The Oxidative Dehydrogenation of N-Butane in A Differential Side Stream
The Oxidative Dehydrogenation of N-Butane in A Differential Side Stream
The Oxidative Dehydrogenation of N-Butane in A Differential Side Stream
Catalysis Today
journal homepage: www.elsevier.com/locate/cattod
a r t i c l e i n f o a b s t r a c t
Article history: The synthesis of butenes and butadiene from the oxidative dehydrogenation of n-butane is a chemical
Available online 27 April 2010 reaction of economic relevance and the choice of catalyst is of considerable importance. In this simulation
exercise a V/MgO catalyst in a differential side-stream catalytic membrane reactor was studied.
Keywords: The Recursive Convex Control (RCC) algorithm was used to determine the operating parameters
Attainable region required to determine the maximum yields of hydrocarbon products. The algorithm, in addition to select-
Recursive Convex Control Policy Algorithm
ing for the duty a single differential side-stream catalytic membrane reactor (DSR) in preference to a CSTR
Butane
and a PFR, also developed the prole for the optimal addition of oxygen along the length of the reactor.
Oxidative dehydrogenation
Differential side-stream reactor
The maximum yield of butenes, all three isomers, was found to be 0.119 carbon mass fraction.
The maximum yield of butadiene from the ODH of n-butane was found to be 0.799 carbon mass fraction.
The rates of formation of hydrocarbon reactants and products are discussed.
Statistical analyses of the ratios of formation rates of the butene isomers and of carbon dioxide to
carbon monoxide are presented.
The interplay between alternative reaction routes for the formation of butadiene is reviewed as well
as the validity of the kinetic data at low oxygen partial pressures.
2010 Elsevier B.V. All rights reserved.
1. Introduction Recently the Recursive Convex Control (RCC) algorithm has been
developed and used successfully to analyse complex chemical reac-
The attainable region (AR) method uses geometrical principles tions. These reactions, by virtue of many possible products, can be
to determine the multi-dimensional spatial region in which all the considered as possessing many dimensions to which can be added
reactants and products for reaction systems with known kinetics the further dimension of residence time. The RCC algorithm has
can be found. The AR method can also be used to nd the maximum been proved to be capable of solving multi-dimensional problems
yield of product as well as the reactor types and congurations nec- of a complexity previously considered too difcult to handle. Again,
essary to yield any slate of products within this spatial region. It has the characteristics of the RCC and its usage shall not be discussed in
been shown by Feinberg and Hildebrandt [1] that only three basic detail in this paper. Rather the attention of the reader is directed to
reactor types, a continuously stirred tank reactor (CSTR), a plug- the work of Seodigeng [6,7] for an extensive description of its capa-
ow reactor (PFR) and a differential side-stream reactor (DSR), in bilities. Sufce it to say that the use of the RCC algorithm requires
suitable congurations, are sufcient to provide any yield of reac- no prior understanding of AR theory to use it successfully. However,
tion products. In the absence of a complete sufciency condition, a knowledge of AR principles is useful in analysing the outputs from
presumed AR is usually termed a candidate AR (ARC ). the RCC method. Another feature of the RCC algorithm is that no
It is not the intention in this paper to describe AR theory in prior reactor assumptions are necessary as the recommended reac-
detail. Instead, the attention of the reader is directed to the work of tor type(s) and their sequencing are determined by the algorithm.
Glasser et al. [2,3] and Feinberg [4,5] for a comprehensive review N-butane and oxygen are supplied together at the tube inlet of
of Attainable Region theory and practice. a shell and tube reactor at atmospheric pressure and at a temper-
ature of 773K. The reactor tubes are packed with a V/MgO catalyst
containing 24 wt% of V2 O5 . A side-stream of oxygen at a pressure of
Abbreviations: AR, attainable region; ARC , candidate attainable region; CSTR, up to 6 atm inside the shell diffuses through the inert porous inor-
continuously stirred tank reactor; DSR, differential side-stream catalytic membrane ganic membrane tube wall to react with the hydrocarbons inside
reactor; IMR, inert membrane reactor; ODH, oxidative dehydrogenation; PFR, plug the tubes which are packed with the catalyst. Tllez et al. [8,9] used
ow reactor; RCC, Recursive Convex Control algorithm.
Corresponding author at: 16 Rooibos Avenue, Allen Grove, Kempton Park 1619, a SiO2 -modied -Al2 O3 membrane tube wall to distribute oxygen
South Africa. Tel.: +27 11 394 7683; fax: +27 11 717 7557. to the V/MgO catalyst inside the tubes. The pressure inside the tubes
E-mail address: [email protected] (D. Milne). was taken to be 1 atm. The partial pressure of oxygen to the reactor
0920-5861/$ see front matter 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.cattod.2010.03.033
238 D. Milne et al. / Catalysis Today 156 (2010) 237245
also were applicable to our earlier papers [10,11] and the relevance
of the results from the RCC tool to those from the earlier papers
depended upon identical operating parameters. Non-isothermal
heat transfer in the determination of attainable regions has been
investigated in particular by Hildebrandt et al. [12], Nicol et al. [13]
and Godorr et al. [14].
In a recent paper, Milne et al. [15] have described in greater
detail the application of the RCC method to the ODH of n-butane
and 1-butene.
3. Results
was used as the primary control variable. The reaction network for Case 1: ODH of n-butane to form butenes (all three isomers).
the ODH of n-butane [8,9] is shown in Fig. 1. Case 2: ODH of n-butane to form butadiene.
This reaction network clearly shows the complexity of the ODH
process. Omitted from the scheme is the formation of water which The partial pressure of oxygen in the feed was varied between
is present ab initio. Of particular interest are the two reaction routes 85 kPa and a very low value and should a DSR be selected by the RCC
for the oxidation of n-butane to butadiene, one direct and the other algorithm as one of the three possible reactors additional oxygen
through the synthesis of butenes and their subsequent oxidation to would be supplied optimally along the length of the DSR so as to
butadiene. We shall discuss this later in the paper. attain the prole of the ARC . This reactor conguration is styled a
The kinetic data for the system, rates of formation, kinetic equa- critical DSR [1,16].
tions and data for the ODH of n-butane, 1-butene and butadiene are The results of these calculations are values of concentrations in
shown in Appendix A. terms of carbon mass fractions of the boundary values of a con-
vex region in a higher dimensional space. In order to present these
2. Experimental results in a way that is understandable to the reader we will present
graphical results of two-dimensional projections in terms of the
Our previous papers [10,11] showed that the maximum possi- variables of interest.
ble theoretical yields of butenes and butadiene from the ODH of
n-butane could be derived from an inert membrane reactor (IMR) 3.1. Case 1ODH of n-butane to form butenes
where the oxygen partial pressure was maintained at a very low and
constant value along the length of the reactor. In line with Attain- In the ODH of n-butane to butenes, we have nine possible chem-
able Region terminology we shall refer henceforth to this IMR as a ical substances. These include oxygen and water as well as the
DSR. Whereas the ensuing IMR, i.e. DSR, was of an impractical size oxidation both of butane and butenes to butadiene, the latter in
what did emerge from these studies were the theoretical maximum this case being considered as an undesirable by-product. To these
yields of hydrocarbon products possible from the ODH process. This nine substances, a tenth variable, residence time, can be added.
then presented the challenge of determining the necessary process Our earlier paper [11] found that a butenes yield of 0.119 carbon
conditions to achieve these maxima in a smaller reactor. Again, mass fraction was the theoretical maximum yield possible from the
in our earlier papers, we began with considering only a DSR and ODH of n-butane.
neglected the possibility that a PFR (or even a CSTR), in some con- Fig. 2 is the two-dimensional projection of the ARC identied by
guration with a DSR, might prove to be a better option. That the the RCC method for the ODH of n-butane to butenes.
optimal reactor conguration, without having to make any prior Fig. 2 shows the extreme points of the proles in mass fraction
assumptions regarding reactor types, emerges from the analysis of space for the yield of butenes (sum of all three isomers) from the
the RCC results is one of the strengths of this tool. ODH of n-butane as derived from the application of the RCC algo-
With the reaction scheme in Fig. 1 there is a very large increase
in the number of moles as the reaction proceeds. To use the AR
method with linear mixing laws, all hydrocarbon concentrations
are expressed in terms of mass fractions of carbon, the number of
carbon atoms remaining constant from the beginning to the end of
the reaction.
In the ODH of n-butane, the reaction system comprises nine
chemical species including oxygen and water. When residence
time is considered, the ODH of n-butane requires a 10-dimensional
space for a complete description. In our earlier papers [10,11]
the kinetic equations were applied to the nine species within
the 10-dimensions from which two-dimensional projections were
abstracted. Where extensions of these two-dimensional concen-
tration spaces were possible, i.e. through the elimination of any
concave areas, they were done solely within the two-dimensional
spaces and not by intrusion into higher dimensional hyperspaces.
In an attempt to keep the paper as focussed as possible on the
results from the RCC tool, we have assumed an isothermal condition Fig. 2. Projection of the set of extreme points derived from the RCC prole for the
of 773 K and a constant tube pressure of 1 atm. These assumptions ODH of n-butane to butenes (sum of all three isomers) plotted in mass fraction space.
D. Milne et al. / Catalysis Today 156 (2010) 237245 239
rithm. Fig. 2 is a two-dimensional projection from a 10-dimensional and trans-2-butene and are associated with a butane concentration
hypersurface. of 0.623 carbon mass fraction.
In our earlier paper [11], the outer bound of the prole shown The selectivity of butane to butenes at any point of the prole
in Fig. 2 was associated with an extremely low and constant value in Fig. 2 is given by the ratio of its y-coordinate to the distance of
of oxygen partial pressure. A very large and impractical residence its horizontal projection on the x-axis from the feed point. Another
time also was required. The RCC method, as shall be shown later way of saying this is that the selectivity at any point of the pro-
in this paper, attained the prole shown in Fig. 2 by judiciously le is equal to the tangent of the angle between the x-axis and a
controlling the addition of the oxygen in the DSR resulting in a line from the feed point to the relevant point on the prole. It is
signicantly lower residence time. clear from Fig. 2 that the maximum selectivity occurs at the feed
The lower bound of the RCC prole in Fig. 2 corresponds to an point and thereafter diminishes. It retains a positive value until the
initial oxygen partial pressure of 85 kPa and the upper bound to a maximum point of the prole is attained and then becomes nega-
low partial pressure value. Our previous paper [11] showed that the tive. The butane-butenes prole is convex over its entire perimeter
upper bounds of this region for constant oxygen partial pressures and a property of a convex prole is the occurrence of maximum
of 0.25 kPa and an extremely low value were indistinguishable so selectivity at the feed point.
close were they together. The ratio of the yields of the three isomers, relative to trans-
Fig. 2 shows that at a low value of oxygen partial pressure the 2-butene, was found to be constant over the spectrum of oxygen
maximum yield of butenes (the sum of all three isomers, 1-butene, partial pressures. The ratios of formation were found to be 1.9:1.2:1
trans-2-butene and cis-2-butene) from the ODH of n-butane is for 1-butene:cis-2-butene:trans-2-butene and are shown in Fig. 4.
0.119 carbon mass fraction and occurs at a n-butane concentra- This is only to be expected. Referring to Table 2 in Appendix A at
tion of 0.623. The maximum yield from the lower bound is 0.103 the end of this paper, the rate expressions for the oxidation of n-
carbon mass fraction. We conclude that the maximum yields of butane to the three isomers of butene, r1, r2 and r3, are functions of
butenes are relatively insensitive to the partial pressure of oxygen. the partial pressure of n-butane and the selective oxidation of the
Despite the reduction of this partial pressure from 85 kPa to a low catalyst sites, 0 . What differentiates these rate expressions are the
value, the maximum yield of butenes increased by less than 16%. respective rate constants, k1, k2 and k3 (Table 3) and the ratios of
Our earlier paper [11] commented that the selectivity of butane these rate constants for 1-butene, cis-2-butene and trans-2-butene
for the maximum yield of butenes is but slightly inuenced by the are 1.9, 1.2 and 1.0 respectively.
oxygen partial pressure. In Fig. 5 we show the optimal control policy derived from the
Detailed analysis of the results from this RCC application (to be RCC algorithm for the distribution of oxygen along the length of the
shown later in this paper) conrmed that the outermost limit was
commensurate with a DSR to which the supply of oxygen was con-
trolled according to a specic regimen (not necessarily a constant
value of 0.25 kPa.)
Fig. 3 is a breakdown of the upper bound of Fig. 2 to show the
isomeric constituents and their contribution to the outer bounds
values. A characteristic of Fig. 3 is the preferential synthesis of 1-
butene followed by cis-2-butene with trans-2-butene having the
lowest rate of formation. In the absence of kinetic data for the
ODH of the individual isomers to carbon monoxide, carbon dioxide,
water and butadiene, we cannot exclude the possibility that the
reaction rate for the subsequent ODH of trans-2-butene exceeds
that for cis-2-butene and the latter, in turn, is greater than that for
1-butene.
The maximum yields of the individual isomers at the low oxygen
partial pressure mandated by the RCC method are 0.055, 0.035 and
0.029 carbon mass fractions respectively for 1-butene, cis-2-butene
Fig. 3. Breakdown of Fig. 2 to show the individual proles of the three butene Fig. 5. RCC operational oxygen control policy for the maximum yield of butenes
isomers plotted in mass fraction space at a very low value of oxygen partial pressure. from the ODH of n-butane.
240 D. Milne et al. / Catalysis Today 156 (2010) 237245
Fig. 6. RCC oxygen control policy as a function of residence time for the maximum Fig. 7. Set of extreme points derived from the RCC prole for the ODH of n-butane
yield of butenes from the ODH of n-butane. to butadiene in mass fraction space.
Fig. 8. Proles of maximum selectivity of butane to butadiene at oxygen partial Fig. 9. Prole of maximum selectivity of butane to butadiene at oxygen partial
pressures between 0.05 kPa and 85 kPa. pressures from 0.05 kPa to 85 kPa.
The butane:butadiene mass fraction prole is concave from the where Y = maximum selectivity of butane to butadiene and
feed point, A, to a point B between the feed point and the zenith of X = oxygen partial pressure (kPa).
the prole. At high values of oxygen partial pressure, point B lies This equation is valid over the range of oxygen partial pressures
close to point A but with the reduction in oxygen partial pressure, from 0.05 kPa to 85 kPa.
point B moves along the prole in the direction of the maximum. The RCC algorithm specied a total residence time of approx-
It is a property of a concave prole that the point of maximum imately 15 000 s to yield a butadiene concentration of 0.8 carbon
selectivity is coincident with point B. As discussed in Case 1, the mass fraction.
maximum value of the selectivity occurs when the tangent at a In our previous paper [11] we showed that a butadiene yield
point on the prole emanates from the feed point. That this is so of 0.665 carbon mass fraction or 83% of the theoretical maximum
was shown in our earlier paper [17]. could be obtained from a DSR with a constant oxygen partial pres-
Fig. 8 below is a detailed analysis of Fig. 7 and shows the rela- sure of 0.25 kPa and with a residence time of 322 s. At a constant
tionship between the maximum selectivity of butane to butadiene, oxygen partial pressure of 2 kPa, the maximum yield of butadiene
i.e. the tangent point, at several values of oxygen partial pressure. was found to be 0.359 (45% of the theoretical maximum yield) with
In Table 1 we show the residence times and butane concentra- an associated residence time of 58 s.
tions for the values of oxygen partial pressure identied in Fig. 8. It is apparent that unlike Case 1, the synthesis of butenes, the
Also shown in Fig. 8 is the locus of selectivity maxima. Initially, as maximum yield of butadiene as illustrated in Fig. 10 is extremely
the partial pressure of oxygen is reduced, the maximum selectivity sensitive to low oxygen partial pressures.
of butane to butadiene increases rapidly with respect to residence In Fig. 10 we show the control policy for the partial pressure of
time. Below 0.25 kPa, this rate of increase slows and in the limit oxygen as a function of the yield of butadiene. Fig. 10, in effect, says
approaches a value of 0.8 with a corresponding increase in the that the initial partial pressure to the reactor conguration should
residence time necessary to attain this value. be approximately 12 kPa and its partial pressure in the tubes should
Fig. 9 shows the relationship between the maximum selectiv- be controlled according to this prole until the yield of butadiene
ity of n-butane to butadiene and the oxygen partial pressure in a has reached 0.8 carbon mass fraction.
differential side-stream membrane reactor. Fig. 11 shows that after a residence time of approximately
At low values of oxygen partial pressure, the prole assumes an 1000 s, the oxygen partial pressure has been reduced to 0.07 kPa, i.e.
asymptotic characteristic and as the partial pressure tends to a very effectively depleted, with a consequential cessation in the further
low value the maximum selectivity tends to a value of 0.8. That this yield of butadiene.
is so can be seen from a scrutiny of Fig. 7. It is clear from Fig. 11 that after approximately 300 s, the oxy-
A sixth-order polynomial curve (solid line) was found to give gen added is only sufcient to keep its partial pressure in the
the best t to the calculated data (shown as circles). tubes at a very low value and in effect only that which is required
The equation of this prole is to compensate for the oxygen depleted during the oxidation
process.
It is concluded that a maximum butadiene yield of 0.8 carbon
Y = 0.2854 + 0.7503X 1 0.6488X 2 + 0.2786X 3
mass fraction can be obtained from a DSR where the addition of
0.0576X 4 + 0.0052X 5 0.0001X 6 oxygen is rigorously controlled. The total residence time for this
Table 1
Maximum selectivity of n-butane to butadiene, residence time and butane concentration for values of oxygen partial pressure.
Oxygen partial pressure (kPa) Maximum selectivity Residence time (s) Butane concentration, carbon mass fraction
85 0.29 34 0.68
2 0.52 34 0.41
1 0.60 66 0.30
0.5 0.67 130 0.20
0.25 0.71 254 0.12
0.1 0.75 613 0.06
0.05 0.77 1 198 0.03
242 D. Milne et al. / Catalysis Today 156 (2010) 237245
Fig. 10. RCC operational control policy for the maximum yield of butadiene from Fig. 12. Rates of formation of butane, butenes, butadiene at a constant oxygen partial
the ODH of n-butane. pressure of 85 kPa.
Fig. 11. RCC oxygen control policy as a function of residence time for the maximum Fig. 13. Rates of formation of butane, butenes and butadiene at a constant oxygen
yield of butadiene from the ODH of n-butane. partial pressure of 0.25 kPa.
D. Milne et al. / Catalysis Today 156 (2010) 237245 243
The main thrust of this simulation study was to use the Recur-
sive Convex Control (RCC) algorithm to determine the attainable
regions, maximum yields and the optimal reactor conguration for
Fig. 14. Molar ratio of yields of carbon dioxide relative to carbon monoxide as the synthesis of butenes and butadiene from the ODH of n-butane.
function of constant oxygen partial pressure.
Another reason for using the RCC algorithm was to establish the
most favourable reactor operating conditions commensurate with
reasonable residence times. What did emerge from his study was
the utilisation of low oxygen partial pressures to achieve these ends.
Several authors have expressed opinions, either directly or indi-
rectly, on the validity of kinetic data when applied at low partial
pressures of oxygen. Dixon [19] commented that for a reactor, typ-
ically of a shell-and-tube conguration, where a reactant is added
to the stream of reactants and products the apparently favourable
kinetics quoted in the literature might well be unfavourable at the
lower partial pressures of the added reactant that seem necessary
for the maximisation of the desired product.
The experiments conducted by Tllez et al. [8] on a V/MgO cat-
alyst seemingly spanned a range of oxygen partial pressures from
2 to 10 kPa. Later papers by Tllez et al. [9,20] quoted molar ratios
for oxygen to n-butane of 0.56 and furthermore pointed out that
oxygen-rich and oxygen-lean conditions can change the nature
of the catalyst signicantly. Assabumrungrat et al. [18] in their
research paper used an air to n-butane ratio of 8 in the reactant
Fig. 15. Standard deviation of the molar ratio of carbon dioxide to carbon monoxide. feed as the standard condition but their experiments were carried
out over a range of air ratios between 1 and 15. This ratio range cor-
responds to oxygen to n-butane molar ratios of 0.23. These limits,
maximum values of the carbon dioxide ratio at each oxygen partial
in turn, correspond to initial oxygen partial pressures of 1876 kPa.
pressure. The minimum and maxima values are shown as broken
Corts et al. [21] studied the kinetics of a modied V/MgO cat-
lines. The fact that the mean values are almost coincident with the
alyst in a uidised-bed reactor under anaerobic conditions and
maximum values despite signicantly lower minimum values is a
claimed an improvement in the selectivity of butane to butenes and
reection of the consistency of the data. The ratio of formation of
butadiene.
carbon dioxide to carbon monoxide over the spectrum of oxygen
We have found that the yield of butenes seemingly is unaffected
partial pressures is 3.5 with a slight increase from 20 kPa to 3.6 at
by low oxygen partial pressures since yields close to the theoretical
1 kPa and thereafter declining to 3.4 at 0.05 kPa.
maximum are feasible at an oxygen partial pressure of 2 kPa, this
As an adjunct to Fig. 14, Fig. 15 shows the standard deviation
being the lower limit used by Tllez et al. (Fig. 5). This is not the case
of the carbon dioxide ratio around its mean value for each oxygen
in the synthesis of butadiene where the yield is highly sensitive to
partial pressure.
the oxygen partial pressure (Fig. 10).
Figs. 14 and 15 show that over the range of oxygen partial
Whereas we are not concerned unduly with the higher values
pressures from 85 kPa to 20 kPa, the mean value of the ratio of for-
of oxygen partial pressure, the kinetic data used in this study at
mation of carbon dioxide to carbon monoxide is 3.5 with a declining
low oxygen partial pressures require corroboration or, if found
standard deviation from 0.055 to 0.026. Below 20 kPa, the ratio
wanting, replaced with data that, like Caesars wife, are above
increases to 3.6 (standard deviation of 0.06) at 1 kPa and reducing
suspicion.
to 3.4 at 0.05 kPa with a standard deviation of 0.166. The minimum
This simulation study, inter alia, has emphasised the importance
standard deviation of the carbon dioxide ratio occurs at an oxygen
low oxygen partial pressures and, a fortiori, the need for reliable
partial pressure of 16 kPa.
kinetic data under these conditions. Consequently, the validity of
Finally, we wish to comment upon the priority of the alternative
our kinetic data at low oxygen partial pressures must be taken as
paths for the synthesis of butadiene. Referring to Fig. 1, butadiene
being unproven and of qualied applicability.
can be produced either by the direct oxidative dehydrogenation
of n-butane or via the production of butenes and their subsequent
ODH to butadiene. The question is posed as to which of the two 5. Conclusions
reaction paths predominates?
The rate expressions for the ODH of n-butane are shown in The attainable region analysis of the ODH of n-butane and 1-
Table 2. For the production of the butene isomers and butadiene, butene has been undertaken using the Recursive Convex Control
244 D. Milne et al. / Catalysis Today 156 (2010) 237245
faster than the direct synthesis of butadiene by a factor of 4.4. Oxidation of butadiene
The validity of the kinetic data at low values of oxygen partial C4 H6 + 7/2O2 4CO + 3H2 O (10) r10 = k10 PC4 H6 0
pressure remains unproven and requires corroboration. C4 H6 + 11/2O2 4CO2 + 3H2 O (11) r11 = k11 PC4 H6 0
2k13 poxygen
0 =
2k13 poxygen + (9k5 + 3k6 )pbutane + (8k8 + 2k9 )pbutenes + (7k10 + 11k11 )pbutadiene
Table 3
Rate constants and activity coefcients from Tllez et al. [8,9] and Assabumrungrat et al. [18].
Cis-2-butene [3] D. Glasser, D. Hildebrandt, C.M. Crowe, Industrial and Engineering Chemistry
Research 26 (1987) 18031810.
r5 = (k3 pbutane k7 pcis-2-butene )0 (k8 + k9 )pcis-2-butene 0 [4] M. Feinberg, Chemical Engineering Science 55 (2000) 24552479.
[5] M. Feinberg, Chemical Engineering Science 55 (2000) 35533565.
[6] T.G. Seodigeng, PhD Thesis, University of the Witwatersrand, Johannesburg,
Butadiene
South Africa, 2006.
[7] T.G. Seodigeng, B. Hausberger, D. Hildebrandt, D. Glasser, Computers and Chem-
r6 = (k4 pbutane + k7 pbutenes )0 (k10 + k11 )pbutadiene 0 ical Engineering 33 (1) (2009) 309320.
[8] C. Tllez, M. Menndez, J. Santamara, Journal of Catalysis 183 (1999) 210221.
Carbon monoxide [9] C. Tllez, M. Menndez, J. Santamara, Chemical Engineering Science 54 (1999)
29172925.
r7 = 4(k5 pbutane 0 + k8 pbutenes 0 + k10 pbutadiene 0 ) [10] D. Milne, D. Glasser, D. Hildebrandt, B. Hausberger, Industrial and. Engineer-
ing Chemistry Research 43 (2004) 18271831 (with corrections subsequently
published in Industrial and Engineering Chemistry Research 43 (2004), 7208).
Carbon dioxide
[11] D. Milne, D. Glasser, D. Hildebrandt, B. Hausberger, Industrial and Engineering
Chemistry Research 45 (2006) 26612671.
r8 = 4(k6 pbutane 0 + k9 pbutenes 0 + k11 pbutadiene 0 ) [12] D. Hildebrandt, D. Glasser, C. Crowe, Industrial and Engineering Chemistry
Research 29 (49) (1990) 4958.
Water [13] W. Nicol, M. Hernier, D. Hildebrandt, D. Glasser, Chemical Engineering Science
56 (2001) 173191.
r9 = (k1 + k2 + k3 + 2k4 )pbutane 0 + 5(k5 + k6 )pbutane 0 [14] S. Godorr, D. Hildebrandt, D. Glasser, C. McGregor, Industrial and Engineering
Chemistry Research 38 (3) (1999) 639651.
+ (k7 pbutenes 0 + 4(k8 + k9 )pbutenes 0 + 3(k10 + k11 )pbutadiene 0 ) [15] D. Milne, T. Seodigeng, D. Glasser, D. Hildebrandt, B. Hausberger, Industrial and
Engineering Chemistry Research 48 (11) (2009) 52115222.
[16] T.K. Abraham, M. Feinberg, Industrial and Engineering Chemistry Research 43
In these equations, p refers to the partial pressure of the sub- (2004) 449457.
scripted hydrocarbon and the rate constants k1 to k13 are those [17] D. Milne, D. Glasser, D. Hildebrandt, B. Hausberger, Chemical Engineering
Progress 102 (3) (2006) 4651 (with corrections subsequently published in
shown in Table 3. 0 and 0 , the selective and non-selective oxida- Chemical Engineering Progress 102(7) (2006) 6).
tion catalysts sites, are as dened earlier. [18] S. Assabumrungrat, T. Rienchalanusarn, P. Praserthdam, S. Goto, Chemical Engi-
neering Journal 85 (2002) 6979.
[19] A.G. Dixon, Catalysis, vol. 14, The Royal Society of Chemistry, 1999, pp. 40
References 92.
[20] C. Tllez, M. Abon, J.A. Dalmon, C. Mirodatos, J. Santamara, Journal of Catalysis,
[1] M. Feinberg, D. Hildebrandt, Chemical Engineering Science 52 (10) (1997) Chemical Engineering Science 195 (2000) 113124.
16371665. [21] I. Corts, O. Rubio, J. Herguido, M. Menndez, Catalysis Today 91 (2004)
[2] D. Glasser, D. Hildebrandt, C.M. Crowe, American Chemical Society (1987) 281284.
18031810.