Distillate hydrotreaters are large energy consumers. As more
stringent sulfur specifications are introduced, refiners must
increase energy usage for hydrotreating purposes. As much as
10% of a refinerys total energy consumption is attributed
to hydrotreating products. Furthermore, these units can be
quite energy inefficient. This case study examines the
economics of retrofitting distillate hydrotreaters options that
can improve energy efficiency: 1) increasing the surface area
of feed preheat exchangers, and 2) installing a hot separator.
The study also addresses the economics of using alternative
heat exchange techniques, such as twisted tubes, plate
exchangers and printed circuit exchangers.
Distillate hydrotreaters are one of the main energy
consumers within a refinery. New environmental rules mandate
even lower sulfur levels for transportation fuels. Thus,
refiners have few options but to increase hydrotreating
severity to remove sulfur-containing compounds from refined
products and product streams. Unfortunately, many existing
hydrotreating operations waste energy.
Room for improvement. The energy efficiency
of a process unit, or of an entire refinery, can be benchmarked and
compared against a selected reference value. Various methods
are used: some are statistical; some are based on historical
performance; and others apply a more solid engineering
Some options apply a best technology (BT) benchmark, whereby
the energy efficiency of an existing unit is compared with the
best design target.1 The ratio of the actual and the
targeted consumption yields a BT index for the process
When benchmarked against such targets, many hydrotreaters
show BT indices of 500%. In other words, these units consume
five times more energy than that of the best design unit. The
BT unit used for comparison would be one with the same
throughput, feed quality and severity of operation as the
Challenged energy performance.
The poor energy efficiency of an average hydrotreater is
primarily due to insufficient heat integration, resulting in the loss
of high-grade waste heat to water and air coolers. The
effectiveness of heat integration is normally assessed by using
pinch analysis techniques and its targeting
Fig. 1 shows a conventionally designed diesel hydrotreater.
In this particular case, the BT index was estimated at 550%.
Part of this inefficiency is immediately attributed to the
units electric power consumption, which is generated at
suboptimal total efficiency (24%). If an assumption is made
that all motive power is generated at the BT efficiency (80%),
so that the effect of using inefficiently generated power is
annulled, the remaining BT index would be 450%.
This unit is heat-integrated to some extent: the reactor
effluent preheats the gasoil (GO) feed in exchanger E-1, before
preheating the stripper feed in exchanger E-3. The unit is hot
fed at 160°C. If the unit were cold fed, there would
normally be another feed/effluent exchanger downstream of E-3
and before the air cooler.
In many existing hydrotreaters, the feed preheat temperature
is considered low, bearing in mind that it is preheated using
very hot reactor effluent. In the unit as shown in Fig. 1, the
feed is preheated to 313°C, although the reactor effluent
is available at 385°C. Surely, the approach between the two
temperatures could have been engineered to be lower, say to
40°C, instead of the actual 72°C.
Obviously, if the sizes of exchangers E-1 and E-3 are
increased, more heat could be transferred between the hot
stream (effluent) and the cold stream (feed). This would
increase the feed-preheat temperature and reduce the furnace
duty, at a constant furnace outlet temperature.
The other major inefficiency is the absence of a hot
separator. The effluent is condensed in the air cooler upstream
of the high-pressure and low-temperature (HPLT) gas/liquid
separator. This heatstill available at a high temperature
of 172°C, and quite valuableis lost to the
atmosphere. It would be more energy efficient to partly
condense the effluent at 172°C in a hot separator, send the
hot liquid directly to the stripper, only cool the remaining
gas against air and send the smaller stream to the existing
The problem of improving the energy efficiency in a
grassroots distillate hydrotreater was addressed
elsewhere.3 It is proposed that a hot separator,
combined with enhanced heat transfer between the feed and
effluent, can lower unit energy consumption to practically a BT
value. At such conditions, the feed would be at a temperature
high enough to be sent directly to the reactor. The furnace
would not be used in normal operation, as all remaining heat
would be supplied by the reaction exotherm. The proposed
enhanced heat transfer is to be accomplished by using
plate-type heat exchangers. Such a solution is feasible, and a
number of units have been designed, built and operated in this
Retrofits prove more difficult. However, in retrofit
situations, which are primarily considered in the present
analysis, it is observed that many hydrotreaters are too small
and are below the critical size for plate-type exchangers to be
retrofitted economically. The problem is further aggravated if
the unit is hot fed, which is otherwise a desirable feature of
energy efficient designs.
With all this in mind, there are two options, not mutually
exclusive, for improving the energy efficiency of an existing,
1. Add an exchanger area to E-1 and E-3 so that more
reactor-effluent heat can be recovered by the GO feed
2. Install a hot separator and modify the exchanger
network, as required.
As may be intuitively expected, the first option is likely
to be less capital intensive, but offering lower energy
savings. The presented study addresses the economics for both
options, and provides conclusions that may direct
Option 1: Increase area of E-1 and E-3.
Both exchangers E-1 and E-3 have large temperature
approaches (72°C and 60°C, respectively) and offer the
opportunity to economically install additional surface area.
Table 1 lists four projects to be evaluated.
Based purely on its simple payback, Project
Ainstalling twisted tubesappears to be the most
attractive. This project could on average save 5.2 GJ/h (from
start-of-run to end-of-run). Twisted tubes are easy to install.
They may cost a fraction more than a new shell, but they allow
duty increase without increasing systems pressure drop,
thus keeping the recycle compressors power unchanged.
Although the shortest of the four options listed in Table 1,
the payback for Project A is not very attractive at 4.5
yearsa return that may not be justifiable unless other
benefits may result.
The cost to add the first new shell (Project B) is high due
to extensive piping modifications. However, once the first
shell is installed, the incremental cost of adding more
exchanger area is reduced. Following this logic, Projects C and
D have been consideredadding 2 and 3 new shells,
respectively. At 4.6 years, Project C shows a slightly improved
payback time than Project B. The payback on the incremental
area is 4.3 years.
As expected, each additional new shell recovers less heat,
as the temperature driving forces in the exchanger are reduced.
Project D (a third new shell in each exchanger) only recovers
an additional 1.3 GJ/h, while the savings are mostly outweighed
by the increase in compressor duty. Project D, alone, has an
incremental payback of 14 years.
The fact that these modified exchangers show a temperature
cross, and have relatively large duties with (now) tighter
temperature approaches, indicates a possible application of
plate-type or similar exchanger designs.
Option 2: Install hot HP separator.
An alternative approach to saving energy is installing a
second separator. The bulk of the liquid downstream of
exchanger E-1 would be sent directly to the stripper, rather
than being cooled to 40°C and then reheated in the reactor
effluent/stripper feed exchanger E-3.
In the example hydrotreater (Fig. 1), this option is
particularly attractive as the pinch occurs precisely in
exchanger E-3. A hot separator eliminates the need for this
exchanger, thus larger energy savings are possible than what is
achievable by adding surface area alone (as in Option 1).
Fig. 1. Flow diagram of a
Since E-3 becomes redundant, it can be reused as an
additional feed/effluent exchanger. There are no other changes
in the exchanger network. The stripper bottom stream continues
to preheat the stripper feed from the cold separator and the
treat gas. If the unit were cold fed, there would be an option
to use more of the stripper bottoms heat and perhaps the gas
stream from the hot separator to preheat the cold feed. The
proposed revised flowsheet is shown in Fig. 2.
Fig. 2. Flow diagram of
the original and retrofit for distillate
Table 2 summarizes the economics of two available hot separator
options. The first option includes installing a new separator,
and reusing E-3 as a further reactor feed/effluent exchanger
(this includes replacing the shells for an increased pressure
rating). The second option considers the incremental benefit of
adding more area to E-3 and adding a recycle-gas heater to
recover additional energy.
The investment cost associated with installing a hot
separator is higher than the cost of simply adding exchanger
area to the feed-preheat train. However, the savings are larger
and the return on investment is improved.
The second option shown in Table 2adding more area to
E-3offers an attractive incremental payback of 2.4 years.
Once the decision is made to install a hot separator, it may be
more cost-effective to increase the size of the feed/effluent
exchangers at the
Due to the higher separation temperature and increase in
hydrogen solubility, installing a hot separator incurs two
important process-related consequences:
1. Reduced hydrogen content in the recycle
gas. A 5% reduction in hydrogen (H2)
concentration is expected for the example hydrotreater. This
will shorten the catalyst life and reduce the cycle length from
about 3 years to 2.2 years. Assuming a catalyst volume of 120
tons, at 16 /kg, the extra catalyst replacement cost
would be around 230,000/yr. Alternatively, the catalyst
life could be restored to three years by increasing the purge
and H2 makeup. About 1,500 Nm3 of
additional H2 will be needed for each 100
m3 of feed. Which of the two alternatives will be
selected depends on the hydrogen cost. Shorter cycle length
would also incur a production loss of 12 days/yr and some
added maintenance costs.
2. Increased H2 loss to fuel
gas. This additional loss is estimated at 175 kg/h. In
this particular case, this additional loss is estimated at 175
kg/h, losing around 275,000/yr as a difference between
the cost of H2 (900/ton) and its values as
fuel (6/GJ). This loss is, however, much reduced if the
purge gas is recycled to the H2 manufacturing
The additional processing cost reduces the benefits of
installing a hot separator from 885,000/yr to a value
between 375,000/yr and perhaps 550,000/yr,
depending on the purge gas routing. This net benefit can be
lower or slightly higher than the 490,000/yr achievable
by revamping the preheat train only. The extra processing cost
renders the hot separation unattractive in this particular
retrofit. Similar results have been reported
elsewhere.4 The conclusion, however, may change if
the effects of unit debottlenecking and/or throughput increase
become substantial. Those benefits would be greater with the
hot separator than with just revamping the feed preheat train,
and may again swing the project economics in hot
Minor processing issues.
Other, minor process issues to be addressed are:
Wash-water and stripper operation.
In the present study, it was possible to maintain the
wash-water consumption and stripper conditions at present
values, so that there would be little or no change to the
downstream operation. This needs to be verified for each
Additional pressure drop from sending the feed
through E-1, although any ∆P increase would be, to a
large extent, offset by the lower flowrate through the cooler,
upstream of the cold separator.
A process study may also address moving exchanger
E-3 upstream of the feed pump, to avoid the need to increase
the pressure ratingincluding the effect of high
temperature on pump operation and cavitation.
Control outlet temperature.
As the furnace duty is reduced by the listed revamp
projects, a question may arise concerning control of the
reactor outlet temperature (which is affected by the feed
temperature and reaction exotherm).
Improved heat integration greatly reduces the duty
of the feed furnace, in some cases to zero. It can be argued
that an important temperature controlling mechanism is removed.
With the feed heater in operation, the reactor outlet
temperature can be controlled simply by turning down the heater
and reducing the feed temperature.
However, with the furnace on minimum firing or on standby,
other controlling mechanisms can be used. These include a
quench-gas flow to the reactor bed, bypassing the feed/effluent
exchanger, using a feed cooler or installing an additional heat
consumer (e.g., a steam generator) in the reactor effluent
To enhance the heat transfer in a large shell-and-tube heat
exchanger with tight temperature approach and a significant
temperature cross, large additional area must be installed. The
materials of construction and pressures involved
add to the costs for such revamps. To lower this cost,
alternative exchanger technologies can be considered. Such
Plate/frame. The plate-and-frame type of
exchangers offer high heat-transfer coefficients, and are
fairly straightforward to mechanically clean. However, the
pressure requirements of hydrotreating units may make some of
these exchangers unsuitable in this application. Plate
exchangers have been proven in large grassroots designs. For
the particular unit considered here, the duty was too small to
Printed-circuit heat exchangers (PCHE).
These exchangers use diffusion-bonded stainless steel plates,
with channels for the fluid etched in them. The channel
geometry leads to high heat-transfer coefficients, while the
design is suitable for high pressures. PCHEs are compact, can
be produced in small sizes and offer several advantages:
Fits in plot area of existing exchangers
Tighter temperature approach can be economically
Lower overall pressure drop
Improved project economics. In the present study,
results found are:
1adding exchanger area. Slightly higher benefits
are obtained (670,000/yr vs. 537,000/yr, due to the
lower pressure drop of PCHE), at 40% lower investment
(2.1 million vs. 2.9 million), offering a 3 year
payback (vs. 5.5 years)
º Option 2using
PCHE in conjunction with hot separator. Again,
slightly higher benefit (930,000/yr vs. 885,000/yr)
are obtained, but at lower investment cost (2.8 million
vs. 3.3 million), and faster payback (3 yr vs. 3.7
Potential disadvantages of using PCHE are:
Mechanical cleaning is not possible (although
materials of construction and low liquid volume
are suited for chemical cleaningsimilar to plate
Small passages (slightly smaller than in plate
exchangers) raising fouling issues
Insufficient experience. PCHE have many offshore
applications, but are rarely used in downstream industry.
Of the two options available for revamping a hot-fed
hydrotreater, 1) adding area to feed/effluent exchanger, and 2)
installing a hot separator with some addition of heat exchange
area, the hot separator option (with exchanger area addition),
offers larger energy savings (13 GJ/h vs. 7.9 GJ/h of furnace
process duty), and larger potential energy benefits
(885,000/yr vs. 490,000/yr). Hot separator requires
a higher investment cost (3.3 million vs. 2.3
million), but offers a more favorable return (3.7 yr vs. 4.6
With a hot separator, the unit energy efficiency, as
measured by the BT index, would improve from 450% BT to around
240% BT. Installing a hot separator incurs additional
processing cost due to reduced H2 concentration in
the recycle gas. This affects the catalyst life and cycle
length, and increases H2 loss to fuel gas. These
factors substantially impact project economics; they can render
the hot separation option economically unattractive in retrofit
situations. In grassroots designs, however, it is expected that
the hot-separator configuration would be chosen.
The analysis and the project economics are based on
energy benefits only. In many cases, improved heat recovery
de-bottlenecks the feed furnace and enables an increase in unit
throughput. Both revamp options are open to this additional
potential benefit. If the unit capacity can be increased, the
refiner may find that the yield-related benefits outweigh the
energy ones, and they more than compensate for the
H2 purity loss in case of the hot-separator
The authors thank Dharmesh Panchal and Joris Mertens for
their valuable suggestions during the preparation of this
1Milosevic Z. and W. Cowart, Refinery energy efficiency and
environmental goals, Petroleum Technology
Quarterly, Summer 2002.
2 For an introduction to Pinch Analysis see:
Hans-Joachim Leimkühler Managing CO2
emissions in the chemical industry, published by
Wiley-VCH, 2010. A more detailed account is to be found in:
Kemp I. C., Pinch analysis and process integration. A user
guide on process integration for the Efficient Use of
Energy, Elsevier, 2007.
3 Barnes, et al, HDS benefits from plate heat
exchangers, Petroleum Technology Quarterly,
4 Mertens, J., Rising to the CO2
challengePart 3. CO2 Emission reduction
options in refineries, Hydrocarbon Engineer,
Milosevic is a senior staff consultant with KBC
Process Technology Ltd., and an internationally renowned
authority on energy optimization and profit improvement
of oil refineries and petrochemical plants. He is
best known through his work on profit improvement and
energy conservation. He has over 40 published papers and
articles on energy efficiency, refinery/petrochemicals profitability
improvement, and energy economics. Dr. Milosevic has
given numerous training courses in energy economics, refinery energy efficiency and
is a senior consultant with KBC Process Technology Ltd, working in the
Energy Optimization group. He manages energy and water
optimization studies in refining, petrochemicals and gas processing industries in
the Far East. Mr. Shire holds BS and MS degrees in
chemical engineering from the University of