Chemical Engineering Science. Vol. 47, No. 9-1 I, pp. 2511--2516. 1992. ooo9-2509/92 55.00+0.

00
Printed in Grear Britain. Pcrgamon Press Ltd

PHENOL-ACETONEPROCESS:CUMENE OXIDATIONKINETICS
AND INDUSTRIAL PLANT SIMULATION
P. Andrigol, A. Cairn?, P Cavalieri d’Orol, A. Fait’, L. Roberti2, M. Tampier?,V. Tartari

1) EniChem ANIC s.r.f., Research Centre, Via S. Pietro 50 - 20021 Bollate MI Italy
2)Ausimont s.p.a., Res. 81 Dew Centre, Via S. Pietro 50 - 20021 Bollate Ml Italy
3) EniChem Polimeri s.r.1..Research Centre, Via della Chimica 5 - 30175 P. Marghera VE Italy

ABSTRACT leaving the reactor was collected and analyzed to quantify the
by-products CHsOOH and HCOOH; their amounts were referred
In order to optimize the performance of the industrial phenol
to the or anic flow to obtain their production rates. Concerning
plant, cumene oxidation and gas liquid loop reactors have been the HCO 8 H determination, it must be pointed out that its measu-
studred at laboratory scale. Both the kinetic and the fluid dyna-
rement is deficient because of its partial decomposition to COs.
mics models were developed and combined to give the indu- By the other hand a suitable correction is not possible owing to
strial reactor model.
the presence of NaHCOs.
An industrial line of EniChem ANIC production was simulated.
Reaction mechanism
INTRODUCTIDN The experimental data were interpreted according to the reac-
tion scheme shown in Fig. 1. The first 8 steps are enough to
The most widely used route for phenol production proceeds
describe the main reaction (Emanuel et al., 1967: Hattori et al.,
through the following steps:
1970) the remaining steps, already described by several authors
a) alkylation of benzene with propylene to cumene (RH);
(Hendry, 1967; Traylor & Russel, 1965), are useful to justify the
b) liquid phase oxidation of cumene to cumene hydroperoxide
by-products formation. As far as we know, it is the first time that
(ROOH); all these steps are considered together in a complete kinetic
c) cumene hydroperoxide acid cleavage to phenol and acetone. model.
EniChem Anic, a world scale phenol producer, studied extensi-
vely all these steps in order to increase the overall process
yield. Classical l-8 steps
In order to reduce the parameters number, the following
In the oxidation of cumene the main by-products (Fig. 1) are:
hypothesis have been made:
phenyl-di-methyl carbinol (ROH), acetophenone (R,COCHs) and
dicumyl peroxide (ROOR).
In the acid cleavage of ROOH, the by-products ROH and ROOR I) k,=2-G
generate secondary reactions with lead to phenol consumption 2) quasi steady-state for radical concentration
and to purification problems.
The kinetic study of the ROOH acid cleavage reaction have been RR,,- = RHO- = RR. = RRoi = 0
already published (Cavalieri d’0ro et al.,).
The present work describes the oxidation step and illustrates by which:
the kinetic study and the fluid dynamics characterization of the
industrial reactors carried out to obtain a mathematical model
able to describe the industrial plant.
rz = rl I r3 = rl
4&4&G ; r4=r5+ 2.G
KlNEllC MODEL
the 4 unknown variables are:
Experimental
The kinetic study was carried out in a micropilot apparatus pro-
vided with a 2 liter CSTR reactor and an on-line oxygen analyzer k2 IRO’I, k3 [HOI, I% [RI, kg [ROil
to obtain a given oxygen partial pressure (Po2) atthe reactor top.
The 3 independent variables explorated were: the resulting equations system is linear and its solution can be
temperature 105,115,125”c; easily found. The expressions obtained for the 4 variables can
oxygen content 1,2,3.6% vol in the off-gases; be substituted in each rate step to give:
space time 7000-28000 s.
For each of the temperatures investigated the experimental r, = kt - [ROOHI; r2 = rl; r3 = rl;
tests were carried out at constanttotal pressure but at different
oxygen content in the off-gases and varying the space time. r4=f.((kSl~).[RHl+2.~.~
Preliminarytests demonstrated that the reaction was kinetical- r5= f. (k&&I. IRHI - $;
ly controlled. The oxygen concentration was estimated on the
basis of the Henry’s coefficient for 0s in cumene. T8=(l-f)2-r,; r7=2.f-(l-f)-rl; r,=P +rl
An acqueous NaHCOs solution was fed continuously in the
reactor to neutralize the acid by-products (essentially HCOOH,
see later on) which can catalyzethe ROOH acid cleavage. In this
way the formation of traces of phenol which can inhibite the where f = (k&&I. [021+ (k&$ [RH] + 2.6
radicalic oxidation reaction can be avoided.
The experimental variables detected were: the reaction volume, The production and consumption rates can be obtained as usual
the inlet and outlet organic liquid flow, the liquid flow of cumene by linear combination of the step rates.
evaporated as well the off-gases flow and Its 0 and CH, con- It is remarkable thatthanks to the assumptions made above the
tent. The effluent organic hase was analyze d to determine mode) parameters are only 3:
ROOH, ROH. R,COCHs and & OOR concentration. The cumene
contentwas calculated by difference. The aqueous liquid phase

css l119111-I 2511

2512 P. ANDRIGO ef al. D12

As it is well known from literature data (Hattori et al.,, 1970). The methyl radical can then abstract an hydrogen atom (step 10)
rising the oxygen concentration or PO2the rate of the mam reac- to give CH& that is also an experimentally detected by-product.
tion approaches a plateau value, This trend can be deduced by In orderto justify R,COCHB production, the introduction of a new
observing the expression of the ratio f. In order to verify this parameter is needed:
trend, the main reaction rate must be written by assuming some
approximations. In the hypothesis of long-chain:
435= kg/ k2

rinitiation << rpropagation i.e. in this case: rl << r,, r5 It’s not necessary to introduce a further parameter to justify
methane production because in the steady-state hypothesis all
RRooH= r5-rl G r5 = f . e3 - [RH] . de, . [ROOH] = the ‘CH3 generated is completely converted to CH4 (i.e. rs = r,,-,).

Hydmgon abstraction from RH and ROOH

In the classical mechanism the radicals RO’and HO’ abstract an
hydrogen atom anlyfrom RH, and thus:
after linearization:
RRIJH = k2 [RN [RO’I RR,COCH~ = kg [RO’I
[RHIW 1 1 .- [RI-II consequentely:
RROOH = I/G+,+ ~.+H PO2
RROH k2
This linear relation has been experimentally verified as shown in
=-_= constant
Fig. 2 in which all the experimental data obtained at the 3 tem- [RN. RR,COCH, ks
peratures investigated have been plotted. In the same figure the
trends obtained in Hattori et al., 1970 are also reported. In order to verify this equation the first term was plotted versus
The parameters so obtained can be used to estimate thef coef- [ROOHMRHI. As shown in Fig. 3 one can easily observe that it is
ficient at different experimental operation conditions. not verified. If we consider the hydrogen abstraction also from
ROOH, as reponed in literature, thus:

92. LO21 1
f= RROH= k2 [RH] [RO’] + k2b[ROOH] [RO’]
e,[O,l + @, 3[RR1 = 1 + (@/I@, . HII. [RHlPo2
and

T “C 105 115 125

RROH k2 [ROOH]
=_+k2b .-
[RHYPo2 mol/m3/atm lo” 115-104 104 115.104 l@ 115.104
f 96 1 .64 .94 1 .52 .92 I 43 [RW - RR,COCH~ kg kg [RN
This equation agrees very well with the experimental data plot-
The f value can be used to calculate some of the by-products ted in Fig. 3 and shows that kdkzb s 1 at 105°C. By analogy the
distribution; for instance for f = 0.8: remaining steps were added to the kinetic model. The new para-
meters introduced are:

RR~ r6
= = (1 - f)2 = 0.04
b2 + RROOR r6 + r7 + r8

Actually Rz was experimentally observed only in traces. Howe- ‘CHs oxidation and corresponding terminations
ver its production rate (step 6)was considered to make possible If the methyl radical could produce exclusively methane then
the analytic solution of the equations system. CH, and RlCOCH3 should be present in the same amount or very
close together (see above steps 9 and 10). Actually CH4 is less
step9a than RICOCHI. A further hypothesis was made about the ‘CHs
According to the classical mechanism, above described, the oxidation to CH,OOH, that also has been experimentally obser-
ratio between the production rates of ROH and ROOR is given ved and quantified.
by: At last all the possible CH 02’ terminations have been conside-
red (steps 12,13 and 14). fhe termination product HCOOH was
1 also experimentally determined, CH OOCH3 and ROOCH were
RROH
-=-=p rl on the contrary never observed butt a ey were introduce d.in the
>l
f .(2-f) reaction scheme to make possible the analytical solution (as
RROOR r7 + r8
described for step 7). In order to consider these new steps
another 2 parameters are needed:
For f values close to 1 (hi h P ) this ratio approaches to 1.
However experimentally BROHyis 20-30 times higher than
[ROORI (see Fig. 5). Therefore the step 8a was introduced to @lo = ktoc 1 k,o ; ~11=W~
reduce the amount of RO; and to rise the RO’ radical. The kinetic
model has now a further parameter: as well another 2 assumptions:

04 = k6r~/ k6 k,2=2di& ; k,,,=2=

The new velocity equations can be easily derived. They are not Kinetic eramete= estimation
reported because their expressions are fairly complex. The mo s al contains 22 parameters; among them 11 are apparent
kinetic constants (9, + 8,r) and 11 are the corresponding acti-
Step 9 and 10 vation energies.
Acetophenone is the second most abundant by-product after The parameters estimation proceeds through the following
ROH. Its generation is due to the p-cleavage: steps:
1) Initial guess evaluation. Graphical methods have been used
starting from approximate equations. Consequentely only an
ke approximate parameters estimation was produced. A suitable
CBH5- C(CH,)2 0’ - C6H5- CO - CH3 + ‘CH3 order of magnitude of the parameters was however obtained, a
(RO’I good start for the following steps.
D12 Phenol-acetone process 2513

2) At this point a non-linear weighted least squares method Industrial mactor

[Buzzi Ferraris, 19681 was used, implemented on a mainframe
(UNISYS 2200/400). The independentvariables were temperatu-
re, IRH), [ROOH) and fOz1.
The dependent variables were the reaction rates. The parame-
ters estimation was good but the method is statistically not cor-
rect because [RHI and [ROOH) should be considered as depen-
dentvariable.
3) The model for the CSTR micropilot reactor was built up in
orderto allow the correct attribution of the independent and
dependent variables, as indicated in the experimental part. The
results of the estimation are good as shown in Fig. 4,5,6.
FLUID DYNAMIC MODEL
In the process described in the present work, cumene oxidation
is industrially performed in bubble-column reactors with an
inner draught-tube. The air is fed at the bottom of the external
annular zone and drives a liquid motion rising up in the external
zone and falling down in the draught-tube. The liquid flow gene-
rated provides a good mixing and an efficient heat transfer to
the cooling system inside the reactor.
Experimental
In order to study the principles governing the fluid dynamics
behaviour of these reactors, a lab-scare 1 m heigh cold model,
made of plexiglas, and operating with the air-water system, was
built scaling down the industrial reactor (Fig. 7). Gas hold-up
both in th external (E .) and internal (E i) zone and superficial
liquid velocity in the’internal zone (I.),$ were experimentally
detected at different superficial gas velocities (U e) in the range
of practical industrial interest (O-O.08 m/set). AIrthe measure-
ments were carried out without liquid feed. Gas hold-up was
obtained by AP measurement with a manometric technique
using an immiscible fluid (CH,t&). The superficial liquid velocity
inside the draught-tube (&) was determined by injecting a tra-
cer at the top of the draugth-tube and detecting the time of pas-
sage between two points at a given distance. The bubble rising
velocitY(U,) was evaluated forthe air-water system by measu-
ring gas hold-up at different U, in a .08 m diameter glass tube.
The value obtained (.30 mlsec) IS m good agreementwith litera-
ture data for bubbles between 1.5 and 5 mm [Motarjemi and
Jameson, 19781.
In order to calculate the industrial reactor performance we
need to know the variation of U, for the air-reaction medium
sYstem and off, atthe industrial dimension,with f&equal tothe
feed flow.
In order to have a point of reference, the value of U, for the air-
cumene system was determined by means of the same experi-
mental technique above described for the air-water system;
value in the range .l-.12 m/s was found.
Measurements of gas hold-up, both in the internal and external
columns were carried out in industrial reactors of different size
using the same experimental technique with a AP cell instead of
the U tube manometer.
In the absence of velocity measurements the data were regres-
sed setting f,, equal to the maximum value, and therefore:
U,, . Ai = max (6)
and evaluating the U, parameterforthe industrial system byfit-
ting experimental and calculated ege and &ci; we found a U,
value near the cumene-air measured value, as expected.
In this way the model should be able to calculate gas hold-up
and hence liquid hold-up values with enough precision. We may
lack some precision in the liquid velocity estimate, but it is more
important to have a better estimate of the liquid volume inside
the reactor from the gas hold-up calculation.
INDURTRIAL REACTORMATtlEMATlCAL MODEL
Both the kinetic and fluid dynamics models were combined in
the reactor model. At the industrial operation conditions the
internal recirculation liquid velocity is very high, so that the
reactor can be described as a perfect mixed one relatively to
the liquid phase. Concerning the gas phase, there are gradients
in Po2, hence in [Oz).
Fig. 9 shows schematicallythe principles of the reactor descrip-
tion. The temperature of the reactor is uniform because of the
intense internal mixing and is controlled acting on the flowrata
of the heat exchanging fluid running in the coils inside the reac-
tor; a part of the heat produced by the reaction is with drawn
also from the water and the organic products stripped by the off
gases. We need therefore to solve only mass balance equations,
with Ri ca tculated at the reactor temperature:

Cold model condensable organic products:

A remarkable characteristic of the system under examination is -
the presence of a gas hold-up inside the draught-tube. Along F~~‘x~~=F~.x~+F,.~~-V.R~ (7)
the whole draught-tube air is withdrawn by the liquid flow that is
moving downwards at a velocity higher than the bubble rising
velocity (U, > U,). The following equations form the approxima-
yj=XjaPp/Pt (8)
te fluid dynamics model:
incondensable products:
gas-liquid slipping for the internal and external zone
F,, . yi, = F, . yi - V . q (9)
U/i
---_ “gi _ u, (1) where:
1-&gi Egi

U
ge be
_---=lJu, (21 + (1 - Egi) . IviRj he P& dV1 wu
he 1-E,e
The presence of an aqueous solution of NaHCOs, immiscible
energy balance with the organic phase and characterized by a hold-up E, with
respect to the liquid volume, lowers the partial pressure of con-
densable and incondensable products:
(3)
Pt = 2ajXj Pp+ Pi& - XH20,C + PO2+ PN2 + PC”4 (111
continuity equations where the organic phase is considered ideal (~~‘1); the same
assumption is made for water in the aqueous carbonate solu-
U/i Ai t a,= U,, A, (4) tion.
UgrAr+$=Ug,A, (5) Inside the reactor, with reference to points 1, __,4 of Fig. 9:

The fraction of gas flow that can recycle (f, = (Ugi *Ai)/ ) can liquid phase
vary between 0 and a maximum value (f,e+,x) correspon 4 mg to
the situation of equilibrium of the forces mvolved (Eci = eve). On F/t = F/4+ F/, (121 F/I = Fiz (13)
the basis of the above equations and given the parameter U, it Fn=Fn+F, (14) (151
54 = Fn
is possible to calculate f, for every set of experimentally mea-
sured values of U/i, E i and E B. Fig. 8 shows the comparison gas phase
between experimentaland calculated data with f, equal to the
60% of frg,max,and obviously with &= 0. Fgt = Fg4 + F,, (161 Fg2 = Fg3 + F, 117)
 2514 P. ANDIUGO et al. D12

F,, = Fgz- (1 -4. (1 -egJ &,RozdV-keRc,,,dV)(18) Table 1 - Outlet liquid compositionin % w/vu

F93 = Fg4-(1 --EC)* (1 -Egi) (IviRozdV-IviRCH4dV)

(19) ReactorN. 1 2 3 4 5 6 7 8
RH exp 83.7 74.9 72.4 69.3 66.4 63.6 66.4 57.2
The fluid dynamics model (eq. 1, .., 6) is than solved taking into
RH talc 81.7 72.3 69.9 67.5 65.2 62.9 86.6 58.3
accountthat:
ROOH exp 9.6 17.9 20.2 23.0 25.6 28.1 31.0 33.8
ROOH talc 11.5 20.1 22.4 24.5 26.6 28.7 30.6 32.6
Fn . MV,
u,j = F/t - MV, u/f?=
ROH exe 1.34 1.88 2.96 2.27 2.51 2.79 3.07 3.36
ROH talc 1.49 2.11 2.28 2.46 2.64 2.64 3.64 3.26
Ai A,
R,COCH,exp 39 46 .43 .47 .51 55 .59 66
and stating that: R,COCH,calc 34 4.S .51 55 .59 .63 .67 .72
ROOR exp .015 944 950 .058 .071 .082 695 .llO
U = Fg3. M’Jg3 + Fg4 . MVg4)
ROOR talc ,023 ,053 961 ,068 ,077 .085 695 .164
91 The complement to 100% is due to inert hydrocarbons.
2. Ai
U _ (Fgl. MVgl + Fgz MVgd
9e - The comparison seems to be fairly good but the calculated con-
2.A. version is higher in the first 6 reactors and lower in the last one.
In order to evaluate the fitting on selectivity, we fixed the con-
Theflowrate atthe top of the external zone (F . MV z) is notfar version (i.e. cumene content) and calculated the products distri-
(510%) from the one at the bottom (F , . MV ,“T:the decrease in bution. Table 2 shows the selectivity calculated without con-
oxygen content is counterbalanced gY the decrease in hydro- straints (talc) and the one calculated by fixing the cumene con-
static pressure. The difference between top and bottom flowra- tent fcalc*); this was achieved varying the reactor volume and
tes inside the drought-tube is about the same value (10-150/o), solving with Newton in order to match the cumene content of
with F,s and Fs4very close together and the hydrostatic pressu- the industrial plant; in this case the agreement is very good.
re playmg the mam role.
The degree of conversion of each species both in the external
and the internal zone is determinated by the mass balance Table 2 - ROOH selectivity in % mol
equations; those for the external zone, written in differential
Reactor N. 1 2 3 4 5 6 7 8
form are:
Set exp 93.9 91.4 90.5 96.5 88.5 87.1 87.6 85.6
Sel. talc 93.0 90.4 90.0 89.4 88.7 87.9 87.1 86.3
3 = A,. (1 --~c). (1 -1~6s). R/j (a, PO*) (20) Set talc” 93.7 91.2 96.7 96.0 89.0 88.2 87.1 85.9
dz

si= A, . (1 - 6,) . (1 - cge) . Rui (x, PO21 CONCLUSIONS

dz
Fo2
PO* = P(z) * -
Zj. Fgj
The integration along the reactor axis L starts from
Z=O Fti = F/j, I Fgj = F,j, I
and ends atz = L, producing the integrals with are used in equa-
tion (10).
An equivalent equations system is written for the internal zone.
The procedure to follow in order to simulate a reactor of known
volume, operated at known values of temperature, feed flowra-
tes and compositions and E, is:
1) from the estimate of integrals J,,RidV, IviRidV and of the hold-
up es,, r+, guessed in the first step or produced in the previous
step, we compute Xi and Yi from (7-10)
2) the molar flowrates F/r. .., F14and F ,, .., F 4 are then computed
from (12-18); this is performed calculating Phe stripping of orga-
nic and aqueous phases on the basis of their partial pressures
(see eq. 11).
3) at this point we solve the fluid dynamics model with Newton
in orderto get a new set Of ssi and &ge.
4) the equations system (20-22) and its equivalent system forthe
internal zone are inte rated in order to get the new values of
j&dV, IvrRfdV; verify iBconvergence is achieved; if not, estima-
te a new set of integrals and hold-ups with Newton (also a sub-
stitution method is suitable1 and go back to 1.
INDUSTRIAL PLANT SIMUIATION
Starting from the single reactor model, the mathematical model
for the industrial plantwith reactors in series was built up. ltwas
used to simulate one of the EniChem Anic industrial production
lines,formed by8 reactors of different size having a total volume
of 400 m3. In order to verify the fit between the model and the
industrial plant, test runs were performed. For one of these,
table 1 shows the comparison between the predicted and expe-
rimental outlet liquid compositions.
(211
The approach of separating the kinetic study from the fluid
dynamics characterization was successfull. Both the chemical
and the physical models, produced by a thoroughly investigation
(22)
at the laboratory scale, were combined in the mathematical
model of the industrial plant. Industrial test runs showedthatthe
model is able to correctly predict selectivity versus conversion.
The productivity prediction is also good but some systematic lit-
tle deviations are still observed. We considerthese results most
satisfactorily with regard to the factthat the kinetic constants
have not be tuned in to fitthe industrial data. Research is in pro-
gress to establish the cause for deviation of the conversion.
ACKNOWLEDGEMENTS
We are grateful to Messrs C. Bertoni, R. Penzo, P. Reggiani and
Spaggiari of EniChem phenol plant for their full cooperation in
organizing and performing experimental measurements and
test runs and for their useful discussion on by-products detec-
tion.
REFERENCES
Buzzi Ferraris, G., 1968, Metodo automatic0 per trovare l’ottimo
di una funzione, Nota I; Ing. Chim. Ital. 4,171-179.
Cavalieri d’Oro, P., Perego, C. and Raimondi, L., 1988, Phenol
from cumene hydroperoxide acid cleava e: reaction kinetics
and industrial plant simulation; III World E ongr. of Chem. Eng.,
Sept. 21-25.1986Tokyo.
Emanuel, N.M., Denison, E.T. and Maizus, Z.K., 1967, Liquid
Phase Oxidation of Hydrocarbons. Plenum Press, New York.
Hattori, K., Tanaka, V., Suzuki, H., Ikawa,T. and Kubota, H., 1970,
Kinetics of liquid phase oxidation of cumene in bubble column;
J. Chem. Eng. Jap. 3 (1). 72-76.
Hendry, D.G., 1967, Rate constants for oxidation of cumene; J.
Am. Chem. Sot. 89,~5438.
Motariemi, M. and Jameson, G.J.. 1978, Mass Transfer from
very small bubbles - The optimum bubble size for aeration;
Chem. Eng. Sci. 33,1415-1423.
Traylor, T.G. and Russel, C.A.. 1965, Mechanisms of autoxida-
tion.Terminating radicals in cumene autoxidation; J. Am. Chem.
Sot. 87‘3698-3706.
D12 Phenol-acetone process 2515

NOTATION V = reactor volume m3

Ai, A, = sectional area m* x = molar fraction in liquid phase -
F = molar flowrate mol s-l y = molar fraction in gas phase
f,, =fraction of recirculating gas - 2 = position on reactor axis m
g = gravity acceleration m s-= er, E. = gas hold-up
h = draughttube height m y = activity -
f-i = Henry coefficient (102] = H Pu2) mol m-3 atm-l Yrj= Stoichiometric coefficient -
kt = reaction rate constant (l”ord ) s-1 T = space time S
of i* reaction (2nd order) mol-l m3 s-l
[I = molar concentration mol m4
L = reactor lenght m
MV = molar volume m3 mol-l Subscripts:
Pr = total pressure (at reactor exit) atm c = carbonate solution
P(z) = pressure at z height atm e = external respectto the draughttube
Pi = partial pressure of jth species atm g=gas
PP = vapourtension of jth species atm i = internal respect to the draught tube
Paz = oxygen partial pressure atm /= liquid
4, Cl,= flow rate m3 s-l 0 = reactor inlet (no notation for reactor outlet)
r, = rate of ith reaction mol m-3 s-l 1, __,4 = points in Fig. 9
Ri = rate of production of j, species = Bi vri. ri _ = vector notation
(of molecules, of radicals) mol m-3 s-l
U, = rising bubble velocity m s-r Superscripts:
-
U/i, tJte, U,,, U,, = superficial velocity m s-l = mean notation

-l

I Fig. 1 - Reaction scheme

classical stess:
initiation
(R- = @ i: ICHJ, ; Al- = @)
side reaction-stew
9a) Rot+ RO,* - .-b RO. + RO. + 0,
I} ROOH w ROW + HO* 9) RO* w R,COCH,+ CHg
21 RO-+RH - R= + ROH 101 CH;+RH D CH, t R*
2bf RO. + ROOH - RO; + ROH lObI CH; + ROOH - CH, + ROT
3) HO*+RH - R* + H,O 10~) CH: + 0, * CH,O;
3bl HO*+ ROOH - RO; + H,O 111 CH,O;+RH - CH,OOH + R-
propagation 11 b) CH,O., + ROOH - CH,OOH + ROt
4) R-+0, w RO; termination
5) ROf+RH * R= + ROOH 12) CH,O;+ ROt - HCHO + ROH + 0,
termination +1/2 0,
6) R*+ R* * R2 I HCOOH
7) R*+RO; ~pb ROOR 131 CH,O;+ CH,O; - CH,OOCH, + 0,
81 ROy+RO; ______) ROOR + 0, 141 CH,O; t R* w ROOCH,

Fig. 2 - Linearization of the ROOH production rate (see text) Fig. 3 - Linearization concerning the step 2b existence
Data of this work: IT= 105°C; AT= 115°C; -T= 125°C
Data from Fig. 10 in Hattori et al., 1970: - - -
(To obtain our units the X axis has been multiplied by
[RHI av= 6510 mol/mJ)
 2516 P. ANDRIGO et al. D12
Fig. 4 - Experimental end calculated ROOH cont. versus space time at Fig. 7 - COLD MODEL
different oxygen partial pressure I P,,= 0.06 atm A P,= 0.12 atm;
l P,= 0.16 atm; * P,= 0.36 atrn T= 115°C water

F=
\
2.5
T= 115OC
n 19602

0
2.0 . 42%02
0 l 3%02
... ... Ai= 28*10-4 m’
.
“E * 6%02 ap A,= 702*10e4 m2
2 1.5 ......
E -calculated
h= 66 m
55
8
E 1.0
CH,CI,

0.5

J(

space time (s) l 10e3

fig. 5- Experimental and calculated ROH, R,COCH and FLOOR Fig. 6 - Hold-up and liquid internal velocity vs gas external velocity
concentrations vs space time at two temperature 7evels in the cold model
. . I : T= 105°C. Pm= 0.20 atm o A o :T= 115°C. Paz= 0.18 atm

/
4- / . Ege x 10’exp
--0. [RDHI / 7_ -EgexlO*calc
AA
---__ [R,COCH,] x 2.5 /
4 Egi x lo2 exp
c [ROORI x 2.5
6 _ -.- Egi x 10’calc
3-
l u,,x 1Oexp
?
z 5- --U,ixlOcalc
‘.
“E 2-
z
E

l-

0 1 I I I I I I I
0 5 IO 15 20 25 0.8 1 1.2 1.4 I .6 1.8 2 2.2 2.
space time (sl l 1O.3
Uge (m/s). 10’

Fig. 6 - Experimental and calculated CH, production rate vs ROOH Fb. 9 - Reactor scheme

=
concentration at different oxygen partial pressure I P,,r= 0.06 atm Ff liquid
outlet
A P,= 0.12 atm; l P,= 0.18 atm; * P,,= 0.38 atm T= 115°C Fg. off-gases
4 4
1.2 L .____________________________
RCH4
t _ I l%OZexp INTERNAL EXTERNAL

Li-
4 2% 02 exp :ILINDRICAl
VOLUME YD?KkR
0.8 - l 3% 02 exp
%. * 6% 02 exp J/i ‘Jgi 2
P
,r - talc data a, x:
5 0.6 - t 5 E ; E t t
1 s s
P .E : g U se U/E
e
0.4 -

0.2 -

PO,
____________________________,
0 I
0 0.5 1 1.5 2 2.5
IROOHl mol/ms* 1O.3