(677c) Regeneration Scale-up Methodology Applied to Bio-Based Isobutanol Dehydration to Linear Butenes

The technology
of isobutanol dehydration to linear butenes has been developed in a joint
collaboration of IFP Energies Nouvelles (France), Total Research &
Technology Feluy (Belgium) and Axens (France). The dehydration reaction takes
place in gas phase with a low water dilution of the alcohol feedstock and is carried
out in a fixed-bed multitubular reactor loaded with a ferrierite (FER)
zeolite-based catalyst. The process has high isobutanol-to-olefins conversion
and yield and strong selectivity to linear butenes. One of the process
drawbacks is the progressive catalyst deactivation due to coke formation and the
respective conversion loss at the end of the catalytic cycle. Therefore, the
process operates in cyclic mode that alternates highly endothermic reaction and
highly exothermic regeneration steps. The proposed presentation describes the
step-by-step regeneration scale-up methodology to reduce the risk involved in
such process scale-up. Although main design criteria are induced by the
endothermic dehydration in order to meet target production, validation of the
adequacy of such scale-up with exothermic regeneration is necessary. Thus, a
sufficient knowledge and control of the various chemical and physical phenomena
occurring during regeneration were acquired. Combustion reactions scheme was
set-up thanks to controlled thermogravimetric experiments at micro scale and
then directly validated by dynamic simulation on pilot unit taking into account
heat transfer phenomena. At last, a specific sizing and rating of process
equipment in non-steady mode was realized to optimize and ensure control of
industrial regeneration protocol. This methodology can be generalized for a wide
variety of chemical processes using heterogeneous catalyst.

The first
step of the regeneration scale-up methodology is the establishment of a
regeneration protocol.  Catalytic tests have
been carried out in a packed bed tubular pilot plant unit consisting of a
tubular reactor filled with catalyst. The regeneration gas temperature has been
maintained by heating the reactor shell with an electrical furnace.  The major
challenge to overcome in the case of FER-based catalyst is its high sensitivity
to regeneration and relatively low capacity to be regenerated. After series of
catalytic tests, we have observed that the optimal composition of the gas for
FER-based catalyst regeneration comprises a variable mixture of nitrogen and
oxygen. Oxygen content varied during regeneration, starting from relatively low
(<1% mol.) and ending at relatively high value (<10% mol.). The temperature
of the regeneration follows a slope-plateau program. It is increased at the end
of protocol in order to complete the coke calcination. Moreover, we have found out
that the presence of a relatively low amount of water in the regeneration gas
provokes performances degradation. The regeneration gas should not contain any
residual free water for better overall catalyst life duration.  As a result,
the regeneration protocol is an operational program that includes a variation
of gas composition and temperature.

The regeneration
protocol described here above has been optimized and tested at the pilot unit
after several catalytic runs with various grades of isobutanol. The protocol
permits to fully restore the catalytic performance and to preserve catalyst
cycle length.

The second
step of regeneration scale-up methodology corresponds to the determination of
coke composition and reactional kinetics of coke combustion via a thermogravimetric analysis (TGA) of coke, in order to enable
control of the exothermic reactions in further extrapolations. Due to this
analysis, we have observed that the coke is mainly composed of two large
hydrocarbon families that burn at different temperature ranges (see Figure 1).

Figure 1. Thermal curves of TGA experiment

Following a
weight loss during a thermogravimetric experiment, activation energies of coke
combustion via Kissinger-Akahira-Sunose equation have been identified. Results obtained
with this equation are coherent with those from bibliographic sources. Moreover,
a TGA experiment allows us to determine partial orders of combustion reactions
by solving a mass balance of reactions. We have compared mass-loss curves
obtained experimentally with those traced with mass balance equation (see Figure 2) and found values of orders of
reaction.

Figure 2. Mass-loss curves of TGA analysis of coke

The third
step of the regeneration scale-up methodology is a dynamic simulation of the catalyst
regeneration in order to confirm the results
obtained experimentally. Reaction kinetic parameters determined by TGA have
been implemented in a 1D plug-flow reactor simulator, simulator with 2D heat
transfer (radiative and conductive) characterization, in order to model coke
combustion in dynamic mode and to trace carbon dioxide/carbon monoxide curves. Two
major hydrocarbon families identified by TGA tests have been represented by mono-aromatic
and poly-aromatic molecules. This simplification does not affect the accuracy
of the results and is globally very robust: curves obtained with the simulator are
coherent with experimental curves. This fact confirms a correctness of dynamic regeneration
modeling (see Figure 3). The
simulator allows to extract a dynamic mass balance data of coke combustion.

Figure 3. Comparison of carbon dioxide/carbon
monoxide curves obtained experimentally and with simulator

Then, the
following hypotheses for process extrapolation to an industrial scale have been
made:

·        
Mass balance data obtained with the dynamic
simulator of the pilot unit is identical for an industrial unit

·        
Mass of coke per mass of catalyst of the pilot
unit is identical for an industrial unit

·        
Mass of catalyst required for an industrial unit
is calculated conserving the same Weight Hourly Space Velocity (WHSV) as for
the pilot unit

The fourth
step of the regeneration scale-up methodology corresponds to transfer of dynamic
mass balance data to static simulation in order to determine heats of coke
combustion. Firstly, we have calculated the
quantity of coke on the catalyst of an industrial unit. Secondly, the regeneration
timeline has been divided into several phases. We have calculated the quantity of
coke burnt at each phase and further determined a mass flowrate of coke per
hour. Lastly, knowing the composition of coke burnt, heats of combustion and
quantities of regeneration gas have been identified.      

The last
step of the regeneration scale-up methodology is the sizing and rating of
process equipment in non-steady modes. Data
obtained in the previous step have been used in order to fix a maximum quantity
of regeneration gas required. It is important to mention that the multitubular
reactor, initially designed to perform an endothermic dehydration reaction,
also allows an efficient heat extraction from the exothermic combustion
reaction during the regeneration phase.

In conclusion, the
above described methodology allows to develop a process of effective FER-based
catalyst regeneration in a multitubular reactor. This methodology can be
applied for other technologies employing a heterogeneous catalyst that requires
periodic regeneration due to coke formation.