KAS method for different reaction degrees
that should be fitted as straight lines. As the slope of regression lines The kinetic triplets are assumed constant in a single-step reaction.
was related to the E values, the E values of pyrolysis at different con- As the value of E by iso-conversional methods had been obtained in
version degrees were calculated. Section 3.2 (Tables 3 and 4), TG data can be simulated by a proper
Table 3 and Table 4 presented the E values for OS and EO, respec- reaction model. Adopting α = 0.5 as an anchor point, Eq. (8) would be
tively. Although the E value calculated by the KAS method was a little described as follow:
lower than that obtained by the FWO method, the differences are
AE
negligible. Hence, the mean values of activation energy estimated by G (0.5) = ( ) P (u 0.5)
βR (10)
FWO and KAS can be adopted by the master-plots method.
Moreover, the activation energy for EO was much higher compared where G(0.5) represents the integral of reaction model at α = 0.5;
with the activation energy for OS. In PH 1, the average E value of EO u0.5 = E/RT0.5; T0.5 is the temperature at α = 0.5.
was 87.13 kJ·mol−1, which was more than three times higher than the Dividing Eq. (8) by Eq. (10), the equation of integral master-plots
values of OS (28.27 kJ· mol−1). And the activation energies increased can be shown as:

Table 4
Activation energy (E) obtained from FWO and KAS method for the pyrolysis of EO.
α PH 1 PH 2 PH 3

FWO KAS FWO KAS FWO KAS

E kJ mol−1 E kJ mol−1 E kJ mol−1 E kJ mol−1 E kJ mol−1 E kJ mol−1

0.2 120.38 120.42 85.19 80.82 163.74 160.43

0.25 129.78 130.20 88.00 83.55 173.52 170.59
0.3 118.79 118.53 93.32 88.94 178.65 175.86
0.35 101.17 99.89 95.64 91.18 186.57 184.10
0.4 92.45 90.62 96.86 92.28 192.44 190.17
0.45 80.64 78.08 101.42 96.92 196.75 194.61
0.5 75.71 72.78 103.45 98.90 195.70 193.43
0.55 75.22 72.16 104.16 99.49 201.53 199.49
0.6 67.16 63.60 109.49 104.95 199.26 197.03
0.65 67.57 63.94 113.83 109.29 202.88 200.77
0.7 67.20 63.45 111.76 107.06 205.69 203.67
0.75 68.35 64.58 113.53 108.79 211.48 207.45
0.8 68.30 64.44 120.95 116.49 207.54 205.48
average 87.13 84.82 102.89 98.36 193.52 191.00

342
Y. Qu, et al.
Fig. 4. (a–c) P(u)/P(u0.5) versus α for various pyrolysis phases [(a) PH 1, (b) PH 2, (c) PH 3] of OS at various heating rates; (d–f) G(α)/G(0.5) versus α for various
reaction models and P(u)/P(u0.5) versus α for various phases [(d) PH 1, (e) PH 2, (f) PH 3] of OS at 60 K min−1.
G (α ) P (u)
=
G (0.5) P (u 0.5)
On the left hand of Eq. (11), the curves of G(α)/G(0.5) versus α were
obtained as the theoretical master plots with differential form G(α)
functions (Table S.2). Moreover, the plots of P(u)/P(u0.5) versus α was
calculated from TG data as the experimental master plots on the other
side of Eq. (11). Compared both sides of Eq. (11), the best fit kinetic
model of different phase in pyrolysis of OS can be estimated by the
integral master-plots method. If using an improper form of G(α), the
experimental master plots would diverge greatly from the theoretical
master [31,32].
As indicated by Eq. (9), the integral temperature P(u) was a function
of α and can be estimated with the previously obtained activation en-
ergy E. Fig. 4(a–c) showed the curves of P(u)/P(u0.5) versus α obtained
Chemical Engineering Journal 374 (2019) 338–346
from TG data of OS at 20, 60, and 100 K min−1. The shape of the P(u)/P
(11) (u0.5) plots at different heating rate was similar, which illustrated that
the pyrolysis processes of OS follows the single kinetic model in each
phase.
The plots of G(α)/G(0.5) were plotted by various kinetic functions as
the theoretical master-plots (Table S.2). The comparisons between the
theoretical master-plots and the experimental master-plots P(u)/P(u0.5)
of different phase in OS pyrolysis at 60 K min−1 were exhibited in
Fig. 4(d–f) as typical examples. The P(u)/P(u0.5) plots of the initial
phase (PH 1) were situated at the theoretical plots F1, and the kinetic
model of F1 was G(α) = −ln(1 − α). This kinetic model indicated that
the reaction mechanism of PH1 was first-order reaction. As for PH 2,
the kinetic model of Rn, G(α) = 1−(1 − α)3/4 fitted best. For the last
phase (PH 3), the kinetic model changed to random nucleation and
nuclei growth mechanism. The kinetic model was G(α) = [−ln
343
Y. Qu, et al. Chemical Engineering Journal 374 (2019) 338–346

Fig. 5. (a–c) P(u)/P(u0.5) versus α for various pyrolysis phases [(a) PH 1, (b) PH 2, (c) PH 3] of EO at various heating rates; (d–f) G(α)/G(0.5) versus α for various
reaction models and P(u)/P(u0.5) versus α for various phases [(d) PH 1, (e) PH 2, (f) PH 3] of EO at 60 K min−1.

(1 − α)]3/4.
The same methods were used to estimate the reaction mechanisms
of EO in different phases and were shown in Fig. 5. As shown in
Fig. 5(a-c), the P(u)/P(u0.5) plots of EO were more consistent than that
of OS, which confirmed that the pyrolysis of EO in each phase was a
single-step reaction. The reason of this distinction between OS and EO
may be the absence of solids make EO more homogeneous.
Fig. 5(d-f) presented the comparison of experimental master-plots P
(u)/P(u0.5) of different phase for EO pyrolysis at 60 K min−1 as an ex-
ample. The kinetic model in PH 1 was D1 suggesting that PH1 of EO is a
diffusion reaction. The expression of kinetic model in PH 1 was
G(α) = α2. Whereas, the kinetic model of PH 2 changed to F3, and the
expression was G(α) = [(1 − α)−2 − 1]/2. For the third phase (PH 3),
the kinetic model was same as PH 1, the kinetic models fitted G(α) = α2
best.
3.4. Pre-exponential factor and kinetic exponent calculation
After the kinetic models had been obtained in Section 3.4, the ki-
netic exponent (n) and the pre-exponential factor (A) can be determined
by Eq. (8) using kinetic models. The calculative process of OS at
20 K min−1 was exhibited as an example. Based on Eq. (8), the most
feasible values of n for PH 1 of OS degradation were calculated by the
linear least square fitting method with plotting G(α) = (1 − α)1−n − 1
versus 10 × EP(u)/βR ranged from 1.0 to 2.0 using a step size of 0.1. In
the same way, the kinetic exponent n of PH 2 and PH 3 can be estimated
by plotting G(α) = [−ln(1 − α)]n against 10 × EP(u)/βR based on Eq.
(8). As shown in Fig. S.4, the most probable values of n for various

344
Y. Qu, et al. Chemical Engineering Journal 374 (2019) 338–346

Table 5
Kinetic triplets (E, A, and G(α)) for various phases of OS and EO pyrolysis at 20, 60, and 100 K min−1 by master-plots method.
Phase β K min−1 E kJ mol−1 A s−1 G(α) Mechanisms R2

OS EO OS EO OS EO OS EO OS EO

PH1 20 28.27 67.72 2.527 × 101

7.386 × 108
−ln(1 − α) α 2
F1 D1 0.9997 0.9987
60 3.076 × 101 3.362 × 107 0.9980 0.9990
100 4.398 × 101 3.313 × 106 0.99818 0.9921

PH2 20 53.37 102.89 5.610 × 103 1.166 × 107 1−(1 − α)3/4 [(1 − α)−2 − 1]/2 R1.5 F3 0.9967 0.9973
60 8.016 × 103 6.845 × 106 0.9970 0.9985
100 9.133 × 103 5.547 × 106 0.9982 0.9967

PH3 20 107.30 193.52 7.460 × 104 1.120 × 1010 [−ln(1 − α)]3/4 α2 A1.75 D1 0.9975 0.9989
60 7.352 × 104 1.240 × 1010 0.9990 0.9988
100 8.799 × 104 1.875 × 1010 0.9977 0.9965

phases of both OS and EO decomposition at 20 K min−1 were de-
termined with the highest correlation coefficient (R2) and the closest
zero intercept. The kinetic triplets (E, A, and G(α)) of different phases in
pyrolysis of OS and EO at various heating rates of 20, 60, and
100 K min−1 were shown in Table 5.
3.5. Effects of solids in OS pyrolysis
As shown in Table 5, the kinetic models of OS and EO were ob-
viously different confirming that solids had significant effects on the
pyrolysis processes of OS. Although other researchers had found that
the solids have positive effects on pyrolysis processes of OS [11], the
mechanisms of solids enhancing the pyrolysis were uncertain. There-
fore, this section revealed the mechanisms of solids in OS pyrolysis by
analyzing the changes of kinetic models between OS and EO. In fact, the
pyrolysis process of OS and EO was an integrated process composed by
a series of thermal degradation reactions, and the kinetic models re-
presented the limiting steps of thermal degradation reactions in dif-
ferent phase. Most frequently used limiting steps included order-reac-
tion (Fn), diffusion (Dn), limiting surface reaction (Rn), random
nucleation and nuclei growth (An), etc.
The reaction model of OS pyrolysis was exhibited in Fig. 6(a) to
interpret the pyrolysis processes. At relatively lower temperature, light
Fig. 6. (a) Primitive reaction model; (b) detailed reaction paths during pyr-
olysis process of oil sludge.
organics and water devolatilized from the oil sludge. With the incre-
ment of reaction temperature, organics in oil sludge began to be ther-
mally cracked into smaller molecular, and most pyrolytic oil was gen-
erated in PH2. In PH3, heavy carbonaceous substance was cracked, and
char was generated simultaneously when the temperature was higher
than 673 K. Some secondary reactions between volatile and char also
occurred. And the reaction paths were also inferred as presented in
Fig. 6(b). Devolatilization was the main process in PH1 where the
physical process dominated. The chemical reactions that may take place
in PH 2 and PH 3 were illustrated, the main chemical reactions in OS
pyrolysis can be classified into: chain scission reaction, aromatization
reaction, aromatic ring releasing, carbonylation reaction, heterocyclic
reaction, cyclization reaction and polymerization reaction. Cyclization
reaction and polymerization reaction which happened in PH3 were the
main reactions of char formation.
The reactive components of OS and EO are similar, thus this reac-
tion model was also applicable to pyrolysis process of EO. The differ-
ence of kinetic models between OS and EO was an evident reflection of
the effect of solids in the oil sludge. The comparison of the kinetic
models was discussed to reveal the influence of solid in the following
parts.
In PH 1, the devolatilization of light organics took place. The kinetic
model of EO in PH 1 was one-way transport diffusion (D1), while the
kinetic model of OS changed to first-order reaction (F1) duo to the ef-
fects of solids. The presence of solids might increase the surface area
and gaps of diffusion, which enhanced the diffusion rate of devolatili-
zation. Once the diffusion rate was high enough, the limiting step was
no longer diffusion, and the limiting step became the order reaction.
PH 2 was attributed to thermal decomposition of medium molecular
weight substances. And in EO the diffusion rate was improved with the
temperature increment. Therefore, the diffusion reaction was no longer
the limiting step of EO in PH 2 corresponding with the kinetic model of
EO (F3). The thermal decomposition of OS in PH 2 was a heterogeneous
reaction, and heterogeneous reaction usually proceeded on the phase
interface. Thus, the surface area of solids became the limiting factor.
The kinetic model of OS in PH2 was surface limiting reaction between
both phases (R1.5), and the reaction dimensions were between one and
two. The SEM of in-situ solids before and after pyrolysis was shown in
Fig. S.5. No obvious pore structure was found in the residue both before
and after the pyrolysis. The poor pore structure meant the specific
surface area of solids was low. Therefore, although the in-situ solids can
provide surface for organic devolatilization the surface area was not
enough. The SEM of solids was consistent with the result of kinetic
analysis.
There were some differences in the last phase (PH 3). Not only
heavy carbonaceous substance was cracked, but also chars formation
occurred in this phase. The kinetic model of OS in PH3 was random
nucleation and nuclei growth (A1.75). The role of solids in PH3 of OS
pyrolysis may act as the nucleating agent in char generation. In con-
trast, the generation of solid chars in EO was a two-phase reaction, in

345
Y. Qu, et al.
which phase change of liquid carbonaceous substances was accom-
plished by the condensation reaction. Hence, mass transfer between
fluid and solid phases may be the limiting step of EO in PH 3. The
kinetic model of EO in PH3 was one-way transport reaction (D1). From
the above, solids were important in OS pyrolysis because they not only
provided surface for devolatilization and phase interface for thermal
decomposition, but also involved in the char generation as the nucle-
ating agent. As shown in Fig. S.5, the surface of residue after pyrolysis
was rougher than that before pyrolysis. The rough morphology fa-
cilitated the nucleation and growth by providing more nucleus, and
char sedimentation on the surface was supported by in-situ solids.
4. Conclusion
The pyrolysis processes of OS and EO showed three stages from the
DTG data, namely devolatilization of light hydrocarbons, thermal de-
composition of middle medium molecular weight substances and
cracking of the heavy carbonaceous substances. The solid fraction in OS
accelerated the pyrolysis of OS from the initial stage. According to the
activation energy calculation, solids had positive effects on OS pyrolysis
because the activation energy of OS was evidently lower than that of EO
in each stage. The presence of solids not only reduced the activation
energy of OS pyrolysis but also changed the reaction mechanisms.
Solids provided surface for the devolatilization of light organic volatiles
(PH 1) and also offered phase interface for thermal decomposition
(PH2). Furthermore, solids not merely provided the reaction conditions,
but also involved in the char formation as the nucleating agent (PH 3).
Acknowledgements
This work was supported by the Open Project Program of State Key
Laboratory of Petroleum Pollution Control (Grant No.: PPC2017001),
and the Fundamental Research Funds for the Central Universities
(Grant No.: DUT18RC(3)036).
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.cej.2019.05.183.
References
[1] G.J. Hu, J.B. Li, G.M. Zeng, Recent development in the treatment of oily sludge from
petroleum industry: a review, J. Hazard. Mater. 261 (2013) 470–490.
[2] C. Lin, et al., Freeze/thaw induced demulsification of water-in-oil emulsions with
loosely packed droplets, Sep. Purif. Technol. 56 (2007) 175–183.
[3] S.M. Al-Zahrani, M.D. Putra, Used lubricating oil regeneration by various solvent
extraction techniques, J. Ind. Eng. Chem. 19 (2013) 536–539.
[4] M. Fortuny, et al., Effect of salinity, temperature, water content, and pH on the
microwave demulsification of crude oil emulsions, Energy Fuels 21 (2007)
1358–1364.
[5] A. Cambiella, et al., Centrifugal separation efficiency in the treatment of waste
emulsified oils, Chem. Eng. Res. Des. 84 (2006) 69–76.
346
Chemical Engineering Journal 374 (2019) 338–346
[6] N. Xu, et al., Effects of ultrasound on oily sludge deoiling, J. Hazard. Mater. 171
(2009) 914–917.
[7] Y.L. Shishkin, Fractional and component analysis of crude oils by the method of
dynamic microdistillation – differential scanning calorimetry coupled with ther-
mogravimetry, Thermochim. Acta 441 (2006) 162–167.
[8] J.B. Chen, et al., TG/DSC-FTIR and Py-GC investigation on pyrolysis characteristics
of petrochemical wastewater sludge, Bioresour. Technol. 192 (2015) 1–10.
[9] O. Drici, S. Vossoughi, Study of the surface-area effect on crude-oil combustion by
thermal-analysis techniques, J. Pet. Technol. 37 (1985) 731–735.
[10] N. Sanaie, et al., Effect of minerals on coke precursor formation, FUEL 80 (2001)
1111–1119.
[11] S. Cheng, et al., Progress in thermal analysis studies on the pyrolysis process of oil
sludge, Thermochim. Acta 663 (2018).
[12] A.G. Barneto, et al., Thermogravimetric monitoring of oil refinery sludge, J. Anal.
Appl. Pyrolysis 105 (2014) 8–13.
[13] P.R. Herrington, G.F.A. Ball, J.E. Patrick, Determination of the relative oxidative
stability of petroleum bitumens by thermogravimetry, Thermochim. Acta 202
(1992) 201–213.
[14] A. Soria-Verdugo, et al., Pyrolysis of biofuels of the future: sewage sludge and
microalgae – thermogravimetric analysis and modelling of the pyrolysis under
different temperature conditions, Energy Convers. Manage. 138 (2017) 261–272.
[15] J.B. Chen, et al., Thermal characteristics and kinetics of refining and chemicals
wastewater, lignite and their blends during combustion, Energy Convers. Manage.
100 (2015) 201–211.
[16] M. Morin, et al., Kinetic study and modelling of char combustion in TGA in iso-
thermal conditions, Fuel 203 (2017) 522–536.
[17] D.I. Aslan, et al., Pyrolysis reaction models of waste tires: application of master-
plots method for energy conversion via devolatilization, Waste Manage. 68 (2017)
405–411.
[18] S. Ceylan, Kinetic analysis on the non-isothermal degradation of plum stone waste
by thermogravimetric analysis and integral master-plots method, Waste Manage.
Res. J. Int. Solid Wastes Public Cleansing Association ISWA 33 (2015) 345.
[19] Y. Duan, et al., Kinetic analysis on the non-isothermal dehydration by integral
master-plots method and TG FTIR study of zinc acetate dihydrate, J. Anal. Appl.
Pyrolysis 83 (2008) 1–6.
[20] D.Y. Chen, J.B. Zhou, Q.S. Zhang, Effects of heating rate on slow pyrolysis behavior,
kinetic parameters and products properties of moso bamboo, Bioresour. Technol.
169 (2014) 313–319.
[21] C. Gai, Y.P. Dong, T.H. Zhang, The kinetic analysis of the pyrolysis of agricultural
residue under non-isothermal conditions, Bioresour. Technol. 127 (2013) 298–305.
[22] K. Liu, X.Q. Ma, H.M. Xiao, Experimental and kinetic modeling of oxygen-enriched
air combustion of paper mill sludge, Waste Manage. 30 (2010) 1206–1211.
[23] C.X. Chen, et al., Oxy-fuel combustion characteristics and kinetics of microalgae
Chlorella vulgaris by thermogravimetric analysis, Bioresour. Technol. 144 (2013)
563–571.
[24] F. Ferdosian, et al., Thermal performance and thermal decomposition kinetics of
lignin-based epoxy resins, J. Anal. Appl. Pyrolysis 119 (2016) 124–132.
[25] Y.H. Ma, J. Wang, Y.S. Zhang, TG-FTIR study on pyrolysis of waste printing paper,
J. Therm. Anal. Calorim. 129 (2017) 1225–1232.
[26] J. Naktiyok, et al., Investigation of combustion kinetics of Umutbaca-lignite by
thermal analysis technique, J. Therm. Anal. Calorim. 129 (2017) 531–539.
[27] Z.Q. Xie, X.Q. Ma, The thermal behaviour of the co-combustion between paper
sludge and rice straw, Bioresour. Technol. 146 (2013) 611–618.
[28] R.M. Wu, et al., Fitting TGA data of oil sludge pyrolysis and oxidation by applying a
model free approximation of the Arrhenius parameters, J. Anal. Appl. Pyrolysis 76
(2006) 132–137.
[29] B.C. Lin, et al., Effects of potassium hydroxide on the catalytic pyrolysis of oily
sludge for high-quality oil product, Fuel 200 (2017) 124–133.
[30] T. Wanjun, et al., New temperature integral approximate formula for non-iso-
thermal kinetic analysis, J. Therm. Anal. Calorim. 74 (2003) 309–315.
[31] S. Vyazovkin, et al., ICTAC Kinetics Committee recommendations for performing
kinetic computations on thermal analysis data, Thermochim. Acta 520 (2011) 1–19.
[32] S.P. Zou, et al., Pyrolysis characteristics and kinetics of the marine microalgae
Dunaliella tertiolecta using thermogravimetric analyzer, Bioresour. Technol. 101
(2010) 359–365.