Fuel 231 (2018) 468–478

Contents lists available at ScienceDirect

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

Full Length Article

Modelling oxy-pyrolysis of sewage sludge in a rotary kiln reactor T

a b,⁎ b c c
Fabio Montagnaro , Claudio Tregambi , Piero Salatino , Osvalda Senneca , Roberto Solimene
a
Dipartimento di Scienze Chimiche, Università degli Studi di Napoli Federico II, Complesso Universitario di Monte Sant’Angelo, 80126 Napoli, Italy
b
Dipartimento di Ingegneria Chimica, dei Materiali e della Produzione Industriale, Università degli Studi di Napoli Federico II, Piazzale Vincenzo Tecchio 80, 80125
Napoli, Italy
c
Istituto di Ricerche sulla Combustione, Consiglio Nazionale delle Ricerche, Piazzale Vincenzo Tecchio 80, 80125 Napoli, Italy

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

Keywords: A mathematical model of a rotary kiln oxy-pyrolyser of sewage sludge is presented. The specific feature of the
Sewage sludge model is the consideration of the effect of axial staging of oxygen feeding to the reactor as one important key to
Rotary kiln the quality and productivity of the syngas, with a focus on the fate of tar and generation of soot. The gaseous
Pyrolysis oxidiser is fed to the reactor at multiple locations along its axis in a way that reproduces the paradigm of the
Gasification
Zwietering reactor. The fate of gaseous components and tar is followed using a simplified lumped-kinetic me-
Oxygen staging
Modelling
chanism that was purposely developed. The model implements submodels for heat transfer among the phases
and with the wall which embody radiative, convective and conductive terms. Homogeneous reactions in gas
phase are modelled with a kinetic submodel that considers the generation of primary tars from devolatilisation of
fuel and subsequent formation of secondary tars and soot by thermal cracking. The model validity has been
confirmed by critical comparison with experimental literature data referring to a similar case. The steady op-
eration of the oxy-pyrolyser is analysed in terms of fluxes of solid fuel and gaseous species, extent of fuel
devolatilisation, temperature profiles of solid and gas phases along the reactor. The performance of the reactor is
characterised in terms of process rate and chemical composition of the produced syngas, along with its heating
value and thermal power. The influence of the distributed feeding is assessed by comparison with a benchmark
case consisting of conventional non-distributed feeding.

1. Introduction
The thermal conversion of waste-derived fuels (e.g., municipal solid
waste, sewage sludge, agricultural residues, automotive shredder re-
sidues) is gaining a clear role in the general frame of the circular
economy (opposed to a linear approach where the resource is extracted,
used and then disposed of) as one pathway to close the recycle loop
when a material/chemical recycle is impossible or economically un-
feasible [1–4]. Even though the calorific value of these materials can be
lower than that of fossil fuels, and the amount of pollutants (e.g., based
on sulphur, chlorine, nitrogen) possibly larger, their thermal conversion
may mitigate disposal problems or respond to specific environmental
legislations. Since a significant fractional content of waste-derived fuels
has a biogenic nature, their use in energy generation can contribute to
the net reduction of CO2 emissions to the atmosphere. Depending on the
physico-chemical properties of the alternative fuel, the technical and
economic feasibility of thermochemical processing of waste-derived
fuels may be improved by co-processing with fossil fuels [5,6].
Sewage sludge derived from the treatment of urban wastewaters is
currently facing rapidly increasing production volumes and severe re-
strictions of the conventional disposal options. Three disposal methods
are used at present: recycling in agriculture, landfilling and combus-
tion. Agricultural utilisation is hindered by the presence of heavy me-
tals and organic micro-pollutants and, as a consequence, this kind of
recycling is rapidly losing acceptance [7–9]. Landfilling is not con-
sidered anymore as an environmentally sustainable option due to liquid
and gaseous emissions in soil, water and air, and is hardly compatible
with the perspective of a circular economy as previously highlighted.
Thermal conversion stems out as the most viable strategy to dispose
sewage sludge, entailing large reduction of sludge volume and thermal
destruction of the toxic organic constituents. In the frame of thermo-
chemical conversion, sewage sludge can be classified in three different
groups: dry sludge (> 80 wt% d.m.), semi-dried sludge (30–55 wt%
d.m.) and mechanically dewatered sludge (20–40 wt% d.m.). In general
dry sludge is not considered for single-fuel combustion, whereas semi-
dried sludge is preferred to the wet one as the process does not need
supplementary fuel.
Concerning thermochemical processing of waste-derived fuels,

⁎
Corresponding author.
E-mail address: claudio.tregambi@unina.it (C. Tregambi).

https://doi.org/10.1016/j.fuel.2018.05.094
Received 20 December 2017; Received in revised form 4 May 2018; Accepted 18 May 2018
Available online 30 May 2018
0016-2361/ © 2018 Elsevier Ltd. All rights reserved.
F. Montagnaro et al.
pyrolysis/gasification presents several advantages over the direct
waste-to-energy combustion path. This is mostly related to the gen-
eration of syngas and condensable species which can be easily trans-
ported, burned or even exploited in gas-to-liquid fuel or chemical
processes. These advantages may be somewhat counterbalanced by
additional complexity and capital cost. Rotary kiln (RK) and fluidised
bed (FB) reactors are frequently considered in the thermochemical
processing of waste-derived fuels. FB converters ensure intimate mixing
of solids and excellent thermal uniformity/stability which enable better
control of thermochemical paths and reduce the occurrence of un-
desired side reactions [10–12]. For these reasons, combustion [13–16]
and gasification [17–19] processes of alternative fuels in FB reactors are
widely investigated. The favourable features of FB converters may be
partly offset by the need for a better control of fuel size and composi-
tion, which may require fuel pre-treatment. Under this respect, the RK
converters are less sensitive to the fuel nature and indeed able to ac-
commodate large variations in fuel size, shape and composition, as well
as calorific value, with minimal or no pre-treatment [20]. As a con-
sequence, RK reactors are largely exploited in combustion [21,22],
pyrolysis [23–25] and gasification [26–29] processes of both conven-
tional and alternative fuels.
The present study addresses the development of a process for oxy-
pyrolysis of sewage sludge in a rotary kiln converter. The aim of the
process is the production of syngas from devolatilisation of a waste-
derived fuel, with oxygen playing the role of promoting autothermal
operation of the pyrolyser by controlled oxidation of volatile com-
pounds. The analysis of literature dealing with thermal conversion of
sewage sludge highlights a lack of understanding about the possible
effect related to the way of oxygen feeding on the properties of the
produced syngas. Therefore, after a section concerning the validation of
the model by critical comparison with experimental literature data, the
specific concern of the study will be the assessment of the effectiveness
of staged oxygen feeding, as opposed to localised feeding at the reactor
inlet, as a tool to selectively promote desired secondary reactions oc-
curring in gas phase, like partial oxidation of tars [30]. The converter
consists of a RK in which the oxidiser (O2-enriched air) is fed at mul-
tiple coordinates along the reactor axis, so as to obtain a reactant
contacting pattern resembling that of a Zwietering reactor [31]. The
reactor is modelled at the steady state using a 1.5D approach. Material
and energy balances are set up considering a semi-lumped kinetic me-
chanism that was purposely developed to represent the complex che-
mical pathways of solid fuel, gaseous compounds, different tar com-
ponents and soot. The model results are analysed with a focus on the
effect of axial staging of the oxidiser on the quality of the produced gas
and on the reactor performance.
2. The model
2.1. The reactor
The oxy-pyrolysis reactor is a rotary kiln equipped with a manifold
for distributed feeding of the oxidiser (O2-enriched air), as schemati-
cally represented in Fig. 1. The reactor has been modelled in a 1.5D
domain by considering the axial motion of solid phase (consisting of
fuel and ballast inert material) and gases, with material and energy
transfers between the two streams. The geometric/dimensional features
of the system reproduce those of a pilot-scale reactor developed by
Centro Sviluppo Materiali [32] in the frame of a joint research project
aimed at the production of syngas by thermal processing of solid re-
sidues. The kiln is 2.64 m long, 0.9 m ID, inclined by 1° with respect to
the horizontal plane and operated at atmospheric pressure. The oxidiser
(O2-enriched air) is fed through a manifold with 7 nozzles equally
distributed along a reactor length of 1.4 m (which represents nearly
50% of the total length). A bed of inert material is present as thermal
ballast to moderate temperature non-uniformities. The ballast (which is
assumed to be calcined bauxite) is loaded batchwise in the RK prior to
469
Fuel 231 (2018) 468–478
Fig. 1. Scheme of the oxy-pyrolyser reactor with uniform gas staging up to a
critical axial coordinate.
operation and is considered isothermal with respect to the solid fuel.
The solids discharge from the RK is designed so that the ballast is re-
tained in the reactor – due to its coarser dimension – and only the
pyrolysed sludge is continuously drained.
The axial motion of the solid fuel particles and of the gas follows the
plug flow pattern. Radial mixing within the solid phase is effective to
the point that segregation phenomena along the radial coordinate of the
reactor could be neglected. Each of the two (solid and gas) phases is
considered radially isothermal, though the difference in temperature
between these two phases has been taken into account. The axial profile
of the reactor wall temperature has been assigned. Intraparticle tem-
perature profiles have been neglected. Table 1 gives the meaning of the
symbols used throughout the article, together with related values.
2.2. The kinetic model
The kinetic mechanism of fuel decomposition and homogeneous
reactions of primary products is reported in Fig. 2.
Immediately after feeding to the converter, the solid fuel undergoes
rapid drying first, then devolatilisation. Any contact between the solid
fuel and oxygen is ruled out in this stage, so that devolatilisation de-
velops along a purely thermally activated – rather than oxygen-en-
hanced – decomposition path [33,34]. The gaseous species generated
from fuel thermal decomposition and those present in the gaseous
feeding eventually undergo homogeneous reactions according to the
kinetic mechanism that is hereby detailed.
The following gaseous species are considered: O2, N2, CO, CO2, H2,
H2O, CH4. Additionally, fuel devolatilisation yields tar. In the context of
the present model, a semi-lumped scheme for tar evolution has been
used as resulting from the reduction of a more comprehensive kinetic
mechanism. Accordingly, tar is assumed to be composed of two com-
ponents: a primary tar, T1, which is directly produced by fuel devola-
tilisation, and a secondary tar, T2, which is the product of secondary
gas-phase pyrolytic reactions of T1. The secondary tar T2 may undergo
further pyrolytic gas-phase dissociation to yield soot, which is assumed
to remain suspended in the gas and is treated as a pseudo-gaseous
compound. Primary and secondary tars differ as to their H/C ratio,
which expresses the degree of aromatisation. For the purpose of the
present computation, hexadecane (C16H34) was assumed as a surrogate
of T1 and naphthalene (C10H8) of T2.
The following homogeneous gas-phase reactions have been con-
sidered:
1
H2 + O2 → H2 O
2 (R1)
1
CO + O2 → CO2
2 (R2)
CH4 + 2O2 → CO2 + 2H2 O (R3)
CO + H2 O ⇄ CO2 + H2 (R4)
O2
T 1 ⎯→
⎯ CO, H2 (R5)
F. Montagnaro et al. Fuel 231 (2018) 468–478

Table 1
List of symbols and related values.
Symbol Meaning Units Value

−1 −1
Cp,B Specific heat of the ballast material [kJ kg K ] f(z) a
Cp,F Specific heat of solid fuel [kJ kg−1 K−1] 1.95
Cp,j Specific heat of the j -th species [kJ kmol−1 K−1] f(z) a
CpT,0j Specific heat of the j -th species calculated at T0 [kJ kmol−1 K−1] Assigned a
Cp,LW Specific heat of liquid water [kJ kg−1 K−1] f(z) a
CpT,SVW Specific heat of water vapor calculated at TS [kJ kmol−1 K−1] f(z) a
D Internal diameter of the reactor [m] 0.9
D Molecular diffusivity [m2 s−1] Assigned a
dB Diameter of the ballast particles [m] 0.01
E Activation energy of devolatilisation [kJ kmol−1] 63,438
EM Activation energy of dehumidification [kJ kmol−1] 28,320
E Rk Activation energy for reactions (R4)–(R8) [kJ kmol−1] Assigned b
Fj Flux of the j -th species [kmol m−2 s−1] c f(z)
Fj0 Flux of the j -th species at the reactor inlet [kmol m−2 s−1] c Assigned d
fchar Mass fraction of char in the raw sludge (dry basis) [–] 0.353
f gG / S Gas/solid geometric view factor [–] 0.27

f gW / G Wall/gas geometric view factor [–] 0.67

fM Moisture fraction for solid fuel (additive to a 100% dry basis) [kg kg−1] 0.4
f oG / S Total emission factor for gas/solid radiation [–] 0.30

foW / G Overall view factor for wall/gas radiation [–] 0.30

foW / S Total emission factor for wall/solid radiation [–] 1

W /S
hcond Conductive coefficient of heat transfer (wall/solid) [kJ m−2 s−1 K−1] f(z)
G/S
hconv Convective coefficient of heat transfer (gas/solid) [kJ m−2 s−1 K−1] 9.1 × 10−4
−2 −1 −1
W /G
hconv Convective coefficient of heat transfer (wall/gas) [kJ m s K ] 8.3 × 10−4
I Molar feeding rate of gas per unit length and unit cross-sectional area of the reactor [kmol m−3 s−1] 6.48 × 10−4 e
j Generic gaseous species and soot [–] 1–10
KT ,B Thermal conductivity of the ballast material [kJ m−1 s−1 K−1] Assigned a
KT ,G Thermal conductivity of gas [kJ m−1 s−1 K−1] Assigned a
k Generic homogeneous (gas phase) reaction [–] 1–8
k0 Pre-exponential factor for devolatilisation [s−1] 79.1
k 0,M Pre-exponential factor for dehumidification [s−1] 350
k 0,Rk Pre-exponential factor for reactions (R4)–(R8) [variable] Assigned f
k KIN Pure kinetic term for reactions (R1)–(R3) [m3 kmol−1 s−1] 2 × 106
k Rk Kinetic constant for reactions (R1)–(R8) [variable] f(z)
k REB Eddy break-up kinetic term for reactions (R1)–(R3) [s−1] Assigned g
k
L Reactor length [m] 2.64
M Overall gas mass in the reactor [kg] Assigned h
n Reaction order for devolatilisation [–] 1.31
nM Reaction order for dehumidification [–] 1
P Operating pressure [kPa] 101.3
Q Overall volumetric flow rate of gas [m3 s−1] f(z)
Q0 Volumetric flow rate of gas at the reactor inlet [m3 s−1] 2.35 × 10−3
Qg Overall volumetric flow rate of feeding gas [m3 s−1] 1.65 × 10−2
R Gas constant [kJ kmol−1 K−1] 8.314
rd,j Formation rate of the j -th species by devolatilisation [kmol m−3 s−1] f(z)
rhk Rate of the k -th homogeneous reaction [kmol m−3 s−1] f(z)
rhk,j Formation rate of the j -th species due to k -th homogeneous reaction [kmol m−3 s−1] f(z)
rM Rate of moisture release [kmol m−3 s−1] f(z)
r Rk Disappearance rate of the first reactant in reactions (R1)–(R8) [kmol m−3 s−1] f(z)
T Temperature of gas phase [K] f(z)
T0 Inlet temperature of gas [K] 298
TS Temperature of solid phase [K] f(z)
TS0 Temperature of solid phase at the reactor inlet [K] 298
TW Wall temperature [K] f(z)
V Reactor volume [m3] 1.68
v Velocity of solid fuel [m s−1] 1 × 10−3
vN Velocity at the nozzle [m s−1] 66.3
W Flux of solid fuel (dry basis) [kg m−2 s−1] f(z)
W0 Flux of solid fuel (dry basis) at the reactor inlet [kg m−2 s−1] 4.37 × 10−2 i
xj Yield of the j -th species by devolatilisation [kg kg−1] Assigned j
yj Molar fraction of the j -th species in the external gas [–] Assigned k
z Axial coordinate of the reactor [m] Variable
z∗ Value of z up to which external gas is added [m] 1.4
α Coefficient of axial dispersion of solid phase [m2 s−1] 9.78 × 10−7
Δ Solid height [m] 0.22
ΔHd Enthalpy change of the devolatilisation reaction [kJ kg−1] 300
(continued on next page)

470
F. Montagnaro et al. Fuel 231 (2018) 468–478

Table 1 (continued)

Symbol Meaning Units Value

−1
ΔHk Enthalpy change of the k -th homogeneous reaction [kJ kmol ] f(z) a
ΔP Pressure drop across feed nozzles [kg m−1 s−2] Assigned h
ε Fractional volume of the solids bed in the reactor [m3 m−3] 0.19
ε′ Power dissipated per unit mass of gas [m2 s−3] 80
λ Kolmogorov minimum scale of turbulence [m] 400 × 10−6
λ H2 O Latent heat of vaporisation [kJ kmol−1] f(z) a
ν Kinematic viscosity of fluid [m2 s−1] Assigned a
ξ Devolatilisation degree [–] f(z)
ξM Dehumidification degree [–] f(z)
ρB Density of the ballast material [kg m−3] 2550
ρg Gas density [kg m−3] Assigned a
ρN Gas density at feeding (nozzle) conditions [kg m−3] Assigned a
σ Stefan–Boltzmann constant [kJ m−2 s−1 K−4] 5.67 × 10−11
ϕ Angle in Fig. 3 [rad] 2.07

a
After Ref. [38].
b
83,700 (R4 forward), 121,800 (R4 backward), 249,400 (R5), 291,000 (R6) and (R8) and 166,300 (R7).
c
But [kg m−2 s−1] for j = soot.
d
6.58 × 10−5 (O2) and 8.55 × 10−5 (N2); for other species it is 0.
e
Up to z = z*.
f
2.75 × 109 m3 kmol−1 s−1 (R4 forward), 2.09 × 1011 m3 kmol−1 s−1 (R4 backward), 2 × 1014 s−1 (R5), 1011 s−1 (R6), 1012 m3 kmol−1 s−1 (R7) and
3 × 109 s−1 (R8).
g
4000 (R1) and 1000 (R2) and (R3).
h
See Section 2.2.
i
Equivalent to 100 kg h−1.
j
Dry mass basis: 0.048 (N2), 0.122 (CO), 0.221 (CO2), 0.017 (H2), 0.033 (CH4) and 0.206 (T1); for other species it is 0.
k
0.435 (O2) and 0.565 (N2); for other species it is 0.

T 1 → T 2, H2 (R6) (R4) is the equilibrium water-gas shift reaction. (R5) and (R7) represent
partial oxidation – in oxygen starving atmospheres – of primary and
O2
T 2 ⎯→
⎯ CO, H2 (R7) secondary tars, respectively. (R6) represents the degradation (cracking)
of primary tar into secondary tar, yielding hydrogen. Finally, (R8) is the
T 2 → C , H2 (R8) pyrolytic decomposition of secondary tar T2 into soot (C), with further
Reactions (R1–R3) are combustion of H2, CO and CH4, respectively. hydrogen release.

Fig. 2. The semi-lumped reaction network.

471
F. Montagnaro et al. Fuel 231 (2018) 468–478

Fuel drying is modelled assuming the following rate (MW stands for EF
kRF4 (z ) = k 0,FR4 exp ⎡− R4 ⎤
molecular weight): ⎢ RT (z ) ⎥ (8-a)
⎣ ⎦
E
W0 fM k 0,M exp ⎡− M ⎤ [1−ξM (z )]nM EB
rM (z ) = ⎣ RTS (z ) ⎦ kRB4 (z ) = k 0,BR4 exp ⎡− R4 ⎤
v MWH2 O (1) ⎢ RT (z ) ⎥ (8-b)
⎣ ⎦
−1 −1
with activation energy EM , pre-exponential factor k 0,M and reactor order with pre-exponential factors: k0,R4F = 9
2.75 × 10 m kmol s 3
and
nM set at 28.32 MJ kmol−1, 350 s−1 and 1, respectively. The moisture k0,R4B = 2.09 × 1011 m kmol s , 3 −1 −1
and activation energies:
fraction of the semi-dried sludge is fM = 0.4 (additive to a 100% dry ER4F = 83.7 MJ kmol−1 and ER4B = 121.8 MJ kmol−1.
basis). Reactions (R5–R8) represent the tar-related chemical network. The
The rates of (R1–R8) are expressed as rRk , i.e. the rate of dis- following expressions for the reaction rates have been taken (the ex-
appearance of the first reactant in each reaction. pression for rR5 (z ) is valid in presence of oxygen, see (R5)):
The rates of the combustion reactions (R1–R3) have been expressed πD2
as the product between a first-order kinetic constant k Rk and the con- FT 1 (z ) 4
rR5 (z ) = kR5 (z )
centration of the fuel gaseous species: Q (z ) (9-a)
πD2 πD2
FH2 (z ) 4 FT 1 (z )
rR1 (z ) = kR1 (z ) rR6 (z ) = kR6 (z ) 4
Q (z ) (2-a) Q (z ) (9-b)
πD2 πD2 πD2
FCO (z ) 4 FT 2 (z ) FO2 (z )
rR2 (z ) = kR2 (z ) rR7 (z ) = kR7 (z ) 4 4
Q (z ) (2-b) Q (z ) Q (z ) (9-c)
πD2 πD2
FCH4 (z ) 4 FT 2 (z )
rR3 (z ) = kR3 (z ) 4
rR8 (z ) = kR8 (z )
Q (z ) (2-c) Q (z ) (9-d)
The kinetic constants, in turn, were expressed as the combination of with kinetic constants:
an “eddy break-up” (EB) and of a purely kinetic (KIN) term, the last one
multiplying the O2 concentration: E Rk ⎤
k Rk (z ) = k 0,Rk exp ⎡− k = 5,6,7,8
⎢
⎣ RT (z ) ⎥
⎦ (10)
1 1 1
= EB + k = 1,2,3 14 −1
k Rk (z ) k Rk πD2 where the pre-exponential factors are: k0,R5 = 2 × 10 s ,
FO (z )
KIN 2
k0,R6 =1011 s−1, k0,R7 = 1012 m3 kmol−1 s−1 and k0,R8 = 3 × 109 s−1,
4
k Q (z ) (3)
and the activation energies are: ER5 = 249.4 MJ kmol−1,
A value of kKIN = 2 × 106 m3 kmol−1 s−1 has been assumed. The EB ER6 = 291 MJ kmol = ER8 and ER7 = 166.3 MJ kmol−1.
−1

approach is based on the assumption that the reaction process is Once the reaction rates have been set, the overall rate of genera-
regulated by the micromixing rate: “what is mixed is burnt”, namely the tion/destruction of all the gaseous species and of soot (C) is expressed
combustion occurs instantaneously when gaseous fuel and oxidant are as:
in contact. It is possible to write [31]:
k
D ⎛ ∑ [rh,j = O2 (z )] ⎞
k EB = ⎜ k ⎟
λ2 (4)
⎜ ∑ [rhk,j = N2 (z )] ⎟
where D is the relevant molecular diffusivity and λ is the Kolmogorov ⎜ k ⎟
minimum scale of turbulence, which was estimated according to the ⎜ ∑ [r k (z )] ⎟
h,j = CO 1 1
classical theory for homogeneous/isotropic turbulent mixing: ⎜ k ⎟ ⎛− 2 − 2 − 2 0 − 8 0 − 5 0 ⎞
⎜ k
∑ [rh,j=CO2 (z )]⎟ ⎜ ⎟ ⎜ 0 0 0 0 0 0 0 0 ⎟ ⎛ rR1 (z ) ⎞
3 ⎜ 0 − 1 0 − 1 16 0 10 0 ⎟ ⎜ rR2 (z ) ⎟
λ= 4 ⎛ν ⎞ ⎜
k
⎟ ⎜ 0
⎜ ⎟
1 1 1 0 0 0 0 ⎟ ⎜ rR3 (z ) ⎟
⎝ ε′ ⎠ (5) k
⎜ ∑ [rh,j = H2 (z )] ⎟ ⎜ − 1 0 0 1 17 53
4 4 ⎟ ⎜ rR4 (z ) ⎟
where ν is the kinematic viscosity of the fluid and ε′ is the power dis- ⎜ k ⎟=⎜ 5 ⎟×
k
⎜∑ [rh,j = H2 O (z )]⎟ ⎜ 1 0 2 −1 0 0 0 0 ⎟ ⎜ rR5 (z ) ⎟
sipated at the feeding nozzles per unit mass of gas. It is:
⎜ k ⎟ ⎜ 0 0 −1 0 0 0 0 0 ⎟ ⎜ rR6 (z ) ⎟
0 0 0 0 − 1 − 1 0 0 ⎜ ⎟
Qg ΔP ⎜∑ [rhk,j = CH4 (z )]⎟ ⎜ 8
⎟ ⎜ rR7 (z ) ⎟
ε′ =
M (6) ⎜ k ⎟ ⎜ 0 0 0 0 0
5
− 1 − 1 ⎟ ⎝ rR8 (z ) ⎟⎠
⎜
⎜ ∑ [rhk,j = T 1 (z )] ⎟ ⎜ 0 0 0 0 0 0 0 10 ⎟
where Qg is the overall volumetric flow rate of feeding gas, ΔP is the ⎜ k ⎟ ⎝ ⎠
pressure drop across the feeding nozzles (ΔP = ρNvN2/2, with ρN the ⎜ k
∑ [rh,j=T 2 (z )] ⎟ ⎟
density of the feed gas and vN = 66.3 m s−1 the gas velocity at the ⎜ k
nozzle), and M the overall gas mass in the reactor (M = Vρg, being V ⎜ k ⎟
⎜ ∑ [rh,j = C (z )] ⎟
the reactor volume and ρg the gas density). Estimations led to the fol- ⎝ k ⎠
lowing values: ɛ′ = 80 m2 s−3 and λ = 400 μm. Accordingly, (11)
kR1EB = 4000 s−1 and kR2EB = kR3EB = 1000 s−1.
The net rate of the water-gas shift reaction (R4) is:
2.3. Mass balance equations: solid phase
πD2 πD2 πD2 πD2
FCO (z ) 4
FH2 O (z ) 4
FCO2 (z ) 4
FH2 (z ) 4
rR4 (z ) = kRF4 (z ) −kRB4 (z ) The rate of change of the dehumidification degree ξM is expressed
Q (z ) Q (z ) Q (z ) Q (z )
through:
(7)
dξM (z ) E
where the kinetic constants of the forward (F) and backward (B) reac- v = k 0,M exp ⎡− M ⎤ [1−ξM (z )]nM
tions are, respectively: dz ⎢
⎣ RTS (z ) ⎥
⎦ (12)

472
F. Montagnaro et al. Fuel 231 (2018) 468–478

with boundary condition: ξM (z = 0) = 0 .

The axial profiles of solid flux and of devolatilisation degree are
∑
j
{ d [Fj (z ) Cp,j (z ) T (z )]
dZ } = (1−f W0 E
char ) ν k 0 exp ⎡− RTS (z ) ⎤ [1−ξ
⎣ ⎦
(z )]n Cp,F TS (z )

expressed through, respectively:

+ rM (z ) CpTS,VM (Z ) TS (z )+
dW (z ) W E ⎤ + ∑ {rhk (z )[−ΔHk (z )]} + I (z ) ∑ [CpT0,j yj (z ) T0]+
= −(1−fchar ) 0 k 0exp ⎡− [1−ξ (z )]n
dz v ⎢
⎣ RTS (z ) ⎥
⎦ (13) k j
4 W /G W /G
+ {f
D g
fo σ [TW4 (z )−T 4 (z )] + f gW / G hconv
W /G
[TW (z )−T (z )]
dξ (z ) E ⎤
v = k 0exp ⎡− [1−ξ (z )]n + f gG / S f oG / S σ [TS4 (z )−T 4 (z )]+
dz ⎢
⎣ RTS (z ) ⎥
⎦ (14)
+ f gG / S hconv
G/S
[TS (z )−T (z )]}
The boundary conditions, set at z = 0 (reactor entrance), are:
W (z = 0) = W0 and ξ (z = 0) = 0 for Eqs. (13) and (14), respectively. (19)
with boundary condition: T (z = 0) = T0 .
2.4. Mass balance equations: gas phase
2.7. Model implementation
The following equations describe the profiles (along the reactor z-
axis) of the flux of the j-th chemical species (involved in k different The core of the model is a system of nonlinear ordinary differential
chemical reactions) and of the overall volumetric flow rate of gas equations. They have been solved in MATLAB® environment using the
(under the hypothesis that the gases are ideal): ode23s function, based on a modified Rosenbrock formula of order 2. A
dFj (z ) relative error tolerance of 10−6 has been used for the convergence of
∀ j ≠ H2 O: = rd,j (z ) + ∑ [rhk,j (z )] + I (z ) yj the solution.
dz k (15-a)

dFH2 O (z ) 3. Evaluation of model parameters

= rd,H2 O (z ) + ∑ [rhk,H2 O (z )] + rM (z ) + I (z ) yH2 O
dz k (15-b) The solid fuel considered in this study is a semi-dried sewage sludge,
already investigated in previous communications [10,14] whence the
dQ (z )
=
πD 2 R
∑ ⎧ d [Fj (z ) T (z )] ⎫ relevant properties were obtained.
dz 4 P j
⎨
⎩ dz ⎬
⎭ (16) The devolatilisation yields, obtained by elaborating data from
Urciuolo et al. [14], are 0.048 (N2), 0.122 (CO), 0.221 (CO2), 0.017
The boundary conditions, set at z = 0, are: Fj (z = 0) = Fj0 ∀ j and
(H2), 0.033 (CH4) and 0.206 (tar) on a dry mass basis. The total yield in
Q (z = 0) = Q0 for Eqs. (15-a)/(15-b) and (16), respectively. The for-
volatile compounds is 0.647, corresponding to fchar = 0.353.
mation rate of the j-th gaseous species due to devolatilisation is ex-
The mass feed rate of sludge (dry basis) is set at 100 kg h−1. Accordingly
pressed as:
the sludge flux at the reactor inlet is W0 = 4.37 × 10−2 kg m−2 s−1. The
W0 E ⎤ xj sludge velocity inside the oxy-pyrolyser has been estimated on the basis of
∀ j: rd,j (z ) = k 0exp ⎡− [1−ξ (z )]n the operational experience of the pilot-scale reactor [32]: it is
v ⎢ RTS (z ) ⎦
⎣ ⎥ MWj (17)
v = 1 × 10−3 m s−1. For devolatilisation, activation energy E , pre-ex-
ponential factor k 0 and reaction order n value 63.438 MJ kmol−1, 79.1 s−1
2.5. Energy balance equation: solid phase and 1.31, respectively [35].
The gasifying agent consists of O2-enriched air: yO2 = 0.435 and
The following equation describes the profile (along the reactor yN2 = 0.565. The overall volumetric flow rate of feeding gas is
z-axis) of the temperature of solid phase (see general hypotheses of the Qg = 1.65 × 10−2 m3 s−1 (at inlet temperature T0 = 298 K). According
model and [27]), and takes into account the thermal dispersion in the to the general hypotheses of the model, a flow rate of Q0 = Qg/
bed, wall/solid and gas/solid radiation, wall/solid conductive and 7 = 2.35 × 10−3 m3 s−1 is fed at the reactor inlet (boundary condition
gas/solid convective phenomena: for Eq. (16)), while the rest is homogeneously distributed through the
other 6 nozzles up to z∗ = 1.4 m. A continuous axial profile of the feed
d ⎡Cp,B (z )
dTS (z )
⎤ rate of oxidising agent was defined, extended to the interval 0 ⩽ z ⩽ z ∗
d [W (z ) Cp,F TS (z )] d {W0 fM [1 − ξM (z )] Cp,LW (z ) TS (z )} ⎣ dz ⎦
dz
+ dz
+ αρB ε dz
= where the feeding manifold develops:
W E
− (1−fchar ) v0 k 0exp ⎡− RT (z ) ⎤ [1−ξ (z )]n [ΔHd + Cp,F TS (z )] ⎧ 67
PQg 1
0 ⩽ z ⩽ z∗
⎣ S ⎦ RT0 πD2 ∗
C (z )
I= z
4
−rM (z ) ⎡|λ H2 O (Z )| + MW
p,LW
TS (z )⎤+ ⎨
⎣ H2 O ⎦ ⎩0 z > z∗ (20)
4
+ D {(1−f gW / G ) foW / S σ [TW4 (z )−TS4 (z )] + (1−f gW / G ) hcond
W /S
(z )[TW (z )−TS (z )] Correspondingly, I = 6.48 × 10−4 kmol m−3 s−1 for z < z∗ = 1.4 m.
+ f gG / S f oG / S σ [T 4 (z )−TS4 (z )] + f gG / S hconv
G/S
[T (z )−TS (z )]} The boundary condition of the system of Eqs. (15-a)/(15-b) is Fj0 = 0 for
all the gaseous species other than O2 and N2, for which instead it is:
(18)
1 PQg 1
with the following boundary conditions: TS = TS0 at z = 0 and
dTS
=0 Fj0 = y j = O2, N2
dz 7 RT0 πD2 j
at z = L. 4 (21)

Accordingly, FO2 (z = 0) = 6.58 × kmol 10−5

and m−2 s−1
2.6. Energy balance equation: gas phase FN2 (z = 0) = 8.55 × 10−5 kmol m−2 s−1.
The semi-dried sewage sludge has specific heat Cp,F = 1.95 kJ kg−1 K−1
The following equation describes the profile (along the reactor [36] and devolatilisation enthalpy ΔHd = 300 kJ kg−1 (average value) [37].
z-axis) of the temperature of gas phase (see general hypotheses of the The volumetric hold up of solids (i.e. the fractional apparent volume of the
model), taking into account radiation and convective phenomena with reactor occupied by the bed solids) is ε = 0.19 . The coefficient of axial
wall and solid phase, and the thermochemistry of the homogeneous dispersion of solid particles in the bed of solids has been evaluated through
reactions: [20]:

473
F. Montagnaro et al. Fuel 231 (2018) 468–478

D⎛ ϕ
Δ= 1−cos ⎞
2⎝ 2⎠ (26)
It is Δ = 0.22 m, while KT ,B has been taken as function of the solid
W /S
temperature [38]. Therefore, hcond is a function of the z-axis. As an
−2
example, it is 9.1 × 10 and 3.7 × 10−2 kJ m−2 s−1 K−1 at a solid
temperature of 500 and 1000 K, respectively.
The wall temperature is assigned on the basis on operational ex-
perience with the pilot plant, and it is expressed as a function of z by
the following polynomial:

Fig. 3. Cross-sectional view of the oxy-pyrolyser, with indication of the geo-
metric features relevant to the view factors for radiative fluxes.
vdB
α=
10 (22)
where dB = 1 cm is the diameter of the ballast solid particles. It is
α = 9.78 × 10−7 m2 s−1. The density of the ballast material is
ρB = 2550 kg m−3.
The wall/gas geometric view factor (f gW / G ) has been calculated as
the ratio between the area of internal surface of reactor in contact with
the gas and that of the entire internal surface of the reactor itself. Both
areas are proportional to the reactor length, so f gW / G can be equivalently
expressed as the ratio between the length of the arc of circumference in
contact with the gas (from A to B clockwise in Fig. 3) and that of the
total circumference. Assuming the solid as entirely constituted by the
ballast material and neglecting the contribution of the sludge fuel, it is:
ϕ−sinϕ
ε=
2π (23)
with the angle ϕ indicated in Fig. 3. Since ɛ = 0.19, it is ϕ = 2.07 rad.
The length of circumference in contact with the solid (from A to B
D
counterclockwise in Fig. 3) is ϕ 2 : thus, the length in contact with the
D tion. A test (No. II in Ref. [29]) with intermediate (moderately harsh)
gas turns out to be πD−ϕ 2 and f gW / G = (π −0.5ϕ) = 0.67 .
The gas/solid geometric view factor ( f gG / S ) has been determined by
considering the solid contained into the reactor as a plate: this factor is
therefore the ratio between the area of the plate surface and that of the
internal surface of the reactor. Again, both areas are proportional to the
reactor length so this view factor can be expressed as the ratio between
ϕ
the length of the chord AB (Fig. 3, equal to Dsin 2 ) and that of the total
circumference. It is f gG / S = 0.27 . The total emission factors for gas/solid
and wall/solid radiation are set at 0.30 [27] and 1, respectively.
G/S
The gas/solid convective heat transfer coefficient hconv has been
calculated through [20]:
G/S
hconv D
Nu = = 0.322 Re 3 Pr
KT ,G (24)
where KT ,G is the thermal conductivity of gas. Under the operating
conditions of the oxy-pyrolyser, it is Re = 1435 and Pr = 0.71: thus
Nu = 11.2 and hconvG/S = 9.1 × 10−4 kJ m−2 s−1 K−1.
W /S
The wall/solid conductive heat transfer coefficient hcond is:
W /S KT ,B
hcond =
Δ (25)
where KT ,B is the thermal conductivity of the ballast material and Δ is
the depth of the solids bed (see Fig. 3):
⎧ C1 = 44.8
⎪C2 = −447
TW (z ) = C1 z3 + C2 z2 + C3 z + C4
⎨ C3 = 884
⎪ C4 = 735
⎩ (27)
with z expressed in [m] and TW in [K]. It is TW (z = 0) = 735 K and
TW (z = L) = 778 K . The wall temperature has a maximum value
(1230 K) at z = 1.21 m. The inlet solid temperature is TS0 = 298 K.
The total emission factor for wall/gas radiation is 0.30 [27]. The
W /G
wall/gas convective heat transfer coefficient hconv has been calculated
through [20]:
W /G
hconv D D 0.055
Nu = = 0.036Re 0.8 3 Pr ⎛ ⎞
KT ,G ⎝L⎠ (28)
W /G −4 −2 −1 −1
It is Nu = 10.2 and hconv = 8.3 × 10 kJ m s K .
4. Model validation
The subsequent Section 5 will be dedicated to detailed discussion of
model results using parameters presented in the previous text and
Table 1, and of the influence of relevant process and design variables. In
this Section validation of the model is accomplished against published
experimental data. To this end, data published by Freda et al. [29] have
been considered. First, details concerning the experiments discussed in
Ref. [29] and related values of the operating parameters, when different
from those listed in Table 1, are commented. The referred work deals
with thermal conversion of sewage sludge in a bench scale rotary kiln.
Experimental tests were carried out under different gasification condi-
tions, from milder to harsher. For each of them, the volumetric com-
position of the syngas at steady state was measured. Such information
allows a direct comparison with results arising from model computa-
gasification conditions was selected as reference case, since the model
here described is mostly focused on oxy-pyrolysis. Moreover, the fol-
lowing model parameters were adapted to allow a reliable comparison
between model computations and experimental data:
– In the work of Freda et al., the sewage sludge is dried prior to
feeding. As a consequence, the moisture fraction in the model was
set to 0.035 in agreement with the value of the experimental work.
On the other side, the ultimate analysis of the sewage sludge here
considered is very close to that of the experimental work. Thus, the
devolatilisation yields were not modified;
– The gasification agent, consisting of O2-enriched air in our simula-
tion, was switched to simple air to meet the specification of the
experimental work;
– Axial staging of the external oxidiser was removed. Instead, all the
gas was fed at the inlet of the reactor such as in the work of Freda
et al.;
– Sewage sludge and gas feed rate were modified to ensure that both
the O2-to-fuel inlet ratio and mean residence time of the solid and
gas phase assumed similar values with respect to those of the ex-
perimental work;
– The initial value of the wall temperature profile was modified to
ensure that the mean temperature of the gas phase was the same of
that used in the experimental tests.

474
F. Montagnaro et al. Fuel 231 (2018) 468–478

Table 2
Comparison between data predicted by model computation and experimental
data from Freda et al. [29] in terms of composition of produced syngas.
Gas species Model computations data Experimental literature data
(%v) (%v)

H2 18% 16%
N2 50% 47%
CO2 17% 15%
CO 11% 17%
CH4 4% 5%

The comparison between the experimental data of Freda et al. and

those obtained from model computations is summarised in Table 2.
Model results are in fair agreement with experimental data. For most of
the gaseous species, the data predicted by the model differ by less
than ± 3% from those obtained from the experiments. The largest error
is observed for CO concentration, whose predicted value differs by
Fig. 5. Axial profiles of ξ (devolatilisation degree) and ξM (dehumidification
nearly 6% from the experimental one. Overall, the comparison between
degree).
model and experimental data can be considered satisfactory and sup-
ports the validity of the proposed model.

5. Results and discussion

5.1. Conversion of solid fuel

Fig. 4 reports the axial mass flux of fuel W (on dry basis) along the
oxy-pyrolyser. Up to z = 0.45 m, W is constantly equal to its inlet value
since the temperature of the solid phase is not high enough to promote
devolatilisation. The thermal decomposition of the sludge becomes
active in the range z = 0.45–0.78 m (see ξ (z ) in Fig. 5), and is asso-
ciated with the steep decrease of W to 15 g m−2 s−1. Beyond z = 0.78
the flux of sludge remains constant and, since the devolatilisation is
complete, consists of sludge char only. For the general hypotheses of the
model, the heterogeneous gasification of the char is neglected. Fig. 5
also reports the degree of fuel drying ξM (z ). The loss of moisture is
active since the reactor inlet and is complete by z = 0.32 m, well before
the devolatilisation onset so that the two phenomena can be considered
in series. Fig. 6. Fluxes of O2 and N2 along the reactor, F .

5.2. Axial profiles of gas-phase composition

Fig. 6 reports the fluxes F of oxygen and nitrogen as a function of z .
In the “gas feeding zone” (0 < z < z∗ = 1.4 m) the N2 flux increases
due to the combination of sludge devolatilisation (active for z in the
range 0.45–0.78 m) and distributed feeding of the oxidising agent.
Fig. 4. Flux of sewage sludge along the reactor, W .
Beyond z∗ = 1.4 m, it remains constant as expected. The concentration
of O2 increases along the reactor in the range 0 < z < 0.52 m. Beyond
z = 0.52 m, O2 is rapidly consumed by combustion ((R1), (R2) and
(R3)) and partial oxidation ((R5) and (R7)) reactions. It is interesting to
note that the O2 flux vanishes already in the nozzle zone (for
z > 0.52 m) as the distributed feeding of the oxidising agent is fully
compensated by oxygen consumption due to homogeneous reactions.
Figs. 7 and 8 report the fluxes F of the gaseous species (on molar
basis) and of soot (on mass basis) along the converter.
Hydrogen is not present up to z = 0.55 m, due to the limited pro-
gress of fuel devolatilisation. Beyond z = 0.55 m the hydrogen con-
centration rapidly increases due to the combined effects of devolatili-
sation and of reactions (R5) and (R6), namely partial oxidation and
cracking of primary tars. Primary tar components, T1, are instead re-
leased as a direct consequence of sludge devolatilisation, to decrease
further on to vanishingly small values before the end of the feeding
zone z ∗ due to the progress of reactions (R5) and (R6). When con-
sidering the chemical pathways to H2, one should also take into account
(R7) and (R8) i.e. partial oxidation and cracking of secondary tars T2,
respectively. On the other hand Fig. 8 shows that, after generation of T2
components by cracking of primary tars (R6), the flux of T2 remains
practically constant. This indicates that the consumption of T2 due to
reactions (R7) and (R8) is negligible when compared with the net flux
of T2. Similar findings are observed for the flux of soot, whose forma-
tion becomes evident at z = 0.55 m as a result of the very slow progress

475
F. Montagnaro et al.
Fig. 7. Fluxes of H2, CO, CO2, H2O along the reactor, F .
Fig. 8. Fluxes of CH4, T1 and T2 (left-hand ordinate) and soot (right-hand or-
dinate) along the reactor, F .
of cracking of T2 (R8). H2 consumption by combustion (R1) in the
nozzle zone gives a modest contribution to the axial profile of H2 flux.
Altogether, a detailed analysis of the reaction pathways indicates that
the key processes for hydrogen production in the distributed feeding
zone are sludge devolatilisation and T1 partial oxidation (R5). Beyond
the feeding zone of the oxy-pyrolyser (z > z ∗), the further increase of H2
flux is due to the progress of the water-gas shift reaction (R4). Actually
a decreasing temperature profile establishes in this reactor zone (see
Section 5.3), reflected by the gradual shift of (R4) toward CO2 and H2.
Similarly to H2, the formation of CO in the feeding zone is dictated
by the combined effects of devolatilisation and partial oxidation of T1.
Beyond z = z ∗ the trend of the CO flux closely mirrors the flux of hy-
drogen, to which it is linked via the water-gas shift reaction.
Carbon dioxide is generated in the zone of distributed gas feeding
due to fuel devolatilisation as well as CO (R2) and CH4 (R3) combus-
tion. Considering the trend of CH4 flux in Fig. 8 (steadily increasing for
the z interval corresponding to the devolatilisation zone, constant
thereafter), it is concluded that CO2 is mainly formed by fuel devola-
tilisation in this zone. Changes of concentrations in the gas phase ac-
cording to the water-gas shift reaction (R4) are responsible for the in-
crease of CO2 flux in the second part of the reactor. Steam is produced
by release of fuel moisture in the early reaction zone, close to the fuel
feeding (Fig. 5). Then, when the operating conditions promote the H2
(R1) and CH4 (R3) combustion, a moderate increase of H2O con-
centration is again observed. Eventually, the concentration of H2O de-
creases to approach the water-gas shift equilibrium (R4).
476
Fuel 231 (2018) 468–478
Fig. 9. Axial profile of temperatures for solid and gas phases. The wall tem-
perature profile is reported for comparison.
5.3. Temperature profiles
Fig. 9 reports the axial profiles of solids, TS , and gas, T , temperatures. The
assigned axial profile of wall temperature is reported as reference. The
temperature of the solids remains close to the inlet value in the reactor zone
where drying is active (see Fig. 5, z < 0.32 m). This finding reflects the
substantial balance between the endotherm associated with sludge drying
and the heat transfer with the reactor walls. Soon after the end of the drying
zone the sludge temperature increases due to the thermal flux transferred
from the reactor walls, following the trend of the wall temperature TW
starting from z = 0.9 m. The rapid increase is related to the moderate
thermicity of devolatilisation as compared with the more pronounced en-
dotherm of fuel drying. In the second part of the reactor, beyond the dis-
tributed feeding zone, the sludge temperature exceeds the wall temperature
(TS > TW ) due to intense thermal feedback from the gas phase. The tem-
perature of the sludge at the discharge port is 875 K.
The temperature of the gas phase rapidly increases soon after the
reactor inlet, displaying a very pronounced peak at z = 0.55 m (1600 K)
due to the cumulative effects of heat transfer with the wall and course
of exothermal gas-phase reactions. At z = 0.75 m, T approaches TW . In
the second part of the reactor, T is constantly larger than the other two
temperatures (solid phase and wall). The temperature of the gas phase
at the oxy-pyrolyser outlet is 1030 K.
5.4. Properties of the syngas
Table 3 (“uniform oxygen staging” column) summarises the main
results of the computations, including the properties of the produced
syngas. A flow rate of syngas of 143 Nm3 h−1 (dry basis and free from
tars) was obtained from model computations. The main gaseous species
present in the syngas are, in the order: H2 (37.0% by volume), N2
(24.1%), CO2 (18.9%) and CO (17.3%), giving rise to a gaseous product
with a remarkably large H2/CO molar ratio (2.1). The additional
amount of water is 51 Nm3 h−1. The concentration of the primary tar at
the outlet of the reactor is almost null, while a flow rate of
0.25 Nm3 h−1 is estimated for the secondary tar. On the other side, the
amount of soot produced according by thermal cracking of T2 (R8) is
0.02 kg h−1. With this composition, the lower heating value of the
produced syngas is 7.1 MJ Nm−3. This value, multiplied by the syngas
flow rate, yields a syngas thermal throughput of 284 kW. Altogether,
the thermal throughput and the quality of the syngas (more than 50% of
the produced syngas is constituted by H2 and CO) provide a good
starting point for the development of a technically and economically
viable exploitation of sludge via oxy-pyrolysis.
F. Montagnaro et al. Fuel 231 (2018) 468–478

Table 3 with a remarkable reduction in tar and soot content. The superior
Comparison of the reactor performance with different feeding of the oxidising performance of delayed staging is explained by considering the much
agent. lower (1350 K) peak temperature in the gas phase as compared with the
Uniform Delayed No oxygen peak temperature in the base case (1600 K). The lower mean tem-
oxygen staging oxygen staging staging perature of gas considerably hinders the progress of secondary pyrolytic
reactions ((R6) and (R8)), promoting tar partial oxidation to syngas
Feed
(R5). Altogether, the delayed staging case yields a thermal
Sludge feed rate 100 kg h−1 100 kg h−1 100 kg h−1
Gas feed rate 54.4 Nm3 h−1 54.4 Nm3 h−1 54.4 Nm3 h−1 throughput > 300 kW with a very limited content of heavy compounds
(tars, soot).
Product
Syngas flow rate a 143.1 Nm3 h−1 148.4 Nm3 h−1 96.7 Nm3 h−1
6. Conclusions
H2 37.0% 37.5% 26.0%
N2 24.1% 23.3% 35.7%
CO2 18.9% 18.0% 26.6% The key result of the model computations is that the distributed
CO 17.3% 18.5% 8.4% feeding of the oxidising agent is beneficial to the productivity and the
CH4 2.7% 2.7% 3.3% quality of the syngas. Staging reduces the peak temperature in the gas
H2/CO 2.1 2.0 3.1 phase, hence hampers secondary pyrolysis and aromatisation of pri-
Lower heating value 7.1 MJ Nm−3 7.3 MJ Nm−3 5.0 MJ Nm−3 mary tars and makes O2 available for their partial oxidation. Primary
Thermal throughput 284 kW 303 kW 135 kW
tars are almost entirely gasified by partial oxidation: directly, or in-
Primary tar (T1) flow rate – – 0.32 Nm3 h−1 directly after thermal cracking to secondary tars. A modest fraction of
Secondary tar (T2) flow rate 0.25 Nm3 h−1 0.01 Nm3 h−1 1.44 Nm3 h−1
tars is eventually converted to soot. These favourable features are not
Soot flow rate 0.02 kg h−1 – 0.90 kg h−1
observed in the case with localised feeding of the oxidising agent at the
H2O flow rate 50.6 Nm3 h−1 48.6 Nm3 h−1 69.7 Nm3 h−1 reactor inlet. The uniform staging case yields a syngas with H2 +
a CO > 50%, H2/CO molar ratio around 2 and LHV of about
Composed by H2, N2, CO2, CO and CH4.
7.1 MJ Nm−3, with a productivity of 1.43 Nm3 per kg of sludge (dry
basis). Even better properties of the syngas are obtained when the
5.5. Effect of oxygen staging converter is operated with a delayed staging configuration, in which the
oxidising agent is uniformly fed in the region of the reactor where the
Table 3 compares results obtained with uniform staging of the fuel devolatilisation is active. On the other hand, the syngas obtained
oxidising agent in the feeding zone with those obtained assuming the with localised feeding (no oxygen staging) has the following worse
usual localised feeding of the oxidising agent at the reactor inlet (“no properties: H2 + CO < 35%, LHV of about 5.0 MJ Nm−3, productivity
oxygen staging” column). The comparison makes it possible to assess of 0.97 Nm3 per kg of sludge (dry basis). Moreover, localised feeding of
the potential of oxygen staging along the reactor as a mean to control oxidising agent yields much higher peak temperature (1800 vs. 1600 K)
the conversion pathway. and much larger (nearly 50-fold) production of soot. The results open to
Table 3 shows that localised feeding of O2 yields a syngas of much further investigation on the potential of staged oxy-pyrolysis of waste-
poorer quality characterised by a worse LHV (around 5.0 vs. derived fuels as a powerful tool to control the gasification pathways,
7.1 MJ Nm−3), reflecting the fact that less than 35% (vs. more than maximise the syngas production and quality, and minimise the PAH and
50% in the distributed feeding case) is constituted by H2 + CO. As soot formation.
previously highlighted, the more relevant path (besides sludge devo-
latilisation) to H2 + CO formation is the partial oxidation of primary Acknowledgments
tars (R5), that in the localised (no oxygen staging) feeding case is
hindered by lack of O2. When O2 is entirely fed at z = 0, in fact, it is The work referred to in this manuscript has been partly funded
preferentially consumed by combustion reactions immediately yielding within the “Tekne-Fluff Project: Innovative technology at high en-
large steam content in the syngas. As oxygen starving conditions es- ergetic efficiency and low environmental impact for production of
tablish, no O2 is available for partial oxidation of T1 (R5) and T2 (R7). electrical/thermal energy from car-fluff”, funded by Centro Sviluppo
This is further confirmed by the larger T1 + T2 contents in the syngas Materiali S.p.A. (Italy). The colleagues involved in this research Project
in the no oxygen staging case. The less favourable composition of are gratefully acknowledged for useful discussion.
syngas here obtained, together with a smaller syngas production rate,
concur in a significantly smaller thermal throughput associated with References
syngas production (135 vs. 284 kW). A deeper scrutiny on temperature
profiles suggests that much higher peak temperatures of gas phase [1] Anthony EJ. Fluidized bed combustion of alternative solid fuels; status, successes
(1880 vs. 1600 K) establish in the localised feeding case, due to the and problems of the technology. Prog Energy Combust Sci 1995;21:239–68.
[2] Werther J, Ogada T. Sewage sludge combustion. Prog Energy Combust Sci
pronounced exothermal effects of combustion promoted by the larger 1999;25:55–116.
O2 concentration at the reactor inlet. This feature largely promotes the [3] Scala F, Chirone R. Fluidized bed combustion of alternative solid fuels. Exp Therm
cracking of T2 (R8) and sooting, as compared with the distributed Fluid Sci 2004;28:691–9.
[4] Saxena SC, Jotshi CK. Fluidized-bed incineration of waste materials. Prog Energy
(uniform oxygen staging) feeding case, yielding a much larger (nearly Combust Sci 1994;20:281–324.
50 times) content of soot at the reactor outlet. [5] Desroches-Ducarne E, Marty E, Martin G, Delfosse L. Co-combustion of coal and
The comparison between the uniformly distributed and the localised municipal solid waste in a circulating fluidized bed. Fuel 1998;77:1311–5.
[6] Cliffe KR, Patumsawad S. Co-combustion of waste from olive oil production with
feeding cases highlights the very positive role that O2 staging may exert coal in a fluidised bed. Waste Manage 2001;21:49–53.
on syngas quality and production rate. This finding stimulated an ad- [7] Dichtl N, Rogge S, Bauerfeld K. Novel strategies in sewage sludge treatment. Clean
ditional step toward the search for optimal staging of oxygen. The 2007;35:473–9.
[8] Singh RP, Agrawal M. Potential benefits and risks of land application of sewage
analysis of model results indicates that the sludge devolatilisation is not
sludge. Waste Manage 2008;28:347–58.
active up to z = 0.45 m. Accordingly a “delayed” O2 staging config- [9] Kelessidis A, Stasinakis AS. Comparative study of the methods used for treatment
uration has been simulated, with 1/7 of the whole oxygen flow fed at and final disposal of sewage sludge in European countries. Waste Manage
the reactor inlet and the rest uniformly distributed in the 0.45–1.4 z- 2012;32:1186–95.
[10] Cammarota A, Chirone R, Salatino P, Solimene R, Urciuolo M. Particulate and
range. Table 3 lists the results obtained in the delayed staging case. A gaseous emissions during fluidized bed combustion of semi-dried sewage sludge:
moderate increase in the overall H2 + CO yield is observed together effect of bed ash accumulation on NOx formation. Waste Manage

477
F. Montagnaro et al.
2013;33:1397–402.
[11] Salatino P, Ammendola P, Bareschino P, Chirone R, Solimene R. Improving the
thermal performance of fluidized beds for concentrated solar power and thermal
energy storage. Powder Technol 2016;290:97–101.
[12] Tregambi C, Chirone R, Montagnaro F, Salatino P, Solimene R. Heat transfer in
directly irradiated fluidized beds. Sol Energy 2016;129:85–100.
[13] Chirone R, Salatino P, Scala F, Solimene R, Urciuolo M. Fluidized bed combustion of
pelletized biomass and waste-derived fuels. Combust Flame 2008;155:21–36.
[14] Urciuolo M, Solimene R, Chirone R, Salatino P. Fluidized bed combustion and
fragmentation of wet sewage sludge. Exp Therm Fluid Sci 2012;43:97–104.
[15] Sänger M, Werther J, Ogada T. NOx and N2O emission characteristics from fluidised
bed combustion of semi-dried municipal sewage sludge. Fuel 2001;80:167–77.
[16] Hartman M, Svoboda K, Pohořelý M, Trnka O. Combustion of dried sewage sludge
in a fluidized bed-reactor. Ind Eng Chem Res 2005;44:3432–41.
[17] Donatelli A, Iovane P, Molino A. High energy syngas production by waste tyres
steam gasification in a rotary kiln pilot plant. Experimental and numerical in-
vestigations. Fuel 2010;89:2721–8.
[18] Petersen I, Werther J. Experimental investigation and modeling of gasification of
sewage sludge in the circulating fluidized bed. Chem Eng Process 2005;44:717–36.
[19] Zhu JG, Yao Y, Lu QG, Gao M, Ouyang ZQ. Experimental investigation of gasifi-
cation and incineration characteristics of dried sewage sludge in a circulating
fluidized bed. Fuel 2015;150:441–7.
[20] Boateng AA. Rotary kilns. Elsevier; 2008.
[21] Nielsen AR, Larsen MB, Glarborg P, Dam-Johansen K. Devolatilization and com-
bustion of tire rubber and pine wood in a pilot scale rotary kiln. Energy Fuel
2012;26:854–68.
[22] Lombardi F, Lategano E, Cordiner S, Torretta V. Waste incineration in rotary kilns: a
new simulation combustion tool to support design and technical change. Waste
Manage Res 2013;31:739–50.
[23] Li AM, Li XD, Li SQ, Ren Y, Shang N, Chi Y, et al. Experimental studies on municipal
solid waste pyrolysis in a laboratory-scale rotary kiln. Energy 1999;24:209–18.
[24] Li SQ, Yao Q, Chi Y, Yan JH, Cen KF. Pilot-scale pyrolysis of scrap tires in a con-
tinuous rotary kiln reactor. Ind Eng Chem Res 2004;43:5133–45.
478
Fuel 231 (2018) 468–478
[25] Appelt J, Heschel W, Meyer B. Catalytic pyrolysis of central German lignite in a
semi–continuous rotary kiln – performance of pulverized one-way ZSM-5 catalyst
and ZSM-5-coated beads. Fuel Process Technol 2016;144:56–63.
[26] Chun YN, Kim SC, Yoshikawa K. Pyrolysis gasification of dried sewage sludge in a
combined screw and rotary kiln gasifier. Appl Energy 2011;88:1105–12.
[27] Hatzilyberis KS. Design of an indirect heat rotary kiln gasifier. Fuel Process Technol
2011;92:2429–54.
[28] Gikas P. Ultra high temperature gasification of municipal wastewater primary
biosolids in a rotary kiln reactor for the production of synthesis gas. J Environ
Manage 2017;203:688–94.
[29] Freda C, Cornacchia G, Romanelli A, Valerio V, Grieco M. Sewage sludge gasifi-
cation in a bench scale rotary kiln. Fuel 2018;212:88–94.
[30] Shi H, Si W, Li X. The concept, design and performance of a novel rotary kiln type
air-staged biomass gasifier. Energies 2016;9(67):1–18.
[31] Westerterp KR, van Swaaij WPM, Beenackers AACM. Chemical reactor and design
operation. Wiley; 1984.
[32] Tolve P, Salvati F, Paravidino M, Petriglieri S, Facciotto V. Process and apparatus
for producing fuel gas obtained by exhausted plastics. Int Patent WO 2015/181713;
2015.
[33] Senneca O, Chirone R, Salatino P. A thermogravimetric study of nonfossil solid
fuels. 2. Oxidative pyrolysis and char combustion. Energy Fuel 2002;16:661–8.
[34] Senneca O, Chirone R, Salatino P. Oxidative pyrolysis of solid fuels. J Anal Appl
Pyrol 2004;71:959–70.
[35] Marias F, Benzaoui A, Vaxelaire J, Gelix F, Nicol F. Fate of nitrogen during fluidized
incineration of sewage sludge. Estimation of NO and N2O content in the exhaust
gas. Energy Fuel 2015;29:4534–48.
[36] Kim Y, Parker W. A technical and economic evaluation of the pyrolysis of sewage
sludge for the production of bio-oil. Bioresour Technol 2008;99:1409–16.
[37] Caballero JA, Front R, Marcilla A, Conesa JA. Characterization of sewage sludges by
primary and secondary pyrolysis. J Anal Appl Pyrol 1997;40–41:433–50.
[38] Green DW, Perry RH. Perry’s chemical engineers’ handbook. 8th ed. McGraw-Hill;
2008.