Heat Exchangers

COMPACT HEAT EXCHANGERS — PART 1:

Designing
Plate-and-Frame
Heat Exchangers
Christopher Haslego,
Alfa Laval Use these design charts for preliminary sizing.
Graham Polley,
www.pinchtechnology.com

C onsidering available heat exchanger technolo-

gies at the outset of process design (at the
process synthesis stage) is not general prac-
tice. In fact, procedures established in some
companies preclude it. For instance, some purchasing de-
partments’ “nightmare” is having to deal with a single sup-
plier — instead they want a general specification that can
ed metal plates fitted between a thick, carbon-steel frame.
Each plate flow channel is sealed with a gasket, a weld or
an alternating combination of the two. It is not uncommon
for plate-and-frame heat exchangers to have overall heat-
transfer coefficients that are three to four times those found
in shell-and-tube heat exchangers.
This three-part series outlines the lost opportunities and
be sent to all equipment vendors, in the mistaken belief the importance of proper heat exchanger selection. This arti-
that they are then operating on a “level playing field.” cle discusses some general aspects of plate-and-frame heat
This omission is both unfortunate
and costly. It results in unnecessary cap-
ital expenditure and in reduced energy
efficiency. It also hinders the develop-
ment of energy saving technology.
Pinch analysis is the key tool used by
engineers to develop flowsheets of ener-
gy-intensive processes, where heat ex-
changer selection is particularly impor-
tant. Yet, this tool is hindering the adop-
tion of a more-progressive approach be-
cause of the way it is restricted to tradi-
tional heat exchangers.
Numerous articles have been pub-
lished regarding the advantages of com-
pact heat exchangers. Briefly, their
higher heat-transfer coefficients, com-
pact size, cost effectiveness, and unique
ability to handle fouling fluids make
them a good choice for many services.
A plate-and-frame heat exchanger
(Figure 1) consists of pressed, corrugat- ■ Figure 1. Cutaway drawing of a plate-and-frame heat exchanger.

32 www.cepmagazine.org September 2002 CEP

 Table. Basis for heat exchanger quotation.
temperature is 115°F.” This simple statement could result
in vastly different configurations compared with the de-
Tubeside Shellside signs that would result from the original specification.
Flowrate, gal/min 500 1,800
Temperature In, °F 280 80 Design charts for plate-and-frame exchangers
Temperature Out, °F 150 92 When it comes to compact heat-transfer technology,
Allowable Pressure Drop, psi 15 15 engineers often find themselves at the mercy of the
equipment manufacturers. For example, limited litera-
exchangers, outlines a procedure for accurately estimating ture correlations are available to help in the preliminary
the required area, and shows how these units can be used to design of plate-and-frame heat exchangers.
simplify processes. Part 2 (which will appear next month) This article introduces a series of charts (Figures 2–7)
covers integrating plate-and-frame exchangers (and other that can be used for performing preliminary sizing of plate-
compact technologies) into pinch analysis for new plants, and-frame exchangers. Examples will help clarify their use.
while Part 3 (which also will appear next month) deals with The following important points should be noted regard-
the application of plate-and-frame exchangers and pinch ing the charts and their use:
technology to retrofits. 1. The heat-transfer correlations apply to single-phase,
liquid-liquid designs.
Specifying plate-and-frame heat exchangers 2. These charts are valid for single-pass units with 0.50-
Engineers often fail to realize the differences between mm-thick plates. The accuracy of the charts will not be
heat transfer technologies when preparing a specification to compromised for most materials of construction.
be sent to vendors of different types of heat exchangers. 3. Wetted-material thermal conductivity is taken as
Consider the following example. 8.67 Btu/h-ft-°F (which is the value for stainless steel).
A process stream needs to be cooled with cooling water 4. The following physical properties for hydrocarbon-
before being sent to storage. The stream requires C276, an based fluids were used for the basis: thermal conductivity
expensive high-nickel alloy, to guard against corrosion; this (k) = 0.06 Btu/h-ft-°F, density (ρ) = 55.0 lb/ft3, heat capac-
metallurgy makes the stream a candidate for the tubeside of ity (Cp) = 0.85 Btu/lb-°F. The following physical properties
a shell-and-tube heat exchanger. The cooling water is avail- for water-based fluids were used for the basis: thermal con-
able at 80°F and must be returned at a temperature no high- ductivity = 0.33 Btu/h-ft-°F, density = 62.0 lb/ft3, heat ca-
er than 115°F. The process engineer realizes that with the pacity = 0.85 Btu/lb-°F.
water flow being placed on the shellside, larger flowrates 5. Accuracy should be within ±15% of the service value
will enhance the heat-transfer coefficient. The basis for the for the overall heat-transfer coefficient, assuming a nomi-
heat exchanger quotation was specified as shown in the nal 10% excess heat-transfer area.
table. According to the engineer’s calculations, these basic 6. For fluids with viscosities between 100 and 500 cP,
parameters should result in a good shell-and-tube design use the 100 cP line on the graphs. For fluids in excess of
that uses a minimum amount of C276 material. 500 cP, consult equipment manufacturers.
A typical plate-and-frame exchanger designed to meet the Equations 1–3 are used to calculate the log-mean tem-
specification would have about 650 ft2 of area, compared to perature difference (LMTD) and number of transfer units
about 420 ft2 for a shell-and-tube exchanger. A plate-and- (NTU) for the hot and cold streams. After the local heat-
frame unit designed to the above specification is limited by transfer coefficients (h) are read from the charts, the over-
the allowable pressure drop on the cooling water. If the cool- all heat-transfer coefficient (U) is calculated by Eq. 4.
ing water flow is reduced to 655 gal/min and the outlet water
temperature allowed to rise to 115°F, the plate-and-frame
LMTD =
(T hot ,in ) (
− Tcold ,out − Thot ,out − Tcold ,in ) (1)
heat exchanger would contain about 185 ft2 of area. The unit T − Tcold ,out 
is smaller and less expensive, and it uses less water. The ln hot ,in 
load being transferred to the cooling tower is the same.  Thot ,out − Tcold ,in 
With shell-and-tube heat exchangers, increasing water
Thot ,in − Thot ,out
flow will minimize heat-transfer area. However, with com- NTUhot = (2 )
pact technologies, the effect is exactly the opposite. The LMTD
larger water flow actually drives up the cost of the unit.
Rather than supplying a rigid specification to all heat Tcold ,out − Tcold ,in
exchanger manufacturers, the engineer should have ex- NTUcold = (3)
LMTD
plained the goal for the process stream. This could have
been in the form of the following statement: “The process 1 1 ∆x 1
stream is to be cooled with cooling water. Up to 2,000 = + + ( 4)
gal/min of water is available at 80°F. The maximum return U hhot k hcold

CEP September 2002 www.cepmagazine.org 33

Heat Exchangers

Pressure Drop, kPa

50 100 150 200
h = Local Heat-Transfer Coefficient,

h = Local Heat-Transfer Coefficient,

3,500 Using the charts
Water-Based Fluids 18,000 Consider the following example.
3,000 1 cP
150,000 lb/h of water is being cooled
15,000
2,500 from 200°F to 175°F by 75,000 lb/h
Btu/h-ft2-˚F

12,000 of SAE 30 oil. The oil enters the ex-

W/m2-K
2,000 2.5 cP changer at 60°F and leaves at 168°F.
9,000 The average viscosity of the water
1,500
5 cP passing through the unit is 0.33 cP
1,000 6,000
10 cP
and the average viscosity of the oil in
3,000 the unit is 215 cP. The maximum-
500 100 cP allowable pressure drop through the
0 plate heat exchanger is 15 psi on the
5 10 15 20 25 30
hot and cold sides.
Pressure Drop, psi Step 1: Calculate the LMTD.
From Eq. 1, LMTD = [(200 – 168)
■ Figure 2. Heat-transfer correlations for water-based fluids, 0.25 < NTU < 2.0. – (175 – 60)]/ln[(200 – 168)/(175 –
60)] = 64.9°F.
Pressure Drop, kPa
Step 2: Calculate NTUhot and
NTUcold. From Eqs. 2 and 3, NTUhot =
50 100 150 200 (200 – 175)/64.9 = 0.38 and NTUcold
h = Local Heat-Transfer Coefficient,

h = Local Heat-Transfer Coefficient,

3,500
Correction: For liquids with average viscosities
18,000 = (168 – 60)/64.9 = 1.66.
3,000 less than 2.0 cP, reduce the local heat-transfer Step 3: Read hhot from the appro-
coefficient by 15% from 3.5 < NTU < 4.0.
1 cP 15,000 priate chart. Use Figure 5, the chart
2,500 Water-Based Fluids
for hydrocarbons when 0.25 < NTU
Btu/h-ft2-˚F

12,000
W/m2-K
2,000 2.5 cP < 2.0. Although there is not a vis-
9,000
cosity line for 215 cP, the line repre-
1,500 senting 100 cP can be used for vis-
5 cP
1,000 6,000 cosities up to about 400–500 cP. The
10 cP
heat exchanger will be pressure-
500 3,000 drop-limited and the heat-transfer
100 cP
coefficient will not change apprecia-
0
5 10 15 20 25 30 bly over this viscosity range for
plate-and-frame exchangers. Read-
Pressure Drop, psi
ing from the chart, a pressure drop
of 15 psi corresponds to hhot ≈ 50
■ Figure 3. Heat-transfer correlations for water-based fluids, 2.0 < NTU < 4.0. Btu/h-ft2-°F.
Step 4: Read hcold from the chart.
Pressure Drop, kPa Use Figure 2, which applies to water-
based liquids when 0.25 < NTU < 2.0.
50 100 150 200
Again, the exact viscosity line needed
h = Local Heat-Transfer Coefficient,

h = Local Heat-Transfer Coefficient,

3,500
18,000 for pure water (0.33 cP) in this case is
3,000 Water-Based Fluids not available. However, the 1.0 cP
15,000 line provides a very good estimate of
2,500 1 cP the heat-transfer coefficient for pure
Btu/h-ft2-˚F

12,000
W/m2-K

2,000 water. Reading from the chart, a pres-

2.5 cP 9,000 sure drop of 15 psi corresponds to
1,500
5 cP hcold ≈ 3,000 Btu/h-ft2-°F.
1,000 6,000 Step 5: Calculate U. Assume a
10 cP stainless steel plate with a thick-
3,000
500
100 cP
ness of 0.50 mm is being used.
0
Type 316 stainless steel has a ther-
5 10 15 20 25 30 mal conductivity of 8.67 Btu/h-ft-
°F. Then from Eq. 4, 1/U = (1/50 +
Pressure Drop, psi
0.000189 + 1/3,000) and U = 49
Btu/h-ft2-°F.
■ Figure 4. Heat-transfer correlations for water-based fluids, 4.0 < NTU < 5.0.

34 www.cepmagazine.org September 2002 CEP

 Pressure Drop, kPa
50 100 150 200

h = Local Heat-Transfer Coefficient,

h = Local Heat-Transfer Coefficient,

Now let’s consider another exam- 1,000
5,400
ple. 150,000 lb/h of water is being Hydrocarbon-Based Fluids
cooled from 200°F to 100°F by 800 1 cP 4,600
150,000 lb/h of NaCl brine. The 3,800

Btu/h-ft2-˚F
brine enters the exchanger at 50°F 600

W/m2-K
and leaves at 171°F. The average 3,000
2.5 cP
viscosity of the water passing 400 2,400
through the unit is 0.46 cP and the 5 cP
average viscosity of the brine in the 1,600
200 10 cP
unit is 1.10 cP. The maximum-allow- 800
able pressure drop through the plate 100 cP
heat exchanger is 10 psi on the hot 0
5 10 15 20 25 30
(water) side and 20 psi on the cold
(brine) side. Pressure Drop, psi
The LMTD is calculated to be
38.5°F. NTUhot and NTUcold are 2.59 ■ Figure 5. Heat-transfer correlations for hydrocarbons, 0.25 < NTU < 2.0.
and 3.14, respectively. From the
charts for 2.0 < NTU < 4.0 (water
based), hhot ≈ 2,000 Btu/h-ft2-°F Pressure Drop, kPa
and hcold ≈ 2,500 Btu/h-ft2-°F. Al- 50 100 150 200
h = Local Heat-Transfer Coefficient,

1,000

h = Local Heat-Transfer Coefficient,

though the material of choice may Correction: For liquids with average viscosities 5,400
be titanium or palladium-stabilized less than 2.0 cP, reduce the local heat-transfer
titanium, the properties for stain- 800 coefficient by 15% from 3.5 < NTU < 4.0. 4,600
less steel are used for preliminary Hydrocarbon-Based Fluids
3,800
Btu/h-ft2-˚F

1 cP
sizing. U is calculated to be 918 600

W/m2-K
2 3,000
Btu/h-ft -°F.
2.5 cP 2,400
400
Implications of size reduction
5 cP 1,600
Alternative technologies offer 10 cP
200
significant size advantages over 800
shell-and-tube heat exchangers. 100 cP

Let’s now consider the implications 0

5 10 15 20 25 30
of this.
The individual exchangers are Pressure Drop, psi
smaller, and the spacing between
process equipment can be reduced. ■ Figure 6. Heat-transfer correlations for hydrocarbons, 2.0 < NTU < 4.0.
Thus, a smaller plot is needed for
the process plant. If the plant is to Pressure Drop, kPa
be housed in a building, the build-
50 100 150 200
ing can be smaller. The amount of
h = Local Heat-Transfer Coefficient,

h = Local Heat-Transfer Coefficient,

1,000
structural steel used to support the 5,400
plant can be reduced, and because 800 4,600
of the weight saving, the load on Hydrocarbon-Based Fluids

that structure is also reduced. The 3,800

Btu/h-ft2-˚F

1 cP
600
W/m2-K

weight advantage extends to the 3,000

design of the foundations used to
support the plant. Since the spac- 400 2.5 cP 2,400
ing between equipment is reduced, 5 cP 1,600
piping costs are lower. 200 10 cP
However, we stress again that the 800
100 cP
savings associated with size and
0
weight reduction can only be 5 10 15 20 25 30
achieved if these advantages are rec-
Pressure Drop, psi
ognized and exploited at the earliest
stages of the plant design.
■ Figure 7. Heat-transfer correlations for hydrocarbons, 4.0 < NTU < 5.0.

CEP September 2002 www.cepmagazine.org 35

Heat Exchangers

Reduced plant complexity

The use of alternative heat ex-
One shell-and-tube exchanger Three shell-and-tube exchangers changer technologies can signifi-
One plate exchanger One plate exchanger cantly reduce plant complexity by
200˚F 200˚F reducing the number of heat ex-
changers through improved thermal
175˚F contacting and multi-streaming.
168˚F 171˚F This adds to the savings associated
with reduced size and weight and
100˚F also has safety implications. The
simpler the plant structure, the easi-
60˚F 50˚F er it is for the process operator to
understand the plant. In addition,
No Temperature Cross Temperature Cross
plant maintenance will be safer,
easier and more straight-forward.
Mechanical constraints play a
■ Figure 8. The situation on the left does not involve a temperature cross, while the one on the right does.
significant role in the design of
shell-and-tube heat exchangers. For
instance, it is common to find that
some users place restrictions on
160˚F 160˚F tube length. Such a restriction can
Styrene, 290,285 lb/h 112˚F have important implications for the
130˚F ∆P = 9.5 psi
design. In the case of exchangers
Water, 248,655 lb/h requiring large surface areas, the
248˚F
∆P = 14.6 psi 80 ˚ F restriction drives the design toward
large tube counts. If such large tube
Styrene, 290,285 lb/h Styrene, 100˚F counts lead to low tubeside veloci-
∆Ptotal = 30 psi ty, the designer is tempted to in-
285˚F
crease the number of tubeside pass-
es in order to maintain a reasonable
■ Figure 9. A single plate-and-frame exchanger handles three process streams … tubeside heat-transfer coefficient.
Thermal expansion considerations
can also lead the designer to opt for
multiple tube passes, because the cost of a floating head
is generally lower than the cost of installing an expan-
Cooling Styrene sion bellows in the exchanger shell.
Water (Hot)
The use of multiple tube passes has four detrimental
effects. First, it leads to a reduction in the number of
tubes that can be accommodated in a given size shell,
thereby leading to increased shell diameter and cost.
Second, for bundles having more than four tube passes,
the pass-partition lanes introduced into the bundle give
rise to an increase in the quantity of shellside fluid by-
passing the tube bundle and a reduction in shellside
heat-transfer coefficient. Third, it results in wasted
tubeside pressure drop in the return headers. Finally,
and most significantly, the use of multiple tube passes
Styrene results in the thermal contacting of the streams not
(Cold) being pure counter-flow, which reduces the effective-
mean-temperature driving force and possibly produces
a temperature cross (i.e., where the outlet temperature
of the cold stream is higher than the inlet temperature
■ Figure 10. … whereas the equivalent shell-and-tube design requires of the hot stream, as shown in Figure 8). If a tempera-
six units. ture cross occurs, the designer must split the duty be-
tween multiple heat exchangers arranged in series.

36 www.cepmagazine.org September 2002 CEP

 Many of the alternative heat-exchanger technologies tube design has 11,344 ft2 of surface area distributed
allow the application of pure countercurrent flow across across six separate exchangers.
all size and flow ranges. This results in better use of the
available temperature driving force and the use of single Budget pricing correlations
heat exchangers. For plate-and-frame heat exchangers with a design pres-
sure up to 150 psi and a design temperature up to 320°F,
Multi-streaming the following cost equations can be used to estimate the
The traditional shell-and-tube heat exchanger handles purchased cost.
only one hot stream and one cold stream. Some heat ex- For areas (A) less than 200 ft2:
changer technologies (most notably plate-fin and print- C = 401 A 0.4887 for Type 316 stainless steel (5)
ed-circuit exchangers) can handle many streams. It is C = 612 A 0.4631 for Grade 1 titanium (6)
not uncommon to find plate-fin exchangers transferring For areas larger than 200 ft2:
heat between ten individual processes. (The principles C = 136 A 0.6907 for Type 316 stainless steel (7)
behind the design of multi-stream exchangers and the C = 131 A 0.7514 for Grade 1 titanium (8)
operability of such units are discussed in Ref. 1.) Such Typical installation factors for plate-and-frame heat ex-
units can be considered to contain a whole heat ex- changers can range from 1.5 to 2.0, depending on the size
changer network within the body of a single exchanger. of the unit.
Distribution and recombination of process flows is un-
dertaken inside the exchanger. The result is a major re- Up next
duction in piping cost. Being able to estimate the area and prices of plate-
Engineers often overlook the opportunities of using a and-frame heat exchangers is an important first step in
plate-and-frame exchanger as a multi-stream unit. As including alternative heat-transfer technology in process
mentioned earlier, this is a common oversight when ex- synthesis. The remaining articles in this series (which
changer selection is not made until after the flowsheet will appear next month) focus on the integration of
has been developed. plate-and-frame (and other compact technology) into
A good example of multi-streaming is a plate heat ex- pinch analyses. CEP

changer that serves as a process interchanger on one side

and a trim cooler on the other. This arrangement is par-
ticularly useful for product streams that are exiting a pro-
cess and must be cooled for storage. CHRISTOPHER HASLEGO is a design engineer with the Process
Another popular function of multi-streaming is to Technology Div. of Alfa Laval (5400 International Trade Dr., Richmond,
lower material costs. Some streams, once they are cooled VA 23236; Phone: (804) 236-1318; Fax: (804) 236-1360; E-mail:
to a certain temperature, pose much less of a corrosion [email protected]). His focus is heat transfer in the
inorganic base chemicals industry. He received a BS in chemical
risk. One side of the exchanger can be made of a more- engineering from West Virginia Univ. A member of AIChE, he also
expensive corrosion-resistant alloy while the other side maintains a website dedicated to chemical engineering entitled “The
can utilize stainless steel or a lower alloy. Chemical Engineers’ Resource Page” at www.cheresources.com.
Figure 9 shows a plate-and-frame unit applied to three
GRAHAM T. POLLEY (Phone: +44-1229-585-330; Fax: +44-1229-585-708;
process streams. A single exchanger with 1,335 ft2 of ef- E-mail: [email protected]) is retired from the Univ. of
fective surface area is used. Figure 10 is the equivalent Manchester Institute of Science and Technology, Manchester, U.K.,
shell-and-tube solution — to avoid temperature crosses, where he was the director of the Centre for Process Integration. To
six individual exchangers are needed: the cooler having promote the application of process integration technology, he has
developed the website www.pinchtechnology.com. His research
two shells in series (each with 1,440 ft2 of effective sur- activities have involved condensation heat transfer, boiling heat
face area) and the heat recovery unit having four shells in transfer, heat-recovery-system design and the development of process
series (each with 2,116 ft2 of surface area). So, the plate- and equipment design methodologies. He is the current president of
and-frame design involves the use of 1,335 ft2 of surface the U.K. Heat Transfer Society. In 1990, he was awarded the Moulton
Medal by the Institution of Chemical Engineers for his work on oil
area in a single unit, whereas the equivalent shell-and- refinery revamping, and in 1992, he and Dr. Nasr were awarded the
Ackrill Trophy by the U.K. Heat Transfer Society for their work on
heat-transfer enhancement. He holds BTech, MSc and PhD degrees
in chemical engineering from Loughborough Univ. of Technology,
Literature Cited and is a member of AIChE.

1. Picon Nunez, M., and G. T. Polley, “Recent Advances in

Analysis of Heat Transfer for Fin Type Surfaces,” Chapters 9 Correlations used to formulate charts were developed by Alfa Laval
and 10, Sunden, B., and P. J. Heggs, eds., WIT Press, (formerly Alfa Laval Thermal) through comprehensive research and
Southampton, U.K. (2000). development and decades of industrial experience.

CEP September 2002 www.cepmagazine.org 37