Mass Transport and Reactions in the Tube-in-Tube Reactor The MIT Faculty has made this article openly available. Please share how this access benefits you. Your story matters. Citation Yang, Lu, and Klavs F. Jensen. “Mass Transport and Reactions in the Tube-in-Tube Reactor.” Org. Process Res. Dev. 17, no. 6 (June 21, 2013): 927–933. As Published http://dx.doi.org/10.1021/op400085a Publisher American Chemical Society (ACS) Version Author's final manuscript Citable link http://hdl.handle.net/1721.1/92948 Terms of Use Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use.
23
Embed
Mass Tr anspor t and Reactions in the Tube-in-Tube Reactor
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Mass Transport and Reactions in the Tube-in-Tube Reactor
The MIT Faculty has made this article openly available. Please share how this access benefits you. Your story matters.
Citation Yang, Lu, and Klavs F. Jensen. “Mass Transport and Reactions in theTube-in-Tube Reactor.” Org. Process Res. Dev. 17, no. 6 (June 21,2013): 927–933.
As Published http://dx.doi.org/10.1021/op400085a
Publisher American Chemical Society (ACS)
Version Author's final manuscript
Citable link http://hdl.handle.net/1721.1/92948
Terms of Use Article is made available in accordance with the publisher'spolicy and may be subject to US copyright law. Please refer to thepublisher's site for terms of use.
Profile: Residence time = 10 s, saturation fraction = 98.17%.
RESULTS AND DISCUSSION
Model Verification
Analytic solutions and numerical simulations were performed under the same operation
conditions as the experiments23 to determine the saturation fraction of hydrogen dissolved in
DCM as a function of residence time (Figure 3). The predictions obtained from the two different
approaches are within 2% difference of each other, supporting the approximations used in
deriving the analytic solution. The analytic approach is less computationally demanding, and it
provides insights into the distinction between two mass transfer compartments, namely, the
entrance region and the fully-developed region. On the other hand, numerical simulations yield
detailed three-dimensional concentration profiles inside the reactor, and can be extended to
alternative reactor configurations and systems with reactions, as is demonstrated below.
11
Figure 3. Gas saturation profile obtained from theory, simulation and two experimental
methods.23 (Orange line: analytic solution; red square: numerical solution; green diamond: digital
measurement result; blue triangle: burette measurement result.)
Our modeling results are also in good agreement with experimentally measured saturation
values. Figure 3 compares model predictions with both the digital bubble counter measurements
and the conventional glass burette measurements, according to the work of O’Brien et al.23 In
addition to reproducing experimental results, the simulations have the advantage of providing
detailed concentration profiles inside the reactor, which are not accessible experimentally. The
three-dimensional concentration profiles elucidate reactant distributions as a result of the mass
transfer characteristics of the tube-in-tube configuration. Such physical insights are helpful in
predicting scaling up behaviors of the tube-in-tube reactor.
Scale-up Behavior:
In continuous flow chemistry systems, production can be increased by a longer operation time,
but that approach is typically only applicable for small amounts. In order to scale up, the
production rate must be increased, which is proportional to the volumetric flow rate and the
product concentration in flow. The concentration of product is limited by the saturation
concentration of gas in the liquid phase, as is subsequently discussed in the context of reactions
in the tube-in-tube reactor. Here in this section, we focus on the scaling-up of the volumetric
flow rate, Q, without sacrificing mass transfer performance.
12
The volumetric flow rate is dependent on three variables:
(Cross sectional area) (Tube Length) =
Residence TimeQ
. (15)
In order to increase the volumetric flow rate, we can (i) decrease residence time, (ii) increase
tube diameter, and (iii) increase tube length. We examined each of the three factors separately
while keeping the other two constant. For comparison, the three approaches will depart from the
same base case: residence time = 10 s, tube diameter = 6 mm, tube length = 1 m, Q = 1.7 mL/min.
Unless otherwise specified, we limit our discussion in the laminar flow regime (Re < 2,000).
(i) Decrease residence time: Decreasing residence time shortens the time for gas permeation.
When the residence time decreases from 10 s (Q = 1.7 mL/min) to 1 s (Q = 17 mL/min), the
saturation fraction plummets from 98% to 35% (Figure 4). Thus, decreasing residence time is not
an appropriate approach to scale up.
(ii) Increase tube diameter: As mass transfer in the tube-in-tube reactor is dominated by the
diffusion process across the radial dimension, increasing the radial dimensions will lead to a
proportional increase in mass transfer resistance, which compromises the mass transfer
performance. For example, when the tube inner diameter increases from 0.6 mm (Q = 1.7
mL/min) to 3 mm (Q = 42.5 mL/min) and the membrane thickness increases proportionally, the
saturation fraction drops from 98% to 32%.
(iii) Increase tube length: If the tube length is increased, the mass transfer performance will
not be compromised as mass transfer resistance (determined by radial dimensions) and residence
time are both invariant. Hence, the saturation fraction remains the same (Figure 4). The trade-off
is a prohibitively high pressure drop, since 2P Q for laminar flow. The flow ultimately turns
turbulent at flow rates beyond 14 mL/min, leading to an even larger pressure drop. For example,
at a flow rate of 50 mL/min, 30 m of Teflon AF tubing is needed. At the same time, the pressure
drop reaches 130 bar, which is beyond the pressure rating for the tube and impractical for most
applications.35
13
Figure 4. Comparison of different approaches to scaling up (Green triangle: increasing tube
length while parallelization; Blue diamond: increasing tube inner diameter; Red square:
decreasing residence time.)
One way to the reduce pressure drops is to parallelize the unit with multiple shorter strands
instead of using one long tube. With the goal of controlling pressure drops below 1 bar, at 50
mL/min, 6 tubes with individual length of 5 m are needed. The Reynolds number is also
effectively decreased after parallelization such that the flow remains laminar for all flow rates
within the range of 50 mL/min. This approach achieves the production goal, but at a cost of
approximately USD$5,000 for Teflon AF-2400 at the current price.36 A compact way of realizing
a parallelized tube-in-tube system would be to seal a bundle of Teflon AF-2400 tubes in a wider
impermeable tube, similar to constructions used for bundled heat exchangers and hollow fiber
dialyzers.33 Securely sealing a bundle of thin Teflon tubes within an outer shell and evenly
distributing the flow across multiple tubes could be potential engineering challenges.
On the other hand, multiphase gas/liquid flow reactors (segmented flow reactors on the ~ 0.1
mL/min scale and Corning Advanced Flow Reactors on the ~ 10 mL/min scale) have equally fast
mass transfer rates as the micro-scale tube-in-tube reactor, and they cover a wide range of scales. 26 Within the full range of flow rates considered here (0 – 50 mL/min), multiphase gas/liquid
flow can always achieve full saturation of gas in liquid at the outlet. The mass transfer coefficient
14
kLa for both the segmented flow reactors and the Corning Advanced Flow Reactors ranges
between 0.1 – 1 s-1 depending on operation conditions, meaning that their mass transfer time
scale is 1 – 10 s.14,37-39 The excess gas at the outlet can be easily removed by using a settling
tank or by passing pressurized gas/liquid flow through an inline degasser made from Teflon AF-
2400.29 The resemblance between segmented flow microreactors and the Corning Advanced
Flow Reactor in terms of mass transfer facilitates the scale up process.
Reverse Tube-in-Tube Configuration:
A reverse tube-in-tube configuration has also been introduced in which the liquid stream is on
the shell side and pressurized gas is in the inner tube.31,32 We performed numerical simulation of
the mass transfer performance of the reverse tube-in-tube configuration to compare with that of
the original configuration. To facilitate the comparison, the same geometrical dimensions were
chosen for the reverse tube-in-tube reactor as the original tube-in-tube (The inner radius of the
Teflon AF tubing is 3 mm, the outer radius is 4 mm). The width of the shell side is 3 mm, which
is the same as the Teflon AF tubing inner radius.
The simulation results are summarized in Figure 5. When operated at the same flow rate, the
reverse configuration provides a higher gas concentration at the outlet. This is because that
flowing the solvent on the shell side provides a larger cross-sectional area than on the tube side,
which, at an equal volumetric flow rate, provides a longer gas/liquid contact time that enhances
mass transfer. The major advantage of this configuration is its ability to directly heat or cool
reactive liquid via a stainless steel outer shell. However, Teflon AF-2400 itself is not designed
for prolonged or aggressive heating. A PTFE-type fluoropolymer has been reported for higher
temperature applications,32 but the gas permeability performance of this new polymer has not
been documented. Another recent investigation attempted to use the Teflon AF-2400 membrane
in the reverse configuration at temperatures up to 80 ̊C, but the temperature effect on membrane
permeability and durability remains to be studied.25
Reactions in the tube-in-tube
The above mass transfer discussion forms the foundation for studying gas/liquid reactions in
the tube-in-tube reactor coupled with mass transfer. The previously published literature on the
tube-in-tube reactor has focused on screening gas/liquid reaction conditions to maximize yield
15
and/or selectivity.19-27,30 Industrially relevant gas/liquid reactions are typically slow, which would
require heating or catalysts to be accelerated to a reasonable rate for flow applications. The tube-
in-tube section affords only several seconds of residence time, and Teflon AF is not designed for
heating or heterogeneous catalyst loading. Thus, in almost all cases, the tube-in-tube section is
used to saturate the liquid stream with dissolved gas before reaction, and the actual reaction takes
place downstream in a heating coil or packed bed. 22, 23
Figure 5. Comparing mass transfer performance of conventional tube-in-tube and the reverse
configuration. (a) Concentration profile of gas in tube-in-tube reactor; (b) 3-D visualization of
tube-in-tube reactor with gas concentration; (c) Concentration profile of gas in reverse tube-in-
tube reactor; (d) 3-D visualization of reverse tube-in-tube reactor with gas concentration; (e)
16
Comparison of saturation fraction of the two configurations. (Blue line: liquid-inside
conventional configuration; Red line: gas-inside reverse configuration.)
One problem in this strategy is the amount of gas dissolved into liquid. Although gas saturation
is fast (~ 10 s), at full saturation, the concentration of dissolved gas is only ~ 24 10 mol/L (at a
gas pressure of 10 bar), making it very likely that the dissolved gas is insufficient unless the
substrate is highly diluted. Thus, it would be useful to predict the extent of gas deficiency at a
given set of conditions, in order to guide experimental design. To this end, we perform numerical
simulations with a newly-added reaction term, and an initial concentration (C0 = 0.5 mol/L) of
substrate is introduced at the inlet instead of pure solvent. It should be noted that more complex
scenarios such as mixed gas, multiple substrates, multiple reactions and complex kinetics can
also be modeled (e.g., hydroformylation reaction, in which a mixture of H2 and CO is employed).
For clarity of demonstration, a single second-order reaction between pure gas and one substrate
is considered here.
For typical reactions (Figure 6a and 6b), the reaction rate is much slower than the gas
permeation rate, as is the case with most gas/liquid reactions. As a consequence, reactions occur
only outside of the tube-in-tube region in downstream tube sections, and the tube-in-tube region
serves only to saturate the liquid stream with gas. Calculating the outlet concentration for Figure
6a and 6b reveals that the gas to substrate molar ratio at the outlet is merely 1:4, indicating that
the downstream reactions will be severely gas-limited, i.e. substrate can achieve only 25% of full
conversion at best (assuming 1:1 substrate to gas stoichiometry). The separation of mass transfer
and reaction in a typical case means that once the flow exits the tube-in-tube region and starts to
undergo reactions, gas can no longer be supplied. The low loading of gas severely limits the
throughput and productivity of the tube-in-tube reactor. In a typical scenario, gas concentration
in liquid can only be as high as ~ 24 10 mol/L (at full gas saturation and a gas pressure of 10
bar), meaning that the inlet substrate concentration cannot exceed 24 10 mol/L if full
conversion were to be anticipated. Assuming that the flow rate can be scaled up to 50 mL/min
(despite the engineering challenges discussed previously), the throughput is only 2 mmol/min.
For very fast reactions with low gas pressure (Figure 6c and 6d), the reaction is also gas-
deficient due to mass transfer limitations. The gas saturation concentration is proportional to gas
pressure according to Henry’s law, and the maximum pressure is limited by the mechanical
17
tolerance of the Teflon AF membrane. The highest gas pressure applied to the tube-in-tube
reactor so far is 30 bar,23 and another source recommends that the maximum gas pressure in the
tube-in-tube reactor not exceed 27 bar.35 Therefore, the dissolved gas concentration in flow is
generally very low.
Both of the above scenarios (which represent almost all experimental conditions) result in gas
deficiency, meaning that the supply of gas is the limiting factor in determining throughput. One
solution for gas deficiency is to perform a total recycle (i.e., connecting the outlet with the inlet)
over a long period of time, as proposed in previous works.23,30 Although more gas can be
introduced into the reactive system after multiple passes through the tube-in-tube section, this
approach essentially turns the system into batch mode rather than continuous mode.
18
Figure 6. Simulated reaction in tube-in-tube reactor: (a) and (b) show gas deficiency under