Some Safety Aspects on the

Design of Sparger Systems for the

Published online 30 June 2006 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/prs.10143

This article summarizes all important parameters cumene to cumene hydroperoxide for phenol and
for a proper and safe design of the sparger system for acetone production, and the oxidation of p-xylene to
large bubble columns, used for the oxidation of or- terephthalic acid. In ref. [1], an overview of the most
ganic liquids with air or oxygen enriched air. As an important processes and the corresponding process
example, this is done for the oxidation of cumene conditions are given. Typically, air or air/oxygen mix-
(isopropylbenzene), which is worldwide used in phenol tures are used, but even processes with the use of pure
processes. Based on data in the literature and newly oxygen are described, for example, in refs. [2, 3].
measured flammability limits at 100° C and 5 barg, the Hazards that may occur during the liquid phase
potential hazards from uncontrolled gas reactions can oxidation can be divided into hazards from uncon-
be avoided, ultimately avoiding an explosion in the trolled reactions in the liquid phase (oxidation and, in
sparger system. For normal operation, a minimum flow the case of peroxides, thermal decomposition) or from
of gas is necessary to ensure a uniform flow through all an explosion in a coherent gas phase anywhere in the
outlets, which are typically holes with a diameter of reactor system. This article provides some general as-
several millimeters. All holes should be directed towards pects on proper design of the sparger system for the
the vessel bottom to self-drain the sparger. For start-up, cumene oxidation to avoid any hazards from gas ex-
a previous inerting of the sparger with nitrogen is plosion.
preferable. Nevertheless, a certain amount of organic
liquid may still be in the sparger system prior to start
OXIDATION PROCESS FOR CUMENE
up. Therefore, the compressed air/oxygen should have a Phenol and acetone are mostly produced worldwide
temperature below the Oxidation Initiation Tempera- from cumene. Cumene is oxidized to cumene hy-
ture to avoid any initiation of a liquid phase oxidation droperoxide at temperatures in the 80 –120° C range at
within the sparger. In addition, the pressure in the a pressure ranging from atmospheric to 700 kPa [4]. The
sparger system should be high enough to keep the gas oxidation of cumene is a radical chain reaction [1]. The
phase concentration of the organic liquid below the peroxide is subsequently cleaved in a second reaction
Lower Flammability Limit. © 2006 American Institute of unit to yield phenol and acetone. Large bubble col-
Chemical Engineers Process Saf Prog 25: 326 –330, 2006 umns are used for the cumene oxidation, with internal
Keywords: oxidation, cumene, explosion limits, or external cooling devices to remove the heat from
bubble column reaction. The kinetics for the oxidation is described, for
example, in ref. [5]. Figure 1 shows schematically a
INTRODUCTION bubble column for the cumene oxidation.
Processes for the oxidation of hydrocarbons by mo- Either air or oxygen-enriched air may be used. In ref.
lecular oxygen are widely used to obtain important [6], pilot plant experiments with oxygen concentrations
chemical products. Such processes are, for example, up to 27 vol% are described. In a production plant,
the oxidation of cyclohexane to cyclohexanone/cyclo- oxygen may be added to the suction side of the air
hexanol for adipic acid production, the oxidation of compressor or on the pressure side, as shown in Figure
1. The off-gas at the top of the reactor has a low final
© 2006 American Institute of Chemical Engineers oxygen concentration of typically below 5 vol%. The

326 December 2006 Process Safety Progress (Vol.25, No.4)

off-gas is further treated to remove organic compo- Figure 2. Explosive range for cumene in oxygen/
nents. The oxidate, containing cumene hydroperoxide, nitrogen mixtures at 100° C and 5 barg. [Color figure
may be further treated in successive reactors to increase can be viewed in the online issue, which is available
the peroxide concentration. at www.interscience.wiley.com.]
Although cumene hydroperoxide is a relatively sta-
ble organic peroxide, the exothermic decomposition
has to be taken into account during a shutdown when
the apparatus is not mixed by aeration [5]. 100° C, the LOC is 9.8 vol%. Again, this figure was
newly measured at the BAM.
LOWER AND UPPER FLAMMABILITY LIMITS OF CUMENE In Figure 2 the explosive range of cumene in air or
In air the LFL and UFL (Lower and Upper Flamma- in a gas phase with lower oxygen content is shown as
bility Limit) for cumene are 0.8 and 6.0 vol%, respec- the hatched area. It is bounded by the LFL, the UFL, and
tively [7]. Any composition in between is explosive. the LOC. With higher oxygen content, going left from
These figures are only valid for standard conditions the y-axis, the explosive range is widened because of
(293°K ⫽ 20° C, 1 atm). the increasing UFL. This is shown schematically in
With increasing pressure the UFL increases, thus Figure 2.
widening the explosive range, whereas the LFL is The air bubbles coming out of the sparger at the
nearly constant [7]. With increasing temperature the LFL bottom (see Figure 1) are spontaneously saturated with
decreases and the UFL again increases [7]. Thus, the cumene, thus lying in the explosive range between the
explosive range is widened with increasing pressure LFL and UFL, as qualitatively shown for 100° C and 5
and temperature. The LFL and the UFL were measured barg in Figure 2. But this is not hazardous. It is a
at a temperature of 100° C and a pressure of 5 barg at common philosophy that single bubbles may “explode”
the Bundesanstalt für Materialforschung und -prüfung for any reason [3]. The released energy is small and
(BAM). Details about the measurement procedure ac- absorbed in the liquid. On their way from the bottom to
cording to the standard DIN EN 1839 are given in ref. the top, oxygen in the bubbles is consumed. The
[7]. In Figure 2, these two values for LFL (0.7 vol%) and cumene concentration in the bubbles increases slightly
UFL (12.5 vol%) are shown on the y-axis. because of decreasing total pressure. Somewhere in the
middle of the reactor height, the gas composition
LIMITING OXYGEN CONCENTRATION leaves the explosive range. The final oxygen concen-
When the oxygen concentration is reduced (going tration in the bubbles and so in the coherent gas phase
right in Figure 2), the UFL decreases, thus narrowing is lower than 5 vol%, maybe close to 0 vol%, and thus,
the explosive range. Both UFL and LFL finally merge at far away from the explosive range. This is an important
the so-called Limiting Oxygen Concentration (LOC) safety aspect for all oxidation processes with organic
(see Figure 2). In ref. [8] a LOC of around 10 vol% was liquids.
measured at atmospheric pressure at around 50° C. In
new tests at the BAM, the LOC was found to be 10.5 AUTOIGNITION TEMPERATURE
vol%, which fits to the figure given in ref. [8]. The tests The Autoignition Temperature (AIT) of cumene is
were made according to the standard prEN 14756. The 420° C under atmospheric pressure [7]. With increasing
LOC decreases with increasing pressure and tempera- pressure, the AIT decreases. For substances with an AIT
ture [9]. At a pressure of 5 barg and a temperature of above 180° C at atmospheric pressure, it was found that

Process Safety Progress (Vol.25, No.4) Published on behalf of the AIChE DOI 10.1002/prs December 2006 327
 concluded that at temperatures below 60 ° C, which
may be called the Oxidation Initiation Temperature
(OIT), the oxidation rate is negligible even with pure
oxygen. Thus, the compressed air/oxygen mixture
should have a temperature of below 60° C. For the
oxidation of cyclohexane, an OIT of 120° C is reported
[11]. Furthermore, the line outside the reactor should
not be insulated to ensure a quick cool-down of warm
liquid in the case of an unexpected backflow just after
a shutdown.
Figure 3. Sparger system for air or air/oxygen Alexander [11] assumes that a gas phase oxidation
mixture. could be initiated at temperatures significantly lower
than the AIT, if a liquid phase oxidation is running. As
previously shown, the AIT at higher pressure will not
be lower than 180° C. But if a liquid phase oxidation
with increasing pressure the AIT decreases to a mini- takes place, volatile radicals may evaporate into the gas
mum level of 180 to 200° C [10]. Thus, the AIT for phase and initiate a gas phase oxidation at tempera-
cumene is much higher than the typical reaction tem- tures below the AIT, which could eventually lead to an
perature in the cumene oxidation process. explosion. But all this could happen only if: (a) a
significant liquid phase oxidation is running, (b) the
SPARGER SYSTEM DESIGN
heat from the gas phase oxidation is not sufficiently
Figure 3 shows a typical design of a sparger. The
removed, and (c) the gas phase is within the explosive
major safety aspect is to avoid any backflow of liquid
range. As already shown before, the temperature of the
into the sparger line, which could cause uncontrolled
air/oxygen should be below the OIT. The line sizes
reactions. As already proposed by Alexander [11], the
should be designed in a way that with the minimum
air/oxygen outlets should be at the bottom of the
air/oxygen flow, any heat from oxidation will be re-
sparger for “self-draining” to avoid any hold-up of liq-
moved, even in the case of any backflow warm liquid.
uid within the sparger. The outlets may be holes of
The sparger in the reactor should also be designed in a
several millimeters in diameter. A minimum flow
way that any heat from reactions will be removed with
through the holes must be maintained to guarantee a
the colder air/oxygen mixture. Finally, the cumene
uniform flow through all the holes and so prevent any
concentration in the gas phase should be kept below
backflow of oxidate into the sparger system [12]. This is
the LFL in the sparger system outside the reactor. That
described by the F-factor criterion:
is possible with a sufficiently high pressure in the
sparger system. For example, at 60° C, the vapor pres-
F ⫽ v h䡠 冑 ␳ G ⱖ 25 (1) sure of cumene is 40 mbar. Thus, at a pressure of more
than 4.7 barg, the vapor concentration of cumene is
with vh as the gas velocity through the holes and ␳G as below the LFL.
the gas density. The typical gas flow rate at operation In Figure 1 it is shown that either air or oxygen-
conditions should be somewhat higher than the mini- enriched air may be used. Oxygen may be added to the
mum flow. suction side of the air compressor. This seems to be the
During a shutdown, it may happen that some liquid most reliable method because if the compressor fails,
flows back into the sparger line. Even with check no oxygen will be fed to the reactor. If pressurized
valves (non-return valves) or automatic valves close to oxygen is fed to the pressure side of the air compressor,
the reactor wall, this could not be prevented com- a reliable control system must be installed to ensure
pletely, as all valves are subject to random failure. As correct mixing ratios between air and oxygen. An in-
shown in Figure 3, all lines of the sparger system terlock on low airflow should shut down the oxygen
should be horizontal. Vertical lines should have a supply.
downward gas flow. This ensures that during startup all In Figure 1 a bubble column is shown without any
liquid will be drained back into the reactor immediately inserts. It is very well known that sieve trays may be
by the (minimum) gas flow. As already proposed in ref. used to redisperse the rising bubbles and to reduce the
[11], an automatic nitrogen flow should be installed for vertical backmixing of the liquid phase [13]. But it
inerting the lines after shutdown and just before should be taken into account that below a sieve tray, a
startup. continuous gas phase may be formed [14]. In the lower
Before startup, the nitrogen flow may not com- part of the reactor, this gas phase will still have oxygen
pletely drain out all liquid. When the air/oxygen flow is concentrations above the LOC, such that the gas phase
started, some liquid in the line is in contact with the composition is in the explosive range. As in the case of
coherent gas phase. The design shown in Figure 3 will liquid oxidation that takes place at higher tempera-
not lead to uncontrolled liquid oxidation hazards be- tures, the hazard according to what Alexander [11]
cause the liquid will be drained out immediately. If, in postulates could no longer be excluded.
addition, the air/oxygen has a low temperature, the
liquid oxidation rate will be low or even negligible. OXIDATION WITH PURE OXYGEN
Figure 4 shows lab results from cumene oxidation at The use of pure oxygen raises initial concerns. As a
relatively low temperatures with pure oxygen. It can be first estimation, the design principles for the sparger

328 December 2006 Published on behalf of the AIChE DOI 10.1002/prs Process Safety Progress (Vol.25, No.4)
Figure 4. Concentration of cumene hydroperoxide over time during oxidation with pure oxygen, starting with
12% cumene hydroperoxide.

system should be the same as mentioned above. But, as air (see Figure 1), is more common as it is easier to
the reactivity of pure oxygen is higher, great care operate.
should be taken to dry the sparger system free from any
organic liquid just before startup.
CONCLUSIONS
If only pure oxygen is used, the coherent gas phase As shown for the oxidation of cumene, the sparger
at the top of the reactor will be explosive. In ref. [3] a system for air or oxygen-enriched air should be de-
stirred reactor is described, the so-called Liquid Oxida- signed in a way that no organic liquid can flow back
tion Reactor (LOR). The top of the liquid is closed with into the sparger during normal operation. That is en-
a lid to allow only traces of oxygen to penetrate into the sured by operating the sparger with a minimum flow to
gas phase above. The top is inerted with nitrogen. For fulfill the criterion as shown in Eq. 1. All gas outlets
processes with high production capacities, the use of should be directed towards the vessel bottom to ensure
these stirred reactors does not seem to be economical. a self-draining of the sparger. To avoid any hazards
On the other hand, for large bubble columns, it does during startup (assuming that organic liquid is in the
not seem to be possible to separate the gas/liquid sparger line), the air/oxygen supply should have a low
mixture from the coherent gas phase in such a way. temperature. In combination with that, the pressure in
Thus, the use of pure oxygen for the oxidation in large the system should be high enough to keep the vapor
bubble columns seems to be impossible without taking concentration of the organic liquid below the Lower
any additional measures. Flammability Limit. If all these measures are taken, an
In ref. [2] a process is described to oxidize cyclohex- oxygen enrichment up to 27 vol% in the air, as reported
ane with pure oxygen in the presence of a liquid water in ref. [6], will be a safe procedure even for large
phase. The water vapors act as an inert. At a certain bubble columns.
temperature, the pressure in the gas phase is very well
defined for a heterogeneous liquid mixture of cyclo-
hexane and water. Any higher pressure would arise LITERATURE CITED
from unreacted oxygen. Thus, the final oxygen concen- 1. G. Franz and R.A. Sheldon, “Oxidation,” Ullmann’s
tration could, in principle, be controlled by a certain encyclopedia of industrial chemistry, Vol. A18, 5th
gage pressure above the vapor pressure of the two- edition, B. Elvers, St. Hawkins, G. Schulz, (Editors),
phase liquid. But it should be kept in mind that, as well Wiley, Weinheim, Germany, 1991.
as for the oxidation with air, the partial pressure of 2. J.-R. Chen, An inherently safer process of cyclohex-
oxygen in the rising bubbles is reduced, too. That again ane oxidation using pure oxygen, Process Safety
reduces the mass transfer of oxygen into the liquid and Progress 23 (2004), 72– 81.
may diminish the net effect from using expensive pure 3. A.K. Roby and J.P. Kingsley, Oxide safely with pure
oxygen. oxygen, Chemtech February (1996) 39 – 46.
The oxidation of cumene in the presence of water, 4. M. Weber, M. Weber, and M. Kleine-Boymann,
so-called wet oxidation, has already been known for “Phenol,” Ullmann’s encyclopedia of industrial
more than 50 years [4], but typically air or oxygen- chemistry, 7th edition, B. Elvers, St. Hawkins, G.
enriched air was used. As described for cyclohexane Schulz, (Editors), Wiley, Weinheim, Germany, 2004.
[2], pure oxygen could also be used in principle. The 5. M. Weber, Large bubble columns for the oxidation
dry oxidation, that means the oxidation of cumene of cumene in phenol processes, Chem Eng Tech-
without any water phase using air or oxygen-enriched nol 25 (2002), 553–558.

Process Safety Progress (Vol.25, No.4) Published on behalf of the AIChE DOI 10.1002/prs December 2006 329
 6. H.-D. Obermeyer and H.-J. Reinhardt, Linde-Berichte 11. J.M. Alexander, The hazard of gas phase oxidation
aus Technik und Wissenschaft No. 79 (2000), in liquid phase air oxidation processes, IChemE
7. E. Brandes and W. Möller, Sicherheitstechnische Symposium Series, 39 (1975), 157–169.
Kenngrö␤en, Wirtschaftsverlag NW, Bremerhaven, 12. W. Burkel, Auslegung von Gasverteilern und Gas-
2003. gehalt in Blasensäulen, Chem.-Ing.-Techn. 46
8. J.C. Butler and W.P. Webb, Upper explosion limits (1974), 205–206.
of cumene, Industrial and Engineering Chemistry, 2 13. P. Zehner and M. Kraume, “Bubble Columns,”
(1956), 42– 46. Ullmann’s encyclopedia of industrial chemistry,
9. W.O. Möller, M. Molnarne, and R. Sturm, Limiting Vol. B4, 5th edition, B. Elvers, St. Hawkins, G.
oxygen concentration, Report of the Physikalisch Schulz, (Editors), Wiley, Weinheim, Germany,
Technische Bundesanstalt PTB (2003), see www. 1992.
ptb.de. 14. E. Bla␤ and K.-H. Koch, Strömungstechnische Un-
10. E. Brandes and M. Thedens, Kenngrö␤en des Ex- tersuchungen an einem Blasensäulen-Kaskaden-
plosionsschutzes bei nichtatmosphärischen Bedin- reaktor bei Gleichstrom von Gas und Flüssigkeit,
gungen, PTB-Mitteilungen 113, 2 (2003), 115–121. Chem.-Ing-Techn 44 (1972), 913–921.

330 December 2006 Published on behalf of the AIChE DOI 10.1002/prs Process Safety Progress (Vol.25, No.4)