MASS TRANSFER: DISTILLATION

Trouble-free design of
refinery fractionators
A review of factors most frequently the cause of distillation towers falling short
of design objectives. Analysis of case histories provides guidelines for identifying
potential troublespots in the most important fractionators
Henry Z Kister
Fluor Corporation

A
two-phase survey was recently refinery fractionators,
completed by Fluor of all the case and these form the
histories related to malfunctions in basis for the current
refinery towers that have been document- analysis.
ed over the last 50 years. Altogether, 400 As with other Fluor
case histories were found in the literature. surveys, certain
The first phase identified the most com- ground rules were
mon root causes of problems in refinery applied to limit the
fractionators (towers), but did not exam- scope. Only specific
ine the troublespots in each specific ser- incidents were
vice. This phase yielded general included. For exam-
guidelines for trouble-free design, but did ple, a statement such
not address issues related to each specific as “leakage from
fractionator. chimney trays in
In the second phase, case histories of refinery vacuum tow-
tower malfunctions are analysed specifi- ers can be reduced by
cally for each of the major refinery frac- seal-welding” does
tionators. Each case history teaches a not constitute a case Figure 1 Uplifted packing in wash section of a vacuum tower
lesson. Together, these lessons are the best history. On the other
tool for understanding the potential trou- hand, a statement such as “one vacuum It clearly shows that the vacuum tower is
blespots in each service, and for drawing tower experienced severe chimney tray by far the most troublesome refinery ser-
guidelines for trouble-free design of each leakage at low-rate operation. Seal weld- vice, which is where the survey begins.
service. ing tray sections reduced leakage to
I have previously described the Fluor acceptable levels” does. Vacuum tower malfunctions
survey methodology in Distillation Opera- Also, incidents of corrosion and foul- The 86 case histories reported for the vac-
tion (McGraw-Hill, New York,1990). All ing were included only if a feature unique uum tower is almost double the number
the case histories used as a basis for the to the column design, operation, or con- reported for the atmospheric crude tower,
survey were extracted from the published trol contributed to their occurrence. For which is the next most troublesome refin-
literature. There were 900 total cases, of instance, an incident where the wrong ery tower. When a vacuum tower per-
which 400 were for refinery towers. In corrosion inhibitor or antifoulant was forms poorly, valuable distillate is lost to
about one quarter of these, the specific applied does not qualify as a case history the resid, and poor distillate quality poi-
service was not stated or the service was in this survey. A case where fouling was sons FCC catalyst. The wash section of
one that did not have enough cases caused by insufficient liquid flow, maldis- the fractionator is the most critical sec-
reported on to permit detailed analysis. tribution, or poor process control, does. tion and also one where most of the mal-
This left about 300 cases for the main Finally, optimisation case studies
(where capacity was raised or pressure Top causes of vacuum tower
Fractionator malfunctions drop lowered by replacing trays by pack- malfunctions
ings) are outside of the scope of the sur-
Number of vey. The objective of the current survey is No. Description Cases
cases to identify the issues that make towers fall
1 Damage 27
1. Vacuum towers 86 short of achieving these design capacities.
2 Coking 21
2. Atmospheric crude fractionators 45 There is some overlap in the tabulation
3 Intermediate draws 17
3. Debutanisers 37 of cases for each fractionator. For 4 Misleading measurements 10
4. FCC main fractionators 33 instance, a coked chimney tray case study 5 Plugging 9
5. Deethanisers 23 will be listed once under “coking” and – Installation mishaps 9
6. Depropanisers, C3/C4 splitters 22 another time under “intermediate draws”. – Abnormal operation (startup,
7. Alky main fractionators/isostrippers 17 This means that adding the individual shutdown, commissioning) 9
8. Coker main fractionators 15 malfunctions may yield a number greater 8 Maldistribution 6
9. Naphtha splitters 11
than the number of malfunctions report- – Weeping 6
10. Deisobutanisers 8
ed for the service. Table 1 lists the main 10 Condenser 4
11. Amine towers 8
fractionators surveyed and the concise
Table 1 number of cases reported for each service. Table 2

109
P T Q AUTUMN 2003
w w w. e p t q . c o m
 MASS TRANSFER: DISTILLATION

functions were reported. The wash sec-

tion of the vacuum tower is therefore the
most troublesome tower section in the
refinery (Figure 1, on previous page).
Table 2 shows the common malfunc-
tions reported in vacuum towers. With 27
case histories, damage tops the list. Most
of this damage can be readily prevented.
Table 3 shows the most common causes.
Foremost are water-induced pressure
surges, which account for one third of the
reported damage incidents. In three of the
nine cases reported, the source of water
was poor draining of stripping steam
lines. In another three, pockets of water
lying in the piping of spare pumps
entered the hot tower when these pumps
were connected to the hot tower.
A lesson in this case is that many, pos-
sibly most, vacuum tower damage inci-
dents can be prevented by design and
operating procedures that adequately
drain the tower steam lines in wet towers, Figure 2 Broken, non-standard flange in the spray header supplying wash oil to the
and that positively prevent water from wash bed of a vacuum tower
spare pump piping from entering the
tower. A joint designer/refiner “hazop” flanges and gaskets and properly allowing sory Committee, San Antonio, Texas, Nov
should focus on these troublespots. for thermal expansion in the header 2001].
The next source of damage in Table 3, design. Coking of the wash section (Figure 3) is
insufficient mechanical strength, is also Damage due to high base level con- a close second in Table 2 with 21 reported
readily preventable. It should be recog- tributed three out of the 27 damage case case studies. Table 4 gives a breakdown of
nised (as can be readily seen from Table 2), histories. This again is an issue that can be the causes. Excess stages and vaporisation
that damage is a major issue in a vacuum at least alleviated by good level monitor- occur in wash beds that are either too tall
tower, and that heavy duty internals ing, alarms, and well-designed trip sys- or contain packings that are too efficient.
design should be used. Although the tems. Another three damage-related In either case, the additional stages inten-
heavy duty design would not be able to case-histories were caused by packing sify the vaporisation of the wash oil, leav-
withstand a major pressure surge, it would fires. This type of damage is more difficult ing little liquid to reach and wet the lower
weather the smaller pressure surges. Some to prevent due to the difficulty of clean- sections of the bed. These lower sections
good heavy duty design practices have ing the packings, especially when coked. of the bed dry and coke. Poor modelling
been described by Shieveler [Shieveler G H, Nonetheless, much progress has been and simulation is another cause of coking.
Use heavy-duty trays for severe services; Chem reported in developing preventive mea- Golden et al stress that the heavy ends of
Eng Progr, Aug 1995]. sures, and is discussed in two excellent the crude must be correctly characterised
Special attention should be paid to grid papers by Bouck and Markeloff [Bouck D in the simulation and that the feed entry
installation and tightening. In two of the S, Vacuum Tower Packing Fires; API Operat- to the tower must be modelled by a series
five cases, poorly fastened grids disinte- ing Practices Symposium, 27 April 1999. of flash steps that correctly represent the
grated in service. Through-bolting has Markeloff R, Packing fires; FRI Technical Advi- physical sequence of steps between the
been far more effective than J-bolting for
keeping grid together, and should be rou-
tinely specified.
Spray distributors and their headers are
prone to damage (Table 3). Again, this
damage can be easily prevented by sound
design, good installation, and thorough
inspection and testing. Water testing
spray nozzles and headers can readily
detect damage (Figure 2). Header damage
can be prevented by using standard

Causes of damage in vacuum

fractionators

No. Description Cases

1 Water-induced pressure surges 9
2 Insufficient mechanical strength 5
3 Broken nozzles or headers of
spray distributors 4
4 High bottom liquid level 3
– Packing fires 3

Table 3 Figure 3 Coking of fouling resistant grid in the wash section of a vacuum tower

110
P T Q AUTUMN 2003
 MASS TRANSFER: DISTILLATION

chimney tray that caused entrainment to be troublesome in three cases. With

Causes of coking in vacuum into the wash bed. Three cases were trapout trays being used only in tray tow-
fractionators reported where maldistribution of vapour ers, this issue is experienced mainly in
or liquid led to coking. vacuum lube towers that have trays and
No. Description Cases Intermediate draw malfunctions is a not packing. The remaining case histories
1 Excess stages and vaporisation 4 close third of the most common vacuum describe coking and excessive hydraulic
– Poor modelling and simulation 4 tower malfunctions, with 17 reported gradients.
3 Insufficient wash, reason not case histories (Table 5). Foremost is leak- With 10 case histories, misleading mea-
reported 3 age from total draw chimney trays. Leak- surements are in the 4th spot in Table 2.
– Misleading measurement 3 age of the HVGO chimney tray represents Three of these 10 are the troublesome
– Liquid, vapour maldistribution 3 good distillate degraded into resid with chimney tray level measurements previ-
Not reported 4 no beneficial effects whatsoever, as leak- ously mentioned. Other troublesome
ing liquid poorly distributes in the wash cases have been reported with short coil
Table 4 bed and does little washing. Leakage of outlet thermocouples (two cases), ambi-
LVGO into the HVGO section lowers the ent changes affecting vacuum measure-
heater outlet and flash zone [Golden S W, HVGO boiling point and can reduce heat ments with ordinary gauges (two cases),
Vacuum Tower Troubleshooting; AIChE Spring transfer, even limiting vacuum on the bottom level, heater fuel flow rate, and
Meeting, 1994]. tower. reflux to a packed bed distributor. The
When these principles are overlooked, This leakage needs to be avoided with lessons from these case histories to instru-
the simulation underestimates wash oil totally seal-welded chimney trays. Special ment specifications are self-explanatory.
vaporisation, leading to the drying up and techniques, as recommended by Lieber- With nine case histories, plugging (as
coking reported in four cases. man, are effective and need to be incor- distinct from coking) is in the 5th spot in
In three other reported coking inci- porated to avoid tray buckling due to Table 2. Of the nine cases, five were plug-
dents, it was stated that the wash flowrate thermal expansion [Lieberman N P, Process ging of spray headers. One case was
was insufficient but no specific reason Design for Reliable Operation, 2nd ed; Gulf Pub- reported of plugged packing, plugged
was given. It is likely that in those cases lishing, Houston, Texas, 1988]. quench pipe, plugged instrument line
too, either the number of stages was Level measurement on chimney trays and plugged ejector. In two cases, the
excessive, or the modelling/simulation has been troublesome in three reported plugging was by corrosion products. Mea-
were poor, or both. In three other cases, cases. This may lead to overflow or sures found effective for alleviating plug-
coking was produced by either a mislead- entrainment. While overflow is equiva- ging in the wash spray headers, which is
ingly low coil outlet temperature signal lent to leakage, entrainment from the one of the most common troublespots,
that caused excessive firing, or a faulty overflash chimney tray can induce cok- are to provide good wash oil filtration and
level measurement on the overflash ing. Leaking trapout trays were reported to specify an all-stainless-steel wash oil

111
P T Q AUTUMN 2003
 MASS TRANSFER: DISTILLATION

fluid dynamics (CFD) is essential. Improv-

Intermediate draw malfunctions ing the vapour horn design has eliminat- Top causes of FCC main
in vacuum towers ed operating problems and improved the fractionator malfunctions
performance of several vacuum towers.
No. Description Cases A surprisingly high number of case
histories place tray weeping in the equal No. Description Cases
1 Leaking total draw chimney trays 6 1 Plugging 9
2 Level measurement on total draw 8th spot in Table 2. This is not an issue
with most vacuum towers that are all – Abnormal operation 9
chimney trays 3
3 Liquid maldistribution to
– Leaking trapout tray 3 packed, but appears to be a major issue packing liquid 6
4 Chimney tray coking 2 with the trayed vacuum towers, mostly
4 Intermediate draws 5
– Excessive hydraulic gradients on in lube service. In two of these, leakage
5 Water-induced pressure surges 4
chimney trays 2 at a draw tray made it impossible to draw
6 Pressure control 3
6 Others 1 sufficient product. In two others, weep- – Vapour maldistribution 3
age at pumparound trays starved the
Table 5 pump and reduced heat transfer. In two
more, poor separation between side-cuts Table 7
line downstream of the wash oil filter. resulted. Blanking valves, using high
Sharing the 5th spot in Table 2 is installa- turndown valves, and in the case of the and air premature upon shutdown,
tion mishaps. Three case histories were side cuts, drawing from seal-welded another from a packing fire at the
associated with spray header installation, chimney trays, was the solution. turnaround. The cause of damage in the
and another three with grid or packing Condenser issues complete Table 2. other three cases was not reported.
assembly. Other cases describe problems Condenser problems raise pressure in the Seven abnormal operation (startup/
with tower out-of-roundness, using car- tower and thus reduce distillate recovery. shutdown/commissioning) incidents
bon steel bolts where stainless steel was Two of the four cases reported dealt with were reported. Four of these resulted in
specified and poor installation of strip- excess lights, one with ejector plugging, four of the damage incidents listed, one
ping trays. and one with flash equilibrium at the pre- led to an explosion, another to a fire and
Also sharing the 5th spot in Table 2 is condenser. one to a chemical release. Three of the
abnormal operation incidents. Five of pressure surges listed in Table 6 under
these describe incidents during startup Crude tower malfunctions Damage resulted from poor dehydration
where a pocket of water entered the hot The three most common atmospheric during startup or pump switchover. Poor
tower and created a pressure surge. Poor crude tower malfunctions (Table 6) are blinding and unblinding led to one
blinding/unblinding contributed two case plugging, intermediate draw malfunc- reported case of explosion and another of
histories, one case is related to pressur- tions and damage. chemicals release.
ing/depressuring and one to flushing. There were nine plugging incidents The next four entries in Table 6 are well
Maldistribution problems, other than reported in atmospheric crude towers: below the top four, and have three to
those attributed to coking, plugging or Four in the wash section or gasoil four reported malfunctions. Four installa-
damage, are in the 8th spot in Table 2 pumparound, three in the top section or tion mishaps were reported, all involving
with a surprisingly low number of case top pumparound, and two in the strip- trays or chimney trays. Leaks of a
histories (six). Of the reported six, four ping section. No cases of plugging were pumparound exchanger, a pump seal,
were vapour maldistribution, three of reported in the middle of the tower. In the and resid to atmosphere were the three
these originating in the flash zone and wash section, the most common cause of reported leak cases. Controlling liquid
one in the previously mentioned tray plugging was entrainment from the flash flow to the wash section has been a spe-
chimney. The other two cases reported zone or from the vapour overhead of a cial challenge, contributing three more
liquid maldistribution problems. Fluor’s preflash drum. In the top of the tower, the case histories.
experience has been that maldistribution, plugging was by scale and corrosion prod- Trouble-free designs properly dis-
especially of vapour from the flash zone, ucts, corrosion inhibitors, and salting out. entrain the vapour in the flash zone and
has been far more troublesome than sug- Five of the nine incidents resulted in preflash drum and properly desalt the
gested by the low spot of this item in plugged trays, two in plugged downcom- crude in order to minimise plugging in
Table 2. ers. In one case, a packed bed plugged, in the fractionator. Specifying fouling-resis-
Vapour horn design and good distribu- another, a liquid distributor to the pack- tant hardware in the wash zone and
tion of liquid to the wash bed are central ing plugged. upper trays is good practice. Downcomer
for achieving trouble-free performance of The large number of intermediate draw trapouts and chimney trays are the most
the wash bed. An expert hydraulic analy- incidents is well in line with Fluor’s expe- important internals for ensuring trouble-
sis, often with the aid of computational rience: chimney trays and downcomer free operation. They need to be designed
trapouts make or break fractionators. Of and inspected carefully, not just left to
Top causes of atmospheric crude the nine, seven took place with down-
tower malfunctions comer trapouts, two with chimney trays. Top causes of malfunctions in
Four of these involved choking or restric- debutanisers, incl stabilisers and
tion in the outlet liquid line, while in two depentanisers
No. Description Cases
others, leakage at the drawoff restricted
1 Plugging 9
the recovery of a side cut. No. Description Cases
– Intermediate draws 9
Four of the nine damage incidents 1 Control 10
– Damage 9
4 Abnormal operations 7 reported were due to water-induced pres- 2 Vapour cloud release 5
5 Installation mishaps 4 sure surges. Two of these were caused by – Installation Mishaps 5
6 Condenser Problems 3 undrained stripping steam lines, one by a 4 Feed arrangement, tray towers 4
– Poor control of wash 3 water pocket in a spare pump, and one by – Reboiler draw arrangements 4
– Leaks 3 plugged drainholes in the bottom seal
pan. One case of damage resulted from
Table 6 exposing column internals to cold water Table 8

112
P T Q AUTUMN 2003
 MASS TRANSFER: DISTILLATION

incidents.There were nine plugging inci-

Top causes of malfunctions in dents reported in FCC main fractiona- Top causes of malfunctions in
C3/C4 splitters and depropanisers tors. Of these, four were salting-out deethanisers (absorbers and
(excl those in alky units) incidents that plugged trays near the top strippers)
of the tower, and were overcome by
online water washes. Three were inci- No. Description Cases
No. Description Case dents in which grid in the slurry section 1 Reboiler draw and return
1 Reboiler draw and return coked up due to vapour or liquid-maldis- arrangements 6
arrangements 7 tribution. One incident was plugging due 2 Excessive tower base level 4
2 Tower flooding by excess to catalyst carryover into upper packed – Control 4
base level 6
sections, and another was plugging of a – Component accumulation 4
3 Vapour cloud release 5 line draining the main feed line, both – Side draw arrangements 4
Table 10 during startup.
There were also nine abnormal opera- Table 9
others, as these will make or break the tion (startup/shutdown/commissioning)
fractionator. incidents reported, two of which were FCC main fractionators than in atmo-
Prevention of water entry, by ensuring previously described. These two inci- spheric crude fractionators, so it comes as
adequate drainage on stripping steam lines dents, plus two others, occurred during little surprise to find liquid maldistribu-
and eliminating dead pockets inside the liquid circulation and dehydration. In tion to packings in a prominent spot in
tower, is central for damage prevention. three of these four, a pressure surge and Table 7. Of the six reported incidents, two
Proper inspection of equipment is a must major damage resulted, the other was the involved liquid maldistribution to the
to prevent the installation mishaps. Proper catalyst carryover. The remaining five slurry pumparound section, the others to
control of the liquid flow rate to the wash incidents include poor unblinding caus- various fractionation sections. Intermedi-
section is the prime control consideration ing a toxic release; switching over oxygen ate draws in FCC main fractionator have
in the tower. and nitrogen purge gas causing explo- been troublesome in five reported case
sions; trip failure on the reflux drum caus- histories, more in chimney trays than in
Main column malfunctions ing liquid carryover and major downcomer trapouts. Finally, four cases
There are some similarities with regard to compressor damage; a major leak due to of water-induced pressure surges were
FCC main fractionator malfunctions the thermal shock while opening or clos- reported, three of which led to major
(Table 7) and atmospheric crude tower ing the valve in the tower inlet; and a damage.
malfunctions, but there are also major startup pressure control problem resulting Two other malfunctions are also shown
differences. Two malfunctions top the from steam condensation. in Table 7: Vapour maldistribution, all
list: plugging and abnormal operation Packings are used more frequently in cases dealing with grid in the slurry sec-

113
P T Q AUTUMN 2003
 MASS TRANSFER: DISTILLATION

tion, and pressure control. Plugging and There is a distinct link between the or pan causing liquid to bypass a once-
coking can be alleviated by providing ade- feeds and reboiler draw arrangements. through thermosiphon reboiler (two
quate on-line wash facilities near the top Both constitute “points of transition”, ie, cases); a reboiler tube leak and slug flow at
of the fractionator, by using plugging- where a stream enters or leaves the tower. the reboiler outlet pipe.
resistant trays there, and by preferring These points of transition are some of the Slightly behind, with six case histories, is
shed decks or disk and donut trays to grid major troublespots in a tower. The lesson tower flooding by excess base level. Two of
in the wash section. Shed decks and disk for debutanisers is that all points of transi- these resulted from the type of reboiler
and donut trays are far less sensitive to tion need to be critically examined for problems previously discussed. False level
vapour or liquid maldistribution than potential bottlenecks, both at the design indica tions led to two others, and frothing
grid, and therefore far less prone to cok- (or debottleneck) and when trouble- or foaming at the tower base led to the
ing during upsets. shooting. remaining two. Clearly, the lessons learned
Startups and shutdowns are major are that troubleshooting and trouble-free
issues in FCC main fractionators, and a Deethaniser malfunctions debottlenecks of C3/C4 splitters should focus
good design of these needs to hazop what The strippers and absorbers are included in on the reboiler piping and the bottom
can go wrong and take preventive mea- the deethaniser malfunctions. Topping the sump.
sures. Intermediate draws and liquid dis- list (Table 9) with six case histories are Similar to debutanisers, depropanisers
tributors are the weakest links in the reboiler draw and return arrangements. and C3/C4 splitters have experienced a high
internals design, and need to be designed Three of the six cases report excessive pres- number of vapour cloud releases, mostly
and inspected carefully, not just left to sure drop in the process inlet or outlet due to line rupture (three cases), but also
others. Finally, pressure controls as well as pipes of a kettle reboiler. The high-pressure due to poor blinding or plugging/freeze ups
liquid flow control to the wash section drop either caused the tower base liquid of valves. Some major blasts resulted. The
are major considerations in these frac- level to rise above the reboiler return inlet, vapour cloud lessons described under
tionators. or back liquid up on the chimney tray feed- debutanisers extend to depropanisers and
ing the reboiler to the top of the chimneys. C3/C4 splitters.
Debutaniser malfunctions Insufficient heat during coke drum
Due to similar functions, stabilisers and switchover was reported in two cases, one Other fractionators
depentenisers have been lumped together of them due to weeping from the draw tray For other refinery fractionators, the num-
with debutanisers. Over 70% of the cases, to a once-through thermosiphon reboiler. ber of case studies reported was less than
however, were contributed by debutanis- Four case histories were reported of base 20, a sample too small for a detailed anal-
ers. level exceeding the reboiler return. Two of ysis. Nonetheless, some observations are
Table 8 shows that the most common these were due to high-pressure drop in significant and require more detail, includ-
malfunctions experienced in debutanisers the kettle piping (those previously men- ing coker main fractionators, alky unit
are widely different from those experi- tioned), the other two due to absence of or main fractionators/isostrippers, naphtha
enced in the vacuum, crude and FCC frac- to poor level indication. As with debu- splitters, deisobutanisers and amine
tionators. Topping the list with 10 case tanisers, control issues are also important absorbers/regenerators.
histories is controls, an item that showed in deethanisers, and account for four case A total of 15 malfunction case histories
low down (if at all) on the main fraction- histories. Also, with four case histories, of coker fractionators have been report-
ator malfunctions list. Of the 10 cases, five component accumulation in deethanisers ed. Of the 15, seven described fouling by
reported difficulties with pressure and con- is a problem. coking or carryover of coke, while five
denser controls. In all five, a total con- Either ethane or water or both accumu- others described damage due to water-
denser was used with partial flooding of late and can lead to cycling, capacity bot- induced pressure surges. There is no
the condenser. In two of the five, the prob- tlenecks, and in the case of water, also doubt that coking and water-induced
lem was induced by presence of non-con- corrosion. Finally, choking of side draws pressure surges are the major issues with
densables. Composition control or the with entrained gas bubbles has been a these fractionators.
assembly of a control system contributed problem in four case histories. A total of 17 malfunctions have been
the other control case histories. The lessons learned from this documen- reported for alky unit main
Vapour cloud release and installation tation are that the points of transition in fractionators/isostrippers. Of these, four
mishaps share the second spot in Table 8. deethanisers (the side draws as well as the described plugging, mostly by scale or cor-
Three of the five case histories of vapour region below the bottom tray, including rosion products; three described explo-
clouds ended in explosions, and one more the reboiler draw and return lines) require sions, either due to vapour cloud release or
in a fire. Some of these were accompanied thorough design, review, and inspection, due to HF carryover in the hydrocarbons
by injuries and heavy damage. Line frac- and must not just be left to others. Preven- and a violent reaction in a caustic bed
ture (two incidents), poor blinding (two tion of component accumulation and care- downstream; and three others described
incidents), and freeze-ups in leaking ful review of the control systems are also accumulation of either ethane or water in
valves (two incidents) were some of the prime considerations that make the differ- the overhead system.
contributing factors. Hazops of debutanis- ence between a troublesome and trouble- A total of 11 naphtha splitter malfunc-
ers should consider some of the lessons free deethaniser. tions have been reported. Four of these
learned from previous vapour cloud reported plugging, mainly by scale and cor-
releases to positively eliminate further Splitter malfunctions rosion products; three reported reboiler
accidents. Malfunctions in C3/C4 splitters also include issues; two were a result of poor installa-
With four case histories, poor feed depropanisers other than those in alky tion; and two reported control problems.
arrangements closely follow, leading to a units, which are uniquely different (Table Control problems with other refinery
capacity bottleneck or an efficiency loss in 10). Similar to deethanisers, reboiler draw fractionators have also been reported. For
the feed region. Also with four case histories and return arrangements lead the list with example, of the eight total deisobutaniser
are reboiler draw arrangements, including seven reported cases. Again, the main prob- malfunctions that have been reported, six
vapour entrainment choking the reboiler lems have been excess pressure drop in involved control problems. Five of these
draw lines and liquid leaking from a trapout inlet and outlet lines of a reboiler causing were temperature control issues that can
tray to a once-through thermosiphon base liquid level to exceed the reboiler be particularly troublesome with narrow-
reboiler, thus “starving” the reboiler. return inlet (two cases); leaking draw tray boiling mixtures.

114
P T Q AUTUMN 2003
 MASS TRANSFER: DISTILLATION

Eight case histories were specifically Plugging, abnormal operation issues,

reported for refinery amine absorbers. Of liquid maldistribution to packings and
these, six reported foaming, and three grid, intermediate draws and water-
reported scale and corrosion products induced pressure surges are the key trou-
catalysing the foam and causing plugging. ble spots in FCC main fractionators. The
Seven other case histories of amine main causes of plugging in these fraction-
absorbers and/or regenerators were report- ators are salting out near the top and cok-
ed without stating whether they came ing of grid beds in the slurry section.
from refineries or natural gas plants. Of These can be alleviated by online water
these, foaming was the issue in five. There washes and by avoiding grid in the slurry
is no doubt that foaming, and to a lesser section, respectively. Tower dehydration
degree fouling with scale and and corro- and liquid circulation are important start-
sion products, is the prime issue in amine up operations that can turn troublesome
towers. and lead to pressure surges or catalyst car-
ryover into the less fouling-resistant
Lessons learned regions.
The vacuum tower is by far the most trou- Control issues, especially pressure con-
blesome refinery fractionator. Damage, trols, are the primary source of problems
wash bed coking, and intermediate draws in debutanisers. Vapour cloud releases
are prime trouble spots. Water-induced have led to explosions and major damage
pressure surges are the leading cause of in debutanisers. Installation mishaps,
damage. Most of the damage incidents in tower feed entry arrangements, and
the vacuum towers are preventable. reboiler draw arrangements have also
Hazoping the possibility of water entry, been major trouble spots. Critical review
using heavy duty mechanical designs, of the control system, especially the pres-
inspecting, testing and correctly design- sure/ condenser controls, learning from
ing spray headers, paying attention to past vapour-cloud accidents, using
level measurement, and applying special hazops to minimise the possibility of
designs and procedures to prevent pack- vapour cloud releases, and sound design
ing fires can drastically reduce damage of feed entry piping and of tower base
incidents. arrangements are the key for trouble-free
Wash bed coking can be alleviated by debutanisers.
using short beds of relatively inefficient In deethanisers, the points of transi-
packing by correctly simulating the wash tion (the side draw arrangements and the
zone and by avoiding insufficient wash. region below the bottom tray, including
Intermediate draw malfunctions can be the reboiler draw and return lines) are
alleviated by seal-welding draw trays and prime trouble spots and are key to trou-
water-testing them at the turnaround and ble-free design and operation. Prevention
by paying attention to reliable level mea- of water and ethane accumulation in this
surement on these trays. Other measures tower is also important.
that promote trouble-free operation are Reboiler draw and return arrange-
good instrument specifications, good fil- ments, and tower base level are the key to
tration of the wash oil followed by stain- trouble-free depropanisers and C3/C4
less steel piping downstream of the filters, splitters. Depropanisers also are prone to
and good distribution of vapour and of vapour cloud releases, and lessons
the wash oil to the wash bed. learned from past vapour cloud incidents
Plugging, intermediate draws, damage should be incorporated in the design and
and abnormal operation incidents are the operation of depropanisers and C3/C4
most troublesome malfunctions in atmo- splitters.
spheric crude towers. Plugging is most Trouble-free operation of coker frac-
common near the tower top, at the wash tionators focuses on preventing coking
zone, or in the stripping section. Choking and water-induced pressure surges; of
and leakage are the prime intermediate alky main fractionators focuses on plug-
draw issues. Water-induced pressure ging, vapour cloud and component accu-
surges are the most common cause of mulation prevention; of deisobutanisers
damage. Plugging problems can be allevi- on composition control; and of amine
ated by eliminating the source of fouling absorbers and regenerators on foaming
and/or by using plugging-resistant inter- and plugging prevention.
nals. Downcomer trapouts and chimney
trays need to be designed and inspected
carefully, not just left to others.
Prevention of water entry by proper Henry Z Kister is a Fluor Corporation
draining of stripping steam lines and by fellow and director of fractionation
sound dehydration procedures at startup technology at Aliso Viejo, California, USA.
is critical. Proper inspection of equip- He has over 25 years’ experience in design,
ment and adequate control of liquid control and startup of frationation processes
flowrate to the wash section are also and equipment. He obtained his BE and ME
important for promoting trouble-free degrees from the University of New South
operation. Wales, Australia.

115
P T Q AUTUIMN 2003