Chemical Engineering Science 272 (2023) 118611

Contents lists available at ScienceDirect

Chemical Engineering Science

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

Flexible operations of a multi-tubular reactor for methanol synthesis

from biogas exploiting green hydrogen
Riccardo Rinaldi, Carlo Giorgio Visconti
Politecnico di Milano, Department of Energy, Via La Masa 34, 20156 Milano, Italy

h i g h l i g h t s g r a p h i c a l a b s t r a c t

A 2D heterogeneous transient model

for multi-tubular methanol reactor is
developed.

The model is used to simulate the
operations of a biogas-to-methanol
process.

The flexibility of a multi-tubular
reactor is shown in the presence of
load changes.

A sensitivity analysis is conducted on
reactor inlet flowrate and
composition.

The effects of tube length and gas
hourly space velocity are assessed.

a r t i c l e i n f o a b s t r a c t

Article history: The present modelling work studies the steady state and transient performances of a multi–tubular fixed
Received 13 December 2022 bed reactor for methanol production from biogas. The modelled system comprises a methanol synthesis
Received in revised form 18 February 2023 reactor, a flash unit, and accounts for the unconverted gas recycle. The novel concept herein proposed and
Accepted 2 March 2023
assessed is the possibility to run a multi–tubular methanol synthesis reactor flexibly, i.e., using the car-
Available online 9 March 2023
bon dioxide present in biogas, together with renewable H2, to increase methanol productivity, only when
this is economically convenient. Accordingly, the methanol synthesis multi–tubular reactor under study
Keywords:
is run in two operating conditions: when the cost of green hydrogen is high, the excess of CO2 present in
Methanol reactor modelling
Dynamic model of reactor
biogas is vented and the reactor is fed with CO2-lean syngas only; conversely, in the presence of afford-
MeOH from biogas able renewable H2, CO2 and green H2 are cofed to the reactor. This study focuses in particular on the
Two–dimensional model impacts of the two working conditions on the methanol productivity, as well as on temperature profiles
Heterogeneous reactor model and transient behaviour of the reactor. The effects of tube length and number are also simulated. The
study shows that the multi–tubular reactor design well manages both operating conditions and that
the steady-state conditions can be reached in a few hours after the switch from one condition to the
other. In view of a small scale process, simulations show that it is better to reduce the tube number rather
than their length, as the latter choice would lead to unacceptable temperature hotspots, due to a wors-
ening of convective heat transfer.
Ó 2023 Elsevier Ltd. All rights reserved.

1. Introduction

The worldwide increase in energy consumption and the urgent

need to reduce greenhouse gas (GHGS ) emissions have encouraged
E-mail addresses: [email protected] (R. Rinaldi), [email protected] the scientific community to find alternatives to non-renewable
(C.G. Visconti)

https://doi.org/10.1016/j.ces.2023.118611
0009-2509/Ó 2023 Elsevier Ltd. All rights reserved.
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

Nomenclature
h i
Acronyms k2 Rate constant for reaction (2) (rWGS) kg mol
CAT sbar
[Unit] K ads;1 Adsorption constant ½
BFDM Backward Finite Differences Method ½ h i
K ads;2 Adsorption constant pffiffiffiffiffi
1 ffi
Eos Equation of state ½ bar

 1
GHGS Greenhouse gases ½ K ads;3 Adsorption constant bar
Nm3 W
GHSV Gas Hourly Space velocity hhm i3 kcat Catalyst thermal conductivity mK
€ CAT W
LCOH Levelized cost of hydrogen kg keff ;rad Radial effective thermal conductivity mK
MeOH Methanol ½ W
NR Number of reactions ½ keff ;ax Axial effective thermal conductivity mK
OCFEM Orthogonal Collocation on Finite Elements Method ½ K eq;MeOH Equilibrium constant of methanol reaction ½
PDEs Partial differential equations ½ K eq;rWGS Equilibrium constant of rWGS  W  reaction ½
POCS Periodic open cellular structures ½ kgas Gas thermal conductivity mK
 
RR Recycle ratio (recycled gases / fresh syngas) ½ K m;i Gas – solid mass transfer coefficient ms
rWGS Reverse water gas shift reaction ½ L Tube length ½m
M Fresh syngas module ½ h i
Symbols m Mass flow rate through reactor kg s
h i
Latin letters [Unit]   MMgas Gas mixture molar mass mol kg
AC Single tube cross-sectional area m2 h 2i h i
kg
av Catalyst pellet specific volumetric surface m
m3
MMi Molar mass of component i mol
 Nt Tubes number ½
Ai Pre-exponential factor in reaction rate constants or in Nu Nusselt number ½
h i h i
adsorption constants mol
or kg mol for reac- p Reactor pressure ½bar
kgCAT sbar
2 sbar
h i   CAT
pflash Flash operating pressure ½bar
tion rate constants; ½, p ffiffiffiffiffi
1 ffi 1
, bar for adsorption con- pref Reformer pressure ½bar
bar
psyngas Inlet syngas pressure ½bar
stants h i

J PeM Material Péclet number ½
Bi Activation energy / Adsorption enthalpy mol
h i PeT Thermal Péclet number ½
cp;gas J
Mixture specific heat kgK Pr Prandtl number ½
h i Q_ tube Thermal power released by the reaction per single tube
m2
Deff ;axH2 Effective axial diffusion coefficient of hydrogen s ½W h i
J
Deff ;intri Effective intraporous diffusion coefficient of component R Ideal gas constant molK
h 2i
r Radial coordinate ½m
i ms h 2i h i
mol
Deff ;radi Effective radial diffusion coefficient of component i ms rj Rate of reaction j skg
CAT

rjk Rate hof reaction

i j of point k in diffusion length coordi-
Di;mix Diffusion coefficient of i-th component in the mixture
h 2i mol
nate skg
CAT
m
s h 2i rp Particle radius ½m
DK;i i-th species Knudsen diffusivity ms Re Reynolds number ½
DL Diffusion length ½m Redpa Reynolds number, calculated with dpa ½
dp Pellet diameter ½m Redpe Reynolds number, with dpe ½
dpa Sphere diameter with the same surface as pellet [m]  calculated

Sp Particle surface m2
dpe Sphere diameter with the same volumetric surface as Sci Schmidt number of species i ½
pellet ½m Shi i-th species Sherwood number ½
Dt Tube diameter ½m   t Time coordinate ½s
F_ CO CO molar flowrate (at MeOH reactor entrance) mol s T Gas phase temperature in reactor ½K
F_ CO2 CO2 molar flowrate (at MeOH reactor entrance, mol or Tav Averaged reactor temperature ½K
vented, or recycled instead of being vented) s Tc Centreline reactor temperature ½K
F_ fresh Molar
 flowrate
 of total fresh gases entering MeOH reac- Tcool Cooling medium temperature ½K
tor mol
s Tflash Flash operating temperature ½K
F_ H2 Hydrogen molar flowrate (coming molfrom the electrolyser TIN Inlet temperature ½K
or at MeOH reactor entrance)  s T max1D Maximum temperature profile of the 1D model ½K
F_ i i-th species molar flowrate mol s T max2D Maximum temperature profile of the 2D model ½K
fi Fugacity of species i ½bar Ts Solid phase temperature ½K
fm Friction factor ½   Ts NEW Solid phase temperature after the introduction of H2/
F_ rec Molar flowrate of recycled gases mol s CO2 ½K
  Inlet syngas temperature ½K
F_ syn;in Inlet molar flowrate of fresh syngas mol s
Tsyngas
h i Tw Reactor temperature on the wall ½Kh i
hconv Convective heat transfer coefficient mW U 1D overall heat transfercoefficient W
h i
2 K
 m2 K
v_ Gas superficial velocity ms h 3i
hw Wall heat transfer coefficient mW 2 K Nm
VNORM; IN Reactor inlet normal
 3  volumetric flowrate h
hw;conv h
Convective i contribution of the wall heat transfer coeffi- Vp Particle volume m h i
kg
cient mW
2 K Wt Mass flowrate per unit area sm 2

x Diffusion length coordinate [m]
k1 Rateh constant
i for reaction (1) (MeOH synthe-
mol
xi i-th component mole fraction ½
sis) 2
kgCAT sbar

2
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

z Axial coordinate ½m s Pellet tortuosity ½

z Dimensionless axial coordinate ½ xi Mass fraction in gas phase of component i ½
xi IN Inlet mass fraction in gas phase of component i ½
Greek letters [Unit] h i xi IN NEW Inlet mass fraction in gas phase of component i after the
DHrj enthalpy of reaction j mol J introduction of H2/CO2 ½
h 3 i xs;i Mass fraction in solid phase of component i ½
e m
Bed void fraction m3void

 xs;i IN Inlet mass fraction in solid phase of component i ½
reactor

ep m3
Particle void fraction m3 void xs;i IN NEW Inlet mass fraction in solid phase of component I after
particle

 the introduction of H2/CO2 ½
ji Parameter from Arrhenius law Same as Ai
lgas Gas mixture viscosity ½Pa  s h i Subscripts [Unit]
m3
Ei i-th species diffusion volume mol c Subscripts for centreline []
h i
qcat Catalyst density m kg
3
i Subscripts for chemical species []
h i j Subscripts for chemical reaction []
qgas Gas phase density m kg
3 k Subscript for discretization point of diffusion length ½
s Subscript for solid phase ½
mij Stoichiometric coefficient of i-th component in j-th
reaction ½

energy sources. The transportation sector, mainly relying on excess of carbon dioxide contained in the syngas manufactured
hydrocarbons based fuels, is responsible for approximately 20% through biogas reforming (Vita et al., 2018), resulting in values of
of energy related CO2 emissions (Sims et al., 2014): new fuels, with M that can be even negative (Rinaldi et al., 2023). In the available
lower environmental impact, need hence to be commercialized to literature, such CO2 excess is managed in two different ways. In the
face such issue. works of dos Santos et al. (2018), Chein et al. (2021), and Ghosh
Methanol (MeOH) is one of the major commodities of the chem- et al. (2019), it is proposed to utilize the CO2-rich syngas as such,
ical industry, utilized as a starting material for the production of i.e., with M 2: the adopted solution leads to a lower yield
many organic compounds. However, it has applications also as a towards the desired product (Rinaldi et al., 2023). Other authors
fuel for internal combustion engines, as a H2 carrier, or as a reac- propose to remove the extra carbon dioxide content from the feed-
tant for fuel cells (Laird, 1997). In fact, it is easy to store and trans- stock (mostly through pressure swing adsorption (Augelletti et al.,
port (Gautam et al., 2020), when combusted ensures lower 2017), or physicochemical absorption (Mulu et al., 2021) (Miltner
emissions of carbon monoxide, nitrogen oxides and hydrocarbons et al., 2017)) or to adjust the composition of syngas through a
with respect to conventional hydrocarbons-based fuels and also reverse water gas shift reactor (Ghosh et al., 2019). The main dis-
has a higher octane number with respect to gasoline (Laird, 1997). advantage of the former solution is the fact that a remarkable por-
MeOH is generally synthesized from natural gas or coal, through tion of the carbon contained in biogas is not utilized to produce
well-established processes including the production of synthesis methanol, but it is vented to the atmosphere; conversely, the latter
gas (H2 + CO + CO2, aka syngas) as a first step. In order to maximize solution implies the necessity to have an additional reactor, with
the yield to methanol, the fresh syngas supplied to the synthesis higher fixed and operating expenditures.
reactor, that is operated with recycle to overcome the low per- An alternative solution is the use of an external source of
pass conversion, needs to satisfy some compositional require- renewable H2, aka green hydrogen, to balance the syngas composi-
ments, often expressed through the stoichiometric module, M, tion, i.e., its module M. In this way, also the CO2 contained in biogas
defined in Eq. (1), where F_ i is the molar flowrate in the syngas of (up to 45 % (Calbry-Muzyka et al., 2022)) can be exploited as bio-
the i-th component (Laird, 1997). genic carbon source for the desired product and the higher plant
costs can be compensated by an increase in methanol productivity.
F_ H2  F_ CO2 The main criticism to such solution are the high fixed and operat-
M¼ ð1Þ
F_ CO þ F_ CO2
ing costs of electrolyzers: indeed, the most common devices on the
industry, based on proton exchange or alkaline membranes
The value of the stoichiometric module should be in principle (International Energy Agency, 2019), are characterized by very
equal to 2 (Laird, 1997) (Phillips et al., 1990): indeed, if M is lower, high levelized costs of hydrogen (LCOH, between 3 and 8=kgH2 )
the system undergoes a shortage of hydrogen to fully convert the (Mayyas et al., 2019). Battaglia et. al. (2021), for instance, studied
carbon oxides into methanol; conversely, if it is higher than 2, the possibility to produce MeOH by coupling carbon capture and
there is an excess of H2. In general, it is common practice to have a water electrolysis. The economic analysis they conducted shows
fresh syngas fed with a stoichiometric module slightly above 2 that almost 90% of total fixed capital investment has to be allocated
(Albarelli et al., 2017) (Clausen et al., 2010): this results in very to the realization of water electrolysis. Moreover, due to the vola-
high values of M at the reactor inlet, thanks to the high recircula- tile prices of power and electricity, the operating costs of elec-
tion ratio in the methanol synthesis loop (Montebelli et al., 2013). trolyzers can significantly vary in time, thus remarkably affecting
Notably, the operations in the presence of excess H2 help the cata- MeOH production prices.
lyst stability, prolonging its lifetime. Novel strategies to limit the cost of renewable methanol while
A viable (dos Santos et al., 2018) alternative to natural gas as maximizing the utilization of renewable carbon are thus needed
feedstock for the methanol synthesis is represented by biogas, a to make bio-methanol competitive. While many efforts are
renewable mixture primarily composed of CH4 and CO2, that is devoted to the reduction of the fixed costs of electrolyzers, the nov-
either collected from landfills or derived from the anaerobic diges- elty proposed by the present work is the development of an inte-
tion of agricultural or animal wastes (Badurek, 2020). In recent grated process with green hydrogen production that exploits this
years, efforts have been made in the research on biogas to metha- H2 only when this is economically affordable. More specifically,
nol processes, showing that the main issue one has to tackle is the
3
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

when the LCOH is high, the excess CO2 is separated from biogas Rinaldi et al. (2023), and Fig. 1b is the novelty introduced in the
and vented. When electricity prices are sufficiently low, instead, present study.
an external flux of green hydrogen coming from the electrolyzer The process configuration shown in Fig. 1a is determined by
is inserted at reactor inlet together with the excess CO2, so as to Rinaldi et al., 2023 on the basis of an economic optimum. Fresh
increase MeOH productivity and minimize the CO2 emissions. biogas is partly sent to the CO2 removal section, where the excess
Given these premises, the main goal of the present work is to of carbon dioxide is removed from the raw biogas before feeding
assess the impacts on methanol reactor of composition and flow- the lean gas to the reformer, and party utilized as fuel for the steam
rate changes due to the introduction of CO2 and H2 streams in reformer. CO2 removal occurs through chemical absorption and
the reactor when the LCOH is low. The novelty proposed in the pre- solvent regeneration in a stripping unit (methyl diethanolamine
sent work is the possibility to exploit one single unit in two oper- is used as solvent). The lean biogas is then mixed with steam, con-
ating modes (with and without CO2 / H2 addition): such choice verted into syngas in the reformer, compressed and sent to the
leads to the necessity of studying the reactor behavior at steady methanol synthesis unit. This is a multi–tubular fixed–bed reactor,
state conditions, but also during the transitory phases. cooled down by steam. Unreacted gases, after recompression, are
In the first part of the paper (section 2.1), the process configu- mainly recycled at the entrance of the reactor to enhance conver-
ration of interest is presented. Then, the 2D heterogeneous model sion; a small portion is burnt in the reformer combustor. Purifica-
used to simulate the methanol synthesis reactor is described in tion steps, including a flash unit and two distillation columns,
section 2.2. The 1D reactor model used by Rinaldi et al. (Rinaldi follow to obtain 99% pure MeOH.
et al., 2023) for the techno-economic analysis of methanol plant The process configuration shown in Fig. 1b is that studied in this
from biogas is then compared with the outcomes of the two work. It is assumed that, in the presence of renewable and afford-
dimensional model (section 3.1). The more accurate predictions able H2 produced by an electrolyzer, the excess CO2 separated from
of the latter suggest that one should assess the performances of biogas can be mixed with the green H2 and fed to the methanol
the unit of interest through a 2D model. The effects of the introduc- reactor, together with the syngas produced via biogas reforming.
tion of CO2 and renewable H2 on the reactor performances are then The green hydrogen and the separated CO2 are injected immedi-
discussed in terms of changes in methanol productivity, reactor ately upstream of MeOH reactor entrance. Such a process configu-
thermal profiles, and transient behavior (section 3.2). In view of ration is selected on the basis of two considerations. Firstly, the
a more compact – scale process, the effects of tubes length and excess of carbon dioxide in the raw-biogas, if not removed
number reduction are analyzed (section 3.3), so as to understand upstream the biogas reforming, would lead to a lower methane
the possibility and the limits associated to the use of smaller conversion in the steam reformer and hence to a smaller MeOH
reactors. production, as reported by Rinaldi et al. (2023). Secondly, if the
CO2 removal unit should operate in a different mode depending
on the availability of green hydrogen (i.e. separating the excess
CO2 when green H2 is not available and not doing that when
2. Methods low-cost H2 is available), this would cause alterations of both the
reforming unit and the CO2 removal system, which would also be
2.1. Process configuration subjected to transient operational regimes. As shown by
Koronaki et al. (2015) and by Kvamsdal et al. (2009), transient
The plant taken as reference in this work is the one studied by behavior of CO2 removal systems is often problematic, and these
Rinaldi et al. (2023), that produces around 2500 kg=h of MeOH systems could take up to several hours before reaching reach
from biogas. The base case, if compared to the European landscape, steady state. Similarly, the steam reformer transitory phase does
is compatible with a medium-large sized biogas plant (European not occur quickly and its impact on the process should be carefully
Biogas Association, 2021); the plant treats 3635 m3/h of biogas assessed, as pointed out by Wang et al. (2021).
(20 MW of inlet thermal power), composed of methane (60 % The solution adopted in this paper (see Fig. 1b) tackles both the
mol) and CO2 (40 %mol). The process scheme is depicted in issues described above, as the steam reformer and the CO2 removal
Fig. 1, where Fig. 1a is the process configuration as described by

Fig. 1. (a) plant scheme in the reference case, as described by Rinaldi et al. (2023). (b) Novelty introduced in the present paper to increase MeOH productivity and avoid CO2
emissions.

4
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

Fig. 2. Sketch of the simulated synthesis loop, including methanol reactor, the flash unit, and unconverted gases recycle (with recompression and cooling). The numbers refer
to the streams, whose values are reported in Appendix C.

system operate independently on the availability of green H2, and Accordingly, a two dimensional heterogeneous chemical reactor
the MeOH synthesis reactor is the only unit undergoing a transient model is developed and utilized in this paper to perform steady
phase. In this way, even in the presence of significant changes in state and transient simulations, as well as to carry out a parametric
the streams flowrates and compositions, the impact on the single analysis on reactor dimensions. Considering that the methanol
unit performances is limited. synthesis reactor is operated with recycle, the model utilized in
Notably, since the injection of green H2 between the biogas the present work, schematized in Fig. 2, accounts both for MeOH
reforming and the methanol synthesis reactor only affects metha- synthesis and unconverted gases recycle. Fresh syngas (stream 1
nol synthesis loop performances, the numerical study carried out in Fig. 2), potentially mixed with green H2 and biogenic CO2
in the present work is focused only on this loop, i.e., on the mod- (stream 2), is added with recycled gases (stream 12) before enter-
elling of the MeOH synthesis reactor, the product separation in a ing methanol synthesis reactor. When the mixture (stream 4)
flash unit, and the recycle of unconverted gases recycle (also leaves the reactor, after being cooled to 40 °C (stream 5) and
accounting for the need of a re-compression of the recycled decompressed (stream 6), it is sent to a flash unit for product sep-
stream). aration. Liquid methanol, diluted with water, leaves the synthesis
loop for the bottom of the flash unit (stream 7). The gaseous stream
from the top of the flash unit (stream 8), mainly composed of
2.2. Catalytic reactor model
unconverted reactants and inert species, is partially purged
(stream 9) to prevent the accumulation of inert species, while
When process simulations are performed, catalytic reactors are
the rest (stream 10) is exploited as recycle gas to increase metha-
often approximated through equilibrium models or simple one–di-
nol production. Due to the pressure losses in the recirculation
mensional models (Nguyen & Zondervan, 2019). This approach is
streams, the synthesis loop is provided with a recompression unit.
usually appropriate when one has to account for global plant per-
Further cooling of the recycled and recompressed stream 11 is
formances by means of mass and energy balances but does not
required to achieve fresh syngas temperature. While the chemical
allow to say much on the reactor design and its capacity to operate
reactor, the flash unit, and the gas recirculation are modelled in
at different sets of process conditions, managing the transients
detail, the cooling and recompression processes are modelled as
caused by the switch from one condition to the other. To this
streams temperature or pressure changes (the detailed modelling
aim, a more detailed model has to be developed, also accounting
of this units is reported in Rinaldi et al., 2023).
for micro and local effects inside the unit and the catalyst
The main parameters utilized in the simulations of the metha-
(Montebelli et al., 2013) and changes along the radial coordinate
nol synthesis loop are reported in Table 1. The information about
(Leonzio & Foscolo, 2020).

Table 1
Parameters utilized for methanol reactor modelling.

Symbol Description Value Unit Reference

Process
RR Recycle ratio (molar ratio of recycled gases over fresh gases) 5 ½ (Rinaldi et al., 2023)
pref Reforming operating pressure (‘‘Biogas Reforming” in Fig. 1) 12 bar (Laird, 1997)(Ghosh et al., 2019)
GHSV Gas hourly space velocity 4000 Nm3 (Montebelli et al., 2013) (Chein et al., 2021)
hm3CAT
Methanol reactor
Tsyngas Fresh syngas temperature at reactor inlet (‘‘Fresh syngas” in Fig. 2) 472:8 K (Rinaldi et al., 2023)
Tcool Cooling medium temperature in the reactor 513:2 K (Rinaldi et al., 2023)
psyngas Fresh syngas pressure at reactor inlet (‘‘Fresh syngas” in Fig. 2) 92:8 bar (Laird, 1997)
F_ syn; in Fresh syngas molar flow rate at reactor inlet (‘‘Fresh syngas” in Fig. 2) 76:3 mol
s
(Rinaldi et al., 2023)
L Reactor length (‘‘MeOH reactor” in Fig. 2) 6 m (Rinaldi et al., 2023)
qcat Catalyst density (‘‘MeOH reactor” in Fig. 2) 1712 kg (Montebelli et al., 2013)
m3
e Reactor void fraction (‘‘MeOH reactor” in Fig. 2) 0:4 ½ (Rinaldi et al., 2023)(Montebelli et al., 2013)
Dt Reactor tubes diameter (‘‘MeOH reactor” in Fig. 2) 0:042 m (Montebelli et al., 2013)
Nt Number of tubes in the reactor (‘‘MeOH reactor” in Fig. 2) 1852 ½ (Rinaldi et al., 2023)
Flash unit
Tflash Flash unit operating temperature (‘‘Flash unit” in Fig. 2) 313:2 K (Rinaldi et al., 2023)
pflash Flash unit operating pressure (‘‘Flash unit” in Fig. 2) 90 bar (Rinaldi et al., 2023)

5
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

MeOH reactor and process conditions (gas hourly space velocity or order of 104 (Montebelli et al., 2013; Yagi et al., 1960;
GHSV, tubes diameter, cooling fluid temperature) and the proper- Specchia & Sicardi, 1980);
ties of the utilized catalyst (i.e., its density and void fraction) are
Radial gas thermal conductivity and diffusivity are considered,
taken from a previous paper by some of the authors (Montebelli as these contributions strongly affect the model predictions of
et al., 2013). Such values are typical of industrial methanol reactors thermal profiles (Dixon, 2012). The hw  keff; rad approach is fol-
(Hartig & Keil, 2002) and are widely utilized to design a unit that lowed, as proposed by Specchia and Sicardi (Specchia & Sicardi,
treats a poor in CO2 inlet fresh syngas. The other operating condi- 1980) and by Montebelli et al. (Montebelli et al., 2013);
tions (i.e., recycle ratio, fresh syngas temperature and pressure, and
Radiative heat transfer is neglected, due to the low tempera-
flash operating conditions) are selected on the basis of the out- tures involved (T < 400 CÞ (Montebelli et al., 2013);
comes proposed in our recent paper (Rinaldi et al., 2023), as a
The catalytic pellet is assumed isothermal and isobaric, as it is
result of a tecno-economic assessment. Given the aforementioned proved that temperature and pressure gradients are limited
data, the number of reactor tubes utilized in the simulations, Nt , is inside MeOH synthesis catalytic pellets (Rout et al., 2013);
computed on the basis of the chosen gas hourly space velocity, as
The catalytic pellet is assumed to always be at the steady –
reported in Eq. (2). state, as the transient phase takes place very quickly (Solsvik
_
4  VNORM;IN & Jakobsen, 2011);
Nt ¼ ð2Þ
Conduction across the tubes wall is negligible.
GHSV  p  Dt  L  ð1  eÞ

Methanol reactor performances are assessed trough a two – The kinetic model proposed by Vanden Bussche and Froment
dimensional heterogeneous model, based on the equations (Wakao & Funazkri, 1978) is utilized in the simulation, since this
reported in Table 2. The mass balance in the gaseous phase is model proved to be effective for a multitubular packed bed reactor
described by Eq. (3); the balance accounts for the presence of radial for methanol production (Leonzio, 2020). It is assumed that the
dispersions and gas solid mass – transfer but neglects axial diffu- reactor is packed with a commercial catalyst (Cu/Zn/Al2O3). No cat-
sion. Inter – particle diffusion phenomena are considered through alyst deactivation is assumed in the numerical model, as tests on
Eq. (4), whereas Eq. (5) describes the mass balance at the solid the pellets showed that, because of its high stability, the catalyst
phase. The energy balance in the gas phase is reported in Eq. (6); maintains its high activity level for a lifespan as long as five years
the equation accounts for the presence of radial diffusion limita- MegaMaxÒ 800. The reactions involved are methanol synthesis
tions but disregards axial dispersion. Inter – particle heat transfer from carbon dioxide and reverse water gas shift reaction. Due to
limitations are described by Eq. (7), while pressure variations along the medium–high pressure of the process, the kinetic model pro-
the catalytic bed are accounted through Eq. (8). The equations are posed by the author is corrected through the use of fugacities
written to account for the transient modelling of the system; in the instead of partial pressures (Peter et al., 2012). Peng – Robinson
following, when results are reported at the steady state, the equa- Equation of State (EoS) is adopted for the purpose (Peng &
tions reported in Table 1 are modified by setting the time deriva- Robinson, 1976). The detailed kinetic model is reported in Appen-
tive equal to zero. dix A, along with all the transport correlations.
The main features of the model are described below: Boiling water at 33 bar and approximatively 240 °C (see
Table 1), that is partially vaporized in the reactor shell and leaves

The reactor is a multi–tubular fixed bed unit, which is described it as saturated steam, is used as cooling medium. Its temperature
through the modelling of one of these tubes; the catalyst is is assumed constant throughout the process, as this reflects the
located inside the tubes and the cooling medium (boiling water) actual conditions of industrial methanol reactors such as Lurgi
is in the reactor shell; technology (Laird, 1997; Montebelli et al., 2013; Montebelli et al.,

Axial diffusivity is neglected. This decision is taken by ensuring 2014).
that material Péclet number (PeM , defined as
PeM ¼ Deff;ax H2 =v  L) is always << 1 (see Wakao and Funazkri 2.3. Flash and gas recycle model
(Wakao & Funazkri, 1978) and Montebelli et al. (Montebelli
et al., 2013)). In the simulations of interest, the value of PeM is As shown in Fig. 2, the stream that leaves methanol synthesis
reactor is sent to a flash unit (simulated using the Peng-Robinson
always in the order of 104 ;
EoS) to separate the liquid phase, mainly composed of methanol

Axial gas thermal conductivity is disregarded, as the thermal
and water, from unconverted reactants (H2, CO2, CO), and small
Péclet number, PeT ¼ keff;ax =qgas  cp;gas  v  L, is always in the
amounts of CH4. Most of these species are recycled to methanol
reactor entrance to enhance methanol conversion; a small portion

Table 2
Main model equation for methanol reactor.

Mass Balances
!
Gas phase @ xi @ xi @ 2 xi 1 @ xi
qgas  ¼ Wt  þ qgas  Deff;rad i  þ þ qgas  Km;i  av  ðxs;i  xi Þ Eq. (3)
@t @z @r2 r @r
Gas – solid interphase @x X NR
qgas  s;i ¼ qcat  ð1  eÞ  MMi  j¼1 ðmij  rj Þ þ qgas  Km;i  av  ðxi  xs;i Þ Eq. (4)
@t  
Solid phase @ dxs;i XNR
0 ¼ qgas   Deff; i   þ qcat  MMi  j¼1
ðmij  rjk Þ Eq. (5)
@x dx
Energy Balances
!
Gas phase @T @T @ 2 T 1 @T
qgas  cp;gas  ¼ Wt  cp;gas  þ keff;rad  þ þ hconv  av  ðTs  TÞ Eq. (6)
@t @z @r2 r @r
Gas – solid interphase @Ts X NR
qgas  cp;gas  ¼ qcat  ð1  eÞ  ðDHrj  rj Þ þ hconv  av  ðT  Ts Þ Eq. (7)
@t j¼1
Momentum Balance
!
Gas phase qgas 1 @p 1 @T XNC  1 @ xi 
 þ þ MMgas þ 2  f m  av ¼ 0 Eq. (8)
W2t p @z T @z i¼1 MM @z
i

6
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

is sent to the combustor of the reforming section where it burns to The reactor axial coordinate (z) is discretized through Backward
provide heat for the reformer. The fraction of gases leaving the Finite Differences Method (BFDM), with a polynomial function
methanol synthesis reactor that are recycled to the reactor itself truncated at the first order (60 nodes utilized). A non – uniform
is determined by means of a parameter, called recycle ratio (RR). logarithmic grid transformation is applied to the axial coordinate,
The parameter is defined, for every simulation, as reported in Eq. so as to account for the largest thermal and compositional gradi-
(9). ents at the start of the domain. The radial coordinate (r) is dis-
cretized by Orthogonal Collocation on Finite Elements Method
_
(OCFEM), with a polynomial function truncated at the third order
Frec 
RR ¼ _
ð9Þ (3 elements utilized). Diffusion length (x) discretization is carried
Ffresh out through OCFEM method, assuming 1 element across pellet
coordinate. Time discretization cannot be specified by the user
where F_ rec is the molar flowrate of recycled gas (‘‘Recycled gas”
on gPROMSÒ, as it varies according to the system convergence rate.
in Fig. 2) and F_ fresh is the molar flow of total fresh gases entering
A 10 s reporting time interval is set, so as to properly describe the
MeOH reactor. Since the reactor inlet flowrate changes in the pres-
system transient behavior. The total number of equations to be
ence of extra H2 and CO2, it is necessary to take such variation into
solved is around 3:97  105 for each time interval.
account in the definition of the system recirculation, always ensur-
ing a high yield to methanol. To this aim, F_ fresh is set to change with
the simulation: in the process configuration depicted in Fig. 1a (i.e., 3. Results and discussion
when no extra H2 and CO2 are utilized), F_ fresh is simply the molar
3.1. Selection of the most appropriate reactor
flowrate of fresh syngas (from bio-methane steam reforming)
entering the reactor. In this case, the recycle ratio is defined by
The temperature along the axial coordinate of the 1D model is
Eq. (10), where F_ syn; in is the molar flowrate of fresh syngas entering represented in Fig. 3a, where it is also compared with the wall
the reactor (namely, ‘‘Fresh syngas” of Fig. 2). (Tw ) and centerline (Tc ) temperature profiles computed with the

F_ rec
RR ¼ ð10Þ
_Fsyn; in

Conversely, if renewable hydrogen and carbon dioxide excess

are exploited for the process (process configuration shown in
Fig. 1b), F_ fresh of Eq. (9) is the sum of syngas, CO2, and H2 flowrates
fed to methanol reactor (it corresponds to the sum of ‘‘Fresh syn-
gas” and ‘‘H2 + CO2” streams of Fig. 2). Hence, the definition of recy-
cle ratio is given by Eq. (11), where F_ H is the molar flowrate of
2

renewable hydrogen and F_ CO2 is the molar flowrate of excess car-

bon dioxide.

F_ rec
RR ¼ ð11Þ
_Fsyn; in þ F_ H þ F_ CO
2 2

The recycle ratio is set equal to 5 in all the simulations, as a

trade-off between the higher yield to methanol that is ensured
by a greater recirculation and the higher re-compression costs
upon caused by the increase of RR. The decision to compute the
recycle ratio on the total processed flowrate (and not on fresh syn-
gas only) ensures, as better explained in the next sections, a
methanol production always approaching that predicted by the
thermodynamic equilibrium.
The flash is assumed to operate at 90 bar, since the reactor out-
let pressure is always above such value (pressure drops are limited
at the adopted process conditions and always below 2 % of inlet
reactor pressure). Due to the fact that the flash pressure is lower
than the pressure of the fresh syngas, recycled gases re-
compression is required (as depicted in Fig. 2).
Due to the high pressures involved, non – ideality is accounted
through Peng – Robinson equation of State (Peng & Robinson,
1976).

2.4. Numerical methods

Material and energy balances for the gas and solid phases
(Table 2), together with momentum balance form a set a of partial
differential equations (PDEs) that is solved through gPROMSÒ soft- Fig. 3. (a) one dimensional (1d) and two dimensional (2d) reactor thermal profiles.
Tc is the centerline temperature, Tw is the wall temperature, and Tav is the average
ware by block decomposition and triangularization methods. The temperature. (b) Heat exchange coefficients of the two studied models. U is the
same boundary conditions adopted by Montebelli et al. overall heat transfer coefficient, hw the wall heat transfer coefficient, and keff rad is
(Montebelli et al., 2013), recalled in Appendix B, are utilized. the effective radial thermal conductivity of the gas phase.

7
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

2D model. The average temperature value, Tav , computed as alignment, the corresponding thermal profiles also reveal some dif-
reported in Eq. (12), is also depicted. The profiles are expressed ferences related to the fact that the former is not able to predict the
along the dimensionless axial coordinate, z . local phenomena occurring inside the catalytic bed, such as inter –
R Dt =2 and intra – particle heat and mass transfer. On the other hand, the
ðTðz ; rÞ  p  rÞ dr greater detail of the 2D reactor approach, due to the presence of
Tav ¼ 0
ð12Þ
p  D2t =4 more comprehensive models for gas thermal conductivity and
the capability of estimating axial and radial profiles, more reliably
By looking at 2D thermal profiles, it can be noticed that the predicts its thermal behavior and the corresponding MeOH pro-
value of temperature along the centerline (Tc ) is lower that on ductivity. The former approach can be exploited to gain general
the wall (Tw ) in proximity of reactor entrance (z ¼ 0  0:05). Then, information about the unit process performances, but the latter
the predicted value of Tc increases and overcomes Tw . This phe- is required when more detailed studies, such as analyses of the
nomenon can be explained by recalling that the temperature of reactor dynamic behavior, have to be carried out. For these reasons,
the reacting stream (Tsyngas ) is lower than that of the cooling med- the two – dimensional model is considered suitable for the tran-
ium (Tcool ) at reactor inlet and thus the boiling water flowing in the sient phase simulations and will be used in the next sections.
reactor jacket acts as a heating medium in the first part of the reac-
tor. Points on the centerline heat-up more slowly than those that 3.2. Analysis of the reactor flexibility
are closer to the wall: as a result, the temperature of the centerline
is colder than that of the wall. 3.2.1. Reactor management strategy
The situation swaps when the chemical reaction starts: the cen- Once a suitable steady-state mathematical model has been
terline temperature becomes higher than wall temperature developed for MeOH reactor and its performances have been
(Tc > Tw ). When z > 0:05, because the heat released by the reac- assessed for the baseline case, the performances of the unit in
tion becomes no longer negligible, the gas temperature exceeds the presence of extra H2 and CO2 have been investigated. The pur-
the value of the cooling medium and boiling water actually acts pose of the inquiry is to understand whether the reactor of interest,
as a cooling medium. Once again, due to the greater distance from normally fed with syngas obtained from biogas, can manage the
the steam flux, points lying on the centerline are cooled down less presence of extra hydrogen and carbon dioxide. The introduction
effectively than those that are closer to the wall: this results in a of such streams entailed the presence of a transient regime before
higher centerline temperature. reaching the steady state. For this reason, an adequate analysis of
Fig. 3 an also depicts 1D reactor model temperature profiles MeOH reactor also had to include the study of the unit dynamic
along the axial coordinate. This profiles can be compared with phase. Accordingly, a transient model is implemented, so as to cap-
the average temperature of the 2D model. The two predictions ture the thermal behavior of the unit during the transient phase
are almost overlapped at z ¼ 0  0:05; for z > 0:05, the one – and to monitor the methanol productivity throughout the process.
dimensional case gives a remarkably higher estimate of the tem- The adopted simulation procedure is depicted in Fig. 4 and is
perature peak (T max1D ¼ 539K vs.Tmax 2D ¼ 530 K). Beyond the tem- handled in the following way:
perature peak, the cooling rate of the 1D model is higher than that
of the 2D one, so that the plateau temperature is reached earlier by
At the beginning of the simulation, the system works at the
the one dimensional model. Such outcome can be explained by steady – state with only syngas entering and no H2 or CO2 co-
looking at the overall heat transfer coefficients (U) in the two cases. feed. Such condition is kept represented in Fig. 4 for time values
As far as the 1D model is concerned, a constant value of below 0 h (t ¼ 1 h to 0 h);
280W=m2  K is set along the reactor (Rinaldi et al., 2023). Con-
At t ¼ 0 h, 50 mol=s of H2 and 16 mol=s CO2 are added (step-
versely, the effective radial conductivity (keff; rad ) and the wall heat addition) to the fresh syngas at the reactor inlet. The new con-
transfer coefficient (hw ) estimated with the 2D model allow to dition is kept for a long time (3:8  103 s, around 1:06 h), so as to
compute a fictious overall heat transfer coefficient (calculated a ensure the achievement of new steady – state conditions. In this
posteriori according to the definition given by Crider and Foss phase, the amount of recycled gases is calculated by Eq. (10)
(Crider & Foss, 1965), Eq. (13)), which varies along z .
 1
1 Dt
U¼ þ ð13Þ
hw 6:13  keff;rad

When z < 0:18, the value of U in the 2D model (dashed curve in

Fig. 3b) is higher than that of the 1D case (continuous line of
Fig. 3b). As a consequence, when reactor temperature increases
due to the heat released by the reaction, the cooling rate of the
2D model is faster and thermal peak is achieved at lower temper-
ature values (in other words, the cooling process is more effective).
On the other hand, when z > 0:18, the overall heat transfer coef-
ficient of the 1D model is higher than the estimate for the 2D case:
this leads to a more rapid heat removal. This is the reason why the
flat profile is achieved earlier in the one dimensional simulation.
Fig. 3b also displays the wall heat transfer coefficient of Eq. (13).
It can be observed that hw and U have a similar trend, with a peak
at z 0:04 and a decreasing trend afterwards. This is due to the
fact that the remaining contribution of Eq. (13), the effective radial
conductivity, is almost constant along the catalytic bed and equal
to 3:7 W=m  K.
Fig. 3a shows a general good agreement between the one –
dimensional and the two – dimensional models. In spite of such Fig. 4. CO2, H2, recycle, and purge molar flows trends during the transient phase.

8
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

(hence, RR still depends only on fresh syngas molar flowrate). exploited, methanol production significantly grows: if one keeps
The molar flowrate of recycled gases is constant and equal to the definition of recycle ratio given in Eq. (10), MeOH outlet flow-
381:4 mol=s; rate goes from 29:2 mol=s in the case depicted in Fig. 5a (i.e., with

At t ¼ 1:06 h ð3:8  103 sÞ, the recycle ratio definition is changed 25 mol=s of H2 and 8 mol=s of CO2) to 41:1 mol=s, when 100 mol=s
according to Eq. (11). In order to do so, the purge flowrate is set of H2 and 32 mol=s of CO2 are employed (Fig. 5e).
equal to zero, until the recycled stream is such that it satisfies The adoption of the new recycle ratio definition, given in Eq.
Eq. (11). Steady – state performances are achieved by ensuring (11), enhances such trend even more, as it allows more unreacted
a long time in the new condition (8  103 s, around 2:22 h). The gases to be converted to MeOH. The difference in MeOH productiv-
final flowrate of recycle gases is 711:5 mol=s; ity due to the change of RR definition is small in Fig. 5a, where the
chemical reactor is operated close to the equilibrium (the distance

At t ¼ 3:28 h ð1:18  104 sÞ, the fluxes of hydrogen and carbon
between the dashed and the continuous curve is modest), but it
dioxide are instantaneously turned off and the recycle ratio is
remarkably increases as the amounts of hydrogen and carbon diox-
set back to the value given by Eq. (10). The amount of recycle
ide are incremented. In Fig. 5e, for instance, when the old recycle
gases that is required at reactor inlet increases up to a point
ratio definition is adopted, MeOH productivity is 41:1 mol=s, but
where the available purge is no longer sufficient to satisfy Eq.
it increases to 51:9 mol=s when RR is defined as in Eq. (11). Thus,
(10). When the purge flow approaches zero (1  105 mol=s), it
the introduction of H2 and CO2 is always beneficial for the system,
is automatically set equal to zero to increase the recycle flow-
but it leads to a greater and greater distance from the chemical
rate, up to the moment when Eq. (10) is satisfied again;
equilibrium, due to the higher amount of gases that are processed

The system is now at the initial operating conditions. A suffi-
per mass of catalyst. The issue, which is expected considering that
ciently long time (9:8  103 s, around 2:72 h) is set so that steady the methanol reactor is designed on the basis of the syngas pro-
– state conditions are reached. The final simulation time is set at duced by the steam reforming, can be solved by increasing the
t ¼ 6 h. recirculation inside the system.
It should also be pointed out that the total time required by the
The system of interest undergoes three transient phases: H2 and system to undergo the entire transient phase (i.e., from the intro-
CO2 introduction, RR definition change, and H2 / CO2 co-feed duction of H2 and CO2 fluxes to the recycle ratio definition change
interruption. and back to the initial conditions) is affected by the amount of uti-
The first one is depicted in Fig. 4 from time t ¼ 0 h up to lized hydrogen and carbon dioxide only in a limited manner.
t ¼ 1:06 h. As a consequence of the insertion of H2 and CO2 gaseous Indeed, if one goes from 25 mol=s to 100 mol=s of H2 (i.e.,
streams, the amount of purge rapidly increases and reaches a peak þ300%), the corresponding total time required by the system tran-
of 97:9 mol=s at t ¼ 80 s ð0:02 hÞ. The purge progressively goes sient phase increases from 5 h to 7 h (i.e., þ40%). In other words,
down: after 1 h, its flowrate diminishes up to 11:2 mol=s. The the unit can fast manage the transient behavior, even in process
steady – state behavior is reached at time t ¼ 1:06 h, with a final conditions that are far away from the ones it is designed for. On
purge flow equal to 11:4 mol=s. top of that, a remarkable aspect is the fact that the reactor under
When the new definition of RR is adopted (Eq. (11)), the system study always adapts to the imposed changes in a short time
of interest undergoes the second transient phase. Steady – state (within a few hours).
conditions are achieved fast: the amount of recycled gases goes A key parameter to look at is the evolution of the temperature
from 381:4 mol=s to 711:5 mol=s in around 5200 s ð1:4 hÞ. The cor- profile along the reactor. Temperature profiles are depicted in
responding total simulation time is t ¼ 2:49 h ð8:96  103 sÞ. Fig. 5b, in Fig. 5d, and in Fig. 5f. The depicted trends represent
In the third transitory phase, when the fluxes of hydrogen and the steady – state thermal profiles of the baseline case (no H2/
carbon dioxide are turned off, the amount of condensate with- CO2 introduction, continuous curves), of H2/CO2 insertion with
drawn from the flash unit grows and the quantity of available gases Eq. (10) as recycle ratio definition (dashed curves) and those of
for recirculation diminishes, up to a point when they are no longer H2/CO2 addition with Eq. (11) as RR definition (dotted curves). Dur-
enough to satisfy Eq. (10). MeOH condensate reaches a maximum ing the system dynamic phase, thermal peaks not higher than
equal to 22 mol=s at time t ¼ 3:44 h ð1:24  104 sÞ. The correspond- those depicted in Fig. 5b, Fig. 5d, Fig. 5f are achieved.
ing value of recycled gases is 248:5 mol=s ðRR ¼ 3:2Þ. In order to The further addition of H2 and CO2 streams increases the heat
adjust the recycle ratio (that should be constant and equal to 5), released by the reaction (MeOH production is greater) and hence
the purge is set equal to zero until the recycled gases are enough one would expect temperature peaks to grow as the inlet flowrate
to satisfy Eq. (10) again. augments. This is true if the continuous (i.e., no H2/CO2 utilization)
and the dashed curves (i.e., extra H2/CO2) in Fig. 5b and Fig. 5d are
compared, but does not hold any longer in Fig. 5f. The reason
3.2.2. Reactor sensitivity to reactants flowrates behind such difference is that the extra H2/CO2 induces two other
A sensitivity analysis on CO2 and H2 flowrates is conducted in effects on methanol reactor:
order to understand the reactor behavior in the presence of a dif-
ferent syngas composition and amount. The effect of hydrogen
On the one hand, the increase of flowrate leads to an increase in
and carbon dioxide amounts on the reactor is studied by varying the Reynolds number (Re) and the value of the overall heat
H2 between 25 mol=s and 100 mol=s and CO2 between 8 mol=s transfer coefficient U increases. This makes the cooling process
and 32 mol=s. A ratio of 3 mol of hydrogen per mole of CO2 is kept even more effective than in the base case and reduces the max-
constant, due to the stoichiometry of methanol reaction (Vanden imum temperature achieved. Further details on the increase of
Bussche & Froment, 1996). The results of such analyses are overall heat transfer coefficient are reported in Appendix D;
reported in Fig. 5, in terms of MeOH productivity and temperature
On the other hand, the greater quantity of gases to be processed
profiles along the reactor. causes a redistribution of the heat released along the catalytic
As far as methanol production is concerned, it is noticed that bed. As a consequence, the temperature peak is shifted right-
the increase of inlet flowrate leads to a boost of produced MeOH wards. In this way, the final flat temperature profile is achieved
in any case. When no H2 and CO2 are introduced in the system, farther from the reactor inlet than in the baseline case. Further
the molar flowrate of liquid methanol getting out the flash unit details on heat dilution effect are provided in Appendix E.
is around 21:5 mol=s. Conversely, when the two fluxes are
9
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

Fig. 5. Effects of H2 and CO2 fluxes on the reactor in terms of methanol productivity and temperature profiles. (a) Methanol productivity with 25 mol/s of renewable H2 and
8 mol/s of recycled CO2 and (b) corresponding temperature profiles. (c) Methanol productivity with 50 mol/s of renewable H2 and 16 mol/s of recycled CO2 and (d)
corresponding temperature profiles. (e) Methanol productivity with 100 mol/s of renewable H2 and 32 mol/s of recycled CO2 and (f) corresponding temperature profiles.

Hence, when extra hydrogen and carbon dioxide are introduced, Fig. 5b, c, and d, the maximum temperature in the reactor is lower
the rise in released heat is mitigated by the higher exchange coeffi- than in the other two cases (continuous and dashed curves). How-
cients and the spreading of reaction heat across the reactor. The ever, the prevailing effect of heat dilution increases the distance
phenomenon is even more evident when the new definition of recy- from chemical equilibrium. Indeed, the gap between the cooling
cle ratio (Eq. (11)) is adopted: by looking at the dotted curves in medium and the reacting stream temperature in the exit section

10
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

Fig. 6. Thermal profiles along methanol reactor at different tube lengths. The limit of temperature is set according to Montebelli et al. (Montebelli et al., 2013). (a) case of no
hydrogen and carbon dioxide addition (b) case of H2-rich conditions. MeOH production, thermal apices, and outlet temperatures of (c) case without H2 and CO2 addition (d)
case with H2 and CO2 addition.

of the reactor increases as the amount of hydrogen and carbon diox- employing a smaller scale unit. A threshold temperature limit of
ide goes up, as one can notice by comparing Fig. 5b, d, and f. 540 K is set, as higher values would lead to the formation of unde-
The sensitivity analysis conducted on H2, and CO2 quantities sired species (Montebelli et al., 2014) (Spath & Dayton, 2003); cat-
shows that the reactor well tolerates up to 100 mol=s of H2 and alyst sintering occurs starting from 560 K (Spath & Dayton, 2003).
32 mol=s of CO2. The adoption of the new recycle ratio definition The effects of tubes length reduction in terms of temperature
allows the reactor to be run close to the chemical equilibrium even profiles are depicted in Fig. 6, both in terms of temperature profiles
in the presence of significant amounts of carbon dioxide and (Fig. 6a and b), and methanol productivity, outlet reactor temper-
hydrogen, with temperature hotspots that are always below 535 K. atures, and temperature hot-spots (Fig. 6c and d). Upon shortening
It is believed that the any further increase of the introduced H2 the tubes, both the maximum temperature reached in the reactor
and CO2 could not be compliant with the engineering design of the and the distance from chemical equilibrium increase. Fig. 6a shows
synthesis loop. The plant herein simulated, in fact, is designed the results of the analysis in the case no H2/CO2 are introduced in
(Rinaldi et al., 2023) to have an inlet flow rate equal to the system. It can be noticed that it is possible to halve reactor
457.8 mol/s (76.3 mol/s of fresh syngas and 381.5 mol/s of recycled tubes length without any relevant effect on the process: the max-
gases, see Appendix C): accordingly, the pipelines, the re- imum achieved temperature is 533 K, and the difference from the
compression system, the vessel (flash unit), and the valves are baseline case (that produces 21.5 mol/s of MeOH, Fig. 6c) and the
sized for these conditions. The use of the plant at conditions that equilibrium (21.6 mol/s) is negligible: 0:33% and 0:43%, respec-
are too distant from the ones the plant is sized for, would lead to tively. If the length is shortened up to 1=4 of the original value, on
practical issues and limitations in the management of the system the contrary, the thermal profile reaches the selectivity limit
itself. For example, when 32 mol/s of CO2 and 100 mol/s of H2 (538 K) and MeOH molar productivity remarkably diminishes:
are employed, the total reactor inlet mole flow rate increases up 1:3% w.r.t. the baseline (the outlet predicted methanol mole flow
to 1249.7 mol/s, which is almost three times as much as the inlet is 21.3 mol/s, as reported in Fig. 6c) case and 1:5% w.r.t equilib-
flow the plant is designed for. Moving beyond this limit would lead rium. Due to the risk of selectivity loss, it is thus not possible to
to results that are valid only theoretically, but which could never scale down the reactor of interest by decreasing its tubes length
be feasible in a real plant. For this reason, it is decided not to more than by a factor of four.
extend the sensitivity analysis further. When the flux of hydrogen and carbon dioxide is introduced
(see Fig. 6b and d), the effect of heat distribution along the catalytic
bed leads to lower temperature peaks and to a higher distance
3.2.3. Reactor sensitivity to tubes length and number
from the chemical equilibrium. At 1=2 of original tubes length,
Some parametric sensitivity analyses on reactor downscaling
the maximum achieved temperature is 527 K (it is 533 K in the
are conducted in order to assess the possibility and the limits of
11
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

Fig. 7. Thermal profiles along methanol reactor at different tubes number employed in the unit, when no H2 and CO2 are introduced in the reactor (a,c) and when extra H2 and
CO2 are introduced in the reactor (b,d). The limit temperature is set according to Montebelli et al. (Montebelli et al., 2013). The gas hourly space velocity, GHSV, is measured in
Nm3/m3CAT/h.

corresponding case of Fig. 6a), but the decrease of MeOH produc- cases (GHSV ¼ 8000, i.e., 926 tubes; GHSV ¼ 12000, i.e., 617 tubes;
tivity w.r.t the equilibrium is 1:5% (vs. 0:43%). The situation embit- GHSV ¼ 16000, i.e., 463 tubes) in Fig. 7a in terms of thermal profiles
ters in the case tubes are shortened by a factor of 1=4: in this case, and in Fig. 7c in terms of methanol production, maximum, and outlet
MeOH production diminishes by 3:6% w.r.t. the equilibrium and temperatures. When H2 and CO2 are introduced into the reactor, the
reaches the value of 28.5 mol/s (Fig. 6d), which is more than twice number of tubes is kept constant w.r.t. Fig. 7a, but the augmented
as much as in the absence of H2 and CO2. inlet flow rate leads to a higher GHSV (that is, GHSV ¼ 5730 for
It is worth to notice that, when the length of the tubes is further 1852 tubes; GHSV ¼ 11461 for 926 tubes; GHSV ¼ 17191 for 617
shortened, the thermal peak is achieved at higher and higher tem- tubes, and GHSV ¼ 22921 for 463 tubes), as shown in Fig. 7b and
peratures and the distance from the equilibrium becomes greater in d.
(the gaseous stream is still in kinetic regime when it leaves metha- It can be clearly noticed that the temperature keeps always
nol reactor). At L=16, the hot spot is not even achieved (this is why below the threshold of 540 K in any case. Contrary to what one
the maximum temperature reached for L=16 is lower than that of could expect, the addition of hydrogen and CO2 does not lead to
the L=8 case, Fig. 6d): as a consequence, a remarkable amount of an exacerbation of thermal profiles, because the effect of reaction
unconverted gases needs to be recycled to the reactor to obtain redistribution along the reactor prevails and temperature peaks
acceptable yields, with much higher compression costs. Indeed, are lower and shifted rightwards as the gas hourly space velocity
in these conditions, MeOH mole flow drops down to 17.8 mol/s is increased. As for MeOH production, when no H2/CO2 are
in the absence of H2 and CO2 (17:2% with respect to the 6 m case) employed, a contraction of 0:5% w.r.t. the equilibrium is observed
and to 14.4 mol/s when H2 and CO2 are employed (69:7% with in the case at GHSV ¼ 8000, corresponding to a MeOH production
respect to the 6 m case). of 21.4 mol/s (see Fig. 7c). The reduction increases to 1:7% at
The study of the reactor transient phase shows similar thermal GHSV ¼ 16000 (21.2 mol/s of H2, as per Fig. 7c). In the cases repre-
profiles: in other words, while going from no H2 and CO2 addition sented in Fig. 7b, the loss in methanol productivity is more remark-
to an increased inlet flowrate, the reactor thermal behavior is very able, due to the increase in the inlet mole flow rate: 1:7% w.r.t.
close to the one depicted for the steady state conditions in Fig. 6a the thermodynamic equilibrium at GHSV ¼ 11461 (which corre-
and b. sponds to the reactor configuration of GHSV ¼ 8000 in Fig. 7a, with
In view of a compact – scale process, an alternative to tubes 926 tubes): in this case the produced MeOH is 29 mol/s, as indi-
length decrease is the reduction of tubes number, as reported in cated in Fig. 7d. When the GHSV ¼ 22921 (namely, the reactor
Fig. 7. The reduction of tubes number (and hence the utilized cat- with 463 tubes), the decrease in methanol production with respect
alyst) is accounted by means of the gas hourly space velocity to the equilibrium is 5:9%, and corresponds to 27.8 mol/s as indi-
(GHSV, measured in Nm3/m3CAT/h). The baseline case, corresponding cated in Fig. 7d.
to 4000 Nm3/m3CAT/h and 1852 tubes, is compared with three other
12
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

By comparing the two adopted solutions, it can be observed that cofed H2 and CO2, but the temperature approaches the selectivity
the reduction of tubes length always leads to issues about thermal limit (538 K) in the hot-spot. Production losses are more pro-
management of the reactor (i.e., higher temperature peaks), even if nounced when H2 and CO2 are introduced: the product mole flow
the decrease in MeOH productivity is moderate. On the other hand, goes from 29.4 mol/s in the baseline case to 28.5 at 1=4 tubes
the sensitivity analysis on tubes number shows that temperature is length (3:6%). Conversely, in the latter case (namely, tubes num-
constantly below the selectivity limit (540 K), but higher losses in ber decrease), the temperature is always under control, but the
methanol production are predicted (up to 5:9% w.r.t. the equilibrium). produced methanol decreases more steeply: when 1=4 of the orig-
Issues related to methanol productivity do not allow the reactor inal tubes number is used and H2 and CO2 are introduced, MeOH
to be scaled down more than 1=4 of initial tubes number. Any fur- productivity drops to 27.8 mol/s vs. 29.4 mol/s of the baseline case.
ther downscaling should be addressed through the use innovative Any further reduction of reactor dimension is not possible by
catalyst formulations, such as the use of highly conductive struc- means of packed beds, due to the worsening of the convective heat
tured catalysts (Montebelli et al., 2013), where the scale- transfer. A viable alternative could be the use of highly conductive
intensive conduction within the thermally connected matrix of a structured reactors or novel chemical reactor design.
solid substrate (honeycomb monolith, open cell foams, or POCS)
is exploited (Tronconi et al., 2014) or by means of innovative reac- Data availability
tor configurations.
Data will be made available on request.
4. Conclusions
Declaration of Competing Interest
The study here presented analyses the steady state and tran-
sient performances of a multi–tubular fixed bed reactor for metha- The authors declare that they have no known competing finan-
nol synthesis, installed in a plant designed to synthesize methanol cial interests or personal relationships that could have appeared
from biogas, where the extra CO2 contained in the biogas, instead to influence the work reported in this paper.
of being separated from the biomethane and vented so to achieve
Acknowledgement
a module close to 2 in the bio-syngas obtained via biogas stream
reforming (baseline case), is added with green H2 when the leve-
This research is part of the Energy for Motion project of the
lized cost of hydrogen is low and converted into methanol. This
Department of Energy of Politecnico di Milano, funded by the Ital-
is a novel concept, never studied before.
ian Ministry of University and Research (MUR) through the Depart-
Accordingly, the main goal of this work was that of exploring
ment of Excellence grant 2018 – 2022.
the possibility to operate the methanol synthesis loop in two dif-
ferent operating conditions, i.e., with a feed of fresh bio-syngas
Appendix A
only (case 1, to be employed when the LCOH is high) or with an
inlet stream enriched in renewable hydrogen and excess carbon
Kinetic model
dioxide (case 2, to be employed when the LCOH is low), and to
manage the transient operations during the switch from case 1
MeOH synthesis kinetics are the ones proposed by Bussche et al.
to case 2 and vice versa.
(Vanden Bussche & Froment, 1996). Due to the high pressure pro-
The simulated synthesis loop comprises the chemical reactor,
cess, the kinetic model proposed by the author is corrected through
the separation of the liquid and the gaseous streams by means of
the use of fugacities instead of partial pressures (Peter et al., 2012).
a flash unit, and the recycle of unconverted reactants. A two –
The reactions involved are CO2 hydrogenation (Eq. (A1)) and the
dimensional heterogeneous reactor model is adopted, in order to
reverse water gas shift (Eq. (A2)). f i is the i-th component fugacity,
account for local heat and mass transfer phenomena, while the
k1 and k2 are the reaction rate constants, Keq;1 and Keq;2 are the
flash unit is simulated through a vapor–liquid phase equilibrium.
equilibrium constants and Kads;1 , Kads;2 and Kads;3 are the adsorption
The sensitivity analysis on the reactor loop, conducted upon
constants. The formation of other species, such as ethanol, is
varying the reactor feed in the range 8  32 mol=s of CO2 and
assumed negligible.
25  100 mol=s of H2, show that the system can manage the change
 
f H O f CH OH
from one operating condition to another in around 2 h, even in the
 k1  f CO2  f H2 1  Keq;1
1 2
3
3

presence of inlet flow rates that are remarkably higher than those mol f f H2 CO2
rMeOH ¼ pffiffiffiffiffiffiffi
the plant was originally sized for (up to + 200 % w.r.t. the baseline kgCAT  s fH 3
1 þ Kads;1  fH
2O
þ Kads;2  f H2 þ Kads;3  f H2 O
case). Upon introducing H2 and CO2, the MeOH production always 2

increases, going from 21:5 mol=s in the baseline case, up to ðA1Þ

41:1 mol=s, when 32 mol=s of CO2 and 100 mol=s of H2 are cofed. h i

 f H O f CO
k2  f CO2  1  Keq;2  f H2 f CO
In spite of this, upon increasing the H2 and CO2 flowrates at the mol
reactor inlet, the amount of unconverted reactants at the reactor rrWGS ¼ pffiffiffiffiffiffiffi
2 2

kgCAT  s fH O
1 þ Kads;1  f H2 þ Kads;2  f H2 þ Kads;3  f H2 O
outlet increases as well. To manage this aspect, a new way of 2

managing the recycle (i.e. a new definition of recycle ratio) should ðA2Þ
be adopted. The model shows that the system adapts to the new
The reaction rate constants, k1 and k2 , and the adsorption con-
definition of RR in 1:6 h, resulting in a remarkable boost of MeOH
stants, Kads;1 and Kads;2 (generically indicated as ji ) are expressed as
of productivity (reaching 51:9 mol=s) which is always close to that  
predicted by the thermodynamic equilibrium). reported in Eq. (A3), with Ai pre-exponential factor and Bi activa-
In view of a compact – scale process, the possibility of shorten- tion energy, adsorption enthalpy or a combination of those. R is
J
ing the tubes length or decreasing the tubes number is also ana- the ideal gas constant (8:314 molK)
lyzed. In the former case, MeOH production is subjected to 0
1
moderate losses but high temperature hotspots: when the reactor  B
ji ¼ Ai  exp@ A
i
ðA3Þ
length is decreased to 1=4 of its original value, the produced MeOH RT
is 21.3 mol/s (vs. 21.5 mol/s of the baseline case, 1:3%) without

13
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

The corresponding values of the pre-exponential factors, activa- reported in Table A2. The superficial gas velocity, v; is computed
tion energy and adsorption enthalpies are reported in Table A1. The as the ratio of the mass flow rate per unit area (Wt ) over the gas
equilibrium constants, Keq; and Keq;2 , are taken from Graaf et al. mixture density (qgas ), as reported in Eq. (A9), where m
_ is the mass
(Graaf, et al. 1989) as reported in Eq. (A4) and Eq. (A5), where T flow rate going through the reactor, Nt is the tubes number, and AC
is the temperature. is the cross-sectional area of the single tube.
3066 _
log10 Keq;1 ¼  10:592 ðA4Þ Wt m
T½K v¼ ¼ ðA9Þ
qgas AC  qgas  Nt
 
1 2073
log10 ¼ þ 2:029 ðA5Þ The gas–solid material transfer coefficient, Km;i , is computed as
Keq;2 T½K reported in Eq. (A10), where Shi is the Sherwood number of species
i and dp is the particle diameter.
Transport correlations Shi  Di; mix Shi  Di; mix
Km;i ¼ ¼ ðA10Þ
dpe dp
The transport properties (namely, gaseous mixture viscosity,
lgas , thermal conductivity, kgas , and density, qgas ), together with The last equality holds because the catalyst shape is assumed to
the heat transfer properties (specific heat at constant pressure, be spherical. The Sherwood number is calculated according to the
cp;gas ) utilized in the set of equations to model the chemical reactor correlation proposed by Pfeffer (Pfeffer, 1964), as reported in Eq.
(Eqs. (3)-(8)) are taken from Multiflash 3.9 package of gPromsÒ. (A11).
The viscosity, the density, and the specific heat are modelled
Shi ¼ 1:26  Redpe  Sci1=3
through the Pedersen’s model (Pedersen et al., 1984), while the !
thermal conductivity is based on the STRAPP approach (Huber, 1  ð1  eÞ5=3

2007). 2  3  ð1  eÞ1=3 þ 3  ð1  eÞ5=3  2  ð1  eÞ2
The material balance of the chemical reactor depends upon the
definition of the i-th component effective radial diffusivity of the  Redpe ðA11Þ
gas phase (Deff;rad i ), the gas–solid mass transfer coefficient (Km;i ), Sci is the dimensionless Schmidt number of species i, computed
and the catalyst volumetric surface (av ). The effective radial diffu- in Eq. (A12), where lgas is the gas mixture viscosity, while Redpe is
sivity is computed as reported in Eq. (A6), where e is the reactor
the particle Reynolds number, calculated in Eq. (A13).
void fraction, Di; mix is the i-th component diffusion coefficient in
the gaseous mixture, dpe is the diameter of a sphere with the same lgas
Sci ¼ ðA12Þ
volumetric surface as the pellet (in the case of interest, it coincides qgas  Di; mix
with the pellet diameter dp ) and v is the gas superficial velocity.
pffiffiffi Wt  dpe
Deff;rad i ¼ e  ð e  Di; mix þ 0:1  dpe  vÞ ðA6Þ Redpe ¼ ðA13Þ
lgas
The diffusion coefficient of the i-th component in the mixture is
calculated as reported in Eq. (A7), according to the Blanc’s correla- The catalyst volumetric surface, described in Eq. (A14), is the
tion, where Di; q is the binarydiffusion coefficient and xi is the i-th ratio particle surface (Sp ) to particle volume (V p ).
component mole fraction.
Sp  ð1  eÞ p  dp
2
!1 6
X
NC av ¼ ¼  ð1  eÞ ¼  ð1  e Þ ðA14Þ
xi Vp d3p d
Di; mix ¼ ðA7Þ p 6
p

i¼1ði–qÞ
Di; q
The balance inside the catalytic pellet depends upon the defini-
The binary diffusion coefficient Di; q is calculated according to tion of the intra-particle diffusion coefficient of the i-th species,
the Fuller’s correlation) (Fuller et al., 1969), as written in Eq. (A8) Deff; i . This parameter is calculated according to the Bosanquet for-
mula (Bosanquet, 1944), as reported in Eq. (A15).
0:0143  T1:75
Di;q ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P  ðA8Þ ep
1=3 PNC 1=3 2 s
p 1 þ 1 
2000
i¼1 Ni
NC
þ j¼1 Nj
Deff;intr i ¼ ðA15Þ
MMi MMq
1
DK; i
þ Di;1mix

T is the gas phase temperature, p is the pressure, MMi and MMq ep is the pellet porosity and s the particle tortuosity. The
are the i-th and q-th species molar masses (respectively), and Ei is numerical value of the ep =s ratio is assumed to be equal to
the i-th species diffusion volume. The utilized values of Ei are 0.123, as proposed by (Hartig & Keil, 1993). DK; i is the Knudsen dif-
Table A1
Pre-exponential factors, activation energies, and adsorption enthalpies used to model the kinetics of methanol reactor.
         
Parameter
AMeOH BMeOH ArWGS BrWGS AAds;1 BAds;1 AAds;2 BAds;2 AAds;3 BAds;3

Units mol kJ mol kJ ½ kJ 1 kJ 1 kJ

mol
kgCAT  s  bar2 mol kgCAT  s  bar kmol bar
0:5 mol bar mol
Value 1:07 36:696 1:22  1010 94:765 3453:38  0:499 17:197 6:62  1011 124:119

Table A2
Pure compounds diffusion volumes and molar masses, as utilized to estimate the binary diffusion coefficient of Eq. (A8).

Species CO CO2 H2 H2O CH3OH Ar N2 CH4

Ei 18.9 26.9 7.07 12.7 29.901 16.01 17.9 24.42
MM 28.01 44.01 2.016 18.015 32.04 39.948 14.0 16.04

14
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

fusivity, calculated as per Eq. (A16) (according to the approach pro- Nu ¼ 1:26  Redpe  Pr1=3
posed by (Bosanquet, 1944)). 0 !1=2 12=3
1  ð1  eÞ5=3
sffiffiffiffiffiffiffiffiffiffiffi @  Redpe A
2  3  ð1  eÞ1=3 þ 3  ð1  eÞ5=3  2  ð1  eÞ2
T
DK; i ¼ 9700  rp  ðA16Þ ðA20Þ
MMi
The energy balance at the gas solid inter-phase, describing the
The value of the average pellet porosity, rp , is assumed to be
9 heat transport of the gas to the catalyst edge, depends upon the
equal to 20  10 m, as proposed by (Graaf et al., 1990).
definition of the reaction enthalpies. Since two reactions occur in
The energy balance at the gas phase depends upon the defini-
the system (i.e., methanol synthesis and reverse water gas shift
tion of the effective radial thermal conductivity (keff;rad ) and the
reaction), two values are required. The employed reaction enthal-
convective heat transfer coefficient (hconv ). The effective radial pies are listed in Table A3.
thermal conductivity of the gas phase is derived as suggested by The energy balance boundary condition at the wall (r ¼ Dt =2,
(Specchia & Sicardi, 1980); the outcome is reported in Eq. (A17). see Appendix B) is based on the definition of a wall heat transfer
0 1
coefficient, hw ; its definition is taken from the work by (Specchia
B 1e Redpa  Pr C & Sicardi, 1980), as reported in Eq. (A21)
keff;rad ¼ kgas  B
@e þ þ
 C
A 0 1
0:220  e þ 
2 2 kgas 2
dpa dpv
8:65  1 þ 19:4  
1e
3 kcat Dt dpa
kgas B C
hw ¼  @2  e þ A þ hw; conv ðA21Þ
ðA17Þ dpv 1:58
kgas
0:0024  Dt
dpv
þ 
1
3 kcat
In Eq. (A17), kgas is the gas thermal conductivity, kcat is the cat-
hw; conv is the convective contribution of the wall heat transfer
alyst conductivity (assumed equal to 1 W=m  K), Dt is the tube dia-
coefficient, that can be retrieved from the corresponding correla-
metr, dpv is the diameter of a sphere with the same surface as the
tion proposed by (Specchia & Sicardi, 1980), as shown in Eq. (A22).
pellet. Since, in the case of interest, the catalyst has a spherical
shape, dpa and dpv coincide with the pellet diameter, dp . Redpa is Nu  kgas Nu  kgas
hw; conv ¼ ¼ ðA22Þ
the Reynolds number, calculated with dpa diameter. Due to the dpa dp
spherical shape of the pellet, it is possible to write Redpa as in Eq.
In this case, the Nusselt number, Nu, is defined on the basis of
(A18).
the Re number, as depicted in Eq. (A23).
(
Wt  dpa Wt  dp 0:0835  Redpa Redpa < 1200
Redpa ¼ ¼ ðA18Þ Nu ¼ ðA23Þ
lgas lgas 1:23  Redpa Redpa 1200
The gas–solid transport is accounted by means of a convective The pressure losses are simulated through the Ergun’s equation,
heat transfer coefficient, hconv , whose expression can be retrieved that is based on the definition of friction factor, f m , whose expres-
from the definition of the Nusselt number (Nu), as reported in sion is reported in Eq. (A23). The definition of f m is taken from
Eq. (A19). (Eisfeld & Schnitzlein, 2001).
 2
Nu  kgas Nu  kgas
hconv ¼ ¼ ðA19Þ 1þ 2
ð1  eÞ2
D
dpe dp 3d t ð1eÞ
f m ¼ 23:75  pe
 þ 0:125
dp Wt e3
The last equality holds because dpe ¼ dp . The expression for the lgas
 
Nusselt number is taken from the work of Pfeffer (Pfeffer, 1964), as
1þ 2
reported in. ð1  eÞ
D
3d t ð1eÞ
 pe
2  ðA24Þ
dpe
2 e3
2 Dt
þ 0:77
Table A3
Reaction enthalpies for MeOH synthesis and rWGS reactions.

Reaction Value Unit Reference Appendix B

MeOH synthesis 49430 J (Montebelli et al., 2013)
molK
rWGS 41,120 J
molK
(Laird, 1997) The adopted boundary conditions for the 2D heterogenous
model are reported in Table B1.

Table B1
Boundary and initial conditions for the set of modelling equations of Table 2.

Mass Balances
 
Position t¼0 z¼0 r¼0 r ¼ Dt =2 x¼0 x ¼ DL
Description Initial time Inlet condition Symmetry No flux at the wall Flux continuity No flux
Gas - phase xi ¼ xi IN NEW xi ¼ xi IN @ xi
@r ¼0 @ xi
@r ¼0  
Gas – solid interphase xs;i ¼ xs; IN NEW     
    @ xs;i @ xs;i
Solid phase Km;i xs;i  xi ¼ Deff;intr i    ¼0
@x @x
Energy Balances
 
Position t¼0 z¼0 r¼0 r ¼ Dt =2 x¼0 x ¼ DL
Description Initial time Inlet condition Symmetry Flux at the wall Flux continuity No flux
T ¼ TIN T ¼ TIN @T @ xs;i  
Gas phase @r¼0 hw  ðT  Tcool Þ ¼ keff;rad  
@x
Gas – solid interphase Ts ¼ Ts NEW     
Momentum Balances
Description Initial time Inlet condition Symmetry Flux at the wall Flux continuity No flux
Gas phase p ¼ psyngas    

15
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

When switching from one operating condition to another, the Appendix C

dynamic analysis starts to be conducted, and initial conditions
have to be provided. The corresponding initial conditions of mass The numerical values the streams of Fig. 2 (here reported as
composition of the gas phase (xi IN NEW ), mass composition of the Fig. C1) are given in Table C1, C2, C3 and C4 for the steady-state
gas–solid interphase (xs;i IN NEW ), and temperature at the catalyst cases, with and without H2/CO2 addition.
edge (Ts NEW ) are always represented with the subscript NEW.
The inlet temperature of the mixture never changes and is always
constant (i.e., TIN is always fixed).

Fig. C1. Sketch of the simulated synthesis loop, including methanol reactor, the flash unit, and unconverted gases recycle (with recompression and cooling). The numbers
refer to the streams, whose values are reported in Table C1, Table C2, Table C3, Table C4.

Table C1
Streams of the baseline case, with no H2 and CO2 addition.

Stream T [°C] p ½bar _ CO [%] CO2 [%] H2 H2O CH3OH CH4

F ½mol=s
[Unit] [%] [%] [%] [%]
1 199.6 92.8 76.3 18 11 70 0.16 0 0.8
2 - - - - - - - - -
3 199.6 92.8 457.7 3.4 2.9 71.6 0.1 0.4 21.7
4 240.3 92.8 414.6 0.5 1.3 66.7 2.0 5.6 23.9
5 40 92.8 414.6 0.5 1.3 66.7 2.0 5.6 23.9
6 40 90 414.6 0.5 1.3 66.7 2.0 5.6 23.9
7 40 90 29.9 0.0 0.1 0.5 27.2 72.1 0.0
8 40 90 384.7 0.5 1.4 71.8 0.0 0.5 25.8
9 40 90 3.3 0.5 1.4 71.8 0.0 0.5 25.8
10 40 90 381.4 0.5 1.4 71.8 0.0 0.5 25.8
11 40 92.8 381.4 0.5 1.4 71.8 0.0 0.5 25.8
12 199.6 92.8 381.4 0.5 1.4 71.8 0.0 0.5 25.8

Table C2
Streams of the case with H2 (25 mol/s) and CO2 (8 mol/s) addition.

Stream T [°C] p ½bar _ CO [%] CO2 [%] H2 H2O CH3OH CH4

F ½mol=s
[Unit] [%] [%] [%] [%]
1 199.6 92.8 76.3 18 11 70 0.16 0 0.8
2 199.6 92.8 8 (CO2) + 25 (H2) 0 24.2 75.8 0 0 0
3 199.6 92.8 655.7 2.7 4.6 77.1 0.1 0.3 15.1
4 240.8 92.7 596.9 0.7 2.4 72.2 2.7 5.3 16.6
5 40 92.7 596.9 0.7 2.4 72.2 2.7 5.3 16.6
6 40 90 596.9 0.7 2.4 72.2 2.7 5.3 16.6
7 40 90 45.7 0.0 0.2 0.5 35.0 64.3 0.0
8 40 90 551.2 0.8 2.6 78.2 0.1 0.4 18.0
9 40 90 4.7 0.8 2.6 78.2 0.1 0.4 18.0
10 40 90 546.4 0.8 2.6 78.2 0.1 0.4 18.0
11 40 92.8 546.4 0.8 2.6 78.2 0.1 0.4 18.0
12 199.6 92.8 546.4 0.8 2.6 78.2 0.1 0.4 18.0

16
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

Table C3
Streams of the case with H2 (50 mol/s) and CO2 (16 mol/s) addition.

Stream T [°C] p ½bar _ CO [%] CO2 [%] H2 H2O CH3OH CH4

F ½mol=s
[Unit] [%] [%] [%] [%]
1 199.6 92.8 76.3 18 11 70 0.16 0 0.8
2 199.6 92.8 16 (CO2) + 50 (H2) 0 24.2 75.8 0 0 0
3 199.6 92.8 853.7 2.6 6.3 79.9 0.1 0.3 10.8
4 241.2 92.7 779.4 1.1 3.9 75.0 3.1 5.1 11.8
5 40 92.7 779.4 1.1 3.9 75.0 3.1 5.1 11.8
6 40 90 779.4 1.1 3.9 75.0 3.1 5.1 11.8
7 40 90 61.4 0.0 0.3 0.5 38.8 60.5 0.0
8 40 90 718.1 1.2 4.2 81.3 0.1 0.4 12.8
9 40 90 6.6 1.2 4.2 81.3 0.1 0.4 12.8
10 40 90 711.4 1.2 4.2 81.3 0.1 0.4 12.8
11 40 92.8 711.4 1.2 4.2 81.3 0.1 0.4 12.8
12 199.6 92.8 711.4 1.2 4.2 81.3 0.1 0.4 12.8

Table C4
Streams of the case with H2 (100 mol/s) and CO2 (32 mol/s) addition.

Stream T [°C] p ½bar _ CO [%] CO2 [%] H2 H2O CH3OH CH4

F ½mol=s
[Unit] [%] [%] [%] [%]
1 199.6 92.8 76.3 18 11 70 0.16 0 0.8
2 199.6 92.8 32 (CO2) + 100 (H2) 0 24.2 75.8 0 0 0
3 199.6 92.8 1249.7 2.9 9.9 81.2 0.1 0.3 5.6
4 241.8 92.6 1145.7 1.9 7.4 76.1 3.4 4.9 6.1
5 40 92.6 1145.7 1.9 7.4 76.1 3.4 4.9 6.1
6 40 90 1145.7 1.9 7.4 76.1 3.4 4.9 6.1
7 40 90 91.5 0.0 0.4 0.4 42.4 56.7 0.0
8 40 90 1054.2 2.1 8.0 82.7 0.1 0.4 6.7
9 40 90 12.8 2.1 8.0 82.7 0.1 0.4 6.7
10 40 90 1041.4 2.1 8.0 82.7 0.1 0.4 6.7
11 40 92.8 1041.4 2.1 8.0 82.7 0.1 0.4 6.7
12 199.6 92.8 1041.4 2.1 8.0 82.7 0.1 0.4 6.7

Appendix D reported here corresp to the cases of Fig. 7. The overall heat trans-
fer coefficient is calculated according to Eq. (13).
The behavior of the overall heat transfer coefficient along the It is possible to notice that heat exchange phenomena become
reactor (U) increases as the inlet flowrate augments, as depicted more and more effective with an increasing gas hourly space veloc-
in Fig. D1. Fig. D1A shows the case of no H2 and CO2 addition, while ity, due to the increase of Reynolds number (Re). This leads to
Fig. D1B represents the trend of U in the presence of renewable lower temperature hotspots along the reactor.
hydrogen and carbon dioxide. The heat transfer coefficients

17
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

Fig. D1. Overall heat transfer coefficient at varying gas hourly space velocity. The gas hourly space velocity, GHSV, is measured in Nm3/h/m3CAT. (A) No H2/CO2 addition; (B) H2/
CO2 addition.

Appendix E the integration step by rWGS (in mol=s), and DHrj is the reaction
heat of the j-th reaction (in J=mol).
The trend of reaction heat across the reactor is represented in
Fig. E1. The results reported here correspond to the cases of
dQ_ tube ¼ dF_ CH3 OH  DHr;CH3 OH þ dF_ CO  DHr;rWGS ðE1Þ
Fig. 7. Infinitesimal released thermal power per single tube
(dQ_ ) is computed through, where dF_ CH OH is the net amount of
tube 3
The reaction dilutes along the catalytic bed as the inlet molar
methanol produced in the space integration step (measured in flowrate increases, leading to lower temperature peaks and to a
greater distance from the chemical equilibrium in the exit section
mol=s), F_ CO is the net amount of carbon monoxide produced in
of the reactor.

18
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

Fig. E1. Distribution of reaction heat across the catalytic bed at different gas hourly space velocity. The value refers to the single tube. The gas hourly space velocity, GHSV, is
measured in Nm3/h/m3CAT. (A) No H2/CO2 addition; (B) H2/ CO2 addition.

References Chein, R.Y., Chen, W.H., Chyuan Ong, H., Loke Show, P., Singh, Y., 2021. Analysis of
methanol synthesis using CO2 hydrogenation and syngas produced from
biogas-based reforming processes. Chem. Eng. J. 426,. https://doi.org/10.1016/
Albarelli, J.Q., Onorati, S., Caliandro, P., Peduzzi, E., Meireles, M.A.A., Marechal, F.,
J.CEJ.2021.130835 130835.
Ensinas, A.V., 2017. Multi-objective optimization of a sugarcane biorefinery for
Clausen, L.R., Houbak, N., Elmegaard, B., 2010. Technoeconomic analysis of a
integrated ethanol and methanol production. Energy 138, 1281–1290. https://
methanol plant based on gasification of biomass and electrolysis of water.
doi.org/10.1016/J.ENERGY.2015.06.104.
Energy 35 (5), 2338–2347. https://doi.org/10.1016/J.ENERGY.2010.02.034.
Augelletti, R., Conti, M., Annesini, M.C., 2017. Pressure swing adsorption for biogas
Crider, J.E., Foss, A.S., 1965. Effective wall heat transfer coefficients and thermal
upgrading. A new process configuration for the separation of biomethane and
resistances in mathematical models of packed beds. AIChE J 11 (6), 1012–1019.
carbon dioxide. J. Clean. Prod. 140, 1390–1398.
https://doi.org/10.1002/AIC.690110613.
Badurek, C. A. (2020). biogas | Description, Production, Uses, & Facts | Britannica.
Dixon, A.G., 2012. Fixed bed catalytic reactor modelling—the radial heat transfer
https://www.britannica.com/technology/biogas.
problem. Can. J. Chem. Eng. 90 (3), 507–527. https://doi.org/10.1002/
Battaglia, P., Buffo, G., Ferrero, D., Santarelli, M., Lanzini, A., 2021. Methanol
CJCE.21630.
synthesis through CO2 capture and hydrogenation: Thermal integration, energy
Santos, R. O. dos, Santos, L. de S., & Prata, D. M. (2018). Simulation and optimization
performance and techno-economic assessment. J. CO2 Util. 44,. https://doi.org/
of a methanol synthesis process from different biogas sources. Journal of Cleaner
10.1016/J.JCOU.2020.101407 101407.
Production, 186, 821–830. 10.1016/J.JCLEPRO.2018.03.108.
Bosanquet, C. H. (1944). British TA Report BR-507. https://scholar.google.com/
Eisfeld, B., Schnitzlein, K., 2001. The influence of confining walls on the pressure
scholar?q=C.H. Bosanquet, British TA Report BR-507 .
drop in packed beds. Chem. Eng. Sci. 56 (14), 4321–4329. https://doi.org/
Calbry-Muzyka, A., Madi, H., Rüsch-Pfund, F., Gandiglio, M., Biollaz, S., 2022. Biogas
10.1016/S0009-2509(00)00533-9.
composition from agricultural sources and organic fraction of municipal solid
European Biogas Association. (2021). EBA. GiE Biomethane Map 2020. https://www.
waste. Renew. Energy 181, 1000–1007. https://doi.org/10.1016/J.
europeanbiogas.eu/eba-gie-biomethane-map/.
RENENE.2021.09.100.

19
R. Rinaldi and C.G. Visconti Chemical Engineering Science 272 (2023) 118611

Fuller, E.N., Ensley, K., Giddings, J.C., 1969. Diffusion of Halogenated Hydrocarbons Nguyen, T.B.H., Zondervan, E., 2019. Methanol production from captured CO2 using
in Helium The Effect of Structure on Collision Cross Sections. Ind. Eng. Chem 73 hydrogenation and reforming technologies_ environmental and economic
(11), 3679–3685. https://pubs.acs.org/sharingguidelines. evaluation. J. CO2 Util. 34, 1–11. https://doi.org/10.1016/J.JCOU.2019.05.033.
Gautam, P., Neha, Upadhyay, S. N., & Dubey, S. K. (2020). Bio-methanol as a Pedersen, K.S., Fredenslund, A., Christensen, P.L., Thomassen, P., 1984. Viscosity of
renewable fuel from waste biomass: Current trends and future perspective. crude oils. Chemical Engineering Science 39 (6), 1011–1016. https://doi.org/10.
Fuel, 273, 117783. 10.1016/J.FUEL.2020.117783. 1016/0009-2509(84)87009-8.
Ghosh, S., Uday, V., Giri, A., Srinivas, S., 2019. Biogas to methanol: A comparison of Peng, D.Y., Robinson, D.B., 1976. A New Two-Constant Equation of State. Ind. Eng.
conversion processes involving direct carbon dioxide hydrogenation and via Chem. Fundam. 15 (1), 59–64.
reverse water gas shift reaction. J. Clean. Prod. 217, 615–626. https://doi.org/ Peter, M., Fichtl, M.B., Ruland, H., Kaluza, S., Muhler, M., Hinrichsen, O., 2012.
10.1016/J.JCLEPRO.2019.01.171. Detailed kinetic modeling of methanol synthesis over a ternary copper catalyst.
Graaf, G.H., Scholtens, H., Stamhuis, E.J., Beenackers, A.A.C.M., 1990. Intra-particle Chem. Eng. J. 203, 480–491. https://doi.org/10.1016/J.CEJ.2012.06.066.
diffusion limitations in low-pressure methanol synthesis. Chem. Eng. Sci. 45 (4), Pfeffer, R., 1964. HEAT AND MASS TRANSPORT IN MULTIPARTICLE SYSTEMS. Ind.
773–783. https://doi.org/10.1016/0009-2509(90)85001-T. Eng. Chem. Fundam. 3 (4), 380–383.
Hartig, F., Keil, F.J., 1993. Large-Scale Spherical Fixed Bed Reactors: Modeling and Phillips, V.D., Kinoshita, C.M., Neill, D.R., Takahashi, P.K., 1990. Thermochemical
Optimization. Ind. Eng. Chem. Res 32, 424–437. production of methanol from biomass in Hawaii. Appl. Energy 35 (3), 167–175.
Hartig, F., Keil, F.J., 2002. Large-scale spherical fixed bed reactors: modeling and https://doi.org/10.1016/0306-2619(90)90066-M.
optimization. Ind. Eng. Chem. Res. 32 (3), 424–437. https://doi.org/10.1021/ Rinaldi, R., Lombardelli, G., Gatti, M., Visconti, C.G., Romano, M.C., 2023. Techno-
IE00015A005. economic analysis of a biogas-to-methanol process: Study of different process
Huber, M. L. (2007). NIST Standard Reference Database 4 NIST Thermophysical configurations and conditions. J. Clean. Prod. 136259. https://doi.org/10.1016/J.
Properties of Hydrocarbon Mixtures Database (SUPERTRAPP) Version 3.2 Users’ JCLEPRO.2023.136259.
Guide Based on research sponsored by: The NIST Supercritical Fluid Property Rout, K.R., Hillestad, M., Jakobsen, H.A., 2013. A numerical study of pellet model
Consortium and Standard Reference Data. consistency with respect to molar and mass average velocities, pressure
International Energy Agency. (2019). The Future of Hydrogen – Analysis - IEA. https:// gradients and porosity models for methanol synthesis process: Effects of flux
www.iea.org/reports/the-future-of-hydrogen. models on reactor performance. Chem. Eng. Res. Des. 91 (2), 296–317. https://
Koronaki, I.P., Prentza, L., Papaefthimiou, V., 2015. Modeling of CO2 capture via doi.org/10.1016/j.cherd.2012.09.003.
chemical absorption processes - An extensive literature review. Renewable and Sims, R., Schaeffer, R., Creutzig, F., Cruz-Núñez, X., D’Agosto, M., Dimitriu, D.,
Sustainable Energy Reviews, Vol. 50. Elsevier Ltd., pp. 547–566. https://doi.org/ Figueroa Meza, M.J., Fulton, L., Kobayashi, S., Lah, O., McKinnon, A., Newman, P.,
10.1016/j.rser.2015.04.124. Ouyang, M., Schauer, J.J., Sperling, D., & Tiwari, G., 2014. Climate Change 2014:
Kvamsdal, H.M., Jakobsen, J.P., Hoff, K.A., 2009. Dynamic modeling and simulation of Mitigation of Climate Change -. Chapter 8 - Transport.
a CO2 absorber column for post-combustion CO2 capture. Chem. Eng. Process. Solsvik, J., Jakobsen, H.A., 2011. Modeling of multicomponent mass diffusion in
48 (1), 135–144. https://doi.org/10.1016/J.CEP.2008.03.002. porous spherical pellets: Application to steam methane reforming and
Laird, T. (1997). Ullmann’s Encyclopedia of Industrial Chemistry, 5th Edition VCH: methanol synthesis. Chem. Eng. Sci. 66 (9), 1986–2000. https://doi.org/
Weinheim, Germany. 1996/1997. Section A, 28 vols. Section B, 8 vols. DM 19 10.1016/j.ces.2011.01.060.
400. Organic Process Research & Development, 1(5), 391–392. 10.1021/ Spath, P. L., & Dayton, D. C. (2003). Preliminary Screening – Technical and Economic
OP970020U. Assessment of Synthesis Gas to Fuels and Chemicals with Emphasis on the
Leonzio, G., 2020. Mathematical modeling of a methanol reactor by using different Potential for Biomass-Derived Syngas. http://www.osti.gov/bridge.
kinetic models. J. Ind. Eng. Chem. 85, 130–140. https://doi.org/10.1016/J. Specchia, V., Sicardi, S., 1980. MODIFIED CORRELATION FOR THE CONDUCTIVE
JIEC.2020.01.033. CONTRIBUTION OF THERMAL CONDUCTIVITY IN PACKED BED REACTORS.
Leonzio, G., Foscolo, P.U., 2020. Analysis of a 2-D model of a packed bed reactor for Chem. Eng. Commun. 6 (1–3), 131–139. https://doi.org/10.1080/
methanol production by means of CO2 hydrogenation. Int. J. Hydrogen Energy 00986448008912525.
45 (18), 10648–10663. https://doi.org/10.1016/J.IJHYDENE.2020.01.248. Tronconi, E., Groppi, G., Visconti, C.G., 2014. Structured catalysts for non-adiabatic
Mayyas, A., Ruth, M., Pivovar, B., Bender, G., & Wipke, K. (2019). Manufacturing Cost applications. Curr. Opin. Chem. Eng. 5, 55–67. https://doi.org/10.1016/J.
Analysis for Proton Exchange Membrane Water Electrolyzers. https://www. COCHE.2014.04.003.
nrel.gov/docs/fy10osti/72740.pdf. Vanden Bussche, K.M., Froment, G.F., 1996. A Steady-State Kinetic Model for
MegaMaxÒ 800. (n.d.). Retrieved February 2, 2023, from https://www.clariant.com/ Methanol Synthesis and the Water Gas Shift Reaction on a Commercial Cu/ZnO/
en/Solutions/Products/2019/02/14/13/39/MegaMax-800. Al2O3Catalyst. J. Catal. 161 (1), 1–10. https://doi.org/10.1006/JCAT.1996.0156.
Miltner, M., Makaruk, A., Harasek, M., 2017. Review on available biogas upgrading Vita, A., Italiano, C., Previtali, D., Fabiano, C., Palella, A., Freni, F., Bozzano, G., Pino, L.,
technologies and innovations towards advanced solutions. J. Clean. Prod. 161, Manenti, F., 2018. Methanol synthesis from biogas: A thermodynamic analysis.
1329–1337. https://doi.org/10.1016/J.JCLEPRO.2017.06.045. Renew. Energy 118, 673–684. https://doi.org/10.1016/J.RENENE.2017.11.029.
Montebelli, A., Visconti, C.G., Groppi, G., Tronconi, E., Ferreira, C., Kohler, S., 2013. Wakao, N., Funazkri, T., 1978. Effect of fluid dispersion coefficients on particle-to-
Enabling small-scale methanol synthesis reactors through the adoption of fluid mass transfer coefficients in packed beds: Correlation of sherwood
highly conductive structured catalysts. Catal. Today 215, 176–185. https://doi. numbers. Chem. Eng. Sci. 33 (10), 1375–1384. https://doi.org/10.1016/0009-
org/10.1016/J.CATTOD.2013.02.020. 2509(78)85120-3.
Montebelli, A., Visconti, C.G., Groppi, G., Tronconi, E., Kohler, S., 2014. Optimization Wang, J., Wei, S., Wang, Q., Sundén, B., 2021. Transient numerical modeling and
of compact multitubular fixed-bed reactors for the methanol synthesis loaded model predictive control of an industrial-scale steam methane reforming
with highly conductive structured catalysts. Chem. Eng. J. 255, 257–265. reactor. Int. J. Hydrogen Energy 46 (29), 15241–15256. https://doi.org/10.1016/
https://doi.org/10.1016/J.CEJ.2014.06.050. J.IJHYDENE.2021.02.123.
Mulu, E., M’Arimi, M.M., Ramkat, R.C., 2021. A review of recent developments in Yagi, S., Kunii, D., Wakao, N., 1960. Studies on axial effective thermal conductivities
application of low cost natural materials in purification and upgrade of biogas. in packed beds. AIChE J. 6 (4), 543–546. https://doi.org/10.1002/AIC.690060407.
Renew. Sustain. Energy Rev. 145,. https://doi.org/10.1016/J.RSER.2021.111081
111081.

20