Williams - Oxygen Applications

Oxygen Applications for Organic Liquid Phase Oxidation
Reactors
William R. Williams
Praxair Applications R&D
Tarrytown, NY
1Executive Summary
The following summarizes activity in oxygen applications for organic liquid phase
oxidation (LPO) reactors, and proposes a strategy for future R&D activity.
We propose that we continue activity in three programs: air enrichment, direct injection,
and process re-optimization. The research plan will be conducted in such a way that new and
existing experimental results can be used to evaluate each option using reactor modeling. In this
way, we can take advantage of the technical similarities in each offering to synergistically cover
all applications.
Although the applications may be covered simultaneously with one research program,
unique aspects of the individual chemical processes will require that each chemistry be scouted
separately. Therefore, the chemical processes were prioritized. The prioritization was based on
market analysis, customer benefit, ease of implementation, and technical probability of success.
The chemical processes, in order of priority, are as follows:
1. Purified terephthalic acid (PTA or TPA) and DMT from TPA.
2. Dimethyl terephthalate (DMT) via the Witten Process.
2. (Tie with DMT) Cyclohexane oxidation to cyclohexanol/one (ketone/alcohol oil).
4. Phenol via cumene peroxidation.
5. Propylene oxide co-produced with styrene monomer (PO/SM).
A strategy for implementation is presented which describes an 18-24 month program to
research these processes, determine the best application, and introduce the application to the
marketplace.
Table of Contents
1. Executive Summary.......................................................................................................................
2. Introduction....................................................................................................................................
3. Overview of Applications..............................................................................................................
3.1 Enrichment of air with oxygen (EA).......................................................................................
3.1.1 Plant productivity limited by gas throughput...................................................................
3.1.2 Plant productivity limited by mass transfer in reactor......................................................
3.1.3 Plant productivity limited by reaction kinetics in reactor................................................
3.2 Direct injection of oxygen (DI)...............................................................................................
3.2.1 Plant productivity limited by gas throughput...................................................................
3.2.2 Plant productivity limited by mass transfer in reactor......................................................
3.2.3 Plant productivity limited by reaction kinetics in reactor................................................
3.3 Process re-optimization............................................................................................................
3.3.1 Background.......................................................................................................................
3.3.2 Reducing by-product formation with process re-optimization.........................................
4. Previous and Current Customer Contacts......................................................................................
4.1 Oxo-acids.................................................................................................................................
4.2 TPA..........................................................................................................................................
4.3 DMT (Witten Process).............................................................................................................
4.4 Cyclohexane to K/A Oil..........................................................................................................
4.5 Cumene hydroperoxide for phenol production........................................................................
4.6 Propylene oxide/styrene monomer (PO/SM) co-production...................................................
4.7 Acetic acid...............................................................................................................................
5. Economic Evaluation of LPO Applications..................................................................................
5.1 EA or DI for Capacity Increase...............................................................................................
5.2 Re-optimization for Capacity Increase or By-product reduction............................................
5.3 An example: Enrichment or DI for TPA with re-optimization...............................................
5.3.1 Capacity expansion: Capital improvement versus supplemental oxygen.........................
5.3.2 EA or DI for capacity expansion: Variable cost analysis.................................................
5.3.3 EA or DI for by-product reduction: Variable cost analysis..............................................
5.4 DMT (Witten Process).............................................................................................................
5.5 Caprolactam via cyclohexane oxidation..................................................................................
5.6 Phenol (via cumene hydroperoxide)........................................................................................
5.7 Propylene oxide co-production with Styrene Monomer (PO/SM)..........................................
6. Strategy and Prioritization.............................................................................................................
6.1 Technical Strategy...................................................................................................................
6.2 Market Analysis, Project Evaluation, and Prioritization.........................................................
7. Schedule and Manpower Requirements for Future Work.............................................................
8. Conclusions....................................................................................................................................
Appendices (included only in original).
A. Derivation of benefits of EA for (A-1) mass transfer limited, (A-2) kinetically
limited, and (A-3) DI for kinetically limited systems.
B. Manuscript on liquid phase oxidation kinetics.
C. Project planning tools worksheets and producer overview for major LPO processes.
2
2Introduction
For some time, Praxair has had a wide offering of applications in organic liquid phase
oxidation reactors. Until recently, the focus of R&D was on the Liquid Oxidation Reactor
(LOR), a truly unique Praxair offering which allowed the safe, efficient use of pure oxygen in
LPO1 reactors. Some of the reasons for this strategy were as follows:
· The LOR offered differentiating technology from any offerings of our competitors.
· Because the LOR is the only LPO application that uses pure oxygen, it was assumed
that it would provide the maximum benefit to the customer.
· Also because the LOR uses pure oxygen, it offered PX the largest volume of sales.
The other applications, air enrichment and direct injection, which are both applications to
supplement air with oxygen, were simply offered as commercial applications and little
laboratory or modeling work was conducted on these applications. However, during the course
of LOR studies we have discovered that there are many circumstances where most of the benefit
can be derived by supplementing air with oxygen rather than complete replacement of air with
oxygen. While this reduces the amount of oxygen sales for PX, the lower oxygen sales
requirement reduces the cost to the customer. Thus, the value to the customer is much higher,
and the likelihood of customer implementation of the technology is greater. Further, the
supplemental oxygen technologies of air enrichment and direct oxygen injection involve less
mechanical change to the customer’s process. This also increases the likelihood of customer
adoption.
Below we will summarize each application and the technical work to date. Then we will
review commercial experience with LPO applications. Next we will outline a methodology for
evaluating the economics of the various LPO applications on various chemicals to determine the
best way to use oxygen to enhance processes, and the processes which could most probably
benefit from LPO applications. We will show that our learnings from LPO applications, LOR
experiments and modeling can be applied in unique ways to maximize customer benefit. Our
strategy will be to use this experience and understanding of oxygen in LPO reactors to
differentiate our LPO offerings from our competitors’ technology and more importantly, our
customers’ own in-house technology. Finally, we will outline priorities in chemical processes
and applications and a schedule for implementation of LPO programs.
3Overview of Applications
3.1Enrichment of air with oxygen (EA)
Enrichment of air with oxygen requires the least equipment change to a customer’s
process of any of the LPO applications. It is primarily for this reason that EA has been
commercialized more than the other LPO applications.
In EA, oxygen is injected into the air feed of the LPO reactor (Fig. 1). This provides
several benefits:
1
For purposes of this report, LPO will be used to refer to organic liquid phase oxidation reactors. We will not
consider inorganic oxidations in this report.
3
· More oxygen delivery for same gas volume throughput—This reduces the volumetric
throughput for the air compressor allowing a capacity increase for processes
which are limited by the air compressor capacity. Or if the process gas velocity in
the liquid phase reactor is too high for the reactor design, this also allows a
capacity increase.
· Higher oxygen partial pressure in the gas bubbles—In processes where the reactor
productivity is limited by mass transfer of the oxygen to the liquid phase, or
processes where the kinetics are limited by low equilibrium oxygen concentration,
reactor productivity will be increased.
Vent
Product
Reactor
Air
Oxygen
Figure 1: Schematic of enrichment of air feed with oxygen (EA).
In most of these cases, reactor productivity will be increased by the percentage increase
in oxygen flow to the reactor. Thus the plant can produce more product and profit to this degree.
While EA requires little change, it does require Praxair expertise. We have done EA for
various oxidation reactors including LPO and have experience in this area. The biggest
limitation of EA is materials compatibility. Typically, existing piping and sparger material is
compatible with air, but is not compatible with high oxygen concentrations. PTC has developed
limits on the maximum oxygen concentration acceptable for specific materials. Typically, most
materials in LPO air feed lines and spargers can take a maximum oxygen concentration between
26 and 30% oxygen. Therefore, the amount of additional oxygen that can be introduced to EA is
usually limited to about 25% of the oxygen which is delivered in the air.
4
The benefits of enrichment are a capacity increase. The amount of capacity increase will
depend on the process, and particularly where the bottleneck is. Although each chemistry must
be considered separately, we can group customers needs into three broad categories: Gas
throughput limited, reactor mass transfer limited, and reactor kinetically limited. The analysis of
these scenarios will be treated below, but first we will present some general assumptions used in
this analysis.
LPO reactors typically are run such that the headspace oxygen concentration is fixed.
This is primarily for safety reasons. The oxygen concentration must be kept below the limiting
oxygen value (LOV). The LOV is the maximum concentration of oxygen where no
organic/oxygen ratio will be flammable. Maintaining oxygen below the LOV is generally
preferable, because as organics are treated in the condenser and pressure and temperature are
changed in downstream equipment, the ratio of organic vapor to oxygen will change. Thus, a gas
stream which is initially outside of the flammability limits can move into the flammable region
during subsequent processing. Figure 2 illustrates:
0 100
Operating lines with constant
oxygen/nitrogen ratios as
organic concentration varies
(usually by condensation).
UFL (in O2)
OR
EN
GA
LO
YG
VL
NIC
OX
in e
UFL (in air)

FLAMMABLE
LFL (in air)

LFL (in O2)
Limiting
Oxygen
100 0 Value (LOV)
0 100
INERT
Figure 2: Ternary flammability diagram depicting operating lines as organic composition

varies in downstream reflux columns or condensers.
Practically, this means that in all cases the extra oxygen supplied by EA must be
consumed such that the concentration of oxygen in the bubbles exiting the liquid into the
headspace must be the same as in the air system. With these assumptions the reaction rate in the
reactor in mols per volume per time is given by the following expression:
5
O 2 RV RXR   FOexit
F exit
2 
 FOfeed
2  FO 2  ,or R 
O2
feed
(1)
 O 2V RXR
Notice that the flow of oxygen exiting the reactor is less than the feed flow, but the sign of O2is
also negative so R is positive.
Because the nitrogen is not consumed in the reactor, it is convenient to express oxygen flow as a
function of nitrogen flow which is the same in the exit as the feed by defining the following
ratio:
yO 2 yO 2
  ;and FO 2  FN 2 (2)
y N 2 1  yO 2 
It is also convenient to define the term L as flow per reactor volume = F/V rxr. Thus, assuming
negligible solvent evaporation, equation (1) becomes:
R   Lexit  
O 2  LO 2  /  O 2  L N 2 exit   feed /  O 2
feed
(3)
In evaluating the productivity improvement, we are more concerned with the relative
improvement over the air based process, so in our analysis we will define an enhancement term,
EA, as the ratio of productivity for EA to the productivity for air-based operation. This gives the
following expression:
REA N 2 exit  EA 
LEA
EA=  Air (4)
R Air LN 2 exit  Air 
For convenience, the subscripts “Air” and “EA” imply feed conditions. Notice that we
have assumed that the exit composition is the same regardless of EA or Air operation. This will
be true in all analyses. Also, it is very important to realize that the flow of nitrogen is not
necessarily the same in EA as Air. We shall see that it is often necessary to have less nitrogen
flow in EA so that the reactor can consume enough oxygen to keep the exiting oxygen level at
the design condition. In fact, the analysis of the specific cases below are essentially aimed at
calculating the adjustment to the feed gas flow so that the reactor can consume enough oxygen to
maintain the exit oxygen concentration below the LOV.
Now we shall consider the specific cases:
3.1.1Plant productivity limited by gas throughput

In these situations, customers are limited by the volume of gas that the process can
handle. Typically this is the air compressor, but often it is other processing equipment such as
flooding in downstream distillation columns or the reactor. In all these cases, reactor conditions
can be adjusted to process more material, but the process cannot handle more gas volume.
Because the oxygen is typically introduced after the air compressor (Fig. 1), and the oxygen
added is typically consumed in the reactor, EA allows oxygen addition without adding gas
throughput to most unit operations. In this scenario, we assume that the total flow of gas to the
reactor is the same for EA and Air operation:
6
gas feed  L N 2 1  EA   L N 2 1   Air   LTotal gas feed
EA EA Air Air
LTotal (5)
Next we solve for the ratio of nitrogen flows:
LEA
N2  1  Air 
Air    (6)
L N 2  1  EA 
Next substituting equation 6 into equation 4, we get the following result:

 1  Air   exit  EA 
EA     (7)
 1  EA   exit  Air 
If we assume that sparger material of construction limits enrichment to 25% (= 0.3333), we
calculate a 125% capacity compared to air with 21% oxygen in the feed (= 0.266).
3.1.2Plant productivity limited by mass transfer in reactor

In this scenario, the process productivity is limited by the mass transfer rate of oxygen in
the reactor. Because the mass transfer is proportional to the oxygen concentration in the gas
bubbles, productivity will be proportional to the oxygen concentration in the bubbles. Because
the concentration in the bubbles change as the oxygen is consumed, the analysis is a bit more
complicated. If the reactor is truly mass transfer limited, one can estimate the benefits of EA by
treating the problem similar to an absorption tower. The result of this analysis gives the
following ratio of nitrogen flows2:
exit
1 
LEA
    d
 Air
N2
Air  exit (8)
LN 2 1 
    d
EA
and the following result for enhancement:

exit
1     

exit  EA   
  
 d  exit  EA  ln exit   exit  Air 
 Air   Air  
EA  exit  (9)
1     

exit  Air   
  
 d  exit  Air  ln exit   exit  EA 
EA   EA  
In this case enrichment to 25% oxygen will give a 113% capacity increase over air
operation. Notice that this is less than the above case where EA is used simply to deliver more
oxygen to a reactor. Figure 3 summarizes this effect showing that for most practical cases, using
EA to overcome mass transfer limitations results in less benefit than using EA to overcome gas
throughput restrictions. Also, there is a limit solution for 100% oxygen, and at very high
concentrations the mass transfer limited system actually shows more capacity increase than the
compressor limited system. However, the solution is unrealistic at 100% oxygen feed because it
requires perfect conversion of oxygen to maintain the oxygen concentration in the headspace
2
The full analysis of this problem is contained in Appendix A.
7
below the LOV. Further, it is likely that the system would become kinetically limited (see
below) at higher oxygen concentrations.
9
Enhancement for mass
8 transfer limited
7 Enhancement for
compressor limited
6
5

4
3
2
1
0
0% 20% 40% 60% 80% 100%
Oxygen concentration in feed, mol fraction
Figure 3: Expected capacity improvements for oxygen used to debottleneck compressor or

to eliminate mass transfer limitations.
3.1.3Plant productivity limited by reaction kinetics in reactor

In this scenario, the reactor productivity is limited by the liquid phase reaction rate. This
causes some complications because we now have to concern ourselves with liquid phase oxygen
concentration. In the mass transfer case, this was not a concern because we assumed that the
reaction was fast enough to keep the liquid phase oxygen concentration near zero.
In order to analyze the reaction limited case, we need to understand the degree of
backmixing in liquid. Full hydrodynamic modeling or studies would be required to understand
this, and in fact, there is little consensus on how much backmixing there is in bubble columns or
stirred tanks. For this analysis, we will attempt to bracket the actual behavior of the column by
looking at the two extreme cases: (a) Liquid well-mixed, or (b) liquid and gas traveling together
in a bubble column. A stirred tank reactor (STR) is probably more like case (a), and a bubble
column is probably better modeled by case (b), but we should not expect either model to be
entirely accurate.
If we treat the system assuming the liquid is perfectly mixed, the oxygen level in the
liquid must be solved iteratively by determining a liquid phase oxygen concentration which
satisfies both the overall reactor mass balance and the local mass balance for the gas/liquid mass
transfer. This requires a numerical solution, and the results are not qualitatively different from
case (b) which can be solved analytically, so we will look closer at case (b).
The full derivation of the analytical solution for case (b) is contained in Appendix A. The
essentials are as follows:
8
Because the system is not mass transfer limited, the concentration of oxygen in the liquid
will be proportional to the local concentration of oxygen in the gas raised to the order of the
reaction, . Thus the local rate of reaction at any point of the reactor is given by the following:

  
r  k y 
 k   (10)
O2
1 
The lower case r denotes a local rate of reaction as opposed to an upper case R which denotes the
overall reactor productivity.
The analysis gives a similar result for the enhancement factor EA operation compared to
air operation as in the mass transfer case. The only difference is that the function to be integrated
is raised to the order of the reaction.
exit 
1 
  EA     d
EA 

exit  EA  LEA
N2

exit
 Air 
 
(11)
  Air  LAir exit 
1 
exit N2
 exit  Air   
 
 d
EA 
The integral has an analytical solution for integer orders of reactions, but must be solved
numerically for fractional reaction orders. Zero and first order cases are interesting. For zero
order reactions, equation (11) gives a value of 1 for enhancement regardless of oxygen feed
concentration. No enhancement is observed from increasing feed oxygen concentration for zero
order reactions. (This is also true for case (a), systems where the liquid is well-mixed). First
order reactions have an identical enhancement versus oxygen feed concentration relationship as
mass transfer limited reactions, because equation (9) and (11) are identical if  = 1.
Table 1 gives expected enhancement for 25% oxygen feed concentration and various
oxygen reaction orders. Fig. 4 shows the effect of oxygen feed level for various oxygen reaction
orders.
Table 1: Effect of reaction order on productivity enhancement by EA to 25% oxygen.

Reaction order in oxygen () Enhancement over air ()
0 1
0.5 1.07
1 1.13
1.5 1.19
2 1.23
9
10
9
8 =2
7
6  = 3/2
 5
 = 1 (or mass
4 transfer ltd)
3
2  = 1/2
1 =0
0
0% 20% 40% 60% 80% 100%
Oxygen concentration in feed, mol fraction
Figure 4: Effect of oxygen concentration in feed on productivity enhancement for various

oxygen reaction orders.
The curves in Fig. 4 might at first seem encouraging; exceptional improvements can be
attained by increasing oxygen in higher order reactions. However, liquid phase oxidation
kinetics are typically of low order in oxygen concentration, and are frequently zero order.
Therefore, in most systems, oxygen will provide the maximum benefit in the mass transfer
limited regime (typically at higher temperatures).
3.2Direct injection of oxygen (DI)

In direct injection, supplemental oxygen is injected as a pure component directly into the
LPO reactor instead of into the air feed line as in EA. Figure 5 illustrates.
10
Vent
Product
Reactor
Oxygen
Air
Figure 5: Direct injection of oxygen.

Direct injection of oxygen requires some modifications to the customer’s process, but it
offers several advantages over EA:
· There is no material of construction limitation on the amount of oxygen injected into
the process. Because the oxygen is injected directly into the reactor as a pure
component, the DI reactor sparger material of construction (MOC) is compatible
with pure oxygen. Thus, additional oxygen is introduced by increasing DI oxygen
flow, not by increasing oxygen concentration in the air spargers as in EA.
Whereas the oxygen level in EA is limited by MOC compatibility, the oxygen
level in DI is not.
· Oxygen can be introduced directly where it is needed. Often in reactors (particularly
in bubble columns) there are “dead zones” where very little gas bubbles are
present. Often these regions become oxygen-starved and by-product formation
reactions become favored. Thus, if oxygen is directly injected into these regions,
by-product formation will be reduced.
· In many circumstances, DI will provide more productivity improvement than EA for
the same amount of supplemental oxygen added. This will be seen in some of the
simplified cases below, and has also been verified with more detailed modeling.
Therefore, DI not only offers the productivity improvement of EA, but more flexibility
in terms of being able to add more oxygen than EA, and potentially improved selectivity. It is
important to note that the value of DI largely depends on the amount of coalescence of the DI
bubbles with the primary feed bubbles (from the air). If the DI bubbles mix with the primary
feed bubbles immediately, no advantage will be gained by DI over simply feeding the oxygen
with the air feed stream. However, if the DI bubbles remain separated from the primary feed, we
11
will get the maximum benefit possible. Obviously, the actual case is somewhere in between, and
understanding the amount of coalescence in DI is an active area in our gas-liquid mixing studies.
However for our purposes, we will assume that the DI bubbles do not coalesce with the primary
feed bubbles to determine potential benefit.
Now we will consider the impact of DI on productivity in the three scenarios discussed
for EA and compare.
3.2.1 Plant productivity limited by gas throughput

In this example, DI offers no additional productivity improvement because in this case,
we are trying to deliver more oxygen with the same amount of total gas flow. DI only changes
the location of the oxygen injection, but the same amount of gas flow is required. Given the
additional changes required to the process compared to EA, EA is preferable to DI for this
situation.
3.2.2Plant productivity limited by mass transfer in reactor

In this analysis, we will treat the DI oxygen as a parallel feed to the primary feed. Thus,
most of the analysis above and in Appendix A for EA will be applicable for the air stream.
However, because the oxygen is pure in the DI stream, we must treat the mass balance slightly
differently.
Theoretically, as long as there is little solvent vapor pressure, the oxygen bubbles will
shrink to extinction as the oxygen is absorbed into the liquid. This means that mass transfer
would increase because the interfacial area per bubble volume will increase as the bubble
shrinks. In this analysis, we will not account for the increased mass transfer, so this analysis will
underestimate the mass transfer of the DI oxygen slightly. With this assumption the mass
balance for DI oxygen is as follows:
feed exit * DI *
L DI ,O 2  L DI ,O 2  K MT y O 2  K MT (12)
Note that because the oxygen is pure, the concentration of oxygen in the bubble does not
vary. This allows one to solve without having to integrate over the bubbles as they travel through
the fluid.
This equation is extremely important. Note that if the mass transfer coefficient is
sufficiently large, the exit oxygen flow goes to zero, and equation (12) is invalid. Therefore,
equation (12) represents a criteria for full DI conversion, and should more properly written as
follows:
* feed exit exit feed *
if K MT  LDI ,O 2 , LDI ,O 2  0 , else LDI ,O 2  LDI ,O 2  K MT (12a)
In all DI applications, the most critical issue is whether the oxygen injected can be fully
consumed. Equation (12a) provides the criteria for determining whether the DI oxygen can be
converted. The solution is different depending on whether the DI oxygen is consumed or not.
We shall consider these cases separately:
DI at low supplemental oxygen levels—Complete DI oxygen conversion:
12
In DI, the total reactor productivity will be given by the reactor productivity from oxygen
delivered in the primary air feed, and the DI feed. The amount of reaction due to the DI oxygen
conversion is given by:
feed exit feed
LDI ,O 2  L DI ,O 2 LDI ,O 2
R DI   (assuming complete DI O2 conversion) (13)
 O2  O2
Similar to the EA analysis, it is convenient to express the DI oxygen flow to the ratio of
nitrogen fed in the primary feed, and the total oxygen feed to nitrogen in the primary feed as an
overall oxygen to nitrogen ratio, viz.:
feed feed feed
LDI ,O 2 LDI ,O 2  L Air ,O 2
DI  , overall  (14)
LN 2 LN 2
Solving for RDI,total:
LN 2 L
R DI ,total  R Air  RDI  
O 2
Air  exit  DI    N 2 overall  exit 
O2
(15)
Because the DI oxygen is fully converted, it has no effect on the headspace oxygen
concentration. Thus, the primary air flow (and nitrogen flow) in the DI process is the same as in
the air process, and when we ratio the reaction rates, the nitrogen flows cancel to give the
following expression:
LN 2
R DI ,total R Air  R DI
                
O 2 Air exit DI Air exit DI overall exit
DI      (16)
R Air R Air

LN 2
     Air   exit   Air  exit 
O 2 Air exit
Looking at an overall oxygen feed concentration of 25%, we can see that introducing
oxygen using DI provides a significant advantage over adding the oxygen to the air feed as in
EA. Equation 16 gives a result of 132% productivity enhancement using DI compared to only a
113% productivity enhancement using EA. This is for the same amount of oxygen molecules
delivered from an ASU. Figure 6 compares DI with EA for various overall oxygen feed
concentrations:
13
10
9 EA enhancement for
mass transfer limited
8
DI enhancement for
7
mass transfer limited
6
 5
4
3
2
1
0
0% 20% 40% 60% 80% 100%
Overall oxygen concentration in feed, mol fraction
Figure 6: Comparison of EA and DI at equivalent supplemental oxygen levels for a mass

transfer limited reaction. (Note: analysis assumes that all oxygen in DI feed is consumed).
As can be seen, DI offers significant advantage at the same level of supplemental
oxygen, but more importantly, because the DI system is customized for pure oxygen and the air
spargers are not exposed to elevated levels of oxygen, there is no MOC limitation on DI.
However, there is a limitation in that this analysis assumes complete oxygen DI conversion. A
complete analysis will require the understanding of this limitation which is discussed below.
DI at higher supplemental oxygen levels—Incomplete DI oxygen conversion:
In the above analysis, we have not determined the condition where DI oxygen
consumption is incomplete. The real limitation of DI is the criteria that all the DI oxygen be
consumed. The limitation on the maximum DI flow will depend on the mass transfer rate. Also,
in a bubble column, as oxygen concentration is increased nitrogen flow decreases, and this will
reduce the agitation from the gas bubbles. This means that the mass transfer coefficient will
probably decrease. Therefore, two things work against increasing oxygen concentrations: More
oxygen to consume, and reduced mass transfer coefficient.
The analysis of DI when the oxygen conversion is incomplete is numerically quite
difficult because one must determine how much DI oxygen escapes unreacted, and adjust the
target exit oxygen level for the air stream. A program has been written to do this, and the results
for a particular mass transfer coefficient are given by Fig. 7, for a system with an enhanced k La
of 10 min-1. (The enhancement is from homogeneous reaction in the boundary layer). This
enhanced kLa value corresponds to that observed in LOR experiments for the oxidation of p-
xylene to terephthalic acid (TPA).
14
10
6 No solution for greater O2

 5 feed concentrations, target
oxygen level in headspace
4 cannot be attained.
2 DI oxygen
completely DI oxygen
1 consumed breakthrough
0
20% 30% 40% 50% 60% 70% 80% 90% 100%
Overall oxygen feed concentration, mol%
Figure 7: Limitations on amount of DI oxygen for a system with k La = 10 min-1.
Modeling work shows that there is a very narrow region where incomplete conversion of
DI oxygen is stable. Further, this results in diminished performance and stability. When the DI
conversion is incomplete, often very little increase in DI oxygen can create a situation where not
enough nitrogen is available for maintaining the headspace below the LOV. Therefore, for
practical purposes, one must always make sure that the DI feed rate is such that the oxygen is
fully converted, and an understanding of the maximum DI oxygen level with full conversion of
DI oxygen is critical, and an a priori understanding of this would be preferable.
3.2.3Plant productivity limited by reaction kinetics in reactor

In a reaction limited system, performance is determined by the oxygen level in the liquid.
Therefore, predicting performance of in a kinetically limited system requires an understanding of
how having bubbles with widely varying oxygen concentration will effect the equilibrium
concentration of oxygen in the liquid. At this point, we do not feel that we have a satisfactory
understanding of these effects, and therefore we have not attempted to quantify benefits of DI in
these systems. We feel that fundamental studies of the effect of DI on dissolved oxygen are
necessary in order to understand the effect of DI on kinetically limited systems.
3.3Process re-optimization
3.3.1Background
The final application is really a method to enhance the EA or DI offering. In our
experience, we have seen that EA and DI have two major limitations.
15
First, EA and DI have the greatest advantage when used as needed when demand is
particularly high. (This will be discussed in greater detail in the economics section). In this way
these applications give flexibility—Use the oxygen when you need it to get plant capacity
beyond 100% and don’t use it when you don’t need it. Unfortunately, in order to get favorable
oxygen pricing, and reliable oxygen, the customer generally needs to guarantee that they will use
oxygen continuously. This significantly reduces the flexibility of this application.
Second, most liquid phase reaction kinetics are zero-order in oxygen, and as seen above,
little benefit is attained using oxygen in such a system under kinetically limited conditions. In
these situations, temperature must be increased to drive the reaction into the mass transfer
limited regime to take advantage of the higher concentration of oxygen. Fig. 8 illustrates.
0.01
Increase temperature and add oxygen,

significant increase in productivity.
Reactor Increase temperature with air.

productivity 0.001 Moderate increase in productivity.
(mol/L/min)
Air-based
process
operating point.
No advantage in
adding oxygen. 30% enrichment
Air
0.0001
90 120 150 180 210 240
Temperature, °C
Figure 8: Typical productivity versus reactor temperature curve for EA versus air
operation, showing that oxygen only improves productivity in mass transfer limited regime
(at higher temperatures).
The by-product formation typically increases with temperature, and as Fig. 8 illustrates,
the productivity increases at a slower rate with temperature in the mass transfer limited regime.
Therefore, it typically makes the most sense to run the reactor at the transition between mass
transfer and kinetically limited because increasing temperature at this point probably increases
by-product formation more than it increases productivity, and decreasing temperature
significantly reduces reactor formation. Therefore, this operating condition is probably the most
typical operating condition in the industry.
These two factors, the required commitment to continuous use of oxygen for the
customer, and the requirement that temperature be raised to gain productivity with oxygen use,
16
create significant barriers to entry. The advantage of flexibility is lost because during times when
no capacity increase is required, the oxygen still must be paid for even though it is not needed.
And when capacity increases are needed, the temperature often must be increased to get a
capacity increase causing an increase in by-products.
During the development of LOR for TPA, air and oxygen operation was compared. We
noticed many of the problems discussed above. We could easily get significant productivity
benefits at higher temperatures, but at lower temperatures, we did not see such an advantage. We
then tried to take advantage of the oxygen by increasing the catalyst concentration and this
worked for oxygen. Surprisingly, no improvement was observed in air-based operation. Instead
of increasing productivity, we could also decrease temperature in oxygen operation and get the
same productivity in the air process at a lower temperature. This gave lower by-products.
These observations cannot be explained with the simple mass transfer versus kinetically
limited model of liquid phase reactions. A full explanation is contained in the manuscript in
Appendix B. Briefly, these observations suggest that the reactor can be limited by catalyst. In air
operation, there is enough catalyst to activate the oxygen for oxidation—additional catalyst
merely is in excess. However, in oxygen operation, the concentration of oxygen is higher and
additional catalyst is required to activate the higher oxygen concentration. Thus, additional
catalyst can increase productivity in oxygen-based operation, but cannot improve productivity in
air-based operation. In short, to take the maximum advantage of oxygen, catalyst and oxygen
must be balanced so that neither is in excess. Most importantly, this occurs in kinetically limited
systems, so this allows one to improve kinetically limited systems using oxygen without
increasing temperature.
3.3.2Reducing by-product formation with process re-optimization

Again, in standard EA or DI, with a LPO reaction that has kinetics zero-order in oxygen,
we usually must increase temperature to move the reaction from kinetically limited to mass
transfer limited, so that we may take advantage of higher oxygen concentrations. The discovery
in the TPA project suggests we can instead increase catalyst to make the reactor mass-transfer
limited without increasing temperature. This means that productivity enhancement can be
accomplished without increasing by-product formation by using oxygen with increased catalyst
because the temperature would not need to be increased. More importantly, during times when a
capacity increase is not necessary, oxygen could still be used to allow for lower temperature
operation at the same productivity of the air process resulting in decreased by-product
formation.
Figure 9 illustrates the concept using the model developed for the TPA project. In this
example, the air-based reactor is running at 165°C and has a productivity of about 0.015
mol/L/min. Adding oxygen to 30% in the feed, gives a moderate increase in reactor productivity,
but not remarkable. However, doubling the catalyst loading and adding oxygen gives a
significant productivity improvement up to about 0.023 mol/L/min. When additional capacity is
not needed, one could reduce temperature to about 153°C without a loss in productivity.
However, if only the catalyst is increased and air is used, dropping the temperature to 153°C
would result in a loss of productivity from 0.015 mol/L/min to about 0.013 mol/L/min.
17
0.04
Air at base catalyst
30% EA at base catalyst
0.035 Air at 2X base catalyst
30% EA at 2X base catalyst
0.03
Increase oxygen and catalyst
for increased productivity
0.025
or…
decreased operating temperature
Reactor
Productivity 0.02
(mol/L/min)
Increase oxygen only:
0.015
little productivity
improvement
0.01
Air-based, base catalyst
operating point.
0.005 Increase catalyst only:
decreased operating temperature
gives lower productivity
0
120 130 140 150 160 170 180 190 200
Reactor temperature (°C)
Figure 9: Improving process performance with process re-optimization to take full

advantage of added oxygen.
It is possible that this concept could be extended to other chemistries, particularly
dimethyl terephthalate (DMT) production, which is similar to TPA production, or any catalyzed
LPO process. However, many processes are non-catalytic or autocatalytic. The kinetics of these
processes are often very complex, and each would have to be treated individually, but it is quite
possible that re-optimization of other variables, such as reactant or product concentration, may
yield similar results.
Where applicable, re-optimization solves the two problems of EA and DI discussed
above. It improves the efficacy of oxygen use such that it can be used both in kinetically and
mass-transfer limited systems. More importantly, it allows reduction in by-product formation
instead of productivity improvements. This means that oxygen can still improve the process even
during times of low demand, and the customer will then be able to use oxygen continuously
without the financial penalty of having to pay for unneeded oxygen.
The by-product reduction will be determined by cost of the feedstock, cost of by-product
remediation, and the change in by-product formation with temperature. These will vary with
process, but if one knows the activation energy of the by-product reaction relative to the
activation of the product reaction, the following analysis can be used.
18
We can determine the temperature reduction possible by equating the reaction rate using
oxygen at the lower temperature, T, to the reaction rate at the air process at the normal operating
temperature T0:
T T0
Rtotal ,O2  Rtotal ,air (17)
Next, we assume that the enhancement at the lower temperature is the same as the higher
temperature, so that:
T T
Rtotal ,O2  Rtotal ,air (18)
Then we substitute the difference in reaction rate to solve for the reaction rate in air at T.
  E prod  1 1 
T
Rtotal T0
,air  Rtotal ,air exp
    (19)
 R  T T0  
Combining equations (17), (18), and (19), we can eliminate the reaction rate terms:
  E prod  1 1  
1   exp    (20)
 R  T T0  
Therefore, we can determine the possible temperature reduction by knowing the enhancement
term, which we can determine by any of the methods outlined previously. However, it is more
useful to solve for the term 1/T - 1/T 0:
 1 1   ln() 
      (21)
 T T0   E prod / R 
At this point, we need to consider the byproduct reaction. We must determine the activation
energy associated with the by-product reaction and the relative effect oxygen has on the kinetics
at a given temperature. Initially, let us assume that productivity enhancement of the by-product
reaction at a constant temperature is the same as it is for the main reaction. Thus at the lower
operating temperature we can determine the by-product reaction rate based on the by-product
reaction rate using air at the higher temperature.
T   E byp  1 1  
T
Rbyp T0
,O2  Rbyp ,air  Rbyp ,air exp
    (22)
 R  T T0  
Substituting (21) into (22) we obtain the rather simple relationship:

  E byp 
T
Rbyp T0

,O2  Rbyp ,air exp ln (23)
 E prod 
Rearranging the exponential expression for further simplification gives the following:
 E 
 1 byp 
T  E prod  T0 (24)
Rbyp ,O2   Rbyp ,air
19
Generally, we would be most interested in the reduction of by-products possible from
using oxygen. The percentage reduction in by-product formation is given by the following
expression:
T0 T  E 
Rbyp ,air  Rbyp ,O2
 1 byp
 E prod

Brel  T0  1  
(25)
Rbyp ,air
Equation (25) can be generalized if the by-product reaction has a different order
dependence on oxygen than the main reaction. As seen above, the dependence of the overall
reaction rate with respect to oxygen may not be identical to the order of the oxidation step with
respect to oxygen. However, we would expect the relative order of the byproduct reaction to the
main reaction to approximate the ratio of the order of reactions, byp/. Assuming this in the
derivation above leads to the general expression:
T0 T   byp Ebyp 
Rbyp ,air  Rbyp ,O2
  
  prod E prod 
Brel  T0  1  (26)
Rbyp ,air
Thus, the percentage reduction in by-product formation only depends on three

parameters, , which is determined by the above analysis, the ratio of the order of the oxygen
dependence in the by-product and main reaction kinetics, and the ratio of the by-product to main
reaction activation energies.
Therefore, if one knows the relative order and activation energy of the by-product
reaction to the main reaction, one can estimate the by-product reduction. This will be a rough
estimate because it does not account for transitions from mass transfer limited to kinetically
limited at different temperatures, but it does provide some indication.
Most importantly, the methods for estimating benefits presented here and in Appendix B,
allow one to scout chemistries in simple lab-scale experiments to get kinetics of the main and
by-product reactions as a function of operating conditions, catalyst loading, and temperature.
Once these are determined, the methods described here can be used to estimate benefits to the
customer. If results are encouraging, we should be able to use the results to persuade customers
to do further studies in their own lab or pilot facilities under JD agreements.
4Previous and Current Customer Contacts

4.1Oxo-acids
Union Carbide Corporation: Commercialized LOR (pure oxygen process) for oxo-acids
at Texas City TX. Start-up of commercial facility in June 1997. The following was demonstrated
at lab, pilot, and commercial scale operation:
· 3X reactor productivity increase.
· 6 percentage point increase in selectivity.
· 10X reduction in vent gas emissions.
· Higher mass and heat transfer leading to better process control.
20
· A much simpler process leading to a 35% decrease in capital costs
4.2TPA
Eastman Chemical: A standard air enrichment test was performed the week of November
13, 1995 with Eastman Chemical under the direction of Anne Roby. Eastman Chemical
employs a train of bubble columns for the oxidation of p-xylene to terephthalic acid (TPA).
Praxair assisted Eastman in Columbia, South Carolina with a 25 mol% O2 air enrichment test.
Eastman is interested in increasing TPA capacity without increasing their total off-gas. In
addition, they are interested in the effect of enrichment on product quality and the oxidative loss
of the acetic acid solvent. Although the results of the test are currently being analyzed, Praxair
was able to increase the production capacity by approximately 17% in one of the oxidizers,
however, an increase in burn products was also detected. This increase was not sustainable due
to a problem which developed downstream of the oxidizer approximately 10 hours into the test.
Eastman could be a candidate for re-optimization.
Dow, Temex, Reliance and Amoco have all been interested in LOR for TPA production.
Praxair has had significant JD efforts with both Dow and Amoco in this area. The following has
been demonstrated in these studies:
· Safe operation of the LOR for oxygen-based oxidation for PX oxidation to TPA.
· More efficient oxygen utilization: >98% versus 80% for the air process.
· Lower acetic acid (HOAc) consumption: 2.5 g HOAc per 100 g TPA produced versus
4.1 g HOAc/100 g TPA in the air process. Additionally, it is believed that an even
lower acetic acid consumption rate of 2.0 g/100 g can be attained with further
process optimization of the oxygen-based process.
· Higher selectivity of PX to TPA reaction: 99% versus 97.5% for the air process.
· Up to 65% higher productivity in the air process can be attained in lieu of improved
TPA selectivity and reduced acetic acid consumption. This offers a TPA producer
flexibility to meet fluctuating market demands.
Based on recent JD efforts with Dow, and data generated independently of Dow, we are
currently negotiating terms of an exclusive license for an LOR based TPA process. This license
would not cover DI or EA for TPA, and we could easily use learnings generated outside of the
Dow/PX JD efforts, to create a technology package for EA or DI with process re-optimization.
Reliance may be a good candidate for EA/DI.
4.3DMT (Witten Process)

Huls: A proposal for direct injection was submitted to Huls in June, 1995. Huls produces
DMT in a train of six bubble columns. They were interested in testing direct injection for
increasing production up to 50%. Some tests were conducted at Huls under a JD agreement, and
PX has the reports of the results. Although the direct injection technology was well received by
Huls, they were able to debottleneck their process and achieve a capacity increase using
alternative methods.
GTC, KoSa: Intermittent discussions between Praxair and GTC had been ongoing for
most of 1998 regarding the possibility of DI for DMT. Currently, there are no plans for a JD,
21
although this has been discussed. Anne Roby, Mike Hogan, and Bill Williams met with KoSa in
Wilmington, NC in November of 1998 to discuss the possibility of EA or DI for DMT at the
Wilmington plant. A two-way secrecy is being planned and we have offered to assist them in
oxygen-based laboratory studies to estimate the benefits of supplemental oxygen in their process.
4.4Cyclohexane to K/A Oil

JD with ABB Lummus Global: In lab-scale studies of cyclohexane oxidation to K/A oil, a
nylon intermediate, the following was demonstrated:
· 3-4X productivity improvement.
· 4-10 percentage point selectivity improvement.
Nitrocarbono: LOR for cyclohexane oxidation to K/A oil, a nylon intermediate, has been
seriously considered by Nitrocarbono in Brazil with SAWM. Limited funds and concern of risk
from both parties has prevented commencement of pilot studies. This could be a candidate for
EA or DI as this is much less risky and capital intensive than an LOR retrofit.
4.5Cumene hydroperoxide for phenol production

Mitsui, Georgia Gulf, BTL, and Allied Signal have all expressed interest in EA for
cumene oxidation to cumene hydroperoxide (CHP). UCC also explored this before the
divestiture of PX, and PX is in possession of a model for CHP production. HAZOP studies were
conducted for both BTL and Ga. Gulf, spargers were designed, and both projects were close to
commercialization. However, in the case of BTL, the air compressor was leaking enough oil into
the line to make oxygen addition unsafe, and the cost of replacement or upgrade of the
compressor was prohibitive. In the case of Ga. Gulf, changing business priorities lowered the
priority of this project, and final approval was never attained.
This is an autocatalytic process with complex kinetics so it is not clear that process re-
optimization would be applicable, but regular DI or EA certainly has potential.
4.6Propylene oxide/styrene monomer (PO/SM) co-production

In PO/SM co-production ethyl benzene is oxidized to ethyl benzene hydroperoxide
(EBHP) which is an intermediate in PO/SM production. Repsol, Arco, and Shell have expressed
interest in EA for PO/SM. Repsol is particularly interested and we have recently done some
simulations (Nov. 1998) showing that if a plant is compressor limited, process re-optimization
may allow for improved selectivity, and straight EA may provide improved productivity.
4.7Acetic acid
Hoescht-Celanese is currently using EA for liquid phase acetic acid production. We
could explore re-optimization with them.
5Economic Evaluation of LPO Applications

The key to the success of EA, DI, or re-optimization is to provide enough savings in
capital deferral or in by-product reduction to justify the cost of oxygen. As pointed out in the
previous section, EA and DI predominantly provide increased capacity, and re-optimization
allows for the flexibility to also reduce by-products during times of lower demand.
22
5.1EA or DI for Capacity Increase
For capacity increase, the cost of oxygen is normally insignificant compared to the value
of additional product. Therefore, the net monetary benefit of EA or DI is usually readily
apparent. Rather, the issue generally is one of EA and DI as capital deferment because the
increased capacity can usually be attained by a capital investment (CI) such as a larger
compressor, larger reactor, and/or a larger separation train.
The evaluation of the cost of providing additional capacity will typically come down to a
capital versus operating cost decision. Obviously CI will be mostly a capital investment and EA
or DI will mostly be an operating investment. However, both options have capital and operating
costs. In the CI option, additional energy will be required to compress additional air to produce
the additional product. In EA, a small amount of capital will be required for an oxygen delivery
and control system, and in DI, further investment will be required on reactor modifications such
as the oxygen sparger. Therefore, on the capital side we will credit CI for the investment which
would be required to provide the oxygen in EA or DI, and on the operating side, we will credit
EA or DI for the energy which would have been required for additional air compression.
Therefore the net comparative cost of CI for expansion is given by the following:
Net Cost = (Capital investment for CI) - (Capital investment for EA or DI)
or…
ICI = ICI - IEA or ICI = ICI - IDI (27)
where…
ICI = Net capital investment (after credit for EA investment).
ICI , IEA, and IDI is the capital required to implement a nameplate capacity expansion, enriched air,
or direct injection respectively. Because the economics of DI are so similar to EA, we will only
refer to EA from here on and unless noted, this analysis is valid for DI as well.
In EA and DI, any additional capacity requires the purchase of oxygen. Therefore the
cost of attaining the additional capacity will be the cost of oxygen. In order to evaluate the
economics, we need to determine the cost per metric tonne of additional product, CO, which will
be the cost of oxygen per tonne, RO of oxygen times the tonne of oxygen required to produce an
additional tonne of product, f.
Cost of oxygen per tonne of product  CO  fRO (28)
It should be pointed out that this assumes that the additional oxygen will produce the
additional product. As discussed previously, knowledge of the kinetics and mass transfer
limitations of the project will allow one to determine the optimal oxygen level. This should be
done before this analysis is undertaken, or at least realistic estimates should be made.
As pointed out above, we will take credit for the energy which would have been required
to deliver the additional air for the additional product. We will define CE as the energy cost
23
associated with producing a tonne of product3. Therefore the net cost of oxygen per tonne of
product is given by the following:
Net cost of oxygen per tonne of product for enrichment  CEA  CO  CE (29)
Therefore, the cash flow required for enrichment or DI is proportional to the additional
product produced determine in tonne per annum, P. Because enrichment is an ongoing expense,
to compare the enrichment versus a capital investment, one must use some economic model for
comparing initial investment versus a series of payments. One would choose the option having
the lowest cost to attain a given capacity. For the analysis in this paper, we have chosen to
amortize the capital investment in one year payments over the life of the project (indicated by
the PMT function). Therefore, the annual cost of the oxygen for EA or DI must be lower than
the annual payments on the capital investment as given by equation 30.
C EA P  PMT I CI  (30)
We will normalize the annual expense to the additional tonnes per annum (tpa) to give a cost per
tonne of additional capacity using the modified relationship:
PMT I CI 
C EA  (31)
P
Figure 10 is a comparison of oxygen costs versus the amortized annual payments at 12%
APR for a 15 year project life on the capital for a capacity expansion. The line defines the
inequality (31).
$80
$70
Choose capital
$60 investment
Net oxygen
$50
cost per
$40
tonne of
product. $30
$20
Choose air enrichment
$10
$0
$0 $100 $200 $300 $400 $500 $600
Net capital investment deferred by using air enrichment
per additional mtpa production capacity.
Figure 10: Supplemental oxygen (EA or DI) versus amortized payments on initial capital
investment (15 annual payments @ 12% APR),
3
The energy required to produce a tonne of product will actually change because the air compressor efficiency
will change with volume of gas delivered. For purposes of this paper, we will assume the air compressor
efficiency is the same at varying gas volumes.
24
normalized to additional capacity in metric tonnes per annum (mtpa).
What is not in this analysis is how much of the extra capacity will actually be used. If the
capacity is not used, any capital investment will be idle and thus wasted. In theory, this would
not be the case for EA or DI, but because of long-term oxygen contracts, the customer would
still have to pay even though the supplemental oxygen is not necessary. However, if re-
optimization is used, the oxygen could simply be used for by-product reduction. Thus, re-
optimization is a hedge against market downturns as described below.
5.2Re-optimization for Capacity Increase or By-product reduction.

As shown above, re-optimization allows for either a capacity increase without increasing
by-product formation or a reduction in by-product formation without a reduction in capacity.
Thus, during times of high demand, the cost of oxygen is paid for by increased productivity, and
during times of low demand, the cost of oxygen is paid for by reduction of by-products. Fig. 10
illustrates.
Oxygen used for productivity increase during peak demand.
Increased production $ >> Oxygen $
120%
Demand/Capacity
110%
100%
90%
80%
Time
Oxygen used for selectivity during turndown.

Increased selectivity $ oxygen cost.
Figure 11: Enrichment with re-optimization. Strategy for using selectivity or productivity
to pay for oxygen.
The savings in by-product reduction will be determined by cost of the feedstock, cost of
by-product remediation, and the change in by-product formation with temperature. The
percentage by-product reduction can be determined by experiment or estimated by equation (25)
or (26). Once this is determined, the process savings from by-product reduction can readily
determined by multiplying the cost of feedstock material current material cost of the feedstock
which is lost to by-products in the current air process:
Csavings  Brel Current $ loss of feedstock to by - product  (32)
25
Next, one would add the material savings given by equation (32) to other process
savings, such as lower capital, and lower vent gas treatment costs associated with the reduced
gas flow in an oxygen-based, or oxygen-enriched process.
5.3An example: Enrichment or DI for TPA with re-optimization

To illustrate, let us consider EA or DI to 25% for TPA production with process re-
optimization. There are two extreme scenarios: The high demand scenario is when one uses the
oxygen for productivity improvements, and the low demand scenario where one uses the oxygen
for selectivity improvements.
5.3.1Capacity expansion: Capital improvement versus supplemental oxygen

We believe that by increasing the catalyst concentration, we can drive the reaction into
the mass transfer limited regime, so we will estimate the benefit by using the analyses above. For
EA we will use equation (9) to estimate the productivity enhancement at 113%, and for DI we
will use equation (16) to give a productivity enhancement of 132%. The alternative is a capital
investment for a capacity increase. Table 2 analyzes capital investment versus supplemental
oxygen for both cases:
26
Table 2: Comparison of using supplemental oxygen (EA or DI) versus a capital investment
for plant capacity increase.
Reactor information: EA DI
Nameplate capacity (kmtpa) 250 250
Oxygen as fraction of gas feed (v/v) 25% 25%
Actual productivity enhancement 1.13 1.32
Actual capacity 282.5 330
Total additional product produced (kmtpa) 32.50 80.00
Supplemental oxygen costs:

tonne of oxygen required per tonne of product 0.61 0.61
tonne ASU oxygen required/tonne of product 0.12 0.12
ASU oxygen required (kmtpa) 34.90 40.77
Net oxygen cost (- compressor energy cost, per tonne product) $ (35.00) $ (35.00)
Net annual cost of oxygen (1000s) $ (1,222) $ (1,427)
Cost of supplemental oxygen/tonne additional product $ (38) $ (18)
Capital analysis (new plant design modification):

Capital cost of new plant construction at nameplate capacity (MM) $ (68.60) $ (68.60)
SRI capacity increase exponent 0.69 0.69
Size increase cost factor 1.09 1.21
Capital cost of new plant construction at expanded capacity (MM) $ (74.64) $ (83.08)
Capital cost for supplemental oxygen skid avoided (MM) $ 0.20 $ 1.00
Net additional cost to build plant at new capacity (MM) $ (5.84) $ (13.48)
Net capital cost for each additional tonne/yr in capacity (new plant) $ (180) $ (169)
Amortized cost of capital/tonne of additional product (15 yrs at 12%) $ (26) $ (25)
Net savings of supplemental oxygen over amortized cost of capital $ (11) $ 7
per additional tonne product
Capital analysis (expansion of capacity with additional plant line)

SRI capacity decrease exponent 0.76 0.76
Cost factor: Cost of small additional plant/Cost of construction of new plant 0.21 0.42
Capital cost of expansion (MM) $ (14.55) $ (28.86)
Credit for cost of supplemental oxygen skid avoided (MM) $ 0.20 $ 1.00
Net additional cost to build plant at new capacity (MM) $ (14.35) $ (27.86)
Net capital cost for each additional tonne/yr in capacity (expansion) $ (442) $ (348)
Amortized cost of capital/tonne of additional product (15 yrs at 12%) $ (65) $ (51)
Net savings of supplemental oxygen over amortized cost of capital $ 54 $ 58
per additional tonne product
Following the analysis in the table, first we analyze the cost of oxygen. Each tonne of
PTA requires 0.61 tonnes of oxygen, so a 250 kmtpa PTA plant will use 152.5 kmtpa of oxygen
delivered with air. For 25% EA the predicted expansion is 13% to about 283 kmtpa PTA (33
kmtpa additional product), so the total oxygen demand will be 172.3 kmtpa oxygen.
Approximately 1/5 of the oxygen demand will be provided from the ASU or about 34.9 kmtpa
ASU oxygen required to produce 33 kmtpa additional product.
27
Oxygen delivered as 25% DI has more impact and capacity expands by 32%. The new
capacity will be 330 kmtpa PTA (80 kmtpa additional product). The total oxygen demand for the
expanded plant will be 201.3 kmtpa oxygen. 40.8 kmtpa (1/5 of the total oxygen demand) will
be provided by the ASU to provide the additional 80 kmtpa PTA.
The cost of oxygen is always difficult to estimate, because of regional and circumstantial
differences in oxygen pricing. Given that these will typically be on-site plants, we can estimate
the oxygen price between $33 and $44 per metric tonne. We must also take credit for the
compressor energy savings, but these typically delivering oxygen as air is a small fraction of the
oxygen cost (about $2-3/tonne oxygen delivered), so given the wide fluctuation in oxygen
pricing, we have chosen to take a small credit on oxygen price to account for compressor
savings. For purposes of this report, we will use $35/m.t. of oxygen.
Using these assumptions, the cost of supplemental oxygen is around $38/tonne of
additional TPA for EA and $18/tonne of additional TPA. The difference is due to the fact that
the supplemental oxygen is used much more efficiently in DI than EA.
The cost for supplemental oxygen is then compared to the cost to expand the plant’s
capacity by capital investment. Capital investment estimates were made based on SRI reports.
SRI reports the total installed capital investment (ex. utilities and contingencies) for the
oxidation section of a 250k metric tonne per annum (kmtpa) plant at $68.6MM. The cost of
expanding capacity by equipment investment will depend on whether one is increasing the
design size of a new plant, or adding capital (such as a new reactor or compressor) to expand an
existing plant.
5.3.1.1Capacity increase of a new plant

For a capacity increase for an new plant, we estimate the total new larger plant cost using
the capacity exponent for an increase, 0.69, so the added capital of requirement for 113%
capacity for a net additional 32.5 kmtpa (EA equivalent) or 132% capacity for a net additional
80 kmtpa (DI equivalent) would be calculated as follows:
$68.6MM * 1.130.69 - $68.6MM = $6.04MM for EA equivalent
or $68.6MM * 1.320.69 - $68.6MM = $14.5MM for DI equivalent
We then take a credit for the cost of installing a supplemental oxygen skid which would
not be necessary in a capital investment. Typical cost of EA installation is quite small, about
$200M. For DI, the installation cost will vary depending on how difficult it is to add a sparger to
the reactor, but we might expect the cost to be somewhere around $1MM. This gives the net
capital in Table 2 of $5.84MM to match the capacity expansion of EA and $13.48MM to match
the capacity expansion of DI. Normalizing the net capital investment per additional capacity
gives $180/mtpa for EA equivalent, and $169/mtpa for DI equivalent.
To compare variable cost with capital investment, we assume that the capital is paid off
in yearly installments over the project life. For our purposes, we will assume a 15 year project
life and an APR of 12%. This gives an annual cost of capital which when divided by the annual
additional tonnes of TPA produced, gives a cost per additional tonne of capacity. $26/tonne to
expand to the equivalent of EA capacity, and $25/tonne to expand to the equivalent of DI
capacity.
28
The analysis in Table 2 shows increasing the design size of a new plant before
construction is better than using EA, and DI is only marginally better than a capital expansion
($7/tonne). This is typical; economies of scale at the design stage will typically favor
incremental investment. It is, in fact, remarkable that the analysis shows DI even breaking even
with incremental capital investment.
5.3.1.2Capital expansion of existing plant

This is difficult to analyze because circumstances will differ, but the key is that the
economies of scale are no longer present in these scenarios. As an example, Table 2 considers
adding a small additional line to augment an existing 250 kmtpa plant. Although it would be
unlikely anyone would place a small reactor train aside a the main reactor train, one could
estimate the cost of this using the SRI capacity decrease exponent (0.76) to get a 13% size train
or a 32% size train viz.:
$68.6MM * 0.130.76 = $14.6MM for EA equivalent.
or $68.6MM * 0.320.76 = $28.8MM for DI equivalent.
These are substantially higher numbers and now EA or DI are clearly favored, by $54
and $58/tonne of additional TPA capacity respectively.
While the absolute numbers of these analyses are obviously not accurate, it clearly shows
why expansions are best as additions of entire lines, not small plant additions. In the following
section, we will describe how EA or DI can be used with process re-optimization to bridge the
“in between” times, when demand outstrips capacity, but does not justify the doubling of
capacity by adding an entire line.
5.3.2EA or DI for capacity expansion: Variable cost analysis

Once we decide to use EA or DI over capital expansion, we can see that for a capacity
expansion, additional profits will easily pay the net variable cost of oxygen. The first two
columns of Table 3 show the cost analysis for EA and DI. In this analysis, we have imbedded
the credit for compressor energy savings into the oxygen price as a credit and used a net oxygen
cost of $35/tonne:
29
Table 3: Cost analysis of EA and DI used for capacity increase or by-product reduction for
PTA production4.
EA for DI for EA for by- DI for by-
capacity capacity product product
increase increase reduction reduction
Reactor information:
Nameplate capacity (kmtpa) 250 250 250 250
Oxygen as fraction of gas feed (v/v) 25% 25% 25% 25%
, calculated at constant temperature 1.13 1.32 1.13 1.32
Actual productivity enhancement 1.13 1.32 1 1
Actual capacity with DI or EA 282.5 330 250 250
Ratio of by-product/main activation energies 1.85 1.85 1.85 1.85
By-product reduction (%) 0% 0% 10% 21%
Material information:
tonne of oxygen required per tonne of product 0.61 0.61 0.61 0.61
tonne ASU oxygen required/tonne of product 0.12 0.12 0.12 0.12
p-Xylene requirement/product w/w 0.68 0.68 0.68 0.68
Acetic acid lost to byproduct per product (w/w) 0.058 0.058 0.052 0.046
Product sale price (/tonne) $ 629.00 $ 629.00 $ 629.00 $ 629.00
Net oxygen cost5 (/tonne) $ (35.00) $ (35.00) $ (35.00) $ (35.00)
p-Xylene cost (/tonne) $(313.50) $(313.50) $ (313.50) $ (313.50)
Acetic acid cost (/tonne) $(792.00) $(792.00) $ (792.00) $ (792.00)
Cost analysis:
Total additional product produced (kmtpa) 32.50 80.00 - -
Additional oxygen required (kmtpa) 34.90 40.77 30.89 30.89
Additional p-xylene required (kmtpa) 22.10 54.40 - -
Additional acetic acid required (kmtpa) 1.89 4.64 (1.43) (3.05)
Additional sales of product (1000s) $ 20,443 $ 50,320 $ - $ -
Cost of oxygen (1000s) $ (1,222) $ (1,427) $ (1,081) $ (1,081)
Additional p-xylene cost (1000s) $ (6,928) $(17,054) $ - $ -
Additional acetic acid cost (1000s) $ (1,493) $ (3,675) $ 1,133 $ 2,414
Net additional profit (1000s) $ 10,800 $ 28,164 $ 52 $ 1,333
For the expansion of a world class plant of 250 kmtpa PTA production, EA would
provide 33 kmtpa additional product for $11MM additional annual profits, and DI would
provide 80 kmtpa additional product for $28MM additional annual profits.
Thus EA and DI will produce substantial profits, the only question is whether this is
better than a capital expansion which is covered by the analysis above. However, the evaluation
must not only consider the high demand scenario, but it must also consider the possibility that
the high demand is only temporary and that there will come a time when demand is only at
capacity or even slightly below.
4
All materials cost data in this report are from the January 29, 1999 issue of Chemical Market Reporter unless
otherwise noted.
5
Net oxygen cost includes a credit for compressor energy savings because the cost of the oxygen is for oxygen
delivered at reactor pressure whereas air must be compressed.
30
5.3.3EA or DI for by-product reduction: Variable cost analysis
During times of low demand, the extra capital will be of no use because there would be
no need for expanded capacity, but oxygen can be used to reduce by-products. For simplicity,
we will assume that the demand is equal to capacity but one can easily consider less than 100%
capacity scenarios using this same analysis.
Because we will not be producing any more product than the air process, we cannot take
credit for additional profits on product as above. Rather, we must consider by-product reduction.
In the case of TPA, we have laboratory data which suggests that replacement of air with oxygen
reduces acetic acid losses by nearly 50% from 58 kg/tonne PTA to 30 kg/tonne PTA. Process
modeling estimates that increasing oxygen to 25% concentration and using process re-
optimization could potentially reduce the acetic acid losses by 10% for EA and 21% for DI. This
is a reduction in acetic acid losses of about 6 kg/tonne for EA and 12 kg/tonne for DI. Also,
notice that in the above analysis, we considered only the incremental capacity increase. Because
the reduction in HOAc is a percentage of the entire PTA capacity, the cost of oxygen must be
analyzed over the entire capacity. Thus, only the oxygen that comes from an ASU is an expense.
The analysis is in columns 3 and 4 of Table 3. Notice that the cost savings comes from
by-product reduction, which is indicated as a negative number until the row heading “additional
acetic acid required”. This is the source for the savings to pay for the oxygen.
As can be seen in Table 3, by-product reduction does not provide as much financial
benefit as a capacity increase, but without re-optimization, one would still have to pay for the
oxygen during these times, but it would not be needed resulting in unnecessarily increased costs.
Also, if one chose a capital investment, as shown above, re-optimization with air has limited
value, so the expanded plant size would do no good in this situation.
Thus, one can use EA or DI with re-optimization to provide substantial benefit during
peak demand times by increasing plant capacity, and “ride out” economic downturns by
reducing by-product formation to roughly break even with the cost of oxygen.
This does complicate the question of capacity expansion versus EA or DI. It will depend
predominantly on what one expects from the market. If one expects the high demand to be
temporary, EA/DI with process re-optimization is probably a better choice than a capacity
expansion. If one expects high sustainable growth, a plant expansion is probably warranted.
However, the economies of scale are such that one would probably want to delay expansion until
a larger expansion can be justified such as an entire additional PTA line. In this way EA or DI
could be used as a bridge between times where only a fractional increase is warranted to the time
when an entire new line is warranted.
Figure 11 illustrates a possible approach to a PTA producer with 15% growth using
capacity expansion and EA/DI to meet demand throughout the growth period.
31
4.5
Demand / original capacity
4 Current capacity/original capacity
3.5 EA/DI used for capacity increase
3 EA/DI used for by-product reduction
Relative 2.5
demand or Year 2:
capacity 2 EA/DI use
commences
1.5
0.5
0
0 2 4 6 8 10 12
Year
Figure 12: Use of EA/DI and capital expansions to meet market demand.
In this example, the producer slightly oversizes the initial plant. At year 2, demand
begins to outpace capacity and EA/DI is instituted for a capacity increase. This delays the need
for a capital investment until about year 4, when a full additional PTA line is needed, not just a
partial expansion. At this point, the customer does not need EA or DI for capacity expansion, but
because he is committed to a take or pay contract, he continues to use the oxygen with process
re-optimization. The by-product reduction pays for the oxygen during years 5 and 6. Around
year 7, the expanded plant can no longer meet demand, and EA/DI can again be used for
capacity expansion. This continues with periods of EA/DI for capacity increase, followed by a
capital plant expansion, followed by EA/DI for by-product reduction and so on.
5.4DMT (Witten Process)

Table 4 contains an analysis for DMT production. DMT production is similar to PTA: It
involves p-xylene (PX) oxidation, except that p-xylene is also the solvent, and in the Witten
process, the oxidation is done in two steps with intermediate esterfication with methanol. The
basis for the by-product formation rate is based on KoSa’s estimate that approximately 6% of the
PX is consumed to carbon oxides. By-product reduction estimates are based on analogy to
experience with PTA production. Also, note that because p-xylene is both converted to product
and lost to by-product (unlike in PTA where acetic acid is the predominant component lost to
by-products), in this case by-product reduction is doubly beneficial: Less p-xylene is lost, and
less p-xylene is required to synthesize the same amount of DMT. The methanol used in DMT
synthesis is liberated upon polymerization and recycled. Because the cost of recycle varies with
process specifics, we have not included the cost of methanol in this analysis.
32
Table 4: Cost analysis of EA or DI for DMT production.
Actual productivity enhancement 1.13 1.32 1.01 1.01
Actual capacity 282.5 330 252 253
Reactant overall requirement/product (w/w) 0.63 0.63 0.63 0.63
Reactant converted to product/product w/w 0.592 0.592 0.596 0.600
Reactant lost to byproduct per product (w/w) 0.038 0.038 0.034 0.030
Oxygen cost (/tonne) $ (35.00) $ (35.00) $ (35.00) $ (35.00)
Reactant cost (/tonne) $(313.50) $(313.50) $ (313.50) $ (313.50)
Cost analysis:
Total additional product produced (kmtpa) 32.50 80.00 1.57 3.35
Additional reactant required for prod. (kmtpa) 19.25 47.38 - -
Additional reactant to by-product (kmtpa) 1.2 3.0 (0.88) (1.91)
3 2
Additional sales of product (1000s) $ 17,875 $ 44,000 $ 866 $ 1,845
Cost of oxygen (1000s) $ (841) $ (982) $ (749) $ (754)
Additional cost of reactant to prod. (1000s) $ (6,034) $(14,852) $ - $ -
Additional cost of reactant to by-prod. (1000s) $ (385) $ (948) $ 277 $ 600
Benefits from productivity enhancement are similar to PTA. By-product analysis for
DMT (and the other chemicals) is different from the PTA analysis because the p-xylene
(reactant) is consumed to make both the desired product and the by-product. This means that
when by-product formation is reduced, more reactant is available to make the main product.
Thus in DMT, one not only saves from reactant losses, but as seen in the above table, some
additional product is made with no additional reactant needs. This is seen by the fact that the
capacity is slightly higher than nameplate with by-product reduction. This will also be the case
for the other chemistries analyzed in the following sections.
5.5Caprolactam via cyclohexane oxidation

Table 5 summarizes the economics for the nylon 6 intermediate, caprolactam,
synthesized by cyclohexane oxidation to K/A oil (cyclohexananone/ol Ketone/Alcohol oil)
followed by oximation to caprolactam. EA or DI would be used in the oxidation step. Adipic
33
acid, a nylon 66 intermediate also can be synthesized by this process, but the cost analysis will
be different.
Table 5: Cost analysis for DI or EA for caprolactam production.

Actual capacity 158.2 184.8 140 141
Product sale price (/tonne) $1,980.00 $1,980.00 $ 1,980.00 $ 1,980.00
Reactant cost (/tonne) $ (380.00) $ (380.00) $ (380.00) $ (380.00)
Cost analysis:
Cost of oxygen (1000s) $ (487) $ (568) $ (432) $ (434)
Additional cost of reactant to prod. (1000s) $ (6,147) $ (15,131) $ - $ -
Additional cost of reactant to by-prod (1000s) $ (919) $ (2,261) $ 142 $ 322
As can be seen, substantial profits can be realized by increasing capacity because of the
high value of caprolactam. By-product reduction was estimated by extrapolating LOR results
which suggested doubling productivity could be traded for an 8-10% selectivity improvement.
This was used to estimate the by-product/product activation energy ratio.
5.6Phenol (via cumene hydroperoxide)

Most phenol is co-produced with acetone by oxidizing cumene to cumene hydroperoxide
(CHP), and then dissociating CHP to make acetone and phenol. The oxygen would be used in
CHP synthesis. The below analysis (Table 6) assumes no by-product credit for additional
acetone, as this is usually not the major source of income in a phenol process. The activation
energy ratio of by-product/product reaction was based on internal UCC reports of enrichment
studies for CHP synthesis.
34
Table 6: Cost analysis for EA or DI for phenol production via cumene hydroperoxide.
Actual capacity 101.7 118.8 91 92
Cost analysis:
Cost of oxygen (1000s) $ (238) $ (278) $ (213) $ (215)
Additional cost of reactant to prod. (1000s) $ (4,797) $ (11,807) $ - $ -
Additional cost of reactant to by-prod (1000s) $ (306) $ (754) $ 331 $ 727
It is important to recognize that cumene oxidation to CHP is predominantly practiced

without a catalyst, and even in processes which use a catalyst, the catalyst is not necessary for
CHP production. The process is autocatalytic (the product is the catalyst) and there may be some
opportunity for process re-optimization, but this is not certain. For this reason, we must consider
the probability of success somewhat lower for this program than for catalytic processes such as
TPA, DMT, or K/A oil synthesis.
5.7Propylene oxide co-production with Styrene Monomer (PO/SM)

In the PO/SM process, ethylbenzene is oxidized to ethylbenzene hydroperoxide (EBHP).
This is the step where oxygen would be used. EBHP is then reacted with propylene to make PO
and SM. The major difference in the PO/SM process is that most of the by-products in the EBHP
reactor can be converted to SM, but PO cannot be produced from these by-products. This
complicates the analysis, but the fundamental benefits are similar. Table 7 summarizes:
35
Table 7: Cost analysis of EA or DI for propylene oxide co-production with styrene
monomer (PO/SM)
Actual capacity 226 264 201 202
EB fed/ PO produced (w/w) 2.49 2.49 2.49 2.49
EB fed to PO/PO produced (w/w) 2.151 2.151 2.159 2.169
EB to SM only/PO produced (w/w) 0.339 0.339 0.331 0.321
SM produced/EB fed (w/w) 0.918 0.918 0.918 0.918
PO net sale price (- propylene cost, in $/tonne) $1,161.16 $1,161.16 $ 1,161.16 $ 1,161.16
SM sale price (/tonne) $ 462.00 $ 462.00 $ 462.00 $ 462.00
Cost analysis:
Total additional PO produced (kmtpa) 26.00 64.00 0.73 1.61
Total additional SM produced (kmtpa) 59.44 146.30 (1.22) (2.73)
SM/PO production ratio (w/w) 2.29 2.29 2.27 2.25
Additional EB to PO and SM (kmtpa) 55.94 137.69 - -
Additional EB to SM only (kmtpa) 8.80 21.67 (1.33) (2.98)
Additional sales of PO (1000s) $ 30,190 $ 74,314 $ 847 $ 1,871
Additional sales of SM (1000s) $ 27,459 $ 67,592 $ (566) $ (1,263)
Cost of oxygen (1000s) $ (1,314) $ (1,535) $ (1,167) $ (1,172)
Additional cost of EB to PO and SM (1000s) $ (31,318) $ (77,091) $ - $ -
Additional cost of EB to SM only (1000s) $ (4,930) $ (12,135) $ 747 $ 1,667
Net additional profit (1000s) $ 20,088 $ 51,146 $ (139) $ 1,104
In this analysis, credit needs to be given for the SM produced from the by-products.
Also, PO synthesis from EBHP requires propylene. Rather than adding another line, the PO
price has been reduced to a net price which accounts for the cost of propylene to make the PO.
Like CHP, EBHP synthesis is autocatalytic and no additional catalyst beyond the product
is required. Therefore, it is not clear that process re-optimization will be successful for this
chemistry. Due to a lack of data on relative activation energy, we chose to assume that the
relative activation energy of EBHP is equal to CHP. Therefore, of all the examples discussed
here we are least certain of these economics and probability of technical success.
36
6Strategy and Prioritization
As can be seen above, there is a wide variety of LPO applications available to the
customer. More importantly, we currently have a good fundamental understanding of the
strengths, weaknesses, and applicability of each application. Because processes, economics, and
the specific business of individual customers, each chemical and customer will require a unique
evaluation to determine the best application. However, with our current understanding, we
should be able to conduct generic lab experiments to understand the kinetics of each process, and
make evaluations based on the technical and economic models outlined above. First, we will
detail a generic approach to identifying the best application (or whether there is an appropriate
application) for each customer. This generic strategy will allow us to react to market inquiries.
However, we also would like to take a proactive approach and work on some strategic chemicals
on our own. The remainder of this section will be an analysis of the opportunities in some of the
commodity chemicals. Finally, we will prioritize the chemicals to determine where we have the
best chance of success.
6.1Technical Strategy
On any chemical which we consider for LPO applications the following steps should
occur. Each of these steps should be considered a gate where the merits of the project are re-
evaluated based on new learnings and a decision on whether to proceed is made.
1. Preliminary economics: Using educated guesses based on readily available
information, estimate probable impact of various LPO applications using the
procedure outlined above. Determine whether, assuming favorable reactor
kinetics, there is enough value to the customer to proceed with a project.
2. Determination of simplified kinetics: This may require lab work, a literature search,
customer input, or a combination thereof. The idea is to keep the effort
fundamental and focus on gathering good data for subsequent modeling efforts,
not necessarily detailed development/pilot plant efforts.
3. Modeling of data: Using the simplified models above and more complicated models
which are available. DI, EA, and re-optimization will be evaluated to determine
impact on process performance.
4. Detailed economic evaluation: Use model predictions to determine impact on
process, and detailed economics to determine potential profitability of using
oxygen. If oxygen has potential value, determine best application for customer’s
needs.
5. Process demonstration and/or development of best LPO application: Either conduct a
plant test jointly with the customer, or enter into a joint development (JD) agreement
to determine impact of using oxygen in a pilot test.
The above market-directed programs will be complemented by fundamental studies on
DI. The above discussion clearly shows that DI should provide marked improvement over EA if
the theory is valid. Specifically this requires that DI oxygen bubbles are consumed before
coalescing with the primary feed bubbles. Also, we want to understand how DI will work in
bubble columns versus stirred reactors, and to determine whether DI could provide selectivity
benefits. To better understand these issues we intent to conduct fundamental studies at our
37
European facility, and to fund research at the Chemical Reaction Engineering Laboratory at
Washington University in St. Louis.
6.2Market Analysis, Project Evaluation, and Prioritization

Table 8 contains a summary of research on producers of the target chemicals. This was
based on information from the most recent SRI PEP reports and ChemSystem PERP reports. We
eliminated locations not strategic to Praxair and determined potential worldwide volumes for
each chemistry. These are not market share numbers, rather the entire market in regions where
PX has interest.
Table 8: Worldwide potential oxygen sales for target chemistries.

Chemical: PTA (and TPA DMT Cyclohexane Phenol PO (SM co-
for DMT in (Witten) oxidation process)
Eastman
process)
Total WW volume, kmtpa: 10,541 3,530 3,804 4,216 1,084
WW volume in PX markets, 9,231 2,879 3,804 3,595 779
kmtpa:
Estimated product O2/product: 0.15 0.10 0.10 0.08 0.20
WW potential oxygen volume, 4,813 1,074 1,158 1,027 660
m.t./day:
WW potential oxygen volume 4,215 876 1,158 875 474
in PX markets, m.t./day:
Avg. plant size among in PX 280 144 112 138 260
markets (kmtpa):
Typical plant oxygen demand, 128 44 34 34 158
m.t./day:
Each project was then analyzed using project planning tools which ranks various benefits
of the project including the market size. The projects were EA or DI for PTA, DMT,
Cyclohexane oxidation, Phenol, PO/SM. The following aspects were evaluated:
1. Customer acceptance (1 (poor)-5 (good) rating).
2. Fit with Praxair objectives (1-5 rating).
3. Sustainable advantages (1-5 rating).
4. Liability and safety (1-5 rating).
5. Overall market attractiveness (1-10 rating).
6. Competitive position (1-10 rating).
It should be stressed that no effort was made to normalize these rankings to other ARD
projects. These were evaluated only relative to one another, so conclusions can only be drawn on
the relative quality of these projects versus each other, not on other ARD projects. The results
are summarized in Figure 13:
38
Customer Acceptance
10
6
Competitive Position Fit w ith PX objectives
4
Market Attractiveness Sustainable advantages
EA/PTA
EA/DMT
EA/C-Hex
Liability/Safety
EA/Phenol
a) Enrichment with process re-optimization EA/POSM
Customer Acceptance
10
6
Competitive Position Fit w ith PX objectives
4
Market Attractiveness Sustainable advantages
DI/PTA
DI/DMT
Liability/Safety DI/C-Hex
DI/Phenol
b) Direct injection DI/POSM
Figure 13: Radar plots of project planning and marketing metrics.

As can be seen in these plots, most of the projects have similarly good ratings on criteria
1-4. There are slightly more concerns in the safety area for DI over EA because of the use of
pure oxygen, and Phenol and SM/PO have some more safety concerns because of uncertainty
concerning the response of an autocatalytic reactor to higher throughput.
39
The major differences can be seen in the Market Attractiveness versus Competitive
Position which is highlighted in more detail in Figure 14:
Competitive Position
Market Attractiveness Competititve position and market attractiveness
0 5 10 15 20
EA/PTA 8.1 5.7
DI/PTA 7.6 7.8
EA/DMT 7.3 5.7
DI/DMT 6.8 7.8
EA/C-Hex 7.3 5.7

Technology
DI/C-Hex 6.8 7.8
EA/Phenol 6.3 5.3
DI/Phenol 5.5 7.8
EA/POSM 5.9 5.3
DI/POSM 5.5 7.8
Figure 14: Competitive position and market attractiveness for various technology
offerings.
As can be seen, for all chemistries, EA generally has higher market attractiveness
primarily because of the lower technical uncertainty, and lower development manpower required
whereas DI has higher competitive position primarily because we believe that we can defend this
with pending patents. However, DI has overall higher combined ratings, and most work done to
support DI, supports EA. Therefore, we recommend that both EA and DI be pursued for
identified chemistries.
Given this, the decision becomes one of prioritizing chemistries. Our analysis (Figs. 13
and 14) suggests that PTA is the best opportunity. This is because we have a fair amount of data
which indicates that DI, EA, and re-optimization will provide customer benefits. Our analysis
suggests that DMT and cyclohexane oxidation are the next best candidates. We believe DMT has
potential because it is similar to PTA, and we believe cyclohexane oxidation has potential
because of the encouraging laboratory work done by ABB Lummus Global in support of the
LOR development for cyclohexane oxidation. We believe that the potential of these projects
warrants further with or without up front customer interest.
Phenol and PO/SM are another matter. Although we should expect there to be a way to
use process re-optimization to improve these processes with oxygen, kinetics are more complex,
and the principle has not been demonstrated. The likelihood of success is therefore somewhat
40
less than the above chemistries. Therefore, we recommend that these chemistries be put on hold,
and we only study these chemistries with up front customer interest.
7Schedule and Manpower Requirements for Future Work

In order to generate customer interest in the applications which we decide to pursue, we
feel the following tasks need to be carried out. Some will be done in parallel, particularly the
academic collaboration work.
1. Gather all data on PTA and cyclohexane oxidation to date and model data to estimate
potential benefit and identify any additional experiments to refine estimates.
Present to 2Q99 with 50% engineer time.
2. Publish/Patent work on potential benefits of EA, DI, and re-optimization. 2Q99-
3Q99 with 50% engineer time.
3. Conduct collaborative research on DI bubble/Air bubble interaction with Washington
Univ. Ongoing, but expect significant results by 4Q99 with 10% engineer time
overseeing work.
4. Conduct any necessary experiments to complete PTA understanding and develop
kinetics on DMT in laboratory to support EA/DI for DMT. 3Q99-2Q00 with 50%
engineer time, 50% technician time.
5. Conduct any necessary experiments to complete cyclohexane oxidation
understanding. If necessary, 2Q00-3Q00 for set-up. 3Q00-1Q01 for experiments.
8Conclusions
Until now, LPO applications have been treated as individual programs, EA, DI, and
LOR. We believe that we can better serve customer’s needs with less resources by learning the
fundamentals of each chemistry and using modeling and target laboratory experiments to
identify the most appropriate application. To that end, we have developed a technical strategy
for delivering an offering in a specific chemistry in Section 6.1. In Section 6.2, we used the
project planning tools to prioritize among the five major chemistries as follows:
1. PTA
2. DMT
2. Cyclohexane oxidation (tie)
4. Phenol
5. PO/SM
We recommend that the top three chemistries be pursued regardless of up front customer
interest. We recommend that the remaining two chemistries, Phenol and PO/SM be put on hold
until a customer shows interest. In Section 7 we outlined a schedule that with sufficient
manpower, will provide a technical package for the three targeted chemistries (PTA, DMT, and
cyclohexane oxidation) within 18-24 months. The goal of this technical package will be to
generate sufficient interest from an industrial partner to proceed with a joint development
project.
41

Williams - Oxygen Applications

Uploaded by

Copyright:

Available Formats

Williams - Oxygen Applications

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Williams - Oxygen Applications

Uploaded by

Copyright:

Available Formats

Oxygen Applications for Organic Liquid Phase Oxidation

UFL (in air)

LFL (in air)

Figure 2: Ternary flammability diagram depicting operating lines as organic composition

3.1.1Plant productivity limited by gas throughput

Next substituting equation 6 into equation 4, we get the following result:

3.1.2Plant productivity limited by mass transfer in reactor

and the following result for enhancement:

Figure 3: Expected capacity improvements for oxygen used to debottleneck compressor or

3.1.3Plant productivity limited by reaction kinetics in reactor

Table 1: Effect of reaction order on productivity enhancement by EA to 25% oxygen.

Figure 4: Effect of oxygen concentration in feed on productivity enhancement for various

3.2Direct injection of oxygen (DI)

Figure 5: Direct injection of oxygen.

3.2.1 Plant productivity limited by gas throughput

3.2.2Plant productivity limited by mass transfer in reactor

Figure 6: Comparison of EA and DI at equivalent supplemental oxygen levels for a mass

6 No solution for greater O2

Figure 7: Limitations on amount of DI oxygen for a system with k La = 10 min-1.

3.2.3Plant productivity limited by reaction kinetics in reactor

Increase temperature and add oxygen,

Reactor Increase temperature with air.

3.3.2Reducing by-product formation with process re-optimization

Figure 9: Improving process performance with process re-optimization to take full

Substituting (21) into (22) we obtain the rather simple relationship:

Thus, the percentage reduction in by-product formation only depends on three

4Previous and Current Customer Contacts

4.3DMT (Witten Process)

4.4Cyclohexane to K/A Oil

4.5Cumene hydroperoxide for phenol production

4.6Propylene oxide/styrene monomer (PO/SM) co-production

5Economic Evaluation of LPO Applications

5.2Re-optimization for Capacity Increase or By-product reduction.

Oxygen used for selectivity during turndown.

5.3An example: Enrichment or DI for TPA with re-optimization

5.3.1Capacity expansion: Capital improvement versus supplemental oxygen

Supplemental oxygen costs:

Capital analysis (new plant design modification):

Capital analysis (expansion of capacity with additional plant line)

5.3.1.1Capacity increase of a new plant

5.3.1.2Capital expansion of existing plant

5.3.2EA or DI for capacity expansion: Variable cost analysis

3 EA/DI used for by-product reduction

5.4DMT (Witten Process)

5.5Caprolactam via cyclohexane oxidation

Table 5: Cost analysis for DI or EA for caprolactam production.

5.6Phenol (via cumene hydroperoxide)

It is important to recognize that cumene oxidation to CHP is predominantly practiced

5.7Propylene oxide co-production with Styrene Monomer (PO/SM)

6.2Market Analysis, Project Evaluation, and Prioritization

Table 8: Worldwide potential oxygen sales for target chemistries.

Market Attractiveness Sustainable advantages

Market Attractiveness Sustainable advantages

Figure 13: Radar plots of project planning and marketing metrics.

EA/PTA 8.1 5.7

DI/PTA 7.6 7.8

EA/DMT 7.3 5.7

DI/DMT 6.8 7.8

EA/C-Hex 7.3 5.7