T. Barletta and J .

Nigg, Process Consul ti ng

Servi ces, I nc., Houston, Texas; and S. Ruoss,
J . Mayfield and W. Landry, Tosco Refi ni ng Co.,
Bel l e Chasse, Loui si ana
F
l oodi ng of the crude di sti l l ati on uni t (CDU)
adversel y i mpacts a refi nerys profi ts. Such
condi ti ons force crude charge reducti ons, down-
gr adi ng of pr oduct dr aws, etc. Consequentl y, the
CDU operates under stress. Determi ni ng the root
cause for the fl oodi ng i s another obstacl e that a
refi ner must overcome. I ncorrect di agnosi s of the
fl oodi ng source can be equal l y as damagi ng as the
root cause. Gatheri ng val i d fi el d data
i s v i tal wh en u n der s tan di n g th e
dynami cs occurri ng wi thi n the crude
col umn. Proper i nterpretati on of the
data i s al so requi red to determi ne the
best recourse to correct the probl em.
I n thi s case hi story, a crude col umn
begi ns fl oodi ng. Severe upsets reduced
crude throughput. Fi el d pressure drop
data are used to fi nd the root cause for
the fl oodi ng and oper ati onal upsets.
Wi th thi s i nfor mati on, the r ef i ner
i nstal l s a bypass onl i ne to overri de the
col umns restri cti on and restore the col -
umns performance.
Case history. Tosco Refi ni ng Co., a
di vi si on of Tosco Corp., owns and oper-
ates a l arge, hi ghl y i ntegrated si ngl e-
tr ai n r ef i ner y at Bel l e Chasse,
Loui si ana. (Tosco pur chased the Al l i ance Refi ner y
on Sept. 8, 2000.) Thi s faci l i tys CDU had been oper-
ati ng over si x year s si nce the l ast tur nar ound. I n
earl y 2000, the crude col umn began fl oodi ng, whi ch
l i mi ted maxi mum throughput. Thi s fl oodi ng reduced
maxi mum CDU capaci ty by 20,000 to 25,000 bpd (25
Mbpd). Attempts to i ncr ease cr ude char ge r ate or
process l i ghter crude bl ends at reduced throughput
caused severe crude col umn operati onal upsets.
Fi el d measurements i denti fi ed a l i qui d restri cti on
i n the top secti on of the col umn. A bypass l i ne was
i nstal l ed onl i ne i n June 2000. I t al l owed l i qui d to ci r-
cumvent the restriction. Crude throughput was imme-
diately restored by 2025 Mbpd without any operating
upsets or col umn fl oodi ng. The bypass i nstal l ati on
had a simple payout of less than two weeks; Tosco cap-
tured incremental revenue from additional crude runs
wi thout taki ng a uni t shutdown.
Problem symptoms and c onsequenc es. The
Al l i ance Refi nery crude col umn produces ful l -range
naphtha as over head pr oduct and ker osi ne, di esel
and atmospheri c gas oi l (AGO) as product si de draws.
Fi g. 1 shows an overvi ew of the crude col umns upper
secti on, i ncl udi ng the di ameter swage at the di esel
pumparound return. I n earl y 2000, the crude col umn
experi enced severe upsets when the crude rate was
wi thi n 2025 Mbpd of targeted maxi -
mum. These crude column upsets were
characteri zed by:
A i ncr eas e i n pr es s ur e dr op
between the di esel si de stri pper vapor
return tray and the top of the col umn
An i ncr ease i n k er osi ne dr aw
temperature
Loss of l evel i n the kerosi ne si de
stri pper
Hi gh col umn bottoms l evel s.
Duri ng the upsets, operators woul d
decrease heater outl et temperature to
lower the vapor flowrate in the column.
After 3040 minutes, the pressure drop
i n the upper secti on woul d begi n to
decrease and shortl y thereafter kero-
si ne product fl owrate and draw tem-
perature would decrease rapidly. Even-
tual l y the cr ude tower bottoms l evel
woul d i ncrease, and the cycl e woul d start over. The
ti me between the onset of fl oodi ng and recovery to
stabl e operati on was approxi matel y 3
1
2 hours.
To mi ni mi ze these upsets, the di esel pumparound
duty was maxi mi zed at the expense of the top
pumpar ound duty. Thi s r educed i nter nal r ef l ux
between ker osi ne and di esel , whi ch r esul ted i n a
7,900-bpd downgrade i n kerosi ne to di esel product.
Also, the crude tower pressure was increased to reduce
the superfi ci al vapor vel oci ty i n the top of the col umn.
Hi gher crude tower operati ng pressure decreased the
AGO cutpoi nt and rai sed the feedrate to the vacuum
col umn. Heavy crude runs were l i mi ted to stay wi thi n
the maxi mum capaci ty of the vacuum col umn.
HYDROCARBON PROCESSING /JULY 2001
Diagnose flooding
columns efficiently
Using field pressure data, a refiner made a low-capital modification on a
crude distillation column to increase throughput without a shutdown
D
i
s
t
i
l
l
a
t
i
o
n

R
e
f
i
n
i
n
g
An opera ting
problem is
impossible to
fix until the
specific ca use
is found.

Reprinted from: July 2001 issue, pgs 71- 75. Used with permission.

Finding the root c ause. An oper ati ng pr obl em i s

i mpossi bl e to fi x unti l the speci fi c cause i s found. I t
was cl ear that f l oodi ng was occur r i ng somewher e
between the top of the col umn and the kerosi ne si de
stri pper vapor return tray, but where? The col umn was
fl oodi ng, but the speci fi c cause and exact l ocati on was
unknown. The Al l i ance Refi nery wanted to know i f i t
was possi bl e to ci rcumvent the fl oodi ng and i ncrease
crude rate wi thout taki ng a shutdown.
Pressure drop can be used to i nfer fl oodi ng, just as
mor e sophi sti cated methods such as gamma scans.
However, pressure drop can be much easi er to i nter-
pret than a gamma scan. Pressure can be measured at
col umn nozzl es, stri pper vapor l i ne vents, draw nozzl e
l ow poi nt bl eeders, rel i ef val ve pi pi ng, and si de-stri pper
bottoms l evel bri dl es. Fi g. 2 shows the typi cal pl aces
where pressure can be measured.
The crude column pressure profile was measured when
the crude col umn was stabl e and once agai n duri ng an
upset. The pressure drop was measured by taking simul-
taneous readi ngs wi th two di gi tal smart manometers,
whi ch are accurate to 0.05 psi . When measuri ng smal l
pressure di fferenti al s, a si ngl e-gauge pressure survey i s
not accurate because fl uctuati ons i n col umn pressure
result in inaccurate differential pressure measurements.
Pressure profile stable operation. The crude col -
umn pr essur e dr op was measur ed when the col umn
was i n stabl e oper ati on. Fi g. 3 shows the measur ed
pr essur e dr ops. The pr essur e dr op acr oss the top
pumpar ound was about 0.20 psi and was measur ed
over a 40-mi n peri od. I t was stabl e duri ng thi s peri od,
fl uctuati ng between 0.20 and 0.23 psi . At 0.20 psi pres-
sure drop, the top pumparound bed was fl ooded. But
was i t causi ng the crude col umn upsets?
When a crude col umn top pumparound bed fl oods,
the temperature di fference between the tower overhead
and top pumparound draw wi l l decrease. The temper-
ature di fference shoul d be i n the range of 2535F. The
temperature di fference had been l ess than 10F for the
past two years pri or to the upsets as shown i n Fi g. 4.
Therefore, the bed had been operati ng fl ooded duri ng
the enti re ti me. Even though the top pumparound bed
had been operati ng fl ooded duri ng thi s ti me, the uni t
had been abl e to achi eve maxi mum throughputs.
When a pumpar ound bed fl oods, l i qui d stacks up
unti l there i s enough head to push i t out the bottom
of packi ng. Mi nor pumparound bed fl oodi ng wi l l resul t
i n hi gher pressure drop, but typi cal l y does not cause
major operational problems. I f the flooding is not severe,
then addi ti onal pressure drop can go unnoti ced. How-
ever, i f fl oodi ng i s severe, then l i qui d wi l l stack i n the
bed, pr event l i qui d fl ow down the col umn and cause
severe operati ng probl ems.
The pressure drop measured from above the col l ector
tray to below the kerosine side stripper vapor return tray
was approxi matel y 0.148 psi per tray. Thi s i s hi gh for
standard valve trays. More field data was needed to deter-
mine if these trays were flooding and causing the upsets.
Pressure profileupset condition. Fi g. 5 shows the
pressure drop across the top pumparound bed during an
upset. Pressures were measured at one of the 30-in. over-
head l i nes and at a
3
4-i n. val ve l ocated 6 i n. bel ow the
packed bed. The measured pressure drop was swi ngi ng
between 0.2 and 1.75 psi . Yet, the cal cul ated pressure
HYDROCARBON PROCESSING /JULY 2001
Fig. 1. C ru d e c o lu m n u p p e r se c tio n .
Fig. 2. P re ssu re m e a su re m e n t lo c a tio n s.
drop through the overhead nozzl e i s 0.06 psi . Therefore,
the measured drop was essenti al l y across the packi ng.
Pr essur e dr op acr oss the top pumpar ound bed i s
extremely high. This bed is highly loaded/flooded even at
the l owest pressure drop 0.24 psi . The top pumparound
bed i s appr oxi matel y 9 ft of str uctur ed packi ng. I t i s
nearly full of liquid at a pressure drop of 1.75 psi. A pres-
sure drop of 1.75 psi through the bed represents a l i q-
ui d hei ght i n the packed bed of roughl y 6 ft.
I n Fi g. 6, the (measured) top pumparound pressure
drop and the pressure drop between the overhead and
the di esel si de stri pper vapor return tray are compared.
Duri ng the upset, the pressure drop between the col -
umn overhead and thi s tray cycl es between 3.0 psi and
4.7 psi . The fi el d-measured pressure drop across the
top pumpar ound bed i s al so cycl i ng at the same fr e-
quency by roughl y the same amount. The i ncrease i n
pressure drop during the upset is in the top pumparound
bed, not the trays bel ow i t.
The pr essur e dr op i s i ncr easi ng fr om the poi nt of
the downcomer restri cti on. Pressure drop i ncreases as
the l i qui d hol d-up (l evel ) i n the packi ng bui l ds and con-
versel y, the pressure drop decreases as the l i qui d hei ght
decl i nes. When the l i qui d hei ght i n the packi ng bui l ds
suffi ci ent head, the l i qui d dumps down the tower and
the crude tower bottoms l evel i ncreases. As the l i qui d
works i ts way down the tower, product-draw tempera-
tures drop due to composi ti on changes.
Collector tray performance. A si mpl i fi ed sketch of
the top pumparound col l ector tray i s depi cted i n Fi g. 7.
The tray deck i s l ocated 3 ft bel ow the packi ng and 4 ft
above the tray bel ow. Vapor from the tray bel ow fl ows
up through the col l ector tray ri sers and i nto the bottom
HYDROCARBON PROCESSING /JULY 2001
Fig. 3. S ta b le o p e ra tio n p re ssu re d ro p .
Fig. 4. To p P /A d ra w te m p e ra tu re m in u s o ve rh e a d .
Fig. 6. C o m p a riso n o f to p P /A P a n d c ru d e to we r to p P.
Fig. 5. P re ssu re d ro p d u rin g u p se t c o n d itio n s.
of the packed bed. Li qui d from the bottom of the packed
bed drops onto the col l ector tray. A porti on of the l i qui d
i s wi thdr awn fr om the col umn vi a the dr aw nozzl es
and i s pumped thr ough the top pumpar ound heat
exchanger network. Li qui d overfl ows the col l ector tray
i n center and si de downcomers and provi des i nternal
refl ux for the trays bel ow.
The col l ector tray i s desi gned to operate wi th approx-
i matel y 9
1
2 i n. of l i qui d on the tr ay deck. The l i qui d
l evel shoul d be bel ow the ri sers, whi ch are 12 i n. hi gh.
Fi g. 8 shows what the col l ector tr ay shoul d l ook l i ke
wi th the desi gn l i qui d l evel .
Pressure can be used as a l evel i ndi cator. One foot of
water wi l l exert 0.4329 psi of pressure head. Therefore,
i f the el evati on between two pressures and the l i qui d
densi ty i s known, then the l i qui d head can be cal cu-
l ated. Fi g. 9 shows the col l ector tr ay pr essur e mea-
surement locations and the elevation difference between
them. The measured pressure drop was 0.7 psi , there-
fore, the col l ector tray i s 100% ful l of l i qui d and i s back-
i ng l i qui d up i nto the packi ng.
Interpreting field data. Li qui d was backi ng up onto
the col l ector because of a restri cti on i n one or more of
the downcomers. The downcomer cl earance i s 2 i n. The
restri cti on was most l i kel y due to sal ts, pi eces of cor-
roded packi ng and/or scal e that has l ai d down on the
tr ays and par ti al l y bl ocki ng off the ar ea under the
downcomer. The restri cti on was most l i kel y i n the col -
l ector tr ay downcomer s or i n the downcomer s of the
thr ee tr ays di r ectl y bel ow the col l ector tr ay. Fi g. 10
shows a l i qui d col l ector ful l of l i qui d and the probabl e
l ocati on of downcomer restri cti ons.
The col l ector tr ay measur ements wer e cr i ti cal i n
determi ni ng the l ocati on of the restri cti on. One or more
downcomer restri cti ons bel ow the col l ector tray were
preventi ng l i qui d from fl owi ng down the col umn. Wi th-
HYDROCARBON PROCESSING /JULY 2001
Fig. 10. F lo o d e d c o lle c to r tra y.
Fig. 7. To p p u m p a ro u n d c o lle c to r tra y.
Fig. 8. C o lle c to r tra y d e sig n liq u id le ve ls.
Fig. 9. C o lle c to r tra y p re ssu re su rve y.
out thi s i nformati on, one coul d have erroneousl y con-
cl uded ei ther top pumparound bed hydraul i c fl oodi ng
or fl oodi ng of the trays between the col l ector tray and
the kerosi ne si de stri pper vapor return tray was caus-
i ng the upsets.
Pressure drop can be used to i nfer tray or packi ng
performance. Hi gh-pressure drop i mpl i es fl ooded trays
or packi ng, whi l e l ow or no pressure drop i mpl i es dam-
aged or mi ssi ng i nter nal s. A pr oper l y desi gned tr ay
wi l l have a typi cal operati ng pressure drop of 0.070.12
psi per tray.
The measured pressure drops across secti ons of the
col umn ar e shown i n Fi g. 3. The measur ed pr essur e
drop across the secti on from the col l ector tray to the
kerosi ne si de stri pper vapor return tray i ncl udes 0.39
psi of l i qui d head above the ri sers. Therefore, the actual
pressure drop across these trays was 0.116 psi per tray.
The measured pressure drop across the trays between
the ker osi ne si de str i pper vapor r etur n tr ay and the
di esel si de str i pper vapor r etur n tr ay was appr oxi -
matel y 0.08 psi per tray. Thi s does not i ndi cate fl ooded
or mi ssi ng trays.
When trays fl ood from downcomer restri cti ons, as
opposed to acti ve area fl oodi ng, hi gh-pressure drop can
onl y be measured when the downcomers are fl oodi ng.
Thi s i s why measuri ng the col umn pressure drop when
the uni t was i n an upset hel ped di agnose the probl em
and why gamma scans of the col umn were i nconcl usi ve.
Solution. A bypass line was installed online. The instal-
l ati on r equi r ed two hot taps, one on the top
pumparound return l i ne and one on the crude col umn
shel l . The bypass l i ne r outes a sl i pstr eam of top
pumpar ound r etur n l i qui d onto acti ve panel s of the
thi rd tray bel ow the col l ector tray. Fi g. 11 shows the
bypass confi gurati on.
Si nce the bypass l i ne was i nstal l ed, the crude charge
rate has been i ncreased by 2025 Mbpd and crude pro-
cessing flexibility has improved. The bypass has allowed
opti mi zati on of the cr ude col umn pumpar ounds.
Decr easi ng di esel pumpar ound whi l e i ncr easi ng top
pumparound has i mproved fracti onati on between kero-
si ne and di esel pr oducts. Ker osi ne pr oduct yi el d has
i ncreased by 7,900 bpd.
F /5 M /7 -2 0 0 1 A rtic le c o p yrig h t 2 0 0 1 b y G u lf P u b lish in g C o m p a n y. A ll rig h ts re se rve d . P rin te d in U S A .
Tony Barletta is a chemical engineer for Pro-
cess Consulting Services, Inc., Houston,
Texas. The company provides revamp, opti-
mization and troubleshooting services to the
refining industry. He previously worked as a
production planner and process specialist
for BP Oils Alliance Refinery and a process
engineer for Hess Oil Virgin Islands Corp. He
holds a BS degree in chemical engineering
from Lehigh University.
J ason M. Niggis a chemical engineer for Pro-
cess Consulting Services, Inc., Bedford, Texas.
His responsibilities include test run execu-
tion and analysis, revamp process design
and process equipment evaluation. Prior to
joining Process Consulting Services, he
worked for Koch Engineering Co. (now Koch-
Glitsch) as a process design engineer and a
refinery engineer for Koch Refining Co. in
Corpus Christi, Texas. He holds a BS degree
in chemical engineering from Kansas State University.
Stephanie P. Ruoss is currently planning & coordination super-
intendent at the Alliance Refinery. She has worked in the process
engineering and planning groups for the 10 years that she has been
at Alliance. She holds a BS degree in chemical engineering from
Mississippi State University and a MBA from University of New
Orleans.
J erry K. Mayfield is currently a shift superintendent for TOSCO.
He has worked in the operations department at the Alliance
Refinery for 25 years. In operations, he has worked as an oper-
ator, chief operator and asset coordinator.
Walter D. Landry is currently a field shift supervisor for TOSCO.
He has worked in the operations department at the Alliance
Refinery for 25 years. Previously, he has worked as an operator,
chief operator and training coordinator.
Fig. 11. H o t ta p b yp a ss lin e .