Design and Rating of Packed Distillation Columns

GBH Enterprises, Ltd.

Process Engineering Guide:
GBHE-PEG-MAS-612

Design and Rating of Packed
Distillation Columns
Information contained in this publication or as otherwise supplied to Users is
believed to be accurate and correct at time of going to press, and is given in
good faith, but it is for the User to satisfy itself of the suitability of the information
for its own particular purpose. GBHE gives no warranty as to the fitness of this
information for any particular purpose and any implied warranty or condition
(statutory or otherwise) is excluded except to the extent that exclusion is
prevented by law. GBHE accepts no liability resulting from reliance on this
information. Freedom under Patent, Copyright and Designs cannot be assumed.

Refinery Process Stream Purification Refinery Process Catalysts Troubleshooting Refinery Process Catalyst Start-Up / Shutdown
Activation Reduction In-situ Ex-situ Sulfiding Specializing in Refinery Process Catalyst Performance Evaluation Heat & Mass
Balance Analysis Catalyst Remaining Life Determination Catalyst Deactivation Assessment Catalyst Performance
Characterization Refining & Gas Processing & Petrochemical Industries Catalysts / Process Technology - Hydrogen Catalysts /
Process Technology – Ammonia Catalyst Process Technology - Methanol Catalysts / process Technology – Petrochemicals
Specializing in the Development & Commercialization of New Technology in the Refining & Petrochemical Industries
Web Site: www.GBHEnterprises.com

Process Engineering Guide:

Design and Rating of Packed
Distillation Columns

CONTENTS

SECTION

0

INTRODUCTION/PURPOSE

3

1

SCOPE

3

2

FIELD OF APPLICATION

4

3

DEFINITIONS

4

4

DESIGN PHILOSOPHY

4

5

PERFORMANCE GUARANTEES

5

6

DESCRIPTION OF PACKED COLUMN INTERNALS

6

7

DESIGN CALCULATIONS

7

7.1
7.2
7.3

Selection of Packing Size
Rough Design
Detailed Design and Rating

7
7
9


8

LIQUID DISTRIBUTION AND REDISTRIBUTION

12

8.1
8.2
8.3
8.4
8.5

Basic Concepts
Pour Point Density
Peripheral Irrigation - the Wall Zone
Distributor Levelness
Maximum Bed Height and Liquid Redistribution

12
13
13
13
14

9

PRACTICAL ASPECTS OF PACKED COLUMN DESIGN

14

9.1
9.2
9.3
9.4
9.5
9.6
9.7
9.8
9.9

Packing
Support Grid
Liquid Collector
Liquid Distributor or Redistributor
Packing Hold-down Grid
Reflux or Feed Pipe
Reboil Return Pipe
Liquid Draw-offs
Vapor Draw-offs

14
15
16
17
21
21
23
23
23

10

BIBLIOGRAPHY

25

APPENDICES

A

DEFINITIONS

26

A.1

INTRODUCTION

26

A.2

MECHANICAL DEFINITIONS

26

A.3

PERFORMANCE DEFINITIONS

26

B

PACKING HYDRAULICS - THE NORTON METHOD

30


TABLES

1

PACKING FACTORS FOR THE MORE COMMON
RANDOM PACKINGS

32


FIGURES
1

PACKED COLUMN - GENERAL ARRANGEMENT

6

2

LIQUID AND VAPOUR LOAD CONSTANTS, KL & KV,
FOR TRIAL DIAMETER CALCULATION

8

3

PLATE EFFICIENCY CORRELATION OF O'CONNE

10

4

DIAGRAMMATIC ARRANGEMENT OF GAS-INJECTION
SUPPORT

15

5

LIQUID COLLECTORS - VANE AND CHIMNEY TYPES

16

6

MULTI-SPRAY DISTRIBUTORS

17

7

LADDER OR PIPE LATERAL

18

8

ORIFICE DRIP PANS

18

9

NOTCHED TROUGH DISTRIBUTOR

19

10

ORIFICE TROUGH DISTRIBUTOR

20

11

TWO PHASE FEEDS

22

12

VAPOR SPARGER

24

13

GENERALIZED PRESSURE DROP CORRELATION FOR
RANDOM DUMPED PACKINGS

30

DOCUMENTS REFERRED TO IN THIS PROCESS
ENGINEERING GUIDE

45


0

INTRODUCTION/PURPOSE

GBHE does not manufacture packings or the associated column internals - they
are purchased from specialist manufacturers. The detailed design of the
packings and internals is undertaken by the manufacturers. In most cases,
competitive bids are sought from various manufacturers.
The role of the Process Engineer is to:
(a)

specify the process requirements, in the form of data sheets;

(b)

ensure that what is offered by the bidding manufacturers will meet these
requirements; and

(c)

compare the designs offered on technical merit; outstanding features of a
particular design may outweigh any additional cost which may be incurred.

On existing plant there is often the need to assess the performance of packed
columns for several reasons:
(d)

to assess the reasons for any shortfall in performance, compared with
design or earlier operation;

(e)

to assess uprating capability of the existing packings and liquid
distributors, from high rate plant trials; or

(f)

to explore modifications for operation at higher or lower rates.

Packings are made in metal, ceramic and plastic; the vast majority of distillation
applications use metal packings. Ceramic packings tend to give problems of
breakage in service. They are used in corrosive services. Few plastics can
tolerate the temperatures encountered in distillation.
Three families of packings exist: random, structured and grid packings. Random
packings are the traditional ones and various non-proprietary types are available;
newer high performance random packings have been developed, but all are
proprietary. Structured packings are the newest ones and most of these are
proprietary. Grid packings are not generally used for distillation, because of poor
efficiency, but can be useful in fouling services.


The design for proprietary packings and especially structured packings depends
on the design methods supplied by their manufacturers. These design methods
are sometimes unreliable and expert advice should be sought when considering
proprietary packings.
This Guide has been prepared for GBH Enterprises.

1

SCOPE

This Guide deals with the design and rating of packed distillation columns. It
covers neither guidance on the selection of trays and packings nor some aspects
of their performance characteristics; advice on both of these is given in GBHEPEG-MAS-610 - Selection of Internals for Distillation Columns.
Guidance is also given on the assessment of liquid distributors - the detailed
design of distributors requires specialist knowledge and experience not available
in GBHE, though we are able to check some of the more critical aspects of
design. Also, some of the more important practical issues are discussed.
The scope of the guide is summarized in its clause headings:
4
5
6
7
8
9

Design Philosophy
Performance Guarantees
Description of Packed Column Internals
Design Calculations
Liquid Distribution and Redistribution
Practical Aspects of Packed Column Design

In addition, Appendices give examples of design calculations by various methods
for both random and structured packings, list packing factors for many of the
available random packings, provide definitions of terminology used for packed
columns, and the topics relevant to packings in the FRI Design Practices Manual
(Ref. [1]).

2

FIELD OF APPLICATION

This Process Engineering Guide applies to the design of packed distillation
columns by process engineers in GBH Enterprises world-wide.


3

DEFINITIONS

For the purposes of this Guide the following definitions apply:
Fractionation Research

A co-operative research company. Many of the
definitions they use Inc. for packed column
design are given in Appendix A of this Guide.

With the exception of proper nouns, terms with initial capital letters which appear
in this document and are not defined above, are defined in the Glossary of
Engineering Terms.

4

DESIGN PHILOSOPHY

It is assumed that GBHE-PEG-MAS-610 “Selection of Internals for Distillation
Columns” has been consulted and that, even if a firm decision to use trays has
been made, the selection criteria have been checked to ensure that important
factors have not been overlooked.
GBHE does not manufacture packings or the associated internals for distillation
columns - they are purchased from specialist manufacturers. Generally speaking,
the packing manufacturers have more experience in packed column design than
GBHE process engineers. It is therefore preferable for the manufacturer to take
responsibility for the design. The role of a GBHE Process Engineer then
becomes one of:
(a)

comparing the technical merits of designs proposed by manufacturers in a
competitive bid situation;

(b)

ensuring that what is proposed will work; and

(c)

seeking modifications where proposals appear unsatisfactory.

If modifications are required, the aim should be to agree changes with the
manufacturer which do not diminish his design contingencies, while removing the
risk of poor performance. This is generally achievable since we will usually be
seeking extra contingency in the design.
Until recently, manufacturers' expertise has been mainly in the hydraulic design
(flooding, weeping, pressure drop, etc.), but this is now changing. Most
manufacturers will enter into discussions and share their experience on efficiency
as it relates to our application.

Especially when they have prior experience with the same system or have done
tests on our application (sometimes at their cost, but more usually at ours), they
will be able to predict efficiency.
It is important to distinguish between internals design and column design. Almost
certainly, the vessel will be on a much longer delivery than the packing and other
internals - typically about 1 year compared with 12 weeks. We often need to
decide the column diameter and approximate height long before we want to talk
to packing manufacturers. Usually this can be done using our own design
methods and ensuring sufficient contingency for the manufacturer's final design.
The recommended practice for packed column design is dependent on the
purpose for which the design is required:
(d)

A rough column sizing is usually all that is needed for initial flowsheeting
studies;

(e)

Approximate designs are required for pre-sanction flowsheeting, where
cost estimating is the main requirement;

(f)

Detailed designs are required for:
(1)

(2)

assessment of existing column capacity, either with the existing
internals or by considering modified designs; and

(3)

(g)

in-house column sizing, prior to enquiry on tray manufacturers, to
determine diameter and approximate height;

checking manufacturers proposals, either for a new column or for
debottlenecking an existing column.

For a sanction estimate, enquiry on manufacturers is strongly
recommended. In-house designs are likely to require modification when
manufacturers are consulted, unless generous contingencies have been
incorporated. Such contingencies are likely to give scope for future
uprating, but they may add significantly to the cost.


5

PERFORMANCE GUARANTEES

The question arises of whether to seek guarantees from manufacturers. This
needs to be decided by each project team, because there are pros and cons. If a
guarantee is obtained, the manufacturer is bound to do all he can to resolve any
shortcoming in his design, up to the limit of his liability (which will be stated in the
guarantee). In practice, it may not be clear whether a particular problem has
been caused by the design, the installation, or the subsequent operation.
Since the manufacturer is bound by a guarantee, he will be concerned not to say
or do anything which could be construed as an admission of liability. Experience
shows that, even where there is no guarantee, most manufacturers are very
willing to help us resolve the problem - after all, they stand to learn from a
detailed knowledge of what went wrong, no matter whose fault it is.
Guarantees do not cover consequential losses. If a failure occurs and the plant is
shut down, the consequential loss will far exceed any sum in the manufacturer's
guarantee - in cash terms a guarantee is of little value. Furthermore, the
guarantee will generally contain clauses defining requirements of access for
testing, dismantling and examination in the event of failure. The guarantee may
be unenforceable if we do not meet these requirements.
Before seeking a guarantee, it is important to be clear that it will be helpful to us
when a problem arises and that it can be enforced. An efficiency guarantee is
generally the most difficult to enforce. For example, if we have specified the
number of trays on the basis of our own vapor-liquid equilibrium (VLE) model and
the manufacturer has specified efficiency, he may claim that a failure to meet
design is due to an error in our VLE model. Such a claim is very difficult to refute.

6

DESCRIPTION OF PACKED COLUMN INTERNALS

Most designs of packed column internals are essentially similar and their main
features are described in this Section. In describing internals, customary
terminology is introduced. A list of definitions is given in Appendix A. Most of the
definitions are those used by Fractionation Research Inc and are understood by
the majority of packing manufacturers.


FIGURE 1

PACKED COLUMN - GENERAL ARRANGEMENT

Distillation columns are always cylindrical vessels. Figure 1 shows the general
arrangement of a packed column, with two packed beds and feed between them.
Liquid is fed to the top of a bed through a liquid distributor. A bed limiter is placed
on top of the packing to restrict packing movement in the event of large surges in
vapor flow. The packing rests on a support grid.


Below the support grid is a liquid collector. For some types of liquid distributor,
this may not be necessary, but is recommended for large columns (> 1 m ID) to
ensure that liquid from the bed above and the liquid feed are fully mixed. Liquid
from the collector is then taken to the distributor at the top of the next bed.
Below the bottom bed there is often some form of vapor distributor. It is much
easier to achieve good vapor distribution than good liquid distribution. The reboil
return pipe shown in Figure 1 is a design commonly used; it has a bottom
aperture at the column centre line which directs the vapor and associated liquid
downwards. Liquid falls to the sump and vapor turns round and flows up to the
packing. This type of design also prevents vapor/liquid mixture from the reboiler
impinging on the column wall opposite the inlet nozzle - otherwise erosion may
cause vessel weakening in this area.

7

DESIGN CALCULATIONS

Three categories of design can be identified which fall roughly in line with the
stages of estimate in a project for a new plant:
(a)

Class D - budget - rough design.

(b)

Class C - pre-sanction - preliminary design, still mainly for costing.

(c)

Class B - sanction - detail design.

For the assessment of existing packed columns or for plant modifications (except
where a new column is required) all calculations will be in the detailed category.
This Section is concerned with random and structured packings (grid packings
are not generally used in distillation service). They are considered as two
separate families since their performance differs significantly and cannot be
represented by one method. Generally speaking, the design methods we have
are not sufficiently reliable for final design except for duties where we have prior
experience, either on the full scale or in a pilot unit. This is true both of hydraulic
and efficiency correlations. For proprietary packings we generally have to rely
on manufacturers data which varies from simple charts in sales brochures to
computer programs which contain the manufacturers own correlations, with some
know-how not disclosed.


7.1

Selection of Packing Size

One of the first considerations is the size of packing which should be used. For
random packings, For random packings general practice is to use 38 or 50 mm
rings for all but small columns (< 500 mm diameter). By analogy, for sheet
structured packings, the common sizes are in the range of 200 to 350 m2/m3
specific surface area. Gauze structured packings tend to be in the region of 500
m2/m3.

7.2

Rough Design

For Class D purposes, it is rarely necessary to distinguish between random and
structured packings. Two calculations are required:
(a)

hydraulic capacity to estimate column diameter; and

(b)

packing efficiency to estimate column height.

The following equation provides an estimate of column diameter, assuming metal
random packings, given a packing factor:


FIGURE 2

LIQUID AND VAPOUR LOAD CONSTANTS, KL & KV, FOR TRIAL
DIAMETER CALCULATION


To a first approximation, packing efficiency (expressed as height equivalent to a
theoretical plate, HETP) may be estimated using the plate efficiency correlation
of O'Connell (Ref [2]), Figure 3, since packings and trays respond similarly to the
VLE and transport properties of systems. O'Connell correlated efficiency in terms
of relative volatility and liquid viscosity. This may be used to estimate packing
HETP:
HETP = H/(plate efficiency %),

m

where H is a function of packing size:
Size, mm
H

7.3

15
11

25
11

38
13

51
15

89
26

Detailed Design and Rating

When the detailed design stage has been reached it is strongly recommended to
gather the design data on the appropriate data sheet proformas. This should be
done as a separate step in the design process, to avoid confusion and errors
arising from ad hoc reference.


Detailed design and rating calculations are required in the following
circumstances:
(a)

column sizing, to determine diameter and height, prior to enquiry on
packing manufacturers;

(b)

performance assessment of existing columns;

(c)

exploring design options for uprating an existing column or changing its
duty;

(d)

comparing manufacturers' proposals to establish their suitability and rank
them on technical merit.

Manufacturers' quotations should contain enough detail for assessment. If they
do not, we should request the missing information - the packing and internals
manufacturing business is a highly competitive one and we can choose to go to
another supplier, so we usually get what we ask for, provided our request is
reasonable.
Care is needed in using plant data for either the design of a new column or the
assessment of an existing column for a new duty. For example, plant data may
show, for a given system with a given column design (including packing,
distributors, etc.), that flooding occurs at 105% of the predicted flood point.
Beware:
(a)

How accurate were the plant data? It should be possible to obtain plant
data which show a mass balance accurate to within 5% and a heat
balance within 10%, but this is not often achieved in practice.


(b)

The same system may not perform so well on a different packing
type or column diameter - it may flood at a lower predicted % flood
if the correlation does not properly account for the effects of
mechanical features in the tray design.

(c)

A different system may not perform so well on the same equipment
design - it may flood at a lower predicted % flood if the correlation
does not properly account for the effects of system physical
properties.

7.3.1 Random Packings
The methods described in this Section have been implemented as a Calculation
in some commercially available programs. Details of the methods are given in
Appendix B and may be used for manual calculation.
The four main calculations to be performed are:
(a)

Flood point at constant liquid rate L/V ratio;

(b)

Load point at constant liquid rate or constant L/V ratio;

(c)

Packing pressure drop;

(d)

Packing efficiency, as HETP (height equivalent to a theoretical plate).

For the first three, a generalized pressure drop correlation published by the
Norton Company (Ref [3]) is used - see Appendix B. It is the latest in a long
series of pressure drop correlations based on the original work of Sherwood (Ref
[4]). Many variants have been developed and published in the open literature and
in manufacturers' sales literature.
The flood point is taken to be coincident with a frictional pressure gradient of 1.5
in.H2O/ft (12.3 mbar/m) of packing. In some cases, packings will operate at
higher pressure gradient, but mass transfer is almost certain to be poor.
The load point is defined as the combination of vapor and liquid loads at which
the mass transfer efficiency is a maximum and above which it declines sharply.
The load point may be taken to be 90% of the flood point. Design for 90% of the
load point is generally recommended for well behaved systems.

Foaming systems rarely affect packed columns seriously, but a strong foaming
tendency can reduce capacity. It is believed that the flood point declines towards
the load point in foaming systems and therefore design should be for less than
90% load. Only experience on the same or similar systems can give a firm
indication of a design value, but a value of 70% load should cover most cases.
Packing efficiency (HETP) is predicted using an FRI correlation (Ref [1], Vol 2,
Section 8.4) which was fitted to data on low relative volatility hydrocarbon
systems from 250 mbar to 15 bar, and methanol/water at atmospheric pressure.
Its performance with other types of system, especially with relative volatilities
greater than 5, is not proven.

7.3.2 Detailed Design and Rating - Structured Packings
Structured packings are all proprietary devices. We have no non-proprietary
methods for the design of structured packing columns. We do not know what
methods are used in these programs and although it is clear that the correlations
used are very similar to (or the same as) the ones used by the manufacturers.
We do not know
how accurate they are.
We therefore rely to a greater extent on manufacturers in the supply of structured
packings than is the case for random packings. This is particularly true in two
circumstances:
(a)

the prediction of efficiency for all systems; and

(b)

the prediction of capacity for high pressure systems (> 15 bar).

If neither we nor the manufacturer have prior experience with a similar system we
should not proceed without carrying out tests to measure efficiency, at least at
total reflux. Most manufacturers have access to small scale test units in which
such tests can be performed, but these may not be able to operate under all the
conditions we require. This is a potentially difficult area where expert advice
should be sought.


8

LIQUID DISTRIBUTION AND REDISTRIBUTION

The performance of a packed distillation column is critically dependent on the
quality of liquid distribution to each of the packed beds. Poor liquid distribution is
the most common cause of unexpected poor separation. A full discussion of the
problems is presented in the FRI Design Practices Manual (Ref [1], Vol 5,
Section 2.02). The following is a summary.
Liquid distribution is discussed in relation to random dumped packings, where
there has been thorough research by FRI. The behavior of structured packings is
believed to be similar.

8.1

Basic Concepts

A perfect distributor may be considered as one which lays down liquid uniformly
over the entire cross section of the column on the top of the packed bed. In
practice this is not achievable:
(a)

the liquid is distributed from holes or slots and these must be large enough
not to block,

(b)

the distributor will not be perfectly level,

(c)

space must be allowed for the passage of vapor, and

(d)

process flexibility (turndown) is generally required.

Indeed, perfect distribution is not necessary. Research results and mathematical
simulation both show that packings have a characteristic "natural" distribution
which is less than perfect, and that an excellent initial distribution is quickly
degraded to the natural distribution - it takes only a few layers of packing. On the
other hand, if the initial liquid distribution is not as good as the packing's natural
distribution, it takes a substantial bed depth (meters in extreme cases)
before the natural distribution is achieved.
Although an over-simplification, it is helpful to think of liquid flowing over a
packing as a series of discrete streams rather than a uniform film. The smaller
the packing the greater the number of streams and the better the natural
distribution. Small packings can therefore be expected to be more sensitive to
poor distributor design and this indeed is so.


Clearly, the liquid streams do not flow vertically through the packed bed but are
diverted laterally, in a random manner, by the packing elements. The extent of
this lateral diversion may be expressed as a "spreading coefficient" which has
the characteristics of a diffusion coefficient. The spread factor is easily
determined experimentally. Large packings have higher spread factors than
small packings.

8.2

Pour Point Density

With the stream concept of natural distribution, it is important that the number of
streams leaving the liquid distributor should be at least equal to the number of
streams in the natural distribution. Based upon measurements of the natural
liquid distribution of Pall rings, minimum values of pour point density for random
packings are recommended:
Ring size, mm
Pour points per m2

15
100

25
60

38
40

50
35

89
30

Distributors for sheet metal structured packings tend to have 60 to 100 pour
points per m2, while distributors for gauze structured packings may well have 200
or more.
Pour point density and peripheral irrigation (see 8.3 below) need to be
considered together; in small columns it may be that the requirement for
peripheral irrigation can be met only with a higher pour point density than given
above.
The holes may be on either a triangular or square pitch, but the pattern should be
complete across the entire tray - vapor risers and support beams should fit within
the pattern without eliminating any pour points.

8.3

Peripheral Irrigation - the Wall Zone

It has been thought for many years that a common cause of poor performance in
packed beds (high HETP) was the development of wall flow - liquid reaching the
wall, flowing down it, bypassing the packing and not contacting the vapor
effectively. With this in mind, many commercial distributors were designed so that
a small zone along the column periphery was not irrigated.


FRI research has showed that a peripheral zone with insufficient liquid could
double the effective HETP - this was obtained with 15 mm packing in a 1.2 m dia
column with a 75 mm peripheral annulus which was not irrigated with liquid from
the distributor. Larger packings, with higher spreading coefficients are not so
severely affected by a 75 mm wall zone. The recommended maximum distance
between the pour points in a distributor and the column wall, as a function of
packing size (random packings) is:
Ring size, mm
Max distance, mm

15
25

25
25

38
38

51
51

89
76

Given that the pour points are usually arranged on a triangular or square pitch,
the criterion for the wall zone may require a larger number of pour points than the
table in 8.2, above. This is most likely with small columns. Some manufacturers
produce designs for small columns with the pour points in concentric rings. Such
a design is not generally recommended since it is difficult to get an even
distribution and efficiency may well suffer. Seek expert advice for an assessment
of such a design.

8.4

Distributor Levelness

Ideally, the liquid flow through all the pour points in a distributor should be exactly
the same. Random variations, due to minor differences in hole finish (burrs, etc.)
do not cause problems, but serious problems can arise if the distributor is not
level. A key factor in the consideration of out-of-levelness is the shape of the
orifices which control the liquid flow. Three shapes are in common use and they
have different head/flow characteristics:
(a)

for a V notch, flow is proportional to head to the power of 2.5;

(b)

for a rectangular notch, flow is proportional to head to the power of 1.5;

(c)

for a submerged orifice, flow is proportional to head to the power of 0.5.

The V notch is therefore the most sensitive while the orifice is the least sensitive
to out-of levelness. Out-of-levelness may be tackled by calculation and
specification, but is usually dealt with by distributor testing - see 9.4.4.


8.5

Maximum Bed Height and Liquid Redistribution

Although FRI research in the early 1980s overturned traditional thinking about
the influence of wall flow in packed beds, the phenomenon still exists. At the top
of a packed bed, a modern distributor design is intended to put liquid evenly on to
the packing. As the liquid passes through the bed it spreads and on reaching the
wall some of it runs preferentially down the wall instead of returning to the
packing. The build-up of wall flow in this manner eventually causes
sufficient maldistribution to lower the efficiency of the packed bed.
Packed beds with modern random packings have, in some cases, operated
successfully with beds up to 13 m deep, but in other cases they have failed - for
reasons that are not obvious (or have not been made public). Experience with
beds up to 8 m deep has proved satisfactory and this is suggested as a working
maximum, subject to the condition that the number of theoretical stages in the
bed should not exceed 20.
As liquid travels down a long bed and wall flow begins to develop, it results in
variable liquid/vapor flow ratios across the column at any particular elevation.
Thus a radial composition profile develops in both the vapor and liquid flows. The
purpose of redistribution is to even out both the radial liquid flow profile and also
the radial composition profile. Thus the redistributor needs to fully mix the liquid
leaving a bed before distributing it to the bed below. Some commercial designs
make specific provision for this mixing to take place; others do not. The same is
in principle also true of the vapor, but specific provision is not made for either
mixing or redistribution. It is probable that the tortuous passages through the
risers, and the resultant turbulent eddies at their exit, provide good mixing and
the pressure drop of the bed above ensures good vapor distribution.

9

PRACTICAL ASPECTS OF PACKED COLUMN DESIGN

The FRI Fractionation Tray Design Handbook Vol 5 contains much invaluable
information on the practical aspects of packed column design.

A brief description of packed column internals is given in Clause 6. This Clause
gives more detail and recommendations on some of the more common problem
areas.


9.1

Packing

Structured packing is installed in layers, usually about 250 mm high. For small
columns (< 1 m) each layer may be supplied as one or two pieces. For larger
columns, the pieces are typically 250 mm wide and high by up to 2 m long,
shaped to fit the column. Manufacturers differ in how the packing relates to the
column wall. One approach requires a gap between packing and wall with wall
wiper strips fitted to the packing to bridge the gap. The other approach is to make
the packing a tight fit against the wall. Both approaches are intended to
avoid problems with wall flow, but the former approach is preferred.
Random packing (except in ceramics) is usually tipped into a column from the
bags in which it is supplied through a sock so that the maximum packing fall is no
greater than 2 - 3 m and the packing is not seriously damaged. For ceramic
packings, it is usual to fill the column with water first to minimize breakage.
Breakage is often a problem in the supply of ceramic packings, a significant
proportion being damaged on delivery.

9.2

Support Grid

The support grid for random packings usually takes the form of illustrated in
Figure 4 and known as a 'gas injection' or 'multi-beam' type. The requirement
which leads to such a design is that the apertures in the support plate should be
small enough to prevent packing pieces falling through, but the area for vapor
flow must not be severely restricted by the area taken up by the ligaments
between the apertures. The typical gas injection support plate has an area for
vapor flow equal to the cross sectional area of the column. The design is also
said to segregate the vapor and liquid flow; the vapor flows through the slots in
the risers while liquid falls through orifices at the bottom.


FIGURE 4

DIAGRAMMATIC ARRANGEMENT OF GAS-INJECTION
SUPPORT


9.3

Liquid Collector

As explained in 8.5 there is a practical limit to bed height. When more stages are
required, additional beds must be added. Between these beds, liquid from the
bed above has to be collected, mixed and redistributed to the bed below. Also, if
there is a liquid feed, this too must be mixed with the liquid coming from the
upper bed. The two basic types are a chimney tray and a vane collector,
examples of which are shown in Figure 5. The collector discharges into
an annular and/or a central channel, where the liquid is mixed before feeding to
the liquid redistributor. Liquid feed, if any, is fed into the collector or the channel.
A liquid sidestream may be taken from the collector or its channel, but this
usually entails special design to ensure that the liquid is de-aerated before
withdrawal.
FIGURE 5

LIQUID COLLECTORS - VANE AND CHIMNEY TYPES


9.4

Liquid Distributor or Redistributor

There are four basic types which are in common use: pipe lateral, pan, trough
and spray. Figures 6-9 show simple illustrations of the general principles. The
spray type distributor (see Figure 6) is NOT RECOMMENDED for mass transfer
duties - its usual application is in pumparounds where the main duty is direct
contact heat transfer. A redistributor differs from a distributor in that it takes liquid
from a packed bed above it and distributes it to the bed below. The design of a
redistributor is essentially the same (and often actually the same) as a distributor,
but the main difference is likely to be the provision of "hats" over the vapor risers
to prevent liquid from the bed above falling through the risers to the bed below. A
few general comments follow.


FIGURE 6

MULTI-SPRAY DISTRIBUTORS

The distributor should be located above the bed of packing (typically 150 mm) to
provide sufficient free vapor space in the column to ensure liquid disengagement
before reaching the confined vapor passages through the distributor.
The vapor passages should be evenly distributed over the column cross section.
Designs should be avoided where the vapor is required to move horizontally for
any considerable distance at high velocity so that it causes the descending liquid
to be displaced laterally.
A common error in distributor design is to provide so many distribution points that
the minimum liquid head at which the holes operate is too low. The minimum
liquid head should be such that the maximum out-of-levelness can be tolerated.
The minimum acceptable hole size depends on:
-

the presence of or tendency to form solids;

-

feed filtration facilities, if any;

-

corrosion or erosion, if any (small holes are affected to a greater extent
than large holes by corrosion and erosion).

For non-fouling, non-corrosive systems with a high quality filtration (to say, one
fifth to one tenth of the distributor hole size), a hole size of 2-3 mm may be
considered acceptable. More usually, hole sizes of 5-8 mm are considered the
minimum acceptable for clean non-corrosive systems with no feed filtration.
For fouling systems, corrosive systems, or where solids are present in the feed,
larger holes are required and in some cases notched troughs may be the

only practical choice. Especially in new equipment, beware of rust, fouling, mill
scale, etc. from the shell material or upstream equipment.

9.4.1 Pipe Lateral Distributor (Figure 7)
Primarily, this is designed for use where the feed liquid is available under
pressure and where the turndown does not exceed 2.5:1. It has a high free area
for vapor flow which gives a low risk of local flooding and usually allows the
distributor to be placed close to the packing. It should be used only with clean or
filtered liquids. The maximum liquid rate is in the region of 0.01 m 2/s.
FIGURE 7

LADDER OR PIPE LATERAL

9.4.2 Pan Distributor (Figure 8)
The pan distributor is adaptable to a wide range of liquid flow rates by varying the
size and number of liquid orifices in the pan. A turndown of 3:1 is achievable and
higher values may be obtained with special designs. Pans for small columns (up
to 1.2 m) often take the form illustrated in Figure 8 (a). The pan is mounted on
lugs fixed to the column wall. Gas flows up through the gas risers and the
peripheral gap around the pan. In larger columns the pan (Figure 8 (b)) has no
outer wall and is mounted on a support ring. The gas risers may be of circular,
square, or rectangular cross section. They must be sized and located so that
they do not interfere with the distribution pattern of the liquid pour points.


FIGURE 8

ORIFICE DRIP PANS

9.4.3 Trough Distributors
There are two main types of trough distributor: the notched trough and the orifice
trough – see Figures 9 and 10.
FIGURE 9

NOTCHED TROUGH DISTRIBUTOR

For many years the notched trough was the most popular type for large columns
- above about 1.5 m dia. However, because of the head/flow relationship
mentioned earlier, it is very sensitive to out-of-levelness. Also, the liquid is
discharged into the restricted gas flow area between troughs and at high liquid
rates this may result in premature flooding and maldistribution.


In spite of these problems, it may be the best practical choice for a fouling
service since the build-up of solids cannot totally block the notches as it can the
orifices in other types of distributor. A turndown ratio of 4:1 is generally
achievable and some manufacturers claim 10:1.
Orifice trough distributors are less sensitive to the problems of notched troughs
and are becoming the industry standard for large columns (above 1.5 m dia) with
high liquid loads. In the simplest designs the orifices are in the base of the
trough, but this gives only a limited turndown (< 3:1) and the orifices are prone to
blockage. Often the orifices are either in the sides of the trough or in the sides of
tubes fitted through the floor of the trough. This reduces the susceptibility to
blockage and also allows design for higher turndown by having extra distribution
holes at a higher elevation - see Figure 10. Orifices in the sides of the trough are
fitted with guide tubes to lead the liquid down to the packing and avoid
entrainment in the restricted space between troughs. Provision is sometimes
made for leveling each trough individually.
A major aspect of trough distributor design is the need for predistribution. If liquid
comes through one nozzle (e.g. reflux to the top distributor) it may be fed into a
single large trough (usually known as a parting box) which distributes the liquid
accordingly to the final distribution troughs through large orifices either in the
base or sides of the parting box. In many cases there is a second stage of
parting boxes.
The hydraulics of liquid flow in parting boxes and the final distribution troughs can
cause problems. High lateral velocities occur where the liquid is introduced, the
flow characteristics of orifices with a high transverse velocity upstream are not
well understood, and standing waves tend to develop at particular flow rates. It is
important therefore that trough type distributors should be tested before
installation and preferably at the manufacturers site.


FIGURE 10 ORIFICE TROUGH DISTRIBUTOR

- AT LOW RATES, ONLY THE BOTTOM ROW OF HOLES IS IN OPERATION
- AT HIGH RATES, BOTH ROWS ARE IN OPERATION

9.4.4 Testing of Liquid Distributors
Commercial distributor design technology has improved in recent years as
suppliers and end users have begun to appreciate its importance. However many
problem areas remain, such as the minimum acceptable liquid head, feed
velocity dissipation, and turbulence caused by liquid cascading from a parting
box into the troughs. There are also several mechanical factors which
can cause maldistribution, such as levelness and fit-up tolerances, drill or punch
reproducibility, and liquid tight seams. Therefore distributor testing is often
warranted to ensure proper operation. Detailed recommendations and procedure
are given in Ref [1], Vol 5, Section 2.02.1. Most manufacturers are able to
provide distributor tests.
A distributor should be tested when one or more of the following criteria apply:
(a)

Critical services where poor performance has a major impact on plant
profitability, safety, etc.. Services where the loss of fractionation will cause
a product to go off-spec.


(b)

Low liquid rate services (< 1.4 l/s m2) since uniformity of
liquid distribution is more difficult to achieve under these conditions.

(c)

Large diameter towers (> 2.5 m dia).

(d)

Extremely high liquid rate services (> 45 l/s m2).

The tests are done with water on an open test stand where the operation of the
distributor can be observed and samples taken of liquid rate at many locations
under the distributor. As such the test does not cover all possible problems. Cold
water has different physical properties to the liquids in most distillations. The
actual process fluids may for example exhibit greater froth or foam heights than
water. Nevertheless, the tests can be used to highlight potential operating
difficulties and take preventative measures before installation.
Specifically, tests can be used to:
(e)

permit visual observation of the distributor operation. Typical points of
interest include: overflow of troughs, excessive aeration, wave formation,
vertical discharge of liquid;

(f)

evaluate the turn-up and turn-down capabilities of the distributor;

(g)

expose leaks due to inadequate fit-up of internal flanges or poor
fabrication (welding),

(h)

check final assembly for dimensional tolerances; and

(j)

identify gross maldistribution caused by turbulence, excessive liquid
velocities, or errors in orifice sizing.

See Ref [1], Vol 5, Section 2.02.1. For details.

9.5

Packing Hold-down Grid

In the event of a packed column being run flooded or subjected to sudden vapor
surges, it is likely that the packed bed will be disturbed. This is particularly true
with random packings, but structured packings also can suffer. The purpose of a
hold-down grid is to hold the packing in place.


It is always recommended for random packings, but may be optional for
structured packings, except in vacuum operations, where large pressure surges
seem surprisingly common. Ideally the hold-down should be attached separately
to the column wall, but is often attached to the liquid distributor above. If the
problem is one of massive surges, this latter option could result in disturbance of
the distributor and loss of performance.

9.6

Reflux or Feed Pipe

For a detailed discussion, see Ref [1], Vol 5, Section 2.02.2.
Reflux may be introduced direct on to the top distributor, provided care is taken
to ensure that its inlet velocity is dissipated before it comes near any of the
distribution orifices. With a pan distributor, this may take the form of a ladder type
predistributor. With a trough distributor, a ladder type predistributor may be used
instead of a parting box.
Two phase feeds can be much more difficult to deal with. It is important to avoid
slug flow of liquid into the column, since slugs tend to be travelling at much
higher velocities than normal liquid flow and can cause severe damage to the
liquid collector and/or distributor. Ideally the phases should be separated outside
the column, with each phase introduced to the column through a separate
nozzle. Some manufacturers provide a "flash box" designed to separate the
phases in the column. Similarly, a vane separator can be installed at the column
entry. When the feeds are predominantly vapor or liquid, one of the devices
shown in Figure 11 are often used, the "raceway" is good for a high vapor
fraction (i.e. mist flow) and the baffle design is suitable for a low vapor fraction
(i.e. bubbly flow).
With two phase feeds a problem which needs to be considered is the possibility
of slug flow either in the distributor or in the pipework to it. If slug flow exists at
bends or orifices, equipment will be damaged due to severe vibration.


FIGURE 11 TWO PHASE FEEDS


9.7

Reboil Return Pipe

For a detailed discussion, see Ref [1], Vol 5, Section 2.02.2.
Most reboilers return a two phase mixture to the column. The flow regime will
generally be mist flow. If this is simply fed through the inlet branch, impingement
of the liquid on the column wall opposite will probably cause erosion and may
lead to failure. The most common design of vapor distributor is shown in Figure
12. This directs the vapor/liquid mixture downwards, the liquid falling to the
column base and the vapor turning round, losing much of its velocity and
flowing up to the bottom packed bed. Sometimes further provision for vapor
distribution is required. This can be assessed by comparing the stagnation
pressure rise in the incoming vapor (neglecting the liquid content) with the
pressure drop across the first meter of packing:

The provision for vapor distribution can then be determined according to the
value of R:
(a)

if R < 2.5 no vapor distribution is required; an open nozzle will suffice if
there is no risk of erosion on the opposite column wall;

(b)

if 2.5 < R < 4.5 the simple vapor distributor described above will suffice;


(c)

if R > 4.5 consideration should be given to a more sophisticated vapour
distributor.

For cases where R > 4.5, a chimney tray may be specified and designed on the
basis that its pressure drop should be at least 0.25 times the stagnation pressure
rise. The space between the tray and the packing support plate should be at
least 0.3 m. A conventional tray (e.g. sieve or valve) may be used, in which case
the space between the tray and the support plate will probably be greater but the
tray will achieve some mass transfer as well as vapor distribution.

9.8

Liquid Draw-offs

Liquid removal at an intermediate position in a column may be taken from a liquid
collector below a packed bed. The collector will be of the chimney tray type (see
Figure 5) and will be designed to have a working liquid level to ensure that the
draw-off is liquid as intended. The level can be maintained with a high weir over
which liquid flows to the distributor below.

9.9 Vapor Draw-offs
Vapor draw-off from an intermediate position in the column may be taken from
between beds. Consideration must be given to the possibility of liquid being
taken at the same time. Provision may be needed for removing this liquid and
returning it to the column. The vapor draw-off nozzle should be located below a
liquid collector to minimize the risk of liquid entrainment. Preferably the draw-off
nozzle should not finish flush with the inside surface of the column shell but
should protrude into the column, so that any liquid draining down the column
does not get into the draw-off.


FIGURE 12 VAPOR SPARGER


10

BIBLIOGRAPHY

[1]

FRI Fractionation Tray Design Handbook, Reports Centre, C & P, Wilton

Five volumes as follows:
Vol 1 - Sieve, bubble cap and dualflow trays
Vol 2 - Packings, proprietary tests, baffle trays, FRI report index
Vol 3 - Computer program listings
Vol 4 - FRI experimental data
Vol 5 - Design practices
Volumes 1, 2 & 5 are the most generally useful.
[2]

0'Connell; Trans AIChemE, Vol 42, p241, 1946.

[3]

Strigle R F; Random Packings and Packed Towers; Gulf Publishing
Company, 1987.

[4]

Sherwood T K; IEC, Vol 30, P 765, 1938.

[5]

GBH Enterprises; Absorption and Stripping Towers;


APPENDIX A
A.1

DEFINITIONS

INTRODUCTION

The following definitions are in the main those used by FRI and are understood
by most tray manufacturers; especially those who are members of FRI (see
Appendix E).
The definitions are given in two groups:
(a)

Those relating to the mechanical features of packed columns.

(b)

Those relating to the performance of the packing and internals.

A.2

MECHANICAL DEFINITIONS

A.2.1 Average Bulk Density
Weight divided by the volume of a packed bed. For random packing it is a
function of container or bed size, shape and loading method, in addition to
the size, shape and weight of the individual packing elements.
A.2.2 Pour Point Density
The number of irrigation points provided by a distributor divided by the
cross sectional area of the column.
A.2.3 Specific Surface Area
Packing surface area divided by the volume of a packed bed.
A.2.4 Void Fraction
Volume fraction available for vapor and liquid flow within a packed bed.


A.3

PERFORMANCE DEFINITIONS

A.3.1 Dynamic Pressure Drop
See Frictional Pressure Gradient (A.3.5).
A.3.2 F-Factor
A correlating parameter for hydraulic capacity and pressure drop. It is
similar to the capacity factor, but does not include liquid density and is
applicable primarily to vacuum conditions where pressure drop and
capacity are essentially independent of liquid rate. It is defined as:

A.3.3 Foam
A mass of bubbles stabilized by surface effects.

A.3.4 Flooding
Inoperability due excessive retention of liquid inside the column.

A.3.5 Frictional Pressure Gradient
That part of the pressure gradient in a packed bed which is due to the
frictional forces between the vapor flow, liquid flow, packing and wall of the
column. It is sometimes called the dynamic pressure gradient. The total
pressure gradient is the sum of the frictional and static pressure gradients.


A.3.6 Flow Parameter, X
A correlating parameter used in conjunction with Y (qv) for hydraulic
capacity and pressure drop. Essentially it is the square root of the ratio of
liquid to vapor kinetic energies, based on column superficial area:

where L and G are vapor and liquid mass rates.

A.3.7 Hydraulic Transition
Loading above which the vapor rate influences the liquid rate contribution
to pressure drop.

A.3.8 Lambda
A parameter in efficiency calculations, relating to the relative contributions
of vapor and liquid film resistances. It is defined as:

where G and L are vapor and liquid molar flow rates, and m is the slope of
the equilibrium curve. The slope may be obtained from commercially
available computer programs - the vapor and liquid compositions
on each stage are in equilibrium (by definition) and the gradient may be
estimated from the composition differences of adjacent stages:

where y and x are vapor and liquid mole fractions respectively.


A.3.9 Load Point
The combination of vapor and liquid loads above which there is a sharp
decline in mass transfer efficiency.

A.3.10 Packing Factor
A correlating parameter for hydraulic capacity and pressure drop,
characterizing the packing. It is applicable to random packings only, has
the units of reciprocal length and is most often quoted in 1/ft. Values for
most of the commonly used packings are given in Table 1 of Appendix B.

A.3.11 Radial Spreading Coefficient
The radial spreading of liquid in a packed bed has been shown to follow
the laws of diffusion. The following equation describes the spreading of
liquid from a point source:

A.3.12 Static Pressure Gradient
That part of the pressure gradient in a packed bed which is due to the
static head of the continuous phase, namely the vapor. The total pressure
gradient is the sum of the frictional and static pressure gradients.

A.3.13 System Limit (Ultimate Capacity)
A limiting combination of vapor and liquid loads which is a function of
system properties only. If exceeded, massive entrainment of liquid
droplets will occur. It can only be overcome by increasing the column
cross sectional area available for vapor flow.


A.3.14 Universal Packing Factor
See Packing factor.

A.3.15 Wall Flow
Liquid flowing on the wall in a packed column.

A.3.17 Wall Zone
In a packed bed, the peripheral annulus where the inner diameter is
tangent to the outermost pour points in the liquid distributor.

A.3.18 Wetting Rate
Liquid flow over a packing surface is conveniently expressed as flow rate
per unit periphery, known as the wetting rate. It is given by:

For absorption services, wetting rate so defined is regarded as an
important design parameter (see Ref [5]). Below a minimum value (which
depends on system and packing material) low efficiency is expected.
This concept does not apply to distillation.


A.3.19 Y - A Parameter Relating Capacity and Pressure Drop to Flow
Parameter
It is defined as:


APPENDIX B PACKING HYDRAULICS - THE NORTON METHOD
This appendix details the Norton generalized pressure drop as published in Ref
[3] and gives guidance on how to use it for distillation column sizing. This may be
used as a "hand" calculation.
FIGURE 13 GENERALISED PRESSURE DROP CORRELATION FOR
RANDOM DUMPED PACKINGS

The Norton correlation is expressed in graphical form, Figure 13, as:


Y, the ordinate is the gas velocity term, defined by:

X, the abscissa, is the flow parameter, defined by:

ΔP, the parameter in the graph, is the frictional pressure gradient in the packed
bed, in. H2O / ft. To this must be added the vapor phase static head to give the
total pressure gradient

The flood point is taken to be coincident with a frictional pressure gradient of 1.5
in. H2O/ft (12.3 mbar/m) of packing. In some cases, packings will operate at
higher pressure gradient, but mass transfer is almost certain to be poor.
The load point is defined as the combination of vapor and liquid loads at which
the mass transfer efficiency is a maximum and above which it declines sharply.
The load point may be taken to be 90% of the flood point. Design for 90% of the
load point is generally recommended for well behaved, systems.
Foaming systems rarely affect packed columns seriously, but a strong foaming
tendency can reduce capacity It is believed that the flood point declines towards
the load point in foaming systems and therefore design should be for less than
90% load. Only experience on the same or similar systems can give a firm
indication of a design value, but a value of 70% load should cover most cases.

DOCUMENTS REFERRED TO IN THIS PROCESS ENGINEERING GUIDE
This Process Engineering Guide makes reference to the following documents:
GBHE ENGINEERING GUIDES
GBH Enterprises

Glossary of Engineering Terms
(referred to in Clause 3)

GBH Enterprises

Use of Process Data Sheets for Distillation and
Absorption Columns (referred to in 7.3)

GBHE-PEG-MAS-610

Selection of Internals for Distribution Columns
(referred to in Clauses 1 and 4)


Design and Rating of Packed Distillation Columns

More Related Content

What's hot (20)

Similar to Design and Rating of Packed Distillation Columns (20)

More from Gerard B. Hawkins (20)

Recently uploaded (20)

Design and Rating of Packed Distillation Columns