10 4 SPECIAL EDIT ION Fa l l 2008

Upgrading the Bottom of the Barrel

IN SIDE...

Feed Contaminants in
Hydroprocessing Units

Maximizing
ULSD Unit Performance

New 420DX Catalyst

New 585DX Catalyst

Inlet Diffuser and

Random Packing

SmART Catalyst
System® Series

Hydroprocessing Catalysts from

The Chevron and Grace Joint Venture
Dear Hydroprocessor:

This special issue of the Catalagram® is devoted solely to you and your operations. Hy-
droprocessing engineers are continually faced with challenges, such as the rapidly increasing de-
mand for ultra low sulfur diesel worldwide, as well as the requirement for low sulfur FCC feed to
meet increasingly stringent government regulations.

Advanced Refining Technologies was formed by Chevron USA and Grace Davison to bring
the best possible products and technologies to hydroprocessors. Since our inception in 2001, we
have continuously developed and manufactured catalysts that upgrade the heaviest feeds, deliver
exceptional run length and activity, and produce the cleanest fuels.

Why should you make ART part of your bid process?

• ART is the only catalyst company experienced at minimizing sulfur and heavy metals in
the full boiling range of products;

• More refiners use ART’s high performance catalyst systems while processing difficult feed
in high pressure residuum fixed-bed and ebullating bed hydroprocessing units than any
other catalysts;

• ART’s research and development experience focuses on customizing catalyst solutions to

fit feed characteristics, reactor operating parameters, and desired product slate. This
includes running pilot plant tests with prospective customers' desired feed to optimize the
catalyst system and to gain confidence in commercial performance expectations;

• ART’s distillate catalyst R&D lab and production facility are in the same location, ensuring
you that production quality assurance aligns with R&D design parameters.

Contact your ART sales representative or me to discuss how ART can optimize your hydropro-
cessing operation. We stand ready to deliver the performance you demand.

Sincerely,

Robert H. Bullard
Vice President and Managing Director
Avanced Refining Technologies
 In this Special Issue of the CATALAGRAM®
Feed Contaminants in Hydroprocessing Units
By Josh Siegel and Charles Olsen, ART 2
1 0 4 SPECIAL EDITI ON Fa l l 2 0 0 8

Opportunity feeds can help improve profitability, but there can be some consequences.
Upgrading the Bottom of the Barrel Many of these new feeds may contain unknown levels of common catalyst poisons
IN SIDE ...
such as silicon or arsenic. It is important to be aware of the various contaminants to
Feed Contaminants in
Hydroprocessing Units avoid unfortunate surprises like pressure drop build up or unexpected catalyst deacti-
Maximizing
ULSD Unit Performance

New 420DX Catalyst

vation which can both result in shortened cycle length on the hydrotreater and unex-
New 585DX Catalyst

Inlet Diffuser and

pected turnarounds. In this article, guidelines are suggested to help minimize the
Random Packing

SmART Catalyst
System® Series potential impact.

Hydroprocessing Catalysts from

Maximizing ULSD Unit Performance when Processing LCO and other

The Chevron and Grace Joint Venture

Previously Processed Feeds

7
CATALAGRAM 104 ®
By Brian Watkins, ART
SPECIAL EDITION ULSD has evolved from simply meeting the diesel sulfur specification to a constant
awareness of unit performance in order to process difficult streams such as FCC light
cycle oil (LCO) and other thermally cracked stocks; synthetic crudes; and various other
Managing Editors: pre-processed feed sources. It is important to understand the impact of processing
Charles Olsen new feed streams, and this paper highlights a few examples demonstrating significant
and differences in feed reactivity, which are not necessarily anticipated from the usual bulk
Lauren Blanchard feed analyses.

Contributors: 420DX: A New High Activity CoMo Catalyst for ULSD by ART
Gerianne D’Angelo By Brian Watkins and Charles Olsen, ART 12
Garry E. Jacobs ART recently announced the commercialization of its newest DXTM Platform Catalyst,
Dave Krenzke 420DX. 420DX catalyst will enable refiners to enhance their ULSD operation with either
Charles Olsen increased cycle length or additional use of opportunity feedstocks in order to maximize
Greg Rosinski margin. Commercial samples of 420DX have been sent to several major oil companies
Josh Siegel with positive feedback that “420DX is a top tier pro capable of exceeding refiners
Brian Watkins needs in demanding ULSD applications.”

Please address
Improving FCC Feed Quality with ART’s Newest FCC Pretreat Catalyst, 16
585DX
your comments to By Brian Watkins, ART
betsy.mettee@grace.com ART’s newest generation of ultra high activity NiMo catalyst, 585DX offers refiners a
significant boost in its ability to generate low sulfur FCC products as well as deliver
Advanced Refining benefits of nitrogen removal and poly aromatic saturation.
Technologies
7500 Grace Drive
Combining New and Old Technologies – Inlet Diffuser and Random
Columbia, MD 21044
410.531.4000 Packing Dramatically Improve Reactor Performance
19
By Garry E. Jacobs, Fluor Enterprises, Inc. and Gerianne D’Angelo, ART
www.e-catalysts.com This paper, delivered at the 2008 NPRA Annual Meeting, presents the case history of a
kerosene hydrotreater that failed to meet product sulfur targets, subsequent to a
catalyst changeout.

ART Excels In ULSD Service: Update on Sulfur minimization By ART

By Greg Rosinski, Dave Krenzke, and Charles Olsen, ART 29
ULSD production with the first SmART Catalyst System® Series began early in 2004 at a
North American refinery processing a feed containing 40% of a high endpoint LCO.
Since that time DX™ Platform Catalysts have been selected for over 35 ULSD applica-
tions as either stand-alone catalysts or as components in SmART System. The technolo-
gy has been a great success since its introduction with millions of pounds installed in
commercial units around the world.

The information presented herein is derived from our testing and experience. It is offered, free of charge, for your con-
©2008 sideration, investigation and verification. Since operating conditions vary significantly, and since they are not under our
control, we disclaim any and all warranties on the results which might be obtained from the use of our products. You
W. R. Grace & Co.-Conn. should make no assumption that all safety or environmental protection measures are indicated or that other measures
may not be required.
 Feed Contaminants
in Hydroprocessing Units
Josh Siegel
Technical Services Engineer
Charles Olsen
Worldwide Technical Services
Manager
ADVANCED REFINING
TECHNOLOGIES
Chicago, IL USA
2 www.e-catalysts.com
Introduction
efiners are facing many chal-
R
lenges from new regulations
on sulfur levels in gasoline and
diesel fuel, the rising cost of crude
and other raw materials, and direc-
tives to decrease the cost of produc-
tion. In an effort to lower these costs,
refineries often find opportunities to
purchase lower cost crudes or pur-
chase other feedstocks such as LCO.
While these opportunity feeds can
help improve profitability, there can be
some consequences. Many of these
new feeds coming into the refinery
may contain unknown levels of com-
mon catalyst poisons such as silicon
or arsenic. It is important to be aware
of the various contaminants to avoid
unfortunate surprises like pressure
drop build up or unexpected cata-
lyst deactivation which can both
result in shortened cycle length on
the hydrotreater and unexpected
turnarounds.
In this article, several of the more
common feed contaminants that
may be present in hydrotreater
feeds are reviewed. In some cases
guard catalysts can be employed to
help mitigate some of the problems
caused by catalyst poisons and in
other cases, guidelines are sug-
gested to help minimize the poten-
tial impact.
Silicon Figure 1
Silicon (Si) is probably the most
widespread catalyst poison encoun- CH3
tered in distillate hydrotreater feeds. CH3
The common source of silicon is
[ O Si]n O H2
from a delayed coker operation CH3
which uses an anti-foam agent
[ O Si]n O
based on polydimethylsiloxane to H [ O Si]n O
suppress foaming in the coker O H
drums. The siloxane complex O H H O
breaks down in the coking process
to lighter molecular weight frag-
ments consisting of modified silica
gels. These remnants end up prima-
rily in the naphtha range, although Al2O3 Al2O3 Al2O3
small quantities have also been
found in the kerosene and diesel
fractions. As a result, silicon con-
tamination is a major concern in
units treating coker naphtha. In
these units, the rate of silicon depo-
Figure 2
sition on the catalyst is usually what
determines run length rather than Same Canister Silicon Capacities
coke deactivation. In extreme cases 8.0

of contamination, cycle lengths can 7.0

Pounds of silicon / ft3 of catalyst

be as short as three to six months.

6.0

Under hydrotreating conditions, the 5.0

silica fragments present in the feed
4.0
undergo a condensation reaction
with the alumina forming a strong 3.0
chemical bond as depicted in 2.0
Figure 1. Thus, once the silicon is
bound to the surface it cannot be 1.0

removed and is considered a per- 0.0

manent poison. The modified gels Comp
pet
etit
itor
Competitoror A Compet
etit
itor
or C
Competitor AT724G
AT7
A 724
24G
G AT535
T535
AT535
are associated with the alumina
support as opposed to the active
metal sulfides, and are dispersed
throughout the alumina surface of
the catalyst.
Figure 3
As the silicon builds up on the cata-
StART TM
Catalyst System Offers Longer Cycles
lyst it begins to restrict the catalyst
pores and eventually blocks access 25

to the active sites. This phenomenon

is referred to as pore mouth plug- 20
Cycle Length, months

ging.
15
Silicon is one of the contaminants
which can be trapped by using spe-
10
cially designed catalysts. ART intro-
duced the StART™ Catalyst System
for just this purpose. This technolo- 5

gy combines a state of the art silica

guard material, AT724G catalyst 0
along with the active HDS and HDN Cycle 1 Cycle
Cyclle 2 TTM Cat
StART Catalyst
alyst
l System
Syste

ART Catalagram® 104 Special Edition Fall 2008 3

 Figure 4 Canister data for a variety of ART
Temperature Dependence of Silicon Capacity catalysts indicates that catalysts
containing nickel are more effective
14
for trapping arsenic. Figure 6 sum-
Silicon capacity, lbs Si/ft3 catalyst

12 marizes the arsenic pick up values

for several ART NiMo catalysts. As
10
this data shows, both AT535 and
8 AT575 are quite effective for trap-
ping arsenic. Note that the differ-
6
ence between canisters 1 and 2 are
4 the level of arsenic in the feed (can-
2
ister 2 was in a unit processing
>80% coker naphtha) and the tem-
0 perature of the reactor containing
460 480 500 520 540 560 580 600 620 640 660 the canister.
Temperature, ˚F

The difference between canisters

catalyst AT535. The AT724G is a as use of these crudes, especially indicates that like silicon pick up,
high surface area guard catalyst synthetic crudes from Canada, has the ultimate arsenic pick up is
designed for maximum silicon increased in recent years. The strongly dependent on temperature.
capacity. Figure 2 compares the Si Arsenic is believed to bind with the Figure 7 shows the As pick up as a
capacity of AT724G and AT535 with metal sulfide sites, and in particular function of temperature for an ART
some competitor catalysts. AT535 the active nickel on the catalyst form- NiMo catalyst. These results were
by itself has essentially the same Si ing nickel arsenide. This has a dra- obtained by analyzing spent sam-
capacity as these materials, while matic impact on catalyst activity. To ples of the catalyst from a three
AT724G has over 30% higher Si demonstrate the impact of arsenic on reactor unit processing 100%
capacity. catalyst activity, ART obtained a series cracked naphtha from a synthetic
of catalysts containing different levels crude source. The first reactor was
Figure 3 shows a commercial exam- of arsenic. These samples were care- operated at very low temperature
ple of the how the StARTTM Catalyst fully regenerated in the laboratory, and (~275°F) in order to saturate
System can increase the cycle were then activity tested using a diolefins. The second reactor was
length over competitive silicon trap- diesel feed containing 50% cracked designed to saturate mono-olefins
ping materials. In this case, ART’s stocks under conditions producing and operated at about 430°F. The
custom designed system more than <500 ppm sulfur. Figure 5 summa- last reactor had an inlet of 570°F
doubled the cycle length over the rizes the results of that work. At 1000 and an outlet temperature of
previous cycles with catalyst from ppm arsenic the catalyst shows 5°F approximately 650°F. The arsenic
competitor A. HDS activity loss and nearly 15°F loss content on the catalyst correlated
in HDN activity. The activity loss with the temperature of the reactor
Another important aspect to be quickly increases to over 50°F at 1 as depicted in the figure. The data
aware of with silicon poisoning is wt.% arsenic on the catalyst. demonstrate that a high nickel cata-
that the deposition of the modified
silica gels on the alumina surface is
a catalytic reaction and the ultimate
Figure 5
quantity of silicon pick up depends
on reactor temperature. The tem- Effect of Arsenic on Catalyst Activity
perature dependence of silicon pick 90
Degrees Activity Loss, ˚F

up for AT724G is shown in Figure 4. 80

A catalyst in a guard reactor which 70
is limited to low temperatures will 60
pick up much less silicon than the
50
same catalyst in a main reactor
operating at higher temperature. 40
30
Arsenic 20 HDS

10 HDN
Arsenic (As) is found in many
0
crudes including some from West
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
Africa and Russia as well as many
synthetic crudes. It is frequently wt.% Arsenic on Catalyst (Carbon free basis)
becoming a common contaminant
4 www.e-catalysts.com
 Figure 6 lyst can pick up very high arsenic
Arsenic Pick-Up in Basket Testing levels if the operating temperature
and feed concentration are high
0.7
enough.
Pounds of arsenic / ft3 of catalyst

0.6 Canister 1
Gulf Coast Unit processing 30% coker Sodium and Calcium
naphtha from another crude source.
0.5 Canister 2
Unit processing >80% Sodium (Na) is a severe catalyst
0.4 coker naphtha from poison that can cause significant
Canadian synthetic crude.
activity loss even at low levels. It
0.3 works by promoting the sintering of
catalytic metals and neutralizing
0.2
acid sites. Typical sources of sodi-
0.1 um include a malfunctioning de-
salter, sea water contamination or
0 caustic contamination. Depending
GSK-6A AT724G
AT7
A 7 24 G AT535
AT5
A 5 35 AT575
AT5
A 575 AT724G
AT7
A 7 24 G A
AT535 on the source of sodium, the signs
of poisoning include rapid activity
loss and an increase in pressure
drop. Figure 8 shows the effects of
sodium poisoning on catalyst activi-
ty. The figure indicates that for a
Figure 7 sodium content of 0.5 wt.% the
Temperature Dependence of Arsenic Capacity activity is at most 60% of fresh cat-
2.50 alyst activity. This translates to
roughly 30°F loss in activity for 1
2.00
wt.% sodium on the catalyst.
lbs As/ft3 of Catalyst

Calcium (Ca) is a similar poison to

1.50
sodium and is found in some West
African crudes. There is some evi-
1.00
dence that it is an even stronger poi-
son with roughly 1 wt.% calcium
0.50
resulting in 50°F or more activity
loss. It is therefore critical to keep
0.00
these out of the hydrotreater feed
200 250 300 350 400 450 500 550 600 650 700
Temperature, ˚F
with a suggested maximum of 0.5
ppm of either sodium and calcium.

Phosphorous

Phosphorous (P) contamination in

oil has been traced to fractionation
Figure 8 fluids that are often used in crudes
from the Western Canadian
Impact of Sodium Contamination on Activity Sedimentary Basin. The source is
100 diphosphate esters which are solu-
ble in the crude oil. Refineries that
90
run large percentages of light
Relative Activity, %

80 Western Canadian crude have

reported crude column and crude
70 furnace fouling for many years.
Improvements made to crude
60 columns to minimize fouling have
transitioned the depositing of phos-
50
phorous to the downstream
40 hydrotreaters.
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80
wt.% Sodium on Catalyst

ART Catalagram® 104 Special Edition Fall 2008 5

 Other sources of phosphorous
include gasoline slop tanks, import-
ed feeds and lube oil wastes. If
phosphorous does manage to make
its way into the hydrotreater, it will
poison the active sites of the cata-
lyst causing a loss in activity. A level
of 1 wt.% of phosphorous on the
catalyst results in roughly 10°F loss
in activity. ART recommends that a
feed content of < 0.5 wppm be
maintained whenever possible, as
well as the use of feed filters to
assist in trapping of phosphorous
sediment.
A case study of the detrimental
impacts of feed poisons on
hydrotreater performance involved a
ULSD unit which had recently start-
ed up with ART catalysts. Shortly
after start up the unit began to
experience extremely rapid catalyst
deactivation. It was so severe that
within a few months the unit
required an unplanned turnaround
and fresh catalyst was installed.
Samples of spent catalyst were col-
lected representing catalyst
throughout the bed, and these were
analyzed in the laboratory. The
results are summarized in Table I. It
is apparent from these results that
the catalysts were exposed to high
levels of several poisons including
arsenic, sodium, phosphorous and
iron. The contaminants penetrated
well into the catalyst bed. Catalyst
at the bottom of the reactor was not
yet poisoned, but the coke content
was extremely high for catalyst
which had been onstream such a
short time. The level of contami-
nants indicates the catalyst in the
top half of the bed had lost over
60°F of activity, while the bottom
was providing most of the HDS con-
version. This required very high
temperatures, which is reflected in
the high carbon content for the cat-
alyst at the bottom of the bed.
Iron
Iron (Fe) works its way into
hydrotreater feed as rust and iron
scale from corrosion of upstream
equipment and piping, as well as
from unfiltered particulates present
in the feed. Iron naphthenates can
6 www.e-catalysts.com
form from corrosion due to naphthenic
acid in the feed, and the iron readily
precipitates out in the presence of
heat and H2S. These iron particulates
fill the interstitial spaces in the catalyst
bed which will result in a higher than
expected pressure drop. To help miti-
gate the pressure drop associated
with iron, ART uses a series of grad-
ing materials (GSK-19, GSK-9, GSK-
6A and GSK-3A) which have high void
space to accumulate and ‘store’ these
particulates. GSK-9 is also an iron
trapping material which has high inter-
nal void space for trapping soluble
iron inside its large pore network.
These measures are helpful for delay-
ing pressure drop buildup, but they do
not prevent or eliminate it. Effective
feed filtration to remove particulates
(at least 25 microns) in combination
with high void grading provides a
longer lasting way in helping mitigate
pressure drop buildup from these
sources.
Nickel and Vanadium
Nickel (Ni) and Vanadium (V) contam-
ination have been found in resids and
the heavier fractions of vacuum gas
oils. They are a more significant prob-
lem in FCC pretreat units and other
applications processing heavier
feeds, and not often encountered in
diesel or other light feeds. The deac-
tivation mechanism of these poisons
is pore mouth plugging. Nickel and
Table I
Spent Catalyst Analysis
As, wt.% Fe, wt.%
GSK-19 0.00 0.72
GSK- 9 0.00 1.23
GSK- 6A 0.11 0.11
Top 0.10 0.64
Middle 0.07 0.08
Bottom 0.04 0.00
vanadium are usually contained in
large porphoryin molecules which
are too big to penetrate into the
pores of typical hydrotreating cata-
lysts. Therefore, the nickel and
vanadium end up depositing on the
outside of the catalyst ultimately
blocking access to the active sites
within the pores. Pilot plant testing
with heavy feeds by ART on a wide-
ly used FCC pretreat catalyst,
AT575, indicates that a 1 wt.% Ni+V
on the catalyst results in 5-9°F loss
in activity. AT575 is a catalyst
designed for treating heavy feeds,
so the activity loss will be greater for
a smaller pore (relative to AT575)
catalyst.
Nickel and vanadium are treated by
using specially designed catalysts
for removing metals. These cata-
lysts have pore size distributions
which are tailored to provide very
high capacities for nickel and vana-
dium. ART has vast experience with
treating resid and resid containing
feeds, and offers a wide array of
ICR and HOP series catalysts with
differing metals capacities and HDS
activities. Using these in conjunc-
tion with ART AT or DXTM Catalyst
Platform are an effective way of
dealing with nickel and vanadium in
cases where they are known to be
present.
Na, wt.% P, wt.% C, wt.%
2.43 0.60 0.0
1.71 4.43 0.5
0.21 2.29 3.9
1.66 2.29 6.6
0.20 1.05 7.5
0.02 0.68 11.1
Maximizing ULSD Unit Performance when
Processing LCO and Other Previously
Processed Feeds
W
ith demands for increased
Brian Watkins production of ultra-low sulfur
Supervisor, Hydrotreating diesel (ULSD) and the rise in
Laboratory Services crude prices, unit operation and per-
formance have become significantly
ADVANCED REFINING more challenging. This has prompted
TECHNOLOGIES refiners to look for ways to maximize
Chicago, IL USA their diesel pool by using opportunity
feedstocks. However, use of these
feedstocks may lead to other prob-
lems which impact unit performance.
ULSD has evolved from simply meet-
ing the diesel sulfur specification to a
constant awareness of unit perform-
ance in order to process difficult
streams such as FCC light cycle oil
(LCO) and other thermally cracked
stocks; synthetic crudes; and various
other pre-processed feed sources. It
is important to understand the impact
of processing new feed streams, and
this paper highlights a few examples
ART Catalagram® 104 Special Edition Fall 2008 7
demonstrating significant differ-
ences in feed reactivity, which are
not necessarily anticipated from the
usual bulk feed analyses.
Background
It has been documented for some
time that desulfurization of diben-
zothiophene and substituted diben-
zothiophenes occurs through two
reaction pathways: the direct sulfur
abstraction route and the hydro-
genation abstraction route. The for-
mer involves adsorption of the mol-
ecule on the catalyst surface via the
sulfur atom followed by C-S bond
scission. This path is favored when
using cobalt-molybdenum (CoMo)
based hydrotreating catalysts. The
second pathway involves saturation
of one aromatic ring of the diben-
 Table I feeds. The amount of LCO blended
Base Feed Properties into the diesel hydrotreater has a
much greater effect on catalyst per-
Type SR LCO Ebullating Fixed formance when producing ULSD
Bed Diesel Bed Diesel than when operating under prior,
less restrictive, low-sulfur regula-
tions. Use of diesel range products
˚API 37.72 15.31 32.24 30.9
from ebullating bed resid or fixed
Sulfur, wt.% 1.096 1.041 0.143 0.344 bed resid desulfurizers can also
Nitrogen, ppm 46 837 940 479 have a significant impact on catalyst
Aromatics, lv.% 22.34 68.81 40.38 38.54 activity if not clearly identified as to
Mono-, lv.% 14.43 18.44 32.05 27.28 their origin. The general properties
Poly-, lv.% 7.88 50.37 8.34 11.26 of diesel streams from these units
often indicate that they may be fair-
Dist., D86, ˚F ly easy to hydrotreat due to their
IBP 404 387 366 489 unusually low sulfur content. Table I
10% 470 462 440 538 lists the general properties for
50% 532 542 524 589 examples of each of the base feeds
used in pilot testing at ART.
70% 558 598 552 614
90% 612 660 591 649
Note that Ebullating Bed Diesel rep-
FBP 678 712 637 704 resents a diesel fraction from the
product of an LC-FINER or H-Oil
zothiophene species followed by to a ULSD unit can have varying Unit, and Fixed Bed Diesel is the
the extraction of the sulfur atom. effects depending on the severity of diesel fraction coming from a Fixed
Nickel-molybdenum (NiMo) cata- the pre-treatment. LCO and coker Bed Resid Unit.
lysts have a higher selectivity for diesels have long been common ele-
desulfurization via this route. ments combined with a straight run The diesel product from an
(SR) feed source to produce ULSD Ebullating Bed Resid (EB) Unit and
It is efficient to model ULSD kinetic products. LCO generated from an the Fixed Bed Resid (FB) Unit pro-
schemes by grouping the various FCC can vary depending on the vide very different sulfur distribution
sulfur species into “easy sulfur” and severity of the pre-treatment of the patterns compared to the other
“hard sulfur” categories. The so- FCC feed. However, the common ele- feeds in Figure 1. It is clear that
called easy sulfur is made up of ment is the increase in polynuclear although the total sulfur is much
compounds which are readily aromatic compounds relative to other
desulfurized via direct abstraction
and boil below about 680°F, while
hard sulfur is made up of com-
pounds which are more readily
Figure 1
removed via hydrogenation followed Sulfur Distribution in the Four Feedstocks
by abstraction. These compounds 4000
include 4,6 dimethyl-dibenzothio-
phene and other di- and tri- substi-
tuted dibenzothiophenes. The rela-
tive amounts of easy and hard sul- 3000
fur in a feed are critical properties to
consider since the concentration of
each can vary significantly from
feed to feed depending on crude 2000

source, boiling range and the prior

thermal or catalytic treatment of the
feedstock. Further details can be 1000
found in the 2007 ERTC paper by
ART entitled “SmART Strategies for
Maximizing ULSD Unit Perform- ne
phe -BT
thio C1 -BT 0
zo C2 -BT
ance: Tuning Hydrogen Utilization Be C3 +B
T
C4 ne
p he BT
for Flexibility with Cracked Stocks.” hi o 1 -D DB
T
zot C T
en -D M DB
Dib C2 -DM BT
4,6 +D
C3
The use of thermal or catalytic treat-
SR Fixed Bed LCO Ebullating Bed
ment of feedstocks that will be sent

8 www.e-catalysts.com
lower for the two resid diesel materi- Figure 2
als, a majority of the sulfur species Comparison of EB and FB Diesel Blends
that are present are all the so-called
hard sulfur species.
in ULSD Application
120

Required Temperature increase, ˚F

To help explain how the HDS activi-
ty changes with product sulfur for
100 SR
these feeds, pilot work was com-
pleted using 100% SR diesel as the 25% EB Diesel
base feed. The individual compo- 25% FB Diesel
80
nents were blended into the base
feed to show the individual, as well
as some cumulative effects on cata- 60
lyst performance. Two different
concentrations (15% and 30% by 40
volume) of LCO were used. The
blends containing LCO produced a
five to seven number decrease in 20
API, as well as decreases in total
sulfur of 2500 to 2700 ppm, respec- 0
tively. Feed nitrogen content, how-
0 100 200 300 400 500 600
ever, increases by 150 to 250 ppm,
and the total aromatic content in Product Sulfur, ppm
each of the blends increases 10 to
18 volume percent. The resid diesel ence between the straight run and the EB diesel blend feed at 200 ppm
streams were blended into the SR at blended resid diesel streams. Figure sulfur and 22°F for the FB diesel.
a 25% by volume concentration. 2 is a plot of the EB and FB diesel Clearly, there is a different tempera-
This gave a similar 1.5 to 2 number blends at different levels of sulfur ture response for each feed which is
decrease in API for each of the FB removal. The impact of adding EB most likely an indication that
and EB diesel streams, with a signif- diesel is little more than 4°F higher although the total sulfur is lower, the
icant increase in total nitrogen of temperature relative to the SR, while additional nitrogen from the EB unit
120 to 220 ppm, respectively. Feed the FB diesel required just over 10°F is hindering the catalytic ability to
aromatic content showed a three to for low sulfur products. As the prod- saturate the more difficult sulfur
four number increase in mono-aro- uct sulfur is decreased the required compounds required to produce
matics. temperatures start to diverge: 17°F ULSD. The FB diesel feed also has
higher temperature is required for the an impact, due to the fact that it
At low severity hydrotreating (higher contains additional refractory sulfur
product sulfur), there is little differ- which is harder to remove.
Figure 3
Addition of LCO on ULSD Performance As product sulfur continues to trend
toward ULSD, the difference
160 increases to over 35°F. A compari-
son of the feedstock inspections
Required Temperature Increase, ˚F

140 SR shown in Table I shows that the two

15% LCO feeds are similar with no obvious
120 30% LCO explanation for a 30°F increase in
required temperature relative to the
100 base feed to achieve ULSD sulfur
levels. This underscores the impor-
80 tance of the source (or history) of a
feedstock, and how significant this
60 can be on unit performance.

40 Figure 3 summarizes data using the

15% and 30% LCO feeds. The
20 impact of even a small amount of
LCO is again readily apparent.
0 Initially about 20°F higher tempera-
0 100 200 300 400 500 600 ture is required compared to the SR
Product Sulfur, ppm feed for 15% LCO, and 40°F higher
ART Catalagram® 104 Special Edition Fall 2008 9
 Figure 4 LCO by itself. This impact can be
Cumulative Addition of LCO for HDS seen through the entire range of
operation as shown in Figure 4.

160 One possible option to gain back

some of the lost activity is to
Required Temperature increase, ˚F

140 SR change the end point of the feed to

25% FB Diesel be used. ART was able to conduct
120 duplicate pilot plant testing on this
15% LCO & 25% FB
same LCO, but with a 30°F end
100 30% LCO & 25% FB point reduction to simulate affect on
catalyst performance. Table II lists
80 the major component analysis
between the two LCO feed sources.
60 The decrease in endpoint lowers
the total sulfur by almost 1000 ppm,
where total nitrogen decreases by
40
129 ppm.
20
The impact this reduction has on
ULSD performance is over 30°F in
0 restored catalyst activity which cor-
0 100 200 300 400 500 600 responds to additional life in the
Product Sulfur, ppm hydrotreater. A comparison of the
two LCO feeds blended at 30% into
temperature is required at the 30% already contains 25% FB diesel has the SR base feed is shown in
LCO level as compared to the base only an additional 10°F impact com- Figure 5.
feed at 500 ppm product sulfur. pared to the 20°F increase in required
Although not obvious from the chart, temperature discussed above. The The addition of LCO has a major
there are also small differences in use of 30% LCO blended into the 25% impact on activity for both the low
the temperature response between FB feed has an impact of 28°F and high endpoint materials. The
the SR and LCO containing feeds. increase in required temperature required temperature increase for
Using 15% LCO there is a 20°F loss which is almost 10°F lower than 30% ULSD in going from 0 to 30% LCO
in HDS activity at 500 ppm sulfur
which increases to 50°F at 200 ppm Table II
sulfur and to over 90°F for ULSD.
With a 30% LCO blend, the activity
Comparison of Boiling Point Reduction on LCO
differences are much greater: 37 °F
at 500 ppm sulfur increasing to 70°F Type LCO LCO
at 200 ppm sulfur and over 115°F for (Low FBP) (High FBP)
ULSD. As mentioned above, this is
an indication that as LCO is added
to the feed, the concentration of ˚API 18.31 15.31
refractory, sterically hindered sulfur Sulfur, wt.% 0.948 1.041
compounds increases making it
more difficult to desulfurize. The
Nitrogen, ppm 708 837
presence of more refractory com- Aromatics, lv.% 66.86 68.81
pounds decreases the temperature Mono-, lv.% 22.65 18.44
response relative to the base SR
Poly-, lv.% 44.21 50.37
feed. That means that a much larg-
er temperature increase is required
for the LCO feed to achieve the Dist., D86, ˚F
same sulfur removal as the base IBP 380 387
feed.
10% 450 462
The effects of combining the vari- 50% 522 542
ous diesel sources are not neces- 70% 579 598
sarily a cumulative effect on catalyst
performance. The effect of adding
90% 638 660
15% LCO into a feedstock that FBP 682 712
10 www.e-catalysts.com
 Figure 5 for the lower endpoint material is
Impact on FBP Reduction on Hydrotreating Performance about 1.2°F per percent LCO.
Processing the higher endpoint
LCO increases the required temper-
160 ature to about 1.4°F per percent
Required Temperature increase, ˚F

LCO. Notice from the chart that the

140 SR
activity effects are not linear with
30% Hi FBP LCO increasing LCO content. The first
120 30% Lo FBP LCO 15% LCO has a larger impact on
activity than the next 15%. Figure 6
100 demonstrates this more clearly in
the form of a plot of activity lost as
80 a function of LCO content. It clear-
ly shows much higher activity losses
60 for the first few percent LCO as
compared to the last few percent
40 added to the feed.

20 ADVANCED REFINING TECH-

NOLOGIES can work closely with
0 refining technical staff to help plan
0 100 200 300 400 500 600 for processing opportunity feeds
such as those discussed above.
Product Sulfur, ppm
One of the keys is being aware of
the potential impact processing cer-
tain feeds will have on unit perform-
ance. Feeds which have been pre-
Figure 6 viously processed present unique
Activity Comparisons at Different challenges and ART is well posi-
LCO FBP and Concentration tioned to provide assistance on how
1.7 best to maximize unit performance
and take advantage of these oppor-
1.6
Low FBP LCO tunity feedstocks.
1.5 High FBP LCO
Degrees ˚F / % LCO

1.4

1.3

1.2

1.1

1.0

0.9

0.8
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32

Percent LCO

ART Catalagram® 104 Special Edition Fall 2008 11

 420DX: A New High Activity CoMo
Catalyst for ULSD by ART

RT introduced its line of ultra metal sulfide sites (see Catalagram®

Brian Watkins
Supervisor, Hydrotreating
Laboratory Services
A high activity DXTM Catalyst
Platform in response to refin-
er’s demands for superior technology
96, 2004). ART CDXi and NDXi cat-
alysts have demonstrated the bene-
fits of this technology in ULSD units
that delivers premium performance. around the world both as stand
Charles Olsen This family of catalysts has exceeded alone catalysts and as part of a
expectations with its performance in SmART Catalyst System® Series.
Worldwide Technical Services
Manager demanding ULSD applications pro-
cessing difficult feed blends. One of Building on this great success, ART’s
the keys to the ultra high activity dedicated research and develop-
ADVANCED REFINING
observed with these catalysts is maxi- ment staff has recently announced
TECHNOLOGIES
mizing the utilization of active metals the commercialization of its newest
Chicago, IL USA
on the catalyst through ART’s chelate DXTM Catalyst Platform, 420DX.
chemistry. This impregnation technol- Figure 1 compares the activity of
ogy significantly improves metals uti- several generations of ART cobalt-
lization by allowing ART to manipulate molybdenum (CoMo) catalysts. The
the active metal components on the figure shows that 420DX offers 30%
catalyst. It has been shown that when improvement in both HDS and HDN
applied correctly chelates promote activity over CDXi catalyst.
and enhance the formation of Type II

12 www.e-catalysts.com
 Figure 1 Researchers at ART have identified
ART’s High Performance Catalysts surface acidity as a key property for
improved catalytic performance. It
is generally accepted that there is a
175 strong relationship between the role
of increased surface acidity and
improved reaction rate for reactions
150 controlled through ring saturation
such as nitrogen and hard sulfur
removal. Changes in surface acidi-
RVA, %

125 ty have also been shown to affect

the interaction of active metals with
the alumina surface during impreg-
100 nation. ART is able to exploit this
phenomenon in the design of
420DX. This catalyst utilizes similar
75 impregnation technology as CDXi
AT405 CDXi 420DX where a chelate is used to bind to
HDS
H D HDN the cobalt ions in the impregnation
solution and reduce interactions
with the alumina support. The
Figure 2
chelate/ion complex stays intact on
IR Spectra of the 420DX Support the catalyst which allows the molyb-
26 denum to sulfide at lower tempera-
1451.67
24
ture promoting the formation of
Lewis Acid sites Type II active sites.
22
Normalized Intensity

20 With 420DX, this technology is

18 enhanced through the use of a
16
modified alumina carrier with
1624.43 improved surface acidity. This is
14
demonstrated in Figure 2, which
1616.20
12 1494.12 shows a pyridine infrared (IR) spec-
10
trum of the new support. The spec-
trum has a doublet at 1624 and
8
1616 wave numbers, as well as an
6 absorption peak at 1452, which are
thought to indicate the presence of
1700 1650 1600 1550 1500 1450
Lewis acid sites. This confirms the
Wave numbers (cm-1) incorporation of surface acidity in
the new support. While the acid
Figure 3 sites give 420DX better perform-
Molybdenum XPS Spectra for Non-Chelated Catalyst ance for both HDS and HDN activi-
ty, they are not strong enough to ini-
Mo Spectra tiate any cracking reactions under
typical hydrotreating conditions.

X-Ray Photoelectron Spectroscopy

Normalized intensity

(XPS) was also used to help better

understand the surface chemistry of
Sulfided 420DX. Figures 3 and 4 summarize
at 755˚F some of these results. Figure 3
Sulfided shows XPS spectra taken on a con-
at 575˚F
ventional (non-chelated) catalyst
which has been sulfided at various
Sulfided temperatures. The spectra cover
at 260˚F the binding energies for the expect-
Fresh ed Mo states with the MoS2 state
highlighted by the dotted red line at
about 229 eV. The fresh catalyst
ART Catalagram® 104 Special Edition Fall 2008 13
 shows the presence of MoO3 as Figure 4
expected. It is not until sulfiding Molybdenum XPS Spectra for Chelated Catalyst
575°F that the MoS2 peak becomes
apparent. Figure 4 summarizes
similar XPS spectra for a chelated Mo Spectra
catalysts like CDXi or 420DX. Mo sulfide appears at 260˚F
Notice in this case that the MoS2 for chelated catalyst

Normalized intensity
peak begins to appear at a sulfiding
temperature of 260°F. This indi-
Sulfided
cates that the presence of the at 755˚F
chelate promotes sulfidation of the Sulfided
Mo at lower temperature compared at 575˚F
to the non-chelated catalyst.
Sulfiding the Mo at lower tempera- Sulfided
at 260˚F
ture results in a more fully sulfided
Mo structure which promotes the Fresh
stacking of the MoS2 into Type II
active sites. This explains much of
the improved activity demonstrated Binding energy (eV)
by the DXTM Catalyst Platform.
Figure 5
To understand the improved activity Cobalt XPS Spectra of CDXi
of 420DX, one has to look at the Co
species as a function of sulfiding 3000
temperature with XPS. Figure 5 Co Spectra
summarizes these data for CDXi.
2500
The XPS spectra now show the
Normalized Intensity

binding energies characteristic of

2000
cobalt oxide (781-782 eV) and Sulfided
cobalt sulfide (778-779 eV). The at 755˚F
1500
binding energy for cobalt sulfide is Sulfided
at 575˚F
highlighted in blue in the figure. The Sulfided
1000
fresh catalyst shows the character- at 396˚F
istic peak for cobalt oxide as Sulfided
500 at 260˚F
expected. The cobalt sulfide peak
Fresh
becomes prominent for sulfiding at
0
575°F, but notice how broad the
810 805 800 795 790 785 780 775
peak is for high temperature sulfid-
ing. The wide peak indicates there Binding energy (eV)
are several states of cobalt present.
Figure 6
Figure 6 shows similar spectra for
Cobalt XPS Spectra of 420DX
420DX. Notice the much sharper
peak for high temperature sulfiding
3000
for the cobalt sulfide species com- Co Spectra
pared to the spectra for CDXi. This
indicates that 420DX has a greater 2500
Normalized Intensity

concentration of the active cobalt

sulfide species which translates to 2000
Sulfided
higher activity. at 755˚F
1500
Sulfided
ART has completed pilot plant test- at 575˚F

ing over a variety of conditions to 1000 Sulfided

at 396˚F
demonstrate the performance Sulfided
advantage of 420DX. Figure 7 500 at 260˚F
shows the results of side-by-side Fresh
testing of CDXi and 420DX at 980 0
psi hydrogen partial pressure and 810 805 800 795 790 785 780 775
2500 SCFB H2/Oil ratio. At these Binding energy (eV)

14 www.e-catalysts.com
conditions, 420DX clearly outper- Figure 7
forms CDXi by over 20°F at 10 ppm Comparison of CDXi and 420DX at High Pressure
sulfur on a difficult feed containing
30% cracked stocks.
650

(10ppm sulfur, 1ppm nitrogen)

One of the additional benefits of
420DX is that the improved HDS 640
and HDN activity has a minimal

WABT, ˚F
impact on increased aromatic satu- 630
ration overall, and consequently
does not increase hydrogen con-
620
sumption to a measurable extent.
This offers refiners greater flexibility
in meeting their HDS activity 610
requirements while minimizing
hydrogen consumption using 600
420DX as a stand-alone catalyst or HDS HDN
in combination with ART’s premium
CDXi 420DX Catalyst
NDXi catalyst in a SmART System
for producing ULSD from difficult
feeds.
Figure 8
Additional pilot plant work was com- ULSD Comparison of CDXi and 420DX
pleted to look at the advantages of
420DX for moderate and high pres-
sure applications using a SR feed-
stock. Figure 8 shows the expected
695 psia
SOR activity for seven ppm sulfur of
420DX and CDX. Note that again at
higher pressure, 420DX shows a
greater than 20°F improvement over
CDXi. At moderate pressure the 1026 psia
activity difference is still greater
than 10°F.

420DX will enable refiners to 600 610 620 630 640

enhance their ULSD operation with
either increased cycle length or CDXi 420DX Catalyst
additional use of opportunity feed-
stocks in order to maximize margin.
Commercial samples of 420DX
have been sent to several major oil

that 420DX catalyst is a top tier

companies with positive feedback

product capable of exceeding

refiners’ needs in demanding
ULSD applications.

ART Catalagram® 104 Special Edition Fall 2008 15

 Improving FCC Feed Quality with ART’s
Newest FCC Pretreat Catalyst, 585DX
s clean fuels regulations
Brian Watkins
Supervisor, Hydrotreating
A become more challenging,
ART’s line of ultra high activity
FCC pretreat catalysts continues to
Laboratory Services evolve. Providing refiners with superi-
or technology and first-class perform-
ADVANCED REFINING ance, ART’s AT575 catalyst has long
TECHNOLOGIES proven its performance advantage
Chicago, IL USA with outstanding stability and excep-
tional ability to provide refiners with
consistent, high quality feed for their
FCC units. In keeping with this tradi-
tion, ART is introducing its newest
generation of ultra high activity NiMo
catalyst, 585DX. Figure 1 compares
the activity of several FCC pretreat
catalysts supplied by ART.
16 www.e-catalysts.com
This newest member to the FCC
pretreat family is capable of signifi-
cantly reducing required Start of
Run (SOR) temperatures for both
hydrodesulfurization (HDS) and
hydrodenitro (HDN) response.
585DX offers refiners a significant
boost in its ability to generate low
sulfur FCC products, as well as
deliver the benefits of nitrogen
removal and poly aromatic satura-
tion.
The enhanced ability for HDS and
HDN allow for use of this product in
a wide range of operation. The
results from side-by-side testing of
 Figure 1 AT575 and 585DX show 585DX
ART’s Line of FCC Pretreat Catalysts clearly outperforming AT575 by over
130% at 1400 psi H2 pressure for
140 both HDS and HDN. In refining
applications that have lower unit
130
Relative Volume Activity, %

pressures, 585DX pilot plant work

120 has shown that the advantage is still
in excess of 115% as compared to
110
AT575 for HDN while maintaining a
100 110% RVA improvement for HDS as
shown in Figure 2.
90

80 Additional performance gains are

70
also available to refiners by utilizing
ART’s ApARTTM Catalyst System.
60 Coupling the improved HDS activity
AT575 AT775 AT792 585DX
of AT792, a trimetallic (NiCoMo)
FCC pretreat catalyst, with the
HDS
H HDN
H
improved HDN activity of 585DX,
allows refiners to produce low sulfur
Figure 2 and nitrogen FCC feeds at signifi-
Comparison of AT575 and 585DX in cantly lower SOR temperatures.
Low Pressure FCC Pretreat Figure 3 shows the advantage of
using this system over a previous
120
system of AT575 and AT775. This
Relative Volume Activity, %

pilot work used a feedstock contain-

ing cracked material and 3000 ppm
110 nitrogen. This work was done for a
refiner at 1600 psia hydrogen pres-
sure. From this testing, the 585DX
and AT792 ApART system provided
100 over a 15°F lower start of run tem-
perature.

An additional benefit of 585DX is

90 the improved aromatic saturation
HDS HDN activity. 585DX shows an increased
saturation of multi ring compounds
AT575
A 585DX
with a 7% increase in conversion of
4+ ring compounds. This increased
Figure 3 saturation allows for improved feed
Comparison of AT575 and 585DX in quality and increased yields from
Two ApARTTM Catalyst Systems for FCC Pretreat the FCC operation. The pilot plant
work shown in Figure 4 used a feed-
720 stock with 2.3 wt.% sulfur and 1900
ppm nitrogen. Using ART 585DX as
(500ppm sulfur & nitrogen)

715 a stand alone catalyst or in combi-

nation with ART AT792 provides
710
many alternatives for producing
WABT, ˚F

quality FCC feedstock.

705

700
585DX is a nickel-molybdenum
(NiMo) catalyst that has outstanding
695 HDS activity coupled with extremely
high HDN and HDA activities. This
690 catalyst provides superior HDN and
HDS HDN HDA to augment refiners’ need for
better FCC feed conversion as well
A
AT575/AT775 585DX/AT792
as reduced sulfur providing support
in producing lower sulfur gasoline.
ART Catalagram® 104 Special Edition Fall 2008 17
 Figure 4
Comparison of AT575 and 585DX for PNA Saturation

80.0
78.0
76.0

% Conversion
74.0
72.0
70.0
68.0
66.0
64.0
62.0
60.0
2 rings+
i 3 rings+
i + 4 rings+
i +

AT575
A 585DX
5

Meet the ART Group

Introducing a few of ART’s world-class team members:

Scott Purnell, is General Manager, with daily responsibilities for the ART business unit. Scott joined Grace in 1993 and
held various technical, marketing, sales and management positions in Grace’s FCC business. Scott has a B.S.Ch.E. from
Penn State and a Ph. D. in Chemical Engineering from the University of Delaware.

Chuck Olsen, Worldwide Technical Services Manager, has been selected to 2008 National Petrochemical
and Refiners’ Association Question and Answer panel. Chuck, who has over 15 years of experience in
hydroprocessing, has held a variety of technical service, research and technical management positions in
Chevron and Grace Davison before joining ART. Olsen holds a B.S.ChE degree from the University of
Minnesota, and M.S. and Ph.D. degrees in chemical engineering from the University of Illinois in
Champaign-Urbana.

Gerianne D’Angelo is Senior Technical Service Engineer with responsibility for estimates, troubleshooting, training engi-
neers and operators, attending meetings to explain proposals, sending out data reports, loading/startups and helping in
anyway possible our clients who have questions or issues on their units. Geri, who has a B.S.Ch.E. from the University of
Illinois, joined ART in 2001 after twelve years at Citgo’s, Lemont, IL refinery.

Woody Shiflett, Director of Global Marketing, is based at the Chevron Technical Center in Richmond. Prior
to joining ART in 2001, Woody was employed Criterion Catalysts from 1983 to 2001, holding increasingly
responsible positions in the hydrotreating business. The author of numerous technical papers, he holds a
B.S.Ch.E. from the University of Akron, a Ph.D. in Chemical Engineering from the University of Wisconsin-
Madison and an MBA from Texas A&M University.

Lauren Blanchard serves as Strategic Business Marketing Manager. She received a B.S. Ch.E. from the
University of Massachusetts and an MBA from Loyola University, Maryland. Since joining Grace in 1993,
she has held operations, technical, quality assurance and marketing positions in both ART and Grace’s
Refining Technologies business.

Mark Peterson, Director Marketing Segment, joined Grace in 2001. Prior to that, Mark was National Sales
Manager, Hydrotreating Catalysts for Akzo and General Manager, Process Technology and Licensing for
Unocal. Mark received a B. A. in chemistry from the University of California-Irvine an Ph.D. in Physical
Chemistry from the University of California-Davis.

Bob Fletcher is ART’s Regional Sales Manager based in Houston, TX. Bob, who holds a B.S. and M.S. in Mechanical
Engineering from the University of Cincinnati, joined Grace Davison’s Refining Technologies in 1993 after holding various
technical specialty chemicals sales positions at Betz and Chemlink.

18 www.e-catalysts.com
 Combining New and Old Technologies -
Inlet Diffuser and Random Packing
Dramatically Improve Reactor
Performance
Garry E. Jacobs
Technical Director,
Process Engineering
Fluor Enterprises, Inc.
Aliso Viejo, CA, USA
Gerianne D’Angelo
Senior Technical Service Engineer
ADVANCED REFINING
TECHNOLOGIES
Chicago, IL USA
he performance of hydropro-
T cessing reactor internals has
improved significantly over
the past decade, driven in part by
ultra low sulfur product specifications.
Yet, some reactors continue to operate
with internals that are inadequate for
the operating conditions. This inade-
quacy may be caused by the intro-
duction of heavier feedstocks or by
constraints that preclude installation of
the appropriate hardware.
This paper presents the case history
of a kerosene hydrotreater that failed
to meet product sulfur targets, subse-
quent to a catalyst changeout. The
reactor operates with a mixed-phase
ART Catalagram® 104 Special Edition Fall 2008 19
feed, but is lacking a liquid distribu-
tion tray. Troubleshooting work
included pilot plant testing of the
catalyst retains, leak testing of the
feed/effluent exchangers, and
radioactive tracer tests and gamma
scans of the reactor. A charge rate
test was also performed which sug-
gested significant underutilization of
the catalyst. Evidence of liquid
maldistribution was also obtained
by visual inspection of the catalyst
bed surface and analysis of “spent”
catalyst samples.
Using computational fluid dynamic
(CFD) modeling, a new inlet diffuser
was designed to improve liquid dis-
 Figure 1
Existing Inlet Diffuser
tribution to the top of the catalyst
bed and was installed during a cat-
alyst changeout. In addition, high
void fraction random packing was
loaded at the top of the bed to pro-
mote radial liquid dispersion. After
restart of the unit, the reactor was
able to meet the refiner’s 10 ppmw
sulfur target for ultra low sulfur
kerosene (ULSK) product.
Background
Subsequent to a catalyst changeout
in the spring of 2007, a kerosene
hydrotreater failed to meet ULSK
specifications. The unit feed con-
sists of a mixture of straight-run
kerosene and coker naphthas.
During the previous catalyst cycle,
the reactor initially treated only fully
vaporized naphtha. A couple of
years into that eight-year cycle, the
straight-run kerosene was intro-
duced, creating a mixed phase
feed. During the final year of that
cycle, the reactor was able to pro-
duce ULSK. The catalyst change-
out was dictated by concerns of sil-
icon breakthrough into the reformer
feed.
The reactor did not have a liquid
distribution tray. The distribution
hardware consisted solely of a
basic inlet diffuser, shown in
20 www.e-catalysts.com
Troubleshooting Activities
A simplified comparison of the
operating conditions and perform-
ance for the two catalyst cycles is
presented in Table I. Of particular
note was the approximately 1½-PSI
reactor pressure drop through the
post 2007 T/A catalyst load. With
uniform vapor/liquid distribution, the
pressure drop should be at least
two times higher. This was the first
evidence of significant liquid mal-
distribution. The ensuing trouble-
shooting activities further explored
this and other possible causes of
the higher-than-expected product
sulfur levels.
Feed/Effluent Exchangers
When producing ultra low sulfur dis-
tillates, extremely small and previ-
ously unnoticeable leaks in
Figure 1. The inlet diffuser is insert- feed/effluent exchangers can result
ed in the reactor inlet nozzle and in off-spec product. Several tech-
helps dissipate the momentum of the niques were used by the refiner to
incoming feed. The addition of a check for such leakage, including
tray was considered by refinery per- feed and product sulfur speciations
sonnel, but could not be justified and tracer tests. If raw feed is leak-
based on acceptable performance ing into the reactor effluent, then
during the catalyst cycle leading up easily hydrotreated sulfur com-
to the 2007 turnaround. pounds (e.g., mercaptans and
alkylthiophenes) will be present in
the product. The speciation results
Table I
Comparison of Operating Conditions and Performance
Catalyst Cycle 1999-2007 Post- 2007 T/A
Naphtha Only (~2 Yrs)
Feedstock Naphtha + Kerosene (~6Yrs) Naphtha + Kerosene
Loading Method Sock Dense
Severity ULSK (Final Year) Inadequate HDS
Axial Temp Profile Typical Typical
Pressure Drop 7 - 9 PSI (Final Year) ~1 1/2 PSI
T/A Observations No Agglomeration
confirmed the presence of mercap-
tans, but not propyl- and lighter thio-
phenes that were abundantly pres-
ent in the feed. The absence of
these thiophenic compounds, which
can not be produced by recombina-
tion reactions, provided strong evi-
dence that feed/effluent exchanger
leakage was not the cause of high
sulfur levels in the kerosene prod-
uct. This result also indicates that
the mercaptans were formed by
recombination reactions, as dis-
cussed subsequently.
Two tracer tests were also per-
formed on the feed/effluent
exchangers. In separate tests, a
high volatility sulfiding agent and a
radioactive noble gas tracer were
injected on the feed side of the
exchangers. The effluent side of
the exchangers was then monitored
for presence of tracer. Both tests
indicated no leakage across the
exchangers, within detection limits.
Implicitly, the absence of sulfiding
agent in the reactor effluent was
also an indication of liquid phase
maldistribution, as the fully vapor-
ized sulfiding agent was completely
converted to hydrogen sulfide in the
reactor.
Catalyst Activity
Since the reactor does not have a
distribution tray, liquid phase sulfid-
ing could not ensure proper activa-
tion of the entire catalyst load. For
this reason, vapor phase sulfiding
was performed using dimethyl
disulfide (DMDS). The sulfiding
step was uneventful, with the low-
and high-temperature hydrogen sul-
fide breakthroughs occurring as
expected.
The catalyst Certificate of Analysis
(COA) was reviewed and found in
compliance with the Process and
Quality Assurance Specifications.
Catalyst retains from the commer-
cial reactor were loaded in a pilot
plant reactor and tested using a
representative blend of the refiner’s
feedstocks. For comparison, a stan-
dard sample of the same catalyst
type was also tested in the pilot
plant. The results confirmed that
Figure 2
Radioactive Tracer Test Results
0
1
2
3
Distance from Top Tangent Line (ft.)
4
5
6
7
8
9
10
11
12
13
14
15
100 1000 10000 100000
Radiaton Intensity (Counts per Time Interval)
the activity of the retained sample was services company to perform two
essentially equal to that of the stan- separate tests.
dard sample, within pilot plant testing
tolerances. Furthermore, the temper- Radioactive Tracer Test
ature required to meet the ULSK sulfur
target in the pilot plant was actually The radioactive tracer test was
10°F lower than the SOR projection for intended to quantify the liquid distri-
the commercial reactor. bution within the catalyst bed. A
radioactive halogenated hydrocar-
The pilot plant results indicated an bon was injected into the reactor
activity disparity of approximately feed. As the feed mixture passed
60°F, compared with commercial unit through the bed, the radioactive
performance. This is roughly equiva- halogen adsorbed onto the catalyst.
lent to a 16-fold difference in space Eight uniformly spaced detectors
velocity. At first glance, this disparity were then lowered about the periph-
may seem too dramatic to be ery of the reactor, and radiation
explained by liquid maldistribution. measurements were obtained at
This point will be addressed subse- multiple elevations.
quently, with the benefit of further
observations. The results of this radioactive tracer
test are shown in Figure 2. The rel-
Vapor/Liquid Distribution Within the ative uniformity of the eight scan
Reactor lines was interpreted as an indica-
tion of “…good liquid distribution
To explore the possibility of liquid within the bed…”, with the caveat
maldistribution within the reactor, the that the test could not identify
refiner utilized a process diagnostics axisymmetric maldistribution.
ART Catalagram® 104 Special Edition Fall 2008 21

Normalized Radar Plots of Radioactive Tracer Test Results
A B
N 2.00 N
2.00
NW 1.50
1.00
0.50
W 0.00
SW
S
4.0 ft. 5.0 ft. 6.0 ft.
Unfortunately, this tracer test was
significantly flawed. A simple flash
calculation at the reactor inlet oper-
ating conditions revealed that the
halogenated hydrocarbon tracer
was approximately 50% vaporized.
Therefore, the results did not pro-
vide an accurate indication of the
liquid distribution, as tracer from the
vapor phase also adsorbed onto the
catalyst. A closer examination of
the data suggests the results are
actually indicative of the vapor
phase distribution.
At each elevation, the raw data was
normalized and graphed on radar
plots (see Figure 3), to facilitate
visualization and interpretation.
• Figure 3A indicates that the
peak tracer deposition shifted from
the E/SE region to the S/SW region.
This shift occurred between the 5’-
0” and 6’-0” elevations, which is
consistent with the transition from
¼” grading to 1/10” silicon guard
catalyst.
• Figure 3B indicates that
another shift in the peak tracer dep-
osition occurred (from the S/SW
region to the West region) between
the 8’-0” and 9’-0” elevations. This
shift is consistent with the transition
from 1/10” silicon guard catalyst to
1/20” HDS catalyst.
• Figure 3C indicates that the
peak tracer deposition stabilized in
the West region through the 12’ 0”
22 www.e-catalysts.com
Figure 3
NW 1.50
NE
1.00
0.50
W 0.00
E E
SW
SE
S S
7.0 ft. 7.0 ft. 8.0 ft. 9.0 ft.
elevation. Minimum tracer deposition
stabilized in the SE region.
These shifts in peak tracer deposition
occurred rapidly, within elevation
changes of one foot or less. In this
reactor, vapor is the continuous
phase. Liquid trickles over the grad-
ing and catalyst, and as such, can not
make rapid radial transitions.
Therefore, the rapid shifts shown in
Figure 3 are likely attributable to
changes in the vapor flow path,
caused by porosity variations. If this
interpretation is correct, the tracer
data indicates locally high and low
vapor-phase space velocities, respec-
tively, in the West and SE regions.
These deviations from uniform flow,
however, are not of the magnitude
necessary to explain the poor reactor
performance. Most notably, the tracer
results provided no meaningful infor-
mation on the liquid-phase distribu-
tion.
Gamma Scan
The gamma scan was intended to
quantify the vapor and liquid distribu-
tion prior to entering the catalyst bed.
A 9 x 9 horizontal fan-patterned scan
was performed, utilizing 9 source
placements and 9 detector place-
ments. The 81 data points were used
to produce the results shown graphi-
cally in Figure 4. This figure indicates
very high densities near the reactor
wall and very low densities in the cen-
C
N
2.00
NE 1.50
NW NE
1.00
0.50
E W 0.00
SE SW SE
10.0 ft. 10.0 ft. 11.0 ft. 12.0 ft.
ter of the reactor. The results were
interpreted as evidence of an
“…annular flow pattern…” with
“…virtually no (liquid) flow in the
center of the bed…”.
An alternate, more probable inter-
pretation of Figure 4 is that the reac-
tor wall was too thick (~3”) to pro-
vide any meaningful information
about the phase distribution above
the bed. As presented below, evi-
dence gathered during the subse-
quent turnaround confirmed liquid
flow primarily in the central region of
the bed. The limitations of gamma
scan technology are discussed
elsewhere.
Charge Rate Test
A charge rate test was performed to
provide an indication of catalyst uti-
lization. The test is performed at
two different reactor feed rates (i.e.,
space velocities), holding feed
quality, reactor temperature, and
gas-to-oil ratio constant. Catalyst
activity is dictated primarily by the
latter two variables. Therefore, the
rate constants for both space veloc-
ities can be equated, as shown in
Equation 1. The feed and product
sulfur concentrations, along with the
two space velocities, are inserted
into this equation to determine the
apparent HDS reaction order, n.
 Figure 4 Intrinsic kinetics for individual sulfur-
Gamma Scan Above Catalyst Bed bearing species (e.g., thiophene)
are generally first order with respect
to the species concentration.
However, petroleum-derived oils
contain a broad spectrum of sulfur
species, with widely varying reactiv-
ities. This wide variation increases
the apparent (i.e., observed) HDS
reaction order to 1.5 – 2.0 for bulk
desulfurization.

The processing objective of the

kerosene hydrotreater is ~99.5%
HDS. This level of treating is beyond
bulk desulfurization, with only alkyl-
benzothiophenes remaining in the
reactor effluent. Therefore, the
apparent reaction order is likely in
the range of 1.0 – 1.5. In contrast,
the charge rate test indicated an
apparent reaction order of 4.3,
which corresponds to 25 – 30% of
the catalyst not being utilized,
depending upon the assumed reac-
tion order (see Figure 5).

Equation 1 The charge rate test has shortcom-

Analyzing Charge Rate Test by Equating Rate Constants ings, though. It does not consider
whether the reacting sulfur species
are present in the vapor phase, liq-
uid phase, or both phases. In this
regard, it seems more applicable for
troubleshooting reactors in which
the reacting species remain pre-
dominantly in the liquid phase at the
reactor operating conditions (e.g.,
gas oil hydrotreaters). This was not
Figure 5
the case for the kerosene
Estimate of Unused Catalyst (Based on Charge Rate Test) hydrotreater, as approximately 35%
of the feed was in the vapor phase
during the charge rate test.
Observed HDS Reaction Order

Subsequent to these troubleshoot-

ing activities, the refiner requested
that Fluor design a new inlet diffuser
to improve liquid distribution to the
catalyst bed.

Inlet Diffuser and Random

Packing Improve Liquid
Distribution

The vapor and liquid phases enter-

ing a reactor can be highly maldis-
tributed due to the impact of the
elbows in the reactor feed line, just
upstream of the inlet nozzle. The
flow paths taken by both phases are
% of Catalyst Not Utilized dependent upon their relative
ART Catalagram® 104 Special Edition Fall 2008 23
 momentums and the piping geome- Figure 6
try, and are not easily predicted Vapor/Liquid Maldistribution at Reactor Inlet
without a tool such as CFD. The
potential extent of maldistribution is
illustrated in Figure 6.

This picture is the result of CFD

modeling of the refiner’s reactor
inlet line. The model extends from
near the heater outlet, through the
combining tee, and along the com-
mon line to the reactor inlet. In the
picture, the interface between the
vapor phase and the liquid phase is
green. At the inlet to the vertical
run, the liquid collects on the far
side of the elbow. In the horizontal
run above the reactor, the flow
begins transitioning into a stratified
regime, with the majority of the liq-
uid flowing along the bottom of the
line. At the final elbow, the liquid’s
momentum carries it to the far side Figure 7
(i.e., east side) of the reactor inlet Fluor Inlet Diffuser
nozzle.

The inset circular image is a cross-

sectional view, just as the flow pass-
es through the reducing flange and
enters the top of the reactor. In this
image, the liquid is red and the
vapor is blue. The two phases are
quite segregated, with liquid heavily
biased toward the east side of the
reactor. The function of the Fluor
inlet diffuser is to correct this mald-
istribution.

New Technology: U.S. Patent No.

7,281,702

Fluor’s patented inlet diffuser is

illustrated in Figure 7. It has been
designed to accommodate the wide
variations in vapor and liquid mal-
distribution that may occur due to
different operating conditions and
reactor inlet line routings. This dif- Figure 8
fuser utilizes three stages, con-
tained in a cylindrical cartridge, fol- Inlet Diffuser Liquid Spray Pattern
lowed by a deflector plate. Each
stage consists of two sets of vanes.
Flow is outward through the upper
set of vanes and inward through the
lower set of vanes.

The upper vanes induce swirling,

which forces the liquid toward the
cartridge wall, creating hold up (i.e.,
allowing the vapor to slip past the
24 www.e-catalysts.com
liquid). The lower vanes act as vor- Figure 9
tex breakers, reducing the swirl of 50mm Ceramic Raschig Rings
the inflowing fluid. The fluid flows
downward in the center of the car-
tridge to pass from one stage to the
next. The swirling action redistrib-
utes the liquid circumferentially.
Exiting the third stage, the liquid
and vapor discharge onto a deflec-
tor plate, creating a hollow cone
spray pattern (see Figure 8).

With this spray pattern, liquid wet-

ting at the top of the catalyst bed is
in the form of an annular ring. The
benefits of an annular ring wetting
pattern have been revealed and uti-
lized , previously. Prior to reaching
the active catalyst, liquid distribu-
tion within the bed can be further
improved using old technology.

Old Technology – Raschig Rings (Courtesy of Saint-Gobain NorPro)

Raschig rings (see Figure 9) were

invented at the end of the 19th cen-
tury by Fritz Raschig to improve the packings to be objectively compared. refiner install a custom designed
fractionation of carbolic acid. The radial dispersion coefficient is inlet diffuser, along with 3 feet of 50-
Since then, higher efficiency ran- dependent upon the packing diame- mm raschig rings, loaded on top of
dom packings have been devel- ter. Figure 10 illustrates that 50-mm the existing catalyst. This combina-
oped for distillation applications. raschig rings are approximately three tion is illustrated in Figure 11. The
Interestingly, though, raschig rings times more effective than the 15-mm reactor feed liquid is treated in an
can still play an important role in spheres commonly used as grading in annular portion of the catalyst bed,
process engineering. These high hydroprocessing reactors. while the vaporized feed is treated
void fraction rings promote rapid liq- primarily in the unwetted portion.
uid dispersion in trickle flow appli- Combining Technologies
cations. This characteristic is quan- Unit Shutdown Activities
tified by a radial dispersion coeffi- To improve liquid distribution in the
cient, which allows various random reactor, Fluor recommended that the In the fall of 2007, the kerosene
hydrotreater was taken off-line. The
refiner felt that a catalyst changeout
Figure 10 was warranted, to maximize the
Radial Dispersion Coefficients probability of achieving ULSK sulfur
levels upon restart. Prior to unload-
6 ing, the top of the bed was visually
Radial Dispersion Coefficient (mm)

inspected. The inspection revealed

an ~3-foot diameter, ~8-inch deep
crater below the inlet diffuser. This
4 crater was unambiguous evidence
Raschig Rings that liquid had “channeled” down
the center of the reactor and
Pall Rings caused localized settling. Due to
2 the settling, the grading materials
had become intermixed. The
Spheres inspection confirmed the inaccura-
cy of the gamma scan results.
0 Approximately 4½ feet of grading
0 5 10 15 20 25 30 35 40 45 50 and catalyst were then vacuumed
Nominal Diameter (mm) from the top of the catalyst bed.
ART Catalagram® 104 Special Edition Fall 2008 25
 Figure 11
Improved Liquid Distribution
HDS catalyst samples were
obtained by coring several feet fur-
ther into the bed. The remaining
catalyst was dumped and the reac-
tor was sock loaded with new cata-
lyst. Because the catalyst did not
pack as densely as planned, only 2
feet of 50-mm raschig rings were
loaded on top of the bed. The Fluor
inlet diffuser was installed prior to
closing the reactor.
Post-Shutdown Performance
Activities
Subsequent to sulfiding activities,
the unit processed only the straight-
run kerosene for several days, after
which the coker naphtha was intro-
duced and operating severity was
slowly increased. The reactor was
able to produce ULSK at a temper-
ature level below the SOR projec-
tion. The refiner was extremely
pleased with the reactor perform-
ance.
Cored Catalyst Samples – Silicon
Loading
The cored catalyst samples were
analyzed for trace contaminants. Of
particular note were the silicon
loadings, as silicon can be used as
26 www.e-catalysts.com
a good marker of the liquid distribu-
tion. As shown in Figure 12, the cen-
ter and east samples contained signif-
icantly more silicon than the other
samples. This silicon distribution is
consistent with the observed crater
and with the CFD modeling (which
indicated an easterly bias to the
incoming liquid).
Cored Catalyst Sample – Activity
Testing
Catalyst from the east core sample
(with 0.68 wt.% silicon loading) was
loaded in a pilot plant reactor and
Figure 12
Silicon Loadings on Cored Catalyst Samples
tested side-by-side with a standard
sample of the same catalyst type.
The tests were performed using a
much heavier (API = 11°) distillate
feedstock, as the refiner’s feed
blend was no longer available. The
results indicated that the east core
sample was 5 – 7°F less active than
the standard sample, at ULSD prod-
uct sulfur levels. Therefore, within
the reproducibility of the pilot plant
testing, this core sample was only
marginally less active than the fresh
retains (tested during the perform-
ance troubleshooting). As such, it is
very inappropriate to refer to this
sample as ”spent” catalyst.
Observed Kinetics Revisited
With visual confirmation that liquid
had indeed channeled down
through the central region of the
reactor, the kinetic performance
during one of the troubleshooting
days was revisited. The basis for
this analysis is presented in Table II.
At the reactor operating conditions,
approximately 70% of the feed was
vaporized. The catalyst volume was
divided into two portions, with one
portion treating the vaporized feed
and the other portion treating the
unvaporized feed.
Gas distribution between the two
portions was estimated based on
the requirement of equal pressure
drop across both portions. The sig-
nificantly reduced gas flow in the
liquid wetted portion most likely
caused localized hydrogen starva- Table II
tion and “hot spots”, an environment Kinetic Analysis of Kerosene Reactor
which promotes mercaptan forma-
tion. Interestingly, mercaptan con-
(Accounting for Feed Vaporization and Liquid
centrations decreased at higher Maldistribution)
operating temperatures, which is
inconsistent with conventional wis- Unvaporized
Vaporized Feed
dom regarding recombination. Feed
Typically, recombination is attrib-
uted to very high operating temper- Unit Feed Rate (BPD) 13,100
atures , rather than localized hydro- Phase Split (BPD) 9,300 3,800
gen starvation. In this instance, API Gravity 41.6 38.8
higher reactor inlet temperatures
Total Sulfur (ppmw) 0.19
increased feed vaporization, which
TBP Distillation (˚F)
increased the gas-to-oil ratio (i.e.,
hydrogen partial pressure) in the 10% 262 310
wetted portion of the catalyst bed. 50% 369 420
90% 481 511
With an LHSV ≈ 3hr-1, the sulfur con-
tent of the vaporized portion is LHSV (hr1)
reduced to less than 1 ppmw. Overall 2.25
Based on an overall product sulfur Each Phase (Note 1) ~3 (Note 2) ~35
content (excluding mercaptans) of
~150 ppmw, the unvaporized por-
Gas-to-Oil Ratio (SCFB)
tion had to contain ~500 ppmw.
Overall ~1100
This concentration requires an
LHSV of approximately 35hr-1. The Each Phase (Note 3) > 1400 < 300
quantity of catalyst involved is
equivalent to the volume of a cone Catalyst WABT (˚F) Same for Both Phases
with a 3-ft. diameter base (i.e., the
diameter of the observed crater) Product Sulfur (ppmw)
and a height equal to that of the Total ~150 (excluding mercaptans)
HDS catalyst bed. Each Phase nil ~500

Repeat Charge Rate Test Notes:

1. Determined by kinetic analysis, using catalyst activity derived from pilot plant testing of
commercial retains.
The refiner is interested in repeating
2. LHSV≈3hr-1 required to treat vaporized feed to <1 ppmw sulfur. Actual LHSV is less
the charge rate test. This test
than 3hr-1.
requires that the coker naphtha be
3. Estimates - based on requirement of equal pressure drop for both the unwetted portion
removed from the unit feed. With no
(vaporized feed) and wetted portion (unvaporized feed) of the catalyst bed.
other convenient destination avail-
able, the test will not be conducted
until the coker is shut down for
maintenance.
It is also important to follow the trail For reactors operating with mixed
that the evidence provides. With cata- phase feed, an appropriately
Concluding Remarks
lyst activity confirmed by testing of the designed distribution tray maxi-
commercial retains and the possibility mizes catalyst utilization and result-
When troubleshooting reactor per-
of exchanger leakage eliminated by ing reactor performance. When
formance problems, it is very impor-
two independent tests, maldistribution installation of a distribution tray is
tant to understand the fundamentals
became the prime suspect. Very low not possible in a revamp applica-
behind the diagnostic tests being
pressure drop – about 1½ PSI – tion, liquid distribution can be
performed. Nothing hampers an
across the reactor was the first direct improved through the use of a new
investigation more than misconcep-
evidence of severe liquid maldistribu- technology, a unique inlet diffuser,
tions and misinformation. In the
tion. The mercaptan-containing prod- and an old technology, raschig
case history presented here, the
uct and the high apparent HDS reac- rings. Together these two elements,
“liquid” phase radioactive tracer
tion order provided further indications at very low relative cost, improve liq-
test results and the gamma scan
of liquid maldistribution. uid distribution and catalyst utiliza-
results both produced misinforma-
tion when producing ultra-low sulfur
tion regarding the vapor and liquid
products in a trayless reactor. The
distribution within the reactor.
ART Catalagram® 104 Special Edition Fall 2008 27
 alternative, neglecting liquid distri-
bution and hoping for adequate per-
formance, should be considered
too great an engineering and eco-
nomic risk to accept.
References
Maiti, R. N., and Nigam, K. D. P.,
“Gas-Liquid Distributors for Trickle-
Bed Reactors”: A Review, Ind.
Eng. Chem. Res. 2007, 46, 6164-
6182.
Alvarez, A., Ramirez, S., Ancheyta,
J., Rodriguez, L. M., Key Role of
Reactor Internals in
Hydroprocessing of Oil Fractions,
Energy & Fuels 2007, 21, 1731-
1740.
28 www.e-catalysts.com
Vidrine, S., Hewitt, P., “Radioisotope
Technology – Benefits & Limitations in
Packed Bed Tower Diagnostics”,
Ethylene Producers Conference at
2004 Spring AIChE Meeting, New
Orleans, LA, 26-28 April 2004.
Unpublished.
Carberry, J. J., Chemical and
Catalytic Reaction Engineering,
McGraw-Hill, New York, 1976; pp
588-589.
Hoftyzer, P. J., “Liquid Distribution in a
Column with Dumped Packing”,
Trans. Instn Chem. Engrs 1964, 42,
T109-T117.
Jacobs, G. E., Milliken, A. S.,
“Evaluating Liquid Distributors in
Hydroprocessing Reactors”,
Hydrocarbon Processing, November
2000, 76-84.
Jacobs, G. E., Krenzke, L. D.,
“Insights on Reactor Internals for
ULSD – Performance of Existing
and New Hardware”, NPRA 2003
Annual Meeting, AM-03-92.
http://www.raschig.com/index.php?
node=208&/~=cb89f5851d60aae8
b6c72ce2779814d5
McCulloch, D. C., “Catalytic
Hydrotreating in Petroleum
Refining.” In Applied Industrial
Catalysis, Volume 1; Leach, B. E.,
Ed.; Academic Press: New York,
1983; pp 113-115.
ART Excels In ULSD Service: Update
on Sulfur minimization by ART

RT first introduced the SmART components in SmART System. The

Greg Rosinski
Technical Service Engineer
A Catalyst System® Series for
ultra-low sulfur diesel (ULSD)
in 2001. Since that time the technolo-
technology has been a great suc-
cess since its introduction with mil-
lions of pounds installed in com-
gy has been widely accepted by the mercial units around the world.
Dave Krenzke refining industry as top tier for ULSD. Several of the first refiners to utilize
Regional Technical Services As detailed previously (Catalagram® a SmART System are still enjoying
Manager 99, 2006), the SmART System is a benefits today, or have reloaded
staged catalyst system customized to another SmART System based on
meet individual refiners objectives the exceptional first cycle they
Charles Olsen received from the technology.
with the performance of the system
Worldwide Technical Services
driven by ART’s DXTM Catalyst
Manager This article contains several case
Platform.
studies highlighting the perform-
ADVANCED REFINING ance of ART catalysts in a variety of
ULSD production with the first SmART
TECHNOLOGIES ULSD units around the world.
System began early in 2004 at a North
Chicago, IL USA These are summarized in Table I.
American refinery processing a feed
containing 40% of a high endpoint
Light Cycle Oil (LCO). Since that time Refiner A is an initial user of a
DX Catalyst Platform has been select- SmART System in Asia Pacific. This
ed for over 35 ULSD applications as refiner conducted in-house testing
either stand-alone catalysts or as for their ULSD catalyst selection.
ART Catalagram® 104 Special Edition Fall 2008 29
 Table I They requested catalyst samples
Summary of ULSD Case Studies from ART and a “market leading”
domestic supplier. The DXTM
Catalyst Platform tested as a sub-
Region Feed Inlet Pressure psig LHSV % CoMo/ % NiMo stantially more active catalyst than
A AP 40% cracked stock 930 0.7 70/30 the others in the program. As a
B AP Straight Run 770 0.7 90/10 result, ART was selected as the cat-
C AP Straight Run 850 1.2 55/45 alyst supplier for their unit, which
D AP Straight Run 1130 1.1 30/70 started up in the last half of 2004.
E NA 40% coker/LCO 1070 1.3 35/65 This refiner was completing a
F NA 50% LCO N/A 1.1 70/30 + Dewax revamp, which involved the addition
G NA 70% LCO 1250 1.0 35/65 of a new reactor in front of an exist-
H NA 45% LCO/LCGO 1300 0.8 25/75 ing reactor. They decided to use
fresh catalyst in the new lead reac-
tor while keeping used ART catalyst
in the lag reactor. The unit condi-
tions are characterized by an inlet
Figure 1 pressure of 930 psig and a LHSV of
Asia Pacific Refiner A 0.7 hr-1. The feed contains up to
720 40% cracked stocks, and the prod-
5 ppm product sulfur uct sulfur has averaged 5 ppm. The
unit ran for three years before the
680 refiner decided to change out the
“used” catalyst in the lag reactor
with fresh catalyst from ART. The
WABT, ˚F

640 performance of the unit is summa-

rized in Figure 1 and, as the figure
depicts, stability has been excep-
600 tional.

Actual Figure 2 shows data for Refiner B.

560 This was another major Asia Pacific
Normalized refiner using a SmART System. This
520 refiner needed to produce 10 ppm
sulfur diesel for two years in a newly
0 200 400 600 800 1000
revamped unit. The operating con-
Days on Stream ditions for this unit are 770 psig inlet
pressure with an LHSV around 0.7
hr-1. The feed was a high end point,
straight run diesel with typical sulfur
content of 1.25 wt.%. This refiner
Figure 2 was concerned with minimizing
Asia Pacific Refiner B hydrogen consumption as hydrogen
800 availability was limited at the refin-
ery. This was a situation well suited
775 High severity mode for the flexibility of the SmART
Product Sulfur: 85% <8ppm System to tune the hydrogen con-
750
sumption/activity relationship. The
WABT, ˚F

725 Low severity mode optimum loading for this unit was
Product Sulfur: 15% <8ppm 90% cobalt-molybdenum (CoMo)
700
and 10% nickel-molybdenum
675 (NiMo) catalyst. This design was
expected to be significantly more
650 Return to low severity mode: active than an all CoMo loading and
WABT dropped about 14˚F
625 would consume the same amount of
hydrogen as an all CoMo loading.
600
The final selection of the SmART
0 50 100 150 200 250 300 350 400 450 500 550 600
System was based on competitive
Days on Stream pilot plant testing by the refiner.
Actual Normalized to 10 ppm

30 www.e-catalysts.com

About eight months into the ULSD
portion of the cycle, the refiner sig-
nificantly increased the operating
severity and began producing very
low sulfur diesel. Typically, 8 ppm is
the recommended product sulfur
target when producing <10 ppm
ULSD. In the higher severity mode,
the product sulfur was well below 8
ppm 85% of the time, whereas in
the lower severity mode the product
sulfur was below 8 ppm only 15% of
the time. The higher severity led to
a much higher deactivation rate and
potentially jeopardized the two year
cycle length. After discussions with
the refiner, the severity was reduced
and the two year cycle length was
achieved.
ART was chosen for the second
cycle which again was based on
competitive testing and the out-
standing performance demonstrat-
ed in the first cycle. The current sys-
tem is performing very well.
Refiner C, another Asia Pacific refin-
er, selected ART catalysts for their
ULSD unit based on the strong ref-
erence from the refiners mentioned
above. This refiner needed to pro-
duce 8 ppm sulfur diesel for two
years and hydrogen availability was
not a constraint. The operating con-
ditions included an inlet pressure of
850 psig and LHSV of 1.2 hr-1. The
feed was also a high endpoint,
straight run diesel with a sulfur con-
Figure 4
Asia Pacific Refiner D
720
700
680
WABT, ˚F
660
640
620
600
0 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375
Days on Stream
Actual
Figure 3
Asia Pacific Refiner C
720
700
680
WABT, ˚F
660
640
620
600
0 50 100
Days on Stream
Actual
tent of 1.8 wt.%. This was a more
demanding operation than those dis-
cussed previously, and a required sys-
tem designed for maximum activity.
The catalyst loading in this case was
55% CoMo and 45% NiMo. The per-
formance is summarized in Figure 3.
This unit met all performance targets
during the two year cycle and has just
been reloaded with a new SmART
System.
Operating data for Refiner D is shown
in Figure 4. This is a grass roots ULSD
unit in Asia Pacific. ART was chosen
to participate in this project because
of the excellent performance of the
SmART System from previous refer-
ences in the region. ART worked
Normalized to 8 ppm
ART Catalagram® 104 Special Edition Fall 2008 31
150 200 250 300
Normalized to 8 ppm
closely with the refiner and engi-
neering construction firm on the
design basis for this project. A 30%
CoMo and 70% NiMo SmART
System was designed for this unit in
order to deliver maximum activity.
The operating conditions for this unit
included an inlet pressure of 1130
psig and LHSV of 1.1 hr-1. This unit
also processes a straight run feed
which has a relatively high level of
nitrogen and an endpoint of 730°F
(by D86).
The unit is running well and meeting
all performance expectations with a
somewhat lower than estimated
deactivation rate, and is well on
track to meet the targeted three-
year cycle length.
Refiner E is another grassroots
ULSD unit, which started up in the
second quarter of 2006 in North
America. ART worked with the
licensor on the unit design and pro-
posed a SmART System loading
consisting of about 65% NiMo. This
loading was designed for maximum
activity as the design feed con-
tained a high percentage of
cracked stocks. The conditions of
the unit included an inlet pressure of
1070 psig with an LHSV of 1.25 hr-1.
This refinery processes mostly
sweet crudes, and the feed to the
unit typically contains about 40%
FCC LCO and coker LGO. The
activity of the system has been
extremely high, allowing this refiner
 to operate at 20% over design Figure 5
charge rate. Figure 5 shows the North American Refiner E
performance for the cycle thus far,
and it is evident from the figure that 720 50
the catalyst stability has been excel- 700
45
lent. In fact, the unit was designed 40

Kerosene, vol.%
for a 24-month cycle, but 48 months 680 35

WABT, ˚F
appears possible even at higher 660 30
than design feed rates. 25
640 20
This unit also typically processes 15
620
10% kerosene with occasional
10
increases to as much as 30%. As 600
5
can be seen in Figure 5, the addi- 580 0
tion of kerosene to the feed has no 0 100 200 300 400 500 600 700 800
negative impact on the activity of Days on Stream
the system, and may even improve
the performance. Notice how the WABT Kerosene
WABT decreases when processing
higher amounts of kerosene. This
suggests that the increase in feed
vaporization (decrease in H2 pres-
Figure 6
sure) is offset by the decrease in
North American Refiner F
hard to treat, substituted dibenzoth-
iophene sulfur species. 720
700
Normalized WABT, ˚F

Figure 6 summarizes data from

Refiner F in North America. This
680
refiner had completed a project to
revamp an existing hydrocracker to 660
ULSD service. The objectives were
to increase feed capacity from 640
30,000 to 45,000 BPSD, while
620
ensuring the capability to process
50% or more LCO, as well as pro- 600
vide for cold flow improvement dur-
ing the winter months. This refiner 580
0 100 200 300 400 500 600
also wanted to minimize the hydro-
gen consumption so that feed rate Days on Stream
could be maximized within make-up
hydrogen constraints. Along with
Süd Chemie, ART designed a
dewaxing/ULSD catalyst system
meeting the unit objectives. The
Figure 7
unit started up successfully in the North American Refiner G
2nd quarter of 2006 and processes 740
both sweet and sour crude derived
Normalized WABT, ˚F

5 ppm average product sulfur

feeds in block operation. The feed 720
is also comprised of 40-50 vol.% of 700
a high endpoint LCO. The perform- 680
ance of the unit is summarized in
Figure 6. The unit came on stream 660
with higher than expected activity, 640
and is well on its way to exceeding 620
the target three year cycle length.
600
Additional details on this unit can be
found in Catalagram® 103, Spring 580
0 100 200 300 400 500 600
2008.
Days on Stream

32 www.e-catalysts.com
 Figure 8 36-month cycle. Also shown in Fig-
North American Refiner G Hydrogen Activity ure 9 is the API uplift from the unit.
From the chart, the API typically in-
1400 100 creases 6-10 numbers depending
95 on the feedstock.

Recycle H2 Purity, vol.%

1300
Inlet Pressure, Psig

1200
1100
1000
900
0 100
Refiner G is another grassroots unit
which started up in the fourth quar-
ter of 2006. The performance of this
unit is summarized in Figure 7. ART
again worked closely with the licen-
sor on the new unit design and pro-
posed a SmART System loading
using 65% NiMo catalyst for maxi-
mum activity. The unit typically
processes >70% LCO, including
both LCO produced within the refin-
ery and purchased externally. The
initial deactivation rate of the unit
was significantly higher than ex-
pected, and it was determined that
this was due to lower H2 partial pres-
sure than design combined with
lower H2/Oil ratio. This can be seen
in Figure 8 which shows the trend of
reactor inlet pressure and the recy-
cle hydrogen purity. Early in the
cycle there were control issues and
the inlet pressure showed a steady
decline. The recycle hydrogen pu-
rity also decreased steadily and had
periods where it fell dramatically.
Once the H2 partial pressure con-
200 300 400 500
Days on Stream
Pressure Recycle H2
SmART System at a U.S. Gulf Coast re-
finery. This was another grassroots unit
that started up in the Fall of 2006. SOR
activity met expectations, and since
that time the unit operating severity has
steadily increased. The current fee-
drate is 22% over design and the feed-
stock endpoint has increased by
30-40°F. Typical unit conditions include
0.8 hr-1 LHSV and 1300 psig inlet pres-
sure, and, on average, the unit
processes a feed containing 15% FCC
LCO, 30% LCGO and 5% coker naph-
tha. At times the unit has processed
as much as 80% cracked stocks in the
feed, and still the product sulfur has av-
eraged less than 6 ppm for the cycle.
The target cycle length was 24 months,
and the unit is currently on track for a
North American Refiner H
700
680
90
As this sampling of case studies
85 demonstrates, the SmART System
80 has been employed in a wide variety
of ULSD applications around the
75 world. The technology has been
70 successfully operating over a broad
range of operating conditions from
65
low to high pressure with feeds rang-
60 ing from straight run to 80% cracked
600 stocks. In each application, all ex-
pectations have been met or ex-
ceeded, and in a number of cases
ART catalyst has been selected for
the second cycle based on the ex-
cellent performance. ADVANCED
REFINING TECHNOLOGIES contin-
ues to develop higher performance
products for ULSD as evidenced by
the recent introduction of the
newest DXTM Catalyst Platform,
420DX. The addition of this catalyst
to the ULSD portfolio is an example
of the commitment that ART will
continue to deliver state-of-the-art
technology for ULSD.
Figure 9
20
18
16
API Gravity Increase

cerns were addressed, combined

660 14
with better bed temperature man-
WABT, ˚F

agement, the deactivation rate de- 640 12

creased significantly. The operation 10
620 8
became much more consistent, and
since that time the deactivation rate 600 6
indicates that the unit will easily meet 4
580
the expected 24-month cycle length. 2
560 0
Finally for Refiner H, Figure 9 sum- 0 50 100 150 200 250 300 350 400 450 500 550
marizes the performance of a Days on Stream

WABT API Uplift

ART Catalagram® 104 Special Edition Fall 2008 33

 Columbia, Maryland USA 1.410.531.8282 Fax 1.410.531.4540

Houston, Texas USA 1.281.449.9949 Fax 1.281.442.0447

Singapore 65.6831.4132 Fax 65.6737.5826

Richmond, California USA 1.510.242.1314 Fax 1.510.242.1412

Worms, Germany 49.6241.4030 Fax 49.6241.403455

Toda, Japan 81.48.431.1952 Fax 81.48.431.1949

The information presented herein is derived from our testing and experience. It is offered, free of charge, for your consideration, investigation and
verification. Since operating conditions vary significantly, and since they are not under our control, we disclaim any and all warranties on the results
which might be obtained from the use of our products. You should make no assumption that all safety or environmental protection measures are
indicated or that other measures may not be required.

Catalagram®, D-PrISM®, Grace®, Grace Davison®, GSR®, SmART Catalyst System®, and SuRCA® are trademarks, registered in the United States
and/or other countries, of W. R. Grace & Co.-Conn. NEPTUNE™ is a trademark of W. R. Grace & Co.-Conn.

ApART™, AT™, DX™ and StART™ are trademarks of Advanced Refining Technologies, LLC. ART® and ADVANCED REFINING TECHNOLOGIES®
are trademarks, registered in the United States and/or other countries, of Advanced Refining Technologies, LLC.

Chevron® and ICR are trademarks, registered in the United States and/or other countries, of Chevron Intellectual Property LLC. LC-FININGTM is a
trademark of Chevron Intellectual Property LLC.

HOP is a trademark, registered in the United States and/or other countries, of Japan Energy Corporation licensed to Advanced Refining Technolo-
gies, LLC and W. R. Grace & Co.-Conn. H-OIL® is a trademark, registered in the United States and/or other countries of Axens North America, Inc.

This trademark list has been compiled using available published information as of the publication date of this brochure and may not accurately
reflect current trademark ownership. This brochure is an independent publication and is not affiliated with, nor has it been authorized, sponsored,
or otherwise approved by any of the aforesaid companies.

© 2008 W. R. Grace & Co.-Conn.