REVAMPS

Vacuum Unit Revamps

Following vacuum unit revamps, it is not uncommon to have operating problems. Case studies
show how the root cause can be identified using differential pressure measurements.

Daryl W. Hanson
Process Consulting Services Inc., Houston, Texas

Norm P. Lieberman
Process Improvement Engineering, Metairie, LA

T
roubleshooting refinery vacuum unit Column internals generate pressure drop mmHg absolute, most digital gauges read a
revamps(Photo 1) begins with accu- if they are functioning properly, but so will few mmHg low. Special high precision
rate differential pressure measure- liquid level. Because pressure drop gener- gauges are required if accurate pressure
ments. Specific pressure measurement ele- ated by packed column internals is very readings are required. When measuring
vations and detail drawings of the column low, two gauges must be used with the pressure drops, this is not a problem.
internals are essential to properly interpret readings taken simultaneously. To ensure
the data. Without complete information there is no offset between the two instru- Rules of Thumb
wrong conclusions can be drawn from the ments, they should both be connected to Column internals generate pressure drop
differential pressure measurements. Three the same process point and each should and so does liquid head. Properly func-
case studies will show how accurate field have the same reading. By measuring dif- tioning stripping section trays will generate
measured differential pressure can be ferential pressure simultaneously the read- between 3.5-5 mmHg per tray. Liquid level
interpreted incorrectly. These cases present ings are not affected by normal pressure also produces
P, therefore 10” of 0.75
only a few of the ways to misinterpret data. variations. specific gravity liquid generates approxi-
When measuring pressures less than 10 mately 13-14 mmHg. Since packing gener-
Measuring Low Pressure
Being able to accurately measure pressures
under deep vacuum requires proper equip-
ment. Acceptable measuring devices are
absolute pressure mercury manometers or
absolute pressure electronic devices.
Moreover, tubing from the instrument to
the process connection must not leak or
measured pressure will be higher than true
process pressure. To test for leaks the pro-
cess isolation valve should be first opened
so that the measuring device reads the
process pressure, then the valve must be
blocked in. If the measured pressure grad-
ually increases then the tubing is leaking, if
it remains constant there is no leak.
If the pressure does not fall to a steady
value within a few seconds, then the con-
nection is partially plugged. The steady
state reading will then read high because a
very minor air leak cannot be completely
off-set by a relatively larger opening into
the connection. Photo 1. Vacuum unit

Reprinted from PTQ Revamps & Operations®, Autumn 2003

 REVAMPS
Case 1:
Black HVGO Product –
“Low” Wash Section
Pressure Drop
Cat feed hydrotreater (CFHT)
run lengths were being dramati-
cally reduced by rapid metals
Measured deposition on the catalyst.
20 during survey
Recent refinery configuration
Structured changes allowed heavy Mexican
Packing heavily
packing
coked in middle and Venezuelan crudes contain-
Grid of bed
ing very high vanadium to be
Slop wax tray
processed. While the vacuum
unit had a history of coking after
a 3 year run, switching to heavy
Actual flash crude caused downstream unit
zone pressure
31 + 8mmHg higher contamination problems in only
Flash zone than inside horn 18 months. CFHT catalyst had
Transfer Vapor 23 Measured reached end-of-run and an
line horn during survey unscheduled outage was
planned. A decision needed to be
made to go into the vacuum col-
umn and replace the packing or
Absolute pressure, mmHgA continue running.
A pressure survey was con-
Figure 1. Measured pressures and locations ducted because coke reduces
packing open area and generates
ates only 0.375 mm Hg pressure drop per higher
P as the coke builds up in the
foot of packing, pressure must be mea- packing. Once pressure drop reaches
sured simultaneously to determine differ- approximately 1 mmHg per foot of pack-
ential pressure. Five feet of packing gener- ing, heavy vacuum gas oil (HVGO) prod-
ates only 2 mmHg when operating proper- uct will begin to turn black due to vacuum
ly and 5 mmHg when coke is being formed. tower bottoms (VTB) entrainment.
Small variations in operating pressure can Pressures were measured simultaneously
materially change the conclusions about in the flash zone and above the wash sec-
the condition of a packed bed when a sin- tion because of convenient platform
gle gauge survey is performed. access. Figure 1 shows the measured pres-
Photo 2. Coked grid Photo 3. Coked structured packing
Reprinted from PTQ Revamps & Operations®, Autumn 2003
sures and the locations. Measured
P
across the slop wax (dirty gas oil or over-
flash) collector tray and five feet of wash
section packing was only 3 mmHg.
Interpreting pressure drop correctly
requires complete knowledge of column
internals and how they influence pressure.
Pressures were taken above the wash bed
and at the same location as the flash zone
pressure instrument. Detailed review of
vendor drawings showed a vapor horn had
been installed; yet it did not appear on the
vessel elevation drawing. The flash zone
pressure measurement was taken inside
the vapor horn.
Pressure should never be measured
inside vapor horns because vapor and liq-
uid velocities are very high. Kinetic energy
can make the pressure reading lower or
momentum from liquid impinging on the
pressure point can produce readings high-
er than the flash zone pressure. Pressure in
the flash zone was 8 mmHg higher than
inside the horn. Therefore, the kinetic
energy component of system pressure was
very high causing the measured pressure
to be lower than the flash zone. Because
the true pressure drop across the packing
was very high, it confirmed the bed was
badly coked (Photo 2 and 3) and a shut-
down was needed to replace the packing.
Furthermore, high column pressure drop
reduced HVGO product yield.
Conversely, pressures outside the vapor
horn and below the overflash pan can be 3
to 6 mmHg lower than the measurement
inside the horn due to liquid impingement
 REVAMPS

Flash 25 Measured
760
zone
Measured
16 during survey

Structured Packing clean when

packing Five (5)
inspected during sieve Initial
P
20 mmHg
Grid turnaround trays
10" Actual
P
0 mmHg
liquid With steam 755
No steam 757
Measured 18 18 No liquid head Stripping steam

P across 31 Measured 15" liquid
collector 21 during survey Quench
Slop wax tray LC 45 With steam
(in liquid level) flow 680 LC 45 No steam
25 Reduced
FI liquid level
21 Actual flash zone
15,000 With steam
Transfer line Flash zone pressure + 10mm Hg
lower than at 15,000 No steam
slop wax tray
Pressure, mmHgA

Absolute pressure, mmHgA Temperature, °F

BPD

Figure 2. Pressure measurement at the upper-level tap on the slop
wax collector, below the wash bed and in the vapor
space above
on the pressure tap. Using this reading
alone, will lead the troubleshooter to a
conclusion that the wash zone packing
and overflash tray have high pressure
drop, but this is often not so when the
refiner opens the tower. The lesson is clear:
a pressure tap immediately below the wash
zone packing should be used to measure
critical wash zone differential pressure.
Case #2:
Black HVGO Product –“High”
Wash Section Pressure Drop
Black HVGO product contained 1.6 wt %
microcarbon residue that increased FCC
catalyst consumption by more that 4
tons/day. Black HVGO is a symptom of
wash bed coking, yet the refiner had no
history of coking. Moreover, flash zone
temperature was only 730°F and the wash
bed packing depth was only 4 feet.
Experience shows it is nearly impossible to
form coke in short crude vacuum column
(versus visbreaker or hydrocracker) wash
zones. Because entrainment from the flash
zone and wash liquid from the top keep the
packing sufficiently wetted to avoid “dry-
out” in short packed beds and operating
temperature was only 730°F, coking was
not likely. Yet in spite of low temperature
Figure 3. Measured pressures above and below the stripping
section
and a short bed, black HVGO product
pointed to coke.
In addition, the pressure survey mea-
sured 15 mmHg across the wash bed. High

P indicated that coke was likely causing
the black HVGO. Figure 2 shows pressure
was measured at the upper level tap on the
slop wax collector, below the wash bed and
in the vapor space above. Furthermore,
slop wax tray level measured 60% which
indicated the upper instrument tap was in
the vapor space. But, slop wax collector
trays have no level glass to confirm the
instrument reading is a true level and these
instruments often do not work. Therefore,
pressure was measured in the flash zone as
a check of slop wax tray liquid level.
Measured flash zone pressure was only
21 mmHg versus the 31 mmHg measured
on the slop wax tray. Level on the tray was
reduced until the instrument reading was
15%, at the same time the measured pres-
sure decreased from 31 to 18 mmHg. Thus,
the initial pressure reading on the tray
included 10 inches of liquid, which
increased the pressure by 13 mmHg.
Further review showed that the top of the
collector tray risers were 10 inches higher
than the instrument top tap. Consequently,
slop wax was overflowing the riser during
normal operation measured at 60% of
level. Vapor flowing through the risers
entrained slop wax that contained vacuum
tower bottoms from flash zone entrain-
ment. Also when tray level was lowered,
HVGO product color became a translucent
dark green. The packed bed was not coked
(Photo 4)!
Case #3:
Low HVGO Product Yield-
“Normal” Stripping Section
Pressure Drop
HVGO product yield had dropped by
4,000 bpd following start-up even though
the same crude rates and blends were
being processed. Problems with column
boot level instrumentation were common
during start-up and throughout normal
operation. High liquid level is the most
common cause of stripping section tray
damage. Yet in the most recent turnaround
no tray damage was found. Until the refin-
er switched to heavier crude, very little
emphasis was placed on stripping section
performance. But as crudes get heavier,
stripping section performance plays an
increasing important role in meeting
HVGO product yield targets. Properly
functioning stripping trays generate 3-5

Reprinted from PTQ Revamps & Operations®, Autumn 2003


Photo 4. Grid without coke
mmHg pressure drop per tray.
Figure 3 shows measured pressures
above and below the stripping section.
Measured differential pressure was 20
mmHg or approximately 4 mmHg per tray
so the stripping section appeared to be
working. Stripping trays work by reducing
oil partial pressure, which vaporizes a por-
tion of the liquid. Heat for vaporization
comes from the liquid stream; therefore,
liquid temperature drops as light material
is stripped. Yet temperature measurements
between the flash zone and the bottom of
the stripping section showed no tempera-
ture drop. Further testing was needed to
clarify the inconsistent data.
Photo 5. Stripping tray damage
Reprinted from PTQ Revamps & Operations®, Autumn 2003
REVAMPS
Because pressure is generated by strip-
ping steam and stripped hydrocarbon
flowing through the sieve holes and liquid
level on the trays, if steam is blocked in,
then the measured pressure drop will nor-
mally decrease to about 10% or less of nor-
mal values. However, when the steam was
blocked in there was no change in the
measured pressure drop. This indicated
that measured pressure included liquid
level. By decreasing column bottom level
the measured pressure drop decreased to
zero. When stripping steam was put back
in service, measured pressure drop was
still zero. Hence the trays had been dis-
lodged by high liquid level (Photo 5).
When troubleshooting vacuum column
strippers, the quench flow rate should also
be monitored. If stripping trays are work-
ing properly there will be temperature
drop. Thus when steam is blocked in, the
temperature of the liquid leaving the strip-
ping trays increases, hence quench flow
rate must increase to maintain constant
temperature in the boot or the columns
bottom temperature will increase. When
stripping trays are not working, there is lit-
tle or no change in quench flow rate or bot-
toms temperature when steam is blocked
in.
High liquid level also results in loss of
vacuum due to liquid phase cracking. If
lowering the pool level restores vacuum,
liquid phase cracking is occurring.
THE AUTHORS
Daryl W. Hanson is a chemical
engineer with Process Consulting
Services in Houston, TX. His respon-
sibilities include process and equip-
ment design. He specializes in all
phases of refinery distillation from
process simulation through field
inspection. Previously he was lead
process specialist for Koch-Glitsch
Inc. where he was involved with more
than 100 column revamps including
heavy oils and light-ends recovery towers. Hanson has
authored more than 20 technical papers on revamping
and distillation. He holds a BS degree in Chemical
Engineering from Texas A&M University.
Norman P. Lieberman is an independent refinery
troubleshooter for his company Process Improvement
Engineering in Metairie, Louisiana. He specializes in
revamping distillation equipment and correcting process
operating problems. He has authored several books and
numerous articles.
Process Consulting Services, Inc.
3400 Bissonnet
Suite 130
Houston, Texas 77005
U.S.A.
Phone: [1]-(713)-665-7046
Fax: [1]-(713)-665-7246
E-mail: info@revamps.com
Website: www.revamps.com