Organic Process Research & Development
Article
2 for nitration is relatively higher than that of 3. Furthermore,
isomer 3 was found to have a greater tendency to yield oxidative
products, although the rates of these oxidation reactions were
relatively lower than those of nitration. In order to see the effect
of dispersion and the nature of the flow (with 3 mol of FNA and
1 mol each of water and 2, it becomes a two-phase flow), experi-
ments in which the extent of mixing was varied by changing the
flow rates and using longer residence time tubes were performed.
It was observed that the extent of conversion was independent of
the nature of mixing and that at identical residence times the
conversion was almost identical ( 2.4% variation) (Figure 7B).
This implies that the micromixer followed by a tubular reactor
offers a system without any mass transfer limitations and is
suitable for studying the reaction kinetics.
Upon analysis of the data obtained at different temperatures in
continuous-flow mode at different residence times, it was
observed that the selectivities for different dinitro derivatives
from both 2 and 3 showed different rate behaviors (Figure 8).
In nitration of 2, increasing the temperature results in a reduction
in the selectivity for 5 while the selectivities for 7 and 6 increased.
These observations indicate that while the pre-exponential factor
for the reaction giving 5 is higher than those for the other two
isomers, the activation energies for the latter two are lower, which
allows their selectivities to increase with increasing temperature.
Similar trends were observed for the case of nitration of 3. These
observations are important if one wants to exclusively synthesize
any of these dinitro derivatives of 1. On the other hand, they
also indicate that at a given reaction temperature for the
mononitration of o-xylene, the residence time should be
restricted to ensure that a complex mixture of dinitro derivatives
is not formed. The literature shows that the melting points of
the dinitro derivatives 7 (89 °C), 6 (82 °C), 5 (75 °C), and 4
(115 °C) vary closely and also over a wide range. This will require
specific and controlled crystallization protocols to recover the
pure mononitro derivatives, which will add more equipment as
well as higher operational costs.
adequate heat transfer coefficient is achieved by providing the
necessary heat transfer area as well as the Nusselt number on the
tube as well as the shell side. Thus, it is always possible to design a
flow reactor that can take care of location-specific heat duties as
the reaction proceeds along the length of reactor. In the present
exercise, it was assumed that for a fixed capacity and for a given
set of conditions the selectivity is independent of the reactor
design, and hence, the downstream processing costs will remain
the same. Similarly, for a given throughput the pumps have a wide
pressure range, and hence, the same pumps are used independent
of the reactor design.
Considering the above criteria, here we give a quick approach
that allows one to evaluate the possible design of a flow reactor.
On the basis of the volume fraction of the reactor that yields a
specific heat duty (or heat generation rate), its length can be
estimated from its diameter, which eventually yields the cost of
the specific section of the reactor. Thus, depending upon the
fractions of the reactor used for different heat loads, the total
capital cost of a reactor (including the cost of a jacket depending
upon the geometrical configuration of reactor, viz., coil, double
coil, triple coil, etc.) can be obtained. The cost of the fittings and
connectors can be added to the capital cost (tubular reactor +
jacket). The operating costs are primarily the cost of pumping
due to the pressure drops at a given flow rate in different sections
of the reactor and the cost of utility and its pumping through the
jacket. Usually, single long tubular reactors are uneconomical
because of the very high pressure drop, and hence, the
numbering up approach is used, which reduces the flow rate
through each reactor and therefore also the resulting pressure
drop. Thus, while the capital cost of the reactor can be retained
more or less the same (distributor cost gets added depending
upon the numbering-up strategy), the operating costs can be
reduced. Also, the capital costs and peripherals, including the
control system, are a one-time investment, which usually
depreciates and needs maintenance, adding to the recurring
costs of a plant.
In view of the above observations on nitration of 1 and
independent nitrations of 2 and 3, it was thought desirable to use
the optimal conditions to check the economic feasibility of the
flow reactor.
We have followed the approach discussed in detail by Joshi and
Doraiswamy22 for the design of a flow reactor for production of 3
at 100 and 500 kg/h. Back of the envelope calculations showed
that a tubular reactor when split into N parallel units (i.e.,
numbering-up) would yield a much lower pressure drop because
of the lower velocities in individual tubes. The details of the
reactor design procedure, optimization approach, integration of
mixing, heat transfer, and reaction will be reported separately. In
Figure 9, we have shown the observations from the calculations
for reactors having 10, 20, and 50 units running in parallel to
achieve the desired production rates at varying volume fractions
of 3.175 mm diameter tube (the rest being the 6.3 mm diameter
tube). It can be seen that the capital cost (CAPEX) of the reactor
continued to increase significantly as the volume fraction of the
reaction mixture through the smaller-diameter tube increased
(Figure 9A,B). With 50 parallel reactors, this increase was not
significant. A plot of the total CAPEX of the reactor (including
reactor cost, peripherals, control system, etc.) versus the total
operating cost (OPEX) showed a positive correlation, with the
steepness increasing with the number of parallel units used to
fabricate the reactor (Figure 9C,D). Also, the steepness in the
trend decreases at higher OPEX, which indicates that the OPEX
plays a significant part in the overall cost right from the first year
of reactor operation. Thus, having more continuous reactor
units running in parallel is always economical as it decreases the
OPEX compared with having a single flow reactor. A few such
assemblies of tubular reactors can be seen in the literature.23−27
3.3. Economics of a Continuous-Flow Reactor for
Nitration of o-Xylene. Considering that the continuous-flow
nitration of 1 is to be carried out for the production of 3 at 100
and 500 kg/h (i.e., about 72 TPM and 360 TPM, respectively),
here we present a simple analysis of the economic feasibility
of this nitration reaction in a jacketed tubular reactor. The
conditions (residence time, temperature, and mole ratio of 1 to
FNA) that yield the maximum selectivity for 3 were used as basis
for the reactor design. As a simple system, it was considered that
the reactor is to be fabricated using a combination of different
sections made from SS316 tubes of 2.5 mm i.d. (i.e., 3.175 mm or
1/8 in. o.d.) and 4 mm i.d. (i.e., 6.3 mm or 1/4 in. o.d.). In order
to produce a fixed quantity of 3 in a given time, knowledge of
the specific residence time fixes the total reactor volume.
The smaller-diameter tube (3.175 mm) is usually cheaper than
the larger-diameter (6.3 mm) tube, and it provides 1.89 times
higher heat transfer area. However it needs a 2.54 times longer
tube length to occupy the same volume and hence would offer a
higher pressure drop at identical flow rates to produce a fixed
quantity of 3. The tube with smaller diameter also yields
relatively better mixing and higher mass transfer rates. Thus,
depending upon the heat generation rate in a specific region
along the length of the reactor, it is necessary to ensure that an
H
Org. Process Res. Dev. XXXX, XXX, XXX−XXX