You are on page 1of 7

Ultra Deep Desulfurization of Diesel: How an understanding of the

underlying kinetics can reduce investment costs

Kim G. Knudsen and Barry H. Cooper


Haldor Topsøe A/S

Abstract

This paper presents an overview of the critical factors governing the production of ultra low sulfur die-
sel (ULSD) i.e. diesel fuel with less than 50 ppm sulfur. To produce ULSD it is necessary to remove
the most refractive sulfur compounds, which are certain alkyl-substituted dibenzothiophenes. Alkyl-
substituted dibenzothiophenes are desulfurized via one of two routes: by direct extraction of the sulfur
atom, or by hydrogenation of one of the aromatic rings followed by sulfur extraction. Factors affecting
the relative rates of reaction for the two routes are discussed, in particular the inhibiting effect of cer-
tain nitrogen containing components of diesel oils on the hydrogenation route. CoMo catalysts are
generally more active for the direct desulfurization route, whereas NiMoP catalysts show relatively
higher activity for the h ydrogenation route. The consequences for ULSD are demonstrated through a
number of cases studies, which serve to illustrate the effect of catalyst choice on required catalyst vol-
ume, hydrogen consumption and product properties. The case studies are used to discuss the merits
of revamps versus grassroots units.

Introduction

Diesel fuel specifications are being tightened throughout the world as part of efforts to improve air
quality. At the same time, the demand for diesel is increasing necessitating use of lower quality feed-
stocks. The combination of these factors places a heavy burden on the refiner’s hydroprocessing ca-
pabilities. New hydrotreating capacity and revamp of existing facilities are needed to meet the future
diesel specifications. The present emphasis is on the reduction of sulfur, but future requirements may
include improvement of cetane number, reduction in polyaromatic content and reduction in density.

For the production of ultra low sulfur diesel (ULSD), the refiner has to decide whether to revamp an
existing hydrotreater or to build a new, grassroots unit. A revamp is less costly but will often be less
flexible with respect to changes in feedstock and in required product properties. Many factors need to
be considered in choosing the most cost-effective solution, but in all cases it is essential to have a
thorough understanding of the kinetics for removal of the most refractive sulfur compounds.

The kinetics of deep desulfurization is governed by the extent to which desulfurization (HDS) occurs
by direct sulfur extraction, or by hydrogenation of the sulfur -containing molecule. The direct route is
primarily inhibited by hydrogen sulfide, and the hydrogenation route by specific nitrogen-containing
compounds. CoMo and NiMoP catalysts exhibit different preferences for the two routes. Detailed un-
derstanding of the kinetics of deep desulfurization is used to select the most suitable catalyst for a
given service and to evaluate the relative advantages of a revamp versus a grassroots unit with re-
spect to investments, hydrogen costs and product properties.

Whether or not the refiner decides on a revamp or a grassroots unit, careful attention needs to be paid
to the design of the reactor internals. Poor distribution of the reactants over the catalyst can contribute
to channeling through the catalyst bed resulting in inefficient utilization of the catalyst, development of
hot spots, and catalyst deactivation due to coke formation. The problem of poor distribution becomes
more acute as the requirement for desulfurization increases.

Changes in Diesel Specifications and Demand

In recent years, the development and use of “environmentally friendly” fuels has had high priority
throughout the world. The driving force is improvement in air quality especially in metropolitan areas.
Reductions in vehicle emissions require improvements in both engine and exhaust technologies, and
-2-

in fuel quality. The vehicle manufacturers have issued a “World-Wide Fuel Charter” 1, which stipulates
minimum requirements for fuel quality to meet future emission standards.

In the EU, current diesel specifications limit sulfur to a maximum of 350 wppm. By the year 2005, the
sulfur content must be reduced to 50 wppm, and diesel containing a maximum of 10 wppm S must
also be available. By Jan 1, 2009 all on-road diesel in EU will (subject to review) have a specification
of max. 10 wppm sulfur. In Germany 10 wppm sulfur diesel will be mandatory from 2003.

In the United States, the EPA has announced proposals to reduce diesel fuel sulfur levels by mid 2006
to 15 wppm. In California, transit bus fleets must use a 15 wppm sulfur diesel by October, 2002 and it
is expected that this limit will be extended to more fleets shortly afterward. In Japan the government is
considering plans to reduce sulfur content to below 10 wppm by 2008.

The demand for diesel is growing in both Asia and Europe. Although the growth rate for middle distil-
late in Asia Pacific was relatively stagnant from 1996-1998, it rose to 4.6% in 19992 and future d e-
mand is forecast to be equivalent to the pre-1997 rate of approximately 5-8% per year 3 . In the EU, the
increase in demand for mid-distillates has been at about 1.2%/annum over the past 5 years. It is pre-
dicted that diesel demand in the EU will increase by 30% over the next 15 years4 . There is a reduced
demand for home heating oil and fuel oil cutter stocks which can help meet the increase need for die-
sel but these lower quality stocks will require severe hydrotreatment to meet the future specifications.

The Selection of Catalyst for Ultra Deep Desulfurization

To appreciate the basis for the selection of catalyst for ultra deep desulfurization it is important to un-
derstand the nature of the most refractive sulfur compounds in diesel pool streams, and understand
how these sulfur compounds react. Many studies have shown that the most refractive sulfur com-
pounds in diesel streams are alkyl-substituted dibenzothiophenes in which the substituents are in the
4 and 6 positions (see Fig.1). In the feed to the hydrotreater, the proportion of these refractive sulfur
compounds in relation to the total sulfur content varies depending on feed origin and boiling -range but
in products from the hydrotreater containing 100 ppm sulfur or less essentially all the sulfur com-
pounds will be compounds of this type.

1 9

2 8

3 7
4 6
S
R R'
Figure 1 - Most Refractive Sulfur Compounds in Diesel Fuels

The normal reaction route for desulfurization is via extraction of the sulfur atom (see Fig. 2). CoMo
catalysts remove sulfur primarily via this route. When the dibenzothiophene (DBT) contains alkyl
groups in the 4 and 6 positions, access to the catalyst site becomes “sterically-hindered” and the rate
of reaction drops considerably.

There is an alternative reaction route for desulfurization called the hydrogenation route in which one
ring of the DBT is hydrogenated before the sulfur is extracted. This route is much slower for most al-
kyl-substituted DBTs, but is faster for the sterically-hindered DBTs. NiMoP catalysts exhibit a relatively
high activity for removal of sulfur via the hydrogenation route. NiMoP catalysts are generally less a c-
tive for removal of the bulk of the sulfur (via direct extraction) but are better than CoMo catalysts for
removing the sterically-hindered DBTs via the hydrogenation route.

Unfortunately, inhibitors present in the oil poison the hydrogenation route. Recent work has identified
these inhibitors as certain basic nitrogen compounds4 -7 . Due to different reactivities towards removal of
the inhibitors and the refractive sulfur molecules, a NiMoP catalyst behaves differently than a CoMo
catalyst. The NiMoP catalyst does not become very active until almost all of the inhibitors have been
removed, at which point it becomes more active than a CoMo catalyst. NiMoP catalysts are preferable
to CoMo catalysts at high pressure and when the content of the inhibiting basic nitrogen compounds is
low. Furthermore, the hydrogenation route is equilibrium limited, which results in a lower reactivity (and
-3-

low apparent activation energy), so that at low pressures and high temperatures, a CoMo catalyst can
therefore be a better choice than a NiMoP catalyst.

Hydrogenation reactions such as aromatic saturation also proceed via the hydrogenation route and

Figure 2 - Reaction Pathways for HDS of DBT


are inhibited by the same nitrogen compounds. If NiMoP catalysts can be applied, more aromatic satu-
ration occurs (compared to using CoMo catalysts) resulting in a greater improvement in cetane num-
ber and density, and a lower product PAH content. There will also be higher hydrogen consumption,
which may have consequences for hydrogen availability and recycle rate, and for compression costs.

Case Studies for the Production of Ultra Low Sulfur Diesel

As mentioned earlier, the refiner needs to decide whether the desired reduction in diesel sulfur content
can be achieved by revamping an existing unit (e.g. adding more hydrotreating reactor capacity, re-
vamping reactor internals, revamping the treat gas system etc.) or whether it is necessary to build a
grassroots unit. The decision will depend on many factors including the quality of feedstocks, the d e-
sign criteria for the existing unit, and whether or not improvements in other diesel properties are re-
quired. The optimum solution with respect to overall investment and operating costs will depend on the
correct choice of catalyst based on the considerations outlined above. In order to illustrate how the r e-
quired extra reactor volume depends on operating conditions and feedstock properties, we present b e-
low three case studies. The case studies also serve to illustrate the effects on other key parameters
for diesel fuel quality and the extent to which there is a need for more hydrogen.

The three cases are:

• Case 1:Hydrotreater producing 500 wppm sulfur operating at 32 bar hydrogen pressure on a
straight-run feedstock
• Case 2:Hydrotreater producing 500 wppm sulfur operating at 32 bar hydrogen pressure on a
blended feedstock containing 25% LCO and 75% straight-run.
• Case 3:Hydrotreater producing 500 wppm sulfur operating at 54 bar hydrogen pressure on a
blended feedstock containing 25% LCO and 75% straight-run.

For each case, calculations have been made to estimate the incremental reactor volume needed to
reach a 50 wppm and a 10 wppm sulfur specification, and key product properties have been esti-
mated. The basis for the calculations and estimates is extensive testing performed in our laboratories
at ultra deep desulfurization conditions. Comparisons are made between two high activity catalysts for
ULSD: TK-574 (CoMo) and TK-573 (NiMoP).

Case 1: Straight-run Feed at 32 bar


-4-

This case is based on straight-run gas oil containing 1.2 wt% sulfur operating at a hydrogen partial
pressure at inlet of 32 bar. Feedstock properties are given in Table 1. A comparison is made between
TK-574 and TK -573. Catalyst cycle length is fixed at 2½ years.

Table 2 shows the reactor volumes required to reach 50 wppm and 10 wppm sulfur relative to the re-
actor volume required to reach 50 wppm on the CoMo catalyst, and the relative chemical hydrogen
consumption. The table also shows changes in other key product properties.

Density, kg/m3 850


Sulfur , wt% 1.2
ASTM D86, °C
10v% / 50v% 243/287
90v% / 95v% 342/360
Total Aromatics Content, wt% 25.5
PAH content, wt% 9.0
Calculated Cetane Index, ASTM D4737 52.8
Table 1: Feed Properties for Straight-run High Sulfur Gas Oil

For Case 1, reducing the product sulfur content from 50 wppm to 10 wppm whilst maintaining cycle
length requires almost a doubling of reactor volume for the CoMo catalyst, and 75% increase in reac-
tor volume for the NiMoP catalyst. The NiMoP catalyst can remove the inhibitors to the hydrogenation
route at the conditions needed to reach the two product sulfur levels, and as a result performs better
than the CoMo catalyst. Similarly, the NiMoP catalyst exhibits more hydrogenation and this result in
improve product properties with respect to PAH content, density and Cetane Index, and a correspond-
ing higher hydrogen consumption.

Product S = Product S = Product S = Product S =


50 wppm 10 wppm 50 wppm 10 wppm
TK-574 TK-574 TK-573 TK-573
Relative Reactor Volume 1.0 1.9 0.85 1.5
Relative Hydrogen
Consumption (SOR) 1.00 0.96 1.27 1.33
Product Properties at Start-of-Run
PAH Content, wt% 3.2 3.8 2.4 2.9
Density, kg/m3 841 842 839 838
CCI, ASTM D4737 55.9 55.7 56.9 57.2
Table 2: Relative Reactor Volume and Hydrogen Consumption, and Product Properties (Case 1)

Case 2: Blended Feed at 32 bar

This case is based on a blend of 25% light cycle oil and 75% straight-run gas oil containing 1.5 wt%
sulfur operating at a hydrogen partial pressure at inlet of 32 bar. Feedstock properties are given in Ta-
ble 3. The catalyst is TK-574, a high activity CoMo type catalyst. Catalyst cycle length is fixed at 2½
years.
3
Density, kg/m 870
Sulfur , wt% 1.5
ASTM D86, °C
10v% / 50v% 243 / 287
90v% / 95v% 342 / 360
Total Aromatics Content, wt% 30.0
PAH content, wt% 14.0
Calculated Cetane Index, ASTM D4737 45.1
Table 3: Feed Properties for High Sulfur Blended Gas Oil
-5-

Table 4 shows the reactor volumes required to reach 50 wppm and 10 wppm sulfur relative to the re-
actor volume required to reach 500 wppm, and the relative chemical hydrogen consumption. The table
also shows changes in other key product properties.

Product S = Product S = Product S =


500 wppm 50 wppm 10 wppm
Relative Reactor Volume 1.0 1.9 3.4
Relative Hydrogen
Consumption (SOR) 1.0 0.94 0.91
Product Properties
PAH Content, wt% 4.2 (SOR) 5.6 (SOR) 6.3 (SOR)
9.1(EOR) 8.5 (EOR) 8.3(EOR)
3
Density, kg/m 855 (SOR) 856 (SOR) 856 (SOR)
CCI, ASTM D4737 49.2 (SOR) 48.9 (SOR) 48.7 (SOR)
Table 4: Relative Reactor Volume and Hydrogen Consumption, and Product Properties (Case 2)

For Case 2, reducing the product sulfur content from 500 wppm to 50 wppm whilst maintaining cycle
length requires almost a doubling of reactor volume, and further reduction from 50 wppm to 10 wppm
requires the addition of approximately 80% more reactor volume. The hydrogen consumption is similar
for all product sulfur levels.

The large extra volume required to meet the low product sulfur levels will result in an increase in pres-
sure drop and may necessitate a revamp of the recycle gas system. It is also possible that the low
LHSV required at the lowest product sulfur levels will result in significant hydrocracking (to naphtha
and gas) thereby reducing diesel yield and necessitating extra investment in product fractionator over-
head equipment. If there is significant hydrocracking, there will be an increase in hydrogen consum p-
tion compared to Table 4.

There is a deterioration in the other key product properties as product sulfur level is lowered. Further-
more, the quality of the product with respect to the three properties shown in Table 4 deteriorates
throughout the run as temperature is raised to counteract catalyst deactivation. This is illustrated in
Table 4 by comparing the PAH content at start-of-run (SOR) and end-of-run (EOR). The PAH content
in all cases is very high at EOR, because the equilibrium between mono-ring aromatics and poly-ring
aromatics becomes less favorable at the temperatures used at EOR. The PAH concentration affects
product density and cetane index (and cetane number), and at EOR the density will be higher in all
three cases and the cetane index (and cetane number) will be lower.

Case 3: Blended Feed at 54 bar

This case is based on the same blend of 25% light cycle oil and 75% straight-run gas oil used in Case
2, but the unit operates at a hydrogen partial pressure at inlet of 54 bar. The feedstock properties are
the same as those given in Table 3. The CoMo catalyst, TK -574, is chosen for the 500 wppm product
sulfur case, and the NiMoP catalyst, TK-573, for all the low sulfur cases. Catalyst cycle length is fixed
at 2½ years.

Table 5 shows the reactor volumes and chemical hydrogen consumption for the three product sulfur
levels. The reactor volumes are relative to the reactor volume required for Case 2, 500 wppm
sulfur. The table also shows changes in other key product properties.

For Case 3, reducing the product sulfur content from 500 wppm to 50 wppm whilst maintaining cycle
length requires a 70% increase in reactor volume, and further reduction from 50 wppm to 10 wppm re-
quires the addition of approximately 45% more reactor volume. The percentage increase is less than
for Case 2, and the absolute increase in reactor vo lume is only 40% that for Case 2.

The effect of pressure on the performance of TK-574 is seen by comparing the results for 500 wppm
product sulfur in Tables 4 and 5. Increasing pressure by almost 70% results in a reduction in required
reactor volume by 35%. At the higher pressure, even the CoMo catalyst exhibits a higher hydrogena-
tion activity as reflected in higher hydrogen consumption, and improved product properties.
-6-

For the remaining product sulfur levels, the difference in required reactor volumes for Case 2 and
Case 3 may be explained by the higher reactivity of the NiMoP catalyst in Case 3. At the higher pres-
sure in Case 3, the NiMoP catalyst can remove the inhibiting species enabling desulfurization of the
sterically-hindered DBTs via the (quicker) hydrogenation route.

Product S = Product S = Product S =


500 wppm 50 wppm 10 wppm
TK-574 TK-573 TK-573
Relative Reactor Volume 0.65 1.1 1.6
Relative Hydrogen
Consumption (SOR) 1.15 1.5 1.8
Product Properties
PAH Content, wt% 1.4 (SOR) 1.8 (SOR) 2.1 (SOR)
6.3 (EOR) 5.9 (EOR) 5.3 (EOR)
3
Density, kg/m 853 (SOR) 848 (SOR) 843 (SOR)
CCI, ASTM D4737 49.7 (SOR) 51.6 (SOR) 52.5 (SOR)
Table 5: Relative Reactor Volume and Product Properties (Case 3)

The higher hydrogenation activity of the NiMoP cata lyst compared to the CoMo catalyst is reflected in
the key product properties for the three product sulfur levels. Although the diesel quality with respect
to these properties will deteriorate at EOR (for the same reason as given above), the extent of deterio-
ration will be less because the aromatics equilibria are more favorable at the higher pressure in Case
3. This has been illustrated in Table 6 for the PAH content at both SOR and EOR.

The higher level of aromatic saturation obtained in Case 3 may be desirable, depending on future di e-
sel specification requirements. It does, however, give higher operating costs due to higher chemical
hydrogen consumption, especially for the low sulfur product levels. Higher hydrogen consumption re-
sults in a lower partial pressure of hydrogen at the reactor outlet, adversely affecting the rate of cata-
lyst deactivation and the key product properties. It is therefore necessary to increase the recycle gas
to oil ratio in order to maintain outlet hydrogen partial pressure, and this will very often require revamp
of the recycle gas system.

Yield loss due to cracking is less at the higher space velocities used in Case 3 than in Case 2, and is
furthermore countered by volume swell due to a higher hydrogen addition.

Revamp vs. Grassroots Unit

Case 2 and 3 can be used to illustrate the considerations involved when deciding between revamping
a unit or building a grassroots unit. Take the case of an European refiner who has a unit producing a
diesel with 500 wppm sulfur at 32 bar hydrogen partial pressure using the feed shown in Table 3. The
refiner is faced with having to meet 50 wppm sulfur in 2005, and 10 ppm in 2009. Should the refiner
choose to revamp the existing unit by adding extra reactor volume i.e. Case 2, or build a new g rass-
roots unit operating at higher pressure, i.e. Case 3?

If the only consideration is to reduce sulfur, the preferred solution would probably be to add reactor
volume as in Case 2, but this assumes that the existing gas recycle system can manage the increased
pressure drop and that the diesel yield loss due to cracking is acceptable. It may be necessary to re-
vamp of the recycle gas system and the overhead fractionator. This solution if possible will be much
cheaper than building a grassroots unit, but the incremental cost of going from 50 ppm to 10 ppm may
be higher.

If PAH content, density or cetane number are also important, then the refiner may choose to build a
new high pressure unit to meet the 50 wppm product sulfur level i.e. the Case 3 solution. This involves
more costly investments and the intelligent use of existing equipment can help reduce overall costs8. If
the product sulfur requirement is lowered at a subsequent date, the incremental investment will be
lower.

In our experience refiners may opt for either solution, depending on their specific needs and possibili-
ties9 . Topsøe has over the past three years designed nineteen ULSD units with specifications of less
than 50 wppm sulfur. Twelve of these units are designed to produce less than 10 wp pm sulfur diesel,
-7-

of which six are revamps.

Conclusion
Refiners will need to invest heavily in hydrotreating units as legislation for ultra low sulfur diesel fuel is
adopted. For optimum design in which the cost of new units is kept to a minimum and the best use is
made of existing equipment, it is necessary to have a deep understanding of the factors governing the
removal of the most refractive compounds in diesel fuels. Topsøe has identified the compounds pres-
ent in diesel feedstocks that inhibit the removal of the most refractive sulfur compounds, and this
knowledge can be used to define the correct choice of catalyst for a given service. The choice of cata-
lyst also affects other key diesel properties such as the density, PAH content and Cetane Index, as
well as hydrogen consumption and the required treat gas-to-oil ratio.

References

1. http://www.bp.com/worldenergy/oil/index.htm
2. Yamaguchi, N.D., “Update on: Asian Product Demand and Quality”, Hart’s World Fuel Confer-
ence, Washington, DC, September 1999.
3. Purvin and Gertz, Hart’s 4t h Annual World Fuel Conference, Brussels, May 19-21,1999
4. Whitehurst, D.D., Knudsen, K.G., Nielsen I.V., Wiwel, P., Zeuthen, P. Preprints, ACS Div. Petr.
Chem. 45, No.4, p.692.
5. Knudsen, K.G., Whitehurst, D.D., Zeuthen, P. “A detailed understanding of the inhibition effect of
organic nitrogen compounds for ultra deep HDS and the consequences for the choice of catalyst”,
Presented at the AIChE Spring National Meeting, Atalanta, GA, March 5 -9, 2000
6. Whitehurst, D.D., Knudsen, K.G., Wiwel, P., Zeuthen, P. , Preprints, ACS Div. Petr. Chem., 45,
No.2, p.367
7. Cooper, B.H., Knudsen, K.G., "The importance of Good Liquid Distribution and Proper Selection of
Catalyst for Ultra deep Diesel HDS", JPI Petroleum refining Conference, October 19-20, 2000,
Tokyo, Japan
8. Bingham, F.E., Christensen, P. “Revamping HDS Units to Meet High Quality Diesel Specifica-
tions”, Asian refinery Technology Conference, March 8-10,2000, Kuala Lumpur, Malaysia
9. De la Fuente, E., Low, G., "Cost-effectively improve hydrotreater designs", Hydrocabon Process-
ing, November, 2001, p 43-46

You might also like