How to TROUBLESHOOT TOWER FLOODS

MARCH 2016 | The Chemical Engineer | page 42

troubleshoot HOW to tower floods Non-radioactive methods to diagnose, locate, and identify the root cause of flooding in distillation towers HENRY KISTER DIRECTOR OF FRACTIONATION TECHNOLOGY, FLUOR; ICHEME COURSE LEADER ON PRACTICAL DISTILLATION TECHNOLOGY

F

LOODING is accumulation of liquid in a distillation tower.

or by air leak into the pump suction.

This accumulation propagates from the lowest flooded

It is equally important to correctly identify the location of

region upward, until the entire tower fills with liquid or

the flood and its nature. We have seen many experiences where

until an abrupt change in hardware design or flow conditions

misdiagnosing the location or the root cause of tower flooding

(eg, feed point) is reached. Flooding may or may not propagate

led to ineffective solutions or debottlenecks. When misdiag-

above that point.

nosed, the problem persists, even worsens, and engineers lose

Flooding is by far the most common capacity limitation in

their credibility, if not their jobs.

distillation and absorption towers. When a tower floods, tray

Once the flood, its cause and location are adequately iden-

efficiency diminishes, separation deteriorates, and products go

tified, proper corrective actions can be devised. The solution

off-spec. Massive liquid entrainment may occur from the top

usually involves re-optimisation (eg changing feed preheat,

of the tower and reach downstream units, causing contamina-

tower pressure) or hardware modifications to tower internals

tion and sometimes equipment damage. Tower flooding often

or auxiliary equipment.

destabilises the tower, with intermittent buildup and dumping

This article will review the tools needed to distinguish

of liquid. This instability may be transferred to downstream

flooding from other issues, and to determine the nature of

or thermally-coupled units. Liquid accumulation in the tower

the flood, its location, and its likely mechanism. Only non-

reduces the bottom flow rate, sometimes starving the bottom

radioactive techniques for flood diagnosis are covered here.

pump and causing it to cavitate. To avoid the onset of flooding,

Radioactive techniques such as gamma scans and neutron

operators cut throughput, and the plant loses capacity.

backscatter, while highly effective for diagnosing flood, are covered in detail elsewhere1 due to space limitation.

Flooding is by far the most common capacity limitation in distillation and absorption towers. When a tower floods, tray efficiency diminishes, separation deteriorates, and products go off-spec

Flood Symptoms Flooding can be recognised by one or more of the following symptoms: 1. Excessive column pressure drop

In existing towers, the onset of flooding limits the tower

2. Sharp rise in column pressure drop

throughput, which often bottlenecks the entire plant. New

3. Reduction in bottoms flow rate

tower diameter is set to sustain operation a comfortable margin

4. Rapid rise of entrainment from column top tray

(typically 20%) below the flood point. In tower revamps, the

5. Loss of separation (as can be detected by temperature

maximum design throughput is restricted by the approach to

profile or product analysis)

(typically 10%) the flood point. Troubleshooting for flood is mandatory for diagnosing the root cause of poor performance in a tower and/or to debottle-

Pressure drop measurements across various column sections are the primary tool for flood point determination.

necking it. Poor separation is caused not only by flooding, but also by inefficient trays or packing, poor distributor design, excessive weeping, or internals damage. Liquid entrainment

Excessive column pressure drop

may be caused by poor reflux piping. Instability may result from

This is due to the liquid accumulation which occurs upon

inadequate control, poor controller tuning, or fluctuating feed

flooding. Typically, pressure drop per tray is 100–130 mm of

rates. Pump cavitation may be due to undersized suction piping

liquid. With most organic and hydrocarbon systems (specific

MARCH 2016 | The Chemical Engineer | page 43

How to TROUBLESHOOT TOWER FLOODS

2.0

INCREASING LIQUID FLOW RATE 1.0

TOP TOWER FULL TOWER LAST 24 FULL TOWER LAST 24 TOP TOWER NOW FULL TOWER NOW TOP TOWER

FLOOD INITIATION

1,500

1,600

1,700

1,800

2,000

2,100

2,200

2,300

ΔP INCHES WATER/FT PACKING

TOWER dP

FLOOD CURVES 17 16.5 16 15.5 15 14.5 14 13.5 13 12.5 12 11.5 11 10.5 10 9.5 9 8.5 8 7.5 7 6.5 6

0.8 0.6

0.4

0.2

0.1

VAPOUR TRAFFIC (REFLUX + PRODUCT)

1

2

4

VAPOUR LOAD (lb/ft s2) 0.5

Figures 1–2 (Left to right): Pressure drop vs internal vapour load from plant operating data5 (Reprinted with permission); Typical pressure drop vs internal vapour load of packed columns10

gravity around 0.7–0.8), this gives 7–9 mbar per tray. If

cases, once flood starts, the pressure drop will keep rising even

the measured pressure drop per tray rises to 15–20 mbar,

when vapour loads are not raised further.

flooding should be suspected. With packed towers, the flood

The flood point can be inferred from the change of slope

pressure drop is given by the Kister and Gill Equation2–4:

in the plot of pressure drop against the tower internal vapour

ΔPflood = 4.17 F

0.7 p

flow rate (see Figures 1 and 2). The internal vapour rate can be , where

inferred from tower instrumentation. For instance, if reflux

ΔPflood = Flood pressure drop, mm water per m of packing Fp = packing factor, m

is not subcooled, the internal vapour rate near the top of the tower equals the reflux plus products. In tray towers, the slope change can be mild or steep. It is not unusual to find a vertically-rising pressure drop once the flood point is reached1,6.

-1

Packing factors for this equation should come from the 8th Edition of Perry’s Handbook4. Packing factors from other sources can lead to inaccurate or even incorrect predictions. Measured pressure drops significantly higher than possible flooding.

ΔPflood

suggest

A tower may flood, yet the pressure drop remains low.

in many packed towers a rapid drop in efficiency occurs well before the hydraulic flood. Here, throughput is limited by loss of separation, and the hydraulic flood point may be of little practical value

The high pressure drop indicates liquid accumulation. When the liquid accumulation is small, the pressure drop may not

In packed towers, the slope change may be continuous (see Figure

significantly rise. Typical scenarios include flooding near

2) rather than abrupt. Further, in many packed towers a rapid

the top of the tower (only a few trays or a short packing

drop in efficiency occurs well before the hydraulic flood. Here,

length

vacuum-packed

throughput is limited by loss of separation, and the hydraulic

towers (accumulation is channelled and the vapour bypasses

flood point may be of little practical value. There are some cases,

the accumulation region); and flooding at low liquid rates (slow

especially in vacuum distillation, where flooding occurs but no

liquid accumulation).

point of inflection is observed7.

accumulate

liquid);

flooding

in

For best results, differential pressure measurements should

Sharp rise of pressure drop

be taken across each section of the tower6,8,9, at least for the flood tests. These can identify the location and conditions

Pressure drop rises with vapour loads. Upon flooding the

when flooding starts, how the flooding propagates, and which

pressure drop rise escalates due to liquid accumulation. In many

remedial action is working.

MARCH 2016 | The Chemical Engineer | page 44

How to TROUBLESHOOT TOWER FLOODS

Figure 310 shows the information that multiple pressure drop recorders can convey. In this tower, gamma scans diagnosed flood initiating at tray number 20 (1 being the bottom tray). Since tray 20 was not highly loaded, this pointed to an installation issue, so a unit shutdown was planned to fix the problem. Figure 3 shows differently; the flood initiated in trays 1–17, and from the relatively large pressure drop rise, probably near the

LC

bottom, and only then propagated to the upper section. The realisation that the flood started at the bottom, not on tray 20,

REFLUX

changed the diagnosis from that of an installation issue to that of a process design issue. A futile and costly unit shutdown to look for a non-existent installation issue was thus prevented. DISTILLATE

Reduction in bottoms flow rate Upon flooding, liquid accumulates in the column, and less reaches the bottom, so the bottom level falls. Most frequently, the bottom level is controlled by manipulating the bottom flow rate, so the level stays constant but the bottom flow rate declines. While a reduction in bottom flow indicates flooding, many

Figure 4: Typical distillation tower overhead system. Reflux drum level control is connected either to distillate valve (shown connected) or to reflux valve (not connected in this diagram)10 (copyright 2013; reprinted with permission)

towers may flood without a significant decline in bottom flow. For instance, when flooding occurs near the top, the bottom section may continue to operate with no significant decline of

When tower overhead flows to a knockout drum, or to another

bottom flow rate. Generally, reduction of bottoms flow rate is a

tower, this entrainment can be recognised as rise of liquid level

good indicator of flooding in towers that flood near the bottom,

in or liquid flow from the drum or the downstream tower. In

and in relatively short towers, particularly if flooding occurs

most distillation towers, the overhead goes to a condenser, then

below the feed.

to a reflux drum. The reflux drum usually has a level control that manipulates either the distillate rate (see Figure 4) or the reflux rate.

Rapid rise in entrainment

When the reflux drum level controls the distillate rate, the

Liquid accumulating in the tower can build up to the top, and is

entrainment rise is often indicated as a significant increase

entrained in the tower overhead stream.

in distillate rate for no apparent reason. When the drum level controls the reflux rate, the entrainment is often indicated as a rise in reflux flow rate for no corresponding increase in boilup. The increased reflux is unable to descend down the tower due to

PREMATURE FLOOD CASE – ΔP CHART

the flooding near the top, so it entrains back into the overhead, returns as additional reflux, and so on. The reflux valve often

UNLOADING TAKING PLACE

opens widely due to the recirculation of entrainment around the tower overhead loop.

#18–62 0–10 PSI

rise of distillate for no apparent reason, rise of reflux for no corresponding boilup increase, or the onset of pressure fluctuations often indicate massive entrainment from the tower

UPPER FLD BEGINS

#18–#1 0–10 PSI

When liquid drainage from the condenser is constrained, which often occurs even under normal operation6, the additional

LOWER FLD BEGINS END

entrained liquid may be unable to drain. It will accumulate in START

Figure 3: Pressure drop profile obtained with multiple pressure drop recorders10 (copyright 2013; reprinted with permission)

the condenser, flooding condenser tubes and reducing condensation rate, which in turn will cause the tower pressure to rise. Eventually, the tower pressure may become high enough to expel the accumulating liquid out of the condenser, exposing tube area and causing the pressure to fall. Once the pressure

MARCH 2016 | The Chemical Engineer | page 45

How to TROUBLESHOOT TOWER FLOODS

falls, the process will repeat. This interaction between the liquid

proved sensitive for detecting early signs of flooding, in turn

carryover and the condenser drainage leads to the pressure fluc-

assisting in pushing towers to their limits11.

tuations sometimes experienced when a tower floods.

Figure 5 shows temperature profiles under normal and

The rapid rise in entrainment indicator is particularly useful

flooded conditions. The tower was uninsulated, and the points

for flooding near the top, when the pressure drop rise is small.

are pyrometer measurements of wall temperatures taken from

This indicator does not detect a flood that does not propagate to

the access ladder. The ladder was on the left of the tower in

the top of the tower.

Figure 5, so all the temperatures are for the even-numbered trays (the odd numbered trays were obscured by the side downcomers descending from the even numbered trays). The crosses map

Loss of separation

the normal temperature profile, showing a discrete reduction in

As flooding is approached, liquid entrainment by the vapour

temperature every two trays. The circles map the temperature

rises sharply. At high pressures and/or high liquid rates, vapour

profile when the bottom 3–4 trays were flooded. Upon flooding,

entrainment in the downflowing liquid also rises. Either type of

the temperature spread across the bottom four trays completely

entrainment lowers efficiency, so separation deteriorates. Since

disappeared, indicating poor separation. It also shows hotter

loss of separation begins before the tower is fully flooded, using

temperatures above, induced by heavier components ascending

it as a flooding indicator can suggest a lower flood point than

due to poor separation in the flooded region.

other indicators. The column temperature profile often provides a good indicator for the separation loss. Liquid accumulation often incurs a temperature rise because the accumulating liquid is richer in heavies, the flooded region separates inefficiently, downflow of cooler liquid from the flooded region is reduced, and the higher

In a troubleshooting investigation there are usually many theories. The troubleshooter’s challenge is to narrow down the theories to a manageable number

pressure drop raises boiling points. Since the loss of separation precedes the hydraulic flood, tracking key temperature changes

Caution must be exercised when curves of this type are interpreted, because they may also indicate a pinch (ie, poor separation due to insufficient reflux or reboil). To distinguish, reflux and reboil can be raised. If separation improves, pinching is indicated. If it deteriorates or stays the same, flooding is indicated.

TRAY 1 #1 #2

Temperature gradients are an effective, low-cost method of determining the flood point, but the success of the method depends on having a large enough number of measurement points and on having a sufficiently large temperature gradient

TRAY 2 #3 #4 TRAY 4 #5 #6

TRAY 6

under normal operating conditions. If the normal tray-to-tray temperature difference is small, as in close separations, the flooded temperature profile will not vary a great deal from the normal profile, and temperature profiles will be poor indicators of flooding.

#7 #8

TRAY 8

TRAY 10

Sight glasses Sight glasses give visual indication of flooding. Sight glasses

#9

are expensive, increase the leakage potential, and may induce

#10

that will permit observation can also be an issue. For these

a chemical release if the glass breaks. Supplying a light source reasons this technique is unfavoured in commercial towers. It is sometimes used with non-hazardous materials near ambient

80

90

100

110

120

130

140

150

pressure.

SURFACE TEMPERATURE, ºF NORMAL OPERATION TEST, UNFLOODED HIGH RATE TEST, FLOODED ON TRAYS 7–10

Figure 5: Flooded and unflooded temperature profiles12 (reprinted with permission)

Determining Flood Mechanism: Vapour and Liquid Sensitivity Tests In a troubleshooting investigation there are usually many theories. The troubleshooter’s challenge is to narrow down the theories to a manageable number. With flooding, one of

MARCH 2016 | The Chemical Engineer | page 46

How to TROUBLESHOOT TOWER FLOODS

the best ways of invalidating incorrect theories is by field tests that show whether the flood is sensitive to vapour, to

OHD

liquid, or to both. For instance, if a flood proves to be sensitive

FC

to vapour but not to liquid, any theory that argues a liquidREFLUX

sensitive flood is denied. In our experience, good vapour and liquid sensitivity tests eliminate about half the theories. In one case10, vapour and liquid sensitivity tests reduced the

FLOOD

number of theories from 12 to five, and eventually led to the root cause. In another case , such tests ruled out a theory that was 13

considered a certainty, thus preventing an incorrect diagnosis

FEED

and solution.

Flood Mechanisms

TC

All floods are characterised by liquid accumulation. There are four different mechanisms that cause this liquid accumulation in trays. 1. Entrainment (jet) flood. Froth or spray height on trays rises with vapour velocity. As the froth or spray approaches the tray above, some of the liquid is aspirated into the tray above as entrainment. Upon further increase in vapour flow rate, massive entrainment of the froth or spray begins, causing liquid accumulation and flood on the tray above.

BOTTOMS

2. Downcomer backup flood. Aerated liquid backs up in the downcomer because of tray pressure drop, liquid height on the tray, and frictional losses in the downcomer apron. All of these increase with increasing liquid rate. Tray pressure drop also increases as the vapour rate rises. When the backup of aerated liquid exceeds the (tray

Figure 6: Distillation tower with temperature control manipulating boilup and reflux entering on flow control. Flood shown near the top in this example10 (copyright 2013, reprinted with permission)

spacing + weir height), liquid accumulates on the tray above, causing downcomer backup flooding. 3. Downcomer choke flood (also called downcomer entrance

.

flood or downcomer velocity flood) A downcomer

2. Flood in the liquid-rich region. At high liquid loads

must be sufficiently large to transport all of the liquid

and high vapour densities, liquid holdup in packed beds

downflow. Excessive friction losses in the downcomer

rises and frothiness increases, impeding liquid drainage.

entrance, and/or excessive vapour venting from

Upon further raising vapour or liquid loads, large liquid

the downcomer in counter-flow, will impede liquid downflow, initiating liquid accumulation on the tray above.

accumulation and flooding initiates. 3. System limit flood (also called ultimate capacity flood). This is the same as in tray towers.

4. System limit flood (also called ultimate capacity flood). This is an ultimate jet flood, and takes place when the

Entrainment (jet) floods, packing vapour-rich floods, and

vapour momentum force acting to lift the large liquid

system limit floods are induced by excessive vapour loads and

drops above the tray exceeds the gravity force. This flood

are therefore highly vapour-sensitive. If there is any sensitivity

is independent of tray geometry and tray spacing.

to liquid loads with these floods, it is small. In contrast, downcomer choke floods, packing liquid-rich floods and floods due

In packed towers, there are three flood mechanisms:

to packing distributor overflows are caused by excessive liquid loads and are therefore highly liquid-sensitive. If there is any

1. Flood in the vapour-rich region. As vapour loads are

sensitivity to vapour loads, it is small. Downcomer backup flood

raised, a point is reached where the vapour rate interferes

can be induced by either excessive vapour load or excessive

with the free drainage of liquid. The bed starts loading up

liquid load, depending on the dominant term in the downcomer

with liquid. Upon further increase in vapour rate, large

backup equation2, 4, and can be sensitive to either.

liquid accumulation takes place and floods initiate.

A common flood test starts with the tower under normal

MARCH 2016 | The Chemical Engineer | page 47

How to TROUBLESHOOT TOWER FLOODS

operation. Reflux is gradually raised until symptoms of flood

from equipment vendors, and in tower simulation software.

are observed. In the tower in Figure 6, this test cannot tell

The available procedures may vary in reliability, so caution is

whether the flood is vapour-sensitive, liquid-sensitive, or both.

required in selecting the appropriate method.

The increase in reflux rate increases the liquid load, but also

First, the tower is simulated to provide the internal vapour

cools the tower, so the temperature controller increases boilup,

and liquid loads and physical properties for each stage in the

raising the tower vapour load. It is impossible to tell whether

tower at the highest throughput before the tower runs into

the flood was due to the initial increase in reflux (therefore,

trouble. These are used in the hydraulic equations.

liquid-sensitive), or due to the subsequent increase in boilup (therefore, vapour-sensitive), or due to both.

Calculating the proximity of flood limits is invaluable in diagnosing the root cause of a tower flood. In one tower,

To determine whether the flood is liquid-sensitive, the

flooding initiated upon feed rate increase. The tower was simu-

temperature control needs to be disconnected, so the boilup

lated at the maximum throughput. Based on the simulation and

is kept constant (on flow control or in manual). The reflux is

tray/downcomer geometry the capacity limits were calculated

raised. If the tower floods, then the flood is liquid-sensitive.

(see Table 1). A value exceeding or approaching 100% for one of

The drawback of this test is that since reflux is raised without a

these parameters indicates flooding by this mechanism.

matching increase in boilup, lights are induced into the bottom product, making it off-spec. Similarly, for a vapour sensitivity test, the reflux is kept constant (in Figure 6 it already is constant as it is flow controlled) and the boilup is raised. In this test the temperature controller can remain in auto and the temperature set point is

With flooding being a major source of tower problems and bottlenecks, correct diagnostics of the flood, its location, and its root cause are central to a correct cure

raised. Flooding in this test indicates a vapour-sensitive flood. This test induces heavies into the distillate and gets it off-spec.

In this tower, the gamma scans indicated flooding initiating

The good news is that these are usually quick tests; if

near the feed. Table 1 shows that near the feed all trays operated

performed correctly each test normally will yield the answer

a comfortable margin away from jet and downcomer backup

within 2–3 hours. Once these two tests are performed, all the

floods. Two different methods were used to evaluate down-

theories that did not predict the observed sensitivities are

comer choke. Method 1 (accurate data interpolation) showed

invalidated.

that no trays approached downcomer choke flood. Method

Overall, the trick here is to test the response of the tower to one variable at a time.

2, a conservative correlation, showed the bottom trays near downcomer choke flood, but the trays near the feed operated a comfortable margin from flood. Since the gamma scans showed

Determining Flood and Flood Mechanism: Hydraulic Analysis

flood near the feed, and no flood near the bottom, downcomer choke flood was ruled out as the root cause. This analysis directed the troubleshooting efforts towards

Hydraulic procedures are available in distillation texts2–4 to

the feed arrangement, leading to the diagnosis that a

calculate the proximity to the points of initiation of the various

poorly-designed feed entry caused the flooding14. This in turn

flood types. In addition, hydraulic calculation software is avail-

led to a successful fix and debottleneck.

able from technology suppliers like Fractionation Research,

Concluding Thoughts Table 1: hydraulic analysis that helped diagnose root cause of tower flooD just above just below at bottom feed feed % jet flood

55

39

56

% froth in downcomer

56

56

69

% downcomer choke, interpolation

71

55

72

% downcomer choke, correlation

90

79

103

Troubleshooting distillation towers is analogous to the conventional practice of medicine, in which the doctor employs a variety of techniques and tests to diagnose a problem. Based on the diagnosis the doctor implements a cure. A correct diagnosis often leads to an effective cure that heals the patient quickly, while a poor diagnosis may prolong patient suffering. The same applies to distillation. An incorrect diagnosis and an ineffective cure retard, even prohibit, recovery from a tower problem. With flooding being a major source of tower problems and bottlenecks, correct diagnostics of the flood, its location, and its root cause are central to a correct cure. Many tools and techniques can be applied to correctly diagnose flood. Like the doctor, the troubleshooter must be versed in these techniques and their correct application. This article aims to familiarise engineers with techniques

MARCH 2016 | The Chemical Engineer | page 48

How to TROUBLESHOOT TOWER FLOODS

useful for diagnosing flood, and with the practical experience

7. Kister, HZ, Rhoad, R and Hoyt, KA, “Improve Vacuum-Tower

learnt at the school of hard knocks. It is our hope that this expe-

Performance”, Chem Eng Progr, p36, September 1996.

rience will help troubleshooters arrive at the correct diagnosis

8. AIChE Equipment Testing Procedure, Trayed and Packed Columns: a

and reach an effective solution to the problem in hand.

Guide to Performance Evaluation, AIChE, January, 2014. 9. McLaren, DB, and Upchurch, JC, “Guide to Trouble-Free Distillation,” Chem Eng, 1 Jun, 1970, p139.

References

10. Kister, HZ, Practical Distillation Technology, course manual,

1. Kister, HZ, “Common Techniques for Distillation Trouble-

2013.

shooting”, Chapter 2, in A Gorak and H Schoenmakers (Eds),

11. Dzyacky, G, and Carlson, S, “Improve Column Performance:

Distillation: Operation and Application, Elsevier, 2014.

Operate Closer to the Hydraulic limit without Flooding”, in

2. Kister, HZ, Distillation Design, McGraw-Hill, New York, 1992.

Distillation 2009: Proceedings of Topical Conference, AIChE National

3. Strigle, RF Jr, Random Packings and Packed Tower, 2nd ed, Gulf

Spring Meeting, p369, Tampa, FL, April 2009.

Publishing, Houston, Tx, 1994.

12. Kister, HZ, Larson, KF, Burke, JM, Callejas, RJ and Dunbar, F,

4. Kister, HZ, Mathias, P, Steinmeyer, DE, Penney, WR and Fair,

“Troubleshooting a Water Quench Tower”, Proc of the 7th Annual

JR, “Equipment for Distillation, Gas Absorption, Phase Disper-

Ethylene Producers Conference, Houston, Texas, Mar 1995.

sion, and Phase Separation”, Section 14, in RH Perry and D

13. Ponting, J, Kister, HZ, and Nielsen, RB, “Troubleshooting

Green, Chemical Engineers’ Handbook, 8th Ed, 2008.

and Solving a Sour Water Stripper Problem”, Chemical Engineer-

5. Kister, HZ, Clancy-Jundt, B, and Miller, R, “Troubleshoot-

ing, p28, Nov 2013.

ing a C3 Splitter Tower, Part 1, Evaluation”, PTQ, p97, Q4, 2014.

14. Kister, HZ, Grich, DE and Yeley, R, “Better Feed entry Ups

6. Kister, HZ, Distillation Operation, McGraw-Hill, NY, 1990.

Debutanizer Capacity”, PTQ Revamp and Operations, p31, 2003.

Practical Distillation Technology 25–27 July, Singapore 12–14 September, London, UK Presented by Henry Kister, a recognised specialist with a vast background in all phases of distillation, this course examines distillation technology in detail. With particular emphasis on the problems that can occur and how to solve them, you’ll discover key techniques for trouble-free operation and reduced distillation cost. Book Singapore – www.icheme.org/pdtsing Book London – www.icheme.org/pdt This course can also be run in-house, email [email protected] to request a quotation.

MARCH 2016 | The Chemical Engineer | page 49 1362_16 PDT ad 125x175.indd 1

16/02/2016 16:20