CW

Operations & Maintenance

Distillation Clean
solvent
A

Reboiler Startup Feed FIGURE 1.

93% A The extractive
7% B distillation tower
discussed in this article
employs two reboilers

Can Pose Challenges 108°C

#3 ∆Pout
140°C
Kettle
hs, REB reboiler

1 5 Steam
Future designs can benefit from the lessons 109°C
3

H
that were learned via a well-planned Thermosiphon
reboiler
∆Pin 2

#1 109°C
investigation of this startup problem #2 140°C 4

Contaminated solvent
2% A 1% B 97% solvent
Etienne Rubbers, Kirsten Green TABLE 1. DESIGN STREAM COMPOSITIONS
and Terry Fowler, Stream #1 Stream #2 Stream #3
Sasol Technology (Pty.) Ltd.
Liquid to Liquid Vapor
Flashed Flashed
kettle ex kettle ex kettle
Henry Z. Kister and Walter J. Stupin, Temperature,°C 109 124 137 140 140
Fluor
A, wt, % 30 3 2 2 62
B, wt, % 14 2 1 1 29

D
uring the initial startup of a
new extractive distillation C, % wt 56 95 97 97 9
tower* at a Sasol plant in Se- Vaporization, mol% — 69.0 71.7 72.3 —
cunda, South Africa, the Mass flow, kg/h 134,000 72,000 62,000
chimney tray feeding liquid to the
tower’s once-through kettle reboiler tractive distillation, with the boiling ney tray at 109˚C, and gravity-flows to
unexpectedly overflowed. The plant liquid consisting of a large concentra- the kettle reboiler. Vapor generated in
quickly implemented a successful fix, tion of very high boilers and a small the kettle reboiler at 140˚C is re-
but the cause of the overflow remained concentration of volatile components. turned to the tower above the lower
obscure. Further joint troubleshooting The mulitifeed and multidraw kettle chimney tray. Unboiled liquid from
by a Sasol/Fluor team cleared up the arrangement, and possible foam for- the kettle reboiler, also at 140˚C, over-
mystery, and provided lessons for how mation, may also play a role. flows the reboiler weir into the kettle
to avoid similar problems in other ex- draw compartment, from where it
tractive distillation units. Process description gravity-flows into the tower bottom-
The investigation utilized pressure The process feed to the tower (Figure product sump.
drop measurements, neutron back- 1) contains: 93% Component A, an or- Table 1 shows the rapid change in
scatter, surface temperature surveys, ganic with an atmospheric boiling component concentration that takes
and hydraulic calculations to estab- point of 80°C, which is recovered as a place near the bottom of the tower.
lish the force balance that led to the high-purity overhead product from the The tower bottoms contain 2% by
overflow. Among other things, the top of the tower; and 7% Component B, weight of Component A, 1% of Compo-
measurements offered strong evi- an organic that boils at 84°C under at- nent B, and 97% solvent (Component
dence of boiling maldistribution in the mospheric pressure. B has a higher C). One stage up, at the bottom chim-
reboiler. The team came up with a affinity to the solvent and leaves at the ney tray, the liquid composition
theory that is consistent with all the tower bottom. The solvent boils at changes to 30 wt.% A, 14 wt.% B, and
measurements and force balance; it above 250˚C at atmospheric pressure. only 56 wt.% solvent. This difference
subdivides the reboiler into a stag- In the unusual boilup supply sys- is the reason for the steep tempera-
nant region, an intense boiling region, tem for the tower (Figure 1), liquid at ture change near the tower bottom.
and a kettle region. 108˚C from the bottom trays is col- Such steep composition and tempera-
This regional maldistribution pat- lected on an upper chimney tray, from ture changes are typical of extractive
tern is believed to be unique to mix- where it gravity-flows to a once- distillations, using a high-boiling sol-
tures such as those encountered in ex- through thermosiphon side-reboiler. vent to separate relatively low-boiling
The liquid from the outlet of the ther- organics.
*The tower was not designed by the authors’ em-
ployers. mosiphon is collected on a lower chim- (Continues on p. 56)
CHEMICAL ENGINEERING WWW.CHE.COM FEBRUARY 2004 55
 Operations & Maintenance

The startup problem

Upon startup, the lower chimney tray
operated at the design temperature of
109˚C. The kettle reboiler vapor was at
145˚C, somewhat hotter than the de-
sign 140˚C. However, the bottom sump
temperature was 109˚C, much colder
than the design temperature of 140˚C.
This low temperature meant a much
larger concentration of A (around 30
FIGURE 2. Temperature measurements made during troubleshooting of the kettle
wt.%) in the sump, compared to the de-
reboiler constituted a key input for the analysis of its performance problem
sign (around 2 wt.%). Because of the
discrepancy, the recovery of A in the Based on the overflow theory, the the liquid level on the lower chimney
product stream was much poorer than tower was shut down after one day in tray. The level was found at 300 mm.
the design recovery. What’s more, service, and the chimney heights on This finding confirmed that, with the
flooding was apparent in the solvent the lower chimney tray were extended initial 200-mm chimneys, the liquid
stripper downstream of the extractive from 200 mm to 900 mm. Furthermore, level would have been above the top of
distillation tower, due to the significant the chimney into which the side re- the chimneys and overflow would
variation in its feed composition. boiler could have been blowing liquid have occurred. It also confirmed that
Operators first tried to improve perfor- was blanked. Upon return of the sys- increasing the chimney height to 900
mance by manipulating some of the tem to service, the sump temperature mm would have indeed eliminated the
tower operating variables. Raising the reached the design 140˚C and stayed overflow. Thus, this test fully con-
steam rate as far as possible only had a there during all periods of normal oper- firmed the startup team’s analysis
small effect, raising the sump tempera- ation. The reboiler temperature de- and solution.
ture from 109 to 112˚C. Reducing the feed clined from 145˚C to the design 140˚C. The neutron backscatter scans also
flowrate had a much greater effect, rais- The chimney tray modifications were argue against the liquid trajectories
ing the sump temperature as high as successful in preventing any further from the side reboiler causing the by-
130˚C, but at the penalty of lower pro- liquid bypassing from the lower chim- passing. With the trajectories fully
duction rates. The situation was not tol- ney tray into the bottom sump. eliminated during the startup fix, the
erable, and changing operating variables chimney tray liquid level still exceeded
did not produce a satisfactory solution. Taking the pulse the original chimney height of 200 mm.
The higher reboiler temperature Although the problem thus went Entrainment: Entrainment of liquid
suggests that the flowrate through the away, understanding its root cause in the vapor leaving kettle reboilers
kettle reboiler was lower than as de- posed an interesting challenge. Fur- has been known to occur when the
signed. The only plausible explana- thermore, such understanding would vapor space above the tube bundle is
tion for the low sump temperature is be valuable, both for dealing with sim- small [1]. To explore this possibility,
that the liquid from the lower chim- ilar systems and for application dur- the team assessed the design of the
ney tray was bypassing the kettle re- ing any future debottlenecking of the existing kettle in light of good-design
boiler and reaching the bottom sump. same Sasol distillation unit. So, the criteria listed in the literature [1]. The
Such bypassing can be due either to following checks and measurements ratio of kettle diameter to overflow
liquid leaking from the lower chimney were performed: baffle height in the existing kettle is
tray, or to liquid overflowing into the Pressures: The team measured pres- 1.4, which is near the recommended
chimneys. The observation that the sures at the vapor space above the ket- minimum (1.3 to 1.6). The ratio of tube
sump temperature rose at lower feed tle reboiler and in the vapor above the bundle to kettle diameter is 0.63,
rates argued against tray leakage lower chimney tray in the tower. The which slightly exceeds the recom-
being the root cause, thus supporting measured pressure drop at startup was mended maximum 0.6.
liquid overflow into the chimneys. 4–8 kPa, higher than the design 2.5- Foaming: The team conducted some
There was another possible route kPa maximum, even though the limited tests to check for foaming. The
for liquid to bypass the lower chimney flowrates were lower. The outlet-line tray-pressure drop across the column
tray. Vapor-liquid effluent from the pressure drop was recalculated, using did not provide any evidence of foam-
thermosiphon side reboiler enters the three alternative procedures: a Fluor ing on the trays. So, if foaming took
vapor space above the lower chimney method, a commercial simulation, and place, it would be confined to the kettle,
tray (Figure 1). Though the bottom of a simple hand method based on Perry’s where the solvent concentration was
this nozzle is 300 mm above the hat of Handbook [2]. All three methods gave high and where vigorous boiling takes
the chimneys, it was possible that liq- numbers in the 4–6-kPa range, which place. A sample of reboiler liquid was
uid trajectories from the side reboiler were slightly lower than, yet well in shaken, and was observed to produce
effluent were finding their way into a line with, the field measurements about 25 mm of froth above the liquid,
chimney that had its opening oriented Liquid level: Neutron backscatter which took 1–2 min. to disappear.
towards the effluent nozzle. scans were performed to determine Because of the high fluid tempera-
56 CHEMICAL ENGINEERING WWW.CHE.COM FEBRUARY 2004
tures, it was not possible to simulate in millimeters of liquid head): reboiler floor (Point 2) in Figure 1.
the operating conditions. So foaming Based on the Equation (1) force bal-
H = ⌬Pin + hs,REB + ⌬Pout (1)
could have played a role, even though ance, the chimney tray overflowed be-
no stable foam was observed. Older- where H is the liquid driving cause the pressure drop in the reboiler
shaw tests [1] during the piloting of the head (top of chimney tray liquid level outlet line was high and because the
column indicated some wall-supported to reboiler floor), ⌬Pin is the pressure full static head of the non-aerated re-
foaming, but the foam was unstable. drop in the inlet piping, hs,REB is the boiler liquid acted to raise the driving
Temperatures: Laser-guided infrared static head in the reboiler and ⌬Pout is head on the chimney tray. Entrain-
pyrometers measured the surface tem- the pressure drop in outlet piping. ment from the kettle could have also
peratures around the kettle reboiler The neutron scan measurements contributed. This entrainment would
shell. Several sets of these measure- gave H as 15.5 kPa at normal operat- further increase the pressure drop in
ments were made. While the tempera- ing conditions. The ⌬Pin was calculated the reboiler vapor outlet line, and ac-
ture varied from one set of readings to at about 1 kPa. The ⌬Pout at normal count for the outlet line pressure-drop
another due to changes in steam pres- operating conditions was calculated at measurement being slightly higher
sure, the trends observed did not vary 5–7 kPa; this range was slightly higher than calculated. Calculations show
and were completely repeatable. Figure than the startup ⌬Pout, which was 4–6 that about 10-30% entrainment in
2 shows the results of the most compre- kPa and agreed with the plant data at the vapor would account for the slight
hensive temperature survey. its original startup. With the assump- pressure drop difference. However,
The kettle reboiler has four liquid tion of no aeration in the reboiler liq- this difference can also be explained
inlets and four vapor outlets, a con- uid, and allowing for the head over the by simple inaccuracies in measure-
figuration that is often used to provide weir as recommended [3], hs,REB was ment and calculation.
good fluid distribution in horizontal calculated at about 8 kPa. The right Although the pressure balance ex-
shell-side reboilers. The average sur- side of Equation (1) thus gives 14-16 plains the chimney tray overflow, it
face temperatures at the two vapor kPa, which is slightly on the low side is unable to explain the observations
outlet nozzles furthest away from (but well within calculation accuracy) from our temperature measurements.
the overflow baffle were low (122 of the 15.5 kPa head on the left side of These are discussed below.
and 128°C). The surface temperature Equation (1). Vapor disengagement between
at the next vapor-outlet nozzle was Basic to this force balance is the as- tubesheet and first vapor outlet noz-
141˚C, and at the vapor nozzle closest sumption that the static head inside zle: The tubes are 6,100 mm long (this
to the overflow baffle, the temperature the reboiler, hs,REB, is equal to the head measurement does not appear in Figure
was 144˚C. of actual liquid from the top of the 2). Of this length, about 1,500 mm (or
The surface temperatures at the overflow baffle (plus the small head 25%) lies between the tubesheet and the
reboiler liquid pool displayed the op- over the weir) to the reboiler floor. This beginning of the first vapor outlet noz-
posite trend. Temperatures were high assumption makes hs,REB the head of zle. So if the heat exchanger were to va-
(176 and 186°C) near the flange, and non-aerated liquid, which may be ques- porize the liquid uniformly, practically
were progressively lower along the tionable for a pool of boiling liquid. all the vapor exiting at the first vapor
final half of the reboiler length. The This assumption is made in some outlet nozzle should initiate in the ex-
surface temperatures at the liquid key literature references on the sub- changer section between the tubesheet
pool were 160–168°C in the third ject [3, 4], and is often the conserva- and the beginning of that nozzle.
quadrant away from the overflow baf- tive assumption made for design. The kettle overflow baffle is 1,150
fle, 157–160°C in the second quadrant More-comprehensive models [5, 6] mm high. A Francis weir formula cal-
away from it, and 154°C at the quad- replace the non-aeration assumption culation [2] shows that a 72,000-kg/h
rant just before the overflow baffle. by a model that takes into account the liquid overflow over the outlet weir will
Heat transfer: The heat transfer generated vapor volume, as well as the incur an additional liquid head of 50
coefficient was calculated to be 839 tube-bundle two-phase pressure drop. mm. So, the liquid level is at least 1,200
kW/(m2)(˚C), which is lower than the Application of these models to kettle mm above the kettle floor (without al-
design 1,026 kW/(m2)(˚C). reboilers is further complicated by the lowance for hydraulic gradients or for
recirculation of liquid inside the kettle aeration of the liquid, which will tend to
Assessing reboiler operation and the existence of fairly clear liq- make the liquid height even taller).
The measurements and determina- uid zones between the bundle and the The first 236 mm of tubes are in the
tions made as outlined above enabled sides of a kettle reboiler [5]. 1,000-mm-dia portion of the kettle,
the team to assess the operation In the Sasol reboiler, there is an ad- with no vapor disengagement space.
within the reboiler. ditional justification for the non-aera- Over the next 1,090 mm of the tubes,
Pressure balance: The pressure bal- tion assumption. As is discussed below, the kettle diameter expands from
ance between the elevation at Point there is evidence that the nozzle clos- 1,000 to 1,628 mm, at a 30-deg angle.
1 at the vapor space above the lower est to the tubesheet enters a stagnant Geometry dictates that the first 350
chimney tray and that at Point 2 at liquid region. In that case, this inlet mm of that expansion will be fully
the reboiler floor (see Figure 1) is as will see the full hydrostatic head be- submerged by the 1,200 mm of liquid.
follows (with all pressures expressed tween the overflow (Point 3) and the So the total tube length at which there
CHEMICAL ENGINEERING WWW.CHE.COM FEBRUARY 2004 57
 Operations & Maintenance

is no vapor disengagement whatso-

ever is about 590 mm. In this length,
40-50% of the vapor reaching the first
vapor nozzle should be generated.
Further, in light of the expected vapor
generation, vapor velocities should be
locally high. And, there will not be
much disengagement of liquid in the
next 500 mm or so of tube length, as
this section has a restricted vapor-
FIGURE 3. Boiling behavior varied significantly across regions of the kettle reboiler
passage area due to the smaller diam-
eter. (However, as we have concluded quadrants of the kettle. This means versely, in the intense boiling region, the
below, the actual vapor generation in that there is less flow resistance, due boiling aerates the liquid, which lowers
this section is very low.) to the lower static head of the froth in the liquid head. This induces high
Stagnant liquid region (Figure 3): the central and baffle regions. Conse- flowrates of liquid into the region until
For the liquid in the tubesheet region, quently more incoming liquid will the pressure-drop terms balance the dif-
the surface temperatures measured channel toward the central region. ference between the aerated-liquid head
were high. This is also the region that Our calculations indicate only a few in this region and the non-aerated-liquid
has little disengagement space above. hundred millimeters of head differ- head in the stagnant region.
Liquid temperatures in the range of ence are enough to account for the es- The high liquid flowrate induced
168 to 186°C imply concentration of sential absence of flow in one feed noz- into the vigorous-boiling central re-
Components A and B of less than 0.5% zle while the adjacent nozzle feeds the gion means that the fraction of the liq-
by weight; in other words, very low. bulk of the liquid to the reboiler. As uid vaporized in this region is lower,
Under the assumption that these less feed flows towards the tubesheet so less of the low-boiling components
temperatures apply to the mixture end, the liquid in the tubesheet region are removed from the liquid. This
close to the tubesheet throughout the will stagnate more and vaporize less. lights-rich liquid could possibly per-
tube bundle, the material in this re- The end result is a self-accelerating sist all the way to the two vapor noz-
gion thus consists almost entirely of mechanism that promotes the dead zles closest to the tube sheet.
the non-volatile solvent, Component zone in the tubesheet region and a The lower temperatures observed
C. It then follows that in this region, high-liquid-flux central region. near the vapor nozzles closer to the tube
very little boiling and heat transfer The high-L/V, intense boiling, cen- sheet are the result of the high liquid
take place. This absence will occur if tral region (Figure 3): This region flux in the tube field and the resulting
there is only a small flux of material has a higher-than-design liquid fee- lower temperatures through the re-
flowing around the tubes in this area. drate (and a high liquid/vapor, or L/V, gions providing the vapor to these noz-
The small flow generates only a ratio), to compensate for the reduced zles. The vapor nozzle closest to the
small quantity of vapor, and the liquid flow to the stagnant region. The high tubesheet is the coldest, because that is
around the tubes will get hotter and rate results in lower liquid tempera- where the intense boiling starts and is
hotter as its low-boiling components tures, but with intense boiling and more vigorous, and there is a smaller
flash off. All that will be left is the higher than design vapor generation. fraction vaporized. Also, some of the liq-
heavy liquid, which tends to sit there. A pressure balance across the re- uid is likely to be projected into the
This behavior turns most of the first boiler (Figure 1) equates the pressure vapor outlet nozzles in the form of en-
quadrant into a stagnant zone or dead difference between the common liquid trainment. The action in this region has
pocket of hot, heavy liquid, mostly inlet line (Point 4) and the common re- some similarities to the action in hori-
Component C. boiler vapor space (Point 3), no matter zontal thermosiphon reboilers, in that
It is worth noting that temperatures which region one travels through. The the quantity of vapor generated results
of 168 to 186°C were surface measure- pressure difference is given by Equa- in a low-density froth, which promotes
ments at the shell, some distance away tion (2), again with all pressures ex- liquid flow to this region, in contrast to
from the hot tubes. The stagnant layer pressed as millimeters of liquid head: the behavior with the more dense liquid
between the tubes and shell is a good in the more stagnant zones.
P4 - P3 = hs +
Pn +
PREB (2)
insulator, and it is possible that near One might wonder what caused
the tubes the temperatures will be where hs is the static head in a given the measured wall temperatures of
higher, probably approaching that of reboiler region,
Pn is the pressure the liquid in this region to remain
the condensing steam. drop in the liquid line leading to the hot. The key is that our survey mea-
Since not much liquid vaporizes in region, and
PREB is the reboiler pres- sures the wall temperatures, not the
the stagnant zone, the froth density in sure drop in that region. bulk temperatures. The intense boil-
this region is high, much closer to that In the stagnant region discussed ing takes place at the bulk of the
of the liquid than to that of a vapor/liq- above, hs approaches the full head of the bundle, and may not reach the wall.
uid mix, and much higher than that of non-aerated liquid, and the
P terms are Equation (2) shows that along the
the froth or aerated liquid in the other quite small due to the low flow. Con- wall in this region, there is horizon-
58 CHEMICAL ENGINEERING WWW.CHE.COM FEBRUARY 2004
tal flow of hot liquid from the higher- tion situation: a single zone can han- space, and is the only region behaving
static-head stagnant region towards dle substantially more than the de- like a kettle.
the overflow baffle. This flow is some- sign liquid flowrate and produce much This pattern explains all the ob-
what tempered by the boiling. On the more than the design amount of vapor, servations made: the high pressure
other hand, the vapor temperatures based on a much improved tempera- drop, the measured temperatures, the
were measured right above the bulk ture driving force. Further, the model lower heat transfer coefficient (caused
of the bundle, where the boiling is confirms that this high vapor produc- by the stagnant region), and even the
most intense. tion comes about because of relatively likely entrainment.
The kettle boiling region (Figure small gradients in head within the The regional maldistribution pat-
3): There is a net flow of liquid from tube field, brought about by regions tern reported here gives complete
the central region to the overflow of low froth density due to high vapor support to the assumption that the
baffle, because this is the only manner production. The relatively inactive reboiler liquid head in Equation (1),
in which the solvent can permanently zones explain the lower-than-design hs,REB, is the static head of non-aer-
exit the kettle reboiler system. As one overall heat transfer. ated liquid in the reboiler. This head
moves closer to the overflow baffle, The regional maldistribution pattern is set by the static head of the liquid
there are two effects: this net flow reported here may be universal among in the stagnant region. This explana-
increases, and the composition of the kettle reboilers. But in most cases, it tion closes the loop on the reason for
liquid can be expected to contain less will not result in significant perfor- the chimney tray overflow. The high
of low-boiling material. As a result, mance issues. Instead, this maldistri- reboiler static head, plus the high re-
less vigorous boiling and a higher aer- bution is more significant in reboilers boiler outlet line pressure drop led to
ated-liquid density can be expected in for mixtures such as those used in ex- the high liquid heads in the chimney
this region. This higher aerated-liquid tractive distillation, where the boiling tray. Entrainment from the intense
density will provide more resistance liquid consists of a large concentration boiling central zone could also have
to liquid feed from the inlet nozzle in of very high boilers and a small concen- been a contributor.
this region and, in turn, reduce the tration of volatile components. The low temperatures at the vapor
inlet feed to the kettle boiling region. The mulitifeed, multidraw arrange- outlet nozzles right above the intense-
This is the one region of the reboiler ment plays a major role. A single feed, boiling region were caused by the high
that actually operates as it should, single draw kettle would have experi- L/V ratios in this region, possibly as-
like a kettle. Not much entrainment enced a different maldistribution pat- sisted by entrainment. Above the kettle
takes place here, and there is much tern, possibly even more severe. region, the L/V ratios were much lower,
disengagement space in this region, A stagnant region or dead zone is and so was entrainment, leading to the
both of which promote the true kettle formed in the tubesheet quadrant by a higher than expected temperatures.
action. As the reboiler-feed liquid flow self-accelerating mechanism. With little The high temperatures measured
through this region is lower than the disengagement space, not much boiling near the walls of the exchanger were
design and the percent vaporization is takes place. The low-boiling compo- caused by horizontal flow of hot liquid
close to or higher than design, the tem- nents tend to preferentially vaporize from the stagnant zone along the wall
perature of vapor flowing to the last out of the liquid, so the liquid becomes towards the boiling zones.
nozzles is close to or above design. rich in the non-volatile component. The The wall-supported foaming ob-
If liquid entrainment from the in- nonvolatile liquid, with little aeration, served on the Oldershaw tests, and
tense boiling region is significant, it develops a high hydrostatic head, which the slight foaming observed in the
will generate a higher pressure loss resists movement of fresh liquid into foaming tests, could also be a contrib-
in the reboiler outlet lines above the this region and promotes stagnation. uting factor that could aggravate the
intense-boiling region. Since the pres- This region could constitute a signifi- intense boiling effect.
sure difference between Points 5 and cant performance issue, with lower than The malfunction reported here,
3 in Figure 1 is constant no matter expected heat transfer rates, or lead to namely, excess pressure drop in a
which path is taken, this higher pres- fouling in heat sensitive materials kettle reboiler circuit when compared
sure loss in the nozzles above the in- The high hydrostatic head in the to the available head, was identified
tense boiling zone will induce a higher stagnant zone, and the enhanced [7] as the most common malfunction
vapor flow into the outlet nozzles amount of fresh feed diverted into the experienced in reboilers. The employ-
closer to the outlet baffle. central region, initiate a region of high ment of good design practices, coupled
L/V ratio, with vigorous boiling in the with compiling a valid reboiler pres-
Mysteries explained bulk of the bundle. The boiling is prob- sure balance, can readily circumvent
This overall assessment of the reboiler ably accompanied by entrainment into such malfunctions.
operation led us to the development of the overhead vapor line, low fractional In contrast, the maldistribution
a model. Described in greater detail vaporization, and a relatively high pattern reported here has not previ-
in Reference [8], the model quantifies concentration of lights. ously been reported. It is unique to
the theory presented above, and it ex- Finally, the region near the overflow mixtures such as those encountered in
plains all of the observations. baffle receives close to the design feed extractive distillation, where the boil-
The model confirms a maldistribu- flow and has ample disengagement ing liquid consists of a large concen-
CHEMICAL ENGINEERING WWW.CHE.COM FEBRUARY 2004 59
 Authors
Etienne Rubbers is a corpo-
rate finance consultant for
Operations & Maintenance Sasol (Rosebank, Johannes-
burg, South Africa; Phone:
+27 11 441 3472; e-mail: eti-
[email protected]), in-
volved in merger and acquisi-
tration of high boilers and a small con- reboiler as an alternative for mix- tion activities. Previously, he
was a process engineer in the
centration of volatile components. tures containing a large concentra- concept development group,
This maldistribution affects both ket- tion of components boiling at high developing a number of the
firm’s projects from concept to
tle heat transfer and pressure drop, temperatures together with compo- basic engineering, construction and commission-
ing. Despite his career change, he still has keen
and it, too, needs to be accounted for nents that are volatile. ■ interest in process engineering. He holds a B.Sc.
in the kettle design. Edited by Nicholas P. Chopey (chemical engineering) from Wits University
(Johannesburg) and a master’s degree in indus-
trial management from Belgium’s University of
Lessons for design Leuven, and is a CFA charterholder.
References
Kettle reboiler designs should: Kirsten Green is a senior
1. Kister, H. Z., “Distillation Operation,” Mc- process engineer for Sasol
1. Strictly adhere to the reboiler pres- Graw-Hill, New York, 1990. Technology (Phone: +27 11
2. Perry, J. H., “Chemical Engineers' Hand- 344 0082; e-mail: kirsten.
sure-balance equation. Four-fifths book,” 7th Ed., McGraw-Hill, New York, 1998. [email protected]). She started
of kettle reboiler malfunctions are 3. Lieberman, N. P., “Process Design for Reli- her career at Sasol’s coal
able Operation,” 2nd Ed., Gulf Publishing, preparation plant, gaining ex-
caused by excess pressure drops in Houston, 1988. perience in materials han-
reboiler circuits [7]. 4. Palen, J. W. and Small, W. M., Kettle and Internal dling. She then joined the con-
Reboilers, Hydrocarb. Proc., 43 No. 11, p. 199, 1964. cept development group,
2. Be based upon the clear liquid head 5. Fair, J. R. and Klip, A., Thermal Design of where she has been involved
in the reboiler, not on aerated-liquid Horizontal Reboilers,” Chem. Eng. Prog., p. in feasibility studies for the
86, March 1983. company’s chemical projects. She spent the past
head. 6. Ishihara, K., others, Critical Review of Correla- year on rotational training with MWKL in Lon-
don, doing basic engineering in petroleum refin-
3. Take account of froth density gradi- tions for Predicting Two-Phase Flow Pressure
ing. She holds a B.Ing. from the University of
Drop Across Tube Banks, Heat Transfer Engi-
ents within the reboiler, and the re- neering, 1, No. 3, p. 23, January–March 1980. Stellenbosch and an M.Sc. from University of
Cape Town, both in chemical engineering.
sulting maldistribution. 7. Kister, H. Z., What Caused Tower Malfunc-
tions in the Last 50 Years?, Trans. IChemE, Terry-Ann Fowler is a senior
4. Trade off pressure loss through inlet Vol. 81, Part A, p. 5, January 2003. process engineer for Sasol
nozzles (which provide liquid distri- 8. Rubbers, E., Green, K., Fowler, T., Kister, H. Z., (Phone: +27 11 344 0092; e-
and Stupin, W. J. Once-Thru Reboiler Startup mail: [email protected]).
bution to the reboiler) against the Can be Exciting, in “Distillation 2003: on the She forms part of a team in-
Path to High Capacity, Efficient Splits,” Proceed- volved in conceptual develop-
head and column height required. ings of Topical Conference, AIChE Spring Meet- ment and design of a number
5. Consider a horizontal thermosiphon ing, New Orleans, La., March 31–April 3, 2003. of the firm’s chemical and re-
finery projects. She has
gained experience on projects
from pilot plant studies,
through feasibility and con-
ceptual development, basic engineering, and on
to construction and commissioning. Her B.Sc.
and M.Sc., both being in chemical engineering,
are from the University of Cape Town, and her
Ph.D., also in chemical engineering, is from Wits
University.
Henry Z. Kister, a Fluor
Corp. senior fellow and direc-
tor of fractionation technology
(Phone: 1-949-349-4679; e-
mail: [email protected]),
has over 25 years experience
in design, troubleshooting, re-
vamping, field consulting, con-
trol and startup of fractiona-
tion processes and equipment.
Previously, he was Brown &
Root’s staff consultant on frac-
tionation, and worked for ICI Australia and Frac-
tionation Research Inc. (FRI). The author of text-
books “Distillation Design” and “Distillation
Operation,” as well as 70 published technical arti-
cles, he has taught the IChemE-sponsored “Prac-
tical Distillation Technology” course 250 times. A
recipient of Chemical Engineering’s 2002 Award
for Personal Achievement in Chemical Engineer-
ing, Kister holds B.E. and M.E. degrees from the
University of NSW in Australia. He is a Fellow of
IChemE and a member of AIChE, and serves on
the FRI Technical Advisory and Design Practices
Committees.
Walter J. Stupin is an exec-
utive director for Fluor
(Phone: 1-949-349-5209; email:
walter.stupin @fluor.com). He
manages and executes process
engineering for petroleum re-
fining and chemical plants.
His over 40 years of process
engineering experience
ranges from vice president of
technology at C F Braun Inc.
to research at Fractionation
Research Inc. (FRI). The distillation field has
been an area of key interest, in which he has
published over 20 technical papers. Dr. Stupin
holds B.S., M.S. and Ph.D. degrees, all in chemi-
cal engineering, from the University of Southern
California.

60 CHEMICAL ENGINEERING WWW.CHE.COM FEBRUARY 2004