REFINING

Fundamentals of ULSD
production
Determining the best course for producing ULSD depends on four factors –
chemistry, composition, reaction conditions and catalysis. Guidelines are offered
for effective strategies and for deciding how to optimise future ULSD operations
Salvatore Torrisi, Jr, and P Michael Gunter
Criterion Catalysts & Technologies LP

C
anadian and US refiners will industry must overcome to reliably and 500ppm to 50ppm. Producing ULSD
begin producing ultra-low sul- cost-effectively meet the future on-road will be very similar to producing a
phur diesel (ULSD) fuel (defined and off-road ULSD demand. The most chemical grade product, but with a
as diesel with less than 15ppmw sul- critical issues are: much higher degree of upstream vari-
phur) for on-road use in June 2006. Achieving the design basis for the ability. In this new environment, pro-
They will be the first major markets, ULSD operation (anticipating and duction errors will be costlier than ever.
outside a few European countries, to avoiding operational problems after Operating effectively in this new
convert the on-road diesel fuel supply to ULSD production begins) environment will require a thorough
this “near zero” sulphur specification. Going beyond the ULSD design basis to understanding of the fundamentals of
The general European market is also pro- create added economic benefits ULSD operation. The reaction mecha-
gressing to a near-zero diesel fuel. The Integrating off-road ULSD production nisms involved in extracting the last
sulphur level in much of Europe’s on- into the supply capability remaining sulphur molecules from the
road diesel fuel supply has already been Increasing ULSD supply to meet grow- various diesel boiling range streams are
reduced to 50ppmw, and limited mar- ing diesel demand. different from those required to achieve
kets in this region are receiving the The industry has learned a great deal a 500ppm threshold or even a 50 ppm
10ppmw sulphur diesel that has been from European refiners’ recent experi- level. Understanding these differences
proposed for both on-road and off-road ences. The increased desulphurisation will make it possible to identify more
diesel in 2008/9. Many other countries requirement for the 2005 European on- practical solutions to achieve the ULSD
are planning to implement major reduc- road diesel fuel specification (99.5%) is specification and help generate margin-
tions in the diesel pool sulphur level causing a major change in the way dis- enhancing ideas.
over the next few years. tillate hydrotreating operations are
As with past fuel specification managed. This stage of the diesel fuel ULSD chemistry
changes, such as the 500ppm diesel sul- evolution is giving refiners and technol- To produce ULSD cost-effectively, it is
phur change in the mid-1990s, refiners ogy providers insights into the issues important to understand the nature of
are finding resourceful solutions to com- and best options to make ULSD. It is also the sulphur-containing compounds
ply with these new diesel fuel specifica- driving developments in unit/equip- found in the diesel feedstock pool and
tions. Many European refiners have ment designs, control systems, and the the chemistry involved in removing the
successfully reduced the sulphur content catalysis area that will help create more sulphur. The distribution of sulphur
in the on-road diesel pool from 500ppm robust ULSD designs and provide addi- compounds in a diesel feedstock
to 50ppm, well in advance of the 2005 tional leverage to deal with the previ- depends on many factors such as the
deadline. ously noted critical issues (Figure 1). feedstock’s boiling range and the crude
The majority of Canadian and US Even with this help, the next stage oil type from which it was derived.
refiners are on-track to meet the new (less than 15ppmw sulphur or 99.9% Since their relative reaction rates can
15ppm specification. Even so, there are a desulphurisation) will be even more be vastly different, identifying the spe-
number of significant hurdles that the challenging than the transition from cific sulphur-containing compounds

Figure 1 Increasing desulphurisation requirements speeding catalyst

development Figure 2 Typical sulphur compounds found in distillate feeds

29
P T Q SUMMER 2004
w w w. e p t q . c o m
 REFINING

that exist within the diesel feedstock is

the first critical step in defining the
appropriate operating severity to
achieve ULSD specifications.
Figure 2 illustrates the common types
of sulphur-containing compounds
found in a typical full-range diesel as a
function of boiling point range. As
expected, the structure of the com-
pounds is more complex at the upper
end of the boiling range where the sul-
phur is often bound in substituted steri-
cally-hindered polyaromatic rings,
making it extremely difficult to break
the carbon-sulphur bonds. Sulphur con-
tained in the lower boiling-point com-
pounds (such as sulphides, thiophenes,
and dibenzothiophenes) is easily
removed using traditional hydrotreating
systems and catalysts.
On the other hand, removing sulphur
from the substituted polyaromatic (SPA)
compounds boiling above 645°F is
much more difficult and requires a dif-
ferent approach. This difference in sul-
phur removal difficulty is reflected by Figure 3 Sulphur distribution dependent on crude source and upstream processing
the relative reaction rate scale shown as
the y-axis of the chart in Figure 2. The from an Arabian crude. Despite having delineate the sulphur species that fall
reactivity difference manifests itself in nearly the same endpoint, Feed B has a within the same nominal boiling range
the types of sulphur species that must be greater percentage of SPA sulphur com- and quantify the amount of each species
removed as the product sulphur specifi- pounds (48% versus 37% for Feed A) and across the feedstock’s full range. These
cation is lowered (chart’s upper bar). For so is more difficult to desulphurise than data provide a much more accurate pic-
example, for a product sulphur specifi- Feed A. ture of the amount of sulphur tied up in
cation of less than 500ppm, low conver- It is also important to note that the sul- the SPA compounds and thus a more
sions of SPA compounds are observed. phur species that are found in the upper realistic idea of the severity required to
However, to achieve a product sul- boiling range of diesel feedstocks are not produce ULSD.
phur specification of less than 50 ppm just the SPA-type compounds. This cut of In summary, the types and concen-
with a full range diesel feedstock, a the feedstock can also contain the less dif- trations of sulphur-containing com-
much greater percentage of sulphur ficult sulphur species (e.g., disulphides pounds that are found in a diesel
must also be removed from the group of and benzothiophenes). Newer analytical feedstock can vary significantly, even
difficult SPA compounds that includes techniques such as GC x GC-SCD can for feeds with identical boiling ranges,
molecules such as 4,6-dimethyldiben-
zothiophene (DMDBT).
Most refiners process a wide variety of
crudes and purchased derivatives, and
routinely adjust the operations of major
processing units, like the crude unit and
FCC unit, according to the changing
feeds and overall product needs. Addi-
tionally, the types and volumes of
streams feeding into the diesel pool may
change due to seasonal swings in prod-
uct demand and temporary upstream
unit constraints. These changes in
upstream operations will cause the types
and concentrations of sulphur com-
pounds in the diesel feedstock to vary
[Torrisi, S; Centinel catalysts make it easier to
achieve ultra low sulfur diesel; AIChE Spring
National Meeting, New Orleans, March
2002.].
Feeds A and B in Figure 3 are both
blends of straight-run (SR) and cracked
distillate feedstocks, but were derived
from different crudes and have different
initial boiling points. Feed A is a 70/30
blend of SR plus visbreaker/light cycle
oil (LCO) derived from a Nigerian crude.
Feed B is an 80/20 SR/LCO blend derived Figure 4 Two primary reaction pathways for removal of SPA sulphur species

30
P T Q SUMMER 2004
 REFINING

and are influenced by many variables.

The difficulty of achieving the product
sulphur specification depends on the
molecular structure of the sulphur-con-
taining compounds that must be treated.
Therefore, having an accurate detailed
tally of all the sulphur species is necessary
to correctly set the operating severity.

Desulphurisation mechanisms
Desulphurisation reactions can follow a
number of paths (Figure 4), but two
routes are generally favoured under typ-
ical hydrotreating conditions. The most
common route for removing sulphur is
the direct extraction, or hydrogenolysis,
reaction (carbon-sulphur bond break-
age). This mechanism is predominant
when the product sulphur requirement
is above 350ppm. At this level, most of
the sulphur in the more reactive com-
pounds (up to and including non-sub-
stituted dibenzothiophene) has been
removed.
Figure 5 Distillate products having less than 10ppmw sulphur show a concentration of
Typically, conventional cobalt-
SPA sulphurs
molybdenum (CoMo) catalysts and low-
to-moderate reactor pressure are applied tions and hence the optimum condi- Because of the catalyst’s structure,
to favour this reaction mechanism. tions for achieving the ULSD specifica- these nitrogen compounds are adsorbed
Removing sulphur from the most dif- tion [Remans, T J, et al; Catalytic solutions more strongly to the catalyst than the
ficult compounds, such as 4,6- for sustainable ULSD production: Science and sulphur-containing SPAs. Therefore,
diethyldibenzothiophene, generally application of Centinel technology catalysts; they can block the sulphur-containing
follows the hydrogenation route shown European Catalyst and Technology Confer- SPAs from accessing the catalyst’s active
in Figure 4. It is easier to break the steri- ence, Prague, February 2003]. sites and thereby reduce the rate at
cally-hindered carbon-sulphur bond if Diesel feedstocks also contain a vari- which sulphur is extracted from the very
one of the aromatic rings is first saturat- ety of nitrogen compounds that com- compounds (SPAs) that are the limiting
ed. Saturating the ring changes the pete with the sulphur compounds for factor for ULSD production. To over-
molecule’s spatial configuration and the catalyst’s active sites. Additionally, come this limitation, the ULSD design
makes the sulphur atom more accessible aromatic species that contain no sul- must enhance the hydrogenation reac-
to react with the catalyst’s active sites. phur or nitrogen can also interfere with tion to accelerate the nitrogen removal
After the aromatic ring is hydrogenated, the desulphurisation process; so these process.
the sulphur atom can be removed via factors must also be considered when Removing nitrogen species, either by
hydrogenolysis. Nickel-molybdenum weighing options for producing ULSD. feed selection or by enhancing the reac-
(NiMo) catalysts and higher pressure A high concentration of nitrogen tion environment, can significantly
help promote the hydrogenation reac- compounds in the diesel feedstock, increase desulphurisation performance.
tion mechanism. which often occurs when treating signif- The impact that sulphur-free polyaro-
Figure 5 shows the GC analysis for a icant amounts of coker distillate or LCO, matics (SFPAs) have on desulphurisation
diesel product that was hydrotreated to can severely hamper the desulphurisa- reactions depends on the type of sul-
reduce sulphur below 15ppmw. Not sur- tion reactions. Mostly, these nitrogen phur-removal mechanism. Studies show
prisingly, the residual sulphur is con- compounds are SPA type compounds that the amount of SFPAs in the feed has
tained in SPA compounds, primarily that react with the catalyst’s active sites no significant impact on the direct
multi-substituted dibenzothiophenes. In via the hydrogenation/hydrogenolysis extraction (hydrogenolysis) reaction
fact, for feeds with “long tails” (above reaction path. route. However, because the SFPAs
680ºF by D-2887), the majority of sul-
phur species are the higher molecular
analogues of 4,6-DMDBT. These data
suggest that the most cost-effective
approach for extracting the last remain-
ing sulphur will be an environment that
promotes the hydrogenation reaction
mechanism.

Constraining factors
The nature of the sulphur-containing
compounds and the mechanisms by
which the sulphur can be extracted are a
crucial part of ULSD chemistry, but not
the full story. Other reactions occurring
in the diesel hydrotreater have a major Figure 6 Nitrogen is a significant inhibitor of key SPA sulphur removal reactions. Both
impact on the sulphur removal reac- feed nitrogen content and HDN rate determine total inhibition effect

31
P T Q SUMMER 2004
 REFINING

strongly adsorb onto the catalyst’s

hydrogenation sites, they reduce avail-
ability of these sites and therefore
reduce the rate that sulphur can be
removed via the hydrogenation route –
the critical reaction mechanism for
achieving the ULSD specification.
Hence, it is important to create an envir-
onment that favours the saturation of
these compounds to reduce the compe-
tition for the catalyst’s active sites.
Staying within the thermodynamic
window that promotes polyaromatic
saturation is a critical factor in designing
the ULSD unit that will process feed-
containing SFPAs. Decreasing liquid
hourly space velocity and increasing
unit pressure are two ways to enhance Figure 8 Operating temperature window maximised at higher hydrogen partial pressure
the operating window (Figure 6).
ty value (highest number) of any feed which entraps higher boiling point
ULSD feedstock composition component, partially due to its extreme- hydrocarbons containing high concen-
The components forming the diesel feed ly high concentration of sulphur-con- trations of SPA sulphur and nitrogen
blend (SR, coker, and LCO) will have dif- taining SPAs. Additionally, nitrogen compounds. Thus, careful control of the
ferent reactivity values (defined in terms levels ranging from 300ppmw to well feed cut point is an essential part of
of the relative catalyst volume required over 1000ppmw make this component improving the relative reactivities of
to achieve the ULSD specification) for even more difficult to treat. As a result, these types of component.
the reasons discussed in the preceding LCOs are typically two to four times Having determined the relative reac-
chemistry section. The components’ harder to desulphurise than the average tivities of each potential diesel feed
reactivities can differ by as much as a SR feed. Refineries that have an FCC pre- component, it is possible to establish a
factor of three (Figure 7). Therefore, treater can improve the LCO’s relative processing priority for the components
evaluating the reactivity of each compo- reactivity by operating the pretreater for to meet the ULSD production require-
nent is a critical step in defining the maximum polyaromatic saturation. This ments. Then, the optimum conditions
right ULSD strategy [Torrisi S, Krenzke D, approach not only reduces the volume for processing the selected blend can be
Distillate catalyst technology; NPRA Clean of LCO produced in the FCC unit, but decided. Alternatively, for a defined set
Fuels Challenge, Houston, August 2003]. also greatly reduces the amount of sul- of process conditions, the optimum
Historically, SR feeds have been con- phur and nitrogen in the LCO. blend can be chosen based on the rela-
sidered the easiest diesel feed compon- Thermally processed distillates such tive reactivities.
ent to desulphurise. However, the crude as visbreaker, thermal cracker and coker
source and cut-point of the SR feed com- distillates have been widely viewed to Conditions and reactivity
ponent can significantly change its reac- have similar reactivities as an SR compo- The key to cost-effective, robust and reli-
tivity. For example, an SR feed that is nent. In truth, these distillate feed able ULSD operation is properly balanc-
derived from a synthetic crude typically streams can be as difficult as LCO to ing the feed’s reactivity and the
is more difficult to desulphurise than an desulphurise since they usually contain hydrotreater’s operating severity. Oper-
SR component from a conventional large amounts of nitrogen. ating severity is defined by the operat-
crude. This is because the SR component The difficulty in removing sulphur ing conditions, or variables, that affect
from the synthetic crude usually has from these types of feed component is the desulphurisation kinetics. These typ-
much higher concentrations of SPA sul- often exacerbated by poor product sepa- ically include hydrogen and H2S partial
phur and nitrogen species. ration in the thermal conversion unit’s pressures, catalyst volume, catalyst type
LCO normally has the worst reactivi- product fractionation equipment, and temperature.
Hydrogen partial pressure is one of
the most important operating parame-
ters for ULSD production. First, increas-
ing hydrogen partial pressure accelerates
the hydrogenation-limited reaction rate
of the SPA sulphur species that are espe-
cially present in high boiling point
and/or cracked feeds. Second, higher
hydrogen partial pressure enables the
hydrotreater to go to higher tempera-
tures before reaching the thermodynam-
ic equilibrium limits associated with
polyaromatic saturation reactions. Last-
ly, raising hydrogen partial pressure can
reduce the catalyst’s deactivation rate
and thereby increase the hydrotreater’s
run length.
The combination of these factors
Figure 7 Desulphurisation difficulty affected by nitrogen levels, boiling point and expands the overall operating tempera-
amount of cracked feedstocks ture window for the ULSD unit, which

33
P T Q SUMMER 2004
 REFINING

produces a more economic operation 50ppmw sulphur cap, catalyst selection species) and feeds with a large amount
(Figure 8). Reactor total operating pres- must now consider the need to extract of a cracked component like coker distil-
sure has the most influence on hydro- sulphur from the sterically-hindered SPA late (large amount of nitrogen and poly-
gen partial pressure, but treat gas species. aromatics and the potential for a long
hydrogen purity, treat gas rate, and The choice between CoMo and NiMo tail containing very difficult SPA sul-
hydrogen consumption are also impor- catalysts for ULSD production has phur and nitrogen species).
tant operating variables that affect the sparked considerable debate. In some In these cases, the scale could be
hydrogen partial pressure throughout cases, the choice is fairly straightfor- tipped towards a CoMo catalyst or a
the reactor. ward. There have been efforts to simpli- stacked load in the event that the NiMo
Even with improved hydrogen partial fy the selection decision (such as catalyst pushes the hydrogen consump-
pressures, many units will still need creating a single graphical representa- tion above the hydrogen supply capabil-
increased catalyst volume to achieve reli- tion that depicts “catalyst crossover” ity. Then, it may be necessary to reduce
able ULSD operation. On average, con- points as a function of hydrogen partial feed rate to compensate for the CoMo
verting a low-sulphur diesel hydrotreater pressure and product sulphur levels), catalyst’s lower hydrogenation activity
to produce ULSD takes more than dou- but because many other factors influ- (or consider investments to increase cata-
ble the existing catalyst volume. Apply- ence the catalyst’s performance, thor- lyst volume or increase hydrogen sup-
ing the right catalyst, or catalyst system, ough evaluation of the chemistry, ply). Finally, other product properties
will minimise the reactor size require- composition and reactor conditions is such as cetane, densitt and aromatics
ment and capital investment. the only sure way to make the proper content, may be important factors in
ULSD operations can be grouped into catalyst choice. choosing between CoMo and NiMo cata-
three reactor pressure severities: low, The easiest decisions on catalyst lysts for any operating pressure.
medium and high (Table 1). Low-pres- choice fall into the boxes where there is
sure units are typically best suited to an obvious match in the previously Catalyst advances
produce ULSD from feeds that have low mentioned chemistry, composition and ULSD catalyst performance has
reactivity values, such as SR feeds, using conditions. For example, a CoMo cata- improved dramatically in recent years.
CoMo catalysts to promote the direct lyst would be the clear choice in the case Today’s catalysts are over 50% more
extraction desulphurisation reaction. where the operating pressure is low, and active than the top catalysts of five to
Medium-pressure units can handle some the feed is an SR distillate with very lit- six years ago. The rate of development
cracked feeds with higher reactivity val- tle nitrogen. On the other hand, if the has accelerated due to industry need,
ues and can operate with CoMo or NiMo feed has a high percentage of cracked better analytical techniques that have
catalysts, depending on the desulphuri- feed that contains a lot of nitrogen and improved understanding of both the
sation and denitrogenation require- the operating pressure is high, a NiMo organic and inorganic chemistry, and
ments and the amount of hydrogen catalyst would be preferred. the discovery of improved catalyst manu-
available. Not surprisingly, the intermediate facturing technologies.
High-pressure units are effective for operating pressures are where the cata- Until recently, increasing the amount
dealing with feeds that have the highest lyst selection becomes more complex. of promoter metals on the catalyst, opti-
reactivity values, such as cracked feed- For these cases, achieving the proper mising the catalyst’s pore structure and
stocks that contain high concentrations balance in the chemistry, composition improving the catalyst’s surface area
of SPA sulphur and nitrogen species. and conditions might lead to selecting a were the primary methods for improv-
The NiMo catalysts are preferred in this CoMo or NiMo catalyst, and the final ing catalyst performance. New catalyst
case to facilitate the removal of SPA sul- decision may be influenced more by manufacturing technology develop-
phur compounds via the indirect how much future flexibility is desired. It ments in the latter 1990s produced a
hydrogenation reaction mechanism is also possible that circumstances might breakthrough in metals’ dispersion on
and to more effectively remove nitro- dictate using a CoNiMo catalyst or a the catalyst and in sulphiding the metal
gen species that can inhibit the desul- stacked load of NiMo/CoMo catalysts to sites. This discovery produced metal
phurisation reaction. achieve the optimal balance of activity complexes that are now commonly
and selectivity for ULSD production. referred to as Type II sites (Figure 9).
Catalyst selection Selecting the right catalyst for inter- The main benefit of this innovation
Traditionally, CoMo catalysts have been mediate pressure ULSD units must was to create significantly more active
the dominant choice for desulphurising address all the factors previously dis- sites on the catalyst carrier and thereby
diesel feedstocks. This is logical, given cussed. A NiMo catalyst will generally be raise the reaction rate by making more
that current diesel sulphur specifica- the better choice when the indirect, sites available for the reactants [Remans,
tions in most countries have not hydrogenation reaction mechanism will T J, et al; Catalytic solutions for sustainable
required any significant sulphur be the rate-limiting step or when there is ULSD production: Science and application of
removal from SPA type compounds. a significant constraining factor. Exam- Centinel technology catalysts; European Cat-
However, with the implementation of ples include SR feeds with a high boiling alyst and Technology Conference, Prague,
ULSD specifications, and even the point (large amount of SPA sulphur February 2003].
The result is much greater activity,
Increases in operating severity requirements such as with the proprietary Centinel
Gold catalysts for ULSD, FCC pretreater,
Severity Low Medium High hydrocracker-pretreater and lube appli-
cations [Remans T J et al, Centinel Gold
Pressure, psig 400 – 800 500–800 700–1300
Technology for hydroprocessing applications:
H2 pressure, psia 200–450 400–650 650–950
The new catalyst line of Criterion Catalysts &
LHSV, hr–1 0.5–1.5 0.5–1.5 0.5–1.0
TGR, SCF/B <1000 1000–3000 2500-6000 Technologies; European Catalyst and Tech-
H2 cons, SCF/B <250 200–500 >500 nology Conference, March 2004].
Catalyst CoMo CoMo/NiMo/CoNiMo NiMo These catalysts surpass previous cata-
lysts in desulphurisation, denitrogena-
Table 1 tion and aromatic saturation activity.

34
P T Q SUMMER 2004
 REFINING

less of the situation, these catalyst

advancements offer significant opportu-
nities to improve operations reliability
and increase operating margin.

This article is based on a paper presented at

the 2004 NPRA Annual Meeting in San
Antonio, Texas, USA.

Salvatore P Torrisi is clean fuels technical

specialist for Criterion Catalysts & Tech-
nologies, with responsibility for clean fuels
project coordination in the Americas. He
joined the CRI International family in 1996,
becoming lab manager within the research
and commercial development group where
he was responsible for supporting regenera-
tion and presulphurisation technologies.
Figure 9 Structural composition of metals active sites for conventional and high activity More recently, he has been involved with
HDS catalysts new product market introduction, particu-
larly with application of proprietary Cen-
Thus they are well-suited for ULSD pro- improvement could be used to increase tinel catalysts.
duction, especially for treating feeds feed rate for added ULSD production or E-mail: [email protected]
with large amounts of SPA sulphur and to process difficult feeds such as LCO or P Michael Gunter is marketing manager for
nitrogen species. For refiners who will coker distillate, which were originally Criterion Catalysts & Technologies LP. He has
be revamping or building new excluded because of the high concentra- held a variety of sales, technical service and
hydrotreaters for ULSD production, tions of SPA sulphur and nitrogen business management positions at Criterion
these catalysts may help reduce invest- species. For example, a higher percent- and CRI Catalysts over the past 20 years. He
ment costs by reducing reactor size or age of lower quality distillate streams received a BS in chemical engineering from
eliminating the need to add recycle gas could be upgraded to high quality fuels, Columbia University and a BS in mathemat-
scrubbing facilities. or the overall distillate output (kero, jet, ics from The University of the South.
Alternatively, the performance diesel, etc.) could be increased. Regard- E-mail: [email protected]

35 ENQUIRY NO 320
www.ptqenquiry.com
P T Q SUMMER 2004