You are on page 1of 6

ENERGY SAVINGS BY USE OF DISKIDONUT BAFFLING IN

TUBULAR HEAT EXCHANGERS


J. J. Tabarek, Dale E. Klein and James R. Fair
Centerfor Energy Studies
The University afTexas at Austin
Austin, Texas 78712
ABSTRACT
The use of segmental baffling in shell-and-tube heat
exchangers has become standard design practice, and process
heat transfer specialists recognize that such baffling is the basis
for most design procedures. Data from earlier work at The
University of Texas indicate that disk/donut baffling can provide
equivalent heat transfer rates at significantly reduced pressure
drops. In this paper estimates ar made of the potential for
energy savings by this change of baffle arrangement, and,
methods are provided for executing designs that involve
disk/donut features. '
INTRODUCrJON
The ubiquitous &hell-and-tube heat exchanger represents
about 75% of the heal transfer devices found in the chemical and
petroleum processing industries. In comparison with other
devices its easy adaptability to wide ranges of operating
pressures, temperatures and pressure drops gives it superiority
for many applications, despite its relatively-high volume to
exchange surface ratio.
The design features of the shell-and-tube device have
become fairly standardized, and over the past four decades have
undergone few notable changes. Typically, the device has
baffling on the shell side, and this baffling is of the segmental
type. In this paper, a proposed departure from the standard
design is proposed: the use of disk and donut (or disk and ring)
baffling on the shell side. Actually, this arrangement was used
40 to 50 years ago but, for reasons which do not appear valid by
present knowledge, it was later abandoned in the United States.
As a result of this history, no reliable methods presently exist for
predicting the heat transfer performance of shell-and-tube
exchangers containing disk/donut baffles.
The prospective advantages of the disk/donut baffle
arrangement, compared to the more standard segmental baffle
arrangement, include the following:
1. The complete absence of a bundle-to-shell crossflow
bypas stmlm
2. Minimization of the ineffective baffle-to-shell leakage
stream because of the much lower driving force for that
stream
3. Radial orientation of flow into the tube field, with the,
flow exhibiting much greater uniformity and a minimum'
of dead zones (i.e., an arrangement that improves the
flow effectiveness and minimizes fouling)
4. Higher effectiveness of pressure drop to heat transfer
conversion than for the segmental baffle arrangement,
with the result of less energy demand for a given heat
duty
5. Accommodati n of a "no-tubes-in-window" type layout,
thus assuring a tube vibration-free design.
The potential economic advantage of a disk/donut baffle
arrangement is significant. For 1986, worldwide spending on
131
heat exchangers by the hydrocarbon-processing industry alone
has been estimated at $1.23 billion (McKetta, 1987). The same
source shows that for the United States the spending was
estimated to be $350 million. Recognizing that a portion of the
non-US amount was scheduled for use in the United States, and
recognizing also that the heat exchanger needs of the chemical
processing industry should also be considered, it is reasonable
to use a conservative figure of $800 million as the 1986
spending on heat exchanger equipment, and this amount does
not include furnaces, fired heaters, and the like. In fact, it is
safe to assume that the $800 million figure can represent shell
and-tube heat exchangers.
Based on the work of Taborek and Sharif (1987), use of
disk/donut baffles instead of segmental baffles has the potential
to reduce shell-side pressure drop by about 2 Ibs/in
2
for small
heat exchangers. When one considers the average size of
commercial exchangers and the energy required to overcome
flow resistance on their shell side, the potential energy savings is
quite high. Accordingly, it appears appropriate to revive the use
of disk/donut baffling arrangements.
WORKING PRINCIPLES OF SHELL-AND-TUBE HEAT
EXCHANGERS
For general orientation, a cross-sectional view of a
typical sheU-and-tube heat exchanger is shown in Figure 1.
I. S",uanasy Head--Bonnel
2. S",uanasy Head f1ange--Channel or Bonnet
J. Stauanasy Head Naule
4. S",uanary Tubesheet
5. Tubes
6. Shell
7. Shell Nozzle
8. Expansion Joint
9. Tierods and Sp"c....,
10. Transv""e Barnes or Support Pia...
II. Vent Connection
12. Drain Connection
11. IJls01Jment Connection
14. Support Bracke.
Figure 1. ConslrUction elements of a single tube pan shell-andwbe
heal exchanger.
ESL-IE-88-09-26
Proceedings from the Tenth Annual Industrial Energy Technology Conference, Houston, TX, September 13-15, 1988
This exchanger has a fixed tube sheet, shell expansion joint, and
single tube pass. The flow through the tube bundle with seg
mental baffles is complex and highly nonuniform, as shown
schematically in Figure 2.
Figure 2. Schematic flow distribution diagram for
baffled shell-side flow.
Fundamental understanding of flow through segmental baffles is
essential if one is to evaluate the benefits offered by the
disk/donut baffle arrangement and is therefore presented here as
a brief summary.
The early analysis of shell side flow is due to Tinker
(1958), who suggested a schematic flow pattern as shown in
Figure 2. The shell side flow is considered as divided into a
number of individual streams. Stream A is the leakage stream
in the opening formed by the clearance between the baffle tube
hole and the tube wall (see Figure 3.)
Figure 3. Schematic piclUre of lubelo-baffle
leakage stream A.
Stream B is the main effective crossflow stream, which can be
related to the flow across an ideal tube bank. Stream e is the
tube bundle bypass stream in the gap between the bundle and the
shell wall. Stream E is the leakage stream between the baffle
edge and the shell wall (see Figure 4).
Stream F is the bypass stream in flow channels due to omission
of tubes in the tube pass partitions. This stream has been added
to the original Tinker model by Palen and Taborek (1969). It
behaves similarly to Stream e, but is present only in some tube
layouts.
The basic principle of Tinker's original analysis is that
the ure drop of the crossflow Stream B and the window
turn act as a driving force for the other streams, forcing part of
the flow through the leakage and bypass clearances. For a better
understanding of the physical mechanisms, the flow distribution
model of Figure 2 was reduced by Palen and Taborek to an
equivalent nodal diagram as shown in Figure 5, with the flow
streams represented by arrows and the above-defmed letters.
Figure 5. Schematic model of shell-side flow branches
and their resislances acr.oss one baffle space.
The resistances to flow for each stream are represented by valves
and by the letter K with appropriate subscript
The crossflow, bypass, and pass-partition streams (B,
e, F) are shown flowing in parallel across each baffle space,
joining in the windows and redividing at the next baffled space.
The tube baffle and baffle shell leakage streams (A, E) are
shown flowing in parallel from a hypothetical nodal point in one
baffle space to a corresponding point in the next. Once this
model is accepted, well-known piping network flow techniques
can be applied to solve for the fow rate of each stream in terms
of the cross-sectional area for flow and the flow resistance.
The solution introduced by Palen and Taborek is termed
the "stream analysis method" and has become a generally
accepted method for solving the shell-side flow distribution.
The flow rates are given implicitly by the pressure drop
equation:
(I)
where Mj = flow rate of any stream j (A, B, e, E, F)
=corresponding flow area
Kj = flow resistance of streamj, particular to
any flow channel geometry and Reynolds
number; Kj = f(Mj)
The pressure drop relations may be expressed as follows:
4JC=4JF=4JB=4JX (2)
4J A = 4J E= 4J x = 4J w (3)
where 4J A and 4J w are the crossflow and window pressure
drops shown in Figure 5.
The final equation required is the mass balance of
the total flow rate M:
Figure 4. Schematic picture of shellto-baffle
leakage stream E.
(4)
132
ESL-IE-88-09-26
Proceedings from the Tenth Annual Industrial Energy Technology Conference, Houston, TX, September 13-15, 1988
Once the representative cross-sectional flow areas are
defined, Equations 1-4 can be solved for the flow rates, if the
flow resistance terms Kj are known. The solution is somewhat
simplified by rearranging the pressure drop equations into the
following forms:
(5)
for} =8, C,F
(6)
for} =A or E (7)
4?w
z (8)
- 4?x
M
j
= (FAlM)
(9)
Where F = flow fraction of the total stream
OJ = relative flow stream value, introduced for
simplification of the algebraic manipulations
tJpw, tJpx = baffle window and crossflow pressure drops,
respecti vely.
Since the K} values are functions of the Reynolds
numbers, and therefore of the individual stream flow rates Mj or
flow fractions Fj, the solution of the above set of equations is an
iterative one.
It is of interest to visualize the typical flow fractions of
the various streams. For a well-designed geometry for a seg
mental baffle configuration (baffle spacing/baffle cut, normal
industrial clearances, bypass streams blocked by sealing strips
whenever necessary), the flow fractions calculated for industrial
size exchangers from the Heat Transfer Research, Inc. (HTRI)
data bank are shown in Table I.
Table 1
FLOW fRACTIONS fOR A. TYPICAL INDUSTRIAL HEAT EXCHANGER
(Segmental Baffles)
Turbulent Lamlna r
St ream
Main crossflow 18) 0.40-0.70 0.25-0.50
Bundle bypass Ie + FI 0.15-0.20 0.20-0.30
Shell-to-baffle leakage lEI 0.06-0.20 0.06-0.40
Tube-to-ha.Cfle leakage (III 0.09-0.20 0.04-0.10
From this analysis, the following observations can be
made:
The most effective portion of the total flow, the
B-stream (ideal tube bank), is typically only
about 60%, even if designers take every
precaution (e.g., sealing strips, tight clear
ances) to minimize the parasitic streams.
The driving force for the parasitic streams is
high for segmental baffles, consisting of twice
the crossflow over the tube bundle between the
nodal points of Figure 5 plus the window
turnaround. These losses occur in the rela
tively small area of the segmental window
resulting from seve're con traction and
expansion. Typically the pressure drop
through the window of segmental baffles
accounts for some 50% of the overall value.
As a result of these phenomena, the overall effectiveness
of segmental baffles with respect to pressure drop-to-heat
transfer conversion is rather low, compared to ideal tube banks.
It should be noted that other baffle types may have better
pressure drop conversion effectiveness, but may also possess a
number of disadvantages or inherent limitations. For example,
double segmental baffles have these disadvantages:
Severe maldistribution eXists at the entry and
exit baffle, as discussed by Mueller and Chiou
(1987). The fact that flow effectiveness is very
low in the entry regions makes this baffle
arrangement unacceptable for short units with
few baffles.
Omission of tubes due to impingement plates is
particularly difficult to deal with.
Maldistribution can result in premature tube
vibration. (Halle, et al., 1986).
It is clear that presently-used baffle systems are far from perfect
and that ample room exists for new developments and
improvements.
DISKIDONUI BAFFLE SYSTEM
1, Historical The disk/donut baffles to
be considered are similar to those used occasionally in the
1940's and 1950's. The first edition of The Tubular Exchanger
Manufacturers Association (FEMA) Standards (1941) indicates a
disk/donut baffle arrangement as typical. The original experi
mental data base for this baffle arrangement resulted from
experiments conducted at The University of Texas at Austin in
the late 1930's by Professor B. E. Short and is documented in a
1943 bulletin (Short. 1943). The data series included the study
of segmental and disk/donut baffles of identical tube bundles;
results are shown in Table 2.
Table 2
DIMENSIONS OF TEST EXCHl\NGER Of SHORT
Tube Tube P 1 tch No. of No. of Disk Ring
Diam. Pitch Ratio Tubes Baffles Diam. Diam.
lin. ) lin .) lin. ) (in. }
3/8 11/16 1.833 52 19,11 4.5
1/2 25/32 1. 56 40 19. 11, 7 4.5 4.0'
1/2 25/32 1. 56 40 19, 11 4.95 J.5'
1/2 25/32 1. 56 40 19, 11 5.5 2. 5
.. Disk/donut only
Principal Dimensions of Test Geometry
Shell: 6.06 in. diameter, 60 in. long
Baffles: Segmental: 21' cut
DiSk/donut: as 8pecified below
Baffle Spacing: 19 baffles: 2.33 in.
11 baffie,: 4.24 in.
1 baffles: 1.11 in.
The Short work represents the only known published source of
heat transfer data for disk/donut baffling.
133
4.0
ESL-IE-88-09-26
Proceedings from the Tenth Annual Industrial Energy Technology Conference, Houston, TX, September 13-15, 1988
On the basis of Short's data, Donohue developed a
formal calculation method which remains the only presentation
on this subject (Donohue, 1949). He recognized the advantages
of disk/donut baffles:
It is interesting to compare disk and donut baffles with
segmental baffles from the viewpoint of heat exchanger
design where the rate of fluid flow and allowable
pressure drop are specified....lt was found that when 19
disk and donut baffles were used the pressure drop was
the same as when 11 segmental baffles were used, for a
given rate offlow; however, the (overall) coefficient of
heat transfer obtained with disk and donut baffles was
approximately 15% greater than that obtained with
segmental baffles....The lower values for the coefficient
ofheat transfer obtained with segmental baffles seems to
indicate that some of the kinetic energy of the fluid is
dissipated in eddy motions ocurring in the pockets in the
flow path.
The reasons why disk/donut baffles have rarely been
used in the United States are not clear. The fact that Kern
(1950) barely mentioned such baffles in his classic text, and
included no method of calculation, could be a contributing
factor. Later research at the University of Delaware (Bell, 1953)
and work at fITRI (palen and Taborek, 1969), concentrated on
segmental baffles only. Thus, no dependable methods evolved
for the computer-aided design of heat exchangers with
disk/donut baffles, and accordingly specification of such baffles
in the United States virtually disappeard. On the other hand,
these baffles remained popular in Europe (including Russia),
even though the design methods are still based mainly on the old
Donohue model (1949).
The data obtained by Short are not sufficient for the
development of generalized correlations. Among other
deficiencies, rather nontypical geometries were used, especially
with respect to tube layouts. The data are of high quality,
however, and observations on the effectiveness of the various
baffle types can be derived. A recent broader study by Taborek
and Sharif (1987) on the effectiveness of pressure drop/heat
transfer conversion for a variety of shell-side baffling and flow
arrangements provides some insight into the potential usefulness
of disk/donut baffling.
2. Construction and Operational Principles A sketch
of a typical layout of disk/donut baffles in a shell-and-tube heat
exchanger is shown in Figure 6.
Schematic flow pattern in disk/donul heal
exchanger: a) upper. standard; b) lower,
with no tubes in windows.
Figure 6.
The ring baffle has an outer diameter Db (subject to the same
tolerances as segmental baffles), an inner ring diameter Dr, and a
disk baffle with diameter Dd. The flow penetrates the tubular
field symmetrically in the radial direction.
The tube field in pure crossflow is between Dd and Dr
and has a mean diameter Dr = (Dd + Dr)/2. The mean
developed length of the tube field is typically about twice the
shell diameter, which results in a large crossflow area of
relatively shallow depth. This arrangement, in tum, permits the
use of relatively close baffle spacing for a uniform flow
distribution, without undue increase of pressure drop, as noted
by Donohue (1949).
The pressure drop can be modified by both the baffle
spacing and the baffle edge overlap (Dd - Dr), For assembly
purposes, the baffle edge overlap must have a minimum of one
tube row common to both baffles, but it can be as large as the
turnaround at the inner and outer flow passage will permit
without causing undue pressure drop. The earlier designs of the
disk/donut baffles were usually low pressure drop devices with
very small baffle overlaps. The proposed system would not be
restricted by this criterion.
The layout shown in Figure 6 includes a circumferential
flow distributor, demonstrating the completely symmetrical flow
entry. This plan is particularly important for the prevention of
tube vibration. The shell wall opposite the nozzle serves as
impingement protection. However, any other standard nozzle
entry is possible.
A variation of the design would omit tubes fro!Tl the inner
ring and from the outer annulus, permitting full support of every
tube at each baffle spacing, as shown in the lower part of Figure
6. If flow distributors at the inlet and outlet are used, potential
local bypass streams are also eliminated. Thus, the disk/donut
baffle permits a virtually tube-vibration-free design with high
flow efficiency and less tube omission than the "no-tubes-in
window" segmental baffle layout
3 Potential Benefits. The characteristics and advan
tages of the flow in a disk/donut baffle system, as compared to
the segmental baffle, are summarized below:
1. There is no bypass stream between the tube bundle
and shell.
2. Only every second ring baffle has a parasitic baffle
to-shell leakage stream, while the disk baffle has
none.
3. The tube field in crossflow is long but shallow;
the fore, the pressure drop differential between
baffles is small. Accordingly, the driving force that
causes the leakage stream between the baffle and
shell, and between the tube and baffle hole, is
minimized. As a result, the effectiveness of the tube
bundle with disk/donut baffles approaches that of an
ideal tube bank, unlike the case with segmental
baffles, where the effective flow fraction is typically
less than 60%.
4. The fully symm trical radial flow sweeps the entire
tube field unifonnly. avoiding areas of low flow
("dead zones") that arc typical for segmental baffles.
This flow unifonnity together with the minimum
effect of bypass and leakage streams is the main
reason for the high effectiveness of pressure drop to
heat transfer conversion.
5. The temperature profile distortion is minimized
because of the absence of the bypass and leakage
str ams. .
6. As a byproduct of the flow uniformity, it can be
speculated that preferential fouling deposition in the
"dead" baffle areas will be minimized.
134
ESL-IE-88-09-26
Proceedings from the Tenth Annual Industrial Energy Technology Conference, Houston, TX, September 13-15, 1988
PERFORMANCE CQMPARfSONS
The work by Short covered three types of shell side
baffling: segmental, orifice and disk/donut. Figure 7 shows
typical baffle dimensions. Table 2 provides additional
information. For the Short experiments water was the tube side
fluid and several rent oils as welJ as water were used for
shell side fluids. Representative data are shown in Figure 8; for
the cases shown only the baffling arrangement was changed.
For the water-water system, a comparison between
segmental and disk/donut baffling systems is shown in Figure 9.
The heat transfer coefficients obtained for both types of baffles
are similar, but the disk/donut baffles consume much less
pressure drop. Conversely, for the same pressure drop the heat
transfer coefficient for disk/donut is about 55% higher.
SUMMARY AND CONCLUSIONS
While it is clear that more experimental research is
needed before generalized design models can be developed, the
Short data show convincingly that there can be energy savings
possibilities when disk/donut baffles are specified for shell and
tube heat exchangers.
Disk/donut baffles can be used for a great variety of
services as alternatives to both segmental and double segmental
baffles. They can be very effectively used with finned tubes.
The Short data plus some proprietary information have permitted
intial insight, and analytical calculations confirm the disk/donut
baffle superiority. More systematic data are needed for full
development and utilization of this most promising concept in a
field of practical heat transfer where innovative ideas are
urgently needed.
Until more definitive design models are available, based
on research yet to be done, the approximate method of Donohue
(1949) appears to be the best available for predicting shell side
heat transfer coefficient when disk/donut baffles are used.
LITERATURE CITED
Bell, K. 1., "Final Report of the Cooperative Research Program
on Shell and Tube Heat Exchangers," Univ. of Delaware
Engineering Experiment Station Bull. No.5, January 1953.
Donohue, D. A., "Heat Transfer and Pressure Drop in Heat
Exchangers," Industrial and Engineering Chemistry, Vol.
41, No. 11, pp. 2499-2511 (1949).
Halle, H., Chenoweth, 1. M., and Wambsganss, M. W.,
"Shellside Flow-Induced Tube Vibration in Typical Heat
Exchanger Configurations: Overview of a Research Pro
gram," in Flow Induced Vibration 1986, S. S. Chen, J. C.
Simonis, and Y. S. Shin, eds., ASME Pressure Vessel and
Piping Division, Vol. 104, pp. 161-169, (1986).
Kern, D. Q., Process Heat Transfer. McGraw-Hill, New York,
1950.
McKetta, 1. J., "Hydrocarbon Process fndustry: Capital and
Other Expenditures," Encyclopedia of Chemical
and J. 1. McKetta, Ed., Vol. 26, pp. 374-390,
Marcel Dekker, New York, 1987.
Mueller, A. C., and Chiou, J. P., "Review of Various Types of
Flow Maldistribution in Heat Exchangers," in Maldistribu
tion of I;low and Its Effect on Heat Performance,
1. B. Kitto, Jr., and J. M. Robertson, Eds., ASME Heat
Transfer Division, Vol. 75, pp. 3-16 (1987).
Palen, J. W., and Taborek, 1. 1., "Solution of Shell Side Flow
Pressure Drop and Heat Transfer by Stream Analysis
Method," Chemical Symposium
Series No 92, W. R. Martini, Ed., Vol.. 65, pp. 36-52
(1969).
o 00
000
ft- (p
HAL'-f\IiooN . t $tU
\too:r.t. "u_Ud__"ln ..... " ' ........ II-.c....., .......011 ......
.. __r.u ",.I.j
4.0Dlll.HcI, . oCIDIQ.DUok 4,OD1ol1ol( 3.'Ot...
iOt'1{p ft- ft- l"t- j{p
D,stS- A1iODovc,I"INt,lT ! ....,
i"l-Ihp
fto;o. 0 \.'4
ef'H,8
ft - !fp. ;\ OLe.. C)f"It'tu ft it P
51"010 o.,U<.I l[)io O,.l,t,(A
t"'
Figure 7. Sizes and Iypes of baffles and sizes and arrangements of tubes as used in the tests by Short.
135
ESL-IE-88-09-26
Proceedings from the Tenth Annual Industrial Energy Technology Conference, Houston, TX, September 13-15, 1988
___
Short, B. E., Heat Transfer and Pressure Drop in Heat
Exchaneers, University of Texas Publication No. 4324,
Austin, Texas, 1943.
Taborek, J. 1., Sharif, A., "Effectiveness of Pressure Drop to
Heat Transfer Conversion for Various Shell-Side Flow Con
figurations," paper presented at the ASME/AIChE National
Heat Transfer Conference, Pittsburgh, August, 1987.
Tinker, T., "Shell-Side Characteristics of Shell-and-Tube Heat
Exchangers: A Simplified Rating System for Commercial
Heat Exchangers," Transactions, ASME, Vol. 80, pp. 36-52
(1958).
100 ...----...,.--.-""T'"-r-.--r..,..,."r'"---,-,..---,--,-"...-n
OR':;" Orlftce Bafrles
SEC ,- Segmental Bs ff1 es
DO'" Oisk/Donut baffles
10
N
C

. .0
Q.
o

c
.01 1..--..I..-........--L-...L...................._-...I---l_L.....I........L..4,1,1,-!00 10
2
1
FLOW RATE, 1bs/hr , 10-
3
Figure 8. Pressure drop and heat transfer rate for
three baffle systems. (Data of Shan, 1943).
136
10000,--,-,---,----,'---,-;,-'1---,'----,-,-,
.0
500Cf-
300C
200C_
0"'--- --
OISK,tloNUT ...-----1>
100C
BAFFLES
,--O......
500
300
200
100 '--_..1..--'-_.1--_-' 1_.J--'-_-'-_...J1'--_.L-.J------J
0.1 0.5 1.0 10 50
PRESSURE OROP. p' i
Figure 9. Heat transfer vs. pressure drop for different
baffling arrangements.
ESL-IE-88-09-26
Proceedings from the Tenth Annual Industrial Energy Technology Conference, Houston, TX, September 13-15, 1988

You might also like