You are on page 1of 14

DOI: 10.1002/ente.


Flexible Operation of Fixed-Bed Reactors for a Catalytic

Fuel SynthesisCO2 Hydrogenation as Example Reaction
Mara Iglesias Gonzlez,* Hilko Eilers, and Georg Schaub[a]

It may become attractive in the future to operate fuel-related over Fe catalysts to short-chain hydrocarbons, which can be
chemical reactions, with H2 as reactant, under flexible load used to adjust the heating value of substitute natural gas, is
conditions when H2 is produced through H2O electrolysis investigated as example reaction. A specific reactor design
using fluctuating renewable energy. In this way, the size of for flexible operation with gas recycling is a consequence of
the H2 storage device connected to the electrolysis cell can the constraints resulting from product partial pressures and
be reduced and, consequently, its investment costs. The reaction temperature. The experimental and mathematical
design of a flexible reactor depends on the characteristics of methods developed can be applied to other fuel-synthesis
the catalyst and the chemical reaction. CO2 hydrogenation processes.


In a future energy system, primary energy supply

will be based to a large extent on renewable sour-
ces such as wind and sun, leading to electricity with
fluctuating generation characteristics. As chemical
energy carriers will still be needed in mobile and
other applications, conversion of electricity into
chemical energy may then become important. For
example, fluctuating electricity can be converted
into hydrocarbon fuels, via hydrogen as intermedi- Figure 1. Flow diagram of an electricity-to-fuel process chain with H2 as intermediate and
gaseous hydrocarbons as product. Example H2 flow pattern (FH ) produced by H2O elec-
ate, which can replace fossil fuels and can be attrac- 2

trolysis from fluctuating electricity and example H2 flow patterns at storage outlet.
tive as a long-term storage in excess electricity-gen- Dashed line: large H2 storage size, catalytic reactor operating at steady-state conditions.
eration situations. Potential sources of concentrated Full lines: H2 storage smaller with two different control strategies, catalytic reactor operat-
CO2 are facilities in which biomass is fermented ing under transient conditions.
either to ethanol or methane, biomass combustion
or gasification (leading to CO/CO2 mixtures), and industrial ble conditions is qualitatively outlined based on information
processes such as iron and steel production and calcination from literature (Tables 1 and 2).
in the cement industry. A general flow diagram of an elec- Among the reactions indicated, methanol, dimethylether
tricity-to-hydrocarbon process (Figure 1) includes electrolysis (DME), and methanation synthesis can be seen as relatively
for fluctuating H2 generation, H2 storage, and catalytic syn- easy routes to be operated under flexible conditions due to
thesis processes. In this context, flexible operation of chemi- the high and robust selectivity of the desired product with
cal synthesis reactors may be motivated by the possibility of limited sensitivity towards temperature variations and cata-
reducing the size of the storage and the investment needed.[1] lyst stability. The LTFT synthesis, methanol-to-gasoline con-
version, and CO2 hydrogenation to gaseous hydrocarbons
can be seen as more complex syntheses in this respect
Fuel synthesis reactions
(Table 2). Therefore, based on information found in litera-
The flexibility that a chemical process can achieve is directly ture, they may be considered as more demanding for flexible
related to the characteristics of the chemical reaction (e.g.,
heat of reaction, product selectivity, and reaction rate equa-
[a] M. Iglesias Gonzlez, H. Eilers, Prof. G. Schaub
tion). Table 1 summarizes general information of different Division of Fuel Chemistry and Technology
fuel-related routes using H2 as reactant and potentially fixed- Engler-Bunte-Institut, Karlsruhe Institute of Technology (KIT)
bed synthesis reactors: methanol-to-gasoline (MTG) conver- Engler-Bunte-Ring 1, 76131 Karlsruhe (Germany)
sion, methane synthesis, low-temperature FischerTropsch
This publication is part of a Special Issue on the Energy, Science & Tech-
(LTFT) reaction with CO, and high-temperature Fischer
nology conference in Karlsruhe, Germany. To view the complete issue
Tropsch (HTFT) reaction with CO2. A first indication of the visit:
probable success when operating these reactions under flexi-

Energy Technol. 2016, 4, 90 103 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 90
Table 1. Fuel-related synthesis reactions using H2 as reactant in fixed-bed reactor.

Reaction 9 H0298 K[a]

R Equilibrium Reaction rate equation EA Product selectivity Catalyst deactivation
[kJ mol1] with T [kJ mol1]
methanol 128.1 (CO), at high T p
pCO pH MeOH 75100 high, MeOH, > 99.9 % sintering at T @ 250 8C at high concentra-
r 2 Kp
f pH2 O
86.9 (CO2) tions of CO and/or CO2

DME[c] 144[d] at high H2O partial pressures[e]

51.3 (MeOH) at high T pMeOH 2
high DME
r f pH2 O

MTG[f ] 55.7 (MeOH) no pCO pH

r f pCO ;pH2 O 3590 complex, T sensitive[g] coking
methanation[h] 206.2 (CO), at high T 240 CH4 and CO2 (CO), high coke formation (CO), Ni-carbonyl (CO)[j]

165.1 (CO2) pCO pH

r f pCO2 244[i] CH4, > 99 %
LTFT 115.1 (CO)[k] no 93103[l] complex, T sensitive[g] at high CO conversion[m]
CO2 hydroge- 125.0 (CO2)[n] complex, T sensitive[g]
pCO pH O
no pCO pH 2
139, 95 at high CO2 conversion (high (pH2 O /pH2 )
r1 2 2
f pCO2 ; pCO; pH2 O
nation pCO pH
to C2C4 hydro- r2 f pCO ; pCO; pH O


[a] Ref. [2]. [b] Ref. [35]. [c] Dimethylether. [d] Ref. [6]. [e] Ref. [7]. [f ] Ref. [810]. [g] Mixture of hydrocarbons. [h] Ref. [11, 12]. [i] Ref. [13]. [j] Ref. [14]. [k] As-
suming C10H22 as final product. [l] Ref. [15, 16]. [m] Ref. [4, 17]. [n] Assuming C3H8 as final product.

Table 2. Qualitative evaluation of the feasibility of synthesis reactors to be operated under flexible conditions (with respect to reactor engineering aspects).

Reaction/Synthesis Feasibility Main features

methanol high efficient heat transfer required, high selectivity to the desired product
DME high low heat of reaction and high selectivity to the desired product
MTG moderate complex product selectivity and catalyst deactivation due to coking
methanation high efficient heat transfer required, high selectivity to the desired product
LTFT moderate complex product selectivity and CO conversion limited, efficient heat transfer required
CO2-hydrogenation to C2C4 hydrocarbons moderate complex product selectivity and CO2 conversion limited, efficient heat transfer required

operation. The main reasons for this are the complex product C2C4 hydrocarbons for heating value adjustment of substi-
distribution and the sensitive catalyst stability. tute natural gas (SNG)]; and (ii) develop methods to study
Flexible operation of catalytic chemical reactors is a field transient reactor operation based on a combination of exper-
of research where a lot of aspects may play an important imental work and conceptual studies using mathematical
role, with limited knowledge available at present (e.g., reac- models.
tor design, catalyst and material behavior, control strategies, The approach followed to evaluate the restrictions that
heat integration, etc.). In this context, the present study has catalyst and industrial reactor design could impose on the at-
the following aims: (i) identify general aspects and limita- tainable reactor flexibility is shown in Figure 2. The methods
tions of flexible reactor operation, using an example of a rep- developed combine experimental work in a lab-scale setup
resentative catalytic fuel synthesis reaction and a suitable with mathematical modelling of the lab-scale setup and an
fixed-bed reactor [example reaction: CO2 hydrogenation to industrial-scale reactor. The combination of mathematical

Figure 2. Approach followed to evaluate the flexibility of an industrial-scale fixed-bed reactor.

Energy Technol. 2016, 4, 90 103 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 91
models and experimental work is a tool to evaluate the po- CO2 shift : CO2 H2 CO H2 O
tential to operate chemical reactions in a flexible way. Math- 1
9 R H 0300 C 39:7 kJ mol1
ematical models offer the possibility to analyze flexible oper-
ation of a chemical reactor, with representative parameters
obtained by experimental work (e.g., kinetic parameters, cat- Fischer Tropsch : CO 7=3 H2 ! 1=3 C3 H8 H2 O
alyst behavior under transient operating conditions). 9 R H 0300 C 174:2 kJ mol1
In particular, the present study addresses the preliminary
design of an industrial-scale fixed-bed reactor able to operate In the temperature range 245300 8C CO2 conversion and
under flexible load conditions (i.e., within a load range hydrocarbon yield increase with temperature at a relative
always satisfying that conversion and selectivity are accord- constant CO yield. However, at temperatures above 300 8C it
ing to specifications) based on experimental data collected is not possible to achieve stable conditions (Figure 3). An in-
before and on steady-state and non-steady-state balance
equations. Previous work on the same reaction has already
been published by the authors: Iglesias-Gonzlez et al.[18]
summarized experimental work carried out in lab-scale
fixed-bed reactors to determine the set of kinetic parameters
suitable for steady-state conditions and the effect of operat-
ing conditions on catalyst activity and selectivity. The same
lab-scale setup was used to investigate transient operation,
with special focus on catalyst behavior. Different methods
were developed to determine whether the catalyst activity
and/or selectivity vary due to variations in the reaction mix-
ture composition and/or temperature. Results are presented
in Ref. [19].

Example reaction: CO2 hydrogenation to C2C4 hydrocarbon Figure 3. Effect of temperature over time in CO2 conversion (XCO2 ), CO yield
synfuels (YCO), and yield of organic components (YCorg ) at tmod = 1000 kg s m3 for
100 g Fe/2 g K as catalyst.
Table 3 summarizes the main characteristics of the CO2 hy-
drogenation to gaseous hydrocarbons considered in this
study as an example fuel synthesis reaction. Most relevant crease in temperature leads to a loss of FT activity that could
parameters for reactor design are the heat of reaction and be related to the presence of a high H2O partial pressure due
the deactivation of the catalyst in the presence of products to the high CO2 conversion level achieved.[20, 21] In the pres-
[(pH2 O /pH2 )out > 0.1]. Three main differences appear in com- ence of H2O, iron carbides can be transformed to iron
parison with LTFT: (i) the liquid product is absent, (ii) the oxides, with no activity for the FT synthesis; therefore, CO
heat of reaction is lower due to the endothermic CO2-shift yield increases.
reaction, and (iii) the partial pressure of H2O is higher as it is Figure 4 shows that the ratio (pH2 O /pH2 )out correlates with
a by-product of the CO2-shift [Equation (1)] and FT reac- CO2 conversion and hydrocarbon selectivity. Curves shown
tions [Equation (2)]. The complex hydrocarbon product dis- in Figure 4 are calculated based on stoichiometry for the dif-
tribution typical of the FT synthesis is not taken into account ferent H2/CO2 inlet ratios and are helpful for orientation.
for the calculations and all hydrocarbons formed are as- Due to the characteristically high FT activity of the 100 g Fe/
sumed to be propane C3H8. 2 g K catalyst, the yield of organic products is very close to

Table 3. Main characteristics of the CO2 hydrogenation reaction to produce C2C4 hydrocarbon synfuels. Set of kinetic parameters and catalyst restrictions.[18]

Parameter Characteristics
heat of reaction 9 RH0300 8C,1 = 39.7 kJ mol1; 9 RH0300 8C,2 = 174.2 kJ mol1
catalyst 100 g Fe/2 g K
catalyst restrictions limited range of conditions pH2 O /pH2 0.10, Tmax = 300 8C
reaction kinetics pCO pH
r1 k0;1 e RT pCO a1;H2 O p2 H O b
E A;1 p CO H
K p

2 1;CO pCO 2 2 2

r2 k0;2 e
EA;2 pCO pH
RT 2
pCO a2;H2 O pH2 O b2;CO2 pCO2

activation energy EA, 1 = 138.9 kJ mol1, EA, 2 = 94.6 kJ mol1

pre-exponential factor k0,1 = 1.6106 mol s1 kg1 Pa1, k0,2 = 2.6103 mol s1 kg1 Pa1
inhibition terms a1, H2 O = 128, b1, CO2 = 0.8; a2, H2 O = 152, b2, CO2 = 0.5

Energy Technol. 2016, 4, 90 103 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 92
Figure 4. Influence of (pH2 O /pH2 )out on activity and selectivity in C %. Left: CO2 conversion (XCO2 ), CO yield (YCO) as a function of (pH2 O /pH2 )out. Lines calculated
based on stoichiometry to guide the eye. Solid curve for (H2/CO2)in = 4, dashed curve for (H2/CO2)in = 8 and dashed-dot line for (H2/CO2)in = 22. Right: hydro-
carbon product distribution in weight % as a function of (pH2 O /pH2 )out. Full symbols gray: deactivation observed. White symbols: stable operation. Conditions:
p = 12 MPa; (H2/CO2)in = 4, 5, 8, and 22; T = 245300 8C; and tmod = 1755000 kg s m3.

the CO2 conversion and the partial pressure of H2O increases Results and Discussion
proportionally to CO2 conversion. Gray symbols indicate
that deactivation was observed over time with the same A catalytic fixed-bed reactor is proposed for the hydrogena-
trends as shown in Figure 3. tion of CO2 to gaseous hydrocarbons on large scale under
The (pH2 O /pH2 )out ratio may be responsible for the apparent flexible load conditions. Catalytic fixed-bed reactors are the
impossibility to reach high CO2 conversion values with the most important type of reactors in the chemical industry, pet-
100 g Fe/2 g K catalyst. Based on this finding it is suggested rochemical industry, and petroleum refining.[23] Their main
that there is possibly a limiting value of (pH2 O /pH2 )out, which advantages include simplicity, low cost, minimal mainte-
depends on catalyst composition, temperature, and reaction nance, high pressure operation, and easy scaleup. Long cata-
conditions, at which the composition of the catalyst could lyst life ( 3 months) is required because catalyst replace-
start changing, consequently affecting its activity and selec- ment is an expensive and tedious affair. One of the most crit-
tivity.[19] An analogous suggestion regarding the interaction ical points of fixed-bed reactors is temperature control for
between the gas phase and catalyst composition was present- high exothermic reactions.
ed earlier by Hjlund-Nielsen and Bgild-Hansen.[22] High The adiabatic temperature increase is a decisive parameter
values of the (pH2 O /pH2 )out ratio indicate a high oxidation po- to select between adiabatic or non-adiabatic operation. In
tential of the gas mixture. The molar feed ratio (H2/CO2)in is the case of the CO2 hydrogenation for a stoichiometric feed
the most promising variable for decreasing the oxidation po- (H2/CO2)in of 10:3 at p = 2 MPa and T = 280 8C 9 Tad is
tential of the gas mixture to achieve high CO2 conversion around 1000 K. Reactions with a 9 Tad 1000 K are consid-
and hydrocarbon yield without catalyst deactivation. Indeed, ered to be strongly exothermic. The maximum 9 T allowed is
with a ratio (H2/CO2)in = 22, CO2 conversion values above 50 8C (from 250 to 300 8C). At temperatures lower than
70 % are possible without catalyst deactivation and a hydro- 250 8C the CO2-shift rate is very slow, and at temperatures
carbon product mainly based on methane is obtained higher than 300 8C catalyst deactivation occurs. With such
(approx. 80 %; T = 290 8C, tmod = 1000 kg s m3, p = 2 MPa). a small temperature difference allowed, 20 stages (catalyst
Also the hydrocarbon selectivity is affected by the (pH2 O / beds) will be needed for this reaction to take place adiabati-
pH2 )out ratio (Figure 4, right). At high (pH2 O /pH2 )out ratios the cally assuming interstage cooling. This seems to be uneco-
mass fraction of methane decreases and C2C4 and C5 + frac- nomical (typical adiabatic reactors have between 3 and
tions increase, probably due to the reduced H2 availability 5 stages).
and the higher concentration of H2O. This ratio seems to A non-adiabatic reactor with cooling medium is selected
have a stronger effect than temperature, modified residence to guarantee a good temperature control and heat manage-
time, or pressure on hydrocarbon product distribution for the ment (Figure 5). The selected cooling medium is a heat trans-
100 g Fe/2 g K catalyst. fer oil.[24] Boiling water cannot be used due to the higher
pressures required to reach temperatures of around 280
300 8C. The temperature of the heat carrier medium needs to
be close to the desired catalyst temperature. Large heat

Energy Technol. 2016, 4, 90 103 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 93
Figure 5. Simplified representation of a multitubular reactor for an exothermic synthesis reaction, a reactor tube, and an eggshell catalyst particle.

transfer areas per catalyst volume and sufficiently large mass

flow velocity of the reaction gases are important to ensure
good heat transport from the packing to the heat-exchange
Commercial LTFT multitubular reactor dimensions are
considered as guidance for the design of an industrial-scale
reactor for CO2 hydrogenation despite the difference in the
chemical reaction characteristics. LTFT reactors are operated
under steady-state conditions using synthesis gas (mixture of Figure 6. Recycle reactor used for CO2 hydrogenation. Separation units as-
H2 and CO). However, the design of a reactor is dominated sumed as an ideal system. CO2 is produced in a biogas plant and H2 from
H2O electrolysis using renewable energy. Fv : volumetric flow.
by similar features as in the case of CO2 hydrogenation: heat
removal, product selectivity, and catalyst deactivation in par-
ticular.[4] Commercial HTFT reactors, operated at tempera-
tures around 340 8C, are fluidized beds. Fixed-bed reactors The configuration of a recycle reactor (Figure 6) appears
are not recommended at higher temperatures due to the high to be most attractive to carry out hydrogenation of CO2 to
tendency of carbon deposition on the catalyst (CO partial gaseous hydrocarbons under flexible load. In this reactor,
pressure is high for the Boudouard reaction). In the case of product separation and recycling of H2 and CO2 together
CO2 hydrogenation and for the selected catalyst 100 g Fe/ with the intermediate product CO is considered. This config-
2 g K, deactivation is not due to carbon deposition but due to uration has an extra design variable available, the conversion
a high (pH2 O /pH2 )out ratio. Therefore, fixed-bed reactors per pass or recycle ratio, which is advantageous for variable-
remain an alternative for CO2 hydrogenation using 100 g Fe/ load operation.
2 g K. Moreover, the use of a recycle allows operating the reactor
Low catalyst stability at high pH2 O is a common issue in FT under more favorable conditions compared with a one-pass
synthesis; therefore, for LTFT synthesis, single-pass CO con- reactor: (i) high average reaction rates along the catalyst bed
version is limited to about 5070 %. To reach higher overall can be achieved due to the presence of CO (intermediate
conversions, two or three reactors in series can be used with product) at the reactor inlet and a lower partial pressure of
interstage H2O separation. Another option is internal recy- H2O, (ii) heat removal is more effective due to the higher gas
cling of the gases leaving the FT synthesis reactor after sepa- velocities, (iii) overall high conversion is possible, and
ration (e.g., the ARGE reactor, defined in the abbreviation (iv) the (H2/CO2)in ratio can be adjusted to values above the
list, operates with a recycled/fresh gas volume ratio of stoichiometric ratio. The (H2/CO2)in ratio is the only opera-
1.9).[4, 17] These two alternative configurations allow the con- tional variable that has the potential to enable high CO2 con-
trol of the partial pressures in the catalytic reactor and are version and hydrocarbon yield at low (pH2 O /pH2 )out ratios
also appealing options for carrying out CO2 hydrogenation (Figure 4). The disadvantages of recycle reactors are, howev-
under non-steady-state conditions. The use of membranes to er, related to high operational costs (i.e., high energy
remove H2O in situ under FT conditions have already been demand), large separation units, need of a recycle gas com-
considered in detail by Rohde et al.[25] and Rohde.[26] The au- pressor, repeated heating and cooling, and loss of material
thors concluded that currently available membranes exhibit due to purging.
low stability, permselectivities, and permeances regarding A crucial point of this configuration might be the behavior
H2O separation under FT conditions. Hence, they are not an of the separation units under variable load conditions. The
option at the moment. separation of H2, CO2, and CO from H2O and hydrocarbons

Energy Technol. 2016, 4, 90 103 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 94
is not an easy task: the reactor outlet flow needs to be Example design study: Flexible load (70100 %)
cooled to condense reaction water and higher hydrocarbon
products (in case they are produced). Afterwards, CO2 is sep- The procedure shown in Figure 10 is applied to design an ex-
arated in a CO2 scrubber unit to avoid freezing up the subse- ample multitubular fixed-bed reactor including recycling to
quent cryogenic unit with solid CO2. The gas is decom- be operated between 70100 % load for a fixed tube diame-
pressed, and the water vapor content of the gas is reduced to ter of 3 cm. Selected values for the assumed design variables
70 8C dew point (dehydration). Cryogenic cooling and sepa- are shown in Table 4. The desired minimal overall conversion
ration are needed to separate H2 from light hydrocarbons.[17] of CO2 is 90 % provided that the temperature is lower than
Having chosen the conversion per pass, the recycle ratio is 300 8C and the maximum value of the (pH2 O /pH2 )out ratio is
an immediate consequence to reach the required overall con- below 0.1. The selected basis for calculation is Fv, CO2 , average =
version. When designing reactors for one load operation, 500 m3 h1 (STP), according to the CO2 typically produced in
there is an economical optimum for the conversion per pass a biogas plant. The thickness of active material cshell is chosen
that balances reactor size against the combination of product in such a way that internal mass transfer limitations are
separation load and recycle ratio. However, if the reactor is avoided. The active density (1active, R = 214.1 kg m3) and parti-
operated under variable load conditions and the overall con- cle conductivity (lp = 5.6 W m1 K1) are a consequence of
version needs to be maintained constant, the recycle ratio the selected values for cshell and particle diameter (dp).
will be a consequence of the operation load due to the con- Figure 7 shows the calculated total molar flows of the
stant size of the reactor. Flexibility is obtained for the reac- chemical components along the catalyst bed at minimum and
tor through the recycle ratio. maximum operation load. The molar flow Fn of CO is almost
A potential alternative or complementary strategy to constant along the catalyst bed, implying that the rate of CO
obtain flexibility for variations in load could be to modify formation through the CO2 shift reaction equals the rate of
the reactor temperature (as manipulated variable), with consumption through the FT reaction. Operation at mini-
higher temperatures at high volumetric flows and lower tem- mum load leads to a conversion per pass of 15 % and conse-
peratures at low volumetric flows. The aim is to modify the quently a recycle ratio RF = 5.55 (where the F designation
reaction rate according to the residence time of the gas in indicates fresh, as indicated in the definitions below). The
the reactor (load), trying to keep the conversion per pass overall conversion is 93 % and Rout = 0.988. The partial pres-
almost constant; thus, variations on the recycle ratio between sure restriction, max (pH2 O /pH2 )out = 0.1, is not passed at the
full and part load operation are reduced. The rate of temper- reactor outlet (Lr = 3.14 m). To achieve an overall conversion
ature change of the cooling medium needs to be adjusted to of 90 % when this reactor is operated at maximum load, the
the rate of load change. Temperature adjustment to maintain average reaction rate along the catalyst bed has to increase
a constant conversion is a strategy used in industrial practice from the minimum operating load (at maximum operation
when a catalyst is deactivated during the process (e.g., in cat- load the reaction rate is about 40 % higher than at minimum
alytic reforming of naphtha, where the catalyst is slowly de- operation load, rFT, Lmax = rCO2 shift, Lmax = 3.12 mol m3 s1 versus
activated by coke deposition).[27] rFT, Lmin = rCO2 shift, Lmin = 2.24 mol m3 s1). However, the resi-
The use of reactors in series with interstage H2O separa- dence time of the gas in the catalyst bed decreases, leading
tion is attractive for one-load operation (steady state), but to a decrease in the conversion per pass (from 15 % to 11 %)
the evaluation of its potential for variable-load operation is and an increase in the recycle ratio (RF = 7.56). The pressure
out of the scope of the present work. A high number of reac- drop increases approximately threefold compared with oper-
tors is required to achieve a high flexibility. However, the ation at minimum load due to the higher gas velocity along
possibility to remove only H2O from the product and feed the catalyst bed. The electric power required for the recycle
the gaseous products together with the unconverted H2, CO, compressor increases approximately sixfold (Pel, R compressor =
and CO2 into the next reactor could reduce separation costs. 23.83 kW at Lmax and Pel, R compressor = 3.68 kW at Lmin) due to
An attractive balance between number of reactors and sepa- the recycle flow being twice as high and the pressure drop
ration units might be possible. When the reactors in series thrice as high.
are operated under minimum load, only gaseous hydrocar- Figure 8 left shows the temperature profiles along the cata-
bon products are present in the last reactors. These gases lyst bed for 100 % and 70 % load situations. The hot spot at
need to be inert when in contact with the catalyst. Experi- minimum load reaches 300 8C; however, when the reactor is
mental work is required to prove this requirement. operated at maximum load, the maximum temperature of

Table 4. Operating conditions, reactor dimensions, and catalyst particle design for operation between 70 and 100 % load. Gray columns: result of the calcu-
lations following procedure shown in Figure 10. White columns: values selected for calculations.

Operating conditions Reactor dimensions Catalyst particle properties

p Tin (H2/CO2)in dt Lr NT mcat Vr e cshell dp 1s cps
[MPa] [8C] [] [m] [m] [] [t] [m3] [] [Lm] [m] [kg m3] [J kg1 K1]
2 280 10/3 0.03 3.14 956 0.45 2.12 0.4 150 0.003 1316 1100

Energy Technol. 2016, 4, 90 103 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 95
Figure 7. Total molar flows of reactants and products along the catalyst bed for maximum and minimum operation load (10070 %). Details for recycle reactor
and operating variables: see Table 4.

the hot spot decreases to 297 8C. The reason for a lower hot- the mass of active catalyst is determined for the minimum
spot temperature at maximum load is the higher overall heat fresh flow. The maximum CO2 flow that can be converted
transfer coefficient caused by a higher gas velocity. per pass is calculated by multiplying the maximum reaction
An increase in reactor flexibility (decrease in Lmin) with rate (rr) by the available mass of active catalyst (Fn, CO2 , max, rr ).
a constant tube diameter requires a lower number of tubes, In the case that the molar flow of CO2 in the fresh feed is
therefore a smaller reactor vessel but a higher power higher than Fn, CO2 , max, rr, it would not be possible for the reac-
demand for the compressor when the reactor is operated at tor to reach the desired overall conversion (maximum reac-
maximum load. tion rate corresponds to Xi,pp = 0 and RF = 1) and, conse-
The flexibility of the reactor is directly connected with the quently, the reactor cannot be operated at this Lmin value.
characteristics of the chemical reaction. The maximum CO2 For the reaction kinetics used in the present study and an
molar flow in the fresh feed that can be converted is a func- average CO2 flow of 500 m3 h1, Lmin values lower than 0.05
tion of the maximum reaction rate and the mass of active cannot be used because the required minimum overall con-
catalyst available (kinetic limitation). version for maximum load cannot be achieved.
In the case of CO2 hydrogenation, the maximum reaction Even though the reaction rates for CO2 hydrogenation are
rate is at z = 0 (where no H2O is present). For a certain Lmin, very attractive for flexible operation, the low value of the

Figure 8. Calculated axial profiles of temperature and ratio (Fn, H2 O /Fn, H2 ) along the catalyst bed at different times after a change in reactor load, that is, fresh
feed flow with a rate of 5 % min1 from 100 % load to 70 % load. Gray area: region not acceptable for catalyst stability reasons (see Figure 3 and Figure 4).

Energy Technol. 2016, 4, 90 103 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 96
maximum allowed conversion per pass leads to high energy flow requires a lower RF. Assuming Rout as constant, the over-
requirements for recycle compression (Xi,pp = 14.7 % under all conversion per pass increases almost linearly. The molar
the selected operating conditions). With a reactor designed flow of unconverted CO2 decreases when the reactor is oper-
for Lmin = 0.5, the conversion per pass at the maximum load ated at minimum load Fv,CO2 ,out *(1Rout).
is around 3 % and the required recycle ratio is approximately
Characteristic times
Design variables have a strong effect on reactor flexibility
and on the resulting energy demand. Using the design varia- A chemical plant is an example of a complex, high-order,
bles presented herein (Table 4), only a 3 cm tube diameter nonlinear process that can be described using a group of
can be used to reach a value of Lmin = 0.5. As a consequence, time constants. Processes involving chemical reactions, mass
changing design variables offers the possibility to obtain al- transfer, and heat transfer take place on different time and
ternative recycle system designs for lower Lmin values. length scales (i.e., catalyst particle, reactor, and attached
equipment). The characteristic time t is a measure of the
time it takes to observe a significant change in a given pro-
Simulation of example load change in a recycle reactor
cess resulting from an input disturbance [Eq. (3)].
The reactor conceptually designed to be operated at 100
70 % load is used in this section to describe the transient op- capacity
t 3
eration from maximum to minimum load (from 100 % to rate
70 %) for a selected load-change rate of 5 % min1. The load-
Comparing the characteristic times of different processes
change rate for the chemical reactor is related to the load-
with the characteristic time of the inlet disturbance might
change rate of H2 production and the size of the H2 storage
help to significantly simplify the number of processes that
devices. High rates might be unsafe for the equipment at-
need to be considered in the mathematical model describing
tached to the reactor (i.e., heat exchangers, compressors,
the process. Processes that occur on much shorter time scales
etc.). Rates of 1015 % min1 are recommended by some
(small value of the time constant) than the characteristic
companies working on process design and construction.[28]
time of the inlet disturbance are assumed to be in equilibri-
The value assumed for the calculation presented herein is, in
um or in steady state (i.e., transient portion of the change is
comparison, conservative.
negligible). On the other hand, processes with characteristic
Transition from one operation point to the other does not
times considerably longer than the characteristic time of the
seem to be a problem for a fixed-bed reactor if the reactor is
inlet disturbance (large value of the time constant) are con-
properly designed according to maximum and minimum flow
sidered as being time invariant.
and catalyst limitations are considered [e.g., Tmax = 300 8C,
max. (pH2 O /pH2 )out = 0.1]. Temperature and (pH2 O /pH2 )out limi- In a future flexible chemical plant, characteristic times for
tations are never exceeded (Figure 8). The transient tempera- variations in inlet conditions might vary between minutes
ture profile is strongly influenced by the catalyst density and (102 s) and days (105106 s). Processes with a similar charac-
heat capacity of the catalyst, both of which are responsible teristic time as the inlet variation will operate under transient
for the dynamic thermal behavior of the reactor bed. Be- conditions. Figure 9 shows the characteristic times of the pro-
cause a relatively low rate of change (5 % min1) is used, cesses taking place in the catalyst particle and in the chemi-
changes in the temperature, concentration, etc. occur evenly cal reactor discussed in the previous section. Characteristic
along the catalyst bed. times are evaluated for the maximum (gray) and minimum
The dynamic behavior of the fresh feed flow can be seen (white) operation load, in both cases along the catalyst bed
in the recycle flow but with some delay. At the very begin- (for z = 0 and for z = L; Figure 9). A definition of the charac-
ning of the load change, the amount of fresh feed decreases, teristic times and average characteristic time values can be
which is mixed with the steady-state recycle flow. The change seen in Table 5.
in the reactor inlet flow is small but enough to slightly
change the conversion per pass and, consequently, the recy-
Characteristic times for processes in the catalyst particle
cle flow and hence, the inlet flow to the reactor. It is as-
sumed for simplicity that no dead time caused by the separa- The characteristic times for the processes taking place in the
tion units exists in the recirculation. The gas coming out of catalyst particle are formulated based on a heterogeneous
the reactor at t = t1, in the case of being recycled, will be fed balance of the catalyst particle positioned in a flowing
into the reactor at t = t1 + 9 t. In reality, the time required for stream (the gas phase inside the particle is neglected). Tem-
separation and recirculation is expected to be significant; perature and concentration are considered in three different
however, without knowing the design of the separation units regions: the bulk fluid phase, the boundary layer surrounding
it is not possible to make a good guess. The recycle flow the particle, and the particle itself. Variations at the inlet of
reaches its steady state after 420 s whereas the fresh feed the reactor propagate through the fluid and may change the
flow is constant after 366 s. conditions at the outer edge of the boundary layer and, con-
A minimum overall conversion of 90 % is always satisfied. sequently, in the boundary layer and the particle itself. Most
An increase in the conversion per pass due to the lower feed of the characteristic times with regard to the catalyst particle

Energy Technol. 2016, 4, 90 103 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 97
long-term behavior, making faster processes insignificant by
comparison. According to the calculated values, the mass
and heat transfer processes taking place in the catalyst parti-
cle are faster than the chemical reaction rate (with the ex-
ception of the first 10 cm in the fixed bed where the external
heat transfer rate is slower than the chemical reaction). This
means that mass and heat transport on the catalyst particle
do not need to be considered because they are fast in com-
parison to the chemical reaction rate (pseudohomogeneous
Internal mass and heat transport processes are independ-
ent of the operational load and depend mostly on the charac-
teristics of the catalyst particle (e.g., size of the pores, type
of pores, particle composition). Internal mass transport limi-
tations can be, to a large extent, reduced with the use of
eggshell catalysts. External mass and heat transfer depend
directly on the gas velocity; it might occur that due to
a change in load, the process becomes controlled by mass/
Figure 9. Characteristic times for maximum (gray, 100 %) and minimum load
(white, 70 %) at the catalyst bed for z = 0 and z = L calculated for the example heat transfer rather than by the chemical reaction.
reactor dimensions shown in Table 4 and definitions in Table 5. Region for at- The characteristic times for external and internal heat
tached equipment based on Lerou and Ng.[29] Large gray area: expected char- transport in a catalyst particle connect the accumulation
acteristic times for variation in inlet conditions (assumption). term of the catalyst particle with the heat-transfer coefficient
between gas and particle and the effective heat conductivity
have already been defined in the literature[30, 31] or are used inside the catalyst particle, respectively.[32]
as dimensionless groups to evaluate the necessity to use Due to variations in the operating conditions (i.e., partial
a pseudohomogeneous or heterogeneous model. To calculate pressures, temperature, or other parameters) changes in the
the transfer coefficients, a representative position, tempera- catalytic reaction might occur (i.e., modification of the reac-
ture, pressure, gas velocity, and reaction rate need to be se- tion mechanism, accumulation of species on the catalyst sur-
lected to write the material and energy balance as dimen- face, or a change on the catalyst composition), resulting in
sionless. different catalyst activities and/or selectivities.[30] Experi-
The catalyst particle involves processes taking place in ments within the range of conditions where the steady-state
series (i.e., external mass transfer, internal mass transfer, and kinetics are determined are presented together with experi-
chemical reaction); the slowest process would dominate the ments under extreme conditions in Iglesias et al.[19] A charac-

Table 5. Characteristic times definition for the different processes taking place in the catalyst pellet and in the reactor. Example values calculated for maxi-
mum and minimum operation load of the recycle reactor considering the CO2 hydrogenation as reaction. Characteristic times calculated using average
values along the catalyst bed.

Process Definition Characteristic time [s1]

100 % load 70 % load
e VR
Reactor processes Fv;in

convective mass transport 1.4 2.7

e 1g cp;g 1e1s cp;s
reactor cooling 10.3 14.4

e 1g cp;g 1e1s cp;s

convective heat transport 469.3 888.9
1g cp;g uL

e 1g cp;g 1e1s cp;s

reactor heat-up by chemical reaction 1118.2 1555.9
9 HR Tr i

Catalyst particle processes e Ci

chemical reaction ri
11.9 16.2
external mass transport 1
0.1 0.1
internal mass transport 1 0.1 0.1

1s cp;s
external heat transport ag
2.2 3.1
1s cp;s
internal heat transport lp 2.3 2.3

Energy Technol. 2016, 4, 90 103 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 98
teristic time for changes of catalyst properties can be deter- cooling and reaction it is possible to understand the pres-
mined from experiments, for example, a change of the gas at- ence of hot spots. If the time constant of the reaction is
mosphere from pure H2 to synthesis gas (H2 + CO2), carried lower than that of the reactor cooling, the cooling system
out in a magnetometer. A characteristic time for the result- is not able to remove the heat, produced faster, and hot
ing catalyst formation (iron carbides formation for the FT spots appear (e.g., close to the reactor inlet). On the con-
synthesis) can be calculated based on experimental results trary, if the time constant of the cooling is equal or lower
assuming a first-order process and a step change. The charac- than the time constant of the reaction, the reactor remains
teristic time is calculated as 63.2 % of the total time required isothermic (e.g., close to the reactor outlet).
for the change (tcatalyst = 15.6 h = 5.6104 s).[33] Even though
this process is not very relevant for the chemical plant (it
would only occur at the very beginning or in case of a shut
Characteristic times for processes in the equipment attached to
off or regeneration) it is included as reference in Figure 9.
the chemical reactor
Within the range where the kinetic parameters where deter-
mined, no change in the activity of the catalyst is observed. The characteristic times for the catalyst particle and the
Therefore, it is not possible to determine a characteristic chemical reactor need to be compared with the characteristic
time for changes in the catalyst properties. times of the other equipment units taking part in the chemi-
cal process (i.e., separation units, compression, pipes, heat
exchangers, etc.). Separation units, in the case of a reactor
Characteristic times for processes in the catalytic fixed-bed reac-
with recycle, might be accompanied by long characteristic
times due to its significant size. An alternative could be to
Characteristic times for the one-dimensional fixed-bed reac- use a higher number of smaller units or advanced control
tor are derived from the dimensionless non-steady-state mass strategies; however, both alternatives may also be accompa-
and energy balance. The selection of the dimensionless varia- nied with higher investment costs. It is important to keep in
bles for concentration and temperature can be achieved mind that the increase in costs due to flexible operation of
using the inlet, the average, or the outlet concentration/tem- the chemical plant needs to be less than the reduction on the
perature. The same is applicable for the gas velocity. The dy- investment for H2 storage. If expected characteristic times
namics of the momentum balance in the reactor are neglect- for variations in inlet conditions are between minutes (102 s)
ed because they are much faster than the composition and and days (105106 s), the dynamic behavior of the overall
temperature dynamics (range 103102 s, calculated using chemical plant might depend on the characteristic time of
the speed of sound). the attached equipment as long as the catalytic reactor is op-
erated within the range of conditions where the catalyst is
* The characteristic time of the mass transfer by convection stable. The detailed evaluation of the attached equipment is
in the reactor with recycle is relatively short due to the outside the scope of this work and, therefore, a box without
short catalyst tubes (Lr = 3.14 m) and the high gas veloci- values and a question mark is used in Figure 9.
ties used in the recycle reactor.
* The characteristic time of the reaction varies strongly
along the catalyst bed due to the production of H2O and
resulting decrease in the reaction rate. The ratio between The following conclusions can be drawn from the present
the characteristic times of convective mass transport and study:
chemical reaction equals the conversion per pass. Due to
the short characteristic time of the convective transport, * A procedure has been developed for the design of a recy-
the conversion per pass is low (1115 %). High conver- cle system that can be used to operate a catalytic fixed-
sions are not allowed to avoid catalyst deactivation. bed reactor for an exothermic and temperature-sensitive
* The characteristic time of convective heat transport is di- synthesis under variable load. The characteristics of the
rectly related to the gas velocity. In this case the differ- recycle system are determined by specific limitations of
ence between the characteristic time at maximum and the catalyst system and the example reaction (hydrogena-
minimum load can be clearly observed (Figure 9). tion of CO2 on Fe), for example, the sensitivity of the cat-
* The characteristic time of the heat produced by chemical alyst towards partial pressure of products and tempera-
reaction is the longest. Heat processes in the fixed-bed re- ture. Simulation results of transient reactor operation,
actor are much slower than mass-transfer processes due after changing the load, do not show complications if the
to the difference in the storage terms. reactor is designed consistently with regard to the con-
* The characteristic time of reactor cooling remains almost straints of the chemical reaction. However, information
constant along the reactor tube. The heat-transfer coeffi- about reactor-attached equipment on the dynamic re-
cient is affected by gas velocity; however, due to the low sponse of the system has not been considered. According
conversion per pass, the volume contraction by chemical to the experiments carried out on the lab scale, the at-
reaction is relatively low and variations in gas velocity are tached equipment plays an important role on the dynamic
not significant. By comparing the characteristic times of response of the system, which is also expected for indus-

Energy Technol. 2016, 4, 90 103 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 99
trial-scale processes. Sensitivity analysis is required to Ergun.[27] Pressure was considered to change only in axial direc-
evaluate the effects of operating conditions and design tion, and variations in radial direction were negligible due to the
variables, for example, Tin = Tcooling, catalyst loading, thick- assumption of plug flow. Variation of pressure with time was not
considered because it was assumed to be much faster than
ness of active material cshell, (H2/CO2)in ratio, particle di-
changes in temperature or composition. After a change in inlet
ameter dp, and selection of the catalyst form/shape. Fur-
flow, it was assumed that the outlet flow changed at the same
ther investigation regarding design of the attached equip- time as the pressure profile along the catalyst bed. The gas veloc-
ment is recommended also in connection with investment ity changed along the catalyst bed due to the volume reduction
costs. produced by the chemical reaction (according to the stoichiome-
* According to the presented calculations, flexible opera- try of the reaction 54 % volume reduction should occur on com-
tion would lead to an increase in the size of the reactor plete conversion). For calculation, a global mass balance of the
vessel (higher investment cost) and/or in the power gas phase considering all components was carried out.
The mathematical model consisted of a mixture of ordinary and
demand for the recycle compressor (higher operating
partial differential equations. In total, eight equations needed to
cost) compared with reactors designed for steady-state op-
be solved simultaneously: 5 material balances [Eq. (4)] formulat-
eration. The cost increase caused by the flexible operation ed for each component (these equations are related through the
of the chemical process needs to be lower than the reduc- stoichiometric coefficients of the reaction), 1 energy balance
tion in investment cost for H2 storage. Economical evalua- [Eq. (5)], 1 pressure drop equation, and 1 global mass balance of
tions are required to assess the potential of flexible opera- the gas phase. Temperature, concentrations, pressure, and gas ve-
tion of chemical reactors. locity were the dependent variables. They are functions of the
* The concept of characteristic times can be used to show, two independent variables: the axial coordinate z and time t.
in a simplified way, the effect that individual processes in- The solution was in the form ci = f (z, t), T = f (z, t), p = f (z), and
usf = f (z) with usf, z = 0 = f (t).
volved in a chemical plant have on the dynamic response
The method of lines (MOL), explained in detail by Schiesser and
of the system. Griffiths,[34] was used to replace the derivatives with algebraic ap-
* The methods developed herein can be used to investigate proximations, for example finite differences, so that only one var-
the dynamic behavior of catalytic synthesis reactors in any iable (typically time) was left. The partial differential equations
fuel-related process. were transformed into a system of ordinary differential equations
(ODEs), which were solved using finite differences. The set of
differential equations was solved at each grid point. The reactor
length was divided in axial direction. The size of the grid (discre-
Computational methods tization) was a compromise between the degree of accuracy and
the computational work. This compromise is especially important
Catalyst tube model equations for transient calculations, where the reactor profiles are calculat-
ed for each time step. Values for 9 z = 0.001 m and for 9 t = 1 s
The phenomena taking place in a reactor tube filled with catalyst
were typically used.
particles can be described by the combination of material bal-
For steady-state calculations, the accumulation term was set to
ance [Eq. (4)], energy balance [Eq. (5)], and kinetic rate expres-
zero and the ordinary differential equations were integrated
sions (Table 3). These mathematical expressions do not distin-
along the catalyst bed. Boundary conditions were defined at the
guish between conditions in the fluid and the catalyst (pseudoho-
inlet of the reactor (at z = 0): concentrations ci,z = 0, temperature
mogeneous); however, the catalyst contributes significantly to
Tz = 0, pressure pz = 0, and gas velocity usf,z = 0.
the accumulation term and to the radial heat transfer from the
For transient calculations, the set of differential equations was in-
catalyst bed to the cooling medium [Eq. (5)]. The mass of reac-
tegrated along the catalyst bed for each time step. The same
tor wall was not considered as heat storage material because its
l boundary values as for steady-state calculations were required,
temperature is similar to the cooling medium (awall ! wall s and
together with additional initial values. Concentration, tempera-
awall ! aout). Variations in temperature and concentration occur
ture, pressure, and gas velocity profiles along the catalyst bed
with axial position and time. Radial profiles can be neglected (dt/
had to be first calculated. It is important to calculate the initial
dp 10).
values using the same axial discretization used for the transient
X calculations. Differences between the given initial values and
@C @Ci @u N

e i usf Ci sf uij rj 4 those that MATLAB calculated at the beginning of the simula-
@t @z @z j1 tion for t = 0 had to be avoided. Density, heat capacity, viscosity,
heat-transfer coefficient, and reaction rate were calculated for
@T each grid point as a function of temperature, concentrations,
e1g cp;g 1 e1s cp;s
@t pressure, and gas velocity.
X N 4U The main difficulty of fixed-bed reactors is the removal of reac-
@T @usf eff
1g cp;g usf T 9 R Hj0 rj T Tcooling tion heat. Heat needs to be transported perpendicular to the
@z @z d t
j1 flow to the heat exchange surface. The radial gradient depends
5 strongly on tube diameter, fluid properties, and fluid velocity.
Radial transfer conductivity and wall heat transfer coefficients
1 1 rt s 1 strongly depend on the catalyst form used, its size, the thermal
Ueff awall 4leff;r lwall aout characteristics of the catalyst (lp) and of the gas (lg), and the gas
velocity. The thermal conductivity of the gas is considerably
The variables are defined below. The pressure drop along the lower than the thermal conductivity of the catalyst; therefore,
fixed-bed was calculated using the equation proposed by beds with a low porosity favor an effective radial heat transfer

Energy Technol. 2016, 4, 90 103 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 100
(the porosity of the bed was defined by the selection of the cata- lithic honeycombs) to increase the number of degrees of freedom
lyst particle). Large particles are attractive for a good conductivi- and, in the end, the reactor productivity.[37]
ty but originate a lower wall heat transfer coefficient compared Figure 10 shows the procedure developed to design a recycle
with smaller particles. The selection of a catalyst support materi- system for reactions limited by the partial pressure of products
al with a high conductivity increases the conductivity of the bed and temperature as is the case for the CO2 hydrogenation inves-
and favors radial heat transfer. High gas velocity also favors an tigated herein. This procedure has two degrees of freedom for
efficient heat transfer. Heat transport under transient conditions design, the diameter of the tube dt and the desired flexibility of
is strongly influenced by the catalyst density and heat capacity. the reactor Lmin. An average fresh feed flow needs to be defined;
High catalyst density and heat capacity make changes in temper- consequently, a maximum and minimum flow could be calculated
ature (heat up or cool down) slow. according to Equations (7) and (8). Prior selection of operating
conditions [p, (H2/CO2)in, Tin, and Tcooling] is required. Design var-
iables for the catalyst particle (dp, cshell) are set at the beginning
of the calculation.
Catalyst particle: Eggshell
Fv;CO2 ;min Lmin Fv;CO2 ;max 7
Herein, spheres were selected due to its easier mathematical de-
scription compared with other catalyst shapes even though their Fv;CO2 ;min Fv;CO2 ;max
Fv;CO2 ;average 8
use in industrial reactors is very limited. The size of the catalyst 2
particle was also a compromise between a good internal mass
transfer (low pore diffusion resistance) favored by small particles In a recycle system, Xi,overall, Xi,pp, RF and Rout are connected by
and a low pressure drop favored by large particles. Heat transfer two equations; consequently, when two of them are selected the
conductivity was better when using large particles although other two can be calculated by the corresponding equations. In
wider tube diameters were required to avoid wall effects, which the case of a reaction limited by the partial pressure of products,
in turn had a negative effect on radial heat removal. The use of the maximum conversion per pass is fixed due to the partial pres-
eggshell catalyst pellets introduced design flexibility by decou- sure constraint (pH2 O /pH2 )out ; this limitation is reached using
pling the characteristic diffusion distance in catalyst pellets (re- a minimum volumetric flow. For the maximum volumetric flow,
sponsible for internal mass transfer limitations) from the pres- the overall conversion was specified by design. Rout was constant
sure drop. In the work presented here, eggshell spheres of alu- for the minimum and maximum operation load. An iterative pro-
mina were assumed to be coated with the active material cess using maximum and minimum flow was required to deter-
(iron).[35] The thickness of the active material shell was used as mine the value of Rout (Figure 10).
a new design variable (degree of freedom) for the catalyst parti-
cle (instead of using an effectiveness factor). The heat conductiv-
ity of the particle and the active mass density 1active,R changed as
Symbols used
a function of the thickness of active material. The advantages of ag [W m2 K1] heat transport coefficient gasparticle
eggshell catalysts are a complete use of the catalyst active mass awall [W m2 K1] heat transport coefficient bed-wall
(no internal mass transfer), the low pressure drop due to the use
aout [W m2 K1] heat transport coefficient wall-cooling
of large particles, and the lower heat production per reactor
volume compared to full contact (100 % active material) due to
the presence of inert material. The disadvantages are the re- aj,H2 O [] inhibition coefficient for H2O
quirement of larger reactors and increased manufacturing cost bj,CO2 [] inhibition coefficient for CO2
for the catalyst pellets. b [m s1] mass transfer coefficient gasparticle
In LTFT synthesis catalyst particle diameters were around 1 Ci [mol m3] concentration of component i
3 mm due to pressure drop constraints. Using this particle size, cp,g [] heat capacity of gas mixture
diffusion was limited due to the presence of liquid inside the cp,s [] heat capacity of catalyst
pores and, consequently, the effectiveness factor was less than
Deff [m2 s1] effective diffusion
one. The ARGE reactor used cylindrical particles with Lp =
dt [m] tube diameter
5 mm and dp = 2.5 mm.[36] Catalyst research focuses on the use of
structured catalysts (e.g., coreshell) and reactors (e.g., mono- dp [m] particle diameter

Figure 10. Procedure followed to calculate mcat. , Lt, and Rout for a fixed dt and Lmin assuming a catalyst deactivation at (pH2 O /pH2 )out > 0.1 and a desired overall
conversion of 0.9. Constant values for the design variables shown in Table 4 and an average flow Fv, CO2 , average need to be selected.

Energy Technol. 2016, 4, 90 103 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 101
EA [J mol1] activation energy Acknowledgements
9 HR [J mol1] heat of reaction
e [] porosity This work was financially supported by the German Bundes-
F [] flow rate ministerium fr Bildung und Wissenschaft(M.I.G.) (BMBF,
Fv,CO2 ,average [m3 h1] average flow rate of CO2, see Eq. 8 FKZ: 033RC1010A) and by the German Bundesministerium
i subindex for component fr Ernhrung und Landwirtschaft(BMEL, FKZ 22403711)
j subindex for reaction and by the Helmholtz Research School Energy-Related Cat-
k0 [mol s1 kg1 Pa1] pre-exponential factor alysis at Karlsruhe Institute of Technology (H.E.).
kj [mol s1 kg1 Pa1] reaction rate constant
Kp [] equilibrium constant
L [m] reactor tube length Keywords: CO2 hydrogenation fixed-bed reactor dynamic
mcat. [t] mass of active catalyst operation flexibility heterogeneous catalysis
L [] load
lp [W m1 K1] thermal conductivity of particle [1] M. Iglesias Gonzlez, G. Schaub, submitted, 2016.
leff,r [W m1 K1] effective thermal radial conductivity [2] Wired Chemist, 2015,
of bed entropies-organic, accessed 17 June 2015.
N [] number of reactions [3] J. B. Hansen, P. E. Hjlund Nielsen in Handbook of Heterogeneous
Catalysis, Methanol Synthesis (Eds.: G. Ertl, H. Knzinger, J. Weit-
NT [] number of tubes kamp), Wiley-VCH, Weinheim, 2008.
Pel,R-compressor [W] power required by the recycle com- [4] C. H. Bartholomew, R. J. Farrauto, Fundamentals of Industrial Cata-
pressor lytic Processes, 2nd ed., Wiley, Hoboken, 2006.
pi [Pa] partial pressure of i [5] Methanol: The Basic Chemical and Energy Feedstock of the Future
(Eds.: M. Bertau, H. Offermanns, L. Plass, F. Schmidt, H. J. Wer-
R [J mol1 K1] universal gas constant nicke), Springer, Berlin, 2014.
RF [] recycle ratio fresh [6] G. Bercic, J. Levec, Ind. Eng. Chem. Res. 1992, 31, 1035 1040.
Rout [] recycle ratio out [7] J. S. Kim, S. Lee, S. B. Lee, M. J. Choi, K. W. Lee, Catal. Today 2006,
r [mol m3 s1] reaction rate 115, 228 234.
[8] Exxon Mobil, 2012,
rt [m] tube radius nologies/files/sellsheet_09_mtg_brochure.pdf, accessed 8 December
1g [kg m3] density of gas 2015.
1s [kg m3] density of catalyst [9] S. Kvisle, T. Fuglerud, S. Kolboe, U. Olsbye, K. P. Lillerud, B. V.
1active,R [kg m3] density of active material in bed Vora, Methanol-to-Hydrocarbons in Handbook of Heterogeneous
Catalysis (Eds.: G. Ertl, H. Knzinger, J. Weitkamp), Wiley-VCH,
T [8C] temperature Weinheim, 2008.
t [s] time [10] H. Schulz, Catal. Today 2010, 154, 183 194.
t [s] characteristic time [11] M. Gtz, J. Lefebvre, F. Mrs, A. McDaniel Koch, F. Graf, S. Bajohr,
tmod mcat./Fv,in,STP modified residence time R. Reimert, T. Kolb, Renewable Energy 2015, 85, 1371 1390.
[12] T. Schaaf, J. Grnig, M. Schuster, T. Rothenfluh, A. Orth, Energy
[kg s m3] Sustainability Soc. 2014, 4, DOI: 10.1186/s13705-014-0029-1.
Ueff [W m2 K1] overall heat transport coefficient [13] J. Xu, G. F. Froment, AIChE J. 1989, 35, 88 96.
usf [m s1] superficial flow velocity [14] G. A. Mills, F. W. Steffgen, Catal. Rev. 1974, 8, 159 210.
Vr [m3] reactor volume [15] I. C. Yates, C. N. Satterfield, Energy Fuels 1991, 5, 168 173.
[16] G. P. van der Laan, A. A. C. M. Beenackers, Catal. Rev. 1999, 41,
Xi [] conversion of i 255 318.
Xi,pp [] conversion of i per pass [17] Studies in Surface Science and CatalysisFischer Tropsch Technolo-
cshell [m] thickness of active material gy, Vol. 152 (Eds.: A. P. Steynberg, M. E. Dry), Elsevier, Amsterdam,
Yi [] yield of i 2004.
[18] M. Iglesias Gonzlez, C. de Vries, M. Claeys, G. Schaub, Catal. Today
ym [] hydrocarbon mass fraction 2015, 242, 184 192.
z [m] axial variable [19] M. Iglesias Gonzlez, G. Schaub, Chem. Ing. Tech. 2015, 87, 848 854.
[20] M. Claeys, P. Moodley, A. Harilal, A. Ferreira, Oxidation of Hgg
Carbide at High Temperature Fischer Tropsch Synthesis Captured In-
Abbreviations situ in 10th Natural Gas Conversion Symposium (Doha, Qatar), 2013.
[21] A. P. Steynberg, R. L. Espinoza, B. Jager, A. C. Vosloo, Appl. Catal.
ARGE Arbeitsgemeinschaft (Lurgi + Ruhrchemie) A 1999, 186, 41 54.
[22] P. E. Hjlund-Nielsen, J. Bgild-Hansen, J. Molecular Catal. 1982, 17,
DME dimethylether
183 193.
LTFT low-Temperature FischerTropsch [23] G. F. Froment, Chem. Ing. Tech. 1974, 46, 374 386.
HTFT high-Temperature FischerTropsch [24] Sasol, MarlothermHeat-transfer fluids, 2009, http://www.sasoltech-
MOL method of lines, accessed 17 June
MTG methanol to gasoline
[25] M. P. Rohde, G. Schaub, S. Khajavi, J. C. Jansen, F. Kapteijn, Micro-
ODE ordinary differential equation porous Mesoporous Mater. 2008, 115, 123 136.
SNG substitute natural gas [26] M. P. Rohde, PhD thesis, Karlsruhe Institute of Technology (Germa-
STP standard temperature and pressure ny), 2011.
[27] G. F. Froment, K. B. Bischoff, Chemical Reactor Analysis and Design,
2nd ed., Wiley, New York, 1990.

Energy Technol. 2016, 4, 90 103 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 102
[28] F. Graf, M. Gtz, M. Henel, T. Schaaf, R. Tichler, Technokonomi- [33] N. Fischer, R. Henkel, B. Hettel, M. Iglesias Gonzlez, G. Schaub, M.
sche Studie von Power-to-Gas-KonzeptenAbschlussbericht, Claeys, Upgrade of Substitute Natural Gas (SNG) via CO2 Hydroge-
Deutscher Verein des Gas- und Wasserfaches e.V., DVGW-For- nationAn In Situ Perspective in North America Catalysis Society
schung, 2014, NAM 24 (Pittsburgh, PA), 2015.
bote/forschung/innovation/pdf/g3_01_12_tp_b_d.pdf. [34] W. E. Schiesser, G. W. Griffiths, A Compendium of Partial Differen-
[29] J. J. Lerou, K. M. Ng, Chem. Eng. Sci. 1996, 51, 1595 1614. tial Equation ModelsMethod of Lines Analysis with Matlab, Cam-
[30] G. K. Boreskov, Y. S. Matros, Catal. Rev. 1983, 25, 551 590. bridge University Press, Cambridge, 2009.
[31] B. A. Finlayson, L. T. Biegler, I. E. Grossmann, Mathematics in [35] VFF, 2015,, accessed 12 May 2015.
Chemical Engineering in Ullmanns Encyclopedia of Industrial [36] A. Jess, C. Kern, Chem. Eng. Technol. 2009, 32, 1164 1175.
Chemistry, Wiley-VCH, Weinheim, 2000. [37] R. Gttel, Chem. Ing. Tech. 2015, 87, 694 701.
[32] G. Eigenberger, J. Werther, H. Schoenfelder, A. A. C. M. Beenackers,
Y. S. Matros, G. A. Bunimovich, Reaction Engineering in Handbook Received: July 30, 2015
of Heterogeneous Catalysis (Eds.: G. Ertl, H. Knzinger, J. Weit- Revised: September 4, 2015
kamp), Wiley-VCH, Weinheim, 2008. Published online on December 16, 2015

Energy Technol. 2016, 4, 90 103 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 103