Revamp Distillation
Revamp Distillation
Optimize Distillation
System Revamps
Revamping can
improve most
D inant separation technique in
the chemical process industries
(CPI), accounting for more
than 90% of product recovery and purifi-
cation applications. Its use spans the en-
• eliminating operational bottlenecks
and increasing capacity;
• implementing a new process or
product within existing equipment;
• enhancing product purity, thereby
existing columns tire CPI, from massive continuous chemi- creating possible new markets for an ex-
cal and petroleum complexes to small isting product;
and, indeed, batch facilities manufacturing specialty • boosting product or raw-material
chemicals such as materials additives, recovery;
provides one of the pigments, paints, inks, pharmaceuticals, • purifying raw materials;
flavors, and fragrances. With more than • reducing waste-disposal quantities;
most economically 40,000 distillation columns in operation and
in the U.S., the capital investment in- • removing and capturing materials
attractive opportu- volved exceeds $8 billion (1). from water and air waste streams.
Though revamping of distillation sys- Even though these factors make re-
nities available in tems is not a new concept, it is assuming vamping a worthwhile endeavor, the dis-
greater importance today. Many compa- tillation process sometimes is not consid-
process engineering nies now are trying to maximize their re- ered a part of the primary production line
today. Use this turn from existing sites, rather than in- but instead a low-priority support opera-
vesting in new grassroots facilities. Plus, tion. Consequently, too often, it does not
proven strategy the opportunities for column revamping
abound: An increasing number of distil-
receive the emphasis and attention it
deserves.
to find the lation systems are required to process
variable-feed-composition streams and
Distillation for recovery of solvent or
unreacted raw materials also typically is
most-cost-effective to meet more-stringent product and considered a lesser operation than prod-
byproduct specifications. Many batch uct recovery. For such “secondary” oper-
option. systems are operated in campaigns, ations and within smaller facilities, finan-
which may make additional demands as cial constraints can be particularly sensi-
the feed components and composition tive, especially if it is assumed (mistaken-
John A. Williams,
are perpetually changing. Improving the ly) that revamping the distillation column
Environmental & Production
performance of such processes promises will be prohibitively expensive.
Solutions, LLC
a high economic return. In fact, for most
distillation systems currently in opera- Consider a real revamp
tion, a revamp may provide significant Many companies don’t actually assess
improvements in the competitive posi- the revamping of an existing unit — not
tion of the process. realizing that a high economic return usu-
Revamping opportunities are present ally is easily achievable with a well de-
in most existing distillation systems, with signed and executed revamp. Instead, the
CHEMICAL ENGINEERING PROGRESS • MARCH 1998 ©Copyright 1998 American Institute of Chemical Engineers. All rights reserved. Copying and downloading permitted with restrictions.
D I S T I L L AT I O N R E VA M P S
traditional “revamping” approach 4. How to analyze the economic identify whether existing equip-
often taken is to design and build a impact; ment, instrumentation, and controls
new column, rather than to work In addition, we will provide case actually may have the potential to
within the confines of the existing histories of distillation system re- achieve the necessary results with
unit. This may involve lower risk, but vamps to illustrate the value of the only minimal modifications. All
it almost always is more costly in approach. process and instrumentation dia-
terms of capital investment (including In short, this article will provide grams (P&IDs) and controls should
for equipment that a well-designed an overview of all the factors that be personally field-verified. In
revamp may obviate) and lost pro- must be considered when evaluating short, field assessment is not only
duction due to downtime. a possible distillation-system revamp. an excellent method for understand-
The reasons why improvement of Though the recommended approach ing current operations and problems,
existing equipment doesn’t get the em- involves more upfront effort by the but also is a cost-control measure
phasis it deserves are many and varied production engineer, it will identify that may save a company from pur-
— but they generally boil down to a the most economical way to achieve suing unnecessary expenditures.
lack of solid understanding of what operational objectives. Visual inspection — If the time
can be accomplished and a lack of re- frame of the project allows, visually
alistic analysis of capital requirements. HOW TO ACHIEVE inspect the column internals by ob-
Yet, to reiterate, revamping usually is A SUCCESSFUL REVAMP serving the operation through the ac-
the most cost-effective solution. One of the main issues involved in cessways or handholes. Look for in-
Revamps afford the opportunity to revamping a distillation system is de- ternals that may have been displaced
take advantage of the significant termining what must be done to effect or those that have become mis-
progress recently made in distillation an economic gain for the plant. We’ll aligned over time. Also, check for
systems. Such developments include outline several steps that can serve as plugging within the column, as well
the commercialization of higher-effi- a guide. as deformation or any signs of exces-
ciency trays and packings that sive wear of packing material or dis-
promise high economic returns. Define system objectives tributors. Surfaces that have become
Today’s high-performance mass- The first step is to define the de- fouled or show stress cracking or dis-
transfer internals offer dramatically sired objectives of a revamp. Infor- coloration may indicate such prob-
reduced column pressure drops and mation must be gathered relating to lems as poor wetting, pressure surges
increased turndown ratios — en- the processing requirements, includ- during operation, or excessive flash-
abling operation over a wider range ing determining the full range of po- ing near the reboiler discharge. Addi-
of high and low liquid-flow loadings tential feedstock compositions and tionally, if column performance has
— than was previously possible. stream conditions. It also is helpful to declined over time and signs of mate-
The purpose of this article is to note how many product lines are rials corrosion are apparent, consult a
show production and process engi- campaigned upstream of the distilla- metallurgist. A competent metallur-
neers how to determine, given their tion system. The desired minimum gist can assess the impacts of the cur-
particular objectives, whether or not and maximum processing capacity rent corrosion and specify either op-
revamping is a viable option. It intro- also must be defined, as well as the erational conditions less corrosive to
duces an approach that can be applied product composition wanted (both the the existing materials or materials
regardless of scale of operation or minimum acceptable and “ideal”). It that offer a higher degree of corro-
type of product. We will not get into also may be necessary to specify sion resistance.
specific design practices, which can other product attributes such as color Radioactive tools — Three estab-
be found in several excellent theoreti- and clarity. A report of the product lished column on-line diagnostic
cal and practical texts such as Refs. specifications and distillation system tools involving the use of radioiso-
2–8. Instead, we will outline the requirements should be confirmed topes — grid scanning, radioactive
following: with the organization’s Marketing tracing, and computer-assisted to-
1. How to achieve a successful Product Manager and the Production mography (CAT scans) — can be
revamp; Manager, respectively. useful in assessing the hydrodynam-
2. How to factor in environmental ics of the column under existing oper-
issues; Evaluate existing operations ational conditions. The internals defi-
3. How to develop the right strate- The next step is to assess how the ciencies identified then can be ad-
gy — including assessing the desired existing distillation system works. dressed in the revamping design.
operating conditions, finalizing de- On-site observations are crucial to Grid scanning basically is a mea-
sign parameters, and selecting the the understanding of the current sys- surement of the column cross-sec-
proper equipment; and tem’s operation — they often can tional density profile via a series of
Use computer simulations developed, empiricism will continue The two predominant process-
— wisely to predominate in the calculation or stream materials that cause the most
Computer simulators are readily assignment of stage efficiencies. Once environmental concern are: (1) “still
available and are computationally ac- an internals type is selected, you are bottoms,” which often consist of
curate — making it easy to quickly well advised to work closely with the tarry residues that frequently con-
simulate the full range of potential internals vendor to determine the tain a multitude of molecular
operating conditions. The majority of achievable stage efficiencies for a species and exhibit high viscosity at
simulators employ an equilibrium- given design. In practice, stage effi- ambient temperatures; and (2) high-
stage approach, based on vapor/liquid ciencies frequently are assigned based vapor-pressure materials (material
equilibrium (VLE) data, and model on the operation of similar systems of losses), which may slip through
steady-state operations. Batch and components and internals. condensers into vacuum systems or
dynamic simulators also are avail- Whatever methods of modeling are become fugitive emissions through
able. Process simulation packages are used, perform simple checks on the leaks in flanges and other equipment
offered by, among others, Aspen, simulation output to see if the simula- connections.
Chemstations, Hyprotech, and Simu- tion reflects real-world operation:
lation Sciences. • Assess the simulator results (out- Reduce the amount
The key to an accurate simulation, put) against those obtained from a of still bottoms
especially with respect to the number McCabe-Theile diagram. Still-bottoms or tarry-material
of equilibrium stages, is accurate VLE • Compare the output against data formation frequently is caused by
data. Most commercially available from the original startup and the cur- byproduct formation in the reactor
simulators come with libraries of VLE rent operation. and, if conditions permit, in the still.
data. Look also for in-house VLE data More details on how to make sure Distillation systems used to process
that may have been generated with pre- that you have a realistic and reliable effluents from more than one pro-
vious distillation designs for the com- simulation are given by Kister (18). duction line often are particularly
ponents within the system. Literature Using the process simulator, you susceptible to tar-forming reactions.
sources for VLE data include AIChE’s should do a methodical sensitivity An additional problem with still bot-
Design Institute for Physical Property study covering the full range of feed toms involves disposal. The pres-
Data (11), DECHEMA’s Chemistry and operating conditions and product ence of such high-molecular-weight
Data Series: Vapor-Liquid Equilibrium specifications. This allows you: (1) to products in the waste often results in
Collection (12), and publications by determine what is achievable in the additional waste, as processors try to
Hala and coworkers (13,14). VLE data existing equipment, and how the con- manage their flow by: (1) controlling
also can be measured for the compo- trols and the control algorithms may the discharge composition to include
nents under investigation and obtained need to be modified to achieve the lighter components, such as process
from contract research laboratories for target conditions; and (2) to begin to solvent; or (2) keeping the flow high
a fee. As a last resort, if actual VLE determine what a revamped system or the outlet orifice larger than nec-
data are not available and measure- would be capable of achieving. essary to avoid any chance of plug-
ment is impractical, a predictive model ging. Losses also occur when opera-
such as the UNIFAC group contribu- HOW TO ASSESS tors handle the high-viscosity mate-
tion method and its extensions (Sher- ENVIRONMENTAL ISSUES rial by diluting it with valuable sol-
wood and coworkers (15)) can be used, Process wastes and emissions can vents or raw materials.
but only with extreme caution. Carlson have a major economic impact on For distillation yield maximization
(16) provides a decision-based ap- distillation systems. Therefore, any and waste minimization, it is neces-
proach for evaluating physical-proper- revamping strategy needs to assess sary to obtain as much knowledge as
ty parameters for successful process environmental gains and losses. First, possible about any potential (ongo-
simulations. identify what is causing the solid or ing) chemistry occurring in the still or
Despite the precision of simulators liquid waste components in the cur- upstream that leads to tarry materials
in calculating the number of equilibri- rent system. Determine if there are air and their composition in the distilla-
um stages, translating theoretical to or water emissions from the system, tion feed. Because there may be no
real stages remains an inexact predic- and whether they are fugitive or con- documented research or history of the
tive art based on empirical correlations tained within the process streams. still’s chemistry, the only alternative
such as the Murphree tray efficiency Also, ascertain where the emissions may be to pinpoint as closely as pos-
(17). Further, each internals type, size, are escaping — that is, into the vacu- sible the chemistry of the tar forma-
and configuration exhibits different um system, the cooling water tion and its location. Pay particular
hydrodynamics, and, until detailed un- streams, through the piping flanges, attention to additional chemistry that
derstanding of the hydrodynamics is or somewhere else. may be occurring in multicomponent
Conduct field trials mass balance calculations and com- There are, as we’ll discuss, some
At this point, your knowledge putational modeling, and provide an general differences in performance be-
level and revamping strategy should assessment of the actual number of tween trays and packings, and these
be at least 90% complete. Running separation stages — that is, the stage can serve as a guide for preliminary
tests now with the existing equipment efficiency. selection. The final comparison, how-
should verify the strategy, while pro- The goal, throughout the field trial, ever, should be between optimum
viding the final assessment of the ex- is to clarify and confirm the three trayed-column and packed-column de-
isting equipment under the proposed fundamental questions for the column signs. The optimal tray design is one
conditions. Prior to the field trial, fin- revamp, specifically: that balances bubble and downcomer
ish any remaining mass- and heat bal- 1. What are the required hydrody- areas so that neither becomes a capaci-
ances and system productivity-rate namics to achieve efficient phase ty pinch. Such a design then will max-
calculations. Throughout the field contact? imize the fractionation efficiency and
trial, the data logging should include 2. What are the number of stages column capacity by specification of
a continuous assessment of how close necessary for the separation, based on weir height and geometry, downcomer
the system is running to the target the equilibrium thermodynamics of clearance, and the fractional hole area.
conditions; this is critical for deter- this distillation system? Similarly, an optimal packed-column
mining what modifications are re- 3. What precautions, such as the design is one that properly incorpo-
quired to achieve desired operation. avoidance of plugging, fouling, or the rates liquid and vapor distributors,
Hanson and coworkers provide a like, are needed for system reliability? packing type and size, and supports to
summary of preparation procedures At the conclusion of the field trial, avoid or minimize adverse liquid and
prior to a distillation test run (19). the revamping strategy should be vapor maldistribution.
The field trial should strive to sim- nearing completion. You should have
ulate the full range and combinations determined the operational mode and Choose the right internal
of parameters including feed, liquid, sequence, as well as conditions con- The goal in selecting internals is to
vapor, boilup and reflux rates; tem- ducive for acceptable hydrodynamics, come up with a configuration that
perature and pressure profiles; and startup and shutdown procedures, and maintains effective vapor/liquid con-
the startup and shutdown procedures. the instrumentation and the process tact over the desired operating condi-
If the revamping strategy involves a control algorithm. tions, while minimizing pressure
change of operational mode — for in- With this critical information, you drop, liquid entrainment, weeping,
stance, from batch to semicontinuous now can turn to selection of the spe- corrosion, and fouling or plugging. At
operation — the trials may require cific internals. this point, it should be clear that col-
extensive manual control to achieve umn capacity is largely determined
the operational sequence. Limitations Start with by column diameter, specifically its
of the system should be identified and the existing internals ability to handle the desired liquid
explicitly assessed. Retrofitting better internals into an and vapor flow rates without inducing
The hydrodynamics, both liquid existing column can readily provide adverse flooding. Groups, including
and vapor, of column internals should increases in capacity and fractiona- Fractionation Research, Inc. (Stillwa-
be evaluated at the projected typical, tion capability. Such a revamp isn’t ter, OK), and the Separations Re-
minimum, and maximum flow rates, always needed, however. At times, search Program at the University of
and at the minimum and maximum performance of existing internals can Texas at Austin, have performed fun-
achievable reflux rates. This requires be improved significantly by under- damental research on system parame-
that close attention is maintained to all standing their inherent limitations and ters that affect internals selection.
system indicators such as temperature modifying the operation to overcome In general, most columns less than
and pressure profiles, flows, and re- these limitations. 36 in. dia. tend to favor packing, ei-
boiler and condenser operation. Signs No one internals type, be it trays, ther structured or random. Beyond the
of liquid entrainment (jet flooding), structured packing, or random pack- general 36-in.-dia. division, the di-
liquid flooding, and liquid or vapor ing, is superior to the others in all mensionless flow parameter, FP, can
maldistribution — and corresponding applications; indeed, all can be aid in selecting column internals type.
losses in the separating efficiency — made to fractionate the same chemi- The flow parameter is the ratio of the
should be monitored constantly. cal systems. Therefore, it behooves square roots of the kinetic energy of
As the distillation progresses, get a the revamping engineer to under- the liquid to the kinetic energy of the
compositional analysis of the feed, stand what the existing internals vapor — that is, in mass units:
bottoms, overheads, and, if available, configuration can and can’t accom- FP = (L/V)(ρV/ρL)1 (1)
intermediate stage(s) within the col- plish before opting to retrofit a new For optimized, modern 24-in.-spaced
umn. This analysis will refine the set of internals. trays, 2–21-in.-nominal-dia. packing,
existing steam-jet vacuum with a vac- By using the single solvent, the the lightest components into the sal-
uum pump system having a refriger- distillation was converted from a able component also was occurring,
ant-based condenser system. This batch operation to a semicontinuous and the true distillation yields were
proposed revamp would have cost in one, which achieved several objec- unknown.
excess of $2,000,000. It, however, tives for the revamp. The first was in- To achieve the desired distillate
would not have solved the product creased solvent recovery, which en- purity, the design engineers analyzed
quality issue, improved the environ- abled the plant to process consider- the system and developed two re-
mental aspects of the process, or re- ably higher quantities. Final product vamping proposals. One involved re-
duced the disposal costs. purity also was greatly improved due placing the batch still with a thin
Through careful analysis using the to better purity of the solvent. Addi- film evaporator and a three-column
approach discussed in this article, the tionally, because the distillation bot- distillation train including con-
revamping engineers came up with a toms’ concentration and viscosity densers, reboilers, accumulators,
far better alternative. were controllable, the company was pumps, and controls. The second de-
By identifying what the problems able to pump the bottoms to the boil- sign retained the existing still, fol-
were upstream, the engineers more er house for reliable on-site destruc- lowed by a three-column distillation
realistically determined what would tion, which was significantly less ex- train and its associated equipment
make the most sense in the distilla- pensive (and more politically correct) and controls. Either design would
tion process. The challenges to the than shipping it off-site. have required a capital expenditure
distillation included wide variations In summary, a costly revamp was in excess of $1,000,000.
in the feed to the still due to inconsis- avoided by understanding the limita- Fortunately, the approach dis-
tencies among reaction endpoints. tions of the existing distillation sys- cussed in this article then was used.
This became the focus of a short-term tem, modifying the upstream process, On-site analysis identified several
process-development project. It led to converting the operational mode, ad- problems within the existing opera-
alteration in the reaction conditions dressing the environmental concerns, tion. One of the key trouble areas was
and conversion of the reaction and and then retrofitting only a minimal that liquid maldistribution frequently
separation process from a three com- amount of equipment. occurred throughout the batch distil-
ponent system to a two component lation process, impeding fractiona-
one employing the same lower- Pharmaceutical intermediate tion. The existing column was operat-
volatility solvent for both the reaction purification ed on a variable reflux algorithm; the
and separation. Company Y needed to revamp a equipment, however, had been over-
Laboratory distillation deter- batch distillation operation to meet sized to the point that adequate liquid
mined that the reaction byproducts customer requirements for purity of a distribution was maintained only at
were extremely viscous, reaching particular pharmaceutical intermedi- higher reflux rates. The control algo-
more than 1,000 cP at room tempera- ate. The existing column contained rithm and the structured packing in-
ture, and that a stagnant reboiler structured packing with four receivers ternals, coupled with the batch opera-
mixture exhibited detrimental foam- in parallel and had variable reflux tion mode, also restricted the produc-
ing characteristics. In the laboratory, control based on single-loop con- tivity in the larger fraction cuts (the
the solvent in the still bottoms could trollers. It was operated under various five distillation cuts ranged from a
be limited to a few percent. In the vacuum settings utilizing a two-stage few percent to 45% of the feed); op-
plant, the properties of the still bot- steam-jet system. The process was erators frequently overrode the sys-
toms, including viscosity and solvent very operator intensive, requiring up tem controls to improve boilup rates.
holdup, varied widely even though to 80% of the operators’ available at- The feed composition to the still var-
the same temperature and vacuum tention. Additionally, because product ied considerably — however, most of
endpoints were consistently used. specifications were not met, large the variance could be attributed to
The laboratory distillation proved quantities had to be reprocessed. two components, one of which was
that the single-blade propeller agita- The challenges included variable the highest volatility component.
tor in the still simply didn’t provide feed concentrations from one batch to After the field observations were
agitation and, therefore, the interfa- the next. Also, the operation pro- made, the system was modeled on a
cial transfer quickly declined as the duced six components, and the purity simulator — first, batchwise and,
surface became quiescent in the of the salable one — a component of then, in a semicontinuous mode, ad-
early stages of solvent removal. So, intermediate volatility — was very justing the reflux ratio to maintain
the kettle was retrofitted with a dual inconsistent, typically ranging from adequate internal liquid rates for the
axial impeller, which induced a 90–95%, while the customer demand- existing column internals. Through
high-velocity axial flow and more ed purity levels in the >95% range. the simulations, the revamping engi-
interfacial transfer area. Unacceptable cross-contamination of neers determined that the current