You are on page 1of 42

HT203

EXPRESS USER GUIDE: PART 1. BACKGROUND TO USE OF EXPRESS IN DESIGN


AND OPERATION OF E-SHELL HEAT EXCHANGERS IN SINGLE PHASE DUTIES

CONTENTS
Page

1. NOTATION AND UNITS 1

2. INTRODUCTION 3
2.1 Purpose and Scope 3

3. SHELL SELECTION 5
3.1 TEMA E-Shells 6
3.2 Other Shell Types 7
3.3 Shell Selection Strategy 8

4. SIMPLIFICATIONS MADE FOR EXCHANGER MODEL 9

5. CONSIDERATIONS MADE IN THERMO-HYDRAULIC DESIGN OF HEAT


EXCHANGERS 10
5.1 Specification of Thermal Duty 10
5.2 Capital Cost Optimisation 11
5.3 Specification of Allowable Pressure Drops 12
5.3.1 Identification Of Optimum Pressure Drops 13
5.3.2 Optimisation Of Cooling Water Temperature Rise By Optimisation Of
Pressure Drop 13
5.4 Mechanical Integrity 13
5.4.1 Internal erosion/abrasion of tubes 13
5.4.2 External erosion/abrasion of tubes 14
5.4.3 Tube bundle vibration 14
5.4.4 Stream leakage 16
5.5 Mechanical Design Requirements for Thermo-hydraulic Design 16
5.5.1 Mechanical clearances 17
5.5.2 End zones 18
5.5.2.1 End zones for segmental baffles 18
5.5.2.2 End zones for helical baffles 18
5.6 Operability 19
5.7 Maintenance 19
5.8 Longevity 20

1
HT203
6. PRELIMINARY DESIGN DECISIONS 21
6.1 Hazard Assessment 21
6.2 Selection of Headers 21
6.2.1 Fixed tube sheet arrangements 22
6.2.2 U-tubes 22
6.2.3 Floating head arrangements 22
6.3 Fluid Allocation 23
6.4 Tubes 25
6.4.1 Tube configuration 25
6.4.2 Tube size 26
6.5 Number of Tube Passes 26
6.6 Bundle Layout 26
6.7 Baffle Type 28
6.7.1 Baffle arrangement 28
6.7.2 Baffle geometry - segmental baffles 30
6.7.3 Baffle geometry - helical baffles 30
6.7.4 Baffle geometry - ROD baffles 31
6.7.5 Other baffle types 31
6.8 Shell in Series Criterion 31

7. REFERENCES 33

APPENDIX A BACKGROUND TO METHODS USED FOR PRELIMINARY SIZING

A1. MODEL OF TWO-STREAM HEAT EXCHANGER 35

A2. TUBESIDE HEAT TRANSFER 37

A3. TUBESIDE PRESSURE DROP 38

A4. SHELLSIDE HEAT TRANSFER AND PRESSURE DROP 38


A4.1 Segmental baffles 38
A4.2 Helical Baffles 39
A4.3 ROD Baffles 39

A5. EFFECTIVENESS-NTU EQUATIONS 40

2
HT203
EXPRESS USER GUIDE: PART 1. BACKGROUND TO USE OF EXPRESS IN DESIGN
AND OPERATION OF E-SHELL HEAT EXCHANGERS IN SINGLE PHASE DUTIES

1. NOTATION AND UNITS

Units

SI British

A heat transfer area of tube bundle external surfaces, m2 ft2


NtπDextL
C cost – –
C* thermal capacity ratio, defined by Equation (A1.8) – –

cp specific heat capacity at constant pressure J/kg K Btu/lb°F

D tube internal diameter m ft


Dext external diameter of tube bundle m ft

Ds internal diameter of shell m ft

E heat transfer effectiveness, defined by Equation – –


(A1.9)
Eact activation energy of crude oil – –

f Fanning friction factor _ _

Fi annualisation factor – –

FT Log-Mean Temperature Difference (LMTD) _ _


correction factor
H hours of operation p.a. h h
h length of flange neck (see Equation (5.5)) (mm) –
i interest rate _ _
kd _ _
deposition constant

kr removal constant in Ebert-Panchal equation – –

L tube length m ft

Lb baffle pitch m ft

· mass flow rate kg/s lb/h


M

1
HT203
Nu Nusselt number, α D ⁄ λ _ _

Nb number of baffles _ _

Np number of tubeside passes _ _

Nt number of tubes in tube bundle _ _

Pt tube pitch m ft

Pr Prandtl number, c p η ⁄ λ _ _

rf fouling resistance m2 K/W ft2 hr °F/Btu

∆p pressure drop N/m2 (Pa) lbf/ft2


· rate of heat transfer between streams (heat load) W Btu/h
Q

T temperature K °R ( ≈ °F + 460 )

∆T m Effective Mean Temperature Difference (EMTD) K or °C °R or °F

∆Tlm Log-Mean Temperature Difference (LMTD) K or °C °R or °F

U overall heat transfer coefficient, referred to bare W/m2 K Btu/ft2h °F


external surface area of tube bank

νa average tubeside flow velocity m/s ft/s

νs shellside crossflow velocity (see Section ?) m/s ft/s

x foulant layer thickness m ft

α film heat transfer coefficient W/m2 K Btu/ft2h °F

η dynamic viscosity kg/m s or lb/ft s


N s/m2

η eff pump efficiency (default: 80%) _ _

λ thermal conductivity W/m K Btu/h ft °F

Π power consumption W Btu

ρ fluid density kg/m3 lb/ft3

Subscripts

allowable allowable value

cap capital expenditure

in inlet conditions

2
HT203
lam laminar

max maximum value

nozzles across shell or tube nozzles

op annual operating expenditure

out outlet conditions

s shellside

t tubeside

turb turbulent

unit unit cost

1,2 streams 1 and 2 respectively

2. INTRODUCTION

2.1 Purpose and Scope

This Data Item is Part 1 of a group of User Guides to the ESDU EXPRESS computer program, and provides
background information on the use of the program in the design, rating, revamping and operation of
shell-and-tube heat exchangers.

The various parts of the “EXPRESS User Guide” group are:

Part 1. Background to use of EXPRESS in design and operation of E-shell heat exchangers in single phase
duties (HT203).

Part 2. Rapid design and performance prediction (HT204).

Part 3. Rating and performance prediction (HT205).

Part 4. Revamping for revised duties (HT206).

Part 5. Designing and operating for reduced fouling in crude oil pre-heat exchangers (HT207).

Part 6. Consideration and evaluation of exchanger operability (HT208).

Part 7. Handling uncertainty in design (HT209).

The purpose of the EXPRESS program and these User Guides is to assist the process engineer, heat
exchanger designer or plant operator in the selection, design and use of shell-and-tube exchangers of E-shell
configuration with single-phase flows on both the shellside and tubeside.

EXPRESS is based on the use of the methods of this and other ESDU Data Items. It generates apprimatate
design, costing and performance information which is presented using simple and powerful graphical

3
HT203
techniques and design tables. It provides the engineer with a means of:

(1) generating, in “design” mode, preliminary designs and budget cost estimates of shell-and-tube
exchangers for project appraisals,

(2) identifying, in “design” mode, suitable design alternatives,

(3) incorporation of considerations of exchanger operability into design,

(4) obtaining, in “rating mode”, quick design checks on whether or not existing equipment is suitable
for use on specified duties (existing or new),

(5) rapidly checking, in “rating” mode, designs generated by shell-and-tube heat exchanger design
programs,

(6) evaluating, in “revamping” mode, design changes that may be made to revamp an existing
exchanger for revised duties,

(7) evaluating, in “design”, “rating” and revamping” modes, the response of exchangers to changes in
fouling resistance, inlet temperatures and throughput,

(8) evaluating in “design”, “rating” and revamping” modes, the consequences of over-design,

For exchangers that are to be used on certain fouling duties where in-tube fouling in dependent on flow
velocity (for example, biofouling), tube wall termperature (for example, waxing or coking) or both
(exemplified by chemical reaction fouling in the ‘hot end’ heat exchangers in crude oil pre-heat trains - see
Part 5 and ESDU 00016), EXPRESS allows the following:

• determination of fouling threshold conditions and the development of no-fouling designs,

• evaluation of the fouling response of the exchanger to changes in fouling resistance, inlet
temperatures, throughput and design margin.

EXPRESS is complementary to commercially-available exchanger design computer programs (such as


those provided by Heat Transfer Research, Inc. (HTRI) and HTFS. For example, EXPRESS can help the
user of those programs to use them more efficiently. Another prime use of EXPRESS is in training of
inexperienced engineers within organisations that do not have resident in-house expertise.

The commercial computer programs currently used in exchanger design are excellent exchanger analysis
tools. They contain sophisticated thermal analyses and include important features such as bundle vibration
analyses and mechanical design capability. However, their thermal design algorithms demand a ‘starting
point’ for design that is often based on the experience of the engineer. As a consequences many engineers
use these programs in a ‘rating’ or ‘simulation’ mode when undertaking design. In this mode, an initial
geometry is selected, analysed and then systematically ‘improved’ to the final design. The procedures
outlined in this document and incorporated in EXPRESS provide a good starting point for such an evolution.

4
HT203
3. SHELL SELECTION

Shell-and-tube heat exchangers are almost universally designed according to the Standards of the Tubular
Exchanger Manufacturers Association (TEMA)17, and the configurations laid down by this body are
illustrated in Sketch 3.1. Configurations are indicated in terms of the front end type (A-D), the shell type
(E-K and X) and the rear end type (L-W) as illustrated. Thus, a BES shell has a bonnet (integral cover)
front end, a one-pass E-shell and a floating head with backing device at the rear end.

Selection of the shell type is largely a function of the thermal-hydraulic design. However, an E-shell is
usually chosen unless there are good reasons to do otherwise.

Sketch 3.1 Configuration Nomenclature Adopted By The Tubular Exchangers Manufacturers


Association (TEMA)17. ©1988 By Tubular Manufacturers Association

5
HT203
3.1 TEMA E-Shells

Of the various shell types available the simplest is the E-shell, and E-shells with various front and rear
header arrangements are the most common types of shell-and-tube heat exchanger.

The E-shell configuration with the best thermo-hydraulic performance is a single tube pass unit with good
shellside flow characteristics: this is achieved using an efficient arrangement of conventional segmental
baffles (see Section 6.7.2), helical baffles (Section 6.7.3) or twisted tube bundles. In such a configuration,
the shellside flow can be considered close to pure counter-current. This makes best use of available
temperature driving force and leads to the smallest exchanger size. If the design has more than one tube
pass a correction has to be made to the Log Mean Temperature Difference (LMTD) to compensate for the
flow not being truly counter-current. In some cases, more than one shell in series must be used in order to
obtain a workable design.

Multiple shells in parallel are also required where the number of tubes in a design exceed the maximum
number possible in a shell of given diameter.

As described in ESDU 9494211, most E-shells have one of the following configurations (see Sketch 3.1 for
illustrations):

BEM
This is termed the “fixed tube sheet” type and is commonly used for its simplicity and low relative cost.
The fixed tube sheet type has the advantage of no gasketed internal joints. Disadvantages are that whilst
the inside of tubes are accessible for mechanical cleaning the outside surface is not, and that expansion
bellows often need to be fitted into the shell to absorb differential expansion between tube and shell.

BEU
The “U-tube” is also simple and economical. Again, there are no gasketed internal joints. The bundle is
removable so the outside of tubes can be mechanically cleaned, but mechanical cleaning is difficult for the
inside of the tubes due to the U-bend. The presence of U-tubes eliminates the requirement for bellows but
also leads to a reduced tube count for the same shell size and makes it difficult to replace tubes. As “U-tubes”
are at least two-pass, and a BEU design cannot have pure counterflow, there is reduced effectiveness (that
is, reduced effective mean temperature difference). An alternative option is to use a longitudinal baffle
(BFU) which allows countercurrent flow and is relatively economic, though care has to be taken to minimise
leakage at the baffle edge. If the baffle is welded to the shell to achieve this, the bundle is not removable.
Alternatively a shroud cylinder may be used between the tubes and the shell.

AES
Despite its relatively high cost, the “Floating Head” type is the preferred exchanger for many petrochemical
and process users. The tube bundle is fully removable and maintainable, though maintenance time is longer
than for other types. Tubes can be mechanically cleaned internally and externally. The “Floating Head”
design allows fewer tubes than a fixed tube sheet of the same diameter, but in this case the difference is
minimised by the use of a split backing ring to attach the floating cover to the tube sheet. The design is
robust and versatile but only allows pure counterflow with the additional complexity of an internal bellows.
However, the use of the AES configuration does introduce the need for a gasketed internal joint.

BES
The use of a bonnet rather than a channel and removable cover plate is cheaper in cost but, for large sizes,
this is offset by the higher maintenance costs and the A-type closure is usually chosen in these cases.

Other header types may be required for certain process or maintenance reasons. Removable covers (A or
L type heads) may be desired for access to tubes without disturbing pipework. Integral tube sheets (C or N

6
HT203
type) would be considered for higher pressure or high temperatures where they are cheaper than removable
covers, and where gasketed joints need to be minimized. The T type of floating head is similar to the S type
but has a more robust tube sheet to floating cover joint at the expense of greater bypass area around the
bundles (and, thus larger shells) and, therefore, is more expensive. However, with the T type, the bundle
may be removed without dismantling the floating head assembly, unlike the S type, thus reducing
maintenance time. The W and P type floating heads utilise packed glands and are therefore limited on
application by pressure and temperature. They are suitable for applications where leakage of the gasket
joint is to atmosphere. The P type is considerably more expensive than the W type. The NEN configuration
is sometimes found to be an economic option.

The various configurations may be listed in order of increasing cost (left to right) as follows:

Front end: B, N, C, A
Rear end: U, M, N, L, W*, S, T, P* (* Limited application)

(For high pressure either a U-tube or a special type of D header would be used.)

3.2 Other Shell Types

There are some duties where for thermo-hydraulic reasons an E-shell is either unsuitable or not the best
choice. A key design limitation is the shellside pressure drop, and the pressure drop through any shell can
be considered proportional to the square of the average fluid velocity on the shellside and to the number of
tube rows crossed. The following observations can be made for the most common alternative shell types.

F-shells

The F type has a longitudinal baffle, which allows for an approach to pure counter-current flow with two
tube passes. Given the longitudinal baffle, for the same shell size and baffle count, the shellside velocity
in the F-shell is twice that in the E-shell. The F shell pressure drop is therefore four times greater than that
of an equivalent E-shell.

The costs of an F-shell depend upon the requirements for the baffle seal which varies from simple flexible
strips, to a central baffle welded to the shell, to a completely shrouded section. Potential leakage through
the gap between the longitudinal baffle and the shell may restrict the range of application.

J-shells

The J type, or “divided flow” type, has one inlet and two outlet nozzles (thus dividing the flow in half) or
vice versa. This results in a pressure drop of approximately one eighth that of the E type, and the main use
of the J type is therefore for low pressure applications such as gas coolers and condensers.

X-shells

The X type shell has the shellside flow in pure crossflow without cross baffles. The result is an extremely
low pressure drop and, again, it is used for gases and condensing vapour at low pressures.

G-shells

The G type are often used for vaporising duties where low pressure drop is required, (for example,
thermosyphons). A G-shell has half the pressure drop of an equivalent E-shell.

7
HT203
H-shells

The H type is also often used for vaporising duties where low pressure drop is required, and has one sixteenth
the pressure drop of an equivalent E-shell.

K-shells

A K type shell would be chosen for many vaporisers.

3.3 Shell Selection Strategy

The observations above suggest the following shell selection strategy:

(1) Initially select an E-shell

(2) Move to an F-shell if multiple tube passes are required and such a move removes the need for
multiple shells in series.

(3) If difficulties are encountered in meeting the shellside pressure drop constraint move to a J-shell
(or, subsequently and progressively, to a G or H shell).

8
HT203
4. SIMPLIFICATIONS MADE FOR EXCHANGER MODEL

The simplifying assumption made for the purposes of modelling E-shell heat exchangers by EXPRESS is
that the physical properties of the fluids handled by the exchanger do not vary over the length of the
exchanger. This assumption reduces the amount of data the engineer requires in order to generate a result
and significantly reduces the amount of computation involved. The assumption has a number of important
implications.

Firstly, it means that the film heat transfer coefficients are constant over the length of the unit. This has the
advantage that the design may be based upon the use of a Log Mean Temperature Difference (LMTD)
corrected for any multiple tube pass effects (see Section A1). However, there are situations in which this
assumption can lead to significant errors. The physical property most affected by temperature changes is
viscosity. This property appears in both the Reynolds Number (where it appears in the denominator) and
the Prandtl Number (where it appears in the numerator). The film heat transfer coefficient for the tubeside
of an exchanger (where the fluid is flowing under turbulent conditions) is dependent upon the Reynolds
Number raised to a power of 0.8 and the Prandtl Number raised to a power of one third. Thus, the overall
dependency of the film coefficient to the viscosity involves a power of around one half. So, a twofold
change in viscosity would result in a 40% change in film heat transfer coefficient.

Situations in which changes in viscosity result in changes in flow regime across the length of the unit (e.g.
a change from turbulent flow to transitional or laminar flow as a high viscosity liquid is cooled) present
much higher errors.

For the fluid flowing through the exchanger shell the dependency of film coefficient on viscosity will
involve a power of around one third.

Secondly, it means that the film heat transfer coefficient is independent of temperature difference. This has
major advantages with regard to the amount of computation required. However, the designer should be
aware that the film heat transfer coefficient is affected by differences between local wall and local bulk
temperatures. Typically the coefficient varies with the ratio of bulk to wall viscosity raised to a power of
around 0.14. Thus, a twofold difference between the viscosity in the bulk and at the wall would result in
an error in the prediction of local film heat transfer coefficient of +-10%.

The use of an Effective Mean Temperature Difference (EMTD) approach requires the overall heat transfer
coefficient to be uniform across the unit. Errors are introduced when the approach is used in situations in
which the overall coefficient varies along the length of the exchanger or between tube passes. This is also
the case for the standard Effectiveness-NTU equations (see ESDU 98003 - 98007)14.

9
HT203
5. CONSIDERATIONS MADE IN THERMO-HYDRAULIC DESIGN OF HEAT EXCHANGERS

There are usually two groups of engineers associated with the design of heat exchangers: those responsible
for the thermo-hydraulic design and those undertaking the mechanical design.

At the preliminary design stage, the engineer responsible must first ensure that any physical limitations are
taken in to account. These limitations include:

• Exchanger length. It is common for shell-and-tube heat exchangers in refineries to have a standard
length, such as 6 m or 20 ft. The engineer must also be aware to the necessity to ensure adequate
additional space for the removal of tube bundles for cleaning (for example).

• Exchanger weight. On some applications, particularly offshore, cost factors may drive the designer
to the most compact design possible.

Further guidance on selection and sizing is given in ESDU 9201312 and ESDU 9404211.

Full thermo-hydraulic design covers considerations that extend beyond equipment sizing. These include:

(i) assistance in specification of thermal duty (Section 5.1),

(ii) capital cost (Section 5.2),

(iii) assistance in specification of allowable pressure drops (Section 5.3),

(iv) mechanical integrity (Section 5.4),

(v) mechanical design requirements for thremo-hydraulic design (Section 5.5),

(vi) operability (Section 5.6),

(vii) maintenance (Section 5.7),

(viii) longevity (Section 5.8).

5.1 Specification of Thermal Duty

In many situations the thermal duty for the exchanger can be set without making any consideration of its
effect upon exchanger size and cost. However, there are situations in which the thermal duty can be changed
if a significant cost benefit would result. Such a situation often arises in heat recovery problems, where
close temperature approaches can give rise to a need for complex and expensive exchanger arrangements.
Similarly, high heat loads can also give rise to a need for larger exchanger shells or multiple smaller shells.
In each of these circumstances small changes in duty can give rise to large capital cost savings.

The EXPRESS program can be used to evaluate rapidly any possible duty changes that can be made in
order trade off exchanger complexity against exchanger performance.

Another situation that is of interest to the process engineer is the specification of cooling water temperature
rise. For a given heat transfer requirement, the lower the cooling water temperature rise the higher is the
temperature driving force between the streams. This could result in a need for less heat transfer surface.
However, to lower the cooling water temperature rise, the cooling water flow rate must be increased. This
higher flow rate can give rise to increased tube count and, hence, larger shell diameter. The cost of a

10
HT203
shell-and-tube exchanger is more sensitive to diameter than surface area. Thus, less surface area does not
necessarily mean cheaper exchangers. The specification of cooling water temperature rise is therefore an
optimisation problem (see Section 5.3.2).

5.2 Capital Cost Optimisation

The cost of a heat exchanger is controlled by five main factors:

(i) number of shells-in-parallel,

(ii) number of shells-in-series,

(iii) shell diameter,

(iv) bundle length,

(v) material selection.

In general, the fewer the number of shells used the lower the capital cost. Methods for the estimation of
budget cost of shell-and-tube heat exchanges are given in ESDU 9404211.

The capital cost of an exchanger is much more dependent upon shell diameter than bundle length. Hence,
for a given duty a long thin unit is generally cheaper than a short fat unit.

Long thin units often have operational benefits as the allowable tubeside pressure drop is increased as the
number of tube passes can be kept to a minimum. The engineer should seek to maximise the use of pressure
drop on both sides of the exchanger. This is not always possible. For example, the greater the pressure drop
the greater is the power consumption (and, hence, the operating cost).

The EXPRESS program shows the relationship between shell diameter (expressed by the tube count) and
bundle length for designs delivering the required duty. The user can also obtain an estimate of the capital
cost of a chosen geometry.

The capital cost of a heat exchanger does not relate to heat transfer surface area in a simple manner. Capital
costs tend to be far more sensitive to shell diameter than tube length. So, exchangers having the same surface
area can differ markedly in capital cost.

A methodology for the estimation of the capital cost of shell-and-tube heat exchangers has been described
by Saunders (1988)6. ESDU 9201312 and associated Data Items provide an alternative costing method that
is most suitable in the selection of heat exchangers by cost comparisons between equivalent generic types
(for example, comparing for a given duty the budget cost of a shell-and-tube unit and a plate-and-frame
unit). The methodology of Saunders is most appropriate for the purposes of EXPRESS and has been
incorporated into the program. This method yields a prediction of purchase price of the unit. This should
be multiplied by an ‘installation factor’ in order to obtain an estimate of the installed cost of the unit.

The program also allows the determination of an annualised capital cost. Summation of this term with the
operating costs provides a total annual cost (T.A.C.). This can be used in the optimisation of exchanger
capital cost.

Capital costs and operating costs have differing bases. The capital cost is an expenditure made at a fixed
point in time but operating costs are spread across the life time of the equipment. One way around this
situation is to relate the capital cost to a ‘time’ value of the investment. This can be done by multiplying

11
HT203
the investment cost by an ‘annualisation factor’.

A convenient form of annualisation factor is:

i( 1 + i )n
F i = ----------------------------- (5.1)
( 1 + i )n – 1

where i is the interest rate expressed as a fraction and n is the plant life (expressed in years).

The ‘total annual cost’ of the exchanger is given by:

C T.A.C. = F i C cap + C op (5.2)

where Ccap is the capital cost and Cop is its annual operating cost. The optimum pressure drops (Section
5.3) for the two fluids are those which result in the minimisation of this cost.

The following calculation can be applied to both fluids handled by the exchanger in order to obtain an
estimate of the power consumption, Π , involved in the operation of the exchanger. For flow of an
incompressible fluid, power consumption is a function of mass flow rate, M· , fluid density, ρ, pressure drop,
∆p , and pump efficiency η eff (default: 0.8 or 80%):

M· ∆p
Π = ------------ . (5.3)
ρηeff

This calculation can also be made as an approximation for compressible gases when the pressure drop
through the exchanger is a minor part of the overall pressure loss in a system (e.g. the case of a compressor
inter-cooler).

The overall operating cost is then a function of the number of hours the exchanger is in operation each year,
H, and the unit cost of power, C unit :

C op = ( ∑ Π )HCunit . (5.4)

For any potential design the EXPRESS program provides the user with computation of capital cost, power
costs and total annual cost for specified economic data (interest rates etc.). The engineer can systematically
vary the absorbed pressure drop and quickly identify the optimum value.

5.3 Specification of Allowable Pressure Drops

In some situations the allowable pressure drop of a stream is fixed by factors outside the control of the
engineer: one such example is where it is set by fluid drainage across a fixed physical height and hence
available head. However, in many cases allowable pressure drop can be easily changed (for example, by
changing a pump specification). It is not uncommon to find that the specification of allowable pressure
drop is made quite arbitrarily.

The EXPRESS program can be used to explore rapidly the relationship between the allowable pressure
drops on the tubeside and shellside and exchanger geometry. It therefore allows the designer to consider
exchanger pressure drops in an objective manner, and identify optimum pressure drop (Section 5.3.1).
Another opportunity afforded by the use of EXPRESS is the optimisation of cooling water temperature rise
by the optimisation of pressure drop (Section 5.3.2).

12
HT203
5.3.1 Identification Of Optimum Pressure Drops

Provided pressure drop is absorbed in an efficient manner (for example, for shellside flows an efficient
baffle arrangement is used), higher absorbed pressure drop gives higher stream film coefficients and lower
required heat transfer surface area.

Unfortunately, increased pressure drop often means increased power consumption in the prime movers
(pumps or compressors) driving the fluids through the exchanger. Thus, the reduction in exchanger area is
bought at the price of increased operating costs. The optimisation of pressure drops therefore should involve
a trade-off between exchanger capital cost and operating cost.

5.3.2 Optimisation Of Cooling Water Temperature Rise By Optimisation Of Pressure Drop

The problem of fixing cooling water temperature rise is similar in many ways to that of setting pressure
drop. An increase in cooling water flow rate is accompanied by a reduction in the temperature rise and an
increase in effective mean temperature difference. This, in turn, leads to a reduction in the amount of heat
exchanger area required. However, there is an economic penalty to be paid that offsets this reduction in
exchanger capital cost, which is increased power consumption (assuming fixed pressure drop).

The optimisation of cooling water temperature rise should therefore include optimisation of water side
pressure drop.

It is suggested that the following procedure is used:

(1) optimise pressure drops at the initial water usage.

(2) optimise water consumption at the resultant pressure drops.

(3) re-visit pressure drop optimisation to confirm result.

5.4 Mechanical Integrity

Four aspects of design effecting the mechanical integrity of an exchanger fall into the remit of the engineer
undertaking the thermo-hydraulic design. These are:

(i) the internal erosion/abrasion of the tubes (Section 5.4.1),

(ii) the external erosion/abrasion of the tubes (Section 5.4.2),

(iii) potential damage to tubes through tube bundle vibration (Section 5.4.3),

(iv) potential leakage between tubeside and shellside fluids (Section 5.4.4).

5.4.1 Internal erosion/abrasion of tubes

The possibilities of internal erosion/abrasion are minimised by ensuring that the tubeside velocity is kept
below a pre-specified maximum value. This value is a function of whether the fluid handled is a liquid, a
gas, or gas/liquid mixture. It also depends strongly on the tube material. Recommendations on maximum
velocities for water flowing through tubes are given in Table 5.1. (Reference ?).

The specification of maximum tubeside flow velocity affects the ‘design space’ displayed by the EXPRESS

13
HT203
program, as described in Part 2 of the User Guide (HT 204).

TABLE 5.1 Recommended Maximum Tube-side Velocities for Water

Tube material Recommended maximum


tubeside velocity in m/s (ft/s)
Carbon Steel 1.8 (6.0)
Copper-Nickel (Cu-Ni) 3.0 (10.0)
Stainless steel 3.0 (10.0)
Titanium 3.6 (12.0)
Admiralty brass 2.4 (8.0)

5.4.2 External erosion/abrasion of tubes

Fluid velocities on the shellside through the tube bundle are generally significantly lower than those
encountered within the tubes. However, external erosion/abrasion of the tubes can occur immediately below
the inlet nozzle and impingement plates are used for protection.

TEMA provides guidance on the conditions for installation of impingment plates. The controlling variable
in their guidelines is the momentum of the fluid entering the shell through the inlet nozzle. Impingement
protection is required if the product of the fluid density, ρ ,and the square of the fluid velocity, v, exceeds
the value listed in Table 5.3.

TABLE 5.2 Impingement Protection Criteria

Fluid type ρ v2 in kg/m (lb/ft)

Non-corrosive, non-abrasive liquid 2230 (1500)


All other liquids or liquid close to boiling point 744 (500)
Gases, vapours and gas/liquid mixtures 0

5.4.3 Tube bundle vibration

Detailed guidance on tube bundle vibration is given in ESDU 870195. The combination of high crossflow
(shellside) velocities with large, unsupported tube spans may initiate damaging tube vibrations. The
likelihood of damaging tube vibrations occurring must be considered at the design stage, together with the
constraints of heat transfer and allowable pressure drop, and also when process conditions are changed.

Mechanical failure due to vibration may occur by one of several mechanisms:

(a) baffle damage where the walls of the vibrating tubes are worn by the baffle or tube support,

(b) tube collision damage where the amplitude of vibration is such that adjacent tubes impact with one
another,

(c) tube joint failure, at the tubesheet or header, on account of either vibration opening up expanded

14
HT203
joints or the edge of the tube holes wearing the tubes,

(d) fatigue, possibly enhanced by corrosion and/or erosion.

ESDU 870195 should be consulted for more guidance on the possible mechanisms which initiate flow
induced vibration in tube bundles. ESDU 870195 also presents the available data and formulates procedures
for predicting the likelihood or onset of tube vibrations. An indication of whether vibrations are liable to
cause damage is given and steps that may be taken at the design stage to reduce the likelihood of vibration
problems are described.

In general, the possibility of damage due to tube bundle vibration is generally reduced provided the
following steps are taken.

(i) Unsupported tube lengths are kept below standard maximum allowable values.

(ii) An analysis of the susceptibility of the bundle to tube vibration is conducted, for example using
ESDU 870195.

Different tubes, or groups of tubes, in a bundle may have different natural frequencies of vibration. These
natural vibration frequencies are dependent on the ‘unsupported tube lengths’ (the maximum distance
between any two supports along the entire length of all tubes in the bank), which are in turn dependent on
the baffle spacing and baffle cut.

Within a typical bundle in a segmentally-baffled shell-and-tube heat exchanger, between the end baffles
the unsupported tube length is taken as twice the baffle spacing. At the ends of the bundle the unsupported
tube length is equal to the end zone length plus a single baffle space.

This is illustrated in Sketch 5.1 which shows a typical shell-and-tube heat exchanger segmental baffle
arrangement.

B
1m 0•5 0•5 0•5 0•5 0•5 0•75 m
C

Sketch 5.1 Unsupported tube lengths

The unsupported tube lengths are:


Span, m (ft) Left-hand end Mid-bundle Right-hand end
Group A tubes 1 (3.28) 1 (3.28) 1.25 (4.10)
Group B tubes 1 (3.28) 0.5 (1.64) 0.75 (2.46)
Group C tubes 1.5 (4.92) 1 (3.28) 0.75 (2.46)

15
HT203
Group C tubes will have the lowest natural frequencies and will probably be the most prone to suffer damage.
However, there are situations where group A tubes with their somewhat higher natural frequency may be
most liable to damage. Both possibilities should normally be investigated.

Group B tubes are those supported at every baffle. Natural frequencies are much higher than for other cases
and vibration damage is relatively unlikely.

TEMA17 have specified maximum values for unsupported tube lengths in tubes of various materials, as
shown in Table 5.3.

TABLE 5.3 Recommended Maximum Unsupported Tube Lengths (TEMA17)

All steels Aluminium and its alloys


Copper and its alloys
Titanium
Tube outside diameter, mm (in) Maximum unsupported tube lengths, mm (in)
19.05 (0.75) 1524 (60) 1321 (52)
25.4 (1.0) 1880 (74) 1626 (64)
31.75 (1.25) 2235 (88) 1930 (76)
38.1 (1.5) 2540 (100) 2210 (87)
50.8 (2.0) 3175 (125) 2794 (110)

The EXPRESS program does not undertake detailed tube bundle vibration analysis. The engineer needs to
use exchanger analysis software to perform this task. The EXPRESS program does however incorporate a
check on supported lengths. If this is exceeded the engineer should either reduce baffle cut (thereby reducing
baffle span, see Section 6.7) or increase tube diameter or tube wall thickness. A full check on tube bundle
vibration should always be undertaken at the detailed design stage.

5.4.4 Stream leakage

All heat exchanger users should be aware that even though an exchanger may successfully pass a hydrostatic
test, there can still be some leakage between the two sides of an exchanger. If mixing of streams is
unacceptable a double tubesheet should be used. Such a design provides a physically drained space
positioned between the major fluid containment provided by each tubesheet. This space not only provides
physical separation of the streams but also allows the user to observe the presence of fluid leakage. It is
important that this space is allowed to drain freely and that plant operators look regularly for signs of
leakage.

5.5 Mechanical Design Requirements for Thermo-hydraulic Design

Full mechanical design is not undertaken as part of the thermo-hydraulic design. However, the thermal
designer needs to be aware of the implications that the mechanical design will have upon exchanger
geometry. Mechanical design influences exchanger geometry in two respects:

(i) mechanical clearances (Section 5.5.1),

(ii) end zone lengths (Section 5.5.2).

16
HT203
5.5.1 Mechanical clearances

TEMA17 make recommendations for the various clearances that are needed for exchanger fabrication
(tube/baffle clearance, baffle/shell clearance, bundle/shell clearance and pass partition clearance). The
following values are incorporated in the EXPRESS program.

TABLE 5.4 Bundle-to-shell (diametrical) clearances17

Shell internal diameter, TEMA-recommended clearances, mm (in)


mm
Fixed tubesheet or Split backing ring Pull-through floating
U-tube floating head head
> 736 (29) 13 (1/2) 35 (1 3/8) 102 (4)
735 - 990 (29 - 39) 13 (1/2) 41 (1 5/8) 102 (4)
990 - 1218 (39 - 48) 13 (1/2) 43 (1 11/16) 114 (4 1/2)
1218 - 1523 (48 - 60) 16 (5/8) 48 (1 7/8) 124 (4 7/8)
1523 - 1828 (60 - 72) 19 (2/3) 51 (2) 137 (5 3/8)
>1828 (72) 22 (7/8) 51 (2) 137 (5 3/8)

TABLE 5.5 Baffle-to-shell (diametrical) clearances17

Shell internal diameter, TEMA-recommended clearance


mm (in) (diametrical), mm (in)
< 356 (14) 2.54 (1/10)
356 - 458 (14 - 18) 3.18 (1/8)
458 - 610 (18 - 24) 3.81 (0.15)
610 - 1016 (24 - 40) 4.44 (0.175)
1016 - 1397 (40 - 55) 5.72 (0.225)
>1397 (55) 7.62 (1/3)

17
HT203

TABLE 5.6 Tube-to-baffle clearances17

Baffle spacing, mm TEMA-recommended


clearance (diametrical), mm
< 458 (18) 0.794 (1/32)
> 458 (18) 0.397 (1/64)

5.5.2 End zones

The EXPRESS program contains routines that ensure that the exchanger end zones are consistent with the
needs of the mechanical design. The following methodology is used.

5.5.2.1 End zones for segmental baffles

The end zone for an exchanger using segmental baffles extends to the first baffle plate, which must be
positioned beyond the shellside nozzle. It is therefore set by the following factors:

(1) the length of the neck on the main shell flange,

(2) the clearance between the neck of the flange and the nozzle piping,

(3) the diameter of the nozzle itself,

(4) the clearance between the edge of the nozzle and the first baffle.

The additional factor that needs to be applied is the clearance between the nozzle edge and the first baffle
plate. This is assumed to be 25 mm.

5.5.2.2 End zones for helical baffles

The end zone for an exchanger using helical baffles extends to the centre line of the shellside nozzle. It is
therefore set by the following factors:

(1) the length of the neck on the main shell flange,

(2) the clearance between the neck of the flange and the nozzle piping,

(3) the diameter of the nozzle itself.

The length of the flange neck is dependent upon shell diameter and design pressure. A table of these lengths
is provided in BS4504 Part1: 19691.

The following equation was found to correlate the values given in this table and incorporated in EXPRESS:

h = 46.3 + 0.59p + ( 0.011 + 0.0038p )D , (5.5)

where h is the length of the neck (mm), p is the design pressure (bar) and D is the shell diameter (mm).

18
HT203
A nominal value can be set for the clearance between the flange neck and the nozzle pipe: a value of 75 mm
appears reasonable.

External pipe (nozzle) dimensions are also given in BS45041. These data are incorporated into the
EXPRESS program. Above 500 mm it can be assumed that the external diameter equals the nominal value
plus 20 mm.

5.6 Operability

Exchangers are usually designed to operate at a point condition. However, once installed, they are required
to operate over a range of conditions as plant throughput varies and cooling water temperatures undergo
seasonal variations. The performance of the exchanger itself can be expected to fall off over time as fouling
of the heat transfer surface occurs.

The main factors that can be expected to change are:

(1) flow rates

(2) inlet temperatures

(3) fouling levels.

The effects of such changes on exchanger performance must be evaluated at the design stage in order to
demonstrate that the unit is fit for service over the range of duties to which it is likely to be subjected. The
assessment of exchanger operability during design is discussed in Part 6 of the User Guide (HT 208).

There is much uncertainty associated with exchanger design. The main areas of concern in this context are:

(1) the expected fouling levels, specified using so-called ‘fouling factors’

(2) physical property values,

(3) accuracy of heat transfer and pressure drop predictions.

It has been conventional practice to make allowance for such uncertainties through the addition of design
margin (that is, deliberate overdesign). This practice is, however, often counter-productive and even
dangerous. Where a heat exchanger is one of a number in a network, overdesign can result in significant
under-performance of the overall system. This is due to the oversizing of one unit leading to an unforeseen
reduction in temperature driving force within a second unit. Given this danger, it is important that any
affects of oversizing on subsequent system performance be determined at the design stage. The handling
of uncertainty in design is discussed in Part 7 of the User Guide (HT 209).

5.7 Maintenance

Shell-and-tube heat exchangers are usually highly robust pieces of equipment and maintenance needs are
usually low. The problem giving greatest need for maintenance is usually fouling. Potential fouling
problems must be considered when deciding on the allocation of fluids in the exchanger (see Section 6.3).
However, since fouling rates are usually influenced by flow velocities and/or tube wall temperatures there
is often significant interaction between exchanger geometry, operation and fouling. This is discussed in
Part 5 of the User Guide (HT 207).

Two further maintenance aspects need to be considered at the design stage. Firstly, provision must be made

19
HT203
for annual inspection. This usually involves removal of the end headers and inspection of the tube-sheet.
Secondly, provision needs to be made for cleaning. This is discussed more fully when fluid allocation is
discussed in Section 6.3.

Finally, the process engineer should be aware that significant problems can be introduced through a failure
to introduce safeguards against fouling and ingress of foreign objects during plant commissioning. Problems
can be brought about by

• failure to dry exchangers after hydrostatic testing,

• the deposition of corrosion products extracted from corroded pipes used in construction,

• failure to install up-stream strainers to remove objects left in pipes,

• failure to ultimately remove these strainers.

5.8 Longevity

The engineer must give consideration at the design stage to the full life cycle of the exchanger. In terms of
life expectancy, the major consideration is that of materials of construction. Material selection is highly
specialised and involves a consideration of the composition of the fluids being handled (and trace impurities
can be of importance here) and the temperatures at which the exchanger surfaces may be wetted. In this
context the engineer must consider off-design conditions as well as the extremes at which the exchanger
may be required to operate.

20
HT203
6. PRELIMINARY DESIGN DECISIONS

Having selected an E-shell using the guidelines in Section 3, the following design criteria must be
considered to ensure a safe and economic design that should perform well in service:

(1) hazard assessment (Section 6.1),

(2) selection of headers (Section 6.2),

(3) fluid allocation (Section 6.3),

(4) tube configuration and size (Section 6.4),

(5) number of tube passes (Section 6.5),

(6) bundle layout (Section 6.6),

(7) baffle type (Section 6.7),

(8) shells in series criterion ((Section 6.7).

6.1 Hazard Assessment

The first step in the decision making process must be an assessment of the hazards surrounding the
exchanger operation. This will indicate safeguards that must be introduced and direct some of the design
decisions. For instance, if the mixing of the fluids handled by the exchanger must be avoided, then the unit
must either be of fixed tubesheet or U-tube design, and a double tubesheet should be used. So, hazard
assessment leads to the selection of exchanger header type, as described in Section 6.1.

Information arising from the assessment is then used when a decision on fluid allocation (that is, which
fluid flows through the tubes and which flows through the shell) is made. That decision includes an
assessment of tube type and size (Section 6.2), bundle layout (Section 6.3), bundle/baffle type and layout
(Section 6.4).

Some of these choices are influenced by previous decisions. Most can involve iteration within the design
process. The EXPRESS program is designed to allow rapid evaluation of a wide range of design changes
and thereby allows the engineer to rapidly identify the better designs,

6.2 Selection of Headers

As described in Section 3.1, there are a number of header options available to the engineer. Selection of
header types is dictated by the manner in which the tube bundle is fitted into the exchanger shell. A simple
header selection flowchart is given in Sketch 6.1, which encapsulates some of the considerations that must

21
HT203
be made and which are described below.

Do you need to mechanically


clean both sides of the exchanger?

Yes No

Do inlet temperatures
Must mixing of streams
differ by more
be totally prevented?
than 50C?
No Yes Yes No
Consider other
Use
exchanger type Use Fixed
Floating Use U tube
If not acceptable Tube-sheet
Head
use U tube

Sketch 6.1 Simple header selection flowchart

6.2.1 Fixed tube sheet arrangements

Fixed tube sheet arrangements have the advantage that they can be used with single tube passes. This leads
to thermal contacting that approaches pure counter-current flow. The result can be significant reduction in
system complexity in situations where a temperature cross would give rise to the need for multiple
shells-in-series when more than one tube pass is used (see Section 6.8).

Fixed tube sheets should not be used if access to the bundle is required for purposes of cleaning. Otherwise
they can be the preferred unit when a hazardous fluid is to pass through the exchanger shell because the
welded construction removes the need for shellside flanges.

The stresses resulting from thermal expansion can be high in fixed tube sheet arrangements. If the difference
between the two inlet temperatures exceeds 50°C it is likely that an expansion bellows will have to be
incorporated into the exchanger shell. In these circumstances cost considerations will often lead to the
selection of an alternative header type.

6.2.2 U-tubes

The use of U-tubes has the advantage that only one header is required. However, only the shellside of the
exchanger can be properly cleaned by mechanical means. Only chemical cleaning is suitable for the cleaning
of the bends on the tubeside.

This type of unit only has one shellside flange and no internal gaskets. It is, therefore, the preferred option
when a fixed tubesheet cannot be used and mixing of the fluids being handled would gives rise to serious
hazard.

6.2.3 Floating head arrangements

With floating head exchangers only one of the tube sheets is attached to the shell. The other ‘floats’ freely

22
HT203
within the shell thereby catering for thermal expansion stresses.

It is possible to have single tube pass arrangements, with the advantage of near counter-current flow, with
floating head exchangers. However, this requires the use of either a packed gland or an internal bellows.
Neither of these arrangements are suitable when hazardous fluids are being processed. In the case of a
packed gland leakage can be a problem. In the case of an internal bellows, inspection is not easy.

With the split ring design, the header flange extends beyond the bore of the shell and fits into the rear header
space. This arrangement permits closer clearance between the outer tube row and the shell.

The joint between the tubesheet and the cover of the floating head is gasketted and fits within the shell
space. This design is therefore unsuitable for use in duties in which mixing of the fluids handled within the
exchanger would be hazardous.

With the ‘pull-through’ design, the floating head cover fits neatly within the shell bore and can be drawn
through the shell from the other end (hence its name). This design requires larger clearances between the
outer tube row and the inside of the shell. This has two effects. First, the number of tubes that can be
accommodated within a shell of given size may be reduced. Second, the large bypass lane formed by this
clearance can result in a reduction in shellside heat transfer performance. Sealing strips must be used with
this type of arrangement. Again, the unit uses an internal gasket and is therefore unsuitable when mixing
of fluids must be avoided.

For floating head units of the same size, the ‘pull through’ design is the cheaper to fabricate.

6.3 Fluid Allocation

Fluid allocation generally depends on two factors:

(1) the nature of the fluids handled,

(2) the effect the choice has upon the exchanger size and cost.

The properties of the fluid influencing the choice are:

(1) its fouling propensity,

(2) its corrosive nature,

(3) fluid viscosity.

A simplified design chart is given in Sketch 6.2 which encapsulates most of the these design choices. This
chart also higlights opportunities for the use of compact heat exchangers (see ESDU 9201312).

The following comments can be made regarding the fluid property choices listed above.

If just one of the fluids is likely to lead to fouling of the exchanger surface that fluid would generally be
assigned to the tubeside of the unit because it is easier to clean inside the tubes. If both fluids are likely to
lead to fouling of the unit and both sides need to be cleaned then if one of the fluids is amenable to chemical
cleaning (rather than requiring mechanical cleaning) that fluid should be assigned to the shellside. If both
sides needed to be cleaned mechanically the allocation is made on another basis.

23
Does fluid A Does fluid B Shellside of Shellside of
Shellside of Shellside on Shellside on Inside
have a high Is fluid A have a high Is fluid B segmentally- helically- Inside plain Inside twisted Compact heat
Fluid Twisted Tube low-finned high-finned tubes with
fouling corrosive? fouling corrosive? baffled baffled tubes tubes exchanger
exchanger integral tubes integral tubes inserts
propensity? propensity? exchanger exchanger
A Perhaps/No? Yes with SM Yes with SM No No Yes with SM No Yes with SM
Y No
B Yes with SM Yes with SM Yes with SM No No Yes with SM No Yes with SM
A Perhaps/No? Yes with SM Yes with SM No No Yes with SM No Yes with SM
N No
Y B No Yes Yes Perhaps Perhaps Threshold Perhaps Yes
A Perhaps/No? Yes with SM Yes with SM No No Yes with SM No Yes with SM
Y No
N B Yes with SM Yes with SM Yes with SM No No Yes with SM No Yes with SM
Y A Perhaps/No? Yes with SM Yes with SM No No Yes with SM No Yes with SM
N No
B Yes Yes Yes Yes Yes Yes Yes Yes
A Perhaps/No? Yes Yes Perhaps Perhaps Threshold Perhaps Yes
Y No
B Yes with SM Yes with SM Yes with SM No No Yes with SM No Yes with SM
N A Perhaps/No? Yes Yes Perhaps Perhaps Threshold Perhaps Yes
Y N Perhaps
B No Yes Yes Perhaps Perhaps Threshold Perhaps Yes
Y A Perhaps/No? Yes Yes Perhaps Perhaps Threshold Perhaps Yes
No
N Y B Yes with SM Yes with SM Yes with SM No No Yes with SM No Yes with SM
A Perhaps/No? Yes Yes Perhaps Perhaps Threshold Perhaps Yes
N Perhaps
B Yes Yes Yes Yes Yes Yes Yes Yes
A Yes with SM Yes with SM Yes with SM No No Yes with SM No Yes with SM Perhaps with
Y B Yes with SM Yes with SM Yes with SM No No Yes with SM No Yes with SM SM
A Yes with SM Yes with SM Yes with SM No No Yes with SM No Yes with SM Perhaps with
Y N B threshold Yes Yes Perhaps Perhaps Threshold Perhaps Yes SM
24

A Yes with SM Yes with SM Yes with SM No No Yes with SM No Yes with SM Perhaps with
N N Y B Yes with SM Yes with SM Yes with SM No No Yes with SM No Yes with SM SM
Y A Yes with SM Yes with SM Yes with SM No No Yes with SM No Yes with SM Perhaps with
N B Yes Yes Yes Yes Yes Yes Yes Yes SM
A Yes Yes Yes Yes Yes Yes Yes Yes Perhaps with
Y B Yes with SM Yes with SM Yes with SM No No Yes with SM No Yes with SM SM
N A Yes Yes Yes Yes Yes Yes Yes Yes
Y N Perhaps
B threshold Yes Yes Perhaps Perhaps Threshold Perhaps Yes
A Yes Yes Yes Yes Yes Yes Yes Yes Perhaps with
N Y B Yes with SM Yes with SM Yes with SM No No Yes with SM No Yes with SM SM
A Yes Yes Yes Yes Yes Yes Yes Yes
N Yes
B Yes Yes Yes Yes Yes Yes Yes Yes

Key:
SM = special materials (e.g. titanium, Duplex, Superduplex)
Threshold = use EXPRESS to assess designs below fouling thresholds (crude oil, velocity-dependent and wall temperaure-dependent fouling)

Sketch 6.2 Fluid Allocation Selection Chart

HT203
NOTE TO THE COMMITTEE. TO BE CHECKED (P.S. IS THIS USEFUL? CAN IT BE SIMPLIFIED?)
HT203
Where one of the fluids has a significant viscosity (greater than 5 cps) then the designer is faced with a
number of choices. By assigning the fluid to the shellside higher Reynolds Numbers can be achieved and
there is usually better heat transfer and consequently a size advantage. However, if the use of a tube insert
is acceptable there can be advantage in assigning the high viscosity fluid to the tubeside.

If one of the fluids is corrosive and the other not, capital cost savings can often be made by assigning this
fluid to the tubeside. The number of components wetted by the tubeside fluid is fewer than that wetted by
the shellside fluid. These components are also less complex in form and manufacture.

There is one significant danger associated with assigning the high viscosity fluid to the shellside. If the
difference between the viscosity of the fluid at its highest bulk temperature is significantly lower than that
at the cold fluid inlet temperature the flow through a bundle using segmental baffles can be much lower
than expected.

The mechanical clearances required for the fabrication of the exchanger provide an area surrounding the
bundle that does not contain any tubes. This provides a ‘lane’ for flow of the fluid around rather than through
the bundle. The quantity of fluid bypassing the bundle will depend upon the resistance to flow presented
by this lane and by the bundle itself. The viscosity of the fluid inside the bundle differs from that in the
bypass lane. This also effects the flow distribution. If the fluid flowing through the bundle is low its
temperature drops more than expected (and, at an extreme it approaches that of the cold fluid). The position
can be reached that the differences between the viscosity of the fluid within the bundle and that in the bypass
is so large that little fluid actually flows through the bundle itself.

If the potential viscosity differences are large then the designer should either assign the high viscosity fluid
to the tubeside and use a tube insert or should use a helical baffle rather than a segmental baffle. If a
segmental baffle is to be used sealing strips should be installed in order to increase the flow resistance
presented by the bypass lane. This action should be taken irrespective of the magnitude of the potential
viscosity difference.

If the allocation is not dictated by the nature of the fluids then the EXPRESS program can be used to quickly
determine the effect the allocation has upon exchanger size and geometry. Parameter Plots can be quickly
generated for each allocation. The only change that has to be made to the input is an adjustment of the
allocation button.

6.4 Tubes

6.4.1 Tube configuration

Plain tubes Most shell-and-tube heat exchangers use plain tubes, but the designer has a range of options
that can lead to smaller-more efficient exchangers or exchangers with better fouling performance and hence
reduced total life cycle costs (even where the budget cost of the unit is higher).

Twisted tubes are increasingly used, particularly in retrofit and debottlenecking applications where, for
example, better fouling performance or increased thoughput is required. Further guidance on twisted tube
bundles is given in ESDU 0001619. Twisted tube bundles have the benefit of enhancing heat transfer on
both the tubeside and shellside.

Tube inserts come in a range of types and configurations. They can be used for a variety of purposes (see
ESDU 9700713) ranging from improvement of exchanger operability to the mitigation of fouling. In the
context of the design of new exchangers, the principle interest is capital cost reduction. The important
opportunity occurs when the use of heat transfer enhancement allows the duty to be achieved in fewer
exchanger shells. This can be brought about by permitting the use of single tube pass arrangements at lower

25
HT203
velocities or by reducing the amount of area needed such that it can be accommodated in fewer shells.

The EXPRESS program contains equations for a range of insert types (wire matrix, twisted tapes and helical
coils). Particularly in the case of wire matrix inserts, the equations are only representative of expected
performance. The engineer should consult with the supplier to confirm application. Furthermore, wire
matrix inserts come in a wide range of configurations. Only two representative examples are covered by
the program.

6.4.2 Tube size

Consideration of the influence of tube size upon exchanger size and performance indicates the following
for a given tubeside flow rate

(1) exchanger volume per unit heat surface transfer area increases as the tube outside diameter
increases, leading to less compact exchanges,

(2) the film heat transfer coefficient decreases with increasing tube size,

(3) the product of overall heat transfer coefficient and surface area, UA, decreases as diameter
increases,

(4) the tubeside pressure drop decreases as tube diameter increases.

Where the user has control over tube sizing, these observations suggests the following strategy which should
be used:

(1) start with small tubes (recommendation: ¾ inch o.d.)

(2) increase tube size if difficulty in meeting the tubeside pressure drop constraint is encountered during
design.

6.5 Number of Tube Passes

Single tube pass units make best use of available temperature driving force and are therefore the preferred
option. However, they can only be used with fixed tubesheet units or in units that use a special form of
floating head (see Section 3).

Where an E-shell is used with more than one tube pass there may be a need to use more than one
shell-in-series in order to avoid a temperature cross.

If a single tube pass cannot be used, close to pure counter-flow can be achieved by using two tube passes
in an F-shell. However, efficiency is affected by heat transfer through the longitudinal baffle and by fluid
leakage across this baffle, as noted in Section 3.2.

Since shell diameter is the controlling factor in the cost of the exchanger, the designer should seek to
minimise the number of tube passes used within performance constants and consideration of tubeside
fouling (see Part 5 of the User Guide, HT 207).

6.6 Bundle Layout

Sketch 6.3 illustrates the tube arrays commonly used in shell-and-tube heat exchangers, namely equilateral
triangular, square and staggered square. The triangular arrangement leads to the most compact design

26
HT203
whereas the square array simplifies some maintenance operations (for example, cleaning on the shell side).
A typical tube layout has a pitch of 1.25 × tube diameter. Generally, for clean services, the smallest pitch
and a triangular layout is the preferred choice. Square or staggered square layouts are appropriate when
mechanical cleaning on the shell side is required.

Sketch 6.3 Some examples of tube arrays used in shell-and-tube heat exchangers

The tighter the tube pitch ratio the larger the tube count is for a given size of shell. So, the engineer should
start with the smallest pitch ratio (usually 1.25) and only open out to a larger pitch if difficulties are
encountered in meeting the shellside pressure drop constraint.

The number of tubes that can be fitted into a given shell diameter is higher for staggered bundles. In terms
of heat transfer per unit pressure drop the 30, 60 and 90 degree arrangements show similar performance.
The 45 degree arrangement gives about 10% better heat transfer performance for a given pressure drop
(note: this does not mean that the 45 degree arrangement gives better performance at identical cross flow
velocities – the key consideration is the relationship between heat transfer and pressure drop).

The 30 degree arrangement tolerates higher velocities before the onset of vibration.

Bringing these observations together the following recommendations are made:

(1) if there is a need for mechanical cleaning use a 45 degree arrangement,

(2) if chemical cleaning can be used, use a 30 degree arrangement.

Only staggered tube layouts are used in fixed tubesheet units (see section on bundle layout).

27
HT203
6.7 Baffle Type

6.7.1 Baffle arrangement

A range of baffle types is available, as illustrated in Sketch 6.4. The traditional type is the “single segmental
baffle”. This was subsequently supplemented by the “double segmental baffle”, the “triple segmental baffle”
and by the “disc and doughnut baffle”.

The double and triple segmental baffles were introduced for use in applications in which shellside pressure
drop is a controlling factor. Their use involves reduction in the cross-flow nature of the shellside flow in
favour of longitudinal flow. One disadvantage of their use is increase in the unsupported tube length.

Recent years have seen the introduction of improved designs, namely the “rod baffle”, the “helical baffle”
and “twisted tube bundles”. These more modern arrangements also allow a high component of longitudinal
flow. Beneficially, these designs involve greater tube support (particularly the twisted tube design which
eliminates the need for baffles entirely). Further guidance on helical baffles (the ‘Helixchanger’) and
Twisted Tube exchangers is provided in ESDU 0001619, where particular emphasis is given to their benefits
in crude oil fouling reduction.

The EXPRESS program contains design and performance procedures for segmental baffles, helical baffles
and rod baffles. Since costed designs can be developed rapidly for each type, the engineer can use the
program to undertake a systematic appraisal and comparison of the performance of the differing types on
their specific duty.

28
HT203

Principle of the TWISTED TUBETM exchanger


(Brown Fintube Company)

Segmental baffle designs

Helical baffle (Helixchanger)


(ABB Lummus Heat Transfer)

ROD baffle support system designs


Sketch 6.4 Some important baffle types

29
HT203
6.7.2 Baffle geometry - segmental baffles

The geometry of segmental baffles has a very marked effect upon heat exchanger performance. Sketch 6.5
shows the results of a study dealing with segmental baffles, where the effect of baffle cut upon heat transfer
surface area required for a given duty was studied21. For each evaluation an exchanger of fixed tube count
and fixed baffle pitch was used. Hence, the area requirement translate into additions to both the exchanger
tube length and shellside pressure drop. A distinct optimum was found: for the chosen application this
optimum was a cut of 27%. It was found that the use of a 15% cut baffle resulted in an area requirement
that was 23% higher than the minimum value and the use of a 45% cut resulted in a need that was 17%
greater than the minimum. Most importantly, it was found that the optimum performance coincided with
the point at which the free flow area in the window approximately equated with the baffle cross-flow free
area.

90

85

area required
2 80
(m )

75

70
15 20 25 30 35 40 45
baffle cut (%)

Sketch 6.5 Effect of Baffle Geometry on Exchanger Performance

It was subsequently found that changes in baffle pitch resulted in changes in the position of the optimum
cut. However, the optimum cut always occurred where the window and cross-flow, flow areas were equal.

Using this rule in shell-and-tube exchanger design with segmental baffles, the following should be sought:

(1) full use of available shellside pressure drop,

(2) a baffle arrangement in which the window and cross-flow, flow areas are equal.

The EXPRESS program optimises designs by setting equal window and cross-flow areas.

6.7.3 Baffle geometry - helical baffles

With helical baffles, the angle at which the baffle is set has a marked effect upon both heat transfer coefficient
and pressure drop10. Available data show that the heat transfer coefficient is independent of angle up to a
value of around 25 degrees. Above this, the coefficient increases sharply reaching a peak at around 40
degrees. A similar pattern is observed for pressure drop under turbulent flow conditions.

Baffle angle and baffle pitch are not set independently. The construction is such that once the angle has
been chosen the pitch is set within EXPRESS by the exchanger shell diameter.

30
HT203
6.7.4 Baffle geometry - ROD baffles

(This section to be prepared.)

6.7.5 Other baffle types

(This section to be prepared.

6.8 Shell in Series Criterion

It is normal practice to use shell-and-tube heat exchangers with two or more tube side passes. This can give
rise to the so-called “temperature meet” effect, which is illustrated in Sketch 6.6. Consider first the case
where the fluid in the first tube side pass is flowing in the opposite direction (that is, “countercurrent”) to
the hot side fluid (as shown in Sketch 6.6a). Countercurrent flow gives the highest efficiency and the
interpass temperature, Ti , reaches a high value. However, in the second pass the tube side temperature may
exceed the shell side temperature, as illustrated. This gives a reverse heat flow and a reduction of efficiency.
Now consider the case where the fluid in the first tube side pass is flowing in the same direction as the shell
side fluid (that is, “co-current”, see Sketch 6.6b). Here, with co-current flow, the value of Ti is much smaller
since co-current flow is much less efficient. Even though the second pass is countercurrent, the overall
efficiency is still low relative to a pure counter-flow exchanger. Note that the cold side fluid outlet
temperature, Ti, out is identical for the two cases, as can be shown analytically.

Th,in Th,in

Tc,in
Tc,out
Tc,out
Tc,in

Th,out Th,out
Reverse heat flow
Th,in Th,in

Tc,out Tc,out

Th,out Th,out

Tc,in Tc,in

(a) First tubepass counter-current (a) First tubepass co-current

Sketch 6.6 Temperature meet effects

The phenomenon in which Tc, out > Th, in in multiple pass shells is often referred to as a “temperature meet”
or “temperature cross”. The effects of such temperature meets (or crosses) should normally be minimised,
as follows.

31
HT203
For a single phase flow with constant specific heat capacity, the governing heat balance equation is:

·
Q = M· h c p,h ( T h,in – T h,out ) = M· c c p,c ( T c,out – T c,in ) . (6.1)

For a pure countercurrent flow heat exchanger with constant U and constant specific heat capacities, the
mean temperature difference ∆T m is given as the logarithmic mean temperature difference ∆T lm :

( T h,in – T c,out ) – ( T h,out – T c,in )


∆T lm = ---------------------------------------------------------------------------------------- . (6.2)
( T h,in – T c,out )
log e ----------------------------------------
( T h,out – T c,in )

For exchangers which depart from pure countercurrent flow, typified by the multi-pass shell-and-tube heat
exchanger, the mean temperature difference is given by:

∆Tm = FT ∆Tlm , (6.3)

where FT is a correction factor, whose value can be determined from standard charts (see for instance those
of Taborek15) in terms of the end temperatures of the respective fluids. Typical practical values of FT are
around 0.9 for multi-pass shell-and-tube exchangers and around 0.7 for pure cross-flow exchangers. For
FT values less than about 0.8, multiple shell units are usually recommended.

EXPRESS uses the above crieria to determine the need for more than one shell in series, and allows the
user to either override the default value of FT =0.8 or enter directly the number of shells in series, where
this is known.

Further guidance on the shells in series calculation is given in ESDU 9800314 and 9404211, where it is
recommended that the alternative Effectiveness – NTU method be employed. However, for the purposes
of the EXPRESS program, the FT factor method is considered suitable.

Further guidance on the Effectiveness – N TU method is given in Appendix A5.

32
HT203
7. REFERENCES

1. BSI Circular flanges for pipes, valves and fittings (PN designated).
Specification for copper alloy and composite flanges. BS 4504 Pt. 1.
British Standards Institution, London, UK, 1969. (see
http://bsonline.techindex.co.uk for latest version)
2. ROETZEL, W. Mean temperature difference for heat exchanger design – a general
NICOLE F.J.L. approximate explicit equation, J.Heat Transfer, Vol.97, No.1,
pp.5-8, 1975.
3. JARZEBSKI, A.B. Approximate mean temperature difference for calculation of heat
LACHOWSKI, A.I. exchangers, Canadian J. Chem. Engng, Vol.55, pp.741-743., 1977.
SZPONARSKI, T.
GASIOR, S.
4. ESDU Baffled shell-and-tube heat exchangers; flow distribution, pressure
drop and heat transfer coefficient on the shellside. ESDU Data Item
83038, ESDU International plc, London, UK, Mar, 1984.
5. ESDU Flow induced vibration in tube bundles with particular reference to
shell and tube heat exchangers. ESDU Data Item 87019, ESDU
International plc, London, UK, Oct. 1987.
6. SAUNDERS, E.A.D. Heat Exchangers: Selection, design and construction, Longman
Higher Education, Harlow, UK, 1988. (ISBN 0-582-49491-5)
7. GENTRY C.C. ‘RODbaffle Heat Exchanger Technology’, Chemical Engineering
Progress, pp48-55, July 1990.
8. PODDAR, T.K. Private communication with the Institution of Chemical Engineers,
Rugby, UK, 1990.
9. ESDU Forced convection heat transfer in straight tubes. Part 1: turbulent
flow. ESDU Data Item 92003, ESDU International plc, London,
UK, Aug. 1993.
10. STEHLIK P., ‘Comparison of Correction Factors for Shell-and-Tube Heat
NEMCANSKY J., Exchangers with Segmental or Helical Baffles’, Heat Transfer
KRAL D., Engineering, Vol.115, Pt.1, pp.55-65, 1994.
SWANSON L.W.
11. ESDU Selection and costing of heat exchangers. Shell-and-tube type.
ESDU Data Item 94042, ESDU International plc, London, UK, Dec.
1994.
12. ESDU Selection and costing of heat exchangers. ESDU Data Item 92013,
ESDU International plc, London, UK, Dec. 1994.
13. ESDU Heat transfer enhancement in heat exchanger design and utilisation.
Part 1. Tube inserts in single-phase flow. ESDU Data Item 97007,
ESDU International plc, London, UK, Apr. 1998.
14. ESDU Design and performance evaluation of heat exchangers: the
effectiveness-NTU method. Part 1 to 5, ESDU Data Item 98003 -
98007, ESDU International plc, London, UK, July 1998.

33
HT203
15. TABOREK, J. Charts for mean temperature difference in industrial heat exchanger
configurations. Chapter 1.5 of Heat Exchanger Design Handbook,
(HEDH), (G.F. Hewitt, Ed.), Begell House, Inc., NY, 1998.
16. TABOREK, J. Thermal and hydraulic design of heat exchangers. Shell-and-tube
heat exchangers: single-phase flow. Chapter 3.3 of Heat Exchanger
Design Handbook (HEDH), (G.F. Hewitt, Ed.), Begell House, Inc.,
NY, 1998.
17. TUBULAR Standards of Tubular Exchanger Manufacturers Association, 7th
EXCHANGERS Edition, 1988 and 8th Edition, 1999, TEMA, New York, 1999.
MANUFACTURERS
ASSOCIATION (TEMA)
18. PODDAR, T.K., Optimising the design of shell-and-tube heat exchangers, Chem.
POLLEY, G.T. Eng Prog, Sept. 2000.
19. ESDU Heat exchanger fouling in the pre-heat train of a crude oil
distillation unit. ESDU Data Item 01016, ESDU International plc,
London, UK, Jan. 2001
20. ESDU Forced convection heat transfer in straight tubes. Part 2: laminar and
transitional flow. ESDU Data Item 93018, ESDU International plc,
London, UK, Nov. 2001.
21. POLLEY, G.T. Effect of baffle geometry on exchanger performance. (Reference to
et al. be provided.)
22. UNDERWOOD Reference to be supplied.

34
HT203
APPENDIX A BACKGROUND TO METHODS USED FOR PRELIMINARY SIZING

A1. Model of Two-stream Heat Exchanger

The method of analysing the transfer of heat in heat exchangers is summarised in ESDU 9800314. The
method starts from the consideration of a generalised heat exchanger in which heat is transferred from a
hot stream (stream 1) to a cold stream (stream 2).
.
T1,in, M1, cp,1

.
T2,in, M2, cp,2 HX T2,out

T1,out

Sketch A1.1 Simplified representation of a heat exchanger

In passing through the heat exchanger, the hot stream cools from T 1,in to T 1,out while the cold stream is
heated from T 2,in to T 2,out . By a simple heat balance the rate of heat transfer (which is commonly referred
to as the “thermal duty” or “heat load” of the exchanger) between the two streams is

· · ·
Q = ( M c p ) 1 ( T 1,in – T 1,out ) = ( M c p ) 2 ( T 2,out – T 2,in ) , (A1.1)

· ·
where ( M c p ) 1 and ( M c p ) 2 are the average thermal capacity rates in each stream. Alternatively, the rate
of heat transfer may be expressed in terms of the larger and smaller stream properties as

· ·
Q = ( M c p ) smaller ( T in – T out ) (A1.2)
larger
· ·
or Q = ( M c p ) larger ( T in – T out ) . (A1.3)
smaller

In addition the rate of heat transfer may also be expressed by

·
Q = UA∆T m , (A1.4)

where U is the overall heat transfer coefficient between the streams, A is the effective heat transfer surface
area of the heat exchanger and ∆T m is the effective mean temperature difference between the streams.
Care must be taken when evaluating U and A to ensure that they are compatible. For example, for a
shell-and-tube exchanger, if U is calculated on the basis of the outside tube surface then the area term in
Equation (A1.4) will be that of the outside surface of the tubes.

For a truly counter-current flow heat exchanger, the effective temperature driving force is the log-mean

35
HT203
temperature difference (LMTD), given as

( T 1, in – T 2, out ) – ( T 1, out – T 2, in )
∆T lm = -------------------------------------------------------------------------------------------- . (A1.5)
 T 1, in – T 2, out 
loge  -------------------------------------
 T 1, out – T 2, in 

Most shell-and-tube heat exchangers do not approximate a pure counter-current flow arrangement. For
example, it is common to pass the tubeside fluid along the length of the exchanger more than once (that is,
to use multi-tube pass designs). There are two reasons for this: to allow higher tubeside velocities (and
hence better heat transfer and often less fouling) and to avoid the need for expensive shell bellows. With
multiple tube passes the thermal expansion of the exchanger can be accommodated using either U tubes or
a floating head.

For all geometries, the mean temperature difference based on the same inlet and outlet temperatures is either
equal to or less than the LMTD. The F T factor is used to account for the drop-off in heat transfer
performance from the ideal counter-current case, and defined as

∆T m
F T = ------------- , (A1.6)
∆T lm

so the heat load can therefore be expressed as

·
Q = F T UA∆T lm . (A1.7)

The F T factor is a function of the ratio of the thermal capacities of the streams, C*, where thermal capacity
is defined as the product of the mass flow rate and the average specific heat capacity. In general, in a
two-stream heat exchanger one thermal capacity will be greater than the other. The thermal capacity ratio
is defined as the ratio of the smaller thermal capacity to the larger:

·
( M c p ) smaller
C* = ------------------------------
· -. (A1.8)
( M c p ) larger

From this definition, 0 < C* ≤ 1, and a value close to zero is most likely to occur in the isothermal evaporation
or condensation of a pure substance where the thermal capacity in the two-phase stream is very high
(numerically close to infinity).

The thermal effectiveness of the exchanger is defined as the ratio of the ‘actual heat exchange in the unit’
to the ‘maximum possible heat exchange’ (assuming an infinitely-long counter-current exchanger), and is
given by the following expression:

( T in – T out )
larger
E = ----------------------------------------------- , (A1.9)
( T 1,in – T 2,in )

Where stream 1 is the hot stream and stream 2 is the cold stream. Implicit in all these simple relationships
that are used in the methodology of EXPRESS are the following conditions:

(a) steady-state conditions prevail,

36
HT203
(b) both streams have constant and uniform heat capacity,

(c) each stream is perfectly mixed at each point in the exchanger,

(d) film heat transfer coefficients are constant and uniform

(e) fouling resistance is constant for each stream and uniform throughout the exchanger.

Using the assumptions described above (steady state precails etc.) relationships have been developed for
the varipous exchanger geometries in common use. These are reported by Underwood22, Roetzel et al.2
and Jarzebski et al.3

A2. Tubeside Heat Transfer

Convective heat transfer to and from fluids flowing inside tubes is discussed in ESDU 920039. Following
the recommendations in ESDU 92003, the film heat transfer (expressed by the Nusselt number) in the
turbulent flow regime is given by following equation (presented by Petukov), valid over the ranges
5
4000 ≤ Re ≤ 5 × 10 and 0.6 ≤ Pr ≤ 60 :

( f ⁄ 2 )RePr
Nu = --------------------------------------------------------------------------
½ 2⁄ 3
-, (A2.1)
turb 1.07 + 12.7 ( f ⁄ 2 ) ( Pr – 1 )
1
where f = -------------------------------------------------------------- . (A2.2)
2
4 ( 1.82log 10 Re – 1.64 )

Information on the accuracy of this equation and comparisons of its performance against other
commonly-used equations is given in ESDU 92003 and discussed in Part 7.

Ignoring natural convection affects, the film heat transfer under laminar flow conditions can be estimated
from:

3 3 1⁄ 3 3 1⁄ 3
Nu = 3.66 + 0.7 + ( 1.77 Gz – 0.7 ) , (A2.3)
lam

where Gz is the Graetz number, given by


π D
Gz = --- RePr ---- . (A2.4)
4 L

This equation is recommended in ESDU 9301820, and applied to thermally-developing flows assuming
uniform wall temperature boundary conditions.

The prediction of Nusselt number in the transition flow region is difficult, as the heat transfer process
includes characteristics of both laminar and turbulent flow. However, the available experimental data
indicate that the Nusselt number lies in the range between its limiting values for laminar and turbulent flow.
ESDU 9301820 recommends the use of a linear interpolation of the values of Nusselt number for laminar

37
HT203
flow and turbulent flow over a transition region of 2000 ≤ Re ≤ 8000 . The interpolation function is

Nu = ε Nu lam + ( 1 – ε ) Nu turb (A2.5)


Re
ε = 1.33 – ------------ . (A2.6)
6000

Approximate heat transfer correlations for tubes fitted with inserts are given in ESDU 9700713.

A3. Tubeside Pressure Drop

The pressure drop associated with flow through a straight tube is given by the equation:

L
∆p = 2f  ------ ρva 2 (A3.1)
D 

where f is the Fanning friction factor. Saunders6 recommends the following simple equations for the friction
factor inside plain tubes in shell-and-tube heat exchangers:

For Re < 1311: f = 16/Re . (A3.2)

For 1311 < Re < 3380 : f = 0.0122 . (A3.3)

0.264
For Re > 3380 : f = 0.0035 + --------------- . (A3.4)
Re 0.42

These equations are used in preference to the friction factor term in Equation (A2.1): they were developed
from data for corroded tubes and are thought to be more representative of the conditions in an operating
shell-and-tube heat exchanger. A discussion of the relative performance of these equations is given in Part
7 of the User Guide (HT 209).

Equations for friction factors for flow through tubes fitted with tube inserts are given in ESDU 9700713.

For single pass exchangers, the losses in the exchanger headers is taken to be 0.9 tube velocity heads. For
multi-pass units, a loss of 1.6 velocity heads per pass is assumed.

The pressure loss through each exchanger nozzle is assumed to be 1.5 nozzle velocity heads.

A4. Shellside Heat Transfer and Pressure Drop

A4.1 Segmental baffles

Accurate modelling of the flow on the shellside of a shell-and-tube heat exchanger fitted with segmental
baffles is difficult. Guidance on and equations for the prediction of pressure drop and heat transfer in
conventionally-baffled shell-and-tube heat exchangers is given in ESDU 830384.

Shellside heat transfer is promoted by the flow of fluid across the tubes, but not all of the flow entering the

38
HT203
shell flows through the bundle itself. The mechanical clearances required for the construction of the unit
provide flow paths between the tube bundle and the shell and between the edge of the baffle and the shell.
Fluid also bypasses the tube bundle by flowing down the pass partition lanes. Finally, not all of the fluid
entering the bundle flows across the tubes. Some fluid passes between the baffle spaces by leaking through
the clearance between tube and baffle plate.

Heat exchanger analysis programs contain quite sophisticated models that seek to account for the presence
of these various flow paths, often using the flow stream analysis method of ESDU 830384. However,
EXPRESS provides the opportunity to explore the differences in predicted shellside and heat transfer
between this method and the simpler Bell-Delaware method described by Taborek16. The user must
appreciate that there are considerable uncertainties associated with all methods of predicting heat transfer
and pressure drop. By allowing a comparison to be made between two well-recognised methods, some
assessment of confidence in the design can be made (for the user of the EXPRESS program and other
commercial design programs).

A comparison of the performance of the stream analysis method and the Bell-Delaware method is provided
by Taborek16. An assessment of the performance of the stream analysis method against a data set of
high-quality laboratory measurements and some real exchanger case studies is given in ESDU 830384.

In duties that are either tubeside-controlled or where shellside fouling dominates the calculation of the
shellside heat transfer, the choice of shellside method is often relatively unimportant for preliminary
calculations such as those performed by EXPRESS.

The prediction of the shellside film heat transfer coefficient using the Bell-Delaware method involves the
determination of ‘ideal’ values of heat transfer coefficient and pressure drop, based upon the flow through
the shell, followed by the application of correction factors (based on bundle geometry) to account for the
effects of bundle bypassing and fluid leakage between baffle and shell, between baffle and tube and through
pass partition lanes. A correction to account for the different geometry encountered in the exchanger end
zones is also applied. The value of the bypass correction is dependant upon the number of sealing strips
used in the design, and is a design variable.

In the EXPRESS program the number of sealing strips is calculated to be the number necessary to prevent
the leakage correction factor falling below 0.8.

The shellside pressure drop is assumed to consist of three components: a cross-flow component, a window
component and an end zone component. Ideal values are calculated for each component and correction
factors applied for bypassing and leakage.

A4.2 Helical Baffles

The Bell-Delaware Method has been extended to cover bundles fitted with helical baffles by Stehlik et al.10
and has been incorporated into the EXPRESS program. This new methodology is described in an ESDU
Memorandum.

A4.3 ROD Baffles

Similarly, a methodology for the prediction of the performance of bundles fitted with rod baffles is used
and fully described by Gentry7. This methodology is described in an ESDU Memorandum.

39
HT203
A5. Effectiveness-NTU Equations

The concept of Effective Mean Temperature Difference defined in Equation (A1.4) is easy to apply in
design where exchanger terminal temperatures are known and fixed. However, when exchanger operability
is being considered, exchanger simulation must be employed. Here exchanger area is fixed, inlet
temperatures are known and exit temperatures are subsequently calculated. The Effectiveness-NTU method,
described in ESDU 98003 - 9800714, is ideal for these simulations. It can also be used to select appropriate
geometries for given heat transfer duties. Performance data for a very wide range of geometries is given in
ESDU 98005.

The effectiveness of the exchanger is defined as

( T in – T out )
l arg er
E = ------------------------------------------------- (A5.1)
( T 1,in – T 2,in )
∆T l arg er
= ------------------------------------- , (A5.2)
( T 1,in – T 2,in )

where stream 1 is the hot stream, stream 2 is the cold stream and ∆Tl arg er is the modulus of the larger
temperature ( T1,in – T2,out ) and ( T2,in – T1,out ) . By defining ∆Tsmaller as the smaller of these two
expressions and using the ‘ineffectiveness’, 1 – E , similar expressions can be written:

( T in – T out )
smaller
1 – E = ----------------------------------------------------- (A5.3)
( T 1,in – T 2,in )
∆T smaller
= ------------------------------------- . (A5.4)
( T 1,in – T 2,in )

Since the inlet temperatures are known, knowledge of the Effectiveness leads directly to estimation of
exchanger outlet temperature.

For an exchanger of given type, there are relationships between Effectiveness and exchanger size and
throughput expressed in terms of Number of Transfer Units, NTU, which dependant on the overall heat
transfer coefficient for the exchanger U, the heat transfer surface area, A, the flow rate of the stream having
the lower heat capacity flow rate, M· smaller, and the heat capacity of that stream, ( c p )smaller,.

UA
N TU = --------------
- (A5.5)
( M· c p ) smaller

Explicit equations are available for various exchanger configurations in ESDU 98005. Those applicable to
E-shell exchangers are incorporated in the EXPRESS computer program.

The Effectiveness - NTU method is based upon the same simplifying assumptions used in the Effective
Mean temperature Difference approach.

40

You might also like