Heat Transfer

Does Your Application

Call for an
F-Shell Heat Exchanger?
Rajiv Mukherjee
If the process has a temperature
Consultant cross or a low flowrate,
an F shell might prove beneficial.

EVERAL SHELL CONFIGURATIONS, Shell types

S designated E, F, G, H, J, K and X by the Tubular

Exchanger Manufacturers Assocation, Inc.
(TEMA), are available for shell-and tube heat ex-
In shell-and-tube heat exchangers, there are various
patterns of flow through the shellside, each with its
special features and applications. TEMA has devel-
changers. E shells are by far the most common oped a nomenclature for shell types, as shown in Fig-
throughout the chemical process industries. In certain ure 1 (as well as for front and rear heads, which due to
situations, though, F shells offer advantages. This arti- space limitations are not shown here).
cle discusses two such situations — when there is a A TEMA E shell is a single-pass shell. The shellside
temperature cross and when the shellside flowrate is fluid enters at one end and leaves through the other end.
low — and demonstrates why an F shell would be ben- A TEMA F shell (Figure 2) is a two-pass shell
eficial in those applications. that is divided into two passes, an upper pass and a
lower pass, by a longitudinal baffle. The shellside
stream enters at one end in either the upper or the
lower half (first pass), traverses the entire length of
the shell through that half of the shell, turns around
E
and flows through the other half of the shell (sec-
One-Pass Shell ond pass), and finally leaves at the same end of the
J shell through which it entered. The longitudinal
Divided Flow
F
Two-Pass Shell
with Longitudinal Baffle
K

Kettle Reboiler
G

Split Flow
X

Crossflow
H

Double Split Flow Longitudinal Baffle

■ Figure 1. TEMA nomenclature for shell designs. ■ Figure 2. An F shell has a longitudinal baffle that divides it
Source: www.tema.org/TEMAfaqshtm.htm#nomenclature into two passes.

40 CEP April 2004 www.cepmagazine.org

baffle does not extend to the tubesheet at the far end, The merits of the F shell
but stops somewhat short of it so that the shellside If a temperature cross exists, a single E shell is thermo-
fluid can flow from the first pass into the second pass. dynamically incapable of accomplishing the specifed heat
This construction is used for temperature-cross situa- duty. Depending on the degree of temperature cross, two or
tions, that is, where the cold fluid leaves at a tempera- more E shells in series are required. In most CPI heat ex-
ture higher than the outlet temperature of the hot changer services, the degree of temperature cross is moder-
stream. If a two-pass (F) shell has only two tube pass- ate, so two E shells can generally accomplish the task.
es, it becomes a true countercurrent configuration and In such situations, a single F shell can be used instead
it can handle a large temperature cross. of two E shells in series, thereby leading to savings in ex-
A TEMA J shell is a divided-flow shell typically used changer and piping cost. Should an F shell require only
for minmizing shellside pressure drop. The shellside two tube passes to yield the desired tubeside velocity, it
fluid enters at the center (along the length) and divides represents true countercurrent flow, which can handle any
into two halves, one flowing to the left and the other to degree of temperature cross.
the right; the streams leave separately and are combined Often, when the shellside flowrate is relatively low, the
into a single stream by external piping. This is referred to shellside velocity is also low, even with the smallest baffle
as a J1-2 shell. Alternatively, the stream may be split into spacing. In such situations, the allowable shellside pres-
two halves and enter the shell at the two ends, flow to- sure drop cannot be utilized properly. The shellside heat-
ward the center and leave as a single stream; this is re- transfer coefficient, and thus the overall heat-transfer coef-
ferred as a J2-1 shell. ficient, are low, resulting in an unduly large and expensive
A TEMA G shell is a split-flow shell usually employed heat exchanger. Additionally, if the shellside stream is
for horizontal thermosyphon reboilers. It has only one cen- dirty, the low shellside velocity will result in heavy shell-
tral support plate and no baffles. Because TEMA specifies side fouling, which will translate into high operating costs.
a maximum unsupported tube length of about 60 in. or In such cases, many designers use two E shells in se-
1,500 mm for 1-in.-OD tubes, a G shell cannot be used for ries. This yields a higher shellside velocity and thereby a
heat exchangers having a tube length greater than 120 in. higher heat-transfer coefficient. This leads to not only a
or 3,000 mm, as the unsupported length would then ex- smaller and cheaper heat exchanger, but also lower operat-
ceed the TEMA limit. (The TEMA unsupported span limit ing costs due to the reduction in fouling.
varies with tube OD, thickness and material.) As an alternative, an F shell can often yield a compara-
When a larger tube length has to be employed, a ble shellside velocity and heat transfer area. By virtue of
TEMA H shell is adopted. An H shell is really two G being a single shell, such a design can have a lower capital
shells placed side-by-side, so that it has two full support cost than two E shells in series. There will also be a reduc-
plates. The description for this configuration is double- tion in the piping cost, and the lower overall vertical
split, as the flow is split twice and recombined twice. This height can be an advantage in many situations.
construction, too, is invariably employed for horizontal
thermosyphon reboilers. The advantage of TEMA G and H
shells is that the shellside pressure drop is drastically Cross-Baffle
lower than that in an E shell. Diameter Shell ID

A TEMA K shell is a special crossflow shell em-

ployed for kettle reboilers (K for kettle). It has an inte-
gral vapor disengagement space in the shape of an en-
larged shell. Here, too, full support plates can be em-
ployed as required. Clamp Bar

A TEMA X shell is a pure crossflow shell. The shell-

side fluid enters at the top (or bottom) of the shell and
flows across the tubes, then exits from the opposite side of
the shell. The flow may be introduced through multiple 8 Stainless-steel
nozzles located along the length of the shell in order to sealing strips
achieve a better distribution. Because of the low pressure Longitudinal each approx.
Baffle 0.2 mm thick
drop (in fact, there is hardly any pressure drop in the shell,
the only pressure drop being in the nozzles), this configu-
ration is employed for cooling or condensing vapors at
very low pressure, particularly at vacuum. Full support
plates can be located as required for structural integrity;
they do not interfere with the shellside flow since they are
parallel to the flow direction. ■ Figure 3. Stainless steel strips along the baffle can eliminate leakage.

CEP April 200 www.cepmagazine.org 41

Heat Transfer

Physical leakage Table 1a. Process parameters for the

One of the limitations of the F shell is the potential for process condensate interchanger in case study 1.
leakage. Engineers are generally apprehensive of the possi-
bility of physical leakage of the shellside stream from the Shellside Tubeside
inlet pass to the outlet pass across the longitudinal baffle
and the consequent deterioration in performance due to the Flowrate, kg/h 7,100 6,700
loss in shellside performance and the loss in mean temper- Temperature In/Out, °C 255/80 40/236
ature difference (MTD). Heat Duty, MM kcal/h 1.333 1.333
However, with eight to ten pairs of thin stainless steel Allowable Pressure
strips pressing against the longitudinal baffle in the first Drop, kg/cm2 0.4 0.4
(inlet) pass (Figure 3), it is difficult to appreciate this Viscosity In/Out, cP Same as Water Same as Water
concern. Of course, the strips are very likely to get dam- Density, kg/m3 Same as Water Same as Water
aged when such a tube bundle is removed from the shell
Thermal Conductivity
and should therefore be replaced every time the bundle is
In/Out, kcal/h-m-°C Same as Water Same as Water
taken out. In a cost-effective F shell application, this cost
Specific Heat In/Out,
will be insignificant compared to the cost savings the F
shell generates. kcal/kg-°C Same as Water Same as Water
In order to minimize the possibility of physical leakage Fouling Resistance,
across the longitudinal baffle of an F shell, some licensors h-m-°C/kcal 0.0002 0.0002
specify a maximum pressure drop of 0.35 kg/cm2 on the Nominal Line Size, mm 50 50
shellside. This is because the higher the shellside pressure Material of Construction Carbon Steel Type 304L
drop, the greater will be the tendency of the shellside Stainless Steel
stream to leak across the longitudinal baffle. Thus, limiting
the permissible shellside pressure drop is good practice, al-
though 0.35 kg/cm2 might be somewhat conservative and Note the extremely high temperature cross between the
0.5 kg/cm2 may be more realistic. two streams. This called for a pure countercurrent heat ex-
It follows that for services with a low allowable shell- changer. The heat exchanger had to be designed with a
side pressure drop that conform to either of the conditions TEMA BEU construction and stainless steel tubes having
described above (temperature cross and low shellside 25-mm O.D. and 2.5-mm thickness. Due to the U-tube
flowrate), the use of an F shell will be more advantageous. construction, the tube length was not specified and any
Reactor feed/bottom exchangers with condensation and/or suitable length up to 6,000 mm could be selected.
vaporization are a good example. Because the temperature cross was very high, five shells
in series were required. The design that emerged is de-
Thermal leakage scribed in Table 1b. Although the baffle spacing is 40% of
Another potential problem with F shells is thermal leak- the shell I.D. (which is usually an optimum value), the
age across the longitudinal baffle from the hotter shell pass stream analysis is not very good, with a main crossflow
to the colder shell pass. This will adversely affect both the fraction of 43.6% and a shell/baffle fraction of 15.4%. The
shellside heat-transfer coefficent and the MTD. temperature profile distortion-correction factors in the five
Thermal leakage is usually not appreciable unless the dif- shells, from the hottest shell to the coldest, are 0.937,
ference between the shellside inlet and outlet temperatures 0.905, 0.905, 0.907 and 0.918. Because of this and because
is high. Even if the temperature difference is high, thermal the flow pattern between the two streams is not pure coun-
leakage can be avoided by providing a little extra heat-trans- tercurrent, the overall MTD is only 16.3°C. The MTDs for
fer area. Another option is to employ an insulated longitudi- the individual shells, from the hot end to the cold end, are
nal baffle if thermal leakage is expected to be significant. 13°C, 14°C, 15.7°C, 18°C and 21.9°C.
Commercially available software that is usually em- The heat-transfer coefficients are quite high — 3,903
ployed for heat-exchanger thermal design can evaluate both kcal/h-m2-°C on the shellside and 1,868 kcal/h-m2-°C on
physical and thermal leakage across a longitudinal baffle. the tubeside — thereby leading to an overall heat-transfer
coefficient of 702 kcal/h-m2-°C. The shellside heat-trans-
Case study 1: Temperature cross fer coefficient is much higher than the tubeside heat-
Consider the heat exchanger service specified in Table transfer coefficient, since the allowable pressure drop is
1a. This was a process condensate interchanger, that is, a much better utilized on the shellside. There are two pass-
heat exchanger wherein heat is exchanged between the es on the tubeside, and this cannot be increased to four
feed and effluent streams of a stripper. The streams were because the tubeside pressure drop will then exceed the
virtually water and as such, the physical properties of allowable maximum of 0.4 kg/cm2. The net result is an
water were considered for the thermal design. exchanger with five shells, each having a heat-transfer

42 www.cepmagazine.org April 2004 CEP

 Table 1b. Principal construction and performance parameters for On the shellside, the heat-trans-
the process condensate interchanger in case study 1. fer coefficient is far better, 4,166
kcal/h-m2-°C, as the permitted pres-
E Shell Design, F Shell Design, sure drop is far better utilized. Thus,
TEMA Type BEU TEMA Type BFU the overall heat-transfer coeffcient is
516.5 kcal/h-m2-°C. This is 26.4%
Number of Shells in Series 5 2 less than the overall heat-transfer
Shell ID, mm 305 438 coefficient in the BEU design.
Total Heat-Transfer Area, m2 5 × 23.8 = 119 2 × 51.7 = 103.4
The stream analysis reveals that
Number of Tubes × Tube Passes 50 × 2 118 × 2
both the main crossflow fraction and
the baffle/shell flow fraction are high-
Tube O.D. × Thickness × Length, mm 25 × 2.5 × 6,000 25 × 2.5 × 5,500
er than in the BEU design. The tem-
Tube Pitch, mm 26 Rotated 26 Rotated
perature profile distortion-correction
Square (45-deg) Square (45-deg)
factor is 0.963, which is much better
Baffle Spacing, mm 120 150 than that for the BEU design. More
Baffle Cut, % diameter 25 (Horizontal) 20 (Vertical) importantly, there is no LMTD cor-
Nominal Shellside Velocity, Crossflow/Window, m/s 0.24/0.21 0.3/0.3 rection factor (Ft), as the flow is true
Stream Fraction, Tube-to-Baffle Hole 0.122 0.205 countercurrent flow. The net result is
Stream Fraction, Main Crossflow 0.436 0.464 that the overall MTD for the two
Stream Fraction, Bundle-Shell 0.122 0.055 shells is 26.1°C, which is 60% higher
Stream Fraction, Baffle-to-Shell 0.154 0.227 than that of the BEU design. Conse-
Stream Fraction, Pass Partition 0.166 0.049 quently, the total heat-transfer area is
Shellside Heat-Transfer Coefficient, kcal/h-m2-°C 3,903 4,166 103.4 m2, compared to the 119 m2 of
Tubeside Heat-Transfer Coefficient, kcal/h-m2-°C 1,868 937 the BEU design. Because the heat-
Overall Heat-Transfer Coefficient, kcal/h-m2-°C 702 516.5
transfer area is somewhat lower and
Shellside Pressure Drop, kg/cm2 0.32 0.22
there are only two shells, vs. the five
shells in the BEU design, the BFU
Tubeside Pressure Drop, kg/cm2 0.12 0.036
design will be far less expensive.
Mean Temperature Difference, °C 16.3 26.1
An additional advantage is that
Overdesign 3.5 6.2
while the two shells of the BFU design
can easily be stacked, it might not be
area of 23.8 m2. The total heat-transfer area is 119 m2. possible to stack the five shells of the BEU design because it
Let us now consider a design with an F shell with two might be too difficult to remove the upper tube bundles for
tube passes. This will provide true countercurrent flow, so maintenance work. Thus, the five shells of the BEU design
a single shell may suffice. However, upon analysis it was might require two stacks, one having three shells and the
found that although a single shell would suffice, an inordi- other having two, which requires a larger plot area.
nately long tube length would be required; a larger
number of shorter tubes led to an unacceptably low Table 2a. Process parameters for the
tubeside velocity and a much larger heat-transfer liquid-liquid heat exchanger in case study 2.
area. Thus, the design that emerged has two TEMA
BFU shells in series, each having tubes 5.5 m long Shellside, Tubeside,
(straight length). This is equivalent to having a sin- Stream 1 Stream 2
gle shell with 11.0-m-long tubes. Some of the more
important construction and performance parameters Flowrate, kg/h 22.5 344
of this configuration are indicated in the second
Temperature In/Out, °C 369/300 291.6/296.2
column of Table 1b.
Heat Duty, MM kcal/h 1.11 1.11
The heat-transfer coefficient on the tubeside is
very low at 937 kcal/h-m2-°C, because the tubeside Allowable Pressure Drop, kg/cm 0.5 0.5
2

velocity is only 0.11 m/s. The allowable tubeside Viscosity In/Out, cP 1.85/3.3 3.02/2.49
pressure drop is hardly utilized, but it was impor- Density, kg/m 3 752/780 768/761
tant to have only two tube passes so as to have Thermal Conductivity In/Out, kcal/h-m-°C 0.0612/0.0632 0.1/0.1
pure countercurrent flow. Otherwise, the number of Specific Heat In/Out, kcal/kg-°C 0.739/0.719 0.7/0.7
shells and thereby the capital cost would be consid- Fouling Resistance, h-m-°C/kcal 0.001 0.001
erably higher. Because the stream is clean process Nominal Line Size, mm 100 300
condensate, there are no fouling implications. Material of Construction 5 Cr 1/2 Mo 5 Cr 1/2 Mo

CEP April 2004 www.cepmagazine.org 43

Heat Transfer

Case study 2: Low shellside flowrate passes is increased from two to four in order to increase the
Consider the heat exchanger service specified in Table heat-transfer coefficient, the tubeside pressure will exceed
2a. This exchanger transfers heat between two rather dirty the permissible value. Therefore, the tubeside velocity is
liquid streams (fouling resistance = 0.001 h-m2-°C/kcal) of constrained to be only 0.86 m/s, and as a result, the tube-
moderate viscosity. The tubeside flowrate is rather large, side heat-transfer coefficient is only 317.3 kcal/h-m2-°C.
and the shellside flowrate is very low. The low tubeside and shellside heat-transfer coeffi-
A first design was made with a single shell, as indicated cients produce an overall heat-transfer coefficient of only
in Table 2b. What stands out about this design is that the 85 kcal/h-m2-°C, and the heat-transfer area is unduly
shellside velocity is extremely low and the allowable shell- high at 533 m2.
side pressure drop has been very poorly utilized: while 0.5 Not only is the heat-transfer area high, the shellside ve-
kg/cm2 is permitted, only 0.023 kg/cm2 is consumed. As a locity is unacceptably low. As the shellside stream is dirty,
result of the low velocity, the shellside heat-transfer coeffi- this would lead to severe fouling problems.
cient is only 159 kcal/h-m2-°C. In order to improve the design by having a higher shellside
On the tubeside, too, the allowable pressure drop has not velocity, the use of two shells in series was considered. How-
been properly utilized, being only 0.144 kg/cm2 vs. the al- ever, this is a hopeless alternative, since the tubeside pressure
lowable 0.5 kg/cm2. This is because if the number of tube drop, which is already 0.144 kg/cm2 in the single-shell design,
becomes excessive.
Table 2b. Principal construction and performance parameters The next alternative consid-
for the liquid-liquid heat exchanger in case study 2. ered was interchanging the
two fluid sides, that is, to
E Shell Design, E Shell with Sides F Shell Design, route stream 1 through the
TEMA Type AES Interchanged, TEMA Type AFS tubeside and stream 2 through
TEMA Type AES the shellside. This design is
also depicted in Table 2b. The
Shell ID, mm 1,200 1,200 810 shell diameter was kept the
Heat-Transfer Area, m 2 533 447 233 same, at 1,200 mm. In order
Number of Tubes 1,446 1,210 630 to hamdle the much smaller
Number of Tube Passes 2 16 2
flowrate on the tubeside, the
Tube O.D. × Thickness × Length, mm 20 × 2 × 6,000 20 × 2 × 6,000 20 × 2 × 6,000
number of tube passes was in-
creased to 16. Accordingly,
Tube Pitch, mm 26 Rotated 26 Rotated 26 Rotated
the number of tubes de-
Square (45 deg) Square (45 deg) Square (45 deg)
creased from 1,446 to 1,210.
Baffle Spacing, mm 250 350 170
The baffle spacing and cut
Baffle Cut, % diameter 21 25 21 were increased to handle the
Tubeside Velocity, m/s 0.86 0.82 1.96 higher shellside flowrate.
Nominal Shellside Velocity, While the pressure drop on
Crossflow/Window, m/s 0.078/0.071 0.87/0.82 0.34/0.3 both the tubeside and shell-
Stream Fraction, Tube-to-Baffle Hole 0.128 0.185 0.293 side are within the specified
Stream Fraction, Main Crossflow 0.448 0.434 0.617 limits, the exchanger is still
Stream Fraction, Bundle-Shell 0.067 0.053 0.09 highly undersurfaced. It is
Stream Fraction, Baffle-to-Shell 0.313 0.168 0 clear that it will not be possi-
Stream Fraction, Pass Partition 0.044 0.16 0 ble to catch up with this un-
Shellside Heat-Transfer Coefficient, derdesign by increasing the
kcal/h-m -°C
2 159.3 975 449.9
size of the exchanger. The
Tubeside Heat-Transfer Coefficient,
tubeside film resistance is
highly controlling, and the
kcal/h-m2-°C 317.4 80.8 918.2
number of tube passes is al-
Overall Heat-Transfer Coefficient,
ready 16. This design was
kcal/h-m2-°C 85 63.6 177.3
therefore abandoned.
Shellside Pressure Drop, kg/cm2 0.023 0.44 0.31 Finally, an F shell was
Tubeside Pressure Drop, kg/cm2 0.144 0.45 0.495 considered, which is also
Mean Temperature Difference, °C 24.3 28.1 28.8 shown in Table 2b. The
Overdesign Nil –28.3 7.3 stream analysis for the F
Nominal Nozzle Size, Shellside, mm 100 100 100 shell design is far better
Nominal Nozzle Size, Tubeside, mm 300 300 300 than the original E shell de-

44 www.cepmagazine.org April 2004 CEP

sign. The allowable pressure drops are far better utilized
RAJIV
R MUKHERJEE
AJIV M UKHERJEE is a consultant in unfired heat transfer based in New
with much higher shellside and tubeside velocities,
Delhi, India (Phone: 91-11-25523018; Mobile: 98188-50476E-mail:
thereby leading to a much higher shellside heat-transfer rajiv.mukherjee@vsnl.com). He has over 32 years of experience in the
coefficient (449.9 kcal/h-m2-°C vs. 159.3 kcal/h-m2-°C) thermal design, revamping and troubleshooting of air-cooled and shell-
and a much higher tubeside heat-transfer coefficient and-tube heat exchangers and in the design of heat exchanger
networks. He has written several articles and presented many papers at
(918.2 kcal/h-m2-°C vs. 317.4 kcal/h-m2-°C). Conse-
technical symposia. He has also served as faculty for numerous courses
quently, the overall heat-transfer coefficient is far on heat exchanger design and operation, energy conservation, and heat
greater, 177.3 kcal/h-m2-°C vs. 85 kcal/h-m2-°C, and the exchanger networks, and presently teaches an intensive two-day
heat-transfer area was reduced from 533 m2 to 233 m2 refresher course in the design and operation of heat exchangers that
can be offered at a plant or office location. He is an honors graduate in
— a much less expensive design.
chemical engineering from Jadavpur Univ., Calcutta, India.

Final recommendations
In many services involving temperature cross and in be predicted but must be established during the design
numerous situations where the permitted shellside pres- stage. It would, therefore, be best if the process data
sure drop cannot be properly utilized with a single E sheets for heat exchangers in such services specify a
shell, F shells offer a more cost-effective design with choice in the shell style (E or F) and leave it to the heat
lower first cost and/or lower operating cost due to higher exchanger designer to make the final choice. Process li-
shellside velocity and lower fouling. Whether an F shell censors, engineering contractors and plant owners may all
design is going to be superior in these situations cannot work toward this eventuality. CEP

Americ an Institute Of Chemic al Engineers

www.cepmagazine.org or Circle No. 127

CEP April 2004 www.cepmagazine.org 45