PROCESSING

A recent study for a

grassroots 30,000-b/sd
hydrotreater that can produce ultra-low sulfur
diesel (ULSD) helped assess the critical process
Refining
design issues for meeting
future ULSD requirements.
The feed, a 70:30 mixture of virgin
distillate and FCC light cycle oil, was
the basis for a previous Mustang Engineers & Constructors LP study for revamping an existing facility to meet the
ULSD specification.1
This first of two articles presents a
comprehensive overview of design con-

Ultra-low sulfur diesel

In mid 2006, the US Environmental
Protection Agency will begin enforcing
rules that mandate a maximum sulfur
content of 15 ppm (wt) for on-road
diesel fuel. Refiners are considering several strategies to comply with this rule:
Revamping existing facilities commissioned in the early 1990s that now meet
the low-sulfur diesel specification of 500
ppm (wt) sulfur.
Constructing grassroots facilities to
meet the new specification.
Executing various combinations of
new and revamped facilities.
The problem
Middle distillates contain many sulfur
species, including mercaptans, sulfides, thiophenes, and aromatic sulfur compounds.
Sterically hindered dibenzothiophenes are a group of aromatic sulfur
compounds that are most difficult to remove; they constitute a large fraction
that remains in diesel after hydrotreating to the current specification 500 ppm
(wt). This is particularly true for diesel
fuels that contain significant quantities
of cracked stocks, like FCC light cycle
oil, that contain a large concentration of
aromatic sulfur compounds.
Effective removal of these species requires tailored catalysts and process
conditions. Refiners must also consider
other factors, such as feed nitrogen content and aromatics equilibrium. These
issues have been addressed in detail in
the literature.2-7

siderations for grassroots ULSD hydrotreaters. The chemistry of middle

distillate sulfur species (see box), reaction process variables, and major factors influencing reactor design are covered.
Engineering aspects of hydrotreater
design are discussed, including a suitable operating pressure level that satisfies reaction conditions and the
practical limits of piping mechanical
design.
Alternative process
configurations for
ULSD hydrotreaters
were studied, along
with key design parameters and metallurgical considerations
for major equipment.
Part 2, next week, will present
process simulations for different
process configurations at reactor inlet
pressures of 800 and 1,100 psi. We estimated capital costs and life cycle costs
for three cases.

ULSD
HYDROTREATING1

Study outlines optimum

ULSD hydrotreater design

Design considerations
The ULSD hydrotreater designer
must account for these factors during
the front-end design process:
Feed characteristics and variability.
Other product-quality requirements, especially cetane index.
Catalyst selection.
Optimization of reactor process
variables.
Equipment design requirements.
Reliability.
Future off-road diesel sulfur standards.
Minimizing product contamination.
Handling of off-specification
diesel product.

Les Harwell
Sam Thakkar
Stan Polcar
R.E. Palmer
Mustang Engineers & Constructors LP
Houston
Pankaj H. Desai
Akzo Nobel Catalysts LLC
Houston

Process flow
Fig. 1 shows a simplified process
flow diagram for a diesel hydrotreater.
Stripper bottoms heats fresh feed
from the surge drum, which then mixes with recycle hydrogen. Reactor effluent further heats the combined feed,
which is then heated in the charge
heater.
Reactor inter-bed quench may be required depending on the volume of
Based on a presentation to the 2003 NPRA Annual Meeting, Mar. 23-25, 2003, San Antonio.

Reprinted with revisions to format, from the July 28, & August 4, 2003 editions of OIL & GAS JOURNAL
Copyright 2003 by PennWell Corporation

PROCESSING
cracked stocksFCC light cycle oil and
light coker gas oilin the feed. Fig. 1
shows recycle gas used as quench.
Reactor effluent heats the combined feed and flows to a hot, highpressure separator (HHPS). Vapor from
the HHPS heats the recycle gas and
stripper charge before cooling in the
product condenser and entering the
cold, high-pressure separator (CHPS).
Wash water, injected upstream of the
products condenser, helps remove ammonium hydrosulfide. HHPS liquid
combines with heated CHPS liquid and
flows to the product stripper.
Vapor from the CHPS contacts with
amine in a scrubber for H2S removal
and flows to the recycle compressor

suction drum. Makeup hydrogen is

compressed and combines with recycle
gas in the suction drum. The compressor suction drum can purge some recycle gas to improve recycle gas hydrogen
purity. The quantity depends on the reactors hydrogen-partial-pressure requirements and makeup hydrogen purity.
In the product stripper, superheated
steam feeds the towers bottom and
helps remove H2S. Stripper overhead
vapors condense and flow to the stripper accumulator. Accumulator vapor
and liquid, known as wild naphtha, are
processed in offsite facilities.
Stripped ULSD product supplies heat
to the feed stream and then flows to dry-

ing facilities, which can be a coalescersalt dryer or vacuum drying system.

Feed, product characteristics

Sulfur, nitrogen, and aromatics have
the greatest impact on the ULSD facility
design. Feed nitrogen has a significant
impact on the operating pressure. The
refiner must remove nitrogen to essentially the same level as sulfur to meet
the ultra-low sulfur target.
This means that the catalyst and hydrogen partial pressure must be consistent with a high nitrogen-removal operation.
Light coker gas oil and FCC light cycle oil normally contain most of the
feed nitrogen. Feed aromatics content

ULSD HYDROTREATER

Fig. 1

Makeup H2
LCO

Feed heater
Makeup H2
compression

Straight run diesel

Reactor
Recycle
compressor
Amine
treating

Quench

Feed

ULSD product

Gas to fuel

Sour water
Sour naphtha

Cold
high-pressure
separator

Hot
high-pressure
separator
Steam
Stripper

Wash water

Sour water

hydrogen partial pressure, and reactor

temperature during the process design
phase for a given cycle length and
treating severity.
Catalyst deactivation rate falls with
increasing hydrogen partial pressure;
therefore, the space velocity can increase for a constant cycle length. This
is, however, at the expense of higher
hydrogen consumption.
Another important variable is the ratio of total hydrogen supplied to chemical hydrogen consumption. This value
should be 5-6 for feeds with significant
quantities of cracked stocks. This ratio
is normally less important during the
process design, but it can control the
design hydrogen-circulation rate, especially when a refiner exclusively uses
high-purity makeup hydrogen.
When setting the hydrogen partial
pressure and total operating pressure,
one should be aware that the alloy piping flanges limit the reactor pressure,
especially when the feed contains significant quantities of cracked stocks.
This includes piping between the feed
exchangers and feed heater, between
the heater and reactor, and between the
reactor and combined feed exchangers.
Piping is normally a 300-series
stainless steel; for 600-psi flanges this
corresponds to a maximum operating
pressure of around 800 psi at the reactor inlet. For 900-psi flanges, the reactor operating pressure can increase to
about 1,100 psi. These pressures account for relieving conditions.
These criteria should not necessarily
govern a refiners selection of operating
pressure; however, pressures barely exceeding these limits and requiring the
next-higher-rated ANSI flange class are
Reaction process variables
Key reaction process variables include: costly.
Actual pipe wall thickness is calcu Space velocity.
lated from design conditions; the limit Hydrogen partial pressure.
ing value is used for the piping specifi Makeup hydrogen purity.
Ratio of total hydrogen to chemi- cation in the reactor loops alloy sections.
cal hydrogen consumption.
Actual pipe materials can also be
Cycle length.
316, 321, and 347 grades of stainless
Reactor temperature.
steel. Type 347 has a higher allowable
Refiners can use a nickel-molybdenum catalyst for feeds that have high aro- stress value than the other grades but is
matics or nitrogen; an appropriate selec- normally more expensive. Availability
and delivery time for the various staintion of graded catalysts in the beds top
less components are also a major conwill mitigate reactor pressure buildup.
Refiners must optimize reactor space sideration.
The existing equipment and reactor
velocity, hydrogen treat gas quantity,
governs chemical hydrogen consumption at the low space velocities and
high hydrogen partial pressures required for ULSD production.
Product cetane or gravity generally
determines the amount of cracked
stocks that a refiner can include in the
feed.
There is a small increase in the gravity and cetane index during the hydrotreating reaction.
If a significant improvement in
cetane (3-5 numbers) or gravity is required, a multistage design using aromatics saturation catalysts in the second
stage may be a more economical option. The magnitude of improvement in
gravity and cetane determines this design choice.
Design product sulfur is a key target,
not only from a process design standpoint, but also for offsite storage and
transfer considerations.
Most new and revamp facility designs
are for a product sulfur content of 8-10
ppm (wt).
Refiners should pilot test the feed to
confirm reaction process conditions.
Testing for variations in feed characteristics, especially the FCC light cycle oil
and coker light gas oils back-end distillation, is important because product
separation in these facilities is notoriously poor.
This can result in a temporary spike
in the most difficult-to-treat sulfur
compounds in the hydrotreater feed
and requires a higher reactor temperature. This increases the catalyst deactivation rate and directionally reduces cycle
length.

loops piping limit the hydrogen partial

pressure in revamp designs. A higher
treat gas rate can increase the hydrogen
partial pressure; this is usually limited
due to an associated increase in the reactor loops pressure drop and the corresponding maximum operating pressure of system components.
Lower-purity makeup hydrogen requires higher hydrogen circulation
rates to maintain a target hydrogen partial pressure; it may even require a
purge stream from the cold separator. If
makeup hydrogen purity is too low,
there is no combination of recycle rate
and purge that will achieve the target
reactor outlet partial pressure.
For a revamp design, increasing
makeup hydrogen purity is the most
effective way to increase the hydrogen
partial pressure.
Catalyst cycle lengths of 24-36
months are typical for new designs.
This is because, at some point, factors
other than catalyst activity (such as reactor pressure drop) will limit the cycle. For a fixed space velocity, cycle
length increases with a higher hydrogen partial pressure.
Maximum reactor outlet temperature
at end-of-cycle catalyst conditions is
usually 725-750 F. to avoid aromatics
saturation equilibrium constraints and
to maintain product color. The amount
of cracked stocks in the feed and the
crude source also influence this temperature.
Hydrotreating catalyst performance
correlations for reactor temperature is
usually based on the weighted average
bed temperature. WABT is the reactor
inlet temperature plus two thirds of the
reactor temperature rise. Quenching
limits the temperature rise to 40-50 F.
in each bed.
For a 50 F. temperature rise and a
725 F. maximum reactor outlet temperature, the end-of-run WABT is (72550) + 23(50) = 708 F.
The start-of-run WABT must be high
enough to remove the required amount
of sulfur and nitrogen. The catalyst deactivation rate at the design space velocity and hydrogen partial pressure
determines cycle length. During the cycle, WABT will increase 30-50 F. with
lower deactivation rates at higher hydrogen partial pressures.

PROCESSING
Reactor design
A superficial mass velocity set at
2,000-5,000 lb/hr/sq ft determines
the initial reactor diameter. A value of
3,500 allows for a reasonable turndown rate and some upside allowance.
Mass flow includes all hydrocarbons
and hydrogen at the reactor inlet.
The refiner should consider a twotrain design if the required reactor diameter is larger than could be shop
fabricated and shipped to the plant
siteusually 12-14 ft for overland
shipment. For diesel hydrotreaters, this
is normally 30,000-40,000 b/sd for
each reactor train.
Refiners with adequate water access
and dock facilities can consider larger

reactors and larger single-train capacities.

An appropriate correlation8 estimates
the pressure drop for two-phase flow
through the reactor. The pressure drop
calculation should also include interbed quench.
Catalyst physical properties, including void fraction and equivalent particle diameter, have a significant impact
on reactor pressure drop. To ensure
good distribution in the catalyst beds,
most ULSD reactors are dense loaded
instead of sock loaded. The clean pressure drop should be 0.5-1.0 psi/ft of
catalyst bed.
For a required catalyst volume, bed
heights are either based on the heat of

reaction and maximum temperature

rise or set at 30-40 ft. The designer
then estimates the clean pressure drop
for each bed.
The designer determines the maximum overall pressure drop when designing the bed support and compares
it to the catalyst crush strength, which
is the sum of:
Fouled bed pressure drop at 175%
of calculated clean drop.
Dead weight of catalyst and support material.
Coke deposits at 30% of catalyst
dead weight.
Liquid holdup.
15-psi allowance for depressurization.

P RESSURE PROFILE

Fig. 2

Makeup H2
1,200

LCO

1,198 1,100

Feed heater

1,102

Straight run diesel

1,252

Makeup H2
compression

957 1,054
P = 233

1,152
1,290

Reactor

1,190

Recycle
compressor
Amine
treating

Quench
1,025

1,274
1,120
1,174

1,060

1,123

1,022

963

Feed

1,154

ULSD product

1,002 1,100

1,254
1,186
1,096
998

1,086

1,073

1,065

988

975

968

Gas to fuel

1,176 1,276

1,000

Sour water
Sour naphtha

Pressure, psi
Normal

Relieving

978

965 1,062

Hot
high-pressure
separator

Cold
high-pressure
separator

Steam
Stripper

Wash water

Sour water

units normally have 114-chrome shells

and 400-series tubes. Hot shells are 114chrome with 300-series stainless
cladding and 300-series stainless tubes.

a HHPS to debottleneck an existing hydrotreater that is hydraulically limited

in the reactor loop. This revamp would
require, however, a modest increase in
inlet flow and power for the recycle
compressor.
Charge heater
Other options include adding hot
The feed heaters normal duty
and cold low-pressure separators. These
should not be less than 20% of the
combined feed duty to the reactors in- designs marginally unload the upper
part of the stripper and improve LPG
let temperature at end-of-cycle condirecovery.
tions; this allows for turndown and
HHPS separator base material is 114heat balance control. A design margin
of 10% of the combined feed exchang- chrome 12-molybdenum, normally with
er duty accommodates accelerated exa clad or overlay lining, which is 300
changer fouling.
or 400-series stainless depending on
The number of heater passes should the actual predicted corrosion rate.
match the number of parallel feed-efThe CHPS is carbon steel resistant to
fluent exchanger trains. The refiner can hydrogen-induced cracking. A 300-seflow control feed and recycle hydrogen ries stainless steel mesh pad helps coafor each train upstream of the exchang- lesce the wash water. The boot is also
ers. This prevents maldistribution of the hydrogen-induced-cracking resistant
two-phase stream entering the heater
steel with a 14-in. corrosion allowance.
passes.
The limit for one heater pass is
Reactor products condenser
15,000-18,000 b/sd of fresh feed at
A final condenser provides cooling
typical diesel hydrotreater conditions.
before liquid and vapor separation reThis limit is based on a maximum 8-in. gardless of the reactor product separadiameter heater tube and an overall
tor configuration. An air cooler is typipressure drop of 40-50 psig.
cally used, sometimes supplemented
Heater tube metallurgy is typically
with a water-trim cooler.
Feed exchangers
Piping and equipment metallurgy in
Feed exchangers are generally speci- 347 stainless steel.
this area requires special attention due
fied to optimize heat economy; howevto the presence of ammonia and H2S in
er, enough duty should be allowed so
Reactor-product separation
that the charge heater has a reasonable
Several alternative arrangements can the reactor effluent. When this stream
amount of turndown and overall heat- handle reactor effluent after the comcools, these two compounds combine
balance control.
to form ammonium hydrosulfide,
bined feed exchangers.
Feed exchange should not be more
which condenses as a solid on inlet
In general, refiners must decide
than 80-85% of the total duty needed
whether to use a HHPS in addition to a piping and in cooler tubes.
to heat the feed to the reactor tempera- CHPS, or a CHPS only. The HHPS usualRecirculating wash water, injected
ture at end-of-cycle conditions.
before the cooler, introduces a corroly operates around 500-550 F. and,
The reactor feed-effluent exchangers therefore, the vapors can heat recycle
sive aqueous solution in the stream to
mechanical design requires minimal
prevent plugging and under-deposit
gas and cold separator liquid.
leakage. Many refiners prefer a pullA HHPS improves overall heat econ- corrosion.
through tube bundle with a floating
Tube metallurgy selection is based
omy and results in smaller product
head because it is easier to clean; how- stripper and auxiliary equipment. This
on ammonia and H2S in the effluent.
ever, the floating head cover is a poten- arrangement also improves oil-water
The designer should ensure a target
tial leak source. A U-tube bundle avoids separation in the CHPS.
maximum value for Kp of 0.15 before
this potential problem and is less exFor heavy gas oil hydrotreating, a
choosing alloy metallurgy, where:9
pensive.
hot separator is necessary for adequate
Kp = mole % H2S x mole % NH3
Tubes should be seal welded to the
oil-water separation. This is less of a
The designer should also:
tube sheet to prevent leaks from the
concern for diesel hydrotreating unless
Determine makeup water quantity
rolled tube joints.
there are significant quantities of
to limit ammonium hydrosulfide conMaterial selection for feed-effluent
cracked materials, which decrease API
centration in the condensed water
exchangers is based on predicted corro- gravity.
phase to 4 wt %. One must assume
sion rates. Feed-effluent exchangers
Disadvantages of this design are a 5- equal molar amounts of ammonia and
have several shells in each train with
10% lower recycle gas purity that reH2S in the CHPS water and that essenvarying metallurgies. Cold shells are all quires more horsepower in the recycle tially all ammonia dissolves in the wacarbon steel. Intermediate-temperature hydrogen compressor. Refiners can add ter.

Pressure drops for the feed distributor and redistributors, quench internals, and collector are added to the bed
values to provide the overall fouled-reactor pressure drop. For ULSD, the feed
distributor and redistributor design is
crucial for obtaining the target product
sulfur level.
Several designs provide internal reactor quench to limit the temperature
rise. The normal method is to use hydrogen-rich recycle gas. Quench gas also causes liquid vaporization in the reactor, which provides an extra heat sink
for heat removal.
Another less-common approach is to
recycle CHPS liquid to the reactor. The
quench limits the reactor bed temperature rise to 40-50 F.
Reactor metallurgy is typically 114chrome, 12-molybdenum with 18-in. 347
stainless overlay or cladding. The metallurgy choice often depends on the vessel thickness, which affects the cost.
Cladding is usually more economical
for vessels with a wall thickness up to
about 4 in. An overlay is normally used
with thicker walls.

PROCESSING
weight. This is a potential problem for a
Table 2 shows that the assumed
centrifugal recycle compressor because pressure profile is consistent with 900the lower molecular weight requires
psi flanges in the reactor loops alloy
more head and potentially more stages. piping. The relieving pressure profile
This is not as serious for a reciprosets the design pressure for equipment
cating machine; however, higher hydro- in the reactor loop.
gen purity increases the discharge temEarly in the design process, the deperature, which is limited to 275 F. per signer develops a sized equipment list
API-618.
that allows for plot plan studies. The
Makeup gas normally feeds to the
designer checks preliminary reactorrecycle compressor
discharge unless it P RESSURE ALLOWANCES
Table 1
needs an additionEquipment
item
Pressure
drop,
psi
al stage of makeup
Combined feed exchangers, both sides
40
compression. This
Feed heater
50
has the advantage
Reactor, fouled
75
HHPS
vapor-recycle
hydrogen,
both
sides
20
of maintaining
HHPS vapor-CHPS liquid
10
some hydrogen
Effluent air condenser
10
Amine
system,
knock
out
drums
5
flow through the
Piping
25

reactor loop in
Total reactor loop
235
case of a recycle
compressor emergency outage.
loop pipe sketches and rechecks allowances for piping pressure drops before specifying the recycle compressor.
Reactor-loop hydraulics
The designer checks reactor-loop
A hydraulic profile for the reactor
hydraulics a final time after receiving
loop provides the recycle compressor
equipment vendor information and afhead requirements and establishes the
design pressure of piping and equipment. ter completing piping isometrics for
Reactor inlet pressure is set based on the reactor loop.
Compression
We established design pressures in
conditions that the catalyst supplier esInlet flow and head requirements
the reactor loop assuming the CHPS retablishes, and the pressure limits of
determine the recycle compressor delief valve set pressure is 10% higher than
sign. Adequate suction flow is required standard piping flanges.
the normal operating pressure. The reTwo case studies are presented in
for a centrifugal compressor, and the
head should be low enough to limit the Part 2 for reactor inlet pressures of 800 finer can set this margin at 5% with a
pilot-operated relief valve, which will
and 1,100 psig.
number of stages so that only a single
Table 1 shows a typical allowance for lower the reactor loops design pressures.
body machine is needed; this is usually
A 10% margin in a grassroots facility
reactor-loop equipment and piping
a maximum of 10 stages.
allows for future capacity increases
In the past, most diesel hydrotreaters pressure drop for the 1,100-psig case.
Fig. 2 shows a point-to-point pres- while staying within equipment design
had an inadequate volume of recycle
sure profile for normal operations and pressure limits. API Recommended
gas for a centrifugal compressor. New,
Practice 521, Appendix G recommends
larger ULSD hydrotreaters that circulate for the relieving case in which the
a design pressure of 105% of the setCHPS is 10% greater than the operat3,000-5,000 cu ft of gas/bbl of feed
tling-out pressure, which would make
are candidates for centrifugal compres- ing pressure.
the design pressure about 115% of the
Table 2 shows maximum pressures
sors.
normal operating pressure.
A reciprocating compressor used for at relieving conditions and typical deWhile the designer is establishing
recycle normally also provides makeup sign temperatures for critical piping
equipment design conditions, insuffiservices in the reactor section comservice in a multiple-throw machine.
pared to a maximum allowable pressure cient information exists to calculate the
For reciprocating compressors greater
for 900-psi, type-321 stainless flanges. settling-out pressure accurately. When
than 500 hp, two 60% capacity machines are typically used. For smaller
compressors, two 100% units are used.
PIPING CONDITIONS
Table 2
Makeup compressor discharge feeds
to the recycle compressor suction or
Pipe section
Design pressure,
Design temperature,
Maximum allowable
psi
F.
pressure, psi
discharge.
Exchanger to heater
1,252
700
1,260
Makeup gas flowing to the suction
Heater to reactor
1,200
775
1,242
increases recycle gas purity and volReactor to exchanger
1,123
825
1,233
ume, and decreases the molecular
Determine the quantity of circulating water to ensure that 25% of the
total water injected in the liquid phase
is upstream of the condenser.
Limit the mixed-phase velocities
in the piping and tubes to 20 fps for
carbon steel and 30 fps for alloys. The
minimum velocity should be 10 fps.
Provide symmetrical piping into
and out of the condenser.
Ensure that the wash water source
is a stripper dedicated to hydrotreating
spent wash water if stripped sour water
is used.
Ensure that wash water is injected
into the main effluent line upstream of
the condenser, or independently into
each air cooler nozzle through a distribution device like a spray nozzle or restriction orifice. Injecting into piping
provides more time for mixing and vaporizing the water and is less expensive. Injection into individual air condenser nozzles ensures good distribution in each air cooler bay.
Typical alloys, if required, are Incoloy 800 or 825, and duplex stainless
steel 2205.

the heat and material balances and

process flow diagram are completed,
however, one can use preliminary
equipment sizing information, assumptions about pipe quantities, and the reactor loops pressure profile to estimate
the settling-out pressure.
When actual equipment and piping
design information is available, one can
re-estimate the settling-out pressure to
determine the margin available compared to the CHPS relief valve set point.
If the settling-out pressure exceeds the
set pressure, a stepwise calculation or
dynamic simulation can estimate the
relief rate.

Product stripper
The product stripper removes H2S
and light hydrocarbons from the ULSD
product. Light material removal must
be high enough to meet a flash-point
specification of 140 F. This ensures removal of nearly all H2S.
The stripper requires about 30 trays.
Overhead products include sour gas, unstabilized naphtha, and sour water.The
design also includes equipment for intermittently water washing the condenser.
Steam or vapors from a fired reboiler
serve as the stripping medium. Due to
temperature constraints of about 700
F. maximum, the stripper operating
pressure is limited to 40-50 psig for a
fixed reboiler.
Unless the refinery has a low-pressure gas recovery system, a small compressor and spare are needed to handle
the offgas.
A steam stripper can operate at pressures greater than 100 psig, which allows routing into an existing refinery
sour fuel gas system.
Diesel product from a steam stripper
must have all water removed. Vacuum
drying or coalescing followed by salt
drying can accomplish the water removal. If this type of system already exists, the steam-stripping option has a
lower capital cost vs. a fired reboiler. A
steam-stripped tower is also a little
smaller than a reboiled tower.
If the product drying system is not
in place, the refiner should consider the
fired reboiler option. It will also create
less sour water to process.
Product stripper material is killed carbon steel with 410 stainless steel trays

The authors
Leslie J. Harwell is a senior
consulting process engineer at
Mustang Engineers & Constructors LP, Houston. He is
primarily involved in ultra-low
sulfur projects and midstream
Feed filtration
gas processing. Harwell has also
Although distillate stocks are relatively clean, feed filtration is important worked for Fish Engineering
and Litwin Engineering & Construction Inc.
to mitigate exchanger and reactor plug- Corp.
He holds a BS (1967) in chemical engineering
ging. A cartridge filter with 25-
m re- from the University of Texas, Austin. Harwell is a
tention is typical for this application.
registered professional engineer in Texas and a
Cracked feeds should feed the hymember of the Gas Processors Suppliers Associadrotreater hot from the upstream facili- tion.
ties or from gas blanketed storage.
Shrikant (Sam) Thakkar is a
senior specialist for ChevronReferences
Texaco Corp., Bellaire,Tex. He
1. Palmer, R.E., Ripperger, G.L.,
previously worked for Mustang
Migliavacca, J.M., Revamp your hyEngineers & Contractors LP
drotreater to manufacture ultra low sul(when this article was written).
fur diesel fuel, presented to the 2001
He holds a BS (1981) in
chemical engineering from
NPRA Annual Meeting, Mar. 18-20,
Bombay University.
New Orleans, paper AM-01-22.
2. Leliveld, B., Mayo, S., Miyauchi,Y.,
and Plantenga, F., Elegant solutions for Stan Polcar is a principal
engineer for Mustang
ultra low sulfur diesel, presented to the process
Engineers & Constructors LP.
2001 NPRA Annual Meeting, Mar. 18He has also worked for Litwin
20, New Orleans, paper AM-01-09.
Engineering & Construction
3. Brevoord, E., Gudde, N., Hoekstre, Inc. and The Pritchard Corp.
Polcar holds a BSc (1968) in
G., Mayo, S., and Plantenga, F., ULSD
chemical engineering from Case
real life: Commercial performance of
Stars and Nebula technology, present- Western Reserve University, Cleveland. He is a reged to the 2002 NPRA Annual Meeting, istered chemical engineer in Texas.
Mar. 17-19, 2002, San Antonio, paper
R.E. Palmer is the manager of
AM-02-38.
downstream process engineering
4. Landue, M.V., Catalysis Today, Vol.
for Mustang Engineers & Con36 (1997), pp. 393-429.
structors LP. He is responsible
for process design and market5. Piehl, R.L., Refiners tame effluing support for all refining,
ent air-cooler corrosion, OGJ, Aug. 18,
petrochemical, and chemical
1975, pp. 119-20.
projects. He has led numerous
6.Torrisi, S., Proven best practices for
studies,
technology
evaluations, and projects relatULSD production, presented to the
ing to clean fuels production. Palmer previously
2002 NPRA Annual Meeting, Mar. 17-19, spent 23 years with Litwin Engineers & Con2002, San Antonio, paper AM-02-35.
structors Inc. and 5 years with Conoco Inc. He
7. Schmidt, M., Premium perform- has a BS in chemical engineering from the Uniance hydrotreating with Axens HR 400 versity of Missouri, Rolla.
series hydrotreating catalysts, presented to the 2002 NPRA Annual Meeting, Pankaj H. Desai is the new
business development manager
Mar. 17-19, 2002, San Antonio, paper
for Akzo Nobel Catalysts LLC,
AM-02-57.
Houston. He joined Akzo Nobel
8. Larkins, R.P., White, R.R., and Jef- in 1980 and has held various
fery, D.W., Two Phase Concurrent Flow positions in hydroprocessing
in Packed Beds, AICHE Journal, June
and FCC catalyst research and
development. Desai holds a
1961, pp. 231-39.
BTech degree (1974) in chemical engineering
9. Knudsen, K.G, Cooper, B.H., and
Topsoe, H., Catalyst and Process Tech- from Indian Institute of Technology, Kanpur, and a
PhD in chemical engineering from the University
nologies for Ultra Low Sulfur Diesel,
of Houston.
Applied Catalysis A: General, Vol. 189
(1999), pp. 205-15.

and caps. The overhead condenser and

receiver are hydrogen-induced-cracking
resistant steel. The overhead receiver
boot has a 14-in. corrosion allowance.

PROCESSING
ULSD HYDROTREATING
Conclusion

A recent study for a

grassroots 30,000-b/sd
hydrotreater that can produce ultra-low sulfur
diesel (ULSD) helped assess the critical process
Refining
design issues for meeting
future ULSD requirements.
Part 1 (OGJ, July 28, 2003, p. 50) of
this two-part series discussed alterna-

Study identifies optimum operating

conditions for ULSD hydrotreaters
Les Harwell
Sam Thakkar
Stan Polcar
R.E. Palmer
Mustang Engineers & Constructors LP
Houston
Pankaj H. Desai
Akzo Nobel
Catalysts LLC
Houston

tive process configurations for ULSD

hydrotreaters with a discussion of key
design parameters and metallurgical
considerations for major pieces of
equipment. Engineering aspects of hydrotreater design were covered, includBased on a presentation to the 2003 NPRA Annual Meeting, Mar. 23-25, 2003, San Antonio.

ing a suitable operating pressure level

that satisfies reaction conditions and
the practical limits of piping mechanical design.
Part 2 covers process simulations for
a number of process configurations for
reactor inlet pressures from 800 to
1,100 psi. We estimated capital costs
and life-cycle costs for three of the
most promising cases.
The more important conclusions of
this study are:
A grassroots hydrotreater for ULSD
can have a reactor operating pressure as
low as 800 psi if high-purity makeup
hydrogen is used. This approach lacks
the robustness of a 1,100-psi facility
especially in its ability to handle upsets.
This disadvantage, which is difficult to
quantify, has estimated capital and a
life-cycle cost advantages of 10% and
12%, respectively.
A good feedstock characterization
including off-design variations is essential for proper selection of catalyst, reaction conditions, and process configuration. We recommend pilot testing.
Capital and operating costs are

ULSD HYDROTREATER, CHPS ONLY

Fig. 1

Makeup H2
LCO

Feed heater
Makeup H2
compression

Straight-run
diesel
Reactor
Recycle
compressor
Quench

Amine
treating

Feed
ULSD product

Gas to fuel

Cold highpressure
separator

Sour water
Sour naphtha
Steam

Wash water

Sour water

and, therefore, increases the prelower for a two-separator F EED PROPERTIES

Table 1
dicted cycle length for a fixed
systemhot high-presStraight-run
quantity of catalyst. Table 3
sure separator (HHPS)
Property
gas oil
Light cycle oil
Blend
shows the results with a conand cold high-pressure
Flow rate, 1,000 b/sd
20
10
30
Gravity, API
34.5
20.0
29.4
stant cycle length of 23-25
separator (CHPS)comSulfur, wt %
0.8
2.0
1.2
months and an LHSV that
pared to a single CHPS
N, ppm (wt)
100
500
242
Total aromatics, vol %
20.0
65.0
35.0
changes according to reactor indesign.
Monoaromatics, vol %
12.8
57.2
27.6
let pressure. As the reactor inlet
Minimal chemical
Polyaromatics, vol %
7.2
7.8
7.4
Bromine number
1
10
4.2
pressure increases from 700 to
hydrogen consumption,
D86 90 vol % distillation
temperature, F.
640
640
640
1,100 psi, LHSV increases 35%,
in conjunction with a
Cetane index
49.9
30.3
42.6
which means that less catalyst is
process design that elimineeded.
nates reactor loop purge
For both cases, the makeup
and minimizes solution
ULSD REACTION CONDITIONS, CONSTANT LHSV
Table 2
hydrogen is 85 mole %. Chemilosses, can significantly
Reactor inlet pressure, psig 700
900
1,100
cal hydrogen consumption only
reduce the projects lifeStart
of
run
WABT,
F.
Base
Base

7
F.
Base
13 F.
increases 14-15% for the two
cycle cost.
Chemical hydrogen
extremes in reactor inlet pres Operating cost is the
consumption, cu ft/bbl
Base
Base x 1.075
Base x 1.14
Cycle length, months
20-22
29-31
38-40
sure. This increase, however, can
major component in unit
Product sulfur, ppm (wt)
10
10
10
Product nitrogen, ppm (wt)
<10
<10
<10
have a significant impact on the
life-cycle cost.
Yields, wt %
projects life-cycle cost, especial At different pressures
Start of run Feed
100.00
100.00
100.00
C4 and lighter 0.07
0.07
0.07
ly if the refinery is hydrogen
(700-1,100 psig) and
Naphtha
3.32
3.27
3.24
limited.
feed and process condiDistillate
96.04
96.15
96.22
End of run
Feed
100.00
100.00
100.00
For ULSD applications, the
tions, the increase in
C4 and lighter 0.17
0.17
0.16
Naphtha
3.54
3.46
3.41
quantity of cracked stocks in the
chemical hydrogen conDistillate
95.78
95.97
96.11
feed strongly influences chemisumption is a modest
cal hydrogen consumption. Hy15%. For ULSD designs,
drogen solution loss (nonchemhydrogen consumption is ULSD REACTION CONDITIONS, CONSTANT CYCLE LENGTH Table 3
ical consumption) increases
mainly a function of feed
Reactor inlet pressure, psig 700
900
1,100
with pressure.
characteristics, particularly
Reactor LHSV, hr1
Base
Base x 1.20
Base x 1.35
Tables 2 and 3 LHSVs were
aromatics content, and
Start of run WABT, F.
Base
Base + 5 F.
Base + 7 F.
1.0-1.75 hr1, which is rather
unit operating pressure.
Chemical hydrogen
consumption, cu ft/bbl
Base
Base x 1.08
Base x 1.15
Makeup hydrogen
modest with a feed that contains
Cycle length, months
23-25
23-25
23-25
Product
sulfur,
ppm
(wt)
10
10
10
purity has a major impact
a substantial LCO fraction. This
Product nitrogen, ppm (wt)
<10
<10
<10
on the recycle gas comis due to a somewhat low feedYields, wt %
Start of run Feed
100.00
100.00
100.00
pressor. For a makeup hynitrogen content, a processing
C4 and lighter 0.07
0.07
0.07
drogen purity of 85 mole
objective of lower product sulNaphtha
3.30
3.29
3.28
Distillate
96.05
96.12
96.18
%, the centrifugal recycle
fur only and not cetane imEnd of run
Feed
100.00
100.00
100.00
C4 and lighter 0.17
0.16
0.17
compressor requires two
provement, and the application
Naphtha
3.53
3.46
3.42
or three stages. For highof new generation, high-activity
Distillate
95.8
96.0
96.1
purity hydrogen makeup,
catalysts.
eight or nine stages are required. It is
If the feed contains more nitrogen,
Tables 2 and 3 show various reactor
impractical to design for this range of
which is common with coker light gas
parameters and product yields and
purities.
oil, the volume of catalyst required to
qualities using Akzo Nobel Catalysts
Recycle gas purification via recon- LLCs Stars KF-848 high-activity, nickel- meet the target sulfur specification intact with liquid from a cold low-presmoly catalyst for different reactor inlet creases significantly. Also, changing feed
sure separator (CLPS) is not justified
sulfur quantities and sulfur species
pressures and hydrogen partial presbased on energy savings. For an existmight move the design to a more consures at the reactor outlet. The catalyst
ing unit, however, higher-purity recycle is dense loaded into the reactor.
servative LHSV.
gas can extend cycle length or help
We further analyzed reactor yields
Table 2 shows the different cycle
meet the treating specification.
lengths at each pressure level to obtain and process conditions for a LHSV of
1.25 hr1 and reactor inlet pressures of
a 10-ppm (wt) sulfur product given a
constant liquid hourly space velocity
800 and 1,100 psi. We also considered
Case studies
(LHSV). Cycle length varies from 20-22 other design options, including:
Table 1 shows the feed characterismonths at a 700-psig reactor pressure
A two-separator system with an
tics used for this study. The feed is a
HHPS and CHPS.
blend of two-thirds straight-run diesel to 33-35 months at 1,100 psig.
The higher hydrogen partial pres A single-separator system with a
and one-third FCC light cycle oil
(LCO); the combined nitrogen content sure corresponding to each reactor inlet CHPS only.
pressure decreases catalyst deactivation
An HHPS-CHPS system that enis 242 ppm (wt).

PROCESSING
the various options. For
riches recycle hydrogen
each case, the LHSV and
using a CLPS and recon- M AKEUP HYDROGEN Table 4
product sulfur content
tacts the liquid with re- COMPOSITION
were constant. We set
actor effluent that feeds
Component
Fraction, vol %
the recycle hydrogen
the products condenser.
Hydrogen
85.00
Nitrogen
1.00
rate to achieve a ratio of
High-purity hyMethane
3.21
total hydrogen to the
drogen makeup for the
Ethane
3.10
Propane
3.80
reactor to chemical hytwo-separator case.
i-Butane
1.27
drogen consumption of
High-purity hyn-Butane
1.10
i-Pentane
0.64
5.5 for the 1,100-psi
drogen makeup for the
n-Pentane
0.32
C6+
0.56
case, and 6.0 for the
single separator case.
Total
100.00
800-psi case.
Table 4 shows the
Based on the calculow-purity makeup hylated reactor outlet hydrogen partial
drogen composition. We assumed that
high-purity makeup gas was 99.9 mole pressure, Akzo Nobel estimated the cycle length, chemical hydrogen con% hydrogen.
The low-purity makeup is typical of sumption, and weighted average bed
a blend of hydrogen from a catalytic re- temperature to calculate the final reactor yields and other process conditions.
former and a high-purity source, such
as membrane purification or purchased
hydrogen.
Case descriptions
Table 5 shows the results of simulaWe used the simulation results to
tions for the 13 cases corresponding to prepare heat and material balances,

equipment sizes, and design characteristics for three cases:

Case 1 base case: 1,100-psi reactor
inlet with a HHPS, CLPS, and 85-mole
% makeup hydrogen purity.
Case 2: 1,100-psi separator with a
CHPS only and 85-mole % makeup hydrogen purity.
Case 13: 800-psi reactor inlet with
a HHPS, CLPS, and 99.9-mole % makeup hydrogen purity.
We specified a single, three-bed reactor with inter-bed hydrogen quench
for all the cases. We used a centrifugal
compressor for recycle hydrogen and
two-stage, reciprocating compressor
with a spare for makeup. The charge
heater has two passes and steam generation in the convection section. The
product stripper uses steam for stripping rather than a fired reboiler.
We prepared capital cost estimates
for each of the three cases and estimat-

SIMULATION RESULTS
Case

Table 5

Reactor inlet
pressure, psig
1,100
1,100
Makeup H2
purity, mole %
85
85
Number of
separators
2
1
Cold high-pressure separator
Temperature, F.
110
110
Pressure, psig
965
993
H2 partial pressure
at reactor outlet,
psia
612
674
Cycle length,
months
33-35
36-38
Chemical H2
consumption,
cu ft/bbl
400-500 400-500
Makeup H2
flow, MMcfd
19.4
18.8
H2 at reactor inlet:
Chemical H2
consumption, ratio
5.5
5.5
Total H2 at reactor
inlet, cu ft/bbl
2,673
2,673
H2 purity at reactor
inlet, mole %
70.2
76.5
Purge, MMcfd
1
1
Enrichment
liquid, b/sd
0
0
Recycle gas
Flow rate, MMcfd 114.5
104.4
H2 purity, mole %
67.0
75.0
Molecular weight
8.8
6.6
Recycle gas compressor
Normal bhp
1,902
1,741
Number of stages
2
3
Stripper
Offgas flow, MMcfd 4.0
3.5
Overhead liquid
flow, b/sd
1,117
1,140
Overhead liquid
cut point, F.
350
350
ULSD product
Flow, b/sd
29,841
29,856
Gravity, API
33.0
33.1
Cetane index
45.5
45.5
Sulfur, ppm (wt)
10
10
Nitrogen, ppm (wt) <10
<10

10

11

1,100

1,100

1,100

1,100

1,100

85

85

85

99.9

110
965

110
965

110
965

110
993

12

13

800

800

800

800

800

800

99.9

85

85

85

85

99.9

99.9

110
965

110
693

110
693

110
693

110
665

110
693

110
665

644

661

687

898

893

448

492

508

408

612

611

34-36

36-38

37-39

48-50

47-49

18-20

22-24

22-24

<20

33-35

33-35

400-500

400-500

400-500

400-500

400-500

400-500

400-500

400-500

400-500

19.8

19.8

19.8

15.3

15.3

17.0

20.9

25.9

17.3

13.8

14.5

5.5

5.5

5.5

5.5

5.5

6.0

6.0

6.0

6.0

6.0

6.0

2,673

2,673

2,668

2,668

2,668

2,622

2,622

2,622

2,622

2,622

2,622

73.5
1

75.2
1

77.9
1

98.5
0

98.1
0

72.2
1

79.9
5

82.7
10

65.2
1

97.9
0

98.0
0

3,126

6,117

11,756

109.0
71.0
7.9

106.6
73.0
7.4

103.0
76.0
6.7

81.3
98.1
2.3

81.6
97.6
2.5

109.5
69.4
7.5

98.0
79.0
6.3

95.0
82.0
5.9

122.1
61.0
9.7

80.4
97.4
2.4

80.5
97.3
2.6

1,769
3

1,502
2

1,725
3

1,368
9

1,382
8

2,503
4

2,311
5

2,238
5

2,760
3

1,405
9

1,787
8

3.62

3.4

3.1

1.3

2.1

2.9

2.7

2.7

3.0

0.9

1.4

1,080

1,046

1,000

942

971

1,131

1,182

1,226

1,101

969

995

350

350

350

350

350

350

350

350

350

350

350

29,821
33.1
45.5
10
<10

29,814
33.1
45.5
10
<10

29,810
33.1
45.5
10
<10

29,919
33.1
45.5
10
<10

29,859
33.1
45.5
10
<10

29,834
32.5
44.7
10
<10

29,829
32.5
44.7
10
<10

29,824
32.5
44.7
10
<10

29,837
32.5
44.7
10
<10

29,885
32.6
44.8
10
<10

29,882
32.6
44.8
10
<10

400-500 400-500

the ULSD product L IFE-CYCLE COSTS

Table 7
is steam stripped,
Case
1
2
13
Case
1
2
however, Case 2
Reactor pressure, psi
1,100
1,100
800
Reactor inlet pressure,
requires significant
Arrangement
2 separators 1 separator 2 separators
psig
1,100
1,100
additional heat exHHPS
Yes
No
Utility consumption
CLPS
Yes
Yes
change to raise the
Power, kw
3,500
3,418
2,666
Recycle gas flow rate,
Medium pressure steam, lb/hr
2,770
-9,000
4,400
CHPS liquid temcu ft/bbl
3,803
3,493
Boiler feedwater, lb/hr
8,000
20,890
6,160
Recycle gas flow rate,
Fuel, MMbtu/hr
33.8
70.2
25.3
perature
for
effecMMcfd
114.5
104.4
Cooling water, gpm
1,354
1,500
1,000
Recycle gas horsepower
1,902
1,741
tive H2S removal.
Total power consumption,
First year costs, million $
This practically
kw-hr
3,500
3,418
Utilities
2.3
3.0
1.8
Charge heater fired duty,
doubles the charge
Maintenance1
1.0
1.0
0.8
MMbtu/hr
33.8
70.2
2
Catalyst, chemicals
14.7
14.2
13.0
Investment, million $
62.9
71.1
heaters size,
Capital costs, million $3
62.9
71.1
56.7
which increases
Discounted operating costs4
244.4
256.4
214.2
the size of the fied maintenance costs and the conLife-cycle costs
314.9
336.5
278.1
nal effluent air
sumption of utilities, catalysts, and
1
Turnaround costs every 4 years are included in the life-cycle cost but not annual
chemicals. We discounted the operating cooler.
maintenance costs. 2Catalyst and chemicals include hydrogen costs. 3Capital cost includes initial supplies of catalysts, chemicals, and spare parts. 4Discounted value of
Table 6 summacosts over 20 years and added them to
operating costs based on 15% interest rate and 20-year project life.
rizes the results of
the investment to calculate the overall
ments, horsepower savings allow an inthis analysis. Case 1 has an investment
life-cycle cost for each case. The discount was based on an assumed corpo- and operating cost advantage vs. Case 2, vestment of around $110,000 based on
a 3-year simple payout. The equipment
at least for the design feed and other
rate hurdle rate of 15%.
required for this option may be more
Cases 1-5 are based on 1,100-psi re- factors we chose for this study.
Cases 3-5 include a CLPS and liquid expensive than the savings. This
actor inlet pressure and low-purity hyarrangement might make sense for a
recontact with HHPS vapor flowing to
drogen makeup. Case 1 has an HHPS
revamp in which the recycle compresfor the base option. Case 2 has a single the effluent condenser (Fig. 2). Each
case shows the impact of more enrich- sor is the bottleneck.
CHPS (Fig. 1) that results in a higherCases 6 and 7 use high-purity hypurity recycle gas and a corresponding- ment liquid on the recycle compressor
drogen for the CHPS, and HHPS-CHPS
ly lower horsepower for the recycle gas volume and power requirements.
arrangements at a 1,100-psi reactor
A comparison of Case 1 and Case 5
compressor. The HHPS design is more
outlet pressure.
shows a maximum advantage of 177
energy efficient, which results in a reThe reactor outlet hydrogen partial
duction of 36.4 MMbtu/hr of fired du- hp for the enrichment option. Including recycle liquid pump power adjust- pressure increases significantly with a
ty in the charge heater.
Estimated capital investments for Cases 1 and 2 are
Fig. 2
ULSD HYDROTREATER WITH HHPS, CHPS, CLPS
$62.9 million and $71.1 milMakeup H2
lion, respectively. These costs
LCO
are for inside battery limits
Feed heater
Makeup H2
only and include the initial
compression
Straight-run
load of catalyst, chemicals, and
diesel
spare parts.
Reactor
Construction labor rates
Recycle
and productivity are based on
compressor
Amine
Quench
a US Gulf Coast location. We
treating
prepared these estimates by
obtaining equipment pricing
Feed
and then applying appropriate
ULSD product
factors to get a total installed
cost. One should use these figures carefully because actual
Gas
Cold lowfacility location, owners costs,
pressure
Gas to fuel
separator
and design preferences can
substantially affect final cost.
Cold highHot highpressure
Intuitively, Case 1 with two
pressure
separator
separators and an additional
separator
Sour water
HHPS vapor-recycle hydrogen
heat exchanger should require
Sour naphtha
Steam
more capital than the single
Sour water
Wash water
CHPS case (Case 2). Because

HHPS ADVANTAGES

Table 6

PROCESSING
ULSD 20-YEAR COSTS
Life-cycle cost, million $

400
350
300
250
200
150
100
50
0
Case 1

Case 2

ULSD 20-YEAR OPERATING COSTS

160

Life-cycle cost, million $

140

Case 1
Case 2

120

Case 13

100
80
60
40
20
0
Utilities

charge duty, produces more

medium-pressure steam than
is required for the product
striper.
Overall, the 800-psi case
ISBL capital cost
with two separators has the
Operating costs,
lowest life-cycle costs, which
discounted to
we expected; however, this
net present value
case barely meets the miniTotal life-cycle cost
mum hydrogen partial pres(operating and capital)
sure target even with highpurity makeup hydrogen.
Case 13
Many refiners installed
grassroots hydrotreaters, with
reactor inlet operating pressures of
Fig. 4
around 800 psi, to satisfy the 1993
low-sulfur diesel regulations of 500ppm (wt) sulfur. Owners will revamp
most of these facilities to produce
ULSD even though the hydrogen partial
pressures are significantly less than the
minimum recommended for grassroots
ULSD hydrotreaters. This will require
proportionately more catalyst and perhaps shorter cycle lengths.
For the 1,100-psi cases, the two-separator arrangement has the lowest lifecycle cost.
Fig. 3 shows that operating costs,
Catalyst
Chemicals
even when discounted over the proincluding H2
jects life, are the major contributor to
total life-cycle cost for the three cases.
even lower than the CHPS-only cases.
This stresses the importance of an optiCases 12 and 13 use an 800-psi remum process design and minimized
actor outlet pressure with high-purity
hydrogen makeup, and single and two- utility and maintenance costs. Choosing
separator options. The hydrogen partial the right reactor operating conditions
also can reduce chemical hydrogen
pressure is just greater than the miniconsumption.
mum. Estimated US Gulf Coast capital
Fig. 4 compares the operating cost
cost for Case 13 is $56.7 million.
components. Chemical hydrogen consumption has a larger impact than othLife-cycle cost
We prepared a life-cycle cost analysis er operating costs.
A process design that minimizes or
for Cases 1, 2, and 13.
eliminates the need for a purge stream
For this analysis, life-cycle costs are
the sum of the capital investmentin- from the recycle gas results in less
makeup hydrogen needed and a lower
cluding initial catalyst, chemicals, and
life-cycle cost. One should also considspare partsand the present value of
operating costs during the projects life, er the economics of hydrotreating incrementally cracked stocks for ULSD vs.
excluding operating labor, taxes, and
other uses such as heating oil, cutter
insurance.
We used a discount rate based on an stock, or fuel.
The hydrogen cost in Fig. 4 is
assumed corporate hurdle rate of 15%
and a 20-year project life for this calcu- $2.40/Mcf. We assumed that the refinlation. Table 7 shows the cost results in- ery is hydrogen limited and any recluding estimates for maintenance costs former hydrogen the refiner uses backs
out an equivalent quantity of purchased
and utility, catalyst, and chemical conmaterial.
sumption.
The Case 2 design, due to a higher
Fig. 3

Maintenance

corresponding improvement in cycle

length to 47-49 months from 33-35
months for Case 7 vs. Case 1.
Cases 6 and 7 have a compressor
horsepower almost 30% lower than
Case 1; however, due to a substantial
decrease in molecular weight, the compressor head more than doubles, which
necessitates an 8-stage centrifugal compressor vs. a 2-stage machine for Case 1.
Cases 8-13 have a reactor inlet pressure of 800 psi. Cases 8-11 use lowpurity makeup hydrogen.
The hydrogen partial pressure is 448
psia for Case 8, which is far less than
Akzo Nobels recommended minimum
of 600 psia for a grassroots ULSD hydrotreater.
For Cases 9 and 10, we increased the
reactor loop purge to 5 and 10 MMcfd,
respectively, to try to increase the hydrogen partial pressure. The partial
pressure in Case 10 improved to 508
psia at the high purge rate, still significantly below the minimum value.
Case 11 shows that, for a HHPS, the
hydrogen partial pressure of 408 psia is