Main

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Chemical Engineering Journal 425 (2021) 131509

Contents lists available at ScienceDirect

Chemical Engineering Journal


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

Electrified methane reforming: Elucidating transient phenomena


Sebastian T. Wismann a, b, Jakob S. Engbæk c, Søren B. Vendelbo c, Winnie L. Eriksen b,
Cathrine Frandsen a, Peter M. Mortensen b, *, Ib Chorkendorff a, *
a
Technical University of Denmark, Fysikvej 312, 2800 Kgs. Lyngby, Denmark
b
Haldor Topsoe A/S, Haldor Topsøes Allé 1, 2800 Kgs. Lyngby, Denmark
c
Danish Technological Institute, Gregersensvej 1, 2630 Tåstrup, Denmark

A R T I C L E I N F O A B S T R A C T

Keywords: Increasing implementation of renewable energy requires an infrastructure compatible with the intermittent
SMR production of green electricity. Herein we show the flexibility of electrically heated steam methane reforming
Electrification with integrated ohmic heating, through a combination of CFD modelling and lab scale reactor tests. It is shown
Industrial catalysis
how start-up from an idle state to operation conditions can be achieved with instantaneous application of the full
Transient
power required for a steady state conversion of 80%, with initial heating rates exceeding 50 ◦ C/min. The initial
heating rate is correlated to the thermal mass of the reactor, with the endothermic reaction governing the
temperature profile. Cyclical operation displays no apparent delay between the change in temperature and
methane conversion. The highest thermal gradient across the washcoat is predicted at steady state, with no
increase during start-up despite the higher heating rates. The highest risk of carbon formation is predicted at the
inlet at steady state operation. A temporarily peak in the equilibrated carbon potential is predicted near the
outlet during start-up and shutdown between 500 and 600 ◦ C, governed by the thermodynamics of the feed
composition. Integrated ohmic heating supports steam methane reforming scalable to industrial conditions,
operating closer to thermodynamic limits for carbon formation, and potentially based on the access to inter­
mittent excess of renewable energy.

1. Introduction Today, combustion of fossil fuels provide heat in conventional SMR


plants to drive the strongly endothermic reaction, resulting in emissions
Improving efficiency and a quickly growing global capacity for of large amounts of CO2-containing flue gas. A typical fired SMR process
renewable energy has resulted in economically competitive prices for based on natural gas emits 7–11 kg CO2/kg H2, depending on efficiency
green electricity compared to fossil fuels [1–3]. Without efficient solu­ and fuel [6–8]. The global production of syngas accounts for nearly 3%
tions for large-scale energy storage, the intermittent nature of renewable of global CO2 emissions [9,10].
energy creates a demand for technologies compatible with the excess A typical large-scale industrial syngas plant for hydrogen production
production [4]. Production of synthesis gas (syngas) through the endo­ contains several unit operations; a desulphurization section, a pre-
thermic steam methane reforming process (SMR) is a relevant approach reformer, the primary reformer, and a WGS section [5,11]. The pri­
for utilizing excess renewable electricity in a transition phase. Syngas is mary reformer is by far the largest, most complex, and most expensive
a mixture of hydrogen and carbon oxides, and serves as platform for unit operation for a large-scale syngas plant, and it is the only unit
production of many chemicals, including ammonia, methanol, and operation considered in the following [12]. For a typical large-scale
synthetic fuels [5]. The syngas composition is governed by the methane reformer more than hundred reactor tubes, in one or more rows, are
reforming reaction (Eq. (1)) and the water–gas-shift (WGS, Eq. (2)). heated by combustion in a large furnace box. Performance is typically
limited by heat transfer, and relatively narrow tubes (100 to 200 mm in
(1)

CH4 + H2 O⇌CO + 3H2 ΔHr = 206kJ/mol
external diameter, 10–13 m in length) are used to minimize gradients
across the catalyst bed in the individual tubes [11]. To avoid thermal
(2)

CO + H2 O⇌CO2 + H2 ΔHr = − 41kJ/mol
stress or formation of hotspots in the reactor tubes, it is imperative to

* Corresponding authors.
E-mail addresses: [email protected] (P.M. Mortensen), [email protected] (I. Chorkendorff).

https://doi.org/10.1016/j.cej.2021.131509
Received 17 May 2021; Received in revised form 21 July 2021; Accepted 24 July 2021
Available online 8 August 2021
1385-8947/© 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
S.T. Wismann et al. Chemical Engineering Journal 425 (2021) 131509

balance heat supply to the energy consumption by the endothermic re­


action. Safe start-up procedures can require several days to gradually
increase reactant flowrate and heat flux before reaching a steady state
for operation [5]. It is critical that the heat distribution between the
hundreds of reactor tubes in the furnace box is uniform, as low flux
limits performance, and too high flux for a single tube can lead to hot
spots or hot banding – leading to carbon formation and demanding
shutdown of the entire reformer [5,13].
Electrification of the reforming technology can substantially reduce
CO2 emissions, and in addition address key limiting aspects of the
conventional process. Induction heating of ferromagnetic catalyst par­
ticles enables heat delivered directly to the catalytic site, as demon­
strated by Vinum et al. [14]. However large scale is required to attain
competitive energy efficiency, as showed by Almind et al. [15]. Meloni
et al. demonstrated similar inversion of the thermal gradient using mi­
crowaves to heat a structured Ni/SiC catalyst [16]. This was also
demonstrated by Renda et al., heating a catalyzed SiC element by ohmic
resistance [17]. A highly robust system was demonstrated by Surov et al.
utilizing a high temperature plasma torch to drive a reforming reaction
without a catalyst [18].
Integrated ohmic heating enables substantial process intensification,
reducing reactor volume by up to two orders of magnitude compared to
equivalent fired reformer [10]. The short characteristic length scales
inverts typically limiting heat transfer to become the least limiting re­ Fig. 1. Reactor geometry Experimental reactor cross section, showing the
action step [19]. relevant elements. Each red line represents a thermocouple. The illustration is
This work investigates the potential for cold-starting a resistance- not to scale.
heated reformer, the transient response to fluctuations in supplied
power to mimic intermittent operation, and how it influences system the reactor, using two copper sockets at opposite ends of the reactor
gradients. The work is based on a laboratory scale reformer with inte­ (Fig. 1). The current is governed by the electrical resistance according to
grated ohmic heating, analyzed with detailed CFD model to elucidate Ohm’s law, resulting in a heat loss proportional to the square of the
the occurring phenomena. The typical limiting phenomena, carbon applied potential;
deposition and thermal gradients, are assessed, providing suggestions 2
P = URMS /RΩ
towards the possibility for the transient reforming of methane.
where URMS is the root mean square potential (RMS), and RΩ the elec­
2. Experimental setup trical resistance for the system, including wiring to the transformer coil
and the connections (contact resistance). The heat loss over the reactor
A custom build resistance-heated plug flow SMR bench scale reactor accounts for ca. 98% of the total electrical losses based on the measured
setup was used for the experiment. The principle reactor was made of a contact resistance. This does not include transformation losses. The root-
FeCrAl tube. The resistivity of FeCrAl is practically independent of mean-square (RMS) potential is calculated assuming an ideal sinusoidal
temperature, and a 6 mm tube (Goodfellow) of 50 cm in length was wave.
coated with a 130 µm porous zirconia washcoat on the internal surface. √̅̅̅
An 11 cm section of the coat was removed from each end through URMS = UPeak / 2
drilling, prior to impregnation with a nickel nitrate solution. A thin,
residual layer of catalytically active coat was present towards the outlet 3. CFD implementation
of the reactor (39–50 cm) [10]. The reactor was reduced in H2 and
passivated at room temperature. Several 0.25 mm K-type thermocouples Full fluid dynamic equations for momentum, mass transfer, heat
was spot-welded to the external surface to enable accurate measurement transport and electric fields was implemented in an axisymmetric ge­
of the temperature profile (SR630 Thermocouple monitor). The reactor ometry in Comsol 5.2a (Fig. 1). Flow is at all times laminar (Re less than
was insulated by high temperature granular insulation (FreeFlow). The 1100). Consequently, radial mass transfer depends on molecular diffu­
reactor configuration is shown in Fig. 1. sion. A fully developed laminar flow was defined at the inlet, and
The reactor was reduced in the ohmic heating setup prior to catalytic reference pressure at the outlet. The fluid momentum is defined by the
tests. A feed gas of methane, steam, and hydrogen (30/60/10) was used full Navier-Stokes equation for a compressible fluid.
for all experiments. Flowrate and composition was controlled by flow­ [ ]
meters (SLA850, Brooks digital flow controller). The feed gas was pre­ ∂u 2
ρ + ρ(u∙∇) = − ∇p + ∇∙ μmix ∇u − μmix (∇∙u)I
heated to 105 ◦ C to prevent condensation prior to the reactor. The ∂t 3
applied current was measured using a current clamp kept perpendicular In the porous washcoat, the brinkman modification is used [20].
to the current (Keysight 1146B), and the potential via an oscilloscope Heat transfer is defined by the standard equation [21]
(Agilent infiniiVision DSO-X 2014A). Contact resistance was measured ( )
before and after by a source meter at 1A (Keithley 2400 Sourcemeter). ρC p
∂T
+ u∙∇T = keff ∇2 T + Qh + Qr
Outlet gas composition was measured by gas chromatography (Agilent ∂t
7890, TCD/FID) or mass spectrometry (MS, Spectra Microvision plus). where Qh is the heatsource from ohmic heating and Qr the heat
Timescale for the MS data is calibrated against temperature source of the reactions. The measured temperature was defined for the
measurements. copper sockets, as their geometry was distorted in the axial-symmetrical
The reactor was heated by transforming 220–240 V from the grid approach. Heat losses were defined as a fixed temperature dependent
using a set of coils (N = 23), with the applied AC potential to the reactor flux on external surfaces, with additional radiation losses from the un-
controlled by a variotransformer. The AC potential was applied along

2
S.T. Wismann et al. Chemical Engineering Journal 425 (2021) 131509

insulated, lower part of the reactor (Fig. 1). Radiation from internal on the kinetic balance of adsorbed species against the reaction rates.
surfaces had no influence on the model predictions due to confined view Higher hydrocarbons display a high tendency towards forming carbon
factors in the narrow tube. Electrical heating was defined according to by irreversible decomposition on nickel catalysts [5]. However, a pre-
Ohm’s law; reformer can convert practically all higher hydrocarbons (C2+) at suf­
ficiently low temperatures to kinetically suppress formation of carbon,
Qh = E∙J
and carbon formation from higher hydrocarbons is thus not addressed
where E is the electric field [V/m], and J the current density vector [A/ further in this work [5,13,23].
m2]. A contact resistance was implemented equivalent to heat transfer The prevalent mechanism for carbon formation for the conditions
coefficients [10,19] relevant to methane reforming in the primary reformer is formation of
whisker carbon [24,25]. The characteristic carbon fibers of whisker
n∙J = hc ∙(Vi − Vj ) carbon (nanotubes) nucleate on the more reactive step sites [5,24]. If
formed, the high mechanical strength of the carbon whiskers will break
where n is the normal vector, (Vi − Vj ) the potential difference across the apart the catalyst pellets, leading to an increasing pressure drop, which
surface, and hc a resistance factor. Mass transfer was implemented on a can alter flow patterns leading to formation of hotspots, requiring a
mass fraction basis complete shutdown and replacement of the catalyst bed [26].
∂ωi The risk of carbon forming reactions can be assessed by the carbon
ρ
∂t
+ ∇∙ji + ρ(∇∙u)ωi = Ri activity. If the carbon activity exceeds unity, there is a thermodynamic
potential for the formation of carbon. However, catalysts with high
where Ri is the reaction rate, and ji is the mass flux for the i’th specie, intrinsic activity can operate at conditions where the apparent ther­
defined as; modynamic carbon activity in the bulk phase exceeds unity assuming
( ) the catalytic surface is in equilibrium. A critical carbon activity can be
∇Mi
ji = − ρDi ∇ωi + ρDi ωi estimated from the local concentration of species equilibrated with
Mi
respect to SMR and WGS (Eq. 1–2), reflecting the expected composition
Reaction kinetics was implemented for SMR, WGS, and reverse at the catalyst surface. The critical carbon activity, here for the methane
methanation [22]. The reaction rates were implemented as a function of decomposition reaction, can be calculated as;
approach to equilibrium. For the reforming reaction it is defined as; ⃒
pCH ⃒
pCH4 pH2 O aC,eq = Keq 2 4 ⃒⃒
R1 = k1 (1 − β1 ) pH2 eq
p2.5
H2

aC,eq > 1⇒Carbon


where (1 − β) is the approach to equilibrium, which for SMR is defined
as; where pi,eq is the partial pressure of the equilibrated process gas, and Keq
an equilibrium constant for the formation of carbon which depends on
pCO p3H2
β1 = the mechanism of the formed carbon.[27] The equilibrated carbon po­
pCH4 pH2 O Keq
tential is generally always below the thermodynamic potential for the
bulk composition. While graphitic carbon is the thermodynamically
where Keq is an equilibrium constant [5].
favorable form of carbon, it is kinetically suppressed, and less stable
For fluid domains, a triangular mesh with a maximum resolution of
whisker carbon is formed at typical SMR conditions which can be taken
0.1 mm was employed, with improved resolution in the boundary layer
into account by using a lower equilibrium constant.
domain. A quadratic mesh was used for the catalytic coat and reactor
As the carbon needs to nucleate, a critical particle diameter is
wall, with 12 elements across the coat. Remaining domains employed a
required, completely suppressing formation of carbon for small nano­
triangular mesh with a resolution of 5 mm or less. Further details on the
particles [27,28]. From a practical perspective the high temperatures,
implementation of equations, mesh sensitivity analysis, and model
metal loading, and steam content facilitate sintering, resulting in par­
fitting are described elsewhere [10,19].
ticle diameters quickly exceeding the critical threshold for nickel par­
ticles [5,11]. Consequently, carbon activity in industrial reformers is
4. Carbon activity calculations
controlled by operating at a high S/C and optimized reactor geometry
[11].
Low thermal conductivity across the catalyst bed and reactor wall
creates steep thermal gradients, increasing risk of phenomena detri­
5. Results
mental to reformer lifetime occurring, such as carbon formation, or
thermal stress. A net positive rate for carbon formation at any point in a
5.1. Cold start of SMR reactor
reactor is detrimental to catalyst lifetime and would be incompatible
with industrial operation [5,11]. The prevalent reactions for carbon
The intermittent production of renewable electricity can change
formation are the Boudouard reaction (Eq. (3)), CO reduction (Eq. (4)),
significantly throughout a day. This has generated a demand for
methane decomposition (Eq. (5)), and decomposition of higher hydro­
compatible technologies with a fast temporal response. Safe start-up
carbons (Eq. (6)), here with enthalpy given with respect to formation of
procedures for current industrial fired reformers typically require days
graphitic carbon;
( ) to avoid thermal stress or formation of hotspots.[5] With integrated
(3) ohmic heating, fast temporal response is feasible, as the electrical con­

2CO⇌C + CO2 ΔHr = − 172kJ/mol
ductivity of FeCrAl is practically independent of temperature [29]. This
( )
(4) enables homogeneous heating along the entire reactor length and alle­

CO + H2 ⇌C + H2 O ΔHr = − 131kJ/mol
viates the risk of transient hotspot formation.
( )
(5) Fig. 2 shows the temperature development of the maximum tem­

CH4 ⇌C + 2H2 ΔHr = + 75kJ/mol
perature for the lab scale reactor heated from inlet conditions (105 ◦ C,
( )

Cn Hm →nC + ½mH2 ΔHr > 0kJ/mol (6) S/C 2, 1 atm) to operation conditions (80% methane conversion), by
instantaneous application of the steady state potential to a “cold”
Formation of carbon occurs on the catalyst surface and is dependent reactor. The “cold” reactor starts at 15% of the steady state power to

3
S.T. Wismann et al. Chemical Engineering Journal 425 (2021) 131509

Fig. 3. Transient temperature profiles at ambient pressure. Simulated (full


lines) and experimental (dots) initial temperature profiles for an electrically
heated reformer initiated from inlet conditions (cf. Fig. 2).

of catalyst, as explained in previous work [10,19].


Fig. 2. Cold start of electrified methane reformer at ambient pressure. Fig. 4 shows the radial temperature gradient at the top of the coat (z
Maximum temperature and outlet conversion as a function of time. Power was = 11 cm) where the largest difference is predicted due to an edge effect,
15% of the steady state conditions before t = 0. Open circles show experimental and 5 mm below as indicated in the insert. The gradient across the
data and full lines show model predictions. The temperature position is illus­
washcoat reaches a maximum at steady state where the highest tem­
trated in the insert, 7.1 cm from the outlet. Conditions: inlet temperature
perature, and consequently reaction rate occurs. The slope of the tran­
105 ◦ C, feed composition CH4:H2O:H2 (30/60/10), ambient pressure, 920
sient gradient across the washcoat is correlated to the reaction rate, as it
Nml/min.
can be seen to level off after 3 min, corresponding to the onset of the
endothermic reaction (Fig. 2). The thermal difference across the coat
prevent condensation of steam. Both experimental data and simulations
peaks at the ends of the coat due to edge effects, as relative more cata­
are combined in the chart. At 15% power the maximum temperature is
lytic area is exposed facilitating additional cooling due to endothermic
275 ◦ C, and there is no quantifiable or expected conversion of methane.
reactions.
To limit uncertainty related to the CO and CO2 signal of the MS, Fig. 2 is
Practically no gradient is predicted across the reactor wall, and less
based on the methane content in the dry gas. The transient model pre­
than 2 ◦ C across the catalytic coat at the conditions of Fig. 4. The thermal
dicts a slightly slower heating of the reactor, and consequently lower
gradient is correlated nearly linearly to the inward heat flux, thus
initial conversion, but generally predict the maximum temperature
depending on operation conditions and increasing with flowrate or
within ±10οC.
conversion.
Initial start-up time is correlated to the applied power and the
thermal mass of the reactor wall, catalyst, and insulation, based on the
near linear increase in temperature the first 2 min. Above 600 ◦ C, the
reaction rate of the strongly endothermic reaction becomes significant,
substantially changing the slope. The reactor reaches 95% of the thermal 2.5
steady state within 12 min, while nearly 30 min are required to reach
95% of steady state conversion. The equilibration time is primarily Coat edge
dependent on the thermal diffusivity of the insulation. Initial heating 2
2.5
rate is practically independent of flow rate as the thermal mass of the
2
flow is more than two orders of magnitude lower than that of the reactor
ΔTCoat [K]

and the temperatures are insufficient for a substantial contribution from 1.5
1.5
ΔTCoat [K]

the endothermic reaction. Increasing flowrate or conversion requires a 1 Coat


higher heat flux, and concomitantly an increase in the applied potential, 0.5
Reactor wall
resulting in faster start-up as the thermal mass is nearly constant. 0
1 10 20 30 40
Heating rates coinciding with the characteristic timescales are practi­ Axial position [cm]
cally impossible for the given design, given the high thermal mass of the
reactor [19].
Heating substantially faster does not bring further benefits for ap­ 0.5 5 mm from edge
plications utilizing an intermittent supply of renewable energy, as the
grid is typically regulated on an hourly basis [30].
Due to the temperature-independent electrical resistance, a large
part of the reactor is initially nearly uniform in temperature (Fig. 3).
0
After 1 min, the temperature is too low for the reaction rate to signifi­
0 10 20 30 40 50 60
cantly influence the heat losses, resulting in a near constant temperature Time [min]
along the insulated part of the reactor, due to the inherent uniform
ohmic heating. As the temperature increases, the exponentially depen­ Fig. 4. Modelled transient temperature gradient. Predicted radial temperature
gradient across the catalytic washcoat evaluated at the start of the coat and 5
dent reaction rate will consume more heat, limiting the increase in
mm below. The insert shows the radial thermal gradient at steady state con­
temperature. As the reaction proceeds, higher temperatures are required
ditions along the catalytic washcoat. The arrows indicate the position for the
to drive it further, shaping the temperature profile. The small step at 39 transient profile. The timescale corresponds to the results in Fig. 2. 920 Nml/
cm (Fig. 3) is due to insufficient catalytic activity in a thin residual layer min, CH4:H2O:H2 (30:60:10), 1 atm.

4
S.T. Wismann et al. Chemical Engineering Journal 425 (2021) 131509

5.2. Transient response equilibrium fraction is greater as the reaction is kinetically limited due
to insufficient temperatures.
Adding on the understanding of the transient response of the system, The model predicts slightly faster cooling of the reactor, correlated to
also thermal cycling of the system was simulated to understand the time- an overprediction of the methane fraction, as the temperature decreases.
wise response to electricity availability as e.g. induced by intermittent The peak temperature decreases over time, indicating the system does
production of renewable electricity. Fig. 5 shows the response to rapid not reach a steady state. This is correlated to the thick insulation, as the
changes in energy input for the electrically heated reformer. Starting characteristic thermal timescale for the reactor and catalyst is within
from a steady state at 80% methane conversion at an applied potential of seconds, while hours for the insulation [19].
2.9 V (t = 0 s), the RMS potential is abruptly changed every 2 min, No distinguishable transient deviation between temperature and
between 1.3 and 2.9 V. At 1.3 V the corresponding steady state con­ reaction kinetics (methane fraction) is observed on the given timescale.
version is less than 20%. The transient operation in Fig. 5 was modelled This is not unexpected given the inherent uniform heat source from
in a transient CFD model approach. The model accurately predicts the ohmic heating for a material with electrical conductivity independent of
behavior at high temperature (T ±10 ◦ C, XCH4±2%), with slightly temperature, and characteristic timescales below seconds for the cata­
increasing deviation over time for low conversion (T ±23 ◦ C), as illus­ lytic processes [19]. As the applied potential is decreased, the inward
trated by the difference between full lines and points in Fig. 5. Consid­ heat flux is uniformly lowered, resulting in a downward shift of the
ering the accumulating uncertainties related to time, temperature, entire temperature profile (Fig. 6).
potential, and material parameters, the deviation is considered The lower heat flux is most visible near the inlet, between the copper
acceptable. socket and catalytically activity region (6–11 cm), where the slope de­
Decreasing the potential from 2.9 to 1.3 V corresponds to an 80% creases substantially. The slope along the catalytic washcoat is nearly
decrease in supplied energy (Eq. (9)). The endothermic reaction facili­ the same, indicating the supplied heat is consumed by the reaction,
tates an initial steep drop in temperature consuming the latent energy. keeping the reaction near equilibrium.
As the temperature decreases, the reaction rate slows to the point where
cooling is predominantly through external losses. Fig. 5B shows the
5.3. Carbon formation risk
measured, modelled and equilibrium dry methane fraction, increasing
as the temperature drops, as less methane is converted. The equilibrium
Design and operation of industrial reformers is to a large extend
fraction is calculated based on the measured maximum temperature and
constrained by the need to avoid formation of carbon. Carbon deposition
is at all points below the measured and predicted fraction, illustrating
is detrimental to SMR catalysis, and results in deterioration of the
the endothermic driving force of the reactions towards cooling the sys­
catalyst, large pressure drops, formation of hotspots, and in worst case,
tem. At low temperatures the deviation between experimental and
rupture of the reactor tube [5,26]. Fig. 7 shows the temperature and
carbon activity for the experimental reactor, with the carbon activity
evaluated for the actual gas with respect to graphite (the worst-case
scenario). Of the three carbon forming reactions evaluated (Eq. 3–5),
the conditions in the actual gas determines which is prevalent. The
Boudouard and CO reduction reaction have a similar thermodynamic
behavior, and consequently the CO reduction reaction is not illustrated
in Fig. 7. The carbon activity greatly exceeds unity for the methane
decomposition reaction (Eq. (5)) near the top of the washcoat, corre­
lated to the fast pre-heating in the region between the copper socket and
catalytic washcoat. A similar increase, though significantly lower (aMDC
1.1) is observed below the washcoat, for the region with a thin, residual
washcoat. The un-insulated part of the reactor heat exchanges with the
surroundings towards the lower copper socket, swiftly cooling the CO
rich process gas, resulting in a steep increase in the thermodynamic
potential for the exothermic Boudouard reaction.
The carbon activity in the catalytic washcoat initially exceeds unity
due to preheating, but quickly decrease as the reaction proceeds towards

Fig. 5. Transient response of electrified methane reformer. Measured and


modelled transient response. A) Peak Temperature (z = 44.5 cm). B) Methane
fraction in dry gas. C) Applied RMS potential over reactor. Conditions. Inlet Fig. 6. Transient temperature profiles. Modelled and measured temperature
105 ◦ C, feed composition CH4:H2O:H2 (30/60/10), ambient pressure, 680 Nml/ profiles at 0, 120, and 240 s respectively (cf. Fig. 5). Full lines are model pre­
min. MS time resolution is calibrated against temperature measurements. dictions, with experimental values as circles.

5
S.T. Wismann et al. Chemical Engineering Journal 425 (2021) 131509

conditions of SMR without higher hydrocarbons [33]. Fig. 8 shows the


temperature and carbon activity along the catalytically active reactor
surface. It is observed how the initial peak in carbon activity for methane
decomposition (eq. (5)) decrease along the surface, as the process gas
shifts towards equilibrium. Based on equilibrated gas, there is no carbon
potential along the catalytic surface. The thermodynamic carbon ac­
tivity for the equilibrated gas is independent of reaction. As there is no
CO in the feed gas, the activity of the Boudouard and CO reduction re­
action is zero before the catalytic zone.
The actual gas carbon activity exceeding unity outside the catalytic
region in the top and bottom of the reactor as illustrated in Fig. 8 is a
consequence of preheating taking place in the initial part of the reactor,
and the rapid cooling due to heat exchange of the un-insulated part of
the tube and are a compromise in the experimental setup (See Fig. 1).
However, handling local actual gas activity exceeding unity is done in
modern SMR based plants, and through use of pre-reforming technol­
ogy, proper preheating, and syngas cooling technology, these aspects
can all be controlled as long the equilibrated gas carbon potential stays
below 1 [5,23,34].
Fig. 7. Temperature and actual carbon activity at ambient pressure. Thermal
contours and maximum carbon activity evaluated between the methance 5.4. Transient carbon activity
decomposition (aMDC) and Boudouard reaction (aBoud.) with respect to graphite.
White lines mark the regions for carbon activity exceeding unity. Conditions: To investigate the transient carbon activity, start-up and shutdown of
1700 Nml/min, ambient pressure, CH4:H2O:H2 (30/60/10).
the reactor was modelled. The start-up was initiated from a steady state
simulation with no applied power from a temperature equivalent to inlet
equilibrium as there is no carbon risk for the equilibrated gas. The conditions by an abrupt supply of the desired current for steady state
carbon activity drops across the coat as the reaction is driven close to operation at 80% methane conversion (Equivalent to Fig. 2). The shut­
equilibrium. Despite the initial high carbon activity in the inlet section, down is modelled equivalently by changing the applied current to zero
there is no formation of carbon due to the equilibrated gas carbon ac­ from a steady state solution equivalent to 80% conversion. Fig. 9 shows
tivity is below 1 throughout the reactor (as shown in Fig. 8), which the temperature and equilibrated carbon activity for start-up and shut­
means that the overall reaction scheme is guided towards the syngas down of an electrically heated reformer. Initial change in temperature of
product due to the high partial pressure of steam and the SMR reaction the reactor wall is fast for both start-up and shutdown, where the low
being kinetically favorable [5]. This was further evidenced in end-of-run thermal diffusivity of the insulation results in slow thermal
inspection of the washcoat in the top section, where no traces of carbon equilibration.
were found. The small thermal gradients across the washcoat keeps the At no point during start-up or shutdown is the equilibrated carbon
entire catalytic region near equilibrium, suppressing carbon formation, potential exceeding unity, indicating no formation of carbon (at given
enabling operation at harsher conditions than feasible for conventional conditions). A transient peak in the carbon activity is observed both
pellet-based catalyst [31,32]. during start-up and shutdown near the outlet (z = 43 cm, Fig. 8). The
No formation of graphitic carbon is expected in the gas phase at the peak occurs at 550 ◦ C and is a thermodynamic effect, as reflected in the
carbon potential for the equilibrium composition, indicating the reac­
tion is close to equilibrium at the surface.

Fig. 8. Carbon activity and temperature on the internal surface. Carbon activity
for the methane decomposition (aMDC) and Boudouard reaction (aBoud.) along
the catalytic surface, including the equilibrated carbon activity (aEq.) for the Fig. 9. Transient carbon activity during start-up and shutdown. Modelled
catalytic region. Up to 11 cm, there is no catalytic activity. The equilibrium equilibrated carbon activity and temperature evaluated at inlet and outlet (z =
temperature is calculated w.r.t. SMR (Eq. (1)). Conditions: CH4:H2O:H2 (30/60/ 11 and 43 cm, cf. Fig. 7) during start-up from inlet conditions and shutdown.
10), ambient pressure. Conditions: 920 Nml/min, CH4:H2O:H2 (30/60/10), ambient pressure.

6
S.T. Wismann et al. Chemical Engineering Journal 425 (2021) 131509

Overall, the data evidence that transient carbon activities easily Hough, Global Carbon Budget 2018, Earth Syst. Sci. Data. 10 (2018) 2141–2194.
https://doi.org/10.5194/essd-7-349-2015.
adjust to the chemical system during start-up and shutdown. This
[10] S.T. Wismann, J.S. Engbæk, S.B. Vendelbo, F.B. Bendixen, W.L. Eriksen, K.
demonstrates a large agility of the technology and the prospect of rapid Aasberg-petersen, C. Frandsen, I. Chorkendorff, P.M. Mortensen, Electrified
change of operating conditions. Overall, this points towards that the methane reforming: A compact approach to greener industrial hydrogen
electrically heated steam reformer debottlenecks the slow response of production, Science (80-.). 364 (2019) 756–759. https://doi.org/10.1126/science.
aaw8775.
the reforming step in syngas plants. Consequently, this can help bring [11] P.M. Mortensen, I. Dybkjær, Industrial scale experience on steam reforming of CO2-
syngas plants closer to become a balancing technology in the energy rich gas, Appl. Catal. A Gen. 495 (2015) 141–151, https://doi.org/10.1016/j.
infrastructure. apcata.2015.02.022.
[12] J.R. Rostrup-Nielsen, J. Sehested, J.K. Nørskov, Hydrogen and synthesis gas by
steam- and CO2 reforming, Adv. Catal. 47 (2002) 65–139, https://doi.org/
6. Conclusions 10.1002/chin.200317288.
[13] M. Behnam, A.G. Dixon, 3D CFD simulations of local carbon formation in steam
methane reforming catalyst particles, Int. J. Chem. React. Eng. 15 (2017) 1–17,
Increasing implementation of renewable energy requires technolo­ https://doi.org/10.1515/ijcre-2017-0067.
gies compatible with the intermittent nature of green electricity. Elec­ [14] M.G. Vinum, M.R. Almind, J.S. Engbæk, Søren.B. Vendelbo, M.F. Hansen,
trical heating of the strongly endothermic steam methane reforming C. Frandsen, J. Bendix, P.M. Mortensen, Dual-function cobalt–nickel nanoparticles
tailored for high-temperature induction-heated steam methane reforming, Angew.
process can contribute to a substantial reduction in CO2 emissions, and Chemie - Int. Ed. 57 (33) (2018) 10569–10573, https://doi.org/10.1002/anie.
offers a prospective large off-taker of surplus electricity. Integrated v57.3310.1002/anie.201804832.
ohmic heating inherently provides improved thermal control, and the [15] M.R. Almind, S.B. Vendelbo, M.F. Hansen, M.G. Vinum, C. Frandsen, P.
M. Mortensen, J.S. Engbæk, Improving performance of induction-heated steam
intimate contact between catalyst and heat source, enables fast thermal
methane reforming, Catal. Today. 342 (2020) 13–20, https://doi.org/10.1016/j.
response, such as start-up or shutdown within minutes without transient cattod.2019.05.005.
detrimental phenomena. With a transient response on a timescale of [16] E. Meloni, M. Martino, A. Ricca, V. Palma, Ultracompact methane steam reforming
minutes rather than days, an electrically heated reformer is not a reactor based on microwaves susceptible structured catalysts for distributed
hydrogen production, Int. J. Hydrogen Energy. 46 (26) (2021) 13729–13747,
bottleneck for transient operation, in contrast to conventional fired re­ https://doi.org/10.1016/j.ijhydene.2020.06.299.
formers. With minimal thermal gradients across the catalyst, carbon [17] S. Renda, M. Cortese, G. Iervolino, M. Martino, E. Meloni, V. Palma, Electrically
activity is governed by the equilibrium, enabling operation close to driven SiC-based structured catalysts for intensified reforming processes, Catal.
Today. (2020), https://doi.org/10.1016/j.cattod.2020.11.020.
thermodynamic constrains. Electrically heated reformers consequently [18] A.V. Surov, S.D. Popov, V.E. Popov, D.I. Subbotin, E.O. Serba, V.A. Spodobin, G.
provide substantial flexibility towards intermittent operation and could V. Nakonechny, A.V. Pavlov, Multi-gas AC plasma torches for gasification of
be an enabling factor towards plant designs based on utilizing excess organic substances, Fuel. 203 (2017) 1007–1014, https://doi.org/10.1016/j.
fuel.2017.02.104.
renewable energy for a greener production chemical according to [19] S.T. Wismann, J.S. Engbæk, S.B. Vendelbo, W.L. Eriksen, C. Frandsen, P.M.
availability of renewable electricity. Mortensen, I. Chorkendorff, Electrified methane reforming: Understanding the
dynamic interplay, Submitted. (2019).
[20] A. Cybulski, J. Mouljin, Structured catalysts and reactors, Marcel Dekker, 1998.
Funding [21] R.B. Bird, W.E. Stewart, E.N. Lightfoot, Transport Phenomena, 2nd ed., Wiley,
2002.
This work was supported by Innovation Fund Denmark (IFD) under [22] J.G. Xu, G.F. Froment, Methane steam reforming, methanation and water-gas
shift.1. Intrinsic Kinetics, Aiche J. 35 (1989) 88–96, https://doi.org/10.1002/
file no. 5160-00004B and research grant 9455 from Villum Fonden.
aic.690350109.
[23] T.S. Christensen, Adiabatic prereforming of hydrocarbons - An important step in
Declaration of Competing Interest syngas production, Appl. Catal. A Gen. 138 (2) (1996) 285–309, https://doi.org/
10.1016/0926-860X(95)00302-9.
[24] S. Helveg, J. Sehested, J.R. Rostrup-Nielsen, Whisker carbon in perspective, Catal.
The authors declare that they have no known competing financial Today. 178 (1) (2011) 42–46, https://doi.org/10.1016/j.cattod.2011.06.023.
interests or personal relationships that could have appeared to influence [25] S. Helveg, P.L. Hansen, Atomic-scale studies of metallic nanocluster catalysts by in
the work reported in this paper. situ high-resolution transmission electron microscopy, Catal. Today. 111 (1-2)
(2006) 68–73, https://doi.org/10.1016/j.cattod.2005.10.019.
[26] J. Sehested, Four challenges for nickel steam-reforming catalysts, Catal. Today. 111
References (1-2) (2006) 103–110, https://doi.org/10.1016/j.cattod.2005.10.002.
[27] J.R. Rostrup-Nielsen, J.H.B. Hansen, CO2-reforming of methane over Transition
[1] Lazard, Lazard’s levelized cost of energy analysis - version 12.0, 2018. https:// Metals, J. Catal. 144 (1) (1993) 38–49.
www.lazard.com/media/450784/lazards-levelized-cost-of-energy-version-120- [28] H.S. Bengaard, J.K. Nørskov, J. Sehested, B.S. Clausen, L.P. Nielsen, A.
vfinal.pdf. M. Molenbroek, J.R. Rostrup-Nielsen, Steam reforming and graphite formation on
[2] British Petroleum (BP), BP Statistical Review of World Energy 2019, 2019. https:// Ni catalysts, J. Catal. 209 (2) (2002) 365–384, https://doi.org/10.1006/
doi.org/10.2307/3324639. jcat.2002.3579.
[3] I.E. Agency, Key World Energy Statistics 2017, 2017. https://www.iea.org/ [29] MatWeb - Fecralloy, (n.d.). http://www.matweb.com/search/datasheet.aspx?
publications/freepublications/publication/KeyWorld2017.pdf. matguid=c2427c6297594858bedac2a4e5981d2f&ckck=1 (accessed September
[4] V.M. Fthenakis, T. Nikolakakis, Storage Options for Photovoltaics, in: Compr. 12, 2018).
Renew. Energy, 1st ed., Elsevier Ltd., 2012: pp. 199–212. https://doi.org/10.1016/ [30] European Commission, Study on energy prices, costs and their impact on industry
B978-0-08-087872-0.00106-2. and households, 2020. https://doi.org/10.2833/49063.
[5] J. Rostrup-Nielsen, L.J. Christiansen, Concepts in Syngas Manufacture, Imperial [31] A.G. Dixon, M.E. Taskin, M. Nijemeisland, E.H. Stitt, Systematic mesh development
College Press, 2011. for 3D CFD simulation of fixed beds: Single sphere study, Computers Chem. Eng. 35
[6] OECD/IEA, Tracking Industrial Energy Efficiency and CO2 Emissions, (2007) 83. (2011) 1171–1185, https://doi.org/10.1016/j.compchemeng.2010.12.006.
https://doi.org/10.1007/978-3-642-29446-4_14. [32] D. Pashchenko, I. Makarov, Carbon deposition in steam methane reforming over a
[7] P.L. Spath, M.K. Mann, Life Cycle Assessment of Hydrogen Production via Natural Ni-based catalyst: Experimental and thermodynamic analysis, Energy. 222 (2021)
Gas Steam Reforming, International Energy Agency (IEA), 2001. https://doi.org/ 119993, https://doi.org/10.1016/j.energy.2021.119993.
10.2172/764485. [33] D. Pashchenko, Thermodynamic equilibrium analysis of steam methane reforming
[8] A. Kumar, M. Baldea, T.F. Edgar, A physics-based model for industrial steam- based on a conjugate solution of material balance and law action mass equations
methane reformer optimization with non-uniform temperature field, Comput. with the detailed energy balance, Int. J. Energy Res. 44 (1) (2020) 438–447,
Chem. Eng. 105 (2017) 224–236, https://doi.org/10.1016/j. https://doi.org/10.1002/er.v44.110.1002/er.4943.
compchemeng.2017.01.002. [34] K. Aasberg-Petersen, I. Dybkjær, C.V. Ovesen, N.C. Schjødt, J. Sehested, S.
[9] N. Le Quéré, C., Moriarty, R., Andrew, R.M., Peters, G.P., Ciais, P., Friedlingstein, G. Thomsen, Natural gas to synthesis gas - Catalysts and catalytic processes, J. Nat.
P., Jones, S.D., Sitch, S., Tans, P., Arneth, A., Boden, T.A., Bopp, L., Bozec, Y., Gas Sci. Eng. 3 (2) (2011) 423–459, https://doi.org/10.1016/j.jngse.2011.03.004.
Canadell, J.G., Chini, L.P., Chevallier, F., Cosca, C.E., Harris, I., Hoppema, M.,

You might also like