This preview shows page 1. Sign up to view the full content.
Unformatted text preview: Distillation Component Trapping
in Distillation Towers:
This article explains the principles behind
minor-component trapping, summarizes industry’s
experience, and provides guidelines
for diagnosing and addressing the problem. Henry Z. Kister
Fluor Corp. C OLUMN FEED OFTEN CONTAINS COMPONENTS
whose boiling points are between those of the light
and heavy key components. In some cases, the top
temperature is too cold and the bottom temperature too hot to
allow those components to leave the column as fast as they
enter. Water, because of its non-ideal behavior with organics,
is a common problem. Having nowhere to go, these components accumulate in the column, causing flooding, cycling and
slugging. If the intermediate component is water or acidic, it
may also cause accelerated corrosion; in refrigerated columns,
it may produce hydrates. A large difference between the top
and bottom temperature, a large number of components, and
high tendencies to form azeotropes or two liquid phases are
conducive to intermediate component accumulation.
Although component accumulation has created major
problems in many columns, it has not been extensively
addressed in literature. This article expands the previous work
(1–3) into a more comprehensive review aimed at providing
guidelines on anticipating, troubleshooting and overcoming
component accumulation problems. Principles
From each stage in a tower, the molar upward flowrate of
component i, vi (lb-mol/h), is given by:
vi = Vyi = VKixi (1) where K is the K-value at the stage temperature, L and V are
the total liquid and vapor flowrates (lb-mol/h), and x and y are 22 www.cepmagazine.org August 2004 CEP the component concentrations in the liquid and vapor phases,
respectively. The downward flowrate of the same component,
li, is given by:
li = Lxi (2) When vi > li, there is a net upward movement of the component; likewise, when vi < li, there is a net downward movement. Combining this criterion with Eqs. 1 and 2 gives:
Net upward movement when VKi /L > 1
Net downward movement when VKi /L < 1 (3)
(4) In binary distillation, the light key component (LK) always
obeys Eq. 3, while the heavy key component (HK) always
obeys Eq. 4. Consequently, there is always a net upward flow
of the light key component and a net downward flow of the
heavy key component, that is:
VKLK/L > 1
VKHK/L < 1 (5)
(6) For a component with a boiling point intermediate between
the light key and heavy key, the following generally applies:
KHK < KIK < KLK (7) where IK denotes the intermediate key.
Now consider a situation where the V/L ratio is set by the Concentration, mol% initial binary separation, the tower produces reasonably pure
products near the top and the bottom, and an intermediate key
is added into the feed. For a pure top product, KLK near the top
approaches unity. Since KIK < KLK, the ratio VKIK/L will be less
than 1 when KLK/KIK exceeds V/L, suggesting a net downward
flow (Eq. 4). Similarly, for a pure bottom product, KHK near
the bottom approaches unity. Since KIK > KHK, VKIK/L will
exceed 1 when KIK/KHK exceeds L/V, suggesting a net upward
flow. With a net downward flow near the top and a net upward
flow near the bottom, intermediate components tend to accumulate in the tower.
Figure 1 illustrates the buildup of components in a tower
(4). The feed was introduced at stage 10, with the tower bottom at stage 22. The intermediate key component, which tended to concentrate below the feed, was continuously removed
as a vapor side draw from stage 13. The heavy key component, which was more volatile than the heavy non-keys that
made up most of the bottom stream, steeply concentrated
between the bottom stage 22 and the intermediate draw stage
13, peaking at stages 17–18. Over the four stages between 17
and 13, the HK concentration was meant to drop from 50% to
1% based on the design. With such a steep concentration gradient of HK, a slight shortfall in the number of stages (by as
little as one or two out of eight) caused the
concentration of HK in the side draw to
increase from 1% to 10%. Here the HK
acted as the trapped component.
Intermediate component buildup most
frequently takes place over the entire tower,
but at times, it is confined to a section.
Excessively subcooled feed (or fouling of a
feed preheater) can lead to accumulation of
an intermediate component between the
feed and the bottom. Similarly, excessive
preheat (oversurfaced or clean preheater)
may lead to an accumulation between the
feed and the tower top. upward movement, whereas a low ratio near the top of the
column signifies a large movement downward. When the two
come together in the same column, the concentration can be
tremendous. Concentrations (“bulges”) 10 to 100 times the
feed concentration are common.
• non-ideal equilibrium. A high activity coefficient at the
tower bottom can make an intermediate component particularly volatile. For instance, when the tower bottom is mainly
water, organic components such as n-butanol become highly
volatile (high VKIK/L). The same organic can become highly
nonvolatile if the top is rich in methanol or ethanol. For a
methanol/water separation column, n-butanol in the feed will
tend to accumulate to a large extent (2).
• the number of stages. Each stage intensifies the upward
and downward movement. The more stages, the more the
intermediate component concentrates toward the middle of
• the concentration of the intermediate component in the
feed. The higher the concentration, the greater will be its tendency to concentrate in the tower.
• product specifications. The tighter the specs, the greater
the accumulation. Much of this is due to the larger number of
stages needed to reach the tighter specs, especially stages at
high purity, where the upward or downward movement of the intermediate comHK in
ponent is intensified.
In accumulation situations, there is
always an initial period of non-steadystate intermediate component buildup.
The buildup tends toward the equilibrium
concentration that reinstates the component balance in the tower. This equilibrium concentration, however, may not
always be reached, such as when the
VKIK/L is high near the bottom and low
near the top, and/or the number of stages
is high, and/or the concentration of the
intermediate component in the feed is
high, and/or the product specs are tight.
Instead, unsteady-state cycling may set in. How much accumulation?
The accumulation continues until the
intermediate component concentrations in
the overhead and bottom allow removal of
these components at the rate they enter, or
until a hydraulic limitation is reached.
The factors governing intermediate
component accumulation are essentially
those that govern the split of intermediate
components between the top and bottom
■ Figure 1. Composition profile pinpoints
• the VKIK/L ratio. A high ratio near the the concentration of the heavy key between
bottom of the column signifies a large
the bottom and side draw. Hiccups and cycling
A typical symptom of unsteady-state
accumulation is cyclic slugging, which
tends to be self-correcting. The intermediate component builds up in the column
over a period of time, typically hours or
days. Eventually, the column floods, or a
slug rich in the offending component
exits either from the top or the bottom.
(The end from which the slug leaves CEP www.cepmagazine.org August 2004 23 Distillation often varies unpredictably.) Once a slug leaves, column operation returns to normal over a relatively short period of time,
often with minimal operator intervention. The cycle will then
repeat itself. Several experiences have been reported in the literature, and those are abstracted in Table 1.
Intermediate component accumulation may interfere with
the control system. For instance, a component trapped in the
upper part of the column may warm up the control tray. The
controller will increase reflux, which pushes the component
down. As the component continues accumulating, the control
tray will warm up again, and reflux will increase again; eventually, the column will flood. Three cases describing a similar
sequence have been reported (DT2.13B, 1213, and 228).
One of the most fertile breeding grounds for intermediate
component accumulation is a condition in which some section
in the tower operates close to the point of separation of a sec- ond liquid phase (2). The accumulation itself, or a malfunction
of phase separation equipment in the feed route, or simply a
change in feed composition, can induce separation of a second
liquid phase inside the tower. This changes volatilities and
azeotroping, and can generate or aggravate hiccups.
It, therefore, is not surprising that many of the cases in
Table 1 involve such a system. Examples include small quantities of water in a refinery or natural gas deethanizer (DT2.10,
1038, 1039, 751, 1001), and oil or other hydrocarbon accumulating in a methanol/water tower (DT2.15). In other cases,
such as azeotropic and extractive distillation, the accumulation
interacted with the phase separation in the tower and the
decanter (e.g., DT2.22 and DT2.23).
Figure 2 illustrates a common configuration of a reboiled
deethanizer absorber. The top (absorber) section uses a naphtha
lean oil stream to absorb valuable C3 and C4 components from Table 1. Hiccup experiences reported in literature.
Column Case Ref. Brief Description DT2.10 5 Refinery
deethanizer The tower bottom was cooled in an attempt to improve C3 recovery. This led to water accumulation,
causing hiccups and water slugs out the bottom every 2–3 days. The problem was cured by returning
to the previous operation mode. 1038 6 Refinery
stripper At feed drum temperatures below 100°F, the tower flooded due to water and C2 accumulation. To
unflood, the boilup was cut or the drum warmed, but at the expense of poorer C3 recovery. Gamma
scans showed flood initiation in the middle of the tower, even though the highest hydraulic loads were
at the bottom. A retrofit with high-capacity trays produced no improvement. Replacement of the inadequate water-removal facilities by an internal water-removal chimney tray eliminated the water accumulation. 1039 6 Refinery
absorber The tower had no vapor or liquid recycle to the feed drum, two external water intercooling loops, and
inadequate water-removal facilities. Water trapping during cold weather led to severe flooding and carryover. Warming the feed drum and cutting the intercooler duty were cures, but at the expense of lower
C3 recovery. Later, a properly designed water draw tray was installed to eliminate water entrapment. 751 7 Refinery
stripper The residence time in the downcomer-box water draw was too low to separate water, so no water was
withdrawn. The water built up in the tower, periodically hiccuping, disturbing pressure control and
contributing to condenser corrosion and fouling in the downstream debutanizer. Replacement with a
seal-welded chimney draw tray eliminated the problem. 1001
5 Natural gas
deethanizer Small quantities of water accumulated in a refluxed deethanizer and caused the column to empty
itself out from top or bottom, every few hours. This was cured by replacing reflux with oil absorption. DT2.13A 5 Refinery
debutanizer The tower, processing coker naphtha and straight-run naphtha, experienced a hiccup once every 4 h.
The problem was more severe when the preheater fouled. The cure was to operate the top of the tower
warm, but at the expense of lower recovery. DT2.13B 5 Refinery
debutanizer A tray temperature controller manipulated the reboiler steam. Periodically, the boilup rate rose over
time without setpoint changes and would continue rising. The cure was to lower the control temperature
by 20–30°F for a short time, then return to normal operation. 211
5 Azeotropic Intermediate components accumulated in the tower, causing regular cycling (hiccups). This was cured
by raising the top temperature. Product loss due to the higher top temperature was negligible. 1213 9 Chemicals A trapped component periodically built up in the upper section of a large distillation column. When it built
up, the control temperature rose and increased reflux, eventually causing flooding. Gamma scans
diagnosed the problem. Taking a purge side draw solved the problem.
Table 1 continues 24 www.cepmagazine.org August 2004 CEP gas that goes to fuel. The liquid from the absorber is mixed
with fresh feed, cooled and flashed. Flash drum vapor is the
absorber vapor feed. Flash drum liquid is stripped to remove
any absorbed C2 and lighter components. Stripper overhead
vapor is also combined with the fresh feed prior to cooling.
When the column is properly designed, all the free water in
the feed is removed in the flash drum. Only the minute
amount of water dissolved in the hydrocarbon should enter.
Inside the tower, the water tends to concentrate. The tower top
is cool, and the naphtha tends to absorb water, sending it down
the tower. The bottom temperature is hot, tending to vaporize
the water and send it back up. Overall, the water concentrates
in the middle. If the water removal facilities are inadequate, or
are not in the region where water concentrates, or the absorber
bottoms and/or stripper overhead are internal and not returned to the feed flash drum, hiccups and floods may result.
Means of improving C3 and C4 recovery include cutting
boilup, cooling the feed, or increasing the naphtha rate, each of
which cools the tower and shifts the water concentration downward. The system in Figure 2 (Case DT2.10) had no water
removal facilities below the feed. Attempts to recover more C3
and C4 led to concentration of water below the feed and to hiccups. The hiccups led to slugs of water in the stripper bottoms,
which went into a hot debutanizer downstream, rapidly vaporizing and causing a pressure surge there. In Cases 1038 and
751, the water removal facilities did not have enough residence
time for water separation, and were unable to mitigate the
water accumulation and hiccups. In other cases, poor functioning of the feed water removal facilities caused hiccups.
Figure 3 shows a methanol/water tower (Case DT2.15) Table 1. Hiccup experiences reported in literature — continued. Plant and/or
Column Case Ref. Brief Description 228 10 Multicomponent The capacity of a packed tower dropped to almost zero in 3–4 days. A shutdown and restart
reestablished full capacity and the cycle repeated. The cause was accumulation of a trace
intermediate component in the stripping section. It was cured by a vapor side draw between the beds. DT2.15 5 Methanol/
water Several cases were reported in which oil, hydrocarbon and heavier alcohols caused hiccups in this
separation. In one case, separation of an oil or liquid phase could have played a role. DT2.16 5 Ammonia
stripper Small quantities of methanol accumulated, causing hiccups every 2–3 days. This was cured by
increasing the overhead temperature. DT2.22 5 Azeotropic
dehydration Hiccups every 2–3 days were caused by accumulation of an aromatic alcohol that was recycled into the
tower with the benzene entrainer. This was cured by adding water to the decanter to extract the alcohol. DT2.23 5 Extractive
distillation A pilot-scale tower separating aldehydes and ketones from alcohols using water as a solvent experienced hiccups every 2 h when one heavy ketone was at >1% concentration in the feed and at the same
time the water/feed ratio was low. Modeling showed two liquid phases on most rectification trays.
Foaming could have played a role. The cure was a higher water/feed ratio. DT22.14 5 Solvent
recovery The tower separating organics from water periodically foamed due to accumulation of heavy alcohols
just above the feed. The foam (which was seen in sight glasses) raised the tower dP. The cure was
opening the side draw above the feed and temporarily cutting steam. DT15.2 5 Formaldehyde Premature foaming and flooding occurred near the feed. Cause was products of in-tower reactions
forming an intermediate-boiling azeotrope that induced foaming. Enlarging downcomers, adding an
antifoam, and minimizing feed acidity helped alleviate the problem. 202
5 Natural gas
lean oil still An added preheater that performed better than design caused the column to hiccup and empty itself
every few hours either from the top or bottom. A bypass around the preheater eliminated the problem. 229 6 Refinery
stripper Chronic and severe flooding occurred at low (70–80°F) feed drum temperatures. Gamma scans
showed the flood initiated above the internal water-removal chimney tray, eight trays from the top,
even though hydraulic loadings were higher further down in the tower. C2 accumulation and foaming
were possible causes. Adding a feed preheater cured the problem. 221 12 Refinery
stripper Cold feed temperatures caused ethane condensation and accumulation, leading to hiccups once per
week during the winter. This was solved by bypassing some of the feed around the feed cooler. Note: DT indicates cases taken from “Distillation Troubleshooting,” by Kister, H. Z., to be published by Wiley, Hoboken, NJ, 2005 (5). CEP www.cepmagazine.org August 2004 25 Distillation Gas
L L Deethanizer
Boot H2O C3, C4 and Naphtha
to Debutanizer ■ Figure 2. Reboiled deethanizer absorber system that experienced
water accumulation. (11). Some of the oil phase from the separator was entrained in
the feed, in addition to the oil dissolved in the methanol/water
phase. Oil components rapidly became less volatile as they
went up the tower, where methanol concentrations were higher,
and rapidly became more volatile in the water-rich lower sections. Light oil components escaped overhead and heavy ones
out the bottom, while mid-range oil components accumulated,
leading to intermittent flooding and hiccups.
Figure 4 illustrates a solvent recovery tower that experienced
hiccups even though phase separation did not appear to be an
issue (Case DT2.14). The tower separated a low-boiling organics/water azeotrope from a water bottom stream. The tower
experienced hiccups, at times due to concentration of npropanol (cold cycle), at other times due to concentration of a higher-boiling component designated CS (warm cycle). Both
were very volatile in the water-rich bottom section, and became
non-volatile in the cold ethanol-rich upper part of the tower.
Ross and Nishioka (13) showed that foam stability is at a
maximum at the plait point, i.e., just before the solution breaks
into two liquid phases. In many cases, the intermediate component accumulation will drive the tower toward the plait
point, foaming will break out, and the tower will foam-flood.
Sometimes, the foam flood entrains the accumulating components into the top product, permitting the tower to return to
normal. Foaming due to component accumulation occurred in
Case DT22.14, and possibly also DT2.23 and 229. The interactions between intermediate component accumulation and
foaming are discussed at length in Ref. 2.
As stated earlier, intermediate component buildup may take
place over a section of the tower rather than over the entire
tower. In Cases 202, 221, 229, and possibly DT2.13A, 1038
and 1039, the buildup occurred between the feed and either
the top or the bottom. The buildup between the feed and top
was caused by excessively hot feed, and that between the feed
and bottom by excessively cold feed. Freeze-ups, corrosion and separation issues
Minor-component accumulation may lead to other problems well before the accumulation reaches or even approaches
the hiccup concentration. In some cases, the buildup reaches CW
30 75% H2O
8% Others 25 NO
Oil Vent 178°F TC
185°F Azeotrope to
0.2% CS 20
Sample 15 Reflux Drum Gas
Accumulation Oil LC Three–Phase
Condensate Steam H 2O
Feed Water ■ Figure 3. Oil accumulation in a methanol/water tower. 26 www.cepmagazine.org August 2004 CEP Figure 4. Solvent recovery column that experienced hiccups. steady state, and the problem occurs due to steady-state concentration. The most troublesome problems include hydrates,
corrosion, instability and purity issues (Table 2). Tower hydrates are common in ethylene plant C2 splitters
and gas plant demethanizers. Small amounts of water in the
feed (which is usually dried to less than 1 ppm) concentrate in Table 2. Tower hydrates, corrosion and other accumulation experiences reported in literature.
Case Ref. Plant and/or
C2 splitter 1024 15 Natural gas
An ice plug prevented liquid draw from a chimney tray. The plug location was found by using radiodemethanizer active spot density measurements along the pipe. The ice plug was melted by external heat. DT29.1 5 Olefins
C2 splitter DT2.9 5 HCl and
With less than 3 ppm water in the feed, the 15 middle trays severely corroded, needing replacement
once a month. The top temperature was too cold, the bottom too hot, so the water was trapped and
hydrocarbons concentrated near the feed. The cure was to upgrade the tray metallurgy. 1040 29 Natural gas
(ERC) An upstream constraint caused excessive water vapor in the feed gas to the extractive distillation
column. The water peaked at the side reboiler in the middle of the stripping section, forming carbonic
acid and corrosion. Increasing solvent circulation, or surges in inlet gas, pushed the water up, causing
hydrates in the chilled condenser, with off-spec product for up to a week. Extensive methanol injection
and column thawing dissolved the hydrates. DT2.11 5 Refinery
deethanizer The tower bottom was cooled in an attempt to improve C3 recovery. This led to water accumulation. The
bottom trays in the absorber and most of the stripper trays experienced severe corrosion and required
frequent repair and replacement. 1010 16 Refinery
The column feed contained strongly acidic components, which dissolved in small quantities of water and
caused a severe and recurring corrosion failure problem. The rate of corrosion failure was greatly
depropanizer reduced by adopting an effective dehydration procedure at startup. To dehydrate, acid-free butane was
total-refluxed, while drains were intermittently opened until all water was removed. 15143 17 Refinery
deisobutanizer The distillate was C3/iC4, bottoms C5–C8, and side purge was nC4. The purge rate, adjusted per daily
lab analysis, was minimized to minimize iC4 loss. An insufficient purge led to nC4 buildup. This and
variable C5 in the feed impeded temperature control and destabilized the unit. Stabilization was
achieved by inferential model control. 210
5 Chemical The vapor side-product impurity content was 10% (design 1%) due to a non-forgiving concentration
profile. Over the eight design stages in the bottom bed, the concentration rose from 30% at the bottom,
to 50% four stages below the side draw, then dipped to 1% at the side draw. A miss by 1–2 stages
would bring the concentration to 10%. 216 18 Ethanol
distillation Heavy alcohols (fusel oils) were side-drawn, cooled, then phase-separated and decanted from the
tower’s ethanol/water mixture. Depending on the draw tray temperature and composition, the cooled
side draw did not form two liquid phases. The problem was diagnosed with the help of process simulation, and was solved by adding water to the decanter to ensure phase separation. 217 19 Alcohols/
water An intermediate component buildup caused the formation of a second liquid phase inside a column. The
problem was solved by decanting the organic phase and returning the aqueous phase to the column. 136 20 Ethylene
Small amounts of radon-222 (boiling point between propylene and ethane) contained in natural gas
depropanizer, concentrated several-fold in the C3 fraction. It decayed into radioactive lead and contaminated the
towers and auxiliaries, as well as polymer deposits on trays and wastewater from reboiler cleaning. This
caused problems with waste disposal and personnel entry into the towers. Ref. 20 describes the
contamination survey, how it was tackled and the lessons learned. Brief Description
Hydrates between the feed and the interreboiler eight trays below occurred 2–3 times per week.
Stepping up methanol injection and dryer regeneration gave only limited improvement. Methanol and
dissolved hydrates got trapped in the kettle interreboiler, from which they slowly batch-distilled back
into the splitter. This was mitigated by draining methanol/water from the interreboiler. The tower operating close to its natural flood limit flooded every two months. Natural floods were often
mistaken for hydrates and were countered by methanol injection, which aggravated them. Installing
separate dP recorders over the top and bottom sections permitted distinguishing natural floods from
hydrates, and allowed early corrective action, reducing flood episodes to once a year. Note: DT indicates cases taken from “Distillation Troubleshooting,” by Kister, H. Z., to be published by Wiley, Hoboken, NJ, 2005 (5). CEP www.cepmagazine.org August 2004 27 Distillation the tower. Under the high-pressure, low-temperature conditions in the tower, the water combines with the light hydrocarbons to form solid ice-like crystalline molecules known as
hydrates. The hydrates precipitate, plugging tray holes and
valves and eventually restricting flow through the tray. Liquid
accumulates above the plugged tray and the tower floods. A
common cure is to dose the tower with methanol, which acts
like anti-freeze and dissolves the hydrates.
The presence of an interreboiler (or side reboiler) can interfere with hydrate removal. In the system in Figure 5 (Case
1020), the methanol and dissolved water were trapped by the
interreboiler. Over time, the water batch-distilled back into the
tower, causing the hydrate to return. A blowdown from the
reboiler removed the methanol and dissolved water.
Trapping and concentrating minor quantities of water, as
minor as a few ppm, has caused major corrosion problems in
towers handling hydrocarbons together with acidic components.
It is common in refinery reboiled deethanizer absorbers (Figure
2). In a well-designed tower, the entering hydrocarbons are saturated with dissolved water. Any concentration or water entry
beyond that will lead to a free water phase. If it persists, it will
dissolve acidic components, resulting in weak acid circulating
through the tower, which is death to carbon steel. A typical
symptom is corrosion in the middle of the tower (sometimes
also further down), fouling with corrosion products in the lower
part, while the upper trays remain in good condition. 1
C2 Splitter 99 Ethylene Trapping of lights
In many hydrocarbon towers, where water is an impurity,
the reflux drum has a boot to remove the water (Figure 6). The
heavier water phase descends to the boot, from where it is
removed, typically on interface level control. If the amount of
water is small, an on-off switch is sometimes used. If the boot
level control malfunctions, water can be refluxed to the tower,
causing fouling and corrosion in the tower as described in two
cases (1004 and 15157) in Table 3.
Plugging may be a problem in the water outlet line from the
boot because of low flowrates and because solids and corrosion
products tend to become entrapped in the boot and the water
stream. The converse problem is leakage rates across the water
outlet control valve exceeding the rate of water inflow into the
boot. This makes maintaining the level inside the boot difficult
and causes loss of product in the water stream.
Both the plugging and leakage problems are most troublesome when there is a high pressure difference across the wateroutlet control valve. A high pressure difference promotes valve
leakage; it also tends to keep the valve opening narrow, which
promotes plugging. Both problems can be overcome by adding
an external water stream (which may be a circulating stream)
to the boot outlet (Figure 6). This stream boosts velocity (21,
24) and safeguards against a loss of liquid level. The external
water flowrate should be low enough to prevent excessive
water backup from overflowing the boot during fluctuations. It
is also important to pay attention to good level monitoring.
In some columns, the overhead is totally condensed and then
decanted to form an aqueous stream and an organic product.
The product is sent to a stripper to remove traces of the aqueous
phase. The stripper overhead is recycled to the condenser inlet.
When a light condensable organic enters the column, it will
end up in the organic phase. In the stripper, it will be stripped
and returned to the condenser. Thus, it will become entrapped in
the system, traveling back and forth between the condenser and
the stripper. The stripper temperature controller will act to keep 100 Feed Condensed
Tower Overhead 108 Vapor
Product 109 Reflux Drum
Interreboiler Reflux and
Liquid Product LC
Water Source: (14). ■ Figure 5. C2 splitter with interreboiler that experienced
stubborn hydrates. 28 www.cepmagazine.org August 2004 CEP ■ Figure 6. Reflux drum boot arrangement. External
Water Table 3. Experiences of lights trapping.
Column Case Ref. Brief Description 1004 21 Refinery
debutanizer Column internals and reboiler tubes severely corroded after the water draw-off control valve on the
reflux drum boot plugged. Manual draining was too inconsistent to prevent water (saturated with H2S)
refluxing to the tower. Continuous flushing of the water draw line with an external water source
prevented recurrence. 15157 22 Refinery
fractionator A plugged tap on the boot’s oil/water-interface level transmitter locked the transmitter’s reading at about
50%. Water refluxing to the fractionator over time caused cavitation and damage to the reflux pump,
and deposited salts that plugged the top internals. Water in the naphtha destabilized the gas plant. The
problem was eliminated by blowing the level tap. 111 23 Refinery
The depropanizer overhead went to an HF stripper. The stripper bottoms was the propane product,
HF alkylation while the stripper overhead was recycled to the depropanizer overhead. When ethane entered the
depropanizer depropanizer due to an upstream unit upset, it became entrapped in the overhead system and could
not get out. The depropanizer pressure climbed and excessive venting was needed. This was cured
by dropping the stripper bottom temperature to allow ethane into the propane product. the light in the system, because its presence will reduce the control temperature, which in turn will increase the stripping heat
input. The trapped light will raise column pressure, as well as
the heat loads on the column condenser and stripper reboiler.
To provide an outlet for the light, either venting from the
reflux drum or reducing the stripper heat input (thus allowing
it to leave from the stripper bottom) is necessary. Reducing
stripper heat input is more effective when the light is only
slightly more volatile than the organic product. In Case 111,
reducing the stripper heat input effectively provided an outlet
for ethane trapped in the overhead of an alkylation unit
depropanizer, where venting was relatively ineffective. Trapping by recycle
Table 4 reports cases of component trapping due to recycling of product to the feed, usually to improve product recovery. There are two solutions — adding some removal facilities
to take out the component, or increasing the purge rates.
Diagnosing component trapping
Key to the diagnosis, especially where hiccups are encountered, is the symptom. With hiccups, the symptom is cycling
that tends to be self-correcting, taking place over a long time
period. This is seldom less than 1 h, which distinguishes hiccups from other, shorter-period cycles, such as those associated with flooding or hydraulic or control issues, which typically have periods of a few minutes.
Typical cycle periods for component accumulation range
from about once every 2 h to once every week. The cycles are
often regular, but if the tower feed or product flows and compositions are not steady, the cycles may be irregular.
Drawing internal samples from a tower over the cycle
(or over a period of time) is invaluable for diagnosing
accumulation. Even a single snapshot analysis can show
accumulation of a component. In the tower in Figure 4, a sample drawn from a downcomer was key to the diagnosis. Two snapshot samples had concentrations of the accumulating components (n-propanol and CS) of about 50%,
compared to less than 10% in the feed. Unfortunately,
safety, environmental and equipment constraints often preclude drawing internal samples.
Tracking and closely monitoring temperature changes is
also invaluable for diagnosing component accumulation.
Temperature changes reflect composition changes. Since the
accumulation is that of an intermediate key component, it
tends to warm the top of the tower and cool the bottom of
the tower. This tendency will be countered or augmented by
the control system and by the rise in tower pressure drop,
and these interactions need to be considered when interpreting temperature trends. In any case, one symptom is common. At the initiation stage, temperature deviations from
normal are small, often negligible. As the accumulation proceeds, and the concentration of the intermediate key grows,
systematic temperature excursions become apparent. Close
to the hiccup point, temperature excursions become large.
The solvent recovery tower in Figure 4 was well-instrumented, with a temperature indicator every five trays. The
tower experienced two types of cycles: a cold cycle, predominantly due to the accumulation of n-propanol, and a warm
cycle, mainly due to the accumulation of CS. During the cold
cycle, with the control temperature set at 185°F and bottom at
215°F, temperatures below the control tray began to creep
down. The deviations were largest on tray 20, diminishing
toward the bottom of the tower. Over a 2–3-h period, initially
the deviations on tray 20 were small, but they became larger.
Then, suddenly the column showed flooding signs and the
temperatures dropped throughout. The operators tackled the
problem by reducing the feed to about 40%, while maintaining the steam flowrate, which allowed the accumulated component to be purged from the top. CEP www.cepmagazine.org August 2004 29 Distillation Table 4. Trapping by recycle.
Column Case Ref. Brief Description 1019
5 Natural gas
(in series) Modifications to recover the deethanizer overheads (previously sent to fuel) compressed, chilled, then
recycled it to the absorber feed. Small quantities of water, previously going to fuel, returned to the
absorber feed.The absorber top was too cold, and the deethanizer bottom too hot, to allow the water to
escape. The water built up until freezing at the recycle chiller. For years, the chiller was thawed to flare
once per shift. This was cured by adding a small glycol dryer at the compressor discharge. 139 26 Olefins
C3 splitter The frequency of propylene product going off-spec with methanol increased following installation of
a system that enhanced BTX and methanol recovery out of the plant wastewater. The recovered
materials concentrated in the process. Corrective action was routing some water away from the process
and stripping some with fuel gas into a waste gas burner. DT2.19 5 Ethanol/
water Heads (light non-keys) and fusel oil (intermediate keys) products were mixed with cold water, then
decanted, with the aqueous layer returned to the tower. Every 2–3 days the product alcohol developed
an undesirable smell due to buildup of an impurity. This was cured by cutting the feed rate, while
keeping the fusel oil and heads rates the same. Note: DT indicates cases taken from “Distillation Troubleshooting,” by Kister, H. Z., to be published by Wiley, Hoboken, NJ, 2005 (5). During the warm cycle, the temperatures near the bottom
rose. The initial rise was slow. Then the bottom temperature
suddenly jumped to 230°F (normal 215°F) and, at the same
time, the bottom pressure went up by 2–4 psi, indicating
flooding. The rise in bottom pressure accounts for much of
this boiling point rise (about 3°F/psi). This occurred regardless
of whether the tower temperature controller was in automatic
or manual mode. The overhead temperature went up to about
190°F, indicating that the tower was emptying itself out. At the
same time, the bottom flowrate did not change much. The
operators tackled that by cutting the steam flow by about half
and diverting the bottoms to an off-spec tank. This emptied
the accumulation from the bottom rather than from the top,
preventing problems in the dehydration system downstream
and reducing product losses.
Figure 7 shows an azeotropic distillation system in which an
organics/water azeotrope is dehydrated by injecting a hydrocarbon entrainer. The hydrocarbon is more volatile than the organics, so once the water is gone, it distills up, leaving an organic
bottom stream. The water and hydrocarbon leave in the tower
overhead, are condensed, then phase-separated in the reflux
drum, with the hydrocarbon returned to the tower and the water
(with some organics) removed. In this system, water descended
to about Tray 10, forming two liquid phases on the trays above.
Figure 8 is based on actual operating charts for the system
in Figure 7. The steam flowrate to the reboiler was temperature controlled. Initially, there was a good temperature gap
between Trays 8 and 12, which is typical of the region where
the second liquid phase disappears. As the accumulation proceeded, the second liquid phase descended toward Tray 8. It
was countered and pushed back by the control action, but later
came back. With time, the movement became more intense. In
this case, only the temperatures on Trays 8 and 12 were prob- 30 www.cepmagazine.org August 2004 CEP lematic; the temperatures on Trays 4 and 16 did not change
much. This is typical, and for best diagnostics, the relevant
temperatures need to be monitored. In most situations, this is
readily achievable even if thermocouples are not present, since
with today’s surface pyrometers, column wall temperatures
can be reliably measured (27).
Item 3 of Figure 8 shows what happens when the temperature control was placed on manual — it no longer countered
the accumulation, so the swings stopped. The temperature in
this case was set high enough to push the accumulation up the
tower. This provided temporary relief only. Eventually the
buildup returned, requiring a further increase in steam. The
only way to clear the buildup in the long term was to make
drastic changes (analogous to those described for the Figure 4
system) that allowed the impurity to get out.
Simulations are also invaluable for diagnosing component
accumulation. Steady-state bulges can be readily simulated
and recognized on a plot of component concentration against
stage numbers (Figure 1). But since most accumulation problems are non-steady-state, it is necessary to “trick” the simulation to avoid convergence problems (which may reflect the
physical reality that the conditions specified do not lead to a
stable steady-state solution.)
The main trick to overcome this is to study a related system that can converge. One example is to reduce the concentration of accumulating component in the feed to the point
where convergence is readily reached. Then the concentration
of the accumulating component in the feed is gradually
increased, and the changes tracked by means of a tower concentration versus stages diagram for each step.
In the case of a second liquid phase, many stages in the
simulation may alternate between a single liquid phase and
two liquid phases, making convergence problematic. There, Water
Some Organics CW Decanter Hydrocarbon
to Stripper Organics
Steam Condensate Organics ■ Figure 7. Azeotropic distillation system that experienced hiccups.
1 Initiation Tray #4
Tray #8 Tray #12
2 Buildup Tray #4
Tray #8 Tray #12
Tray #4 3 Temperature Control on Manual either increasing or decreasing the concentration of the second
liquid phase can help stabilize the simulation.
Gamma scans and dP measurements are also useful for
detecting intermittent flooding. Gamma scans showing flood
initiating in mid-column, away from feeds or draw points, provide evidence supporting accumulation, especially if the tower
hydraulic loadings are higher near the top or the bottom.
Gamma scans taken at different points in the cycle can help
trace the accumulation from initiation (at which the trays operate normally) to flood. If enough nozzles are available on the
tower, dP transmitters can be just as informative. Finally, sight
glasses are extremely useful when safety requirements permit.
In each specific situation, one of the above techniques can
be particularly valuable. For instance, Case DT29.1 (Table 2)
describes a situation in which recording the dP across each
section of the C2 splitter permitted excellent diagnosis of
hydrates. The hydrates were encountered below the feed, in a
tower section that was not operated at maximum load, so a
rise in dP of the bottom section signified hydrates, while a dP
rise in the loaded top section signified regular flooding.
There are five classes of cures to hiccup and tower accumulation problems: reducing the column temperature difference; removing the accumulated component from the tower;
removing the accumulated component from the feed; modifying tower and internals; or living with the problem. Reducing the column temperature difference
This can be done either by raising the top temperature or
lowering the bottom temperature, or both. This enables the
accumulating component to escape with a product stream.
The effectiveness of this technique may be limited, and it
can cause off-spec products and/or excessive product losses. It
was successfully applied in several of the experiences reported
in Table 1. In some of these (DT2.10, DT2.13A, DT2.13B),
there was a significant product recovery penalty. In the others
(1001, 211, DT2.16), the penalty was negligible.
A special case is accumulation between the tower feed and
the top or between the feed and the bottom. Here, the feed
temperature is often lowered to prevent accumulation of the
component in the top section or raised to prevent accumulation in the bottom section. Similarly, a feed point change may
encourage the component to leave the column at one end or
another. Proper bypasses around preheaters and precoolers are
invaluable for this purpose (e.g., Cases 202 and 221). Tray #8 Tray #12
Tray #16 ■ Figure 8. Temperature changes accompanying component
accumulation in an azeotropic dehydration tower. Removing the component from the tower
Usually, this technique involves drawing a small liquid or
vapor side steam from the column, removing the intermediate
component from the side stream externally, and returning the
purified side stream to the column. If purification is not economical, the side stream may be processed elsewhere, blended CEP www.cepmagazine.org August 2004 31 Distillation with a slop stream, or just purged. The side stream drawn
should be large enough to remove the amount of the component entering in the feed. Since at the draw-off location the
component is normally far more concentrated than in the feed,
the stream drawn is usually small.
A typical example of this technique is using an external
boot for removing water from inside a hydrocarbon distillation
column (Figure 2). Only a small portion of the tray liquid goes
to the boot. This portion must be large enough to prevent
water accumulation in the tower, but small enough to permit
adequate hydrocarbon/water separation in the boot. Proper
design of the draw box, as well as the piping to and from the
boot, are essential for avoiding siphoning, choked flow, and
excessive downcomer backup. As with the reflux drum boot,
an external water supply may be desirable. Alternatively, the
water/hydrocarbon separation can be performed inside the
tower, but this requires a large chimney tray to provide enough
settling time for the entire liquid flow. Cases 1038, 1039, 751
and 2.10 (normal C3 recovery) report successful water removal
Another typical example is removal of higher-boiling alcohols (“fusel oil”) from ethanol/water columns. Unless
removed, they will concentrate in the column, and upon
reaching their solubility limit form a second phase and cause
cyclic flooding as described earlier. Fusel oil is commonly
removed by a similar scheme to that in Figure 2, except that
the side stream is usually cooled prior to phase separation and
the aqueous phase (rather than the organic phase) is returned
to the column.
Another application of this technique is in the separation of
ethyl ether from aqueous ethanol, where benzene tends to
build up in the bottom section. Removal of a small benzenerich side stream out of the bottom section effectively increased
the column’s overall capacity (28).
The component removal technique need not be confined
to gravity settling. Other separation techniques, such as stripping and adsorption, may also be employed. Ref. 2 mentions
the use of a side dryer to prevent hydrates in a C2 splitter.
The best location for the side draw point can be determined by simulation, but changes in composition and simulation inaccuracies can alter the optimum feed point. Case
2.10 in Table 1 (Figure 2) describes a scenario in which the
optimum draw point shifted down the tower over the years,
as the economics for recovering C3 from refinery fuel gas
improved. A flexible solution often employed is to install
draw facilities on several trays, which permits online optimization. While this is easy to do with tray towers, it may be
more difficult with packings, where side draws are normally
taken from collectors or vapor spaces between the beds,
which limits the alternatives.
Another question that needs to be addressed is what 32 www.cepmagazine.org August 2004 CEP purge rate is required. Often the side draw contains good
product that needs to be recycled after the accumulating
impurity is removed. If most of the good product is recovered from the side draw, there is little issue with drawing a
larger side draw flowrate than needed. But if some of the
product cannot be recovered from the side draw, there is an
incentive to minimize the side draw flowrate. This is
achieved at the risk of not removing enough to fully mitigate
the accumulation. Good control can play an important role
here, as shown in Case 15143.
Cases 2.10 (during normal operation), 1038, 1039,
751, 1213, 228 and DT22.14 (Table 1) report success
with side draws. Removing the component from the feed
It is surprising how often this can be the best solution.
The tower in Figure 4, in its initial years of operation,
experienced no hiccups. The problem started when the solvent composition changed. During the initial operation, the
tower feed had twice the ethanol concentration and half
the n-propanol concentration, making n-propanol accumulation far less.
A similar experience occurred in an ethanol/water separation tower that normally received feed from a process that
hydrated ethylene and was lean in heavier alcohols (like
propanol and butanol). Occasionally, the plant processed a
low-cost feedstock from fermentation rich in the heavier alcohols. With this feedstock, severe hiccups were experienced.
Cases DT2.22 and DT15.2 in Table 1 and 217 in Table 2
are also cases in which an accumulation problem was cured
by removing the accumulating component from either the
feed or the reflux drum. This is also the cure contemplated
for Case 1040 (29).
Modifying the tower and internals
A large number of stages is conducive to accumulation.
In one case, doubling the number of separation trays
tremendously intensified tower hiccups. In another, hiccups
were avoided in a new tower by reducing the number of
stages (more reflux, well above the normal optimum),
which allowed the same product purities to be maintained.
In some cases, (e.g., DT15.2) accumulation led to foaming. Addressing the foaming issue (for instance, by enlarging downcomers or injecting antifoam) helped alleviate the
When the main issue with the accumulation is corrosion,
changing materials of construction has effectively mitigated
the effects in many cases (e.g., DT2.9, DT2.11). Several
refinery deethanizers containing stainless steel internals
experienced no significant corrosion, whereas those with carbon steel internals often do. Living with the problem
In many situations, it is uneconomical to cure the problem. At times the product loss is minimal and the problem
is just an operating nuisance. Steps such as improving
controls, minimizing the accumulating component in the
feed by trimming upstream units, reducing or rerouting
recycle streams, changing product specs, and providing
off-spec and storage facilities have helped to reduce hiccup frequency and intensity and minimize product losses. Epilogue
Regardless of whether you decide to live with an accumulation problem or attempt to solve it, accumulation
remains a potential source of erratic operation, flooding,
cycling, foaming and corrosion. It is therefore important
to understand the principles, be familiar with the industry’s experience, and correctly diagnose and address accumulation each time it appears.
CEP HENRY Z. KISTER is a Fluor Corp. Senior Fellow and director of fractionation technology (3 Polaris Way, Aliso Viejo, CA 92698; Phone: (949) 349-4679; E-mail:
email@example.com). He has over 25 years of experience in design, troubleshooting, revamping, field consulting, control and startup of fractionation
processes and equipment. Previously, he was Brown & Root’s staff consultant on fractionation, and also worked for ICI Australia and Fractionation Research
Inc. (FRI). He is the author of the textbooks “Distillation Design” and “Distillation Operation,” as well as 70 technical articles, and has taught the IChemEsponsored “Practical Distillation Technology” course 260 times. He is a recipient of Chemical Engineering magazine’s 2002 award for personal achievement in
chemical engineering and of the AIChE’s 2003 Gerhold Award for outstanding contributions to chemical separation technology. He obtained his BE and ME
degrees from the Univ. of New South Wales in Australia. He is a Fellow of IChemE, a member of the AIChE, and serves on the FRI Technical Advisory and Design
Practices Committees. Literature Cited
14. 15. Kister, H. Z., “Distillation Operation,” McGraw-Hill, NY
Saletan, D., “Avoid Trapping Minor Components in
Distillation,” Chem. Eng. Progress, 91 (9), pp. 78–81 (Sept.
Stupin, W. J., and F. J. Lockhart, “The Distribution of NonKey Components in Multicomponent Distillation,” Presented at
the AIChE 61st Annual Meeting, Los Angeles, CA (Dec. 1968).
Kister, H. Z., R. Rhoad, and K. A. Hoyt, “Improve VacuumTower Performance,” Chem. Eng. Progress, 92 (9), pp. 36–44
Kister, H. Z., “Distillation Troubleshooting,” to be published,
Wiley, Hoboken, NJ (2005).
Barletta, T., and S. Fulton, “Maximizing Gas Plant Capacity,”
Petroleum Technology Quarterly, p. 105 (Spring 2004).
Golden, S. W., “Troubleshooting a Debutanizer Condenser
System,” Presented at the AIChE Spring Meeting, New
Orleans, LA (Mar. 1998).
Kister, H. Z., and T. C. Hower, Jr., “Unusual Case Histories
of Gas Processing and Olefins Plant Columns,”
Plant/Operations Progress, 6 (3), p. 151 (1987).
Severance, W. A. N., “Advances in Radiation Scanning of
Distillation Columns,” Chem. Eng. Progress, 77 (9), p. 38
Eckert, J. S., “Selecting the Proper Distillation Column
Packing,” Chem. Eng. Progress, 66 (3), p. 39 (1970).
Hollowell, W. R., “Three Phase Distillation Calculations and
Trapped Components,” in Kister, H. Z., “Distillation
Troubleshooting,” to be published, Wiley, Hoboken, NJ (2005).
Laird, D., and J. Cornelisen, “Control-System Improvements
Expand Refinery Processes,” Oil and Gas Journal, p. 71
(Sept. 25, 2000).
Ross, S., and G. Nishioka, “Foaming of Binary and Ternary
Solutions,” Journal of Physical Chemistry, 79 (15), p. 1561
Kister, H. Z., T. C. Hower, R., P. R. de M. Freitas, and J.
Nery, “Problems and Solutions in Demethanizers with
Interreboilers,” Proceedings of the 8th Ethylene Producers
Conference, New Orleans, LA (1996).
Freeman, L., and J. D. Bowman, “Use of Column Scanning
to Troubleshoot Demethanizer Operation,” Presented at the
AIChE Spring National Meeting, Houston, TX (Mar. 1993). 16. Lieberman, N. P., “Drying Light End Towers is Critical for
Preventing Problems,” Oil and Gas Journal, p. 100 (Feb. 16,
17. Friedman, Y. Z., “First Principles Inference Model Improves
Deisobutanizer Column Control,” Hydrocarbon Processing, p.
43 (Mar. 2003).
18. Horwitz, B. A., “Hardware, Software, Nowhere,” Chem. Eng.
Progress, 94 (9), pp. 69–74 (Sept. 1998).
19. Russel, R., “Column Simulation,” discussion of Horwitz (18),
Chem. Eng. Progress, 95 (2), p. 7 (Feb. 1999).
20. Taylor, G. L., and M. Cleghorn, “Naturally Occurring
Radioactive Materials: Contamination and Control in a WorldScale Ethylene Manufacturing Complex,” Presented at the
AIChE Spring National Meeting, New Orleans, LA (Feb.
21. Lieberman, N. P., “Design Processes for Reduced
Maintenance,” Hydrocarbon Processing, 58 (1), p. 89 (1979).
22. Sloley, A. W., “Liquid Levels and Density,” presented at the
Chemical Engineers Resource Page, www.cheresources.com.
23. Lieberman, N. P., and G. Liolios, “HF Alky Unit Operations
Improved by On-Site Troubleshooting to Boost Capacity,
Profit,” Oil and Gas Journal, p. 66 (June 20, 1988).
24. Lieberman, N. P., “Process Design for Reliable Operation,”
2nd ed., Gulf Publishing, Houston, TX (1988).
25. Hower, T. C. Jr., and H. Z. Kister, “Solve Column Process
Problems,” Part 1, Hydrocarbon Processing, p. 89 (May
26. Reid, J. A., and C. Wallsgrove, “Odd But True Stories of
Troubleshooting Olefins Units,” in “Distillation: Horizons for
the New Millennium,” Topical Conference Preprints, p. 141,
AIChE Spring National Meeting, Houston, TX (Mar. 1999).
27. Duarte, R., M. Perez, and H. Z. Kister, “Combine
Temperature Surveys, Field Tests and Gamma Scans for
Effective Troubleshooting,” Hydrocarbon Processing, p. 69
28. Drew, J. W., “Distillation Column Startup,” Chem. Eng., p.
221 (Nov. 14, 1983).
29. Chokkarapu, K., R. Eguren, and R. N. MacCallum,
“Selection and Evaluation of Upgrading a Glycol Dehydration
System in High CO2 Gas Service,” Proceedings of the 54th
Laurence Reid Gas Conditioning Conference, p. 43, Norman,
OK (Feb. 22–25, 2004). CEP www.cepmagazine.org August 2004 33 ...
View Full Document
- Fall '08